Springer Series in Optical Sciences 140

Vesselinka Petrova-Koch
Rudolf Hezel
Adolf Goetzberger Editors

High-Efficient
Low-Cost
Photovoltaics
Recent Developments
Second Edition
Springer Series in Optical Sciences

Volume 140

Founding Editor
H. K. V. Lotsch

Editor-in-Chief
William T. Rhodes, Florida Atlantic University, Boca Raton, FL, USA

Series Editors
Ali Adibi, School of Electrical and Computer Engineering, Georgia Institute of
Technology, Atlanta, GA, USA
Toshimitsu Asakura, Toyohira-ku, Hokkai-Gakuen University, Sapporo, Hokkaido,
Japan
Theodor W. Hänsch, Max Planck Institute of Quantum, Garching, Bayern,
Germany
Ferenc Krausz, Max Planck Institute of Quantum, Garching, Bayern, Germany
Barry R. Masters, Cambridge, MA, USA
Herbert Venghaus, Fraunhofer Institute for Telecommunications, Berlin, Germany
Horst Weber, Berlin, Germany
Harald Weinfurter, München, Germany
Katsumi Midorikawa, Laser Tech Lab, RIKEN Advanced Science Institute,
Saitama, Japan
Springer Series in Optical Sciences is led by Editor-in-Chief William T. Rhodes,
Florida Atlantic University, USA, and provides an expanding selection of research
monographs in all major areas of optics:
• lasers and quantum optics
• ultrafast phenomena
• optical spectroscopy techniques
• optoelectronics
• information optics
• applied laser technology
• industrial applications and
• other topics of contemporary interest.
With this broad coverage of topics the series is useful to research scientists and
engineers who need up-to-date reference books.

More information about this series at http://www.springer.com/series/624

Vesselinka Petrova-Koch
•

Rudolf Hezel Adolf Goetzberger

•

Editors

High-Efficient Low-Cost
Photovoltaics
Recent Developments

Second Edition

123
Editors
Vesselinka Petrova-Koch Rudolf Hezel
Gate East University of Hannover
Garching, Germany Pullach, Germany

Adolf Goetzberger
Fraunhofer Institute for Solar Energy System
Freiburg, Baden-Württemberg, Germany

ISSN 0342-4111 ISSN 1556-1534 (electronic)

Springer Series in Optical Sciences
ISBN 978-3-030-22863-7 ISBN 978-3-030-22864-4 (eBook)
https://doi.org/10.1007/978-3-030-22864-4
1st edition: © Springer-Verlag Berlin Heidelberg 2009
2nd edition: © Springer Nature Switzerland AG 2020
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface to the Second Edition

The second edition of our book on High-Efficient Low-Cost PV appears exactly

10 years after its first edition. In the last decade, the global warming and the climate
change became stronger pronounced, the population on earth increased from 6.7
Billion in 2008 by almost one billion to 7.6 Billion in 2018, and also the demand of
electricity and of food worldwide. In the age of digitalization and e-mobility, the
electricity consumption will increase further, and clean electricity will become
mandatory if we care for the equilibrium of the climate on our planet. 100 renew-
ables do not have an alternative. Industries like the PV managed to grow fast in the
last decade, to reach globally 500 GWp! accumulated volume by the end of 2018.
Unfortunately, Germany and Europe, where the Renewable Energy Law EEG
was first introduced, and the PV industry pioneered, lost its leading position in PV
manufacturing and also in installed capacity. Wacker, headquartered in Germany,
remains a leader, and producing 80 000 t/year solar-grade polysilicon today is the
number one worldwide manufacturer. Also, US position is not much different from
this of Europe.
Asia took leading position and China alone manufactures today the biggest
fraction of the c-Si PV, where companies like Jinko and Trina Solar occupying the
top two positions in 2017 followed by Canadian Solar.
The second edition is reviewing the recent development the high-efficient
low-cost PV, of the next-generation PV cells and modules, which targeted in the
last decade single-junction cells with an efficiency 20 plus %, no matter if wafer
c-Si technology or thin-film technology, while decreasing the costs to 0.30 €C/Wp.
We cover recent cutting edge developments of c-Si wafer technologies like
PERC—Passivated Emitter Rear Cell; Bifacial—Dual-sided panels and cells; Multi
Busbar—Multi-ribbon and wire busbars; Split panels—using half-cut cells; Dual
Glass—Frameless; HJT—Heterojunction cells.
A new chapter reviewing the most widely used passivation layers SiN and
Al2O3 and their prospectives for the next-generation PV to come is offered in the
second edition.
The most recent development of III-V Heterojunction High-Efficient Cells for
Concentrator PV (CPV) is reviewed on its way to reach 50% efficiency.

v
vi Preface to the Second Edition

Among the thin-film technologies, we present a review of the development of

CIGS PV—on glass—and on flexible substrates, targeting 20 plus% efficiencies,
and chapter about Perovskite (inkjet printed). Perovskite was selected because it
became a hype in the last several years, and a promising candidate to reduce the
cost of PV below 30 €C/Wp in the near future.
A chapter about Augsburger Tubular PV (ATPV), which promised better solar
sharing between generation of Solar Electricity and Agro, is also presented. It is
time for a paradigm change in direction of affordable Agro PV.
The chapter devoted Fluorescing Collectors pioneered by Prof. Götzbereger in
1977 (more than 40 years ago) deserves further development and interest, and is
included in this edition.
During the 35th PC SEC Europe in Brussels in September 2018, targets like 30%
plus efficient tandem cells by 2022 and costs less than 10 €C/Wp in 2035 have been
put in focus. We only mention these developments to come but did not include in
the second edition on a c-Si tandem with perovskite.

Garching, Germany Vesselinka Petrova-Koch

Pullach, Germany Rudolf Hezel
Freiburg, Germany Adolf Goetzberger
Preface to the First Edition

Global warming and climate change are a result of excessive emission of green-
house gases in the recent decades. During the same period, we have experienced a
dramatic increase in energy demand to support the global industry and a “modern
life style”. These events, combined with the astronomic increase of the energy
prices since the beginning of the twenty-first century, make it obvious that the
global energy mix needs to be changed in favor of clean renewable energy sources.
At the same time, we must improve the energy efficiency of all technologies.
Solar energy conversion, particularly Photovoltaics (PV), represents one of the
most interesting and dynamically growing branches of the industry at the moment.
Although today the cost of PV is still relatively high, it has dropped considerably in
the past years and will continue to do so. A main driving force behind the devel-
opment of solar energy is its enormous, practically unlimited potential. Unlike other
renewable energy sources like wind or water, solar energy can be utilized every-
where on the globe. The cost of PV will have to continue to drop if the goal of
economic viability is to be reached. Prices of PV modules and systems follow a
well-determined learning curve. If present trends continue, it has been predicted that
grid parity (cost of PV equal to grid electricity at the consumer’s site) will be
reached in about 10 years in favorable locations. An important aspect of the
learning curve is that it includes not only scale effects of mass production but also
technical advances and new concepts. It can be shown that for most applications
high efficiency is a prerequisite for lowering cost. It is the purpose of this book to
give a survey of the most important technologies for increasing efficiency and
lowering cost of PV.
In the last five to six years the PV market has grown exponentially: not all over
the world yet, but in some countries like Germany and Spain in the EU, and Japan
in Asia. Japan and Germany may be considered as front-runners and engines of the
newborn PV Mega Industry. The German preferential feed-in tariff for solar
electricity and the new RES Law (called EEG in German) opened new market
dimensions and made it possible that the German market outgrew a country like
Japan, which was and is still the biggest PV producer worldwide. Germany reached
this state in spite of the fact that it is not the sunniest part of Europe, due to the joint

vii
viii Preface to the First Edition

efforts of politics, industry, and academia. About one-fourth of all PV modules

produced worldwide till the end of 2007 are installed in the southern part of
Germany (mainly in Bavaria and Baden-Wuerttemberg)! The exceptionally suc-
cessful EEG law also influenced many other countries. In the last 5 years, more than
30 countries copied and/or modified it. In this way, a PV network was formed. This
network spawned a very active PV development in several countries of the EU like
Italy, Portugal, Greece, and even new member-states like Bulgaria. PV is becoming
accepted also in countries like USA, Australia, and Russia, as well as in the newly
developing economies like China, India, and Turkey, which are on their way to sign
the Kyoto Protocol and to support the targets for Clean Energy.
So far, the PV market is dominated by the so-called first-generation PV, the
production and installation of which recently reached the gigawatt scale.
First-generation PV, which can be seen on many solar roofs today, is based mainly
on mono- and poly-crystalline Si, and these wafers are not made of low-cost
material. This generation is characterized by relatively low efficiencies of the cells
and modules, considerably below the theoretical limits and below results that have
been obtained in the laboratory. In spite of this fact, the demand for PV is increasing
exponentially so that several crises along the value chain of the PV industry have
occurred, for example, the shortage of solar-grade poly-Si feedstock since 2005, or
the one for solar glass, appearing on the skyline now. The crises of the Si feedstock
are causing a deviation from the learning curve. The fast growth of the PV demand
at one side and the shortage of Si at the other have generated not only supply
problems, but they also led to a change of perspectives and to a search for inno-
vative solutions, based on new materials, cells, and module systems. These new
solutions are based on reduced consumption of expensive semiconductor material
using, for example, thin-film PV technology (second-generation PV), or on
improved efficiency of crystalline cells and systems, while making use of
cost-effective concentrators for solar radiation (third-generation PV).
The aim of this book is to present some important trends and achievements that
are shaping the High-Efficient Low-Cost Photovoltaic (HELCPV) Industry.
Our intention here is not to deal with the basics of PV, which are covered
elsewhere. We also do not focus in this book on thin-film PV. Our ambition is to
present here the most important strategies and developments in the transition period
for the PV industry between 2000 and 2007 on its way to a gigawatt industry, and
to predict some of the tendencies of development of the near future, focusing on
innovative highly efficient wafer-based solar cell and innovative modules and
systems, making use of concentrated sun radiation.
The book consists of 13 chapters, written by 25 authors. In chapter 1, an attempt
is made to put together in a chronologic manner the milestones of the PV since the
discovery of the photovoltaic effect in 1839 and to focus the attention on the
important steps to develop a powerful and sustainable PV industry. In Chap. 2, a
comprehensive review of PV for space applications as a for-runner of the PV
Industry and its impact on the terrestrial PV is demonstrated. Chapter 3 focuses on
Preface to the First Edition ix

the development of the terrestrial PV from a niche market at the beginning of the
21st century to one of the most important mainstream markets for electricity at the
moment.
Chapter 4 reviews state-of-the-art poly-Si technology based on the Siemens
process, which remains the major technology also after the recent shortage for
poly-Si. In Chap. 5, the recent advances of the so-called ribbon technology, which
gives the unique chance to reduce the consumption of poly-Si during wafer man-
ufacturing, are described. In Chap. 6, an unambiguous answer to the question “Why
high efficient, bifacial c-Si PV cells?” is given, and a novel high-efficient rear
contact c-Si cell with bifacial sensitivity and elegant simple manufacturing steps is
described. Chapter 7 reviews the commercially available power c-Si PV cells, like
the Point-Contact Solar Cell of Sun Power, the HIT Solar Cells of Sanyo, and the
Buried-Contact Solar Cell manufactured by BP. This gives opportunity to the
reader to compare the concepts available already on the market with the one pre-
sented and analyzed in Chap. 6.
Chapter 8 reviews the development of III–V multi-junction PV cells with effi-
ciency close to 40%, on relatively less expensive c-Ge substrates. The ability of
such cells to operate under extreme conditions (more than 1,000 suns) is empha-
sized. Cost-effective concentrator systems based on Fresnel mini-lenses arrays,
prepared out of silicone, and sun truckers for such concentrator modules are
described here. In Chap. 9, the economic perspectives of the III–V concentrator PV
when manufactured on megawatt scale is discussed and compared to the
first-generation c-Si PV. It is argued why CPV should be considered as a promising
candidate on the market, capable to compete with the c-Si PV particularly in the
very sunny regions in the solar belt countries.
In Chap. 10, the attention is focused on an alternative device concentrating the
sun radiation, named Fluorescent Concentrator, which is less popular than the one
based on Fresnel optics, but which is unique in its ability to concentrate diffuse
radiation in a sophisticated way. The basics and the more recent advances to
develop further and to simplify this device are reviewed in this part. Chapter 11
deals with a concept of a novel hybrid photovoltaic/thermal concentrator. Chapter
12 points the attention then to some innovative PV installation concepts and future
PV applications.
Chapter 13 reviews the recent advances in the field of Organic Solar Cells and
deals with some design rules, helping to make the organic PV relatively high
efficient. We believe that OPV will continue to develop similar to the OLEDs and
will be able to supply the market with high-efficient low-cost PV in a very near
future.

Garching, Germany Vesselinka Petrova-Koch

Pullach, Germany Rudolf Hezel
Freiburg, Germany Adolf Goetzberger
To the Memory of Zhores Alferov

Zhores Ivanovich Alferov, distinguished Soviet and Russian scientist in the field of
semiconductor physics, winner of the Lenin and state awards, winner of the Nobel
Prize for Physics in 2000, vice-president of the Russian Academy of Science in 1991–
2017, Director of the Ioffe Institute in 1987–2003 passed away on March 1, 2019.
Zhores Ivanovich Alferov was born on March 14, 1930, in Vitebsk. In 1952, he
graduated from the Faculty of Electronic Engineering of V. I. Ulyanov (Lenin)
Electro-Technical Institute in Leningrad. In 1953, he went to work at the Ioffe
Institute, where, since the first half of the 50s, research on physics of semiconductor
devices was carried out. At that time, the task was to grow Ge single crystals (they
had not been produced by the Soviet industry) and create planar diodes and triodes
basing on them. Zhores Alferov joined to its accomplishment. He took an active
part in the development of the first Soviet transistors, and, in 1954, the domestic
power Ge devices were created, which quickly found numerous technical appli-
cations in various industries. Fulfillment of this task had required deep insight into
the physics of the processes in semiconductors and into technological processes for
their production. In those years, he worked as a physicist, as a technologist, and as
an engineer, introducing the new items into production: the devices were transferred
for mass manufacturing.
On the basis of a set of works (for which he received the first government award
in 1959), Zhores Alferov defended his Ph.D. thesis, which drew a line under his
10-year work. He faced the question of choosing a future direction of research. It is
needless to say that with development of Ge and Si valves, in the creation of which
Alferov made such a large contribution, the complex of physical and technical
problems associated with power semiconductor devices had not been exhausted.
And the experience gained by him allowed us to hope for further successful
developments in this field. Their start ensured further rapid progress.
According to Zhores Alferov, “The ability to choose promising areas of research,
courage in this choice, patience and perseverance in the work in the absence of
quick success in its conduct” characterizes his search strategy and his research work
itself.

xi
xii To the Memory of Zhores Alferov

In the early 60s, he addressed the problem of heterojunctions and began research
in this area with a small group of his employees.
Technological bases of the research were epitaxial methods allowing one to
grow single-crystalline layers of semiconductor materials with different chemical
compositions on an initial substrate. He and his group presented opportunity to
control such fundamental semiconductor parameters as the forbidden gap, the
electron affinity, the effective charge carrier mass, refraction index, and so on inside
one single crystal. Creation of perfect heterostructures could lead to a qualitative
leap in semiconductor physics and techniques. This had been understood by not
only Alferov but also scientists in many scientific centers abroad, who deployed
intensive study of heterojunctions. As promptly as by 1967, in the large number of
publications, it was judged on that defects and imperfections of real heterostructures
made it impossible to obtain and realize the effects, which had been expected in the
ideal case. In the same year, he and his staff obtained completely opposite result:
they created a heterostructure in the AlAs–GaAs system, which was close by its
properties to an ideal model.
The main thing in such a structure is not only that materials of a heterojunction
pair are characterized by the proximity of lattice parameters and chemical nature of
atoms comprising them, but also that the energy band structure of these materials is
optimal one from the point of view of electrical, optical, and luminescent properties.
This had determined the possibility to produce a wide range of semiconductor
devices and to create a set of experimental methods for studying physical processes
in heterojunctions.
Successful solution of a problem in heterojunctions by Zhores Alferov was
determined by a number of circumstances. Among them there was a physical
approach typical for him to phenomena being studied, his wide coverage of a
problem, his deep intuition, which helped him often to find optimal versions, and,
the last but not least, his ability to captivate a team of young employees with a quite
interesting albeit very difficult task. Discovery made by Alferov in ideal hetero-
junctions and in such new physical phenomena as the “super-injection” and also the
electronic and optical limitations in heterostructures has allowed improving dra-
matically parameters of the majority of known semiconductor devices and creating
fundamentally new ones, particularly promising for applying in optical and quan-
tum electronics.
This new stage in investigating heterojunctions in semiconductors was sum-
marized by Zhores Alferov in his doctoral thesis successfully defended in 1970. In
1972, he became a corresponding member and, in 1979, academician of the
Academy of Sciences of the USSR.
The fundamental research of Zhores Ivanivish formed the basis for a number of
new scientific and technical directions. Some of them went beyond his laboratory
and continued to be developed under his supervision in some industrial institutes.
Since then, a diversity of such devices with heterojunctions (lasers, light diodes,
light indicators, traffic lights, photodetectors, and sunlight converters) are being
produced by industry. Power diodes, transistors, and thyristors based on hetero-
junctions are also being introduced into production.
To the Memory of Zhores Alferov xiii

In using the technology of high-efficient radiation-tolerant solar cells based on

AlGaAs/GaAs heterostructures developed by Zhores Alferov in 70s, the Scientific
Production Association “Kvant” organized for the first time in the world large-scale
production of heterostructure solar cells for space arrays. One of such arrays
installed in 1986 on the basic module of the Space Station “MIR” had worked on
this space station for the entire lifetime without noticeable reduction in power.
Among works performed under the supervision of Zhores Alferov, special
attention was paid to the search of new heterostructures, to investigation of physical
processes on the interface of heterojunction materials, to application of
heterostructures in investigations of luminescence and quantum-dimensional
effects, and also to development of new principles for designing solid-state devi-
ces for electronics and integral optics.
On the basis of ideal heterojunctions in the InGaAsP compound proposed in
1970 by Alferov with his staff, semiconductor lasers operating in a substantially
wider spectral range than lasers in the AlGaAs system were created. They have
found a wide application as radiation sources in fiber-optic extended range com-
munication lines.
As a result of the Alferov’s investigation of charge carrier radiative recombi-
nation and of photovoltaic phenomena in heterostructures, conditions for obtaining
the 100% internal quantum yield have been established, what means practically
entire internal conversion of electric power into light one and vice versa. This has
opened new prospects in creating high-effective sources for coherent and sponta-
neous radiation and photovoltaic and other power converters. In the mid-70s,
Zhores Alferov together with his staff created spontaneous radiation sources, in
which, owing to the reduction of losses inside a crystal, it was managed to achieve
record values of the external quantum yield of 45% in the IR spectral range and
more than 10%—in the “red” one.
Investigation of processes of generation, propagation, and conversion of light
waves in waveguide heterostructures resulted in the development of new principles
for creating monolithic integral optical schemes capable to control light fluxes
effectively. The semiconductor direction in the integral optics based on
heterostructures, which is associated with the works of Zhores Alferov and his
employees, is now considered as one of the most promising in this area of physics
at the junction of modern optics and electronics.
In 80–90s, Alferov with his staff have proposed extensive investigation of solar
cells based on heterostructures, which resulted in the creation of photovoltaic
converters of the concentrated sunlight with the efficiency of greater than 30% at
sunlight concentration ratio of 100–500. These converters create a base for
developing works at a new level on space and terrestrial power engineering.
In the Center of Nano-heterostructure Physics of the Ioffe Institute organized by
Alferov, different techniques for fabricating heterostructures are being developed
and applied successfully: liquid, metal-organic, and molecular-beam epitaxy, which
has allowed creating a new generation of optoelectronic components based on
heterostructures. Among them, there are injection quantum-dimensional lasers of IR
and visual ranges with the record radiation conversion efficiency, single-mode
xiv To the Memory of Zhores Alferov

lasers with a distributed reverse bias, picosecond lasers and photodetectors, and
optoelectronic integral circuits for computers.
Starting from 1993, one of the main directions of the Center is the investigation of
properties of nanostructures of lessened dimension: quantum wires and quantum
dots. Proposed and realized were new ways for obtaining self-organizing nanos-
tructures with high uniformity in size and with crystalline perfection. Investigated
were processes of charge carrier capture, relaxation, and recombination in
nano-heterostructures. In 1993–1994, for the first time in the world, hetero-lasers on
“artificial atoms” (narrowband nano-dimensional inclusions of In, Ga, and As in the
GaAs matrix) are realized. In 1995, Alferov et al. demonstrate for the first time an
injection hetero-laser on quantum dots operating at room temperature through the
main state of quantum dots. For the first time, a continuous mode of operation of
such a laser was demonstrated. A fundamentally important event was widening
of the laser spectral range with using quantum dots on the GaAs substrates up to
1.3 lm, which is important for application in the fiber-optic communication. For the
first time, surface-irradiating lasers on quantum dots in the In-Ga-N system operating
in the UV range at optical pumping at room temperature were realized. Thus,
Alferov’s investigations laid a basis for fundamentally new electronics based on
heterostructures with a very wide range of application known as “band engineering”.
Contents

1 Milestones of Solar Conversion and Photovoltaics . . . . . . . . . . . . . 1

V. Petrova-Koch
1.1 Prehistory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Milestones of the Photovoltaics . . . . . . . . . . . . . . . . . . . . . . . . 2
2 PV as a Major Contributor to the ~100% Renewably Powered
World and Solving the Climate Battle . . . . . . . . . . . . . . . . . . . ... 9
Winfried Hoffmann
2.1 Today’s Energy Situation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Energy Efficiency Measures . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Future Energy Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Technologies and Market Development for Renewables . . . . . . 13
2.4.1 Photovoltaics (PV) . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4.2 Solar Power—CSP and CPV . . . . . . . . . . . . . . . . . . . 19
2.4.3 Wind Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5 Need for Storage as Solution to Variable Renewable
Energies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 20
2.6 The *100% Renewably Powered World . . . . . . . . . . . . . . ... 23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 30
3 Advanced Solar-Grade Si Material . . . . . . . . . . . . . . . . . . . . ..... 31
Karl Hesse and Ewald Schindlbeck
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 31
3.1.1 Metallurgical-Grade Silicon: Carbothermic
Reduction of Silica as Starting Point for Most
Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 32
3.1.2 Established Production Methods: Purification
of Metallurgical Silicon via the “Silane Route”
Is Dominating . . . . . . . . . . . . . . . . . . . . . . . . . ..... 33
3.1.3 Differences in Utilizing TCS or Silane
as Feedstock . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 34

xv
xvi Contents

3.1.4 Accommodation of the Processes to the PV

Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 36
3.1.5 The Myths of the “High Energy/High Cost” Rating
of Established Silane-Based Polysilicon Deposition
Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 38
3.1.6 Alternative Technologies for the Production
of Solar-Grade Feedstock: Purification
of Metallurgical Silicon via Melt Treatment/
Crystallization Is Dominating . . . . . . . . . . . . . . . . . .. 39
3.1.7 Alternative Vapour Phase Deposition
Technologies? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2 Quality Requirements from the PV Market . . . . . . . . . . . . . . . 40
3.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4 Silicon Nitride and Aluminum Oxide—Multifunctional Dielectric
Layers Crucial for the Progress of Silicon Solar Cells . . . . . . . . .. 43
R. Hezel and K. Jaeger-Hezel
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 Silicon Nitride Layers for Solar Cells . . . . . . . . . . . . . . . . . . . 45
4.2.1 Protection Properties . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2.2 Optical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.3 Passivation Properties . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.4 First Application of Silicon Nitride for Solar
Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2.5 Further Optimization of PECVD Silicon Nitride . . . . . 49
4.3 Plasma-Enhanced Deposition Techniques . . . . . . . . . . . . . . . . . 50
4.3.1 Parallel Plate Plasma Reactor . . . . . . . . . . . . . . . . . . . 51
4.3.2 Remote Plasma Reactor . . . . . . . . . . . . . . . . . . . . . . . 51
4.4 Silicon Nitride Passivated Bifacial Solar Cells . . . . . . . . . . . . . 53
4.5 Aluminum Oxide for Next-Generation Solar Cells . . . . . . . . . . 56
4.5.1 Al2O3—Based Rear Surface Passivation Scheme . . . . . 56
4.5.2 Early Laboratory Results for Al2O3 . . . . . . . . . . . . . . 58
4.5.3 Revival of Al2O3 as Passivation Layer for PERC
Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.5.4 Al2O3/SiNx Stacks for PERC Solar Cell . . . . . . . . . . . 60
4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5 High-Efficiency Industrial PERC Solar Cells for Monofacial
and Bifacial Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 65
Thorsten Dullweber
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 66
5.2 Industrial PERC Solar Cells . . . . . . . . . . . . . . . . . . . . . ..... 68
5.3 Bifacial PERC+ Solar Cells . . . . . . . . . . . . . . . . . . . . . ..... 76
Contents xvii

5.4 AlOx/SiNy Rear Passivation and Local Al Rear Contacts . . . . . 83

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells
and Modules in Mass Production . . . . . . . . . . . . . . . . . . . . . . . ... 95
J. W. Müller
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 96
6.2 Approaches for Efficient Cell Development and Production ... 97
6.2.1 Improved Statistical Resolution of Experiments
via Tra.Q . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 99
6.2.2 Root Cause Finding via Tra.Q . . . . . . . . . . . . . . . ... 99
6.3 Cell Efficiency Development of Q.ANTUM Solar Cells
on p-type Silicon Substrates . . . . . . . . . . . . . . . . . . . . . . . ... 100
6.4 Light-Induced Degradation of PERC . . . . . . . . . . . . . . . . . ... 104
6.4.1 Boron–Oxygen Defect . . . . . . . . . . . . . . . . . . . . . ... 104
6.4.2 Light and Elevated Temperature Induced
Degradation (LeTID) . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.4.3 LeTID Under Real Field Condition . . . . . . . . . . . . . . 107
6.5 Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7 Silicon Heterojunction Technology: A Key to High Efficiency
Solar Cells at Low Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 113
A. S. Abramov, D. A. Andronikov, S. N. Abolmasov
and E. I. Terukov
7.1 High-Efficiency Silicon PV Technologies Overview . . . . . . . . . 113
7.2 Key Features of SHJ Solar Cell Technology . . . . . . . . . . . . . . 116
7.2.1 SHJ Cell Technology at a Glance . . . . . . . . . . . . . . . . 116
7.2.2 Wafers for SHJ Cells . . . . . . . . . . . . . . . . . . . . . . . . . 117
7.2.3 Wafer Texturing and Cleaning . . . . . . . . . . . . . . . . . . 118
7.2.4 Transparent Conductive Oxides . . . . . . . . . . . . . . . . . 121
7.2.5 Metallization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.3 Conversion of “Micromorph” to SHJ Technology . . . . . . . . . . 123
7.4 SHJ Module Technology and Reliability Issues . . . . . . . . . . . . 125
7.4.1 Cell Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7.4.2 Cells Interconnection . . . . . . . . . . . . . . . . . . . . . . . . . 126
7.4.3 Module Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.4.4 Perspective Products . . . . . . . . . . . . . . . . . . . . . . . . . 129
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
xviii Contents

8 III–V Solar Cells and Concentrator Arrays . . . . . . . . . . . . . . . . .. 133

Zh. I. Alferov, V. M. Andreev and M. Z. Shvarts
8.1 Introduction—From Primary Heterostructures to III–V
Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 133
8.1.1 Single-Junction AlGaAs/GaAs Concentrator
Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 138
8.2 Multi-junction Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . .. 141
8.3 Lattice-Matched GaInP/Ga(In)As/Ge Triple-Junction
Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 143
8.4 Lattice-Mismatched (Metamorphic) Heterostructures
for Multi-junction Solar Cells . . . . . . . . . . . . . . . . . . . . . . . .. 145
8.5 Multi-junction Solar Cells: Current Status of High-Efficiency
Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 147
8.6 Concentrator PV Modules and Installations with III–V
Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 151
8.6.1 Design of Fresnel Lens Sunlight Concentrators . . . . .. 154
8.7 Module Efficiency Improvement . . . . . . . . . . . . . . . . . . . . . .. 157
8.7.1 Compensation of Chromatic Aberration Negative
Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 157
8.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 164
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 165
9 CIGS Thin Film Photovoltaic—Approaches and Challenges . ..... 175
F. Kessler, D. Hariskos, S. Spiering, E. Lotter, H. P. Huber
and R. Wuerz
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
9.1.1 Brief History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
9.1.2 Structure of a CIGS Solar Cell . . . . . . . . . . . . . . . . . . 177
9.2 Device Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
9.2.1 CIGS Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
9.2.2 Back and Front Contact . . . . . . . . . . . . . . . . . . . . . . . 189
9.2.3 Buffer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
9.2.4 Substrate-Related Issues . . . . . . . . . . . . . . . . . . . . . . . 197
9.2.5 Series Interconnection of Cells . . . . . . . . . . . . . . . . . . 198
9.3 Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
10 Perovskite Photovoltaics: From Laboratory to Industry . ........ 219
D. Forgacs, K. Wojciechowski and O. Malinkiewicz
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
10.2 Perovskite Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
10.2.1 General Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
10.2.2 The Role of the Metallic Cation . . . . . . . . . . . . . . . . . 223
10.2.3 The Role of the Halide Anion . . . . . . . . . . . . . . . . . . 225
10.2.4 The Role of Monovalent Cations . . . . . . . . . . . . . . . . 226
Contents xix

10.3 Processing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

10.3.1 Solution Processing . . . . . . . . . . . . . . . . . . . . . . . . . . 229
10.3.2 Evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
10.3.3 Hybrid Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
10.4 Device Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
10.4.1 N-I-P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
10.4.2 P-I-N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
10.4.3 Tandem Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . 235
10.5 Device Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
10.5.1 Charge Generation and Transport . . . . . . . . . . . . . . . . 237
10.5.2 Loss Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
10.5.3 Device Measurement and Hysteresis . . . . . . . . . . . . . . 240
10.6 Scale up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
10.6.1 Commercialization Bottlenecks . . . . . . . . . . . . . . . . . . 242
10.6.2 Target Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
10.7 Advancements by Saule Technologies . . . . . . . . . . . . . . . . . . . 246
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
11 Augsburger Tubular Photovoltaic (ATPV) . . . . . . . . . . . . . . . . . .. 257
V. Petrova-Koch and J. Mayer
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 258
11.2 Shaped PV Versus Flat PV . . . . . . . . . . . . . . . . . . . . . . . . . .. 259
11.2.1 Advantages of the Cylindrical Tubular
PV Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 259
11.2.2 Comparison Between an Array Cylindrical
Glass Tubes and a Pair of Flat Glass Plates
(Without PV Cells) . . . . . . . . . . . . . . . . . . . . . . . . .. 261
11.2.3 Cell Designs and Semiconductor Materials, Suitable
to Serve Tubular PV . . . . . . . . . . . . . . . . . . . . . . . .. 263
11.3 Power Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 267
11.3.1 Solyndra Power Tubes, the Forerunner
of the Tubular PV . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
11.3.2 Koch Power Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . 267
11.4 Tubular Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
11.4.1 ATPV Versus Solyndra Modules . . . . . . . . . . . . . . . . 269
11.5 ATPV Rooftop Installation Versus Rooftop Installation
with Flat PV Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
11.6 Agro-PV Under an ATPV Power Pergola . . . . . . . . . . . . . . . . 273
11.7 Other ATPV Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
11.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
xx Contents

12 Fluorescent Solar Energy Concentrators: Principle and Present

State of Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 277
Adolf Goetzberger
12.1 Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 278
12.2 Concentrator Stacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 280
12.3 Light Guiding by Photonic Band Pass Mirrors . . . . . . . . . . . .. 282
12.4 Factors Determining Energy Efficiency of Fluorescent
Concentrators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
12.5 Theoretical Limits of Concentration and Efficiency . . . . . . . . . 285
12.5.1 Limit of Concentration . . . . . . . . . . . . . . . . . . . . . . . . 285
12.5.2 Limit of Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . 286
12.6 Improvements in Basic Design . . . . . . . . . . . . . . . . . . . . . . . . 287
12.6.1 Optical Concentrators at the Collector Output . . . . . . . 287
12.6.2 Combination of Fluorescent Collector with Large
Area Si-Solar Cell . . . . . . . . . . . . . . . . . . . . . . . . . .. 288
12.6.3 Combination of Fluorescent Concentrator
with Up-conversion . . . . . . . . . . . . . . . . . . . . . . . . .. 289
12.6.4 Combination of Collector Stack with Band
Pass Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
12.7 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
12.7.1 Results of the Initial Period . . . . . . . . . . . . . . . . . . . . 291
12.7.2 Recent Experimental and Theoretical Work . . . . . . . . . 292
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
Contributors

S. N. Abolmasov R&D Center of Thin Film Technologies in Energetics,

St. Petersburg, Russia
A. S. Abramov Ioffe Institute, St. Petersburg, Russia;
R&D Center of Thin Film Technologies in Energetics, St. Petersburg, Russia
Zh. I. Alferov St Petersburg Academic University, St.-Petersburg, Russia
V. M. Andreev Ioffe Institute, St.-Petersburg, Russia
D. A. Andronikov Ioffe Institute, St. Petersburg, Russia;
R&D Center of Thin Film Technologies in Energetics, St. Petersburg, Russia
Thorsten Dullweber Institut für Solarenergieforschung Hameln, Emmerthal,
Germany
D. Forgacs Director of Knowledge Management, Saule Technologies, Wroclaw,
Poland
Adolf Goetzberger Fraunhofer Institute for Solar Energy Systems, Freiburg,
Germany
D. Hariskos ZSW – Zentrum für Sonnenenergie- und Wasserstoff-Forschung
Baden-Württemberg, Stuttgart, Germany
Karl Hesse Wacker Chemie AG, Wacker Polysilicon, Burghausen, Germany
R. Hezel University of Hannover and ZAE Bayern, Pullach, Germany
Winfried Hoffmann Hanau, Germany
H. P. Huber Laserzentrum Hochschule München, Fakultät für angewandte
Wissenschaften und Mechatronik, Hochschule für angewandte Wissenschaften
München, Munich, Germany
K. Jaeger-Hezel Munich University of Applied Sciences, Pullach, Germany

xxi
xxii Contributors

F. Kessler ZSW – Zentrum für Sonnenenergie- und Wasserstoff-Forschung

Baden-Württemberg, Stuttgart, Germany
E. Lotter ZSW – Zentrum für Sonnenenergie- und Wasserstoff-Forschung
Baden-Württemberg, Stuttgart, Germany
O. Malinkiewicz Chief Technology Officer, Saule Technologies, Wroclaw,
Poland
J. Mayer Ledvance GmbH, Equipment Engineering, Augsburg, Germany
J. W. Müller Hanwha Q CELLS GmbH, Bitterfeld-Wolfen OT Thalheim,
Germany
V. Petrova-Koch Gate-East, Garching, Germany
Ewald Schindlbeck Burghausen, Germany
M. Z. Shvarts Ioffe Institute, St.-Petersburg, Russia
S. Spiering ZSW – Zentrum für Sonnenenergie- und Wasserstoff-Forschung
Baden-Württemberg, Stuttgart, Germany
E. I. Terukov Ioffe Institute, St. Petersburg, Russia;
R&D Center of Thin Film Technologies in Energetics, St. Petersburg, Russia
K. Wojciechowski Scientific Director, Saule Technologies, Wroclaw, Poland
R. Wuerz ZSW – Zentrum für Sonnenenergie- und Wasserstoff-Forschung
Baden-Württemberg, Stuttgart, Germany
Chapter 1
Milestones of Solar Conversion
and Photovoltaics

V. Petrova-Koch

1.1 Prehistory

Seventh century BC: In ancient Egypt the houses were built so that the solar radiation
could be collected during the day and used during the night.
Fifth century BC: The Greeks oriented their houses so that they could receive solar
energy in the wintertime to heat the buildings.
Third century BC: Archimedes used mirrors to reflect direct sun radiation and to
defend Syracuse from invasion by the Romans.
Second century BC: the first windows made out of transparent mica were inserted
in houses in northern Italy, with the aim to increase the use of solar radiation in
wintertime.
First century AD: the so-called heliocaminos started to be used. These solar baths
with big mica windows oriented to the south found their maximum application in
Italy around the fifth century.
Fourteenth century: the first solar law was introduced in Italy.
1767 in Russia: M.V. Lomonossov suggested the use of lenses to concentrate the solar
radiation.
1767 in Switzerland: Horace de Saussure discovers the amplification and increased
heat efficiency in Matjoshka-type fivefolded glass boxes.
1830 in South Africa: J. Hershel uses the first solar cooker.
Around 1830: H. Repton constructs the first (glass) greenhouses in Europe.

V. Petrova-Koch (B)
Gate-East, Schleissheimer str. 17, 85748 Garching, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 1

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_1
2 V. Petrova-Koch

1.2 Milestones of the Photovoltaics

1839: Alexandre-Edmund Becquerel, a young experimental physicist in France, dis-

covered the photovoltaic effect at age of 19, while assisting his father, experimenting
with electrolytic cells made up of two metal electrodes.
1873: W. Smith, working in the UK, discovered the photoconductivity of selenium,
which led to the invention of the photoconductive cell.
1883: Ch. Frits, an American inventor, described the first solar cell made from Se
selenium wafers.
1887: H. Hertz discovered in Germany that ultraviolet light altered the lowest voltage
capable of causing a spark between two metal electrodes.
1888: Ed. Weston receives the first patent for solar cells (U.S. 389124 and US3891-
25).
1904: W. Hallwachs discovered the photosensitivity in a copper/cuprous oxide pair.
1904: A. Einstein publishes his pioneering theoretical work on the photoelectric effect
(he received the Nobel Prize in 1921 for this work).
1916: R.A. Millikan provides experimental proof of the photoelectric effect.
1916: Y. Czochralski (Polish scientist) develops a new method to grow single-crystal
silicon.
1930: W. Schottky discovers a new cuprous oxide photoelectric cell.
1931: A.F. Ioffe guides a project at the Physico-Technical Institute in St. Petersburg
on thallium sulphide photocells, which reach a record efficiency at that time of more
than 1%. He submitted a proposal to the Soviet government concerning the use of
solar PV roofs for providing electricity.
1932: Audobert and Stora discover the photovoltaic effect in CdS.
1948: W. Schottky presents the first theoretical concept for semiconductor PV.
1951: at Bell Labs the first p–n junction was grown in germanium.
1953: D. Trivich publishes the first theoretical calculations on the conversion effi-
ciency of the solar spectrum with materials of different bandgaps.
1953: G. Pearson at Bell Labs begins research of Li-doped Si solar cells.
1953: D. Chapin; C. Fuller and G. Pearson realized a two cm2 Si solar cell with an
efficiency of 4% (published on NY Times cover page).
1954: D. Chapin, C. Fuller and G. Pearson improve the efficiency of a Si Cell to
6%; first AT&T solar cell demonstration in Murray Hill, NJ.
1 Milestones of Solar Conversion and Photovoltaics 3

1954: At Siemens in Germany, G. Spenke and his team develop an efficient method
for poly-Si growth: Scientists and experts from Wacker and TU Munich participate
in this work as part of a joint team with Siemens. The so-called Siemens Method is
the main technology for the production of solar and semiconductor grade Si.
1954: J.J. Loferski and Jenny at RCA reported a pronounced PV effect in CdS.
1954: The International Solar Energy Society (ISES) was founded in Phoenix, AZ,
USA.
1957–1959: Hoffmann Electronics achieves 8, 9 and 10% efficiency and develops
the grid contact, significantly reducing the series resistance of the device.
1954: J.J. Loferski and Jenny at RCA reported a pronounced PV effect in CdS.
1954: The International Solar Energy Society (ISES) was founded in Phoenix, AZ,
USA.
1970: Its headquarters was later moved to Melbourne, Australia, and in 1995 it was
moved again to Freiburg, Germany.
1978: R. Hezel introduced Silicon Nitride as antireflection and passivation layer
which is presently applied for almost all Commercial Silicon Solar Cells.
1980: M. Riel starts the famous 1,000 solar roof program in Zurich, Switzerland.
1980: BP enters solar business.
1981: The Fraunhofer Institute for Solar Energy ISE in Freiburg, Germany, was
founded by A. Goetzberger.
1981: Reflective solar concentrators used for the first time with solar cells at Ioffe
Institute St. Petersburg.
1982: The PV Production worldwide reaches the value of 10 MW.
1982: A 1-MW PV plant—built by ARCO Solar with 100 Dual-Axis trackers with
c-Si modules—goes online in CA, USA.
1983: The PV production worldwide exceeds 20 MW, and sales exceed 250 million
USD.
1983: International Science and Engineering Conference (the Asian ISE-
Conference) is started.
1984: M.A. Green and S. Wenham introduced the laser-grooved buried-contact solar
cell (LGBC).
1985: M. Green at the University of New South Wales, Australia, breaks the 20%
efficiency barrier for c-Si solar cell under one sun in the research lab.
4 V. Petrova-Koch

1985: R. Swanson founded Sun Power in CA with the goal to commercialize the
high-efficiency c-Si solar cell.
1985: R. Hezel and K. Jaeger introduced Aluminum Oxide as passivation layer for
a new generation of Silicon Solar Cells.
1986: ARCO Solar releases the first commercial thin-film power module.
1987: The Solar Challenge is inaugurated, and PV-powered cars race across Aus-
tralia.
1987: The Institute for Solar Energy Research (ISFH) is founded in
Hamelin/Emmerthal, Germany.
1988: The Centre for Solar Energy and Hydrogen Research (ZSW) is founded in
Stuttgart, Germany.
1989: V.D. Rumyantsev at Ioffe Institute, St. Petersburg, introduces a solar concen-
trator lens-cell system with radically reduced size.
1990: ARCO Solar is sold to Siemens and renamed to Siemens Solar.
1991: Nukem GmbH (later Schott Solar) built 1 MW pilot PV plant out of mono-
and bifacial MIS-inversion-layer solar cells, developed by the team of R. Hezel at
the University of Erlangen.
1991: M. Graetzel invents the dye-sensitized electrochemical solar cell. An efficiency
more than 10% was obtained within 5 years after the discovery.
1992: BP commercializes the Laser Grooved c-Si Solar Cell (patented by M.A. Green
and S. Wenham).
1994: NREL develops and demonstrates a two-terminal, high-efficiency
GaInAsP/ GaAs solar cell, which under 180 suns shows more than 30% efficiency.
The third generation CPV was born.
1997: The biggest PV roof with more than 3 MW is installed in Munich, Germany.
1997: Sanyo starts mass production of its high-efficiency HIT c-Si/ a-Si:H PV cells.
1998: SolarWorld A.G. is founded in Germany, the first vertically integrated PV
company.
1999: M.A. Green and J. Zhao achieved a record efficiency of 24.7% for laboratory
c-Si solar cells.
1999: Total PV power installed worldwide exceeds 1 GW.
2000: In April, Germany introduces the new Renewable Energy Sources Act or EEG
(German: Erneuerbare-Energien-Gesetz) (feed-in law), the law which, in 2008, was
translated into more than 40 languages. Germany becomes the largest PV market
worldwide.
1 Milestones of Solar Conversion and Photovoltaics 5

2002: First Solar Silicon Conference dealing with the crisis of Si feedstock was
organized by Photon in Munich, Germany.
2002: Cypress Corp. and Sun Power in the USA start pilot production of the high
efficiency c-Si Sun Power PV cell. Mass production is set up in the Philippines.
2002: Siemens Solar sold its solar activities to Shell Solar, in 2004 Shell Solar c-Si
solar cell activities were divested to SolarWorld.
2004: General Electric enters PV, after becoming owner of AstroPower.
2005: Sharp remains the biggest producer of PV cells worldwide.
2005: Q-Cells, which was founded in 2002, is the fastest growing PV cell producer
worldwide.
2005: Solyndra, the first manufacturer of tubular PV so far was founded in CA, USA.
2006: SolFocus in USA, Concentrix-Solar in Freiburg, Germany, and SolarTec AG
in Munich, Germany, start pilot-production of concentrator III–V PV (CPV ). CPV
Modules consisting of GaAs triple cells on Ge substrates with efficiency more than
35%, and Fresnel concentrator lenses made out of UV-resistive silicone, capable of
providing up to 800 suns.
2006: Wacker expands the production of solar-grade poly-Si in Burghausen, Ger-
many, up to 16,000 tons per year to become the second biggest company in this field
worldwide. The new investment is about 500 million e.
2006: First Int. Solar Glass Conference is organized by Photon in Munich.
2007: Hemlock announces its mega-expansion of poly-Si production up to 3,600
metric tons per year in MI, USA, and will start producing in 2010. The investment is
about $1 billion U.S. Hemlock remains the biggest poly-Si producer worldwide.
2006: InterSolar, the biggest International Solar Fair takes place for a 10th and last
time in Freiburg, Germany. In 2008 it will continue to operate in Munich.
2007: SunPower and Sanyo both announce the highest efficiency for mass produced
solar cell under one sun solar radiation of 22%.
2007: Al Gore and IPCC receive the Nobel Peace Price.
2007: UN-Conference devoted to climate change takes place in Bali.
2008: Q-Cells bypasses Sharp to become the world biggest PV cell producer.
2009: T. Miyasaka et al. demonstrated first Perovskite cells with 3.8% efficiency.
2009: Hemlock announced PolySi production for a Giga-Scale PV at Intersolar NA.
2010: Oxford PV Ltd. founded to develop and commercialize Perovskite photo-
voltaics.
6 V. Petrova-Koch

2010: ZSW announced efficiency record of 20.1% for CIGS cells (0.5 cm2 ).
2011: Sanyo HIT became Panasonic.
2011: Solyndra’s spectacular foul in the US. Other PV companies went bankrupt.
2011: Avancis and Lasercenter Munich University improve CIGS solar cell efficiency
by 15% with the first all-laser monolithical serial Interconnects and picosecond laser
pulses.
2012: Soitec and ISE Fraunhofer announced 36,7% efficient III-V CPV module.
2012: Hanwha acquired Q-Cells.
2012: First Solar demonstrated cost competitive CdTe modules with 17% efficiency.
2012: Siemens exits completely the Solar Business.
2012: 15.5% efficient flexible lightweight modules by MiaSole.
2012: H. Snaith and M. Lee demonstrated the first solid state perovskite PV cell.
2013: Heliatek in Germany achieved Organic PV Cells with efficiency of 12%.
2013: Perovskite PV cells reach 15% efficiency (O. Malinkiewicz and J. Burschka).
2013: Schott Solar exits PV manufacturing.
2014: Panasonic reports 25.6% world record for c-Si HIT cell technology.
2014: Perovskite cells reached 20.1% efficiency at KRCT (Korea).
2014: Sauletech, Poland pioneered first ink-jet printed perovskite PV.
2015: EPFL in Switzerland reports 21% world record efficiency for Perovskite cells.
2015: Solar Frontier reaches one 1 GWp manufacturing per year of CIGS PV.
2016: Panasonic HIT-Module reached 23.8% efficiency.
2016: Kaneka Corporation achieved 26.33% world record with a heterojunction cell.
2016: First manufacturing site of Oxford PV for c-Si/Perovskite tandem cells and
modules, located in Germany.
2016: UNIST (South Korea) reached 22.1% efficiency with Perovskite cells.
2016: UC Berkeley and Stanford reported first graded junction all perovskite PV.
2017: Panasonic HIT cell turns its 20th anniversary.
2017: Wacker Poly Silicon reaches 80 000 tones/year, to become the number one
manufacturer of solar-grade polysilicon (corresponding to a production of 16–20
GWp PV).
2017: Tesla started production of HE cells with Panasonic for solar roof tiles.
1 Milestones of Solar Conversion and Photovoltaics 7

2017: First industry-sized perovskite/silicon tandem cells produced at Oxford PV in

Germany.
2017: Hevel Solar started manufacturing of Hetero Junction HJT c-Si cells with
efficiency more than 22% and modules in Russia.
2018: Solar World Germany went bankrupt.
2018: Longi Solar (China), the biggest manufacturer of mono c-Si PERC cells
demonstrated a half-cut bifacial PERC modules (120 half cells) with 340 Wp.
2018: Hevel Solar and TFTE (Ioffe Institute) demonstrated 144 Half Cell HJT Module
with 460 Wp.
2018: Energetica in Austria announced a Next Generation Module Manufacturing
Gigafab (the biggest in Europa) with cells based on the Blue Chip Technology.
Chapter 2
PV as a Major Contributor to the ~100%
Renewably Powered World and Solving
the Climate Battle

Winfried Hoffmann

Abstract This paper aims to demonstrate in a comprehensive yet clearly arranged

and short digest how we are able to end an unfair, dangerous, environmentally
unfriendly, and increasingly expensive energy supply and usage towards a fair, clean,
and lower cost energy situation for everyone in the future. The important role which
Photovoltaics (PV) has to play is emphasized but also the necessary portfolio of
other renewables, notably wind, together with the large scale introduction of elec-
tricity storage. With today’s existing power purchase agreements for PV, wind and
also new nuclear power stations it is shown that already with current prices, renew-
able electricity is 5–10 times less expensive compared to new nuclear and clean fossil
power stations. Especially the economic argument will quickly drive the development
towards a 100% renewably powered world within this century and provide energy
at lowest price for everyone, highest security of supply everywhere and helping best
to solve the climate battle.

2.1 Today’s Energy Situation

In the years around 2015, the world’s energy situation is shown in Fig. 2.1 [1, 2]. The
primary energy (PE) was around 160 PWh (P = Peta = 1015 ) and consisted of ~83%
fossil and ~4% nuclear exhaustible energy sources. The contribution of renewables—
mainly firewood and hydro—was around 13%. The conversion of primary energy into
usable secondary energy (SE, also named Final Energy Consumption) is associated
with considerable losses and amounted to ~100 PWh. The highest loss originates
from conversion of fossil and nuclear primary energy sources into electricity and is
around 2/3 on an average level. The last column on the right shows the energy content
for what we really need: hours of light with a given intensity from a light source,
transport with a car from A to B, comfortable living in well air-conditioned houses,
power and process heat for the industry, and many more. The machines and devices,

W. Hoffmann (B)
Bergstrasse 48, 63456 Hanau, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 9

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_2
10 W. Hoffmann

Fig. 2.1 Primary (PE), secondary (SE) and end user energy (UE) in ~2015

which convert secondary energy into what we really need, have energy losses which
decrease the end user energy (UE) to ~50 PWh. There was an old rule of thumb for
the relation between these three energy forms: PE: SE: UE = 3: 2: 1. This relation
has today changed to 3.3: 2: 1. It is the goal of this paper to demonstrate that in a
renewably powered world this ratio will significantly change towards ~1.5: 1.5: 1.
The energy sectors within the SE on a global level are shown in Fig. 2.2 and
are 27% for mobility, 28% for low-temperature heat, 23% industry (process) heat,
9% industry electricity, and 13% rest of electricity. For an easy remembering the
fractions can be approximated to ~1/4 mobility, ~1/4 low-temperature heat, ~1/4
industry (process) heat, and ~1/4 electricity (total).

Fig. 2.2 Energy sectors

within secondary energy
globally (~2015)
2 PV as a Major Contributor … 11

2.2 Energy Efficiency Measures

Although energy efficiency measures have nothing directly to do with renewable

energies they are important as they will be able to increase the fraction of renewables
for a given contribution. Two types of efficiency measures are shown in Fig. 2.3: (1) to
use machines and devices which convert SE much more efficiently to the needed UE.
Examples are LED’s instead of light bulbs, electric cars instead of petrol-driven ones,
phase angle modulated power for the millions of pumps instead of constant power for
all flow rates, and many more. (2) to provide the specific needs with measures which
do not require SE; examples are proper insulation, thoughtful planning of future
urban districts to minimize traffic, etc. The arrow (3) shows what by convention has
been agreed, namely that the SE provided by renewable sources is equal to the PE (the
convention for measuring the primary energy content is either the Physical Energy
Content Method (PECM) or the Direct Energy Equivalency Method (DEEM); as a
physicist I would have equated the PE for electricity from a PV module—or other
renewable technologies—as the energy content within the solar energy received, i.e.,
for a 20% efficient solar module, producing 10 kWh electricity, the PE would then
be 5 times higher, i.e., 50 kWh—but let us continue with what has been agreed).
There is one important consequence with many of the new and higher efficient
devices: in most cases, they are more expensive to buy for the same service compared
with old and low-efficiency products (example LED vs. light bulb). Only the total cost
over the lifetime of the respective product shows that in most cases the service will be
cheaper with the new devices by taking into account lower energy consumption and
in some cases longer lifetime. So in future the decision for a product will be no longer
as easy as it was in the past by only comparing the investment cost—what is still
done in many cases today. Only the total cost over lifetime determines which product
delivers a less expensive service. The same argumentation applies also to renewable
energies as their specific investment is typically more expensive compared to old

Fig. 2.3 Energy efficiency

measures (1 and 2) and
removing the losses (3) from
PE to SE to obtain secondary
energy with renewables only
12 W. Hoffmann

fossil and nuclear technologies but they have no fuel and other necessary running
costs and their cost over the product life cycle are already today less expensive.
It is a straightforward exercise with the examples given and some additional
obvious ones to conclude that with today’s available machines and devices together
with ongoing improvements a significant efficiency increase should be possible. It
has been shown in recent years by von Weizsäcker et al. [3, 4] that an efficiency
increase up to a factor of 5 is feasible. In order to be more realistic, we may define
a factor of only 2.5. This would imply an annual decrease for the needed energy to
power the same service of 3.5% or 1.8%, if we would like to have this accomplished
in 2050 or 2075, respectively.
This “realistic” efficiency increase includes the change from oil for transportation
to electricity and electricity based fuels [hydrogen from hydrolysis of water, methane
by reacting CO2 with hydrogen (2H2 + CO2 = CH4 + O2 ) as well as the production of
methanol (CO + 2H2 = CH3 OH)]. With these assumptions, there will be a significant
change for the relative share of the energy sectors. In particular, the fraction of
electricity will show a substantial increase.

2.3 Future Energy Needs

The important question arises how many billion people we will have to satisfy with
energy. The United Nations [5] have made several scenarios for future development.
The medium one is shown in Fig. 2.4 with an interesting result. Until the 1980s we
had a steep increase for the 10 year average of annual population increase, there
is since the 1990s a decrease observed. With the assumed further decrease in the
coming decades as shown, there is good evidence that the global population towards
the end of this century could level off at around 10 billion people.

Fig. 2.4 History and future estimate for the global population [5]
2 PV as a Major Contributor … 13

One important aspect for the future is also to solve the existing inequity for
energy use globally in different regions. If we do not care for a solution we may
see emigration of nations at a level which will manifold exceed the immigration to
Europe in recent years. A simplified calculation is as follows: With today’s ~8 billion
people using ~100 PWh SE we have ~1/4 of people using ¾ of SE. From this, we
can conclude that 1 billion of people enjoying “high-quality of life” are using ~37.5
PWh of SE. Hence with business as usual and allowing 10 billion people the same
“high-quality life” we would have to provide 375 PWh of SE. With our “realistic”
energy efficiency measures (factor 2.5) this number decreases to 150 PWh.

2.4 Technologies and Market Development for Renewables

2.4.1 Photovoltaics (PV)

One of the most fascinating and quickest developing technologies in recent years has
been PV. After the discovery in the mid-1950s of the first Silicon solar cell it was first
used for powering at lowest cost satellites starting in the 1960s. The replacement of
small button cells in millions of calculators and other small devices by integrating
cheaper small solar cells in the housing of these systems happened in the 1970/80s.
This was also the time to power remote places like repeater stations for industry and
solar home systems in developing countries with the lowest cost of electricity. First
support mechanisms allowed the demonstration of grid-connected PV systems in
households (1,000 rooftop in Germany in 1990, 70,000 rooftop in Japan in 1994). It
was the introduction of the best market support program ever installed, namely the
feed-in tariff starting in Switzerland and Germany, where it was further optimized
in 2000 with the so-called “Renewable Energy Sources Act” (H.-J. Fell, H. Scheer
and others) providing for all renewable technologies an individual fixed payment
for each produced kWh from these systems over 20 years and allowing a fair return
of investment. This catapulted the PV market from only 0.3 GW in 2000 to 17
GW in 2010 and realistically towards 100 GW in 2020. This also because other
European countries first (and more than 50 countries worldwide afterward) adopted
this support scheme. As seen in Fig. 2.5 it was Europe which until 2011 dominated
the global annual installations and created an average market growth of 50% p.a.
Unfortunately, Europe experienced thereafter a strong decrease in annual market but
fortunately other regions, notably US, China, and Japan (part of APAC) helped the
global market to increase by 24% p.a. until 2015.
If only an average 14% growth is assumed for the time between 2015 and 2025
the annual and cumulative PV market is shown in Fig. 2.6. The annual market will
be around 100 GW in 2020 and 200 GW in 2025. With this, the cumulated market
of ~250 GW in 2015 will increase to 500 GW in 2019 and approach the first TW in
2023. Major drivers for this development will be China, India, US, South America
and later Africa, and also Europe again.
14 W. Hoffmann

Fig. 2.5 History of annual growth for PV market (SolarPower Europe [6])

Fig. 2.6 PV market projection in coming years (own research in 2016)

When I sketched the Fig. 2.6 in 2016 it was ambitious to state a market of 100
GW in 2020. But with the unexpectedly strong growth of the Chinese market of 50
GW in 2017, we reached already in this year almost 100 GW. Most associations and
market researchers quickly predicted at this moment more than 130 GW by 2020.
However, these people forgot the vulnerability of the PV market as long as it is
driven by state support. In the past 20 years this “vulnerability” was positive, i.e.,
more support measures were granted in many countries as anticipated, resulting in a
stronger growth compared with expectations. When China published in June 2018 a
2 PV as a Major Contributor … 15

cut of subsidies resulting in ~20 GW fewer installations for this and next year, we had
all of a sudden for the first time in 20 years the situation that most probably the annual
market in 2018 will be smaller than in 2017. Fortunately we have now additional
strong markets like India, Middle East, and others which will help to reach 100 GW
in 2020. I am confident that the market will grow as in Fig. 2.6 and the vulnerability
will fade away as we no longer need state support for our PV systems, because of
their competitiveness compared to new nuclear and clean fossil counterparts. This,
of course, needs a level playing field for all technologies (i.e., no or at least equal
subsidies for each one).
The most important driver for the market increase until today was the decrease
for PV module price, which contributed in 2000 ~70% and in 2015 still, ~40% to
the total installed system price. The price development for a globally traded mass-
produced product is best characterized by the so-called Price Experience Curve
(PEC) [7]. Here the logarithm of the average price is plotted versus the logarithm
of the respective cumulated market volume. Typically a straight line is obtained
with a product-specific slope. The slope gives the decrease in price in % when the
cumulative volume is doubled and is called Learning Rate (LR) or Price Experience
Factor (PEF). Figure 2.7 shows such a graph for PV modules starting in the 1970s
when at the cumulative volume of ~0.5 MW the module price was ~100$/W. 2015
we had 250 GW cumulative volume at prices ~0.5 $/W. The PEF is slightly above
20%.

Fig. 2.7 Price Experience Curve for PV modules (ITRPV [8])

16 W. Hoffmann

As seen in the graph the prices followed nicely the best-fit line until a cumulative
volume of ~2 GW was reached, which was in the year around ~2005. Thereafter, we
had even a slight price increase and then a downturn to below the straight line. The
understanding of this behavior needs no scientific explanations but a good under-
standing of the market mechanisms and how prices are formed. Until 2000 the small
PV industry was able to obtain the polysilicon for the wafers from a scrap of the
electronic industry. In 2005 was the time when all existing polysilicon production
capacities for the electronic industry were fully loaded, but with a quick market
growth there was higher demand compared to production capability. As it takes a
few years until complex chemical companies can be operational, the price for polysil-
icon skyrocketed and the price for modules increased in these times of under capacity
along the complete value chain. This was the advent of Chinese production compa-
nies. The Chinese state government had just made the decision to heavily invest in
the production of goods with high future global potential (wind energy, PV modules,
batteries, etc.). In only a few years Chinese companies could use ~40 billion $ invest-
ment money at very favorable conditions from the Chinese state bank to ramp up
the production capacity quickly with the newest production machines from Europe
and the US. In parallel many Chinese companies also went successfully public at the
New York Stock Exchange and by 2010 Chinese companies had taken over the lead
in global solar cell and module production. This capacity increase even surpassed
the 50% annual market growth by a factor of 2. Consequently in 2010 and later we
had an overcapacity of ~100% which resulted in significant price erosion as seen in
Fig. 2.7. The upside was that many new markets could kick start in these years but
the down side was a quick consolidation of the global production industry with heavy
losses in the years 2011/12. For the future years I wish for the sake of a healthful
industry a return to the straight line as indicated by the green stars in Fig. 2.7.
From a technological point of view the observed massive decrease of price
between the 1970s and today is due to a manifold cost reduction: efficiency increase
for solar modules, decrease for specific material cost, lowering the investment needs
along the value chain, automation, and high volume manufacturing. In the following
some examples are given:
• Efficiency increase
In the 1970s the efficiency for industrially produced c-Si solar cells was around
10% which resulted in ~8% module efficiency. This relative 20% reduction was
due to round wafers leading to a low package density. The cell architecture was
a simple BSF (back surface field)-back side and not passivated front side. Si-
Nitride as AR coating resulted also in good passivation thereby increasing the cell
efficiency. In recent years the state of the art for solar cells was the introduction
of narrow screen-printed front metallization and PERC back structure (passivated
emitter rear cell), which gives for monocrystalline wafers efficiency of above 20%,
resulting in ~18% module efficiency. With Heterojunction (HIT) cells and Back-
contacted solar cells the efficiency for solar cells in mass production is well above
23% with module efficiency above 20%. It is important to highlight the increase
2 PV as a Major Contributor … 17

of efficiency for all applications, where area cost per installed power is important,
as this decreases inversely proportional.
• Decrease for specific material cost
One important material cost component for solar cells is poly-silicon. During the
~40 years the following achievements could be made:
# decrease of wafer thickness from ~700 µm to 150 µm (inside hole saw → wire
saw)
# decrease of kerf loss from ~500 µm to 100 µm (same as above)
The weight reduction could be decreased with these measures from 28 to 6 g/dm2
# Decrease for the price of poly-Si from ~60 to 15 $/kg
The material cost for poly-Si decreased then from ~1.68 to 0.09 $/dm2
Together with the efficiency increase from ~8 to 20% we obtain a reduction for the
specific material cost from 2.10 to 0.045 $/W, which is an impressive reduction
by a factor of ~50!
• Lowering the investment needs
Along the value chain for the production of solar modules the specific investment
needs for a production capacity of 1 GW p.a. could be decreased substantially in
the years from 2000 to 2018 as seen in Fig. 2.8.
Keeping in mind that new capacities are in the range of many GW’s and this
achievement is quite crucial.
• Automation and high volume manufacturing
In the early days the capacity for one production line in a 4 shift operation (24/7)
for solar cells was in the range of ~5 MW p.a. without robots. Meanwhile, this
capacity has increased to ~200 MW p.a. using quick kinematic robots and highly
sophisticated in-line characterization for high uptime and highest yield.

Fig. 2.8 Development of investment needs for Si-feed-stock, waferWafer-, cell- and module man-
ufacturing
18 W. Hoffmann

An interesting question arises as to the end of the shown price decrease in the
Price Experience Curve. If we assume for a single band-gap silicon solar cell the
following strongly simplified parameters for the OPEX cost:
• Feed stock Silicon at 12 $/kg and a 200 × 200 mm2 wafer (thickness 80 µm, kerf
loss 70 µm), producing a back-contacted solar cell with an efficiency of 25% to
~2 ct/W
• Crystallization and other materials for wafer production using wire saws → ~1–2
ct/W
• For the cell production (Cu or Al as contact material) → 1–2 ct/W
• The module consists of 2 glass plates, each 1.4 mm → 4 ct/W
plus encapsulation material, etc. → 2–3 ct/W
finish by cables, back rails, etc. → 2–3 ct/W
In total, the OPEX cost amounts to 12–16 ct/W.
For a mass-produced component, one could envisage the OPEX cost to amount to
~80% of the total cost. Hence this total cost would be ~15–20 ct/W. This would imply
that in our Price Experience Curve the limiting price number would be—dependent
on the assumed margin—around 18–24 ct/W. This price would be expected at a
cumulative volume of around 24 TW, which can be reached with reasonable growth
rates—as shown in the last section—in a few decades from today.
It is often argued that with today’s (2018) spot market module prices approaching
already 20 ct/W the above estimates are too much on the high side. However, the
above numbers are based on full cost—including depreciation, financing cost and no
subsidies for energy, waste treatment, and other items. In addition, a healthy industry
must also have reasonable margins to be able to further invest in new and additional
capacities and to do the necessary R&D—together with institutes—to reach the
ambitious further cost reductions. Unfortunately, the observed low spot market prices
today are driven by overcapacity of the global sector leading to consolidation. Before
going bankrupt smaller companies sell often below their cost and larger corporations
being able to survive such difficult times invest in further cost reductions which—
hopefully soon—will allow a positive margin with today’s low prices.
Is this number really the ultimate cost and price? Fortunately not, as Research is
underway to increase the efficiency for c-Si wafer based solar cells even further by
using tandem structures which will allow efficiencies well above 30% with only a
small increase in the cell-associated cost, but a substantial decrease for the 8–10 ct/W,
stated above for the manufacturing of the module (glass, encapsulation, and finish).
If we increase the cell cost by 50% due to additional thin film layers for the tandem
structure we have 2–3 ct/W. However, if a very optimistic module efficiency of 35%
(instead the above stated 25%) is assumed, the module add-on cost is decreased by
a factor 1.4, i.e., ~6–7 ct/W. Using a kerfless wafer production could decrease the
wafer add-on from 3–4 ct/W from above to 2 ct/W. The total OPEX would decrease to
10–12 ct/W with a total cost of 12–15 ct/W. Hence the limiting price would decrease
towards 14–18 ct/W! These rather simplified thoughts underline the needs for further
Research & Development for quite some decades in order to reach these ambitious
goals. These numbers also demonstrate that it is not necessary—as PV is already
2 PV as a Major Contributor … 19

today the lowest cost electricity producing technology - and even dangerous for the
global producing industry—see above arguments - to discuss publicly prices of 10
ct/W in only a few years from now. A more bottom-up calculation for the future cost-
and price development has concluded that a fair price for modules at 10 ct/W can be
expected at around 2040 [9].
The given price level today and the one expected in the years to come, depending on
the growth rate, seen in the Price Experience Curve, have an important consequence
and any new technology entering the market place has two possibilities:
1. A higher price per power may be paid, if other and/or additional customer needs
can be served compared to the mainstream c-Si modules. Examples are flexible
thin-film modules, including organic cell structures, colored dye solar modules
or tubular solar modules, well suited for agricultural applications. These products
are normally said to serve niche markets—however, at a total market of 100 GW
now and 500 GW p.a. in a few years, a niche market of 10% from the total can
be an interesting market place to be. The resulting 10 and 50 GW represent a
module turnover of ~25 billion and ~100 billion e.
2. Any new product intending to enter the mainstream market can only be success-
ful at a lower price. Even if a new technology would be cheaper at the same
production level compared to c-Si technology it has to start at the beginning
with a small production volume resulting in a high cost and price. Consequently
it will be more and more difficult to replace c-Si wafer based products as the
mainstream product.

2.4.2 Solar Power—CSP and CPV

2.4.2.1 Concentrated Solar Power (CSP)

In the mid-1980s a total capacity of ~350 MW solar parabolic trough power plants
were built in the US which was seen for several years as the preferred technology
for large scale centralized solar power plants. The challenge to increase the steam
temperature for the turbines (increase of Carnot efficiency) could not be solved until
today. As a result the production cost for a kWh from these systems is still ~7 ect/kWh
[10], which is ~4 times higher compared to large PV plants in the same locations.
One big advantage which was already demonstrated is the storage of heat during
sunshine and storing for the hours after the sunset. This increases the operational
time of the system and utilizes the turbines much better. Experiments are ongoing to
improve the steam temperature by replacing the limiting heat exchange fluid (oil) of
today.
Another possibility is the use of Dish sterling units or large Power Tower systems
which receive concentrated light from 1,000 s of 2-axis tracking mirrors. Here the
temperature is very high and a high efficiency can be obtained. First commercial
20 W. Hoffmann

systems have been built and it will be interesting to see the development of the
associated cost/kWh.

2.4.2.2 Concentrated PV

Based on 40% efficient multi-junction III-V compound (GaAs) solar cells, typically
used for satellite powering today, highly concentrated PV modules with a 2-axis
tracking mechanism have been commercially built in recent years. The module effi-
ciency could be demonstrated to be above 30% solar light to electricity. As with all
concentrating devices they can only be used in places with high direct sunshine over
the year. Similar to the CSP systems it will be interesting to see the development of
the reachable cost/kWh.

2.4.3 Wind Energy

The technology for windmills has shown a remarkable development. The size for a
single machine in the 1980s was ~75 kW with a rotor diameter of 17 m and a hub
height of ~25 m. Today commercial systems have a power of ~7.5 MW with a rotor
diameter of 126 m and a hub height up to 140 m. A major drive for this development
is the strong dependence of rotor diameter and average wind speed (which is for a
given site higher at higher hub heights) on the power output (P):
P ~ (rotor diameter)2 and P ~ (average wind speed)3
Parallel to this technology development the annual market increased from ~3.8
GW in 2000 to 51.5 GW in 2014 [11] which gives a CAGR of 20% during this time.

2.5 Need for Storage as Solution to Variable Renewable

Energies

The most important types of renewable energies, PV and wind, are by nature variable,
however, increasingly predictable. Only a minor part has uncertainties, e.g., local
clouds decreasing the PV output in those shaded areas. With today’s weather satellites
there is high predictability at least for the day-ahead forecast. Nonetheless if we are
to provide all electricity needs of a country only with renewables we must provide
some storage for periods when no or too little sun and wind cannot provide the actual
need. In Fig. 2.9 the electricity load curve for Germany is shown (upper blue curve).
The area below corresponds to the annual electricity consumption which is ~600
TWh. If we oversimplifying, assume that this 600 TWh should be produced only
from PV and wind we would need ~200 GW wind and ~250 GW PV installations.
The annual energy would be ~400 TWh for wind and 250 TWh for PV. With this the
2 PV as a Major Contributor … 21

Fig. 2.9 Electricity load curve and residual load for Germany [12]

residual load curve (lower red dash-dotted line) is obtained which also shows that ~
half of the year we produce electricity at times when it is not needed (~150 TWh)
whereas the other half year we need electricity at times when there is no wind and
sun (~100 TWh). The good news is that we have more electricity produced from
wind and PV at times when it was not needed; hence we have energy available to
account for losses associated with storage, e.g., charging/discharging batteries and
conversion of electricity to hydrogen or methane (P2G, Power to Gas).
The ratio between wind and PV is very much dependent on location. For Ger-
many’s load curve we have the best balance for a power ratio of ~1:1. In this case,
we have a monthly average which is almost constant during the year, only in winter
months slightly higher (due to wind contribution) and matching well the German
load pattern. It is important to note that this pattern helps to overcome an old fear:
namely the need for seasonal storage, which is no longer necessary. With Germany’s
irradiation and wind profile this gives energy-wise a ratio of ~2:1 for wind: PV (~2
kWh/Wwind and ~1 kWh/WPV ). Other regions with different load profiles and dif-
ferent climatic conditions may have significant differences. For example Near and
Middle East regions have a pronounced load peak in summer times, much higher
irradiation (~2 kWh/WPV ) and less wind. Therefore, the power ratio would be in
favor of PV. In contrast we may look to Scotland with less sunshine compared to
Germany but much more wind; here the power ratio would be in favor of wind. Of
course specific regions have also additional renewable sources which can be used
to optimize the portfolio for the various renewables to reach the lowermost need for
storage.
The storage of electricity has a diversity of solutions: pumped hydro, chemical
batteries and also the long-term storage via P2G (power to gas: hydrolysis, methane
formation and, when needed, re-electrification with fuel cells). Pumped hydro is only
22 W. Hoffmann

in few countries in Europe a viable solution (Austria, Switzerland, and Scandinavia)

and in most others can only provide a small portion of the needed storage. Example
Germany: here 7 GW are installed and in the “old times” (2006) they produced ~4
TWh electricity. Compared to the needed storage of 100 TWh this will not be a
solution. Discussions to transport surplus electricity from Germany to Scandinavia
have the problem that additional high voltage grids have to be built.
Most likely the solution will be a combination of various technologies. This will
start with electricity storage in individual homes (several kWh with lead-acid or Li-
ion batteries) where the self-consumption of PV can be increased significantly from
~20% up to 80% (especially when an e-car is also present). For offices, stores, and
many SME’s (Small and Medium Enterprises) appropriate storage capacities may go
up to the several 100 kWh ranges. In a municipality there is still surplus electricity
which can be collected and stored in M- to GWh batteries (e.g., NaS, redox flow
batteries). Municipalities may run jointly a P2G unit for the small rest of long-term
storage needs.
It is interesting to analyze the potential of price decrease for kWh storage in
homes and SME’s. For this, it is useful to construct a Price Experience Curve for the
Li-ion battery cells, the most costly part of such a storage unit. The result is shown in
Fig. 2.10 for automotive kWh-units which has a Price Experience Factor of ~15%,
i.e., each doubling of the cumulative volume of sold battery cells will decrease the
price by ~15%. At the time the curve was made we had a cumulative volume at the
end of 2015 of ~40 GWh of sold Li-ion battery cells with a price of ~250 e/kWh.
The interesting question is—when could we expect a cell price of 100 e/kWh? For
this we extrapolate the straight line and obtain at ~1 TWh cumulated capacity this
envisaged price. The next question is—at which time would we like to have this
cumulative volume reached, bearing in mind that a meaningful growth rate should
be in place. If we assume that in 2030 (or 2025) we would like this price, we can
calculate the necessary growth rate, which is 24 (or 38) % p.a. These growth numbers
may seem ambitious but if we compare this with the growth rate for the cumulated

Fig. 2.10 Price experience curve for Li-ion batteries [13]

2 PV as a Major Contributor … 23

PV modules in the time 2000 and 2010 the industry was capable to demonstrate 41%.
So this is a realistic goal to be reached.
To complete a battery we must add the assembly cost which is assumed to be
~25% of the cell price. The assumptions for a simplified calculation to obtain the
storage cost in ect/kWh are as follows:
• Lifetime of battery: 5,000 cycles
• Finance cost: 50% up to ~same as investment
• Usable capacity per cycle: ~90%
• → cost per kWh = (Invest + finance cost)/(5,000 × 0.9)
The storage cost is expected to decrease from ~15 ect/kWh today down to
~(4–5.5) ect/kWh in 2025/2030. Even if we have to add some electronic hardware
for the stationary house storage we have a very favorable price for a stored kWh
of PV electricity when compared to a purchased kWh from the utility. This price
development for battery storage also gives a very good prospect for the running cost
of future e-cars. If we assume a petrol-driven car with 5 l/100 km the cost is 7.5 e
at 1.5 e/l. An e-car using PV electricity with 15 kWh/100 km has a cost of only
(1.5–2.2) e which is 3-5 times less (PV (5–10)ect/kWh in southern and northern
regions, storage cost 5 ect/kWh).

2.6 The ~100% Renewably Powered World

Assuming realistic energy efficiency measures of a factor 2.5 from today to 2050 we
have a completely different energy picture in a 100% renewably powered world as
shown in Fig. 2.11. This also takes into account that the unfair energy distribution
of today is changed to a situation in 2050 where the global population has a similar

Fig. 2.11 Energy picture

from today towards a fair
and 100% renewably
powered world in 2050+
24 W. Hoffmann

Table 2.1 Technical and sustainable potential for various renewable technologies [14]

access to energy. Important to note is that primary energy equals SE, and the elec-
tricity makes up 80% of SE (120 PWh, which also contains the energy for mobility)
and the total losses to convert SE into EE are significantly reduced. This is obviously
very different to the situation as of today and results in the ratio PE: SE: EE = ~1.5:
~1.5: 1.
When it comes to the needed portfolio of the different renewable sources it is
important to understand what the respective technical and—even more important—
the sustainable potential is. A comprehensive review for this topic was done in a
detailed publication [14] and is shown in Table 2.1. The technical potential summa-
rizes all land based areas (and near coastal areas for offshore wind) which would
technically allow the utilization for a given renewable source. In contrast, the sustain-
able potential, which is only ~4% of the technical one, takes into account important
ecological, economical, and environmental aspects to be fulfilled. Fortunately the
150 PWh annual needs are only ~5% of the sustainable potential—hence there is
no obvious or unsurmountable hurdle to have the needs covered only by renew-
ables. Also important is the fraction for solar, wind, and all others (including hydro,
geothermal, biomass, etc.) of the sustainable potential which is 90%, 9%, and only
1%, respectively.
In order to account for local opportunities especially for “all others” we should
increase the share for our 150 PWh from 1% up to ~20% which would imply that 30
PWh should come from these sources. Some thoughts about why this is not likely to
increase significantly:
• Hydropower
The last 50 years had an almost linear increase of ~0.5 PWh per decade to 3.5
PWh today. Even if we assume the same capacity addition—which becomes more
difficult in the future—we arrive in 50 years at a cumulative capacity of ~6 PWh.
• Geothermal
2 PV as a Major Contributor … 25

There are limited places with high surface temperature. In addition, most places
have only a rather low steam temperature level for the turbine resulting in a low
Carnot efficiency. From today’s installed capacity of ~15 GW, producing ~0.09
PWh there is even with the globally available capacity potential of ~200 GW an
annual electricity production of ~1.2 PWh is possible [15]. Even if we double this
number, not more than 2–3 PWh will be produced from this technology.
• Bioenergy
If biogas is produced with the fastest growing plants and electricity made with
a 40% power station there is only ~20 MWh/ha electricity p.a. Compared with
PV in the sun belt 80 times electricity in the same area is produced, in northern
countries like Germany it is still 40 times more.
For biodiesel the situation is even worse: while in an area of 10 m2 with plants
for biodiesel a diesel-fuelled car can only drive 110 km (5 l/100 km), an e-car can
drive with PV generated electricity in the same area as far as 10,000 or 20,000 km
p.a. for northern areas and sun belt regions, respectively.
Nonetheless we have substantial amounts of bio-waste which should be used in
the best possible way, also to produce electricity—but a major portion of the 30
PWh is difficult to imagine.
• Ocean energy (wave and tidal)
Although the theoretical energy production from these sources is huge there are
severe practical limitations for harnessing cost-efficiently this energy resource.
Some favored areas may well be used for this technology but again it is difficult
to see a big part for the 30 PWh coming from here.
For wind, there is also good reason to increase the relative 9% towards ~20% of
our needed SE. Major reasons for this are:
• Wind and solar are balancing very well
• Wind onshore is cost-effective and has a good capacity factor (2–3 kWh/(Wwind inst
x a))
• Wind offshore has an excellent capacity factor (3–5 kWh/(Wwind inst x a)) and shows
the good prospect of low cost (just recently a PPA was awarded to Vattenfall at
<5ect/kWh)
• From today’s installed capacity of ~420 GW in 2015, the annual growth can be
smaller than 10% p.a. to reach the 30 PWh electricity production
For solar we will decrease the relative fraction from 90% down to 60% for our
150 PWh scenario. Three technologies should contribute in this area: (a) solar PV
(decentralized), (b) solar power centralized and (c) solar thermal (process) heat.
• Solar PV (decentralized)
The development of this fascinating technology was described in the preceding
section. The annex “decentralized” should cover all PV systems from the small
W-range (pico-PV) via the kW-installations (rooftop for houses) up to the MW-
and even GW-range (commercial rooftops, community owned green fields). The
question is whether we can reasonably assume that these systems are able to grow
in order to be able to supply for example 30 PWh of electricity let us say around
26 W. Hoffmann

Table 2.2 Necessary growth rates (simplified) to provide 20% PV for the future 150 PWh of SE

2050. Table 2.2 illustrates a quick exercise for such a growth: Coming from the past
two decades with a proven CAGR of 20 and 50% p.a. and as shown earlier that the
running decade is realistic to demonstrate 20% we only need much lower GAGR’s
for the coming three decades to reach a cumulated capacity of 23 TW for such PV
installations. If we take a conservative average number of only 1.3 kWh/(WPV inst
× a) these PV systems would produce in 2050 about 30 PWh electricity.
Of course, this simple approach does not take into account dismounting old instal-
lations after ~30 years and would assume an unrealistic stop of installations in 2051
with a gradual increase again as in the years 2010–2015. More realistic is a logistic
growth curve to reach asymptotically the 23 TW as shown in Fig. 2.12 [16]. Here the
increase for PV systems is shown for the major regions worldwide with adjusting the
respective growth to the local situation and the relative share is taken in proportion
of the expected population [5] in the various regions.
In order to arrive at the cumulative capacity, the annual market growth is shown as
a black dotted line (right ordinate). The peaks and valleys derive from the assumption
of an average lifetime of the PV systems of 30 years. In the steady state after ~2040,
the annual installations are in the range of ~800 GW per year (±100). From our
today’s ~70 GW in 2016 we have still a way to go but as we have neither material
limitations nor manufacturing bottlenecks this ambitious goal could be well reached,
given the track record of the last decade.
There is another very important point why this PV growth will really happen. If the
increasing PV installations could only be achieved with subsidies and support against
less expensive alternatives, I would not dare to make such projections. However,
with recently published power purchase agreements (PPA’s) for new nuclear power
stations and new PV systems, we are now able to compare the price for electricity
for the alternatives in a fair manner.
For new nuclear we refer to the agreed contract between Great Britain and France
to build two Areva EPR 1,600 MW nuclear reactors with a planned investment of
2 PV as a Major Contributor … 27

Fig. 2.12 Logistic growth curve to reach the 23 TW-PV installations (=~30 PWh electricity p.a.)

26.5 bn e. EdF will only build these reactors if GB accepts a guaranteed price for
each produced kWh with the following conditions:
(a) 92.5 BP/MWh = ~12 ect/kWh,
(b) duration 35 years after start of operation,
(c) adjustment according to inflation, starting with the reference year 2012,
(d) expected completion 9.5 years after start of work.
With these numbers the price to be paid (in today’s currency) is calculated and shown
in Fig. 2.13 for two inflation rates assumed (1 and 2.5% p.a.) and an expected start
of operation in 2027 (earliest start in 2017 plus 9.5 years). Although inflation rates
were rather low in recent years, it is expected that in the long run, an average inflation
rate will be around 2%. With this the first kWh in 2027 has a price of ~16 ect/kWh
and will increase towards ~35 ect/kWh at the end of the 35 years. The average or
“effective” price for this new nuclear power is ~25 ect/kWh (at 2% inflation rate).
In contrast to this agreement, there have been several PPA’s for new PV MW-sized
green fields recently signed. In several calls for tender in Germany for green-field
PV plants with a size of ~100 MW, the contracts were awarded (a) at and below 5
ect/kWh guaranteed price, (b) duration 20 years, (c) no adjustment with inflation,
and (d) expected completion ~1 year. Another PPA was signed recently by Dubai
Electricity and Water Authority (DEWA) and Abu Dabi Future Energy Company
(Masdar) for a solar park at 2.99 $ct/kWh with similar boundary conditions as in
Germany. Another PPA was signed in Chile at 2.91 $ct/kWh. The lower PPA in Dubai
and Chile reflect the differences of irradiation for Germany (~1 kWh/WPV ), Dubai
(~1.9 kWh/WPV ) and Chile (~2.3 kWh/WPV ). In the case of PV we can realistically
assume a lifetime of ~30 years. If we conservatively take ~1 ect/kWh for maintenance
and repair for the 3rd decade for those PV systems we can calculate the “effective”
price for the PV systems in Germany as follows: two decades with ~7 ect/kWh and
28 W. Hoffmann

Fig. 2.13 Price development for new nuclear (Great Britain) and new PV (in Germany and Dubai)
based on existing PPA’s (power purchase agreements)

one decade with 1 ect/kWh → “effective” price = (2 × 5 + 1)/3 = 3.7 ect/kWh.

Similarly in Dubai and Chile the “effective” price is ~1.8 ect/kWh. This calculation
is based from the point of view of a user for these systems; an investor running the
systems should remember that for the above mentioned PV PPA’s he will receive the
price after inflation as shown in Fig. 2.12.
In summary, we have already today the situation that with current prices a PV-
kWh in less sunny Germany is 5 times cheaper compared to a new nuclear power
station. This factor increases in sunnier regions like South America and Middle East
to more than 10. Whenever the financial world has understood this interesting result
the 100 s of billions of e investment money will no longer go into new nuclear but
in new PV (and wind, see above PPA for wind). There is no longer any meaningful
reason why a country should decide for new nuclear. Similar arguments are with
clean coal (CCS, carbon sequestration, and storage) as this will add cost and make a
clean fossil kWh also very expensive at above 10 ect/kWh.
Until a mandatory CCS will be in place there may be due to the Paris agreement
a possible move to a reasonable Carbon price. This started in Europe some years
ago with ~30 e/t-CO2 collapsed to ~5 and is today at ~20. It is expected that in the
years 2019–2023 this price will average at ~35–40 e/t-CO2 . With the goal to keep
temperature increase well below 2°C, we may anticipate that a price of 50 e/t-CO2
may become realistic. This would add a cost to fossil power plants of ~6 ect/kWh for
lignite, ~5 ect/kWh for hard coal and 2.2 ect/kWh for gas—plus running the power
plant and buying fuel—for sure more expensive than new renewables as demonstrated
before.
• Solar power centralized
While PV (and wind) systems up to the multi MW-range are well suited to cover
the energy needs regionally for private homes, SME’s, hotels and offices, the
2 PV as a Major Contributor … 29

question remains how to provide the necessary energy for the large and energy-
intensive industry. There has always been a tendency to move such industries
to regions with low energy costs, which in the past was predominantly in areas
with hydropower. Examples are metallurgical silicon plants in Scandinavia and
aluminum production from bauxite in Iceland. For all those countries lacking
hydropower a great opportunity could arise to keep or attract the industrial location
of such energy-intensive industries in their region if they have good solar (or wind)
potentials. Regions around the Mediterranean area, desert regions like Gobi in
China, Sahara in Africa, Atacama in Chile, large areas in Australia and many
more could provide multi TW power stations for lowest price electricity and heat.
First TW-PV plants are already built in China and many more could use either huge
standard PV plants (fixed plate), concentrated PV (2-axis tracking systems with
500x concentration using multi band-gap GaAs solar cells) and/or concentrated
solar power (parabolic trough, solar dish or power tower, which can provide not
only electricity but also process heat for industry). An interesting approach is a
hybrid PV and parabolic trough system at the same site to provide cheap continuous
power for industrial production companies: lowest price during sunshine with PV
and lowest price during the night with cheapest heat storage from the parabolic
trough converted to electricity by the turbines. Such hybrid plants are capable to
provide continuous power at ~5 ct/kWh, cheaper than anything else.
A quick exercise shows that with such large centralized systems energy of 30 PWh
in 30 years can be provided with reasonable growth rates: If we assume an average
size of 5 GW we need an annual growth of ~31% p.a. to reach 30 PWh. In the next
10 years 14 of such systems should then be built. Looking to the plans in India,
Africa, Middle East, and China for similar large scale systems this seems feasable.
• Solar thermal (process) heat
At the end of 2010 we had globally ~270 million m2 cumulative solar thermal
installations, with more than 70% installed in China. This corresponds to ~190
GWthermal and an energy equivalent of ~130 TWh. While most of these installations
are used to cover the low-temperature heat for warm water it is increasingly used
also for district heating (example Denmark).
An additional business opportunity for thermal systems will arise which is medium
temperature process heat for many industrial applications from SME’s up to big
companies. Whether the annual global energy needs for this technology segment
will approach 30 PWh remains to be seen; it all depends on how the customer needs
can be satisfied most efficiently in comparison with electricity based solutions from
a cost and convenience point of view.
In summary, we have summarized the portfolio of renewables to cover the future
SE needs of 150 PWh in the following Table 2.3.
The above split should only emphasize the relative importance for all renewables
and not be taken as a quantitative forecast for 2050. It could well be that Solar
PV becomes bigger and Solar power and/or Solar thermal smaller. But this does
not change the basic message from this exercise: namely that renewables will for
30 W. Hoffmann

Table 2.3 Portfolio and split of renewable technologies to cover 100% global secondary energy
needs
Fraction (%) Energy (PWh)
Solar PV (decentralized) 20 30
Solar power centralized (CPV and CSP) 20 30
Solar thermal (process) heat (CSP and solar thermal collectors) 20 30
Wind energy (on- and offshore) 20 30
All other (hydro, biomass, ocean tidal and wave, geothermal 20 30
etc.)
Total 100 150

economic and environmental reasons take over the global energy supply in the coming
decades from today’s fossil and nuclear technologies.

Acknowledgements I would like to thank C. Pillot from Avicenne for providing the market data
for Li-ion batteries and M. Fischer from Hanwha Q-Cells for calculating the logistic growth curve.

References

1. IEA (International Energy Agency), World Energy Outlook (2017)

2. Own research by Winfried Hoffmann
3. E.U. von Weizsäcker, A.B. Lovins, L.H. Lovins, Factor Four: Doubling Wealth–Halving
Resource Use: The New Report to The Club of Rome (Earthscan, London, 1998)
4. E.U. von Weizsäcker, K. Hargroves, M. Smith, Factor 5. Transforming the Global Economy
through 80% Improvements in Resource Productivity, (Earthscan, London, 2009)
5. UN (United Nations), World Population Prospects: The 2012 Revision, Highlights and Advance
Tables (Working Paper No. ESA/WP.228 2013)
6. SolarPower Europe, Global Market Outlook (2019)
7. W. Hoffmann, The Economic Competitiveness of Renewable Energy, (Scrivener publish-
ing/Wiley 2014) (chapter 7, ISBN 978-1-118-23790-8)
8. ITRPV (International Technology Roadmap for PhotoVoltaics), www.itrpv.net/Reports/
Downloads/2016
9. W. Hoffmann, A. Metz, Modules at a Price of 10 $ct/Wp - Dream or Reality? New Approach
for Future Cost and Price Predictions, (EU PVSEC, Marseille, 2019)
10. Personal information DLR institute (2018)
11. GWEC (Global Wind Energy Council) www.gwec.net
12. Hartmann, Dissertation Stuttgart (2012) and Jülich, Fraunhofer-ISE (2014)
13. C. Pillot, personal communication, avicenne, Paris (2014)
14. WBGU (Advisory Board to the German Government on global change), Flagship report 2011,
World in transition: a social contract for sustainability, ISBN 978-3-936-191-37-0 (2011)
15. Geothermal Energy Association, Annual US and global geothermal power production report,
(2016)
16. Data by W. Hoffmann and calculation of logistic growth curve by M. Fischer, Hanwha Q-Cells
GmbH (2015)
Chapter 3
Advanced Solar-Grade Si Material

Karl Hesse and Ewald Schindlbeck

Abstract Silicon-based technologies account for more than 90% of global PV instal-
lations. Since decades, scientists around the world have been searching for the right
technology to produce silicon feedstock for the photovoltaic industry, which fulfils
the needs for “low cost” as well as the needs for sufficient quality. The urgency of
this task has intensified since 1998, when it became clear that the PV demand could
no longer be met with by-products from the electronic industry.

3.1 Introduction

Silicon-based technologies account for more than 90% of global PV installations.

Since decades, scientists around the world have been searching for the right technol-
ogy to produce silicon feedstock for the photovoltaic industry, which fulfils the needs
for “low cost” as well as the needs for sufficient quality. The urgency of this task has
intensified since 1998, when it became clear that the PV demand could no longer
be met with by-products from the electronic industry. The established producers of
polysilicon developed specific products for the solar industry, and expanded their
production capacities significantly. Despite these efforts, a severe Si shortage was
observed in 2008 triggering very high spot prices for solar-grade silicon. New entrants
entered the field of polysilicon production to capitalize on fast growing demand of
photovoltaics. Production capacities were expanded at fast pace and the market situa-
tion changed from “undersupply” to “oversupply” rather quickly. As a consequence,
spot prices for solar-grade polysilicon continuously declined. Today, the polysilicon
market is characterized by a highly competitive environment. Successful polysilicon
producers must have the capabilities to produce high-quality polysilicon at lowest

K. Hesse (B)
Wacker Chemie AG, Wacker Polysilicon, Johannes-Hess-Straße 24, 84489 Burghausen, Germany
e-mail: [email protected]
E. Schindlbeck
Anton-Riemerschmid-Str. 40, 84489 Burghausen, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 31

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_3
32 K. Hesse and E. Schindlbeck

Fig. 3.1 Development of global polysilicon demand (kt). Source Analysis Wacker Chemie AG
September 2018 production pathways solar silicon feedstock

possible cost. Besides the established Siemens and fluidized-bed technologies, there
are still alternative production methods under discussion which potentially could
even be more suitable to serve the needs of the photovoltaics industry. Here require-
ments and possibilities have to be checked very carefully. Also old knowledge and
experience about the different possibilities to produce silicon should not be forgotten.
The future supply of silicon to the PV industry has to be seen in the tense environment
of quality, costs and time of implementation (Fig. 3.1).

3.1.1 Metallurgical-Grade Silicon: Carbothermic Reduction

of Silica as Starting Point for Most Pathways

Most pathways for the production of solar silicon feedstock are starting with met-
allurgical silicon (“silicon metal”, purity >98%) and ending with hyperpure silicon
(purity in the parts per billion range), so the challenge is to do this purification as
efficient as possible.
The silicon metal is produced in submerged electric arc furnaces (electrical con-
sumption between 10 and 30 MW) according to the formula:

SiO2 + 2C = Si + 2CO
3 Advanced Solar-Grade Si Material 33

Feedstocks are woodchips, high-purity quartz lumps and high-purity coal, char-
coal and coke.
The silicon melt has to be refined and solidified in the casting process to get a
multicrystalline structure and then crushed.
An important by-product is silica fumes, formed in by a side reaction according
to the formulas:

2Si + O2 = 2SiO

2SiO + O2 = 2SiO2

Owing to economy of scale and process improvements, the electric consumption

of 10–11 kWh/kg has already reached compared to more than 14 kWh/kg in the
past [1].
As the worldwide production capacity for metallurgical silicon is far more than 1
mio mto/a (most of it for the metal/aluminium industry), there is no supply bottleneck
to see in the future (cf. [2]).
Regarding the quartz as feedstock, it is certainly true that SiO2 is one of the most
abundant materials available on earth (as “sand”). But if you are looking for high-
purity quartz as it is necessary for high-purity metallurgical silicon, the availability
is much more limited to special mines, for example in Spain or Brazil, which is not
so much an issue of availability but costs (for mining and transport).
The purity of typical metallurgical-grade silicon with respect to metallic impurities
is approximately six orders of magnitude worse than that of typical solar-grade
silicon. So purification for the solar needs is inevitable.

3.1.2 Established Production Methods: Purification

of Metallurgical Silicon via the “Silane Route” Is
Dominating

The extreme purity required for silicon in photovoltaic or electronics applications is

achieved by converting metallurgical-grade silicon into silanes (large-scale industrial
processes are based on trichlorosilane or monosilane), which are then distilled and
decomposed to yield hyperpure silicon. The conventional method is the so-called
Siemens process in which TCS is decomposed in bell jar type reactors in the pres-
ence of hydrogen with the deposition of polysilicon on thin silicon rods [3]. The
silicon tetrachloride (STC) produced as a by-product is hydrogenated back to TCS
or is used as a feedstock for pyrogenic silica. The fluidized-bed CVD process used to
deposit polysilicon appears to be a very attractive alternative to the Siemens deposi-
tion process. Wacker has successfully developed a fluidized-bed deposition process
for granular solar-grade polysilicon based on TCS [4]. In contrast, two completely
34 K. Hesse and E. Schindlbeck

TCS-Synthesis Siemens type

Si+3HCl SiHCl3+H2 depostion

SiHCl3

STC-Conversion Fluidized-bed
Si+3SiCl4+2H2 4SiHCl3 depostion

SiCl4 Redistribution
4SiHCl3 SiH4+3SiCl4
Siemens type
depostion

ETHYL-Process
SiF4+NaAlH4 SiH4
SiH4

Fluidized-bed
depostion

Fig. 3.2 Pathways for polysilicon production

different processes use monosilane (SiH4 ) as the feedstock for deposition of polysil-
icon. As shown in Fig. 3.2, one of these processes uses TCS as a feedstock to afford
SiH4 , whereas the other process is based on SiF4 , a by-product from the production
of superphosphate fertilizers. The latter process was used by SunEdison on an indus-
trial scale until 2016. Semiconductor-grade monosilane can be used in a modified
low-temperature Siemens process or in a fluidized-bed deposition reactor to produce
polysilicon.
Approximately 94% of the global capacity of polysilicon used the Siemens process
and TCS as a feedstock in 2016 [5].

3.1.3 Differences in Utilizing TCS or Silane as Feedstock

Lobreyer and Hesse [6] demonstrated that comparable qualities of semiconductor-

grade polysilicon could be obtained from Siemens deposition of polysilicon, regard-
less of the type of silane feedstock (TCS or monosilane).
There are pros and cons associated with the use of both monosilane and TCS
to deposit polysilicon. Breneman and Dawson [7] published a detailed review dis-
cussing the key advantages and disadvantages.
The deposition of polysilicon from monosilane can be carried out at a temper-
ature as low as 650 °C, whereas temperatures >900 °C are necessary for TCS. If
monosilane is used, the main problem appears to be the production of silicon dust by
3 Advanced Solar-Grade Si Material 35

Metallurgical
HCl Silicon H2

Production Production
of HSiCl3 of POLYSILICON
Trichlorosilane Polysilicon

H2
HCl Hydrogenation
of
SiCl4 SiCl4

Fig. 3.3 Wacker polysilicon: closed loops for hydrogen and chlorine yield significant cost advan-
tages

homogeneous vapour deposition, which decreases yields and limits deposition rates;
it also negatively affects the product morphology (pores, density, microroughness).
If TCS is used, approx. 1/4 reacts to form silicon, 3/4 is converted into STC which
leaves the reactor in the off-gas. On the other hand, if monosilane is used, conversion
rates of nearly 100% are obtained and hydrogen is practically the only by-product.
The main pathway to produce monosilane is redistribution of TCS to monosilane and
STC; here ¼ of the TCS reacts to monosilane, ¾ is converted to STC; so the system
with monosilane deposition has to deal with the same amounts of chlorosilanes as
the TCS-based deposition system. There are also major differences in handling of
the individual silanes. The self-ignition and the very low boiling point of monosilane
necessitate extremely strict safety-engineering requirements and a complex cryodis-
tillation system.
Regarding the process complexity, the Wacker TCS process achieves a closed-
loop production system based on only three chemical reactions (Fig. 3.3):

mgSi + 3HCl → SiHCl3 + H2 metallurgical Si to TCS (3.1)

4SiHCl3 → Si + 3SiCl4 + 2H2 deposition of Si (3.2)
3SiCl4 + 3H2 → 3SiHCl3 + 3HCl hydrogenation of STC (3.3)
___________________________
mgSi → Si sum: purification of metallurgical silicon to hyperpure silicon,
closed loops of hydrogen and chlorine

So the decisive point appears to be how the polysilicon deposition process is

integrated into the manufacturer’s particular product flow. The objective here is to
either recycle all the by-products and waste products or use them to synthesize other
products to minimize losses of silicon, chlorine and hydrogen. It is also important to
use any waste heat arising from the various processes within the integrated system.
36 K. Hesse and E. Schindlbeck

SiH2Cl2
SiHCl3
Si SiCl4 Organosilane
Trichlorosilane Trichlorosilane
HCl Synthesis Distillation Production

SiHCl3
HCl RxSiHyClz

Tetrachlorosilane Chlorosilane SiHCl3

Conversion Distillation

Off-Gas
SiCl4 Condensation

Fumed Silica Polysilicon

H2 H2 Deposition
Production
O2 H2

SiO2 poly-Si

Fig. 3.4 Integrated product flow at Wacker Chemie AG, besides polysilicon the production of other
value-added products is possible

The integrated product system used at Wacker takes the additional advantage of
value-added products by linking the production of polysilicon to that of pyrogenic
silica, organofunctional silanes, silicates and silicones (Fig. 3.4).

3.1.4 Accommodation of the Processes to the PV

Requirements

An alternative to the “silane route” has been sought for many years, particularly for
the production of solar-grade silicon. The purity of solar-grade silicon with respect to
metallic impurities is approximately six orders of magnitude higher than that of typi-
cal metallurgical-grade silicon. It is questionable as to whether there is an alternative
to the “silane route” that can achieve this purity at acceptable costs. A very impor-
tant parameter in photovoltaics is solar cell efficiency (around 19–24% in 2018 [8]).
The highest cell efficiencies with low specific material usage are only possible via
high-quality feedstock. For the production of solar-grade feedstock Wacker Chemie
AG has optimized its existing semiconductor-grade production process taking into
account the requirements of the solar industry:
• The deposition process was optimized for maximum deposition rate.
• The purification of the surface by chemical cleaning or etching after crushing
of the rods could be avoided by a new crushing process. This process leads to
3 Advanced Solar-Grade Si Material 37

Silicon Silicon Silicon-

chips rod (slim)rod

Crushing

Solar grade silicon

(for solar cells) 1000°C

Crushing

HF/HNO3
etching
TCS + H 2

off-gas
Electrical current
Electronic grade silicon
(for semiconductors)

Fig. 3.5 Rod deposition: optimized process of deposition handling and crushing for the solar
Industry

lower contaminations, especially for metals, so that the solar requirements (metals
contaminations in the low parts per billion per atom (ppba) range) can be fulfilled;
the metal contaminations of cleaned prime electronic-grade polysilicon are two
orders of magnitude lower) (Fig. 3.5).
The fundamentals for a cost-effective production of solar-grade silicon at Wacker
polysilicon are:
• Highly integrated, but flexible production system: value-added use of by-products
• Economy of scale: investments, costs, productivity
• Advanced reactor technology: high output, yield, quality
• Flexible reactor technology: solar or electronic polysilicon according to demand
• Over 60 years of experience in polysilicon production.

Deposition of Granular Polysilicon with Trichlorosilane

A fluidized-bed deposition process might be a reasonable supplement to the “main
route” rod deposition technology (Fig. 3.6). In this process, a fluidized bed of silicon
particles is heated to the necessary decomposition temperature of the silicon contain-
ing gas which is passed through the bed. Ideally, elemental silicon is deposited on the
silicon particles that start growing. It is possible to operate this process continuously
by regularly withdrawing particles from the fluidized bed and adding smaller seed
particles to the bed. The withdrawn particles are already the ready-to-use product.
Seed particles can be obtained for example by milling of product granules. Besides
the advantage of continuous operation, the process has the potential of lower spe-
cific energy consumption. For customers granular polysilicon has a lot of advantages
38 K. Hesse and E. Schindlbeck

Fig. 3.6 Granular deposition

regarding handling, for high crucible filling grade (in combination with chunk mate-
rial) and for applications which need a finely tunable continuous recharging. Wacker
polysilicon is establishing this granular product to supplement its product portfolio.

3.1.5 The Myths of the “High Energy/High Cost” Rating

of Established Silane-Based Polysilicon Deposition
Technologies

There is a public understanding that the production of solar-grade feedstock via gas
phase deposition is “too costly” because of the “high” energy consumption of the
deposition process. And purification of silicon via metallurgical processes, that is,
different melting und crystallization steps, slag refining, treatment with different
gases and so on, is much less energy-consuming and therefore less expensive. But is
this really true?
1. Myth: High temperature = high energy consumption? The first misunderstanding
is that high temperature in a process equals high energy consumption. But if
there is no energy-consuming reaction involved, the consumed energy is mainly
influenced by the energy losses—a thermosbottle with hot fluid inside is also
hot inside without being “energy intensive”. It is a similar situation with the
gas phase deposition. Energy consumption can be influenced a lot by scale,
reactor design and process design. So the energy consumption figures given by
O’Mara et al. [3] are much lower now. Also, the general belief that monosilane
deposition is less energy-consuming than trichlorosilane deposition because of
the lower deposition temperature is not correct if you consider the maximum
3 Advanced Solar-Grade Si Material 39

possible deposition rate: Trichlorosilane deposition rate can be much higher

because heterogeneous monosilane deposition rate is much earlier limited by
homogenous gas phase deposition (i.e. dust formation). The faster the deposition
rate, the lower the energy losses. In addition, the high energy consumption for
monosilane cryodistillation and storage have to be taken into account.
2. Myth: The sum of chemical reactions, condensation and distillation steps of the
silicon purification via silanes is more energy intensive than “direct” metallur-
gical purification
Let us compare the trichlorosilane-based purification with typical metallurgical
purification involving several melting/crystallization steps (e.g. JFE-process, Apol-
lon solar process, [9, 10]) to reach the minimum required purity regarding metals
by segregation. At established polysilicon producers the total energy consumption
of the total TCS process is lower than 80 kWh/kg with a total silicon yield of much
more than 90% [11]. In metallurgical refining one melting/directed solidification step
of silicon needs about 10 [12] to 16 [13] kWh/kg for the target product. To reach a
similar metal purity as the TCS process, you need at least three melting/solidification
steps and still some elements cannot be efficiently removed due to their unfavourable
segregation coefficient. In addition, you have energy losses by the melt treatment, for
example, with slag refining, gases, electron beam guns (for elimination of phospho-
rus), plasma torches (removal of boron) and so on to bring the dopant concentrations
to an acceptable level. The remaining purity regarding carbon and dopants will still
be worse compared to the TCS process. And you have high yield losses by slag and
metal contaminated parts after segregation.
So there is no reason to condemn the “chemical” silicon purification methods in
favour of “metallurgical” methods with respect to the needed energy or yield.

3.1.6 Alternative Technologies for the Production

of Solar-Grade Feedstock: Purification of Metallurgical
Silicon via Melt Treatment/Crystallization Is
Dominating

If you look at metallurgical methods you will find that they need energy-intensive
melting and crystallization processes and special treatments for the removal of
dopants which are connected with yield losses. The achievable quality is still far
lower compared to gas phase deposition processes. The situation is similar to the
carbothermic direct reduction of pure quartz with pure carbon as well as for the melt
electrolysis of quartz. Other, already in the past evaluated and practised methods
like the reduction of silicon compounds with metals, for example, SiCl4 with zinc in
vapour phase or aluminothermic reduction proved to be not competitive in the past
regarding the achievable quality and costs. Important, limiting factors are the purities
of the feed materials, contaminations during the process steps, low yields of reaction
steps and complex reprocessing of by-products [14].
40 K. Hesse and E. Schindlbeck

3.1.7 Alternative Vapour Phase Deposition Technologies?

In a closed-loop chlorosilane system, the above-mentioned problems are already

solved in a large production scale; dominating technology is the heterogenous vapour
phase deposition on rods. Thereby maximum purity is achieved. For solar-grade sili-
con also fluidized bed deposition of granules is a possibility. Others tested the depo-
sition of silicon on hot surfaces with temperatures above melting point—“vapour to
liquid” deposition [9]. The extremely corrosive nature of fluid silicon is here limiting
the achievable quality. The lowest energy consumption in deposition is reached in
so-called “free space” reactors; here silanes are decomposed in a homogenous gas
phase reaction. But the resulting silicon is in the form of very fine particles (“dust”),
which was not usable in this form in the past because of the highly reactive, respec-
tively oxidized surface and the low density which led to melting problems and the
contamination problem during handling of the material.

3.2 Quality Requirements from the PV Market

As shown in Fig. 3.7 solar cell efficiencies are expected to increase further. High effi-
ciency approaches such as PERC, HIT or back contact cells are likely to gain market
share over the next years. In addition, the solar industry is working hard to improve
process efficiencies, yields and quality at each step of the value chain (ingot, wafer,
cell and module). Highest charge carrier lifetimes, highest yields and most efficient
production strategies are accessible with high-quality polysilicon feedstock. As a
consequence, high-quality polysilicon enables the PV industry to produce highest

Fig. 3.7 Expected development of solar cell efficiencies [8]

3 Advanced Solar-Grade Si Material 41

quality PV products with reliable long-term stability at lowest possible cost—one of

the key drivers for the success of photovoltaics also in the future.

3.3 Summary

TCS will remain one of the semiconductor and photovoltaic industries’ most impor-
tant feedstock in the future, regardless of whether used as a direct feedstock to produce
polysilicon in the Siemens and granulate processes or as a raw material for monosi-
lane production. For more than 60 years, the Wacker group is using and improving
polysilicon production based on vapour phase deposition of TCS in a closed-loop
process. Hydrogen and chlorine are processed in a cycle, and process heat is used
efficiently in the chemical plant system. This process is safe and environmentally
friendly and has plenty of potential for further improvement. Wacker Chemie AG is
operating three world-scale production sites for polysilicon with a total production
capacity of 80,000 metric tons in 2017 (Fig. 3.8).
In parallel, the downstream users are continuously improving their processes and
product qualities. Besides higher cell efficiencies and reliable long-term stabilities
(based on high-quality feedstock), advanced crystallization techniques, smaller wafer
thicknesses, lower kerf loss during sawing and overall higher process yields, there
is still huge potential to lower costs further—the basis to establish photovoltaics as
one of the major energy sources of our future (Fig. 3.9).

Fig. 3.8 Integrated polysilicon production of Wacker Chemie AG in Burghausen, Nünchritz (both
Germany), and Tennessee (USA)
42 K. Hesse and E. Schindlbeck

Fig. 3.9 Projected development of photovoltaics [15, 16]

References

1. A. Schei, J. Tuset, H. Tveit, Production of High Silicon Alloys (1998)

2. Ferroglobe presentation, Inaugural Investor Day, 17 October 2017, p. 8
3. W.C. O’Mara, R.B. Herring, L.P. Hunt (eds.), Handbook of Semiconductor Silicon Technology
(1990)
4. D. Weidhaus, E. Schindlbeck, K. Hesse, Silicon for the Chemical Industry VII (2004), p. 165.
ISBN 82-90265-25-5
5. Analysis Wacker Chemie AG, June 2017
6. T. Lobreyer, K. Hesse, Silicon for the Chemical Industry IV (1998), p. 93. ISBN 82-90265-20-4
7. W.C. Breneman, H.J. Dawson, Silicon for the Chemical Industry IV (1998), p. 101. ISBN
82-90265-20-4
8. ITRPV Roadmap 2018, www.itrpv.net
9. 3rd Solar Silicon Conference, 2, München. April 2006
10. 21st European Photovoltaic Solar Energy Conference, Dresden, 4–8 Sept. 2006
11. China PV Industry Association, China PV Roadmap (2016), p. 7
12. E. Sirtl, The WACKER approach to low-cost silicon material technology, in Proceedings 3rd
EC PV Solar Energy Conference, Cannes (1980), p. 236 ff
13. J.O. Odden, G. Halvorsen, H.M. Rong, R. Glockner, Comparison of the energy consumption
in different production processes for solar grade silicon, in Silicon for the Chemical and Solar
Industry IX (2008), p. 87. ISBN: 978-82-997357-5-9
14. Flat-Plate Solar Array Project Final Report, vol. II: Silicon Material, JPL Publication 86-31,
October 1986
15. W. Storm, Wacker Chemie AG, Photovoltaik – Vom Silicium zur Sonnenenergie, Presentation
Wacker Wissensforum, 4 April 2017
16. International Energy Agency (IEA), World Energy Outlook 2017. www.iea.org
Chapter 4
Silicon Nitride and Aluminum
Oxide—Multifunctional Dielectric
Layers Crucial for the Progress of Silicon
Solar Cells

R. Hezel and K. Jaeger-Hezel

Abstract The achievement of high efficiencies for crystalline silicon solar cells is
highly dependent on the reduction of carrier recombination at both cell surfaces.
As a new era in PV, this loss of charges can be minimized by simply depositing
silicon nitride or aluminum oxide layers at low temperatures. Due to their fascinat-
ing properties, both dielectric films became key components of present and future
monofacial and bifacial silicon solar cells. As a special highlight, by an innova-
tive aluminum oxide-based rear side scheme, high efficiency industrial cells (PERC)
could be developed which are generally regarded to become next-generation solar
cells.

4.1 Introduction

The achievement of high efficiencies for crystalline silicon solar cells is highly
dependent on the reduction of carrier recombination at both cell surfaces. For high-
efficiency laboratory solar cells thermally grown silicon dioxide (SiO2 ) layers have
been applied to obtain very low effective surface recombination velocities (SRV)
[1]. Despite thermal SiO2 has excellent passivation properties on phosphorous dif-
fused silicon surfaces as well as on n- and p-type silicon wafers it requires high
processing temperatures (>900 °C) and long processing time. This treatment can
severely degrade the bulk minority carrier lifetimeLifetime, particularly of lower
quality silicon material, e.g., multicrystalline silicon. Very clean production equip-
ment, expensive electronic grade gases, and thorough cleaning of the samples are
needed to avoid contamination and diffusion of unwanted impurities into the bulk.

R. Hezel (B)
University of Hannover and ZAE Bayern, Josef-Heppner-Straße 26, 82049 Pullach, Germany
e-mail: [email protected]
K. Jaeger-Hezel
Munich University of Applied Sciences, Josef-Heppner-Straße 26, 82049 Pullach, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 43

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_4
44 R. Hezel and K. Jaeger-Hezel

This is undesirable from production cost and throughput considerations [2]. Sim-
ilar problems arise with the implementation of a back surface-field (BSF) which
provides only moderate surface passivation. This fabrication step is still used for
the rear side of the present mainstream solar cells [3]. Furthermore, thermal SiO2
does not provide efficient reflection reduction because of its low refractive index.
Traditional antireflection layers such as TiOx or Ta2 O3 can be deposited in a rather
cheap low-temperature process but they do not provide good surface passivation and
delamination problems are reported.
In order to obtain reliable high-efficiency solar cells with a simple technology low-
temperature-processed dielectric layers had to be found providing excellent surface
passivation together with good optical and surface protecting properties.
Inspired by our research activity in microelectronics [4, 5] the two multifunctional
dielectric films, silicon nitride (SiNx ) and aluminum oxide (Al2 O3 ), appeared to us to
be also attractive for photovoltaics and in 1978 low-temperature surface passivation
of silicon solar cells could be first introduced at the University of Erlangen (FAU)
by these two dielectric materials [6–8].
As to silicon nitride, our pioneering work was starting with its first application
as a passivating antireflection layer for solar cells, followed by intensive basic and
applied research for further optimization [7]. In 1981 plasma-enhanced chemical-
vapor-deposited (PECVD silicon nitride) was introduced in our lab for silicon solar
cells [9]. In 2004 these activities were culminating in the development of a high-
throughput in-line machine for large-scale low-temperature plasma deposition of
silicon nitride [10, 11]. Providing high-quality SiNx passivation and antireflection
layers on an industrial scale was a decisive contribution to the first mass production
of reliable solar cells. Subsequently, the production volume increased dramatically.
Today, for worldwide 90% of the mainstream silicon solar cells plasma silicon nitride
layers are used. They are gaining even more importance for next-generation solar
cells.
As another highlight, in 1985 Al2 O3 layers were first applied by us for silicon
solar cells in order to substitute the conventional back surface-field to further reduce
carrier recombination at the rear side and their unique properties were investigated
[8, 12]. An innovative rear surface configuration with gridded base contacts and
Al2 O3 as passivation between the line-shaped contact fingers was established and
successfully implemented in silicon solar cells [8, 12–14]. Decades later this early
work was reintroduced and intensively studied in many research labs all over the
world [15, 16]. Our rear surface design was implemented in the so-called PERC
solar cells which are regarded as the next-generation silicon solar cells [3, 17, 18].
Both Al2 O3 and SiNx layers are applied for these devices. Mass production started
in 2013.
In this article, the fascinating properties of both dielectric films are reviewed and
their crucial role for present and future commercial silicon solar cells is outlined.
Among others, the early history and our pioneering contributions at the University
of Erlangen and later at the Institute for Solar Energy Research in Hameln (ISFH)
are extensively discussed.
4 Silicon Nitride and Aluminum Oxide … 45

4.2 Silicon Nitride Layers for Solar Cells

4.2.1 Protection Properties

Long-term stability is a prerequisite for the successful utilization of solar electricity.

Due to their outstanding protective properties, SiNx coatings turned out to be deter-
mining the ultimate reliability of solar cells. At the University of Erlangen much basic
work was done including physico-chemical analysis to characterize and understand
the properties of SiNx films.
By a combination of Auger electron spectroscopy (AES) and low energy electron
energy loss spectroscopy (ELS) quantitative information about the chemical bonding
states in thin amorphous SiNx films were obtained. The ground and excited states
of the Si–N bonds could be characterized and an electron level scheme for silicon
nitride was derived [19–22].
Electron and ion bombardment in the course of electron spectroscopy and scanning
electron microscopy revealed outstanding stability of silicon nitride with chemical
bonds much stronger compared to silicon oxide [23, 24]. This is demonstrated in
Fig. 4.1, where for thermal SiO2 and Si3 N4 the effect of a 3 keV irradiation on the
SiLVV Auger spectrum is shown as a function of the beam exposure time [24]. In

Fig. 4.1 a Effect of electron beam irradiation on the SiLVV Auger spectrum of SiO2 at Epr = 3 keV
and Ipr = 10 µA. b The p-p heights ratio A (92 eV)/B(76 eV) for various electron beam parameters
as a function of irradiation time. The ratio A(92 eV)/B(83 eV) of the SiLVV spectra of Si3 N4 is also
plotted
46 R. Hezel and K. Jaeger-Hezel

all cases in the SiLvv spectrum the irradiation results in the appearance of the 92 eV
line which is characteristic for elemental silicon (Fig. 4.1a). The p-p height A of
the 92 eV line relative to the height B of the main peak of 76 eV in the spectrum is
taken as a measure for the creation of “free” silicon by breaking of Si–O bonds and
subsequent electron-stimulated desorption (ESD) of oxygen (Fig. 4.1b) [25].
For the SiLVV Auger spectra of silicon nitride (not shown in Fig. 4.1) the p-p
height A of the 92 eV line—breaking of Si–N bonds and formation of Si–Si bonds—
relative to the height B of the corresponding main peak at 83 eV is also plotted in
Fig. 4.1b. As a striking result, in comparison to silicon oxide for silicon nitride an
almost negligible irradiation effect can be observed. Qualitatively similar results as
for the case of electron irradiation were obtained for argon ion bombardment [21].
According to these and other experimental results it can be concluded, that silicon
nitride layers, unlike SiO2 , can provide excellent protection of the surface and bulk
of silicon solar cells.
It is well known that silicon nitride provides excellent protection against contami-
nation by blocking all kinds of foreign species from diffusing into the sensitive silicon
substrate [26]. Already extremely small quantities are deleterious for the carrier life-
time and thus for the cell efficiency. As a positive exception, only hydrogen is mobile
in the nitride layer which is beneficial in minimizing charge recombination at the sur-
face and within the bulk of silicon by saturation of dangling bonds (see Sect. 4.2.3).
The excellent barrier of SiNx for moisture together with scratch and crack resistance
and low mechanical film stress have to be mentioned [5]. SiNx also serves as mask
against etching and high-temperature diffusion of dopants during device fabrication.
Progress in contact formation for solar cells could be achieved since SiNx proved
to withstand the firing of metal contacts through the already deposited surface film
[27].

4.2.2 Optical Properties

The power output of a solar cell is directly dependent upon the amount of incident sun-
light which is absorbed by the cell. Therefore, a major consideration in the design of
solar cells is the reduction of losses caused by reflection and shadowing. Minimizing
reflection losses from the non-metallized area of the cell is generally accomplished
by coating the surface with one or more highly transparent dielectric layers of lower
refractive index than the silicon itself and/or by texturing the surface of crystalline
substrates via chemical etching the wafers in anisotropic etchants resulting in an
array of pyramids. PECVD (Plasma-Enhanced Chemical Vapor Deposition) silicon
nitride acts as excellent surface-passivation and antireflection coating, a combination
which is not met by the standard coatings on silicon solar cells [28]. As a special
advantage of PECVD silicon nitride, the refractive index n can be varied over a wide
range (from about 1.6–2.6). It thus can be easily optimized for a single and multi-
layer AR coating by the CVD preparation conditions. The refractive index is linearly
correlated to the Si/N ratio of the films. For stoichiometric Si3 N4 n is about 1.95,
4 Silicon Nitride and Aluminum Oxide … 47

higher indexes are obtained if the films contain excess Si. On the front surface of
solar cells n should be close to the optimum of 2.0 in order to minimize optical losses
due to reflection and absorption. Light absorption by the nitride films in the visible
and near UV region of the solar spectrum can be kept very low. With silicon nitride
on a textured silicon surface reflection of sunlight below 3% is achieved making the
cell appearing dark.

4.2.3 Passivation Properties

One of the most outstanding advantages of the thin multifunctional PECVD sili-
con nitride layers is their ability to drastically reduce the loss of light generated
charge carriers on front and rear side as well as in the bulk of the solar cell and thus
considerably increasing cell efficiency.
Surfaces of silicon represent rather severe defects in the crystal structure owing
to non-saturated (“dangling”) bonds. These “surface states” act as recombination
centers for the light generated electron hole pairs hence reducing the number of charge
carriers. Passivation, i.e., minimizing surface recombination is thus a prerequisite for
achieving high efficiency solar cells. Extrinsic silicon surface passivation is based
upon the Shockley-Read-Hall theory [12, 29]. According to this the recombination
of charge carriers via surface states is dominated by those centers that are located
near the middle of the forbidden energy gap. The recombination rate reaches its
maximum in case of depletion when the electron and hole concentrations at the
surface are equal. If the concentration of one partner is strongly reduced in case
of inversion or accumulation, a much lower recombination rate results. This can
be realized by the formation of an internal electric field below the silicon surface,
e.g., by fixed positive or negative charges in a surface dielectric film thus as SiNx
or Al2 O3 . As a special advantage, the low-temperature processed PECVD films are
combining two very effective mechanisms for passivation of the silicon surface [12].
(i) Reducing the density of surface states by chemical passivation and (ii) field effect
passivation via fixed insulator charges.
Chemical Passivation by Intrinsic Hydrogen
During nitride deposition, e.g., by the reaction of the two gases silane (SiH4 ) and
ammonia (NH3 ) in a glow discharge hydrogen is released and up to 30% are incor-
porated in the amorphous films which are, therefore, often designated as a-SiNx :H,
abbreviated as SiNx throughout this work [30]. By ELS spectroscopy a new electron
energy loss peak at 8.6 eV was detected and was related to the presence of chemi-
cally bonded hydrogen [31]. The very active hydrogen species terminates unsaturated
bonds at the silicon surface thus lowering the density of surface states. The low inter-
face state density of 8 × 109 cm−2 eV−1 for nitride films deposited at 270 °C and
annealed at 500 °C is attributed to the saturation of unsaturated bonds at the silicon
surface by hydrogen originating from N–H bonds [32].
48 R. Hezel and K. Jaeger-Hezel

To a certain degree also other electronic states caused by crystal defects in the
bulk such as dislocations or grain boundaries in multicrystalline silicon including
silicon ribbons [33] are inactivated by hydrogen diffusing into the wafer during film
deposition or by a post-deposition heat treatment. Hence also the bulk diffusion
length can be improved resulting in a further increase of the cell efficiency [34].
Field Effect Passivation by Fixed Insulator Charges
SiNx contains a high density of long-term-stable positive charges causing strong
inversion consisting of electrons at the surface of p-type silicon. Hence, as outlined
above, low surface recombination rates are achievable by field effect passivation. For
PECVD SiNx fixed charge densities up to Qn/q = 7.2 × 1012 cm−2 are observed [9,
32, 35]. By incorporation of sodium [36] or cesium [37] the charge densities can be
further drastically increased up to 1.5 × 1013 cm−2 without significantly affecting
the low surface state density.
The combined effects of charge induced and hydrogen passivation are reflected
in the surface recombination velocity Seff which is discussed further below.

4.2.4 First Application of Silicon Nitride for Solar Cells

In 1981 the first and most demanding application of PECVD SiNx layers for silicon
surface passivation was its use as charged dielectric in a novel type of solar cells,
the so-called inversion layer cell [9]. Simple low-temperature fabrication and the use
of aluminum instead of expensive silver for metallization are attractive advantages
compared to the conventional pn-junction solar cells. Worldwide extensive research
was performed with similar structures whereby as transparent dielectrics the previ-
ously used antireflection layers such as thermal SiO2 [38], TiOx and CVD-SiO2 [39],
SiOx [40, 41] and Tax Oy [42] were applied. Efficiencies up to 16% are reported [42].
However, a major shortcoming of all these solar cells was that they were suffering
under severe stability problems.
For the inversion layer solar cells studied at the University of Erlangen, the sim-
ple fabrication consists of deposition of the highly transparent positively charged
PECVD SiNx layer on undiffused p-type silicon and a tunnel MIS (Metal Insulator
Silicon) contact grid with a tunnel Si oxide of 1.5 nm in thickness [9]. In the area
between the parallel grid lines an inversion layer is induced at the silicon surface,
which is responsible for surface passivation and provides a conducting path to the
MIS contacts for the light generated electrons. Therefore, in addition to a high den-
sity of stable insulator charges and a low density of fast interface states also a low
inversion layer sheet resistance is required [12]. Creation of interface states during
cell operation, in addition to surface recombination, also leads to a reduction of the
electron mobility and thus to an increase of the inversion layer sheet resistance. Both
effects are significantly contributing to the loss of solar cell efficiency.
As a conclusion, inversion layer solar cells are extremely surface sensitive elec-
tronic devices and are, therefore, excellent test tools for the quality and particularly
4 Silicon Nitride and Aluminum Oxide … 49

long-term stability and industrial applicability of passivation layers for silicon solar
cells. In the eighties silicon nitride was further optimized for inversion layer solar
cells and in 1990 the technology was transferred from the University lab to a 1 MW
industrial pilot line [43]. This transfer was honored by the German Industry with the
Innovation Award 1992.
The 10 × 10 cm2 solar cells using plasma SiNx as charged dielectric success-
fully passed the specific endurance tests [44]. Peak efficiencies exceeding 15%
could be obtained [43]. (About 10 years later laboratory cells reached efficiencies up
to 19.6% [45].)
In 1994 as a joint Spanish/German project 450 kW of the cells encapsulated in
novel large area double glass modules were installed in Europe’s largest photovoltaic
plant in Toledo/Spain (1 MW total). Encouraging results could be reported clearly
demonstrating the excellent long-term stability of the inversion layer modules [46].
For the next step towards a 10 MW production line cost-competitive high-rate evap-
oration of aluminum for front and back-contacts were required. However, such an
evaporation machine was not available from the equipment manufactures at this time,
so that the successful project could not be continued [47].
From the above over decades successful field test using the surface sensitive inver-
sion layer solar cell structure the important conclusion can be drawn, that PECVD
SiNx films have been proven to be excellent candidates for the reliable large-scale
passivation of future industrial silicon solar cells.

4.2.5 Further Optimization of PECVD Silicon Nitride

Stimulated by these promising results from 1995 onward further research was focused
on the optimization of the SiNx /Si interface quality, now at the Institute for Solar
Energy Research (ISFH) and elsewhere. Figure 4.2 shows the dependence of the

Fig. 4.2 Measured Seff (n)

dependence at polished and
textured 1.5 cm p-Si wafers
passivated by remote plasma
silicon nitride and thermally
grown silicon dioxide
50 R. Hezel and K. Jaeger-Hezel

surface recombination velocity Seff on the bulk excess carrier concentration n

for polished and chemically textured 1.5 cm p-Si wafers passivated by thermally
grown silicon dioxide and remote PECVD silicon nitride [48]. This configuration
corresponds, e.g., to the non-metallized passivated rear surface regions including
bifacial solar cells. As can be seen, for injection levels of relevance for photovoltaic
applications (i.e., n > 1013 cm−3 ) the remote SiNx film provides far better surface
passivation than the thermal Si oxide. At the highest measured injection levels of 1015
cm−3 the polished SiNx passivated surface yields a surface recombination velocity
(SRV) of 4 cm/s compared to 100 cm/s for the thermally oxidized surface. This is
one of the lowest SRV that has ever been obtained on a low resistivity p-Si wafer pas-
sivated by a solid film. Even for the nitride passivated textured wafer to reduce light
reflection a Seff value of only 19 cm/s is obtained. High-quality surface passivation
could also be obtained in an industrial type in-line remote microwave silicon nitride
deposition system with Seff values of 10 cm/s, making the surface passivation in this
system practically equivalent to that of the best laboratory PECVD reactor [10, 49].
Most of the present laboratory as well as industrial silicon solar cells are based
on p-type wafers with a front-side phosphorous diffused n+ -emitter, screen-printed
metallization, and a full-area Al-p+ back surface-field (Al-BSF). Passivation of these
devices by PECVD SiNx is the topic of several papers [50]. Also a systematic opti-
mization of remote PECVD SiNx films for n+ -emitters was performed in our lab [10,
51]. With remote plasma as well as with HF direct plasma SiNx films very good sur-
face passivation can be obtained if the refractive index is greater than about 2.0, i.e.,
if silicon-rich SiNx films are applied. Furthermore, these plasma SiNx passivated
emitters exhibit an excellent thermal stability and are perfectly stable against UV
exposure. Efficiencies up to 21% could be achieved for lab type diffused junction
solar cells [52, 53].
To summarize, compared to thermal oxides grown at high temperatures, low-
temperature remote PECVD SiNx films provide about equal surface passivation on
phosphorous diffused Si–surfaces and significantly superior surface passivation on
low resistivity p-Si wafers. Together with the higher refractive index, these films can
perfectly be applied as reliable passivating AR coatings for high-efficiency silicon
solar cells.

4.3 Plasma-Enhanced Deposition Techniques

Initially silicon nitride layers prepared by atmospheric pressure (APCVD) and low-
pressure chemical vapor deposition (LPCVD) were studied in our lab at the University
of Erlangen and successfully applied for solar cells [7, 19, 22]. These films were
deposited in a hot wall reactor at temperatures between 630 and 900 °C by the
reaction of silane (SiH4 ) and ammonia (NH3 ). Good interface properties could be
demonstrated [32, 35].
4 Silicon Nitride and Aluminum Oxide … 51

In 1981, for the sake of lower deposition temperatures and several other advan-
tages, as an alternative plasma-enhanced chemically-vapor-deposited (PECVD) sil-
icon nitride layers were first introduced by us as passivation and antireflection films
for silicon solar cells [9, 32]. It could be realized by a variety of experiments that
the properties of PECVD SiNx depend strongly on the type of the reactor as well as
on the deposition parameters. There are two fundamentally different plasma reactor
designs, parallel plate (often referred to as “direct”) and “remote”.

4.3.1 Parallel Plate Plasma Reactor

In this version, the wafer is placed onto one of two electrodes where the process-
ing gases silane and ammonia are excited by an electric field and the wafers are in
direct contact with the plasma. The electromagnetic field has a frequency of either
13.56 MHz (“high frequency” or HF mode) or lies in the 10–500 kHz range (“low
frequency” or LF mode). The systems are made up of an array (“boat”) of disk-like
metal plates in a long furnace tube enabling the simultaneous processing of a large
number of silicon wafers. A special phenomenon could be observed by extensive
investigations in our lab, that the plasma excitation frequency has a strong impact
on the electronic properties of the silicon/silicon nitride interface [54–56]. It was
found that the nitride deposition by low frequency direct PECVD does not signif-
icantly affect the initial performance of the surface sensitive inversion layer solar
cells as well as the interface properties. However, under UV light severe degradation
occurs. This behavior is attributed to the generation of interface states during the
direct LF plasma deposition process accompanied by their simultaneous passivation
with hydrogen. These “latent” states (apparently weak Si–H bonds) are activated
by UV light. As to the interface state generation during film deposition, below the
so-called plasma frequency (about 4 MHz) ions are able to follow the plasma exci-
tation frequency producing significant surface bombardment. In the high frequency
regime (HF) bombardment of the substrate surface is minimized since the ions can-
not follow the excitation of the electromagnetic field. Hence SiNx films prepared by
direct PECVD at high frequency (13.6 MHz) provide better surface passivation and
excellent UV stability than SiNx films prepared at low frequency (10–500 kHz) [57].
Inspite of its success the direct plasma batch method has a number of major
drawbacks and technical limitations with regard to the low-cost mass production of
SiNx -coated silicon solar cells [58].

4.3.2 Remote Plasma Reactor

The second and most successful class of plasma reactors is formed by the downstream
or remote system which was introduced into photovoltaics in 1989 by the author’s
group at the University of Erlangen [59]. An important feature of the remote PECVD
52 R. Hezel and K. Jaeger-Hezel

Fig. 4.3 Schematic of a

static remote PECVD reactor

systems is that the silicon wafer is located outside the plasma region, allowing the
deposition of SiNx films with no surface damage to the wafers and making the
frequency relatively unimportant.
Figure 4.3 shows a schematic of the static laboratory-type remote PECVD sys-
tem with downstream geometry (Oxford Plasmalab 80). A mixture of ammonia and
nitrogen is excited by passing through a 2.45 GHz-microwave cavity and mixes with
silane downstream of the plasma. The reaction takes place above the substrate which
is placed on a heated plate. It has been shown to give excellent surface passivation
on both p- and n+ -type silicon and was successfully used for the coating of high
efficiency solar cells in our lab [49, 51, 52].
Development of an Industrial In-Line Remote PECVD System
For industrial production of SiNx coatings, a high–throughput PECVD system is
required. In 1998 the first industrial solar cell line in Germany began its operation
at the ASE company. Direct PECVD was used for SiNx deposition. Due to several
drawbacks of these parallel plate batch mode reactors for the second production line
a remote PECVD in-line system should be developed [60]. In 2000 this develop-
ment occurred in tight cooperation between our research institute ISFH, the solar
cell producer ASE (later Schott Solar) and the equipment manufacturer Roth&Rau.
Figure 4.4 shows a cross section of the deposition system which consists of a process-
ing chamber and two load looks for the entry and exit of the wafer carriers [10, 11].
Inside the evacuated processing chamber two heating plates form an isothermal tun-
nel as a second preheating stage. In the plasma source two quartz tubes are mounted
perpendicular to the wafer transport direction. The microwave antennae are located
in the center of these quartz tubes. Two rows of magnets are installed alongside the
tubes which stabilize the plasma in the low-pressure regime by forcing electrons on
curved paths.
4 Silicon Nitride and Aluminum Oxide … 53

Fig. 4.4 Schematic cross section of a microwave in-line remote PECVD system

Ammonia is fed into the source above the tubes, whereas silane is added below
at the sides of the plasma source. The plasma is excited around the quartz tubes by
means of two continuous-wave magnetron microwave generators with a frequency
of 2.45 GHz. The carrier rests on rollers, by which it is transported linearly through
the vacuum chambers. A throughput of several thousand wafers/hour can be obtained
in continuous operation.
Due to our long experience with silicon nitride the in-line deposition system has
been set up for process development at the ISFH laboratory. After demonstrating
that industrial scale throughput can be obtained while maintaining good surface
passivation the deposition system was transferred by the ISFH staff to ASE GmbH and
successfully implemented into the solar cell production line. Roth&Rau subsequently
commercialized this tool under the name “SINA” and very soon became the world-
leading supplier for remote in-line PECVD systems [61]. This importantly paved
the way for the economic mass production of silicon solar cells worldwide since
reliable low-temperature passivating antireflection SiNx layers could be provided on
a large-scale [10, 11].

4.4 Silicon Nitride Passivated Bifacial Solar Cells

Excellent passivation of both surfaces is of prime importance for highly efficient solar
cells. This also allows a further increase in power output density by utilizing the light
incident on the solar cell from the front and rear side. By the proper installation of
such bifacially sensitive solar modules directing light also onto the rear surface,
power gains up to 50% is achievable [43, 62, 63, 64].
Bifacial cells have a long history and great efforts have been made regarding
both development as well as the application of these devices for space and terrestrial
54 R. Hezel and K. Jaeger-Hezel

Fig. 4.5 Calculated

dependence of the 1-sun
efficiency of a bifacial Si
solar cell on the
area-averaged rear surface
recombination velocity Sback

systems [64]. In the eighties low-temperature dielectric rear surface passivation was
applied for monofacial and bifacial solar cells in order to substitute the complex
high-temperature-processed back surface-field of the standard silicon cells [12, 13,
65].
In Fig. 4.5 the importance of rear surface passivation is demonstrated by the
calculated dependence of the cell efficiencies on the rear surface recombination
velocity Sback [66]. As can be seen for the special case of Fig. 4.5, the front efficiency
increases significantly with decreasing Sback , however, the improvement of the rear
efficiency is much more pronounced. For a poorly passivated rear surface (Sback >
104 cm/s) the rear efficiency is smaller than 5% whereas for a well passivated rear
surface (Sback < 102 cm/s) rear efficiencies higher than 20% can be reached.
As already shown in Fig. 4.2, much lower surface recombination velocities down
to record values of 4 cm2 /s are experimentally possible for p-Si using remote plasma
silicon nitride layers [49]. These SiNx films together with their excellent optical prop-
erties also provide very good surface passivation on phosphorous diffused emitters
so that highly efficient bifacial n+ p silicon solar cells are obtainable by deposition of
silicon nitride on front and rear side [56].
Based on our successful application of PECVD silicon nitride as charged passi-
vating antireflection film for the front-side of inversion layer solar cells, in 1986 this
multifunctional dielectric layer was applied by us for rear side passivation accord-
ing to the novel scheme consisting of a line-shaped aluminum base contact grid in
combination with the passivation layer between the grid fingers [13, 65, 12].
As the first application for 100 µm thick induced junction bifacial solar cells
efficiencies of 15 and 13.2% for front and rear illumination, respectively, could be
achieved [43, 65]. Extensive outdoor measurements for various albedo-conditions
were performed with 100 µm 10 × 10 cm2 bifacial cells in glass/glass modules of
different size up to 0.43 × 1.15 m2 [67].
With the fabrication of a 100 cm2 CZ-Si n+ p-solar cell in a pilot production line
at ASE 16.5% front and 13.8% rear efficiencies, respectively, were resulting [68].
4 Silicon Nitride and Aluminum Oxide … 55

Another fully nitride passivated 4 cm2 bifacial n+ p-silicon solar cell was intro-
duced in 1997 with efficiencies up to 20.1% under front illumination [69] and up to
18.1% under rear illumination [66]. As can be seen from Fig. 4.6, this cell combines
the excellent surface passivation properties of remote PECVD SiNx /SiOx double
layers with a simple local Al back surface-field. It was the first bifacial silicon solar
cell to exceed 20% front efficiency [69].
A completely different nitride passivated diffused pn-junction device is charac-
terized by a corrugated rear surface and both contact grid systems placed on the
steep flanks of the ridges at the rear side [70, 63]. Bifacial sensitivity is automati-
cally included with extremely low grid shadowing together with the advantages of
back-contact solar cells for efficiency, module fabrication and visual appearance. A
schematic of this back collecting bifacial “BACK OECO” (Obliquely Evaporated
Contacts) cell is shown in Fig. 4.7. Solar cells, 4 cm2 in size, were processed

Fig. 4.6 Schematic representation of a bifacial n+ p-silicon solar cell with double layer AR coatings
[69]. Monolayer SiNx coatings are applied in [66]

Fig. 4.7 Schematic of the

nitride passivated back
collecting and bifacially
sensitive BACK OECO solar
cell. Both contact lines are
placed on the steep flanks of
the ridges at the rear side
56 R. Hezel and K. Jaeger-Hezel

using 0.5 cm silicon wafers with an effective base thickness of 150 µm. Effi-
ciencies of 21.5% and 17.7% could be achieved for front and rear side illumination,
respectively [70].

4.5 Aluminum Oxide for Next-Generation Solar Cells

The future of solar cell production still lies in further increasing cell efficiency
while simultaneously reducing production costs. Upgrading the present standard n+ p-
silicon technology by replacing the efficiency limiting full-area Al-p+ back surface-
field (BSF) is commonly regarded as an effective procedure without much additional
costs. Low-temperature dielectric passivation of the rear surface plays a crucial role.
Already in the eighties, a substitute for the conventional BSF was introduced at the
University of Erlangen with a novel rear side scheme suitable both for monofacial and
bifacial solar cells in order to drastically reduce carrier recombination. As described
above for the rear side of bifacial solar cells this novel design is achieved by gridding
the base contact to minimize its area and depositing a thin passivating dielectric layer
on the silicon surface between the contact fingers [8, 12, 13, 65].
It was demonstrated in the preceding chapter that PECVD SiNx turned out to
be a reliable antireflection and passivation layer and is now used on the front-side
of the mainstream silicon solar cells worldwide. However, if the nitride layers are
incorporated in the above mentioned rear surface passivation scheme parasitic effects
have been identified by us which, to a certain extent, decrease cell performance [71,
66]. The performance loss is mainly due to a lower short-circuit current of the cells.
Our experiments clearly revealed parasitic shunting of the inversion layer induced
by the positively charged SiNx films with the rear contact grid [71]. The coupling of
the inversion layer with the base contacts can be mitigated by several means, e.g.,
lowering the inversion layer sheet conductance by the introduction of a thin plasma
SiOx film beneath SiNx . This step was already applied for the bifacial cell shown
in Fig. 4.6 [69] and later by Agostinelli et al. [72]. In 2011 a SiOx /SiNx stack was
implemented in a production line [73].
Alternatively, a thin Si-rich SiNx layer (refractive index n = 2.5) beneath the
standard nitride (n = 2.05) was introduced in the back-contacted bifacial solar cell
shown in Fig. 4.7 [70, 63]. In all cases by the double layer the fixed positive insulator
charge density and thus the shunting effect is reduced.

4.5.1 Al2 O3 —Based Rear Surface Passivation Scheme

In 1985 significant progress with respect to rear surface passivation and parasitic
shunting could be achieved at the University of Erlangen by the introduction of
aluminum oxide as another outstanding charged dielectric in combination with a
novel rear side scheme based on line-shaped local rear contacts (see Fig. 4.8). The
4 Silicon Nitride and Aluminum Oxide … 57

Fig. 4.8 Basic structure of the innovative rear side scheme of locally contacted and Al2 O3 —
passivated p-type silicon solar cells for monofacial and bifacial application [8, 12, 13]

crucial properties for low-temperature passivation were already investigated [8, 12,
13]. As demonstrated in Fig. 4.9, this early innovation became the basic element of
the industrial PERC device, which is world-wide regarded as the next-generation
solar cell.
Most relevant for rear side passivation are the fixed negative charges present
in the Al2 O3 film which are attracting holes as the majority carriers to form an
accumulation layer on p-type silicon [8]. This positively charged layer acts as a
barrier for electrons so that recombination and hence loss of light generated charge
carriers is significantly reduced (“field effect passivation”). As shown in Fig. 4.8 the
novel rear side scheme is made up of a parallel base contact grid and an Al2 O3 layer
on the silicon surface between the grid fingers. The area of the base contacts with
their high recombination rate has to be minimized without significantly increasing
the contact resistance. By this also grid shadowing for back surface illumination is
reduced. Full-area back metallization with contact openings in the Al2 O3 layer as
well as a local back surface-field at the base contact lines are some of the further
options (see Figs. 4.6 and 4.9).
The crucial property of this back surface system is that due to the presence of the
accumulation layer also the parasitic shunting effect is completely avoided [71]. In

Fig. 4.9 Schematic cross

section of an industrial type
PERC solar cell with
full-area back metallization
and line-shaped local
contacts (picture taken
from [81])
58 R. Hezel and K. Jaeger-Hezel

contrast to the case of inversion present with positively charged SiNx as passivation
layer, the flow of electrons along the rear surface towards the highly recombinative
base contacts is suppressed. This, as one of the main advantages of Al2 O3 together
with excellent surface passivation has a high potential for significantly increasing the
efficiency of conventional solar cells at low cost including the application as bifacial
devices [8, 12, 74].

4.5.2 Early Laboratory Results for Al2 O3

For the first application of Al2 O3 for solar cells we used the simple deposition tech-
nique of atmospheric pressure chemical vapor deposition (APCVD) with an open
tube laboratory furnace at temperatures between 300 and 500 °C. A very good level
of surface passivation of p-type solar cells could already be demonstrated [8, 12].
It was confirmed that the reported fixed negative charge density Qf /q = −3 × 1012
cm−2 and the mid-gap interface state density Dit = 8 × 1010 cm−2 eV−1 lies within
the range of typical values of present-day well passivating Al2 O3 layers indicating
Seff values below 100 cm/s on 2 cm p-type silicon [74]. As in the case of SiNx also
for Al2 O3 , in addition to charge induced (“field effect”) passivation, reduction of
the interface state density Dit via saturation of unsaturated bonds by intrinsic hydro-
gen plays another important role for surface passivation. However, in comparison
to SiNx , Al2 O3 provides an even more effective field-effect passivation since less
recombination occurs due to the absence of a space charge region at the surface of
p-type silicon [75].
As to the reliability of the interface properties, an interesting observation could be
made. Under strong UV light irradiation the fixed charge density Qf /q is increasing
due to an internal photo injection process [12]. Thus by UV light the passivation
properties of Al2 O3 are improved via enhancement of accumulation. As a surprising
result, the interface state density Dit is not affected by the UV light. Similar to SiNx
(Fig. 4.1), also Al2 O3 is known for its excellent radiation hardness and strong chem-
ical bonding. Thus Al2 O3 represents an extremely effective and reliable transparent
insulator film for silicon surface passivation, preferably for the rear side of p-type
solar cells to replace the conventional back surface-field. The effectiveness of the
passivation technique was first demonstrated on the front-side of a new back collect-
ing silicon solar cell, the performance of which is extremely sensitive to front surface
recombination [8]. A dramatic increase of the cell efficiency could be observed by
changing the deposition temperature of Al2 O3 from 290 to 500 °C.
Based on the encouraging results at this early stage together with other outstanding
properties, there was clear evidence that Al2 O3 could play a crucial role for efficient
and long-term stable monofacial and bifacial silicon solar cells [8, 12].
However, in the nineties, our R+D work at the FAU Erlangen and later at the
ISFH was focused on the further optimization of silicon nitride to provide a reliable
passivating antireflection layer for standard solar cells. This was an urgent goal at
4 Silicon Nitride and Aluminum Oxide … 59

that time, also for future solar cells. As a result, silicon nitride paved the way for
mass production of solar cells worldwide.

4.5.3 Revival of Al2 O3 as Passivation Layer for PERC Solar

Cells

With the introduction of the Al2 O3 —based rear surface passivation scheme including
line-shaped local contacts at the University of Erlangen, in 1985 a novel concept was
realized that, according to Fig. 4.8, the rear metal makes direct contact with the
p-type substrate on a restricted area while most of the silicon surface is passivated
by a dielectric layer [8, 13]. In 2006 a reintroduction of this configuration occurred
which is considered a key technology in future industrial high efficiency solar cell
production [17, 18, 74].
Four years after the introduction of the innovative passivation scheme shown in
Fig. 4.8, in 1989 a laboratory cell named PERC (“Passivated Emitter and Rear Cell”)
demonstrated the high-efficiency potential of the abovementioned general concept
[1]. The PERC cell is characterized by a high-temperature thermal SiO2 film at the
rear and a full-area metal layer forming point-shaped local rear contacts by openings
in the passivating oxide. With sophisticated processing, a record efficiency of 22.8%
was resulted, a value which at present is already obtained for Al2 O3 —based industrial
cells. The name PERC is now generally used for these latter cells (see, e.g., Fig. 4.9).
The Al2 O3 films were intensively studied in many research labs all over the world
and an extremely large number of publications appeared during the following years
[3, 15, 16, 74]. Soon the interest of the PV industry grew rapidly due to the encourag-
ing results obtained for the first Al2 O3 —based solar cells. Various types of production
processes were developed together with advanced technologies, e.g., for layer depo-
sition, laser contact opening, screen printing, and firing of Al-contacts. From 2012
onward the majority of solar cell manufacturers are already introducing PERC-like
solar cells into industrial mass production upgrading their conventional BSF produc-
tion lines for improved cell performance. For this purpose the innovative basic rear
side scheme shown in Fig. 4.8 was extended by a conventional local back surface-
field at the Al-contact lines and by a SiNx capping layer on top of Al2 O3 . In most
cases a full-area rear metallization is applied by screen printing and firing, which
contacts the silicon wafer at line-shaped areas where the rear passivation layer has
been removed by laser contact opening (LCO) [76]. An industrial type PERC solar
cell is schematically shown in Fig. 4.9. The crucial role of the two innovative charged
dielectric layers can clearly be seen. It should be mentioned, that for the manufac-
turing process point-shaped local rear contacts turned out to be less suitable than the
line-shaped contact scheme [77, 78].
There is increased interest in PERC solar cells with bifacial sensitivity since due
to the open-gridded back contact structure on the one hand light incident on the
60 R. Hezel and K. Jaeger-Hezel

rear can be used to increase power output and on the other hand aluminum paste
consumption is reduced [13, 65, 79, 80].
A bifacial version based on the open rear surface configuration shown in Fig. 4.8 is
obtained by simply applying a line-shaped Al-grid at the contact openings. Referring
to Sect. 4.4, already since 1986 the novel rear surface scheme was used for the
development of bifacial PERC-like solar cells.

4.5.4 Al2 O3 /SiNx Stacks for PERC Solar Cell

For almost all industrial PERC cells thanks to its unique properties silicon nitride
films play an even more important role compared to the present mainstream solar
cells. As can be seen from Fig. 4.9, in addition to the use of SiNx as passivation
and antireflection coating at the n+ -diffused front-side, the rear is made up of a stack
consisting of thin Al2 O3 capped by a thicker PECVD SiNx layer. The SiNx capping
layer has the following benefits. The Al2 O3 field effect back surface passivation is
enhanced by further chemical passivation provided by the active hydrogen species
released during the SiNx deposition. As outlined in Sect. 4.2.1 SiNx also provides an
outstanding protection layer, e.g., for further improving the firing stability of Al2 O3
against the Al-paste for the surface passivation of industrial screen-printed silicon
solar cells [82]. The rear side covered by the Al2 O3 /SiNx stack acts as a mirror for
incident light and thus enhances its absorption within the solar cell. Therefore, a
higher portion of the light is used for the generation of charge carriers [83]. Both
effects, excellent back surface passivation, as well as the mirror action, are further
key factors for boosting cell conversion efficiency of the PERC cells. Furthermore,
by a proper combination of the two films an optimized antireflection coating for light
incident on the rear side of the cell is provided.
There are various techniques for industrial low-temperature deposition of Al2 O3 .
Despite atomic layer deposition (ALD) yields the best passivation, PECVD is
most widely used since high-throughput and low-cost are the inherent advantages
[3]. Seff values below 10 cm/s are achieved [74]. The Roth&Rau company (now
Meyer&Burger) have adapted their successful SINA microwave PECVD system (see
Fig. 4.4, codeveloped at ISFH) for Al2 O3 deposition resulting in a multiple appli-
cation in-line apparatus (MAIA). Implemented into a standard Al-BSF cell line for
the establishment of a PERC cell line the MAIA equipment is capable of depositing
both the front SiNx AR coating and the rear Al2 O3 /SiNx stack within one machine
and in one run [84].

4.6 Conclusions

Minimizing recombination of charge carriers on both surfaces is a prerequisite for

achieving cost-effective high-efficiency solar cells. In the eighties, an important era
4 Silicon Nitride and Aluminum Oxide … 61

for PV began with the extrinsic low-temperature passivation of silicon solar cells by
the two dielectric layers silicon nitride (SiNx ) and aluminum oxide (Al2 O3 ) intro-
duced at the University of Erlangen. Both multifunctional films became key compo-
nents of present and future monofacial and bifacial solar cells.
Due to its unique properties, the positively charged SiNx -layers clearly outper-
formed all other candidates as efficient long-term stable passivating antireflection
coating. Thanks to the successful development of high-throughput plasma depo-
sition of these films economic mass production of reliable silicon solar cells was
possible. Hence, almost 50 years after the demonstration of a silicon solar cell in
1954 at the Bell Labs in USA the way was paved for the first large area application
of solar cells in the history of photovoltaics, soon culminating in a worldwide boom
of solar electricity.
A similar extraordinary progress as achieved by SiNx for the present mainstream
solar cells can also be predicted for the high-efficiency PERC devices as the next-
generation silicon cells by the outstanding properties of Al2 O3 as a negatively charged
surface passivation layer in combination with a gridded Al rear contact scheme. This
latter structure, pioneered by the authors already in 1985 is generally regarded as
a key technology for future silicon solar cells. But still, SiNx plays an important
role. So it took almost 25 years since their introduction, that both charged SiNx
and Al2 O3 layers are now together implemented in the novel silicon solar cell. By
the modifications for upgrading the mainstream cell production lines for PERC cell
manufacturing large investments are avoided compared to the development of new
technologies.
The international technology roadmap for photovoltaics (ITRPV) forecasts a mar-
ket share of PERC solar cells up to 60% and record efficiencies approaching 24%
within the next few years. For the present Al-BSF mainstream solar cells efficien-
cies up to 20% are achievable. A comprehensive review with further details about
monofacial and bifacial industrial PERC solar cells is presented in the following two
articles of this book.

References

1. A. Blakers, A. Wang, A. Milne, J. Zao, M. Green, Appl. Phys. Lett. 55, 1363 (1989)
2. A.G. Aberle, Crystalline Silicon Solar Cells—Advanced Surface Passivation and Analysis.
Centre for Photovoltaic Engineering, University of New South Wales, Sydney NSW 2052
(1999)
3. T. Dullweber, J. Schmidt, IEEE J. Photovolt. 6(5), 1366 (2016)
4. R. Hezel, Siemens Forschungs- und Entwicklungsberichte, Bd.1, Nr.1/72 (1972), p. 153
5. R. Hezel, E. Hearn, J. Electrochem. Soc. 125, 1848 (1978)
6. R. Hezel, German patent application (1978)
7. R. Hezel, R. Schörner, T. Meisel, in Proceedings of the 3rd. European Photovoltaic Solar
Energy Conference, Cannes (1980), p. 866
8. K. Jaeger, R. Hezel, in Proceedings of the 18th IEEE Photovoltaic Specialist Conference, Las
Vegas (1985), p. 1752
9. R. Hezel, R. Schörner, J. Appl. Phys. 52, 3076 (1981)
62 R. Hezel and K. Jaeger-Hezel

10. J. Moschner, J. Henze, J. Schmidt, R. Hezel, Prog. Photov. Res. Appl. 12, 21 (2004)
11. K. Roth, F. Chen, M. Fritzsche, M. Kirschmann, J. Müller, H. Schlemm, in Proceedings of the
21st European Photovoltaic Solar Energy Conference, Dresden (2006), p. 1137
12. R. Hezel, K. Jaeger, J. Electrochem. Soc. 136(2), 518 (1989)
13. K. Jaeger, R. Hezel, in Proceedings of the 7th European Photovoltaic Solar Energy Conference,
Seville (1986), p. 806
14. R. Hezel, K. Jaeger, Fall Meeting of the Electrochemical Society, Honolulu, Extended Abstracts,
vol. 87(2) (1987), p. 1134
15. Photon International (March 2011)
16. Sun and Wind Energy (Nov. 2010)
17. B. Hoex, S. Heil, F. Langereis, M. van de Sanden, W. Kessels, Appl. Phys. Lett. 89, 042112
(2006)
18. G. Agostinelli, A. Delabie, P. Vitanov, Z. Alexieva, H. Dekker, S. Wolf et al., Sol. Energy
Mater. Sol. Cells 90, 3438 (2006)
19. N. Lieske, R. Hezel, Thin Solid Films 61, 217 (1979)
20. N. Lieske, R. Hezel, in Proceedings of the Conference in Insulating Films on Semiconductors
1979, vol. 50 (1980), p. 206
21. R. Hezel, N. Lieske, J. Appl. Phys. 51, 2566 (1980)
22. R. Hezel, Habilitationsschrift Universität Erlangen-Nürnberg (FAU) (1981)
23. R. Hezel, Solid St. Electron., 735 (1979)
24. R. Hezel, Radiat. Eff. 65, 101 (1982)
25. J. Johannessen, W. Spicer, Y. Strausser, J. Appl. Phys. 47, 3028 (1976)
26. A. Sinha, H. Levinstein, T. Smith, G. Quintana, S. Haszko, J. Electrochem. Soc. 125(4), 601
(1978)
27. B. Lenkeit, S. Steckemetz, A. Mücklich, A. Metz, R. Hezel, in Proceedings of the 16th European
Photovoltaic Solar Energy Conference, Glasgow (2000), p. 1332
28. R. Hezel, in Silicon Nitride in Electronics, Gmelin Handbook Si Supplement, vol. B 5c (1991),
p. 321
29. W. Shockley, W. Reed, Phys. Rev. 87, 835 (1952)
30. W. Lanford, M. Rand, J. Appl. Phys. 49, 2473 (1978)
31. R. Hezel, N. Lieske, J. Appl. Phys. 53(3), 1671 (1982)
32. R. Hezel, K. Blumenstock, R. Schörner, in Proceedings of the Symposium on Silicon Nitride
Thin Insulating Films, vol. 83-8, ed. by H.S.V.J. Kapoor (1983), p. 280
33. G.Schwuttke, K. Yang, R. Hezel, ERDA/JPL Contract Nr. 954144-76/05, Technical Progress
Report, USA, 5 (1976), pp. 1–34
34. A. Cuevas, M. Kerr, J. Schmidt, in Proceedings of the 3rd World Conference on Photovoltaic
Solar Energy Conversion, Osaka (2003), p. 913
35. R. Hezel, Solid St. Electron. 24, 863 (1981)
36. R. Hezel, in Insulating Films on Semiconductors, Springer Series in Electrophysics, vol. 7
(Springer, Berlin, 1981), p. 219
37. R. Hezel, in Proceedings of the 16th IEEE Photovoltaic Specialists Conference (1982), p. 1237
38. T. Salter, R. Thomas, Solid St. Electron. 20, 95 (1977)
39. P. van Halen, R. Mertens, R. van Overstraten, R. Thomas, J. van Meerbergen, IEEE Trans. El.
Dev. ED-25, 507 (1978)
40. M. Green, R. Godfrey, M. Willison, A. Blakers, in Proceedings of the 14th IEEE Photovoltaic
Specialists Conference, San Diego (1980), p. 684
41. R. Godfrey, M. Green, Appl. Phys. Lett. 34, 790 (1979)
42. R. Thomas, C. Norman, R. North, in Conference Record of the 14th IEEE Photovoltaic Spe-
cialist Conference (1980), p. 1350
43. R. Hezel, W. Hoffmann, K. Jaeger, in Proceedings of the 10th European Photovoltaic Solar
Energy Conference, Lisbon (1991), p. 511
44. W. Hoffmann, Metall 44(7), 658 (1990)
45. C. Peters, R. Meyer, R. Hezel, in Proceedings of the European Photovoltaic Solar Energy
Conference, Rome (2002), p. 127
4 Silicon Nitride and Aluminum Oxide … 63

46. M. Alonso, R. Pottbrock, R. Voermans, J. Villard, B. Joshi, in Proceedings of the 12th European
Photovoltaic Solar Energy Conference, Amsterdam (1994), p. 1163
47. W. Hoffmann, The Economic Competitiveness of Renewable Energy (Scrivener Publishing
Wiley, 2014)
48. J. Schmidt, T. Lauinger, A. Aberle, R. Hezel, in Proceedings of the 25th PVSC, Washington
D.C. (1996), p. 413
49. T. Lauinger, J. Schmidt, A. Aberle, R. Hezel, Appl. Phys. Lett. 68, 1232 (1996)
50. D. Ruby, W. Williams, C. Fleddermann, in Proceedings of the 1st World Conference on Pho-
tovoltaic Energy Conversion IEEE, New York (1994), p. 1355
51. B. Lenkheit, T. Lauinger, A. Aberle, R. Hezel, in Proceedings of the 2nd World Conference on
Photovoltaic Energy Conversion (1998), p. 1434
52. A. Metz, R. Hezel, in Conference on Record of the 26th IEEE PVSC (1997), p. 283
53. S. Dauwe, L. Mittelstädt, A. Metz, J. Schmidt, R. Hezel, in Proceedings of the 3rd World
Conference on Photovoltaic Energy Conversion, Osaka (2003), p. 1395
54. K. Blumenstock, R. Hezel, in Insulating Films on Semiconductors ed. by J. Simone, J. Buxo.
Elsevier Science Publications B.V. North Holland (1986)
55. R. Hezel, R. Auer, M. Rammensee, Progress in Plasma processing of Materials (1999), p. 891
56. A. Aberle, R. Hezel, Progr. Photov. Res. Appl. 5(1), 29 (1997)
57. T. Lauinger, A. Aberle, R. Hezel, in Proceedings of the 14th European Photovoltaic Solar
Energy Conference, Barcelona (1997), p. 853
58. A. Aberle, T. Lauinger, R. Hezel, in Proceedings of the 14th European Photovoltaic Solar
Energy Conference, Barcelona (1997), p. 684
59. M. Rammensee, Ph.D. thesis, University of Erlangen (1995)
60. J. Moschner, A. Metz, R. Hezel, in Proceedings of the 17th European Photovoltaic Solar
Energy Conference, Munich (2001), p. 1917
61. S. Roth, Themen 2007: Jahrestagung des Forschungsverbunds Solarenergie in Kooperation
mit dem Bundesverband Solarwirtschaft e. V. (BSE Solar), Hannover, Leibniz Univ. (2007)
62. R. Hezel, Prog. Photovolt. Res. Appl. 11, 549 (2003)
63. R. Hezel, in High-Efficient Low Cost Photovoltaics, vol. Springer Series in Optical Science
(Springer, Berlin, 2009), p. 86
64. A. Cuevas, in Proceedings of the 20th European Photovoltaic Solar Energy Conference,
Barcelona (2005), p. 801
65. K. Jaeger, R. Hezel, in Proceedings of the 19th IEEE Photovoltaic Specialist Conference, New
Orleans (1987), p. 388
66. A. Hübner, A. Aberle, R. Hezel, Appl. Phys. Lett. 70, 1008 (1997)
67. K. Jaeger, G. Bende, W. Hoffmann, R. Hezel, in Conference Record 23th IEEE Photovoltaic
Specialist Conference, Louisville (1993), p. 1235
68. K. Jaeger-Hezel, W. Schmidt, W. Heit, K. Rasch, in Proceedings of the 13th European Photo-
voltaic Solar Energy Conference, Nice (1995), p. 1515
69. A. Hübner, A. Aberle, R. Hezel, in Proceedings of the 14th European Photovoltaic Solar
Energy Conference, Barcelona (1997), p. 92
70. J. Müller, A. Merkle, R. Hezel, in Proceedings of the 20th European Photovoltaic Solar Energy
Conference, Barcelona (2005), p. 1020
71. S. Dauwe, L. Mittelstädt, A. Metz, R. Hezel, Prog. Photov. Res. Appl. 10, 271 (2002)
72. G. Agostinelli, P. Chonlet, H. Dekkers, E. Vermerien, G. Beancarne, in Proceedings of the 4th
World Conference on Photovoltaic Energy Conversion (2006), p. 1004
73. K. Münzer et al., Proceedings of the 26th European Photovoltaic Solar Energy Conference,
Hamburg (2011), p. 2292
74. J. Schmidt, F. Werner, B. Veith, D. Zielke, S. Steingrube, P. Altermatt, S. Gatz, T. Dullweber,
R. Brendel, Energy Procedia 15, 30 (2012)
75. J. Schmidt, A. Merkle, R. Bock, P. Altermatt, A. Cuevas, N. Harder, B. Hoex, R. Sanden, E.
Kessels, R. Brendel, in Proceedings of the 23rd European Photovoltaic Solar Energy Confer-
ence, Valencia (2008), p. 974
64 R. Hezel and K. Jaeger-Hezel

76. R. Preu, S. Glunz, S. Schäfer, R. Lüdemann, W. Wettling, W. Pfleging, in Proceedings of the

16th European Photovoltaic Solar Energy Conference, Glasgow (2000), p. 1181
77. T. Lauermann, A. Zuschlag, S. Scholz, G. Hahn, B. Terheiden, in Proceedings of the 26th
European Photovoltaic Solar Energy Conference, Hamburg (2011), p. 1137
78. J. Müller et al. (2011). IEEE Trans. Electron Devices 58(10), 3239
79. T. Dullweber, C. Kranz, R. Peibst, U. Baumann, H. Hannebauer, A. Fülle et al., Prog. Photov.
Res. Appl. 24, 1487 (2016)
80. K. Krauß, F. Fertig, J. Greulich, S. Rein, R. Preu, Phys. Stst. Solidi. A 213(1), 68 (2016)
81. H. Hannebauer, T. Dullweber, U. Baumann, T. Falcon, R. Brendel, Phys. Status Solidi RRL 8,
675 (2014)
82. J. Schmidt, B. Veith, R. Brendel, Phys. Stat. Sol. RRC, 287 (2009)
83. D. Kray, M. Hermle, S. Glunz, Prog. Photov. Res. Appl. 16, 1 (2008)
84. H. Sperlich, D. Decker, P. Saint-Cast, E. Erben, L. Peters, in Proceedings of the 25th European
Photovoltaic Solar Energy Conference, Valencia (2010), p. 1352
Chapter 5
High-Efficiency Industrial PERC Solar
Cells for Monofacial and Bifacial
Applications

Thorsten Dullweber

Abstract The passivated emitter and rear cell (PERC) concept is currently rapidly
being introduced into industrial mass production and is expected to be the new silicon
wafer based solar cell technology standard in the photovoltaics industry. In 2018,
PERC-type solar cells accounted for approximately 40% of the worldwide produced
solar cells and their share is expected to rapidly increase up to 70% within the next few
years. Compared to the previous industrial silicon solar cell technology which applied
a full-area aluminum rear contact, PERC cells passivate the rear silicon wafer surface
with a dielectric layer and only locally contact the silicon wafer with an aluminum
metal contact, which reduces charge carrier recombination and hence increases the
conversion efficiency. Present record conversion efficiencies up to 22.8% of industrial
PERC cells hence exceed the efficiency of conventional Al-BSF silicon solar cells by
more than 2%abs . In addition, PERC solar cells can be made bifacial by substituting
the full-area rear aluminum layer with an aluminum finger grid design. This so-called
PERC+ solar cell design enables large volume industrial manufacturing of bifacial
silicon solar cells which absorb stray light from the rear side and hence increase the
energy yield by 5–25% depending on the detailed module installation conditions. This
chapter describes the most important research results and technology developments
of the past decades as well as the present status of industrial PERC and PERC+ solar
cells.

The passivated emitter and rear cell (PERC) concept is currently rapidly being intro-
duced into industrial mass production and is expected to be the new silicon wafer
based solar cell technology standard in the photovoltaics industry. In 2018, PERC-
type solar cells accounted for approximately 40% of the worldwide produced solar
cells and their share is expected to rapidly increase up to 70% within the next few
years. Compared to the previous industrial silicon solar cell technology which applied

This chapter contains text adapted from [1] and [2] with permission of The Institution of Engineering
and Technology.

T. Dullweber (B)
Institut für Solarenergieforschung Hameln, Am Ohrberg 1, 31860 Emmerthal, Germany
e-mail: [email protected]
© Springer Nature Switzerland AG 2020 65
V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_5
66 T. Dullweber

a full-area aluminum rear contact, PERC cells passivate the rear silicon wafer surface
with a dielectric layer and only locally contact the silicon wafer with an aluminum
metal contact, which reduces charge carrier recombination and hence increases the
conversion efficiency. Present record conversion efficiencies up to 22.8% of indus-
trial PERC cells hence exceed the efficiency of conventional Al-BSF silicon solar
cells by more than 2%abs . In addition, PERC solar cells can be made bifacial by sub-
stituting the full-area rear aluminum layer with an aluminum finger grid design. This
so-called PERC+ solar cell design enables large volume industrial manufacturing
of bifacial silicon solar cells which absorb stray light from the rear side and hence
increase the energy yield by 5–25% depending on the detailed module installation
conditions. This chapter describes the most important research results and technol-
ogy developments of the past decades as well as the present status of industrial PERC
and PERC+ solar cells.

5.1 Introduction

When the diffusion length of the minority charge carriers exceeds the silicon wafer
thickness, charge carrier recombination at the rear side of the silicon solar cells
becomes important. During the past decades, around 90% of the solar cells worldwide
have been manufactured applying silicon wafers and a screen-printed full-area alu-
minum (Al) layer which contacts the complete rear silicon wafer surface by forming
an aluminum back surface field (Al-BSF). The carrier recombination at the full-area
aluminum rear contact is only marginally suppressed by the local Al doping of the
Al-BSF. Also, the full-area Al layer partly absorbs the infrared light which reduces
the light absorption of the solar cell. Due to these two loss mechanisms, the con-
version efficiency of industrial silicon solar cells with full-area Al-BSF is limited to
around 20%. Already in 1989, Blakers et al. [3] in the research group of M. A. Green
at the University of New South Wales (UNSW), Australia, introduced the “Passivated
Emitter and Rear Cell” (PERC) which was designed to overcome the efficiency lim-
itations of the Al-BSF solar cell by introducing a dielectric passivation of the silicon
wafer rear side using only local metal contacts. The PERC solar cell in 1989 did not
apply a production-feasible manufacturing process but instead mostly laboratory-
type processes and wafer materials which increased the conversion efficiency to a
world record value of 22.8% at that time. Very expensive 0.2 cm p-type float-zone
silicon wafers were used. The rear passivation consisted of a thermally grown silicon
dioxide layer which was locally opened by photolithography and etching followed
by evaporation of a full-area aluminum layer which resulted in point-shaped local
aluminum rear contacts. The solar cell’s front side was processed with several pho-
tolithography steps forming inverted pyramids, a selective phosphorus emitter and
evaporated TiPdAg front contacts. The resulting 4 cm2 laboratory-type PERC solar
cell of 1989 is shown schematically in Fig. 5.1. In contrast to Al-BSF cells, the rear
side of PERC cells is largely covered by SiO2 thereby suppressing the recombination
of minority charge carriers (electrons) at the metal rear contact. Majority charge car-
5 High-Efficiency Industrial PERC Solar Cells … 67

Fig. 5.1 Schematic drawing of the laboratory-type PERC solar cell published by Blakers et al. in
1989 [3] applying a thermal silicon dioxide rear passivation, rear-side point contacts patterned by
photolithography, and a full-area evaporated aluminum rear metal layer. This figure is reproduced
from [3] with the permission of AIP Publishing

riers (holes) are transported through the local Al contacts at the rear thereby ensuring
the current flow through the PERC solar cell. Also, the SiO2 /Al layer stack at the
PERC rear side serves as a very good optical mirror reflecting infrared light which
has not yet been absorbed in the silicon wafer back into the solar cell instead of
being absorbed in the Al layer. In the following years, the PERC concept was fur-
ther developed into the passivated emitter and rear locally diffused (PERL) solar
cell resulting in an efficiency of 25.0% by present standards [4] which remained the
world record efficiency for silicon solar cells for almost two decades. The PERL cell
applied a high local p+ doping below the Al rear contacts thereby reducing carrier
recombination at the local rear contacts which increased the conversion efficiency.
The PERC and PERL solar cell designs of the UNSW provided the foundation and
basic concept of today’s industrial PERC solar cells. However, as outlined in the next
sections, a broad variety of process and technology developments regarding surface
passivation and metallization were required to develop an industrial PERC manu-
facturing process sequence applying novel low-cost and high-throughput processes
with sufficient quality for high conversion efficiencies. Since a few years, the new
production processes, in particular, for chemical rear surface polishing, aluminum
oxide/silicon nitride rear surface passivation, laser contact opening, and aluminum
screen printing are available with sufficient quality and throughput and now most
solar cell companies are manufacturing PERC solar cells in mass production with a
present market share of the PERC technology of 40% [5]. The International Tech-
nology Roadmap for Photovoltaics (ITRPV) predicts a further strong increase up
to 70% of the market share of the PERC technology within the next years [6, 7] as
shown in Fig. 5.2a. Several review papers and book chapters [1, 8–10] provide an
overview of the research and development and the present status of the PERC solar
68 T. Dullweber

Fig. 5.2 a Forecasted market share of different solar cell technologies according to the International
Technology Roadmap for Photovoltaics (ITRPV) as published in 2017 [6]. PERC cells are expected
to gain approx. 60% market share within the next 10 years. b The ITRPV roadmap predicts a market
share of bifacial silicon solar cells of 40% until 2028. The bifacial PERC+ cell concept (see Fig. 5.3c)
combines both trends, PERC and bifacial, by changing the full-area aluminum rear layer of PERC
to an aluminum finger grid layout. This figure is reproduced from [6] with the permission from
VDMA

cell technology. The following sections will describe the development, status, and
prospects of industrial PERC solar cells in detail.
Since a few years, bifacial solar cells and modules are gaining attention by the
photovoltaics industry due to the increased energy yield by converting the solar
irradiation to electricity not only with the solar cell’s front side but by also utilizing
diffuse light which illuminates the solar cell’s rear side. As shown in Fig. 5.2b, the
International Technology Roadmap for Photovoltaics (ITRPV) forecasts a steadily
increasing market share of bifacial silicon solar cells of up to 40% till 2028. As will be
described in this chapter, PERC+ solar cells enable a straightforward implementation
of a bifacial solar cell concept into mass production. Hence, several companies are
already producing bifacial PERC+ solar cells and modules. Review articles of the
development and present status of PERC+ solar cells are published in [2, 11].

5.2 Industrial PERC Solar Cells

As described in Sect. 5.1, the first PERC solar cell fabricated at the UNSW in 1989 [3]
applied mostly lab-type processes such as photolithography for structuring, ther-
mal oxidation for masking and surface passivation as well as evaporation of metals
for contact deposition. One important step toward an industrial PERC production
sequence was introduced by the research institute IMEC, Belgium, in 2005 [12, 13]
proposing a process combination of PECVD SiNx as rear surface passivation, rear
contact structuring by laser contact opening, and rear metallization via aluminum
screen printing. However, also the PERC process sequence proposed by IMEC still
applied several lab-type plasma processes, e.g., for texturing and rear emitter removal
[12].
5 High-Efficiency Industrial PERC Solar Cells … 69

Today, a mainstream process flow for industrial production of PERC solar cells
is applied by most of the PERC cell manufacturers as summarized in Table 5.1 and
described in detail in [10]. First, the boron-doped multi- or monocrystalline silicon
wafers with around 180 µm starting thickness and 156 × 156 mm2 area are unisotrop-
ically etched in an acidic or alkaline chemical solution thereby forming pyramids on
the wafer surface, a so-called texture, which reduce the reflection of light. Afterward,
the phosphorus emitter is diffused applying phosphorus oxychloride (POCl3 ) at tem-
peratures around 850 °C in a quartz tube furnace. Since the texturing and POCl3
diffusion process both, the front and the rear wafer surface, a wet chemical etching
and polishing process removes the rear-side phosphorus doping and smoothens the
rear pyramids [14, 15]. Afterward, the wafer surface is cleaned by alkaline and acidic
chemical steps which also remove the remaining phosphorus silicate glass. Then,
the wafer surfaces are coated by surface passivation layer stacks. The rear surface
receives a layer stack consisting of 5–20 nm aluminum oxide (AlOx ) and 80–120 nm
of silicon nitride (SiNy ) which are typically deposited by plasma-enhanced chemical
vapor deposition (PECVD) [16, 17]. AlOx has been shown to obtain very low surface
recombination velocities of minority charge carriers on p-type wafer surfaces due
to its built-in negative fixed charges [18, 19]. The SiNy capping layer is required in
order to protect the AlOx layer against the following Al screen printing and firing
processes, which otherwise would dissolve the AlOx layer. The phosphorus-doped
front surface is passivated by a SiNy layer, which due to its built-in positive charges
[20] is well suited to minimize carrier recombination of n-type wafer surfaces. The
front SiNy layer is typically around 80 nm thick which minimizes the reflection of

Table 5.1 Typical production process sequence of industrial PERC and PERC+ solar cells. Whereas
the principal process flow of monofacial PERC and bifacial PERC+ solar cells is identical, the
individual recipes vary between PERC and PERC+ in order to optimize the front-side efficiency
for PERC and the bifaciality for PERC+
PERC recipes PERC/ PERC+ process PERC+ recipes
Texturing
POCl3 diffusion
3–5 µm polishing removal Rear chemical polishing Reduced polishing removal
PSG etch
Thickness ca. 130 nm AlOx /SiNy rear passivation Thickness ca. 80 nm optimized
optimized for passivation and for antireflection
internal reflectance
Front SiNy
Contact pattern optimized for Laser contact opening Contact pattern optimized for
front-side efficiency bifaciality
Ag rear screen printing
Full-area print Al rear screen printing Fingerprint
Ag front screen printing
Firing Reduced set peak temperature
70 T. Dullweber

light from the front surface to below 3% in average over the whole visible wavelength
range due to destructive interference. The rear metal contact pattern is defined by
local laser ablation of the AlOx /SiNy stack, a process named laser contact opening
(LCO) [21]. Typically, either line-shaped [22] or dashed-line-shaped [23] pattern is
used in production. Then, the front and rear metal contacts are deposited by printing
silver (Ag) or aluminum (Al) containing pastes through a mesh screen. The metal
pastes consist of an Ag or Al powder with a particle size of a few micrometers
(µm) and organic binders and solvents which define the viscosity and printability of
the paste. The mesh screen consists of around 15 µm thin metal wires with around
300–400 wires per inch, which are coated by an organic emulsion. At areas where
the Ag or Al paste should be printed on the wafer, the emulsion is removed from
the mesh screen allowing the metal paste to penetrate through the screen. First, the
rear Ag pads which are later required for module interconnection are screen-printed.
Afterward, the Ag paste is dried at 200 °C thereby removing the solvent. The Al
rear layer is screen-printed full area at the rear side and is dried. Then, the Ag front
contact is screen-printed and dried. The typical Ag front contact layout consists
of around 100 very thin Ag conductor lines (fingers) with a finger width around
40 µm, which are interconnected by four or five perpendicular Ag busbars which
are required for module interconnection. After drying the Ag paste, the whole PERC
process sequence is completed by a rapid thermal process called firing which heats
the wafers for a few seconds up to around 800 °C. At this temperature regime, the
glass frites of the Ag paste etch through the SiNy front layer allowing the Ag to alloy
with the phosphorus-doped silicon forming an electrical contact on the front side.
At the same time, the Al paste on the rear-side alloys with the silicon wafer at areas
where the rear passivation has been removed by LCO thereby forming an electrical
contact at the rear side. The Al pastes contain a very small amount of glass frites
and hence do not etch the rear passivation layer stack. The resulting industrial PERC
solar cell is shown schematically in Fig. 5.3b.
The production process sequence described above applies to both monofacial
PERC solar cells and bifacial PERC+ solar cells. However, as listed in Table 5.1,
several process recipes are optimized differently for PERC versus PERC+ solar cells.
Most importantly, the full-area Al layer print of PERC cells is changed to an Al finger
grid print for PERC+ solar cells [24]. The LCO pattern for PERC is optimized for a
high front-side efficiency by balancing wafer bulk spreading resistance losses versus
recombination losses at the Al rear contacts. For PERC+, an additional optimization
criterion is a high rear-side efficiency which favors larger LCO pitches and hence
higher spacing of the Al fingers in order to minimize reflection losses of the PERC+
cell when illuminated from the rear side. For the same reason, PERC+ cells apply a
thinner rear AlOx /SiNy stack around 80 nm [24] which is optimized for antireflection
and the rear polishing removal maybe reduced in order to maintain part of the rear
texture, both minimizing reflection losses. For PERC cells, the polishing process
and rear passivation thickness are optimized for high front-side efficiencies which
typically favors slightly higher polishing removals and thicker passivation layers to
maximize the internal reflection at the rear. Finally, the set peak temperature of the
firing process for PERC+ is typically around 20 °C lower as for PERC since the
5 High-Efficiency Industrial PERC Solar Cells … 71

Fig. 5.3 Schematic drawings of a an industrial Al-BSF solar cell where the screen-printed Al layer
alloys with the silicon wafer forming a full-area Al-doped back surface field (BSF or Al-BSF);
b an industry-typical PERC solar cell with AlOx /SiNy rear passivation, local BSF formed by laser
contact opening, and a full-area rear Al layer; c a bifacial PERC+ solar cell applying a screen-printed
Al finger grid on top of the laser contact openings on the rear side thereby enabling light absorption
from the front and rear sides

reduced amount of aluminum paste requires less external heating power to obtain
the same physical wafer temperature. Figure 5.3c shows a schematic cross section
of a PERC+ solar cell, which is described in detail in the next section.
For comparison, Fig. 5.3a shows a schematic drawing of a full-area Al-BSF solar
cell, which has been dominating industrial production for many years as illustrated
in Fig. 5.2a. The Al-BSF cell applies the same front side as an industrial PERC
cell, whereas at the rear side the Al layer is printed directly to the silicon wafer
surface without any dielectric rear surface passivation. Hence, the Al-BSF cell pro-
cess sequence does not require AlOx /SiNy rear passivation deposition and no laser
contact opening making it shorter and hence cheaper compared to the PERC process
sequence in Table 5.1. Consequently, PERC cells need to demonstrate at least 1%
72 T. Dullweber

absolute higher energy conversion efficiencies compared to Al-BSF cells in order

to enable lower photovoltaic electricity generation costs on module level (Euro per
Watt peak; e/Wp) or electricity level (Euro per kilo Watt hour; e/kWh). Figure 5.4
shows a numerical simulation published in the year 2011 [25] applying the solar cell
simulation software PC1D of the conversion efficiency versus the rear surface recom-
bination velocity S eff,rear and the internal rear reflectance Rrear . The full-area Al-BSF
forms a p/p+ doping gradient and hence a built-in electric field which limits S eff,rear
to around 400 cm/s. The internal reflectance Rrear of the silicon/ aluminum interface
is around 65%. As can be seen in Fig. 5.4, both parameters limited the conversion
efficiency of Al-BSF cells to about 18.5%. Due to the dielectric rear passivation,
PERC cells exhibit a much lower S eff,rear around 50 cm/s and a higher Rrear around
90% and hence demonstrated conversion efficiencies close to 20% already in the year
2011. The simulation in Fig. 5.4 assumed values for the Ag finger, emitter, and wafer
properties which were typical in the PV industry in the year 2011. In the following
years till today, the solar cell process technology has been constantly improved which
increased the efficiencies of Al-BSF cells toward 20% and the efficiencies of PERC
cells beyond 22% as shown in Fig. 5.5.
Due to continuous optimization of all process steps in Table 5.1 such as improved
Ag pastes enabling narrower Ag fingers, selective emitters with advanced doping pro-
files, improved front and rear surface passivation layers, further developed Al pastes,
LCO pattern, and busbar designs, the conversion efficiency of industrial PERC solar

Fig. 5.4 PC1D simulation from the year 2011 [25] of the conversion efficiency in dependence of
the rear surface recombination velocity S eff,rear for different internal rear reflectance values Rrear .
The full-area Al-BSF with S eff,rear around 400 cm/s and Rrear around 65% limited the conversion
efficiency of Al-BSF cells to about 18.5%. Due to the dielectric rear passivation, PERC cells exhib-
ited a much lower S eff,rear around 50 cm/s and a higher Rrear around 90% and hence demonstrated
conversion efficiencies close to 20%. In the following years, the efficiencies of Al-BSF improved
toward 20% and of PERC cells beyond 22% as shown in Fig. 5.5
5 High-Efficiency Industrial PERC Solar Cells … 73

Fig. 5.5 Independently

confirmed record conversion
efficiencies of industrial-type
PERC solar cells with
screen-printed metal contacts
using monocrystalline
large-area p-type Cz silicon
wafers. Typical efficiencies
of full-area Al-BSF solar
cells are indicated as well.
Graph updated from [11]

cells has been constantly increased over the years. Figure 5.5 shows the evolution of
record conversion efficiencies during the years 2010–2018 obtained with industrial
PERC solar cells (large area >148 cm2 p-type monocrystalline Czochralski-grown
(Cz) silicon wafers, screen-printed metal front and rear contacts with conventional
busbar layout) [22, 26–37]. Starting with an efficiency of 19.2% [26] reported by
Centrotherm in 2010, the benchmark in 2012 has been 21.0% [28, 29] as demon-
strated by Schott Solar. In 2014, the efficiency has been further increased to 21.2%
by ISFH [30] and in July 2015 to 21.7% by SolarWorld [32]. Trina Solar was the
first to surpass the 22% efficiency barrier by reporting a 22.1% efficient industrial
PERC cell in December 2015 [33]. The present champion efficiencies of 22.7% and
22.8% have been obtained by LONGi Solar and JinkoSolar in October 2017, respec-
tively [36, 37]. As can be derived from Fig. 5.5, the record efficiencies of industrial
PERC cells increased by approximately 0.5%abs. per year as average value over the
past 8 years. Simply extrapolating this trend predicts that in 2020 record PERC cell
efficiencies will approach 24% applying industrially feasible production processes.
It took a lot of research and development effort to transfer the lab-type PERC
concept of the UNSW into an industrially manufacturable PERC solar cell concept.
A quite detailed summary of the historic development steps toward industrial PERC
cells is given in [11]. Here, we just want to highlight the most important R&D
achievements which enabled industrial PERC solar cells. As can be seen in Fig. 5.2a,
for many years the so-called BSF or Al-BSF silicon solar cell was the dominant cell
concept in industrial production. In order to upgrade a conventional industrial full-
area Al-BSF production process into a PERC production process, mainly the rear
passivation layer and the laser contact opening have to be introduced. Also, the
phosphorus diffusion in the PERC process should only dope the wafer front surface
and not the rear surface in order to avoid a floating rear junction or shunting of the
electrodes. Hence, most of the R&D efforts for industrial PERC processes where
directed toward low-cost and high-performance processes for the rear passivation
74 T. Dullweber

layer, for the rear contact structuring as well as for the wet chemical polishing process
as outlined in the following paragraphs.
As described in the previous section, high-efficiency laboratory-type PERC solar
cells often passivate the silicon wafer surfaces by means of silicon dioxide (SiO2 )
grown at high temperatures (≥800 °C) in a furnace [3]. However, this process is
rather long and expensive and the SiO2 passivation is dissolved by screen-printed Al
pastes during firing. Hence, high-throughput low-cost surface passivation processes
were required for industrial high-efficiency silicon solar cell manufacturing. Hezel
and Jaeger [38] evaluated silicon nitride (SiNx ) grown by plasma-enhanced chemical
vapor deposition (PECVD) at temperatures around 400 °C for surface passivation of
silicon solar cells. PECVD obtains a high deposition rate resulting in short process
times of just several seconds per wafer enabling high-throughput and low-production
costs. Hence, PECVD SiNx is already since many years applied as a front-side
passivation layer of the phosphorus-doped emitter in industrial Al-BSF solar cells
and seemed to be a promising candidate for the rear-side passivation of industrial
PERC solar cells [12, 39, 40]. However, it was shown that on p-type rear silicon
surfaces of PERC cells SiNx passivation layers induce an inversion of the surface
due to the positive fixed charges within the SiNx layer leading to an enhanced carrier
recombination which was identified and denoted as “parasitic shunting” by Dauwe
et al. [20] in 2002. In contrast, aluminum oxide (AlOx ) was found to exhibit a
high negative fixed charge density thereby being well suited to passivate p-type
silicon surfaces [18, 19]. Hezel and Jaeger [38] reported already in 1989 that AlOx
passivates crystalline silicon surfaces quite well. Almost two decades later in 2006,
Agostinelli et al. [41] and Hoex et al. [42] demonstrated excellent surface passivation
properties of AlOx single layers on p-type silicon wafers. The initial research applied
lab-type low-throughput atomic layer deposition (ALD) of AlOx layers [19, 43].
After demonstrating the high surface passivation quality, high-throughput deposition
processes and tools have been developed such as PECVD [16, 17, 44–46] and spatial
ALD [47, 48] which are capable of processing several thousand silicon wafers per
hour and per production tool. For application of AlOx as rear passivation in industrial
PERC solar cells, the AlOx layer has to be capped with a SiNy layer in order to
prevent the dissolution of the AlOx layer by the Al paste during firing. Additionally,
the SiNy capping layer increases the internal optical reflectance which in turn leads
to improved light trapping [49].
Regarding low-cost rear contact patterning instead of very expensive photolithog-
raphy, in the year 2000 Preu et al. [21] demonstrated that laser ablation of the rear
passivation layer is a promising industrial process to define the rear contact pattern
as it removes the passivation layer with only minimal damage of the silicon crys-
tal structure. Also, laser ablation allows to pattern the whole wafer rear side within
just a few seconds enabling a throughput of several thousand wafers per produc-
tion tool. The laser ablation approach was further developed by Agostinelli et al.
[12, 13] in 2005 into the so-called “i-PERC” concept which proposed for the first
time to combine laser ablation of the rear passivation layer with screen-printed Al
layers, thereby forming local Al-BSF regions by alloying of the silicon and the alu-
minum through the laser contact openings (LCO) during the furnace firing step. This
5 High-Efficiency Industrial PERC Solar Cells … 75

approach resulted then in large-area i-PERC solar cells with efficiencies up to 17.2%
[13] which outperformed the corresponding full-area Al-BSF cells by 0.7%abs . Gatz
et al. [22] were the first to apply line-shaped instead of point-shaped LCOs in com-
bination with improved aluminum pastes which resulted in PERC cell conversion
efficiencies of 19.4% in 2011. The company Sunrise proposed to combine the ben-
efits of LCO lines for the Si–Al alloying process with the benefit of LCO points for
minimizing the contact area fraction by applying for the first time dashed-line LCO
contact geometry [23]. In parallel to the aforementioned LCO research activities,
the composition of Al pastes has been further developed to improve Al–Si alloy-
ing and local Al-BSF formation resulting in better Al contact properties. In 2006,
Meemongkolkiat et al. [50] demonstrated that a key requirement is to use a low con-
tent of glass frits in the Al paste in order to prevent etching of the rear passivation
layer during furnace firing. Also, it was shown that the addition of certain materials
to the Al paste resulted in a strongly increased Al-BSF depth [50]. However, at that
time it was not disclosed neither what type of material was added to the Al paste nor
what the physical mechanism was that enabled the deeper Al-BSF. In 2011, Rauer
et al. [51] demonstrated that the addition of up to 24 wt% silicon to the Al paste
increases the Si concentration in the Al–Si melt during furnace firing resulting in a
thicker epitaxially regrown Al-BSF.
In order to obtain a single-sided phosphorus doping of the PERC cell front side,
the company Centrotherm developed an industrial PERC process flow where the rear
passivation layer (in this case a PECVD SiOx /SiNy stack) is deposited prior to the
phosphorus diffusion thereby acting as a diffusion barrier and protecting thecells
manufacturers rear side from being phosphorus doped [52]. This so-called “Cen-
taurus process” was the first turnkey solution for industrial PERC production lines
becoming available in 2011 [53]. A technological challenge of the Centaurus process
is that the rear passivation layer has to maintain its passivation properties after being
exposed to a high-temperature POCl3 diffusion. Centrotherm achieved this by apply-
ing an optimized PECVD SiOx /SiNy passivation stack, whereas AlOx /SiNy stacks
severely degrade during POCl3 diffusion. A second approach to obtain a single-sided
phosphorus doping relied on wet chemical etching of the rear wafer surface. Early
work in 2009 on rear side etching [54, 55] applied the process sequence texturing
- wet chemical polishing of the rear wafer surface—POCl3 diffusion—wet chemi-
cal junction isolation. This process sequence allowed very long polishing of the rear
surface resulting in smooth wafer rear surfaces without impacting the front-side phos-
phorus emitter by the reactive gas phase of the chemical polishing bath. However,
it was shown by Kranz et al. [56] that, in particular, when applying an AlOx /SiNy
passivation stack the rear surface roughness can be quite high without detrimental
impact on the conversion efficiency. Accordingly, further work by RENA and ISFH
in 2013 [14, 15] simplified the process sequence to texturing—POCl3 diffusion—
wet chemical rear polishing by developing a less reactive wet chemical polishing
chemistry resulting in less emitter front etching and a higher rear surface roughness.
Compared to the Centaurus process, one advantage of the polishing PERC process is
that it is compatible with the favorable AlOx /SiNy rear passivation stack. This maybe
76 T. Dullweber

one reason why the polishing process sequence as shown in Table 5.1 became the
most widely used process sequence for industrial PERC solar cell production.

5.3 Bifacial PERC+ Solar Cells

As described in the previous section and displayed in Fig. 5.6, industrial PERC cells
apply p-type wafers and a full-area screen-printed aluminum (Al) rear layer which
only locally contacts the silicon wafer at areas where the rear passivation has been
removed by laser contact opening (LCO). The full-area aluminum layer prevents
any transmission of diffuse sunlight from the rear side into the silicon wafer and
hence any bifacial applications of these industrial PERC cells. On the other hand,
bifacial solar cell concepts receive increasing interest for applications, e.g., in PV
power plants where the produced electricity can be increased by up to 25% using
bifacial instead of monofacial solar modules due to the absorption of diffuse light by
the module’s rear side [57, 58]. Accordingly, the photovoltaic technology roadmap

Fig. 5.6 Photographs of the front and rear sides as well as schematic cross-section drawings of
industrial PERC and PERC+ solar cells. PERC+ cells enable bifacial applications and reduce the Al
paste consumption while applying the same processing sequence in Table 5.1 as industrial PERC
solar cells. Applying a busbar-less metal grid design ISFH demonstrated a PERC+ cell with 22.1%
front-side efficiency. Figure reproduced from [11]. Copyright (2018) The Japan Society of Applied
Physics
5 High-Efficiency Industrial PERC Solar Cells … 77

ITRPV predicts a market share of bifacial solar cells of 40% until 2028 as shown in
Fig. 5.2b [8].
The concept of a bifacial solar cell absorbing sunlight from the front and rear
sides has first been published in a patent by H. Mori in 1960 [59]. At that time,
the idea was to use pn-junctions at the front and rear sides in order to improve the
collection of photogenerated charge carriers at both wafer surfaces. Bifacial solar
cells with industrially applicable manufacturing technologies have been developed
from the year 2000 onward mostly by applying n-type wafers. The passivated emitter
and rear totally diffused (PERT) solar cell concept has been adopted by ECN and
Yingli to the bifacial PANDA solar cell [60, 61] and by ISC and Bosch to the BiSoN
solar cell [62]. The so-called heterojunction solar cells which apply n-type wafers
and doped amorphous Si surface passivation layers have been developed by Sanyo
which now belongs to Panasonic as HIT solar cell [63] and by Meyer Burger and
CSEM to a turnkey process solution [64]. A challenge of these n-type PERT and
heterojunction cell concepts is that they typically apply screen-printed silver (Ag)
finger grids on both wafer sides and hence consume a high amount of expensive Ag
paste. Also, the n-PERT cells apply single-sided boron and phosphorus doping which
requires additional or other process steps compared to p-type PERC cells. Industrial
bifacial silicon solar cell concepts applying p-type wafers and a screen-printed Al
rear finger grid have been investigated as well. In 2001, ISFH introduced a bifacial
p-type solar cell where the rear Al grid fired through the SiNx rear passivation layer
without using any laser contact opening or rear-side boron doping [40]. ECN further
developed this approach to the so-called p-PASHA cell concept [65]. However, the
published efficiencies of these bifacial p-type solar cells with fired through Al fingers
were just similar to full-area Al-BSF cells and significantly lower compared to typical
industrial PERC cell efficiencies probably due to insufficient local Al-BSF formation.
ISFH and SolarWorld first published a bifacial PERC solar cell in 2015 by apply-
ing a screen-printed rear Al finger grid instead of the conventional full-area aluminum
(Al) rear layer [24, 66], see Figs. 5.3c and 5.6. As shown in Table 5.1, the bifacial
PERC cell used the same manufacturing sequence compared to monofacial PERC
cells with only minimal recipe modifications for rear passivation, laser contact open-
ing, and Al screen printing in order to produce bifacial PERC solar cells without
investing into new or different production tools. Dullweber et al. named this novel
cell concept PERC+ and demonstrated several advantages in the initial publications
[24, 66]. The Al finger grid enabled bifacial applications of the PERC+ cells with
rear-side efficiencies up to 16.7% while still obtaining very high front-side effi-
ciencies above 21% comparable to monofacial PERC cells [66]. The corresponding
bifaciality was up to 80.0%. Also, as outline in the next sections, PERC+ improves
the Al-BSF contact properties, and when combined, e.g., with a white back sheet,
obtains a higher rear-side reflectance [66]. Another advantage is that Al paste con-
sumption of the PERC+ cells with Al finger grid is reduced up to 90% compared to
PERC cells with full-area Al layer [24]. In 2015, two additional papers addressed the
concept of bifacial PERC+ cells. Trina Solar developed glass/glass modules applying
bifacial PERC+ solar cells with colored rear-side passivation designed for aesthetic
optical appearance in building integrated photovoltaic applications [67]. These mod-
78 T. Dullweber

ules where, however, not designed and optimized for maximizing bifacial energy
yield, e.g., in PV power plants. Krauß et al. performed numerical simulations of the
potential front and rear conversion efficiencies of bifacial PERC+ cells and their
corresponding bifacial gains [68].
Meanwhile, PERC+ solar cells have been introduced into mass production by
many solar cells manufacturers. Table 5.2 summarizes published PERC+ conversion
efficiencies when illuminated from the front or rear side. ISFH sets the benchmark
in 2015 for high PERC+ front and rear-side efficiencies as outlined above. First pub-
lished PERC+ conversion efficiencies of solar cell manufacturers were 20.3% [67]
in 2015 and 20.7% [69] in 2016 which then continuously improved to 22.0% in 2017
as reported by LONGi Solar [70]. The highest PERC+ efficiency so far has been
reported by ISFH by developing a busbar-less PERC+ solar cell [11] by screen print-
ing only the Ag fingers on the front side without the Ag busbars as shown in Fig. 5.6,
thereby reducing the shading losses of the Ag front grid. Industrially, this approach
is attractive when combined with wire-based module interconnection technologies
as will be described in a later paragraph. The busbar-less PERC+ cell exhibits an
independently confirmed conversion efficiency of 22.1% [11] when illuminated from
the front side, see the last line in Table 5.2. Due to the continuous efficiency improve-
ments of industrial PERC solar cells as shown in Fig. 5.5, very likely even higher
PERC+ front-side efficiencies will be obtained soon. At the same time, the conver-
sion efficiency when illuminated from the rear side of PERC+ cells produced by cell
manufacturers improved from 13.9% [69] in 2016 to 17.5% [70] in 2017 as shown in
Table 5.2. For the 22.1% efficient PERC+ cell fabricated by ISFH in 2017 [11], the
rear-side efficiency was not published as the rear Al finger grid has been optimized
for monofacial applications and not for high bifaciality. Unfortunately, several lead-
ing solar cell manufacturers such as SolarWorld and Trina Solar which are producing

Table 5.2 Published efficiencies of industrial PERC+ solar cells when illuminated from the front
or from the rear side. Several leading solar cell manufacturers such as SolarWorld and Trina Solar
are producing bifacial PERC+ cells and modules (see Table 5.3) but did not publish any or recent
PERC+ cell efficiencies
Year Efficiency [%] front/rear Organization Source/comments
2015 21.5/16.7 ISFH [66]; industrial process
flow. No rear Ag pads
2015 20.3/n.p. Trina Solar [67]; optimized for optical
appearance in BIPV
2016 20.7/13.9 Big Sun Energy [69]
Technology Inc
2017 21.5/16.1 JinkoSolar [71]
2017 21.4/n.p. Neo Solar Power [72]
2017 22.0/17.5 LONGi Solar [70]
2017 22.1a /n.p. ISFH [11] busbar-less Ag front
grid
n.p.: not published; a independently confirmed
5 High-Efficiency Industrial PERC Solar Cells … 79

Table 5.3 Commercially available bifacial modules applying PERC+ solar cells. Maximum rating
power stated for front-side illumination under standard test conditions (STC, AM1.5, 25 °C) and
without rear-side illumination
Company/Product Maximum rating power [Wp] Comments Source/year
SolarWorld/Bisun 290 5 BB, 60 cells [73]/2017
Neo Solar Power/Glory Bifi 300 4 BB, 60 cells [74]/2017
Trina Solar/DUOMAX 300 5 BB, 60 cells [75]/2017
LONGi Solar/LR6-60PD 305 4 BB, 60 cells [76]/2017
Jinko Solar/Swan bifacial 60H 330 5 BB, 60 cells [77]/2019
Canadian Solar/Biku Module 315 5 BB, 60 cells [78]/2019
LONGi Solar/Hi-MO4 365 6 BB, 60 cells [79]/2019
bifacial

bifacial PERC+ cells and modules (see Table 5.3) did not publish any or no recent
PERC+ cell efficiencies and hence do not appear at all or only with initial results in
Table 5.2.
One key issue with the development of bifacial PERC+ cells is the very high spe-
cific resistivity of 20 µcm [25] of screen-printed Al fingers which is approximately
six times higher compared to screen-printed Ag fingers. This is probably the main
reason why most alternative industrial bifacial solar cell concepts focus on Ag finger
grids on both wafer sides as outlined at the beginning of this section. Hence, in case
of PERC+, the rear Al finger grid has to be designed in a way to minimize series
resistance losses caused by the Al finger lines. Dullweber et al. calculated the series
resistance contribution Rs,L of the Al finger grid [11] in dependence of the number
of busbars/wires and the Al finger width applying the equation

2 lf
Rs,L = ρf · Auc (5.1)
3 Af

with the specific resistivity ρ f of the Al paste, the finger length lf , and the finger
cross-section area Af . The area of the unit cell of the Al finger grid layout (half finger
pitch times half busbar distance) is described by the parameter Auc . The calculated
series resistance contribution Rs,L of the Al finger grid is shown in Fig. 5.7 [11]. In
order not to significantly reduce the front-side efficiency when changing from PERC
to PERC+, as a rule of thumb Rs,L should remain below 0.05 cm2 . As this is not pos-
sible with a three-busbar configuration due to the high Al finger length lf in between
the busbars, the five-busbar design can be considered as an enabling technology of
bifacial PERC+ cells when applying wide Al fingers around 150 µm. When mov-
ing to narrow Al fingers below 100 µm width, multi-wire module interconnection
technologies with, e.g., 10 or 20 wires per PERC+ cell drastically minimize resistive
losses of the Al fingers below 0.02 cm2 .
The fine-line-print capability of Al finger pastes strongly improved over the past
few years when Al paste vendors started to develop PERC+ specific Al pastes which
80 T. Dullweber

Fig. 5.7 Calculated series resistance contribution Rs,L of the Al finger grid in dependence of the
number of busbars/wires and the Al finger width. In order not to reduce the front-side efficiency
when changing from PERC to PERC+, the series resistance increase caused by the Al finger grid
should remain below 0.05 cm2 . As this is not possible with a three-busbar configuration due to
the high Al finger length, the five-busbar design can be considered as an enabler of bifacial PERC+
cells. When moving to very narrow Al fingers below 100 µm width, multi-wire technologies with,
e.g., 10 or 20 wires help to minimize resistive losses of the Al fingers. Figure reproduced from [2,
11] with permission. Copyright (2018) The Japan Society of Applied Physics

reduce the spreading behavior of Al pastes during screen printing. Figure 5.8 shows
the measured Al finger widths in dependence of the screen opening widths obtained
with different Al pastes [12]. Whereas conventional PERC Al pastes designed for
full-area Al screen print show an extreme spreading behavior resulting in wide Al
fingers, and the novel PERC+ Al pastes enable 50–100 µm narrow Al fingers when
applying 50–75 µm screen openings as shown in Fig. 5.8. The latest Al finger pastes
with 50 µm Al finger width approach the fine-line printing capability of actual Ag
finger pastes with printed Ag finger widths of only 35 µm as sketched in Fig. 5.8.
When printing very narrow Al fingers, it becomes challenging to precisely align
the Al fingerprint on top of the laser contact openings (LCO). In case of extreme
misalignment when the Al finger does not overlap the LCO area, the open and laser-
damaged silicon surface of the LCO area leads to very high surface recombination
of minority charge carriers [11]. Accordingly, the alignment tolerance between Al
fingerprint and LCO of approximately ±30 µm depending on the detailed Al finger
and LCO geometries needs to be fulfilled for high PERC+ conversion efficiencies.
This requires high precision laser processes and Al screens as well as camera-based
alignment schemes between LCO and Al screen print.
First commercial bifacial PERC+ glass/glass modules have been introduced by
SolarWorld at the Intersolar 2015 [80] followed by Neo Solar Power, Trina Solar, and
LONGi Solar in 2017 [74–76]. Meanwhile, almost all major solar cell manufacturer
offer commercial bifacial glass/glass modules applying PERC+ solar cells as sum-
marized in Table 5.3. The interconnection of PERC+ cells to strings is accomplished
5 High-Efficiency Industrial PERC Solar Cells … 81

Fig. 5.8 Measured aluminum finger widths in dependence of the screen opening widths. Whereas
conventional Al pastes designed for full-area print of PERC solar cells exhibit an extreme spreading,
recently developed Al pastes for PERC+ fingerprints enable Al finger widths close to 50 µm. In
comparison, Ag pastes obtain final finger widths around 35 µm with screen opening widths of
25–30 µm. Graph updated from [11]

using conventional tabbing–stringing technology where the Cu ribbons are soldered

to the Ag front busbars and to Ag pads on the PERC+ rear side. All manufacturers
apply four or five busbar designs in order to minimize Al finger resistance losses as
shown in Fig. 5.7 and Table 5.3. Since the first publication by Hannebauer et al. in
2014 of a PERC cell applying a five-busbar design [30], the PV industry has migrated
from three busbars to four or five busbars which is beneficial to the industrial adoption
of PERC+ cells. As shown in Table 5.3, the commercially available maximum rating
powers range between 315 and 365 Wp with 60 PERC+ cells per module [77–79].
The rating power is stated for front-side illumination only. When applying addi-
tional rear-side illumination, the output power increases accordingly. For example,
with 10% additional rear-side illumination, the output power will increase by approx.
8% approaching 400 Wp total output power. The resistive losses caused by PERC+
rear Al fingers can be further reduced by moving from four or five busbar designs
to PERC+ solar cells without any busbars. In this case, the module interconnection
is accomplished by, e.g., 10 or 20 wires which reduce the Al finger lengths to a few
millimeters, and hence drastically minimizing its grid line resistance as shown in
Fig. 5.7. The reduced grid line resistance allows a further reduction of the Al finger
width to well below 100 µm which minimizes shadowing losses of the Al finger grid
and increases rear efficiency and bifaciality. In 2017, ISFH has developed a wire-
interconnected PERC+ prototype module [81] where 18 wires are soldered directly
to the Ag front and Al rear fingers applying the smart wire connection technology
(SWCT) of Meyer Burger [82] without using Ag busbars or Ag pads. The PERC+
82 T. Dullweber

SWCT module exhibits an independently confirmed front-side efficiency of 19.8%,

a rear-side module efficiency of 16.4%, and a bifaciality of 83% [81].
When applying bifacial PERC+ modules to outdoor field installations, in partic-
ular, the albedo of the ground and the mounting height of the module are important
parameters to maximize the additional diffuse illumination of the module’s rear side
and hence the additional energy yield of bifacial PERC+ modules compared to mono-
facial PERC modules. Numerical simulations predict that the energy yield of PERC+
modules can increase up to 25% when mounted 0.5 m above ground with an albedo of
80% [24]. SolarWorld took the effort to measure the energy yield of Bisun modules
in two different small-scale outdoor test installations. A 3.2 kWp installation on top
of a flat roof with a high albedo of 74% due to white ballast stones demonstrated an
increased energy yield of 13% compared to a monofacial PERC reference module
which compared well to the calculated energy yield of 13.3% [83]. When mounted to
a two-axis tracker system, the Bisun modules obtained 10% additional energy yield
compared to monofacial PERC modules mounted on a two-axis tracker where the
sandy ground had an albedo of 30% [83]. Trina Solar measured the energy yield of
bifacial DUMAX modules either with a fixed mounted installation or mounted on
a one-axis tracker and compared the results to a fixed mounted monofacial PERC
module [84]. The fixed mounted bifacial modules demonstrated a 10% increased
energy yield and the bifacial DUOMAX modules mounted on a one-axis tracker
demonstrated a 24% increased energy yield compared to the fixed mounted monofa-
cial modules [84], both on sandy ground. Hence, the combination of bifacial PERC+
modules with one-axis trackers is a very promising option to strongly increase the
energy yield of photovoltaic power plants. Meanwhile, several large scale PV power
plants with electricity generation capacity up to 224 MWp applying bifacial PERC+
solar cells are under construction [85–88]. Latest solar bids for new PV power plants
approach an electricity price of $0.02/kWh where the combination of bifacial PERC+
cells with one-axis-trackers strongly contributes to reducing the PV electricity price
[89] (Fig. 5.9). Accordingly, a recent market research study predicts that bifacial

Fig. 5.9 Photograph of a

large scale 20 MWp PV
power plant installed by
LONGi in Golmud, China,
applying bifacial PERC+
solar cells in combination
with one-axis trackers
thereby maximizing the
energy yield from the
modules front and rear side
[91]
5 High-Efficiency Industrial PERC Solar Cells … 83

PERC+ module production capacity will strongly increase to 17% of the global solar
module production capacity in 2021 [90].

5.4 AlOx /SiNy Rear Passivation and Local Al Rear

Contacts

In order to obtain high conversion efficiencies with PERC and PERC+ solar cells,
as shown in Fig. 5.4, the effective minority charge carrier recombination velocity
S eff,rear at the rear side has to be as low as possible. S eff,rear can be calculated from
the surface recombination velocities S pass in the AlOx /SiNy passivated area and in
the Al contacted area S cont according to an analytical model proposed by Fischer
[92] for contact patterns where the Al contact spacing is large compared to the wafer
thickness:
 −1
Rs − ρW 1 Spass
Seff,rear = + + (5.2)
ρD fScont 1−f

The parameters in (2) are the wafer bulk spreading resistance contribution RS and
the diffusion coefficient of the minority charge carriers D = 26.9 cm2 /s. The sili-
con wafer related parameters are the specific resistivity which is typically between
1 and 2 cm for PERC cells and the wafer thickness W around 180 µm. The area
fraction f of the Al rear contacts is typically between 1 and 5%. In order to obtain
high conversion efficiencies by a low S eff,rear , a low surface recombination velocity
S pass in the AlOx /SiNy passivated area of below 10 cm/s is required as well as a
low surface recombination velocity S cont below 500 cm/s at the local Al rear con-
tacts. Applying these parameter settings to (2) results in S eff,rear < 50 cm/s which
is a typical value for high-efficiency PERC solar cells. An easy to use and hence
practical approximation of (2) calculates the saturation current density of the whole
rear-side J 0,rear based on the saturation current densities of the rear passivation J 0,pass
and the local Al contacts J 0,cont applying the equation J 0,rear = (1 − f ) × J 0,pass + f
× J 0,cont . Both equations favor small contact area fractions f which can be obtained
either by reducing the laser contact opening line widths or dot diameter or by increas-
ing the spacing of the laser contact openings. However, both approaches have their
limitations. When reducing the laser contact opening line width or dot diameter
too much, the Al–Si alloying is affected resulting in narrower Al-BSF depths and
increased contact recombination as will be described in the following paragraphs.
When increasing the spacing of the laser contact openings, the bulk spreading resis-
tance increases since the majority charge carrier has to travel a longer distance to
reach the rear contacts. Hence, optimizing the rear contact geometry of industrial
PERC and PERC+ solar cells is a complex task since many parameters have to be con-
sidered. In any case, reducing the specific surface recombination velocities S pass and
84 T. Dullweber

S cont of PERC solar cells was and still is an important task for further improvements
as outlined in the following paragraphs.
Aluminum oxide (AlOx ) was first published by Hezel and Jaeger [38] in 1989
to passivate crystalline silicon wafer surfaces with measured surface recombination
velocities around 100 cm/s. However, it took until 2006 when Agostinelli et al. [41]
and Hoex et al. [42] demonstrated excellent surface recombination velocities below
15 cm/s by applying atomic layer deposited (ALD) AlOx layers on p-type silicon
wafer surfaces. Meanwhile, many publications have reported excellent surface pas-
sivation properties of AlOx layers [19, 43, 92–96] by applying many different depo-
sition processes such as atomic layer deposition (ALD) [19, 43], plasma-enhanced
chemical vapor deposition (PECVD) [16, 17], inductively coupled plasma (ICP)
deposition [44, 45], atmospheric pressure chemical vapor deposition (APCVD) [38,
97], or reactive sputtering [94]. AlOx is ideally suited to the rear passivation of p-type
PERC solar cells, as the parasitic shunting observed with SiNx layers is completely
absent—thanks to the fixed negative charges [18, 19]. Figure 5.10 displays the sur-
face recombination velocity S of AlOx layers deposited by ALD and ICP versus an
external charge Qc applied to the AlOx layers by corona discharge [45]. Without
external charges, both layers obtain S values below 10 cm/s. At a positive corona
charge around 4 × 1012 cm−2 , the internal negative charge is compensated resulting
in highest S values of around 200 cm/s. Hence, this measurement nicely demon-
strates the importance of the negative charges of AlOx layers which repel minority
charge carriers (electrons in case of PERC cells) from the rear side thereby mini-
mizing carrier recombination. Industrial-type PERC solar cells use screen printing

Fig. 5.10 Surface recombination velocity S of AlOx layers deposited by ALD and ICP versus an
external charge Qc applied by corona discharge. Without external charges, both AlOx layers obtain
excellent S values below 10 cm/s. At a positive external corona charge around 4 × 1012 cm−2 , the
internal negative charge is compensated resulting in highest S values of around 200 cm/s. Figure
taken from [45]
5 High-Efficiency Industrial PERC Solar Cells … 85

of metal pastes and firing in a belt-line furnace at high temperatures (750–850 °C)
for contact formation, which makes an excellent thermal stability of the surface pas-
sivation indispensable. As demonstrated by Schmidt et al., thin AlOx layers tend to
degrade during firing; however, a hydrogen-containing SiNy capping layer deposited
by PECVD improves the firing stability considerably [95]. The SiNy capping layer
also serves as a protection layer since the AlOx is not stable against the aluminum
paste during firing. The annealing step which is required for an excellent passivation
quality of the AlOx layer can be integrated into the SiNy deposition by heating the
wafers in the PECVD chamber before depositing the SiNy [17, 45, 98]. Meanwhile,
AlOx /SiNx stacks deposited by PECVD are the preferred choice as rear passivation
of industrial PERC solar cells in mass production [6].
In addition to the AlOx /SiNx rear passivation, the local aluminum (Al) rear con-
tacts strongly influence the current voltage parameters of industrial PERC solar cells.
As can be seen in (5.2), the surface recombination velocity S cont of charge carriers
at the local Al contacts has to be small. The carrier recombination at the Al contact
can be reduced by the local aluminum p+ doping called the aluminum back surface
field (Al-BSF) surrounding the contact. The Al-BSF forms a p/p+ doping gradient
and hence a built-in electric field which repels the electrons from the Al contact,
whereas the hole transport is not blocked by the Al-BSF. Hence, understanding the
Al–Si alloying process and Al-BSF growth during furnace firing is the key in order
to optimize the electrical Al contact properties. When the wafer temperature during
furnace firing exceeds 660 °C, the aluminum melts and according to the Al–Si phase
diagram [99], silicon from the wafer dissolves in the aluminum paste at areas where
the rear passivation has been removed by LCO. If the firing temperature and time are
sufficiently high, the silicon concentration in the aluminum melt exceeds the eutectic
concentration of 12.6 wt% of silicon in aluminum [99]. During cooldown of the firing
process, excess silicon in the Al paste beyond the eutectic concentration epitaxially
recrystallizes at the silicon wafer surface thereby incorporating aluminum up to its
solid solubility limit which is in the range of 5 × 1018 cm3 . Aluminum acts as p-type
dopant in the silicon crystal lattice thereby forming the local Al-BSF. The specific
challenges of optimizing local Al–Si contacts of PERC solar cells as well as screen
printing Al fingers in PERC+ cells are outlined in the following sections.
Whereas the lab-type PERC cell of Blakers et al. [3] applied point-shaped local
rear contacts in combination with evaporated Al, in 2011 it was shown by Lauer-
mann et al. [100] that point contacts in industrial-type PERC cells processed with
local removal of the passivation layer, Al screen printing and firing cause severe void
formation similar to Fig. 5.11a resulting in a scattering of the Al-BSF depth from 0
to 6 µm. In contrast, according to [100], line-shaped local rear contacts decrease the
amount of voids and increase the Al-BSF depth to 4–7 µm thereby reducing contact
recombination and increasing PERC conversion efficiencies. Further investigations
applying carrier lifetime measurements demonstrated that compared with point con-
tacts line contacts achieve a lower surface recombination velocity down to 800 cm/s
[101]. Gatz et al. [102] revealed that the Al-BSF depth increases with increasing LCO
line width thereby reducing the contact recombination velocity down to 300 cm/s.
The application of an LCO line width of 80 µm, full-area Al screen printing and
86 T. Dullweber

Fig. 5.11 Typical scanning electron microscope (SEM) images of locally alloyed aluminum con-
tacts of a PERC and b PERC+ solar cells. Whereas the local Al contacts of PERC cells often show
voids and a thin back surface field (BSF), PERC+ cells exhibit filled contacts and a deep BSF.
Al contact geometries used for physical models in Fig. 5.12 are shown in image b such as the Al
contact width a, the contact height h, and the Al-BSF depth W p+ . Figure (a) is reproduced from [2]
with permission

firing to industrial-type PERC solar cells resulted in a record efficiency of 19.4%

in 2011 [22]. In the same year, Urrejola et al. [103] systematically investigated the
diffusion of silicon from the wafer through local line-shaped contact openings into
the aluminum paste during furnace firing. One main result was that the maximum
spread of silicon in the aluminum paste ranges between 300 and 1000 µm depending
on the peak firing temperature and on the opening width. Hence, the silicon diffusion
and following Al–Si alloy formation extend several hundred µm beyond the local
rear contact edges into the aluminum paste. Taking this finding into account, in 2012,
Müller et al. [104] proposed an analytical model which calculates the Al-BSF thick-
ness in dependence of the contact geometry and several other process parameters.
The model assumes a box-shaped silicon distribution in the aluminum paste with
eutectic concentration after firing. The width of the box is used as a fit parameter to
adjust the model to experimental Al-BSF values. The assumption of a box profile
rather than the more realistic Gaussian silicon concentration profile greatly simplifies
the model since no differential equations for the silicon diffusion have to be solved.
The model further assumes that the excess silicon in the aluminum paste above the
eutectic concentration recrystallizes as Al-BSF during cooldown of the firing step.
The model reproduces the experimental finding that a certain minimum line width
or a minimum point diameter is required to obtain several µm deep Al-BSFs as
needed for low contact recombination and high-efficiency PERC cells. According to
the model, the physical reason is that in case of thin lines or small points the large
silicon spread beyond the LCO edges in the aluminum paste significantly reduces
the silicon concentration in the aluminum paste and hence the amount of silicon
which is available for the Al-BSF formation. In the extreme situation of very small
LCO contacts, the silicon concentration in the aluminum paste is reduced below the
eutectic concentration and hence no silicon is available for the Al-BSF growth. The
model also describes accurately that line-shaped LCO contacts limit the Si diffu-
sion to one direction, and hence obtain a high Si concentration and deeper Al-BSF
5 High-Efficiency Industrial PERC Solar Cells … 87

compared to point-shaped contacts, where the Si diffuses in all directions from the
LCO into the Al layer. In addition to optimizing the LCO geometry another approach
to deeper Al-BSF is to modify the composition of the Al paste, e.g., by adding Si
powder in order to increase the Si concentration available for Al-BSF regrowth. In
2006, Meemongkolkiat et al. [50] showed that the addition of certain materials to
the Al paste resulted in a strongly increased Al-BSF depth, however, without dis-
closing what type of materials was added to the Al paste. In 2011, Rauer et al. [51]
demonstrated that the addition of up to 24 wt% silicon to the Al paste increased the Si
concentration in the Al–Si melt during furnace firing resulting in a thicker epitaxially
regrown Al-BSF.
All investigations of the Al-BSF formation as described above favor either wide
line-shaped rear contacts or point-shaped rear contacts with very large diameter in
order to fabricate a deep Al-BSF and hence to reduce the surface recombination
velocity at the Al rear contacts. Nevertheless, for optimum PERC conversion effi-
ciencies, the contacted area fraction should be small in order to keep the absolute
contact recombination small as shown in (5.2). Also, the wafer bulk spreading resis-
tance increases with wider contact spacing, hence limiting the efficiency. These con-
siderations favor rear contact geometries which achieve a sufficiently deep Al-BSF
with as small as possible contact area. An interesting approach to achieve this was
introduced in 2013 by the company Sunrise [23]. They proposed a dashed-line rear
contact geometry which potentially achieves a deep Al-BSF due to the line-shaped
LCO contacts and simultaneously a small metalized area fraction in combination
with low bulk resistance contribution due to the arrangement of the dashed lines
similar to point-shaped contact patterns [23]. Another reason why industrial PERC
cells favor as small as possible LCO contact areas is to limit the LCO process time
to a few seconds per wafer enabling throughputs of several thousand wafers per hour
per LCO production tool. This is the main reason why a typical industrial LCO line
width today is in the range of 20–40 µm [105]. Thanks to huge progress in optimizing
PERC-specific Al pastes as discussed above, today several µm deep Al-BSFs can be
obtained with industrially feasible 30 µm narrow LCO openings.
In case of PERC+ solar cells, the Al finger width confines the diffusion of silicon
in the Al paste thereby potentially increasing the Si concentration resulting in deeper
Al-BSFs compared to full-area Al layer in PERC cells [24]. Figure 5.11 shows typical
SEM images of local Al contacts of PERC and PERC+ solar cells. Whereas the local
Al contacts of PERC cells often show voids and a thin BSF, PERC+ cells exhibit
filled contacts and a deep BSF [24, 106]. Figure 5.12a shows measured Al-BSF
depths demonstrating that PERC+ cells obtain up to 8 µm deep Al-BSFs compared
to PERC with up to 6 µm. The black line represents a fit applying the model of
Mueller [104] to the PERC data. The red line shows a fit to the PERC+ data using
an extended model that takes the limited Al volume of the Al fingers into account
[106]. According to the model, the limited Al volume of the Al fingers leads to a
higher silicon concentration in the screen-printed aluminum during furnace firing
causing thicker Al-BSFs during the epitaxial regrowth in the cooldown phase [106].
This effect becomes more pronounced for narrow Al contact widths around 50 µm
which are industrially preferred in order to increase the throughput of the LCO tool.
88 T. Dullweber

Fig. 5.12 a Maximum Al-BSF depth of PERC (black) and PERC+ (red) solar cells in dependence
of the Al contact width a. PERC+ cells obtain a deeper Al-BSF compared to PERC since the Al
fingers confine the Si diffusion during firing according to experimental data (dots) and analytical
models (lines). b Voided Al contacts symbolized by open circles are often found for large contact
heights were the Al alloys deeply into the silicon wafer. This finding is explained by modeling
the surface energies of the Si wafer, the Al particles, and the liquid Al–Si melt. The Al contact
parameters W p+ , h, and a are defined in Fig. 5.11b. Both graphs are reproduced from [107] with
permission

Figure 5.12a, b uses different symbol styles where filled circles indicate a filled
local Al contact, e.g., similar to Fig. 5.11b, half-filled circles a partially voided Al
contact similar to Fig. 5.11a and open circles a fully voided Al contact where the
void extends across the whole contact width a. According to Fig. 5.12b, voids, in
particular, occur for Al contacts with large contact height h where the Al–Si eutectic
extends more than 20 µm deep into the Si wafer [107]. This finding is explained by
an analytical model proposed by Kranz et al. that calculates the surface energies of
the liquid Al–Si melt, the Si wafer surface, and the screen-printed Al particle surface
[107]. According to this model, voids form for large contact heights since then a
sufficient amount of Al–Si melt is available in order to wet the large surface area of
the Al particles rather than the small Si wafer surface area thereby minimizing the
total surface energy of the Al–Si system. The PERC+ Al fingers reduce the Al contact
height by about 7 µm which is the reason why PERC+ cells do not exhibit voids
in contrast to PERC cells [107]. The increased Al-BSF thickness and the reduced
amount of voids of PERC+ cells compared to conventional PERC cells result up to
3 mV higher open-circuit voltages V oc due to reduced rear contact recombination
[24].
5 High-Efficiency Industrial PERC Solar Cells … 89

References

1. T. Dullweber, J. Schmidt, Surface passivation of industrial PERC solar cells, in Surface Pas-
sivation of Crystalline Silicon Solar Cells, ed. by J. John (The Institution of Engineering and
Technology, 2018, in press)
2. T. Dullweber, p-PERC+ solar cells, in Bifacial Photovoltaics: Technology, Applications and
Economics, ed. by R. Kopecek, J. Libal (The Institution of Engineering and Technology,
2018), pp. 46–54
3. A.W. Blakers, A. Wang, A.M. Milne, J. Zhao, M.A. Green, 22.8% efficient solar cell. Appl.
Phys. Lett. 55, 1363–1365 (1989)
4. J. Zhao, A. Wang, M.A. Green, F. Ferrazza, Novel 19.8% efficient “honeycomb” textured
multicrystalline and 24.4% monocrystalline silicon solar cells. Appl. Phys. Letters 73, 1991–
1993 (1998)
5. IHS Markit - PV Module Supply Chain Tracker 2018
6. International Technology Roadmap for Photovoltaic (ITRPV.net), 2017 Results, March 2018,
http://www.itrpv.net/Reports/Downloads/
7. International Technology Roadmap for Photovoltaic (ITRPV.net), 2018 Results, March 2019,
http://www.itrpv.net/Reports/Downloads/
8. A. Metz, D. Adler, S. Bagus, H. Blanke, M. Bothar, E. Brouwer, S. Dauwe, K. Dressler,
R. Droessler, T. Droste, M. Fiedler, Y. Gassenbauer, T. Grahl, N. Hermert, W. Kuzminski,
A. Lachowicz, T. Lauinger, N. Lenck, M. Manole, M. Martini, R. Messmer, C. Meyer, J.
Moschner, K. Ramspeck, P. Roth, R. Schönfelder, B. Schum, J. Sticksel, K. Vaas, M. Volk,
K. Wangemann, Industrial high performance crystalline silicon solar cells and modules based
on rear surface passivation technology. Sol. Energy Mat. Sol. Cells 120, 417–425 (2014)
9. M.A. Green, The Passivated Emitter and Rear Cell (PERC): From conception to mass pro-
duction. Solar Energy Mat. Solar Cells 143, 190–197 (2015)
10. T. Dullweber, J. Schmidt, Industrial silicon solar cells applying the Passivated Emitter and
Rear Cell (PERC) concept—a review. IEEE J. Photovolt. 6, 1366–1381 (2016)
11. T. Dullweber, H. Schulte-Huxel, S. Blankemeyer, H. Hannebauer, S. Schimanke, U. Baumann,
R. Witteck, R. Peibst, M. Köntges, R. Brendel, Y. Yao, Present status and future perspectives
of bifacial PERC+ solar cells and modules. Jpn. J. Appl. Phys. 57, 08RA01 (2018)
12. G. Agostinelli, P. Choulat, H.F.W. Dekkers, S. de Wolf, G. Beaucarne, Screen printed large
area crystalline silicon solar cells on thin substrates, in Proceedings of the 20th European
Photovoltaic Solar Energy Conference, Barcelona, Spain, 2005, pp. 647–650
13. G. Agostinelli, J. Szlufcick, P. Choulat, G. Beaucarne, Local contact structures for indus-
trial PERC-type solar cells, in Proceedings of the 20th European Photovoltaic Solar Energy
Conference, Barcelona, Spain, 2005, pp. 942–945
14. C. Kranz, S. Wyczanowski, U. Baumann, K. Weise, C. Klein, F. Delahaye, T. Dullweber, R.
Brendel, Wet chemical polishing for industrial type PERC solar cells. Energy Procedia 38,
243–249 (2013)
15. D. Pysch, C. Schmitt, B. Latzel, J. Horzel, R. Sastrawan, O. Voigt, B.-U. Sander, S. Patzig-
Klein, A. Pediaditakis, S. Queißer, A. Heeren, H. Nussbaumer, H. Kühnlein, X. Mao, X.
Qu, S. Yang, I. Melnyk, X. Gay, F. Souren, R. Görtzen, H. Heezen, Implementation of an
ALD-Al2O3 PERC-Technology into a multi- and monocrystalline industrial pilot production,
in Proceedings of the 29th European Photovoltaic Solar Energy Conference, Amsterdam,
Netherlands, 2014, pp. 612–616
16. P. Saint-Cast, D. Kania, M. Hofmann, J. Benick, J. Rentsch, R. Preu, Very low surface recom-
bination velocity on p-type c-Si by high-rate plasma-deposited aluminum oxide, Appl. Phys.
Lett. 95, 151502-1–151502-3 (2009)
17. H.-P. Sperlich, D. Decker, P. Saint-Cast, E. Erben, L. Peters, High productive solar cell passi-
vation on Roth & Rau MAiAs MW-PECVD inline machine—a comparison of Al2 O3 , SiO2
and SiNx –H process conditions and performance, in Proceedings of the 25th European Pho-
tovoltaic Solar Energy Conference, Valencia, Spain, 2010, pp. 1352–1357
90 T. Dullweber

18. G. Agostinelli, A. Delabie, P. Vitanov, Z. Alexieva, H.F.W. Dekkers, S. De Wolf, G. Beaucarne,

Very low surface recombination velocities on p-type silicon wafers passivated with a dielectric
with fixed negative charge. Sol. En. Mat. Sol. Cells 90, 3438–3443 (2006)
19. B. Hoex, J. Schmidt, P. Pohl, M.C.M. van de Sanden, W.M.M. Kessels, Silicon surface pas-
sivation by atomic-layer-deposited Al2O3. J. Appl. Phys. 104, 044903/1–12 (2008)
20. S. Dauwe, L. Mittelstädt, A. Metz, R. Hezel, Experimental evidence of parasitic shunting in
silicon nitride rear surface passivated solar cells. Prog. Photovolt. Res. Appl. 10, 271–278
(2002)
21. R. Preu, S.W. Glunz, S. Schäfer, R. Lüdemann, W. Wettling, W. Pfleging, Laser ablation—a
new low-cost approach for passivated rear contact formation in crystalline silicon solar cell
technology, in Proceedings 16th European Photovoltaic Solar Energy Conference, Glasgow,
United Kingdom, 2000, pp. 1181–1184
22. S. Gatz, H. Hannebauer, R. Hesse, F. Werner, A. Schmidt, T. Dullweber, J. Schmidt, K. Bothe,
R. Brendel, 19.4%-efficient large-area fully screen-printed silicon solar cells. Phys. Stat. Sol.
RRL 5, 147–149 (2011)
23. B. Tjahjono, M.J. Yang, V. Wu, J. Ting, J. Shen, O. Tan, T. Sziptalak, B. Liu, H.-P. Sperlich, T.
Hengst, B. Beilby, K.-C. Hsu, Optimizing CELCO technology in one year of mass production,
in Proceedings of the 28th European Photovoltaic Solar Energy Conference, Paris, France,
2013, pp. 775–779
24. T. Dullweber, C. Kranz, R. Peibst, U. Baumann, H. Hannebauer, A. Fülle, S. Steckemetz, T.
Weber, M. Kutzer, M. Müller, G. Fischer, P. Palinginis, D.H. Neuhaus, PERC+: industrial
PERC solar cells with rear Al grid enabling bifaciality and reduced Al paste consumption.
Prog. Photovolt: Res. Appl. 24, 1487–1498 (2016)
25. T. Dullweber, S. Gatz, T. Falcon, H. Hannebauer, High efficiency rear-passivated screen-
printed silicon solar cells. Photovolt. Int. 13 (2011)
26. K.A. Münzer, J. Schöne, A. Teppe, R.E. Schlosser, M. Hein, D. Hammer, S. Hüls, M. Hanke,
S. Keller, P. Fath, Advanced rear side technology for industrial high efficiency solar cells, in
Proceedings of the 25th European Photovoltaic Solar Energy Conference, Valencia, Spain,
2010, pp. 2314–2318
27. T. Boscke, R. Hellriegel, T. Wutherich, L. Bornschein, A. Helbig, R. Carl, M. Dupke, D.
Stichtenoth, T. Aichele, R. Jesswein, T. Roth, C. Schollhorn, T. Geppert, A. Grohe, J. Lossen,
H. Krokoszinski, Fully screen-printed PERC cells with laser-fired contacts—an industrial
cell concept with 19.5% efficiency, in Proceedings of the 37th IEEE Photovoltaic Specialists
Conference, Seattle, USA, 2011, pp. 3663–3666
28. Schott Solar AG, press release August 24th 2011, http://www.presseportal.de/pm/23114/
2101260
29. A. Lachowicz, K. Ramspeck, P. Roth, M. Manole, H. Blanke, W. Hefner, E. Brouwer, B.
Schum, A. Metz, Nox-free solution for emitter etch-back, in Proceedings of the 27th European
Photovoltaic Solar Energy Conference, Frankfurt, Germany, 2012, pp. 1846–1850
30. H. Hannebauer, T. Dullweber, U. Baumann, T. Falcon, R. Brendel, 21.2%-efficient fineline-
printed PERC solar cell with 5 busbar front grid. Phys. Status Solidi RRL 8, 675–679 (2014)
31. Trina Solar, press release, November 17th 2014, http://www.prnewswire.com/news-releases/
trina-solar-announces-new-efficiency-records-for-silicon-solar-cells-282907361.html
32. SolarWorld AG, press release July 17th 2015, http://www.solarworld.de/en/group/investor-
relations/news-announcements/corporate-news/single-ansicht/article/solarworld-ag-sets-
new-world-record-for-solar-cell-efficiency/
33. Trina Solar, press release, December 16th 2015, http://www.prnewswire.com/news-releases/
trina-solar-announces-new-efficiency-record-of-2213-for-mono-crystalline-silicon-solar-
cell-300193727.html
34. Trina Solar, press release, December 19th 2016, http://www.prnewswire.com/news-releases/
trina-solar-announces-new-efficiency-record-of-2261-for-mono-crystalline-silicon-perc-
cell-300380906.html
35. W.W. Deng, F.Y. Ye, R. Liu, Y. Li, H. Chen, Z. Xiong, Y. Yang, Y. Chen, Y. Wang, P.P.
Altermatt, Z. Feng, P.J. Verlinden, 44th IEEE Photovoltaic Specialists Conference, 2017, in
press
5 High-Efficiency Industrial PERC Solar Cells … 91

36. LONGi Solar, press release, October 23rd 2017, https://www.pv-tech.org/news/longi-solar-

plans-22-record-perc-cell-in-production-at-end-of-2017
37. JinkoSolar, press release, October 25th 2017, https://www.pv-tech.org/news/jinkosolar-
pushes-p-type-monocrystalline-perc-solar-cell-to-new-efficiency
38. R. Hezel, K. Jaeger, Low-temperature surface passivation of silicon for solar cells. J. Elec-
trochem. Soc. 136, 518–523 (1989)
39. A. Hübner, A.G. Aberle, R. Hezel, Novel cost-effective bifacial silicon solar cells with 19.4%
front and 18.1% rear efficiency. Appl. Phys. Lett. 70, 1008–1010 (1997)
40. S. Steckemetz, A. Metz, R. Hezel, Thin Cz-silicon solar cells with rear silicon nitride passi-
vation and screen printed contacts, in Proceedings of the 17th European Photovoltaic Solar
Energy Conference, Munich, Germany, 2001, pp. 1902–1905
41. G. Agostinelli, A. Delabie, P. Vitanov, Z. Alexieva, H.F.W. Dekkers, S. De Wolf, G. Beaucarne,
Very low surface recombination velocities on p-type silicon wafers passivated with a dielectric
with fixed negative charge. Sol. Energy Mat. Sol. Cells 90, 3438–3443 (2006)
42. B. Hoex, S.B.S. Heil, E. Langereis, M.C.M. van de Sanden, W.M.M. Kessels, Ultralow surface
recombination of c-Si substrates passivated by plasma-assisted atomic layer deposited Al2O3.
Appl. Phys. Letters 89, 042112 (2006)
43. J. Schmidt, A. Merkle, R. Brendel, M.C.M. van de Sanden, W.M.M. Kessels, Surface passiva-
tion of high-efficiency silicon solar cells by atomic-layer-deposited Al2O3. Prog. Photovolt.
Res. Appl. 16, 461–466 (2008)
44. B. Veith, T. Dullweber, M. Siebert, C. Kranz, F. Werner, N.-P. Harder, J. Schmidt, B.F.P. Roos,
T. Dippell, R. Brendel, comparison of ICP-AlOx and ALD-Al2O3 layers for the rear surface
passivation of c-Si solar cells. Energy Procedia 27, 379–384 (2012)
45. T. Dullweber, C. Kranz, B. Beier, B. Veith, J. Schmidt, B.F.P. Roos, O. Hohn, T. Dippell, R.
Brendel, Inductively coupled plasma chemical vapour deposited AlOx /SiNy layer stacks for
applications in high-efficiency industrial-type silicon solar cells. Sol. Energy Mat. Sol. Cells
112, 196–201 (2013)
46. F. Schwarz, R. Beckmann, N. Kohn, S. Nölker, S. Kastl, M. Hofmann, R. Ferré, T. Pernau,
H. Wanka, J. Rentsch, Advanced anti-reflection and passivation layer systems produced by
high-power plasma in the new Manz PECVD system, in Proceedings of the 27th European
Photovoltaic Solar Energy Conference, Frankfurt, Germany, 2012, pp. 1938–1941
47. F. Werner, W. Stals, R. Goertzen, B. Veith, R. Brendel, J. Schmidt, High-rate atomic layer
deposition of Al2O3 for the surface passivation of Si solar cells. Energy Procedia 8, 301–306
(2011)
48. P.G. Vermont, V. Kuznetsov, E.H.A. Granneman, High-throughput solar cell passivation on in-
line Levitrack ALD Al2O3 system—demonstration of process performance, in Proceedings
of the 26th European Photovoltaic Solar Energy Conference, Hamburg, Germany, 2011,
pp. 1644—1647
49. D. Kray, M. Hermle, S.W. Glunz, Theory and experiments on the back side reflectance of
silicon wafer solar cells. Prog. Photovolt: Res. Appl. 16, 1–15 (2008)
50. V. Meemongkolkiat, K. Nakayashiki, D.S. Kim, S. Kim, A. Shaikh, A. Kuebelbeck, W.
Stockum, A. Rohatgi, Investigation of modified screen-printing Al pastes for local back
surface field formation, in Proceedings of the 4th World Conference Photovoltaic Energy
Conversion, Hawaii, USA, 2006, pp. 1338–1341
51. M. Rauer, R. Woehl, K. Rühle, C. Schmiga, M. Hermle, M. Hörteis, D. Brio, Aluminum
alloying in local contact areas on dielectrically passivated rear surfaces of silicon solar cells.
IEEE Electron Device Lett. 32, 916–918 (2011)
52. K.A. Münzer, M. Hein, J. Schöne, M. Hanke, A. Teppe, R.E. Schlosser, J. Maier, A. Yodyun-
yong, S. Krümberg, S. Keller, P. Fath, Technical performance and industrial implementation
in favour of centaurus technology. Energy Procedia 27, 631–637 (2012)
53. K.A. Münzer, J. Schöne, M. Hein, A. Teppe, R.E. Schlosser, M. Hanke, J. Maier, K. Varner,
S. Keller, and P. Fath, Development and implementation of 19% rear passivation and local
contact centaurus technology, in Proceedings of the 26th European Photovoltaic Solar Energy
Conferene, Hamburg, Germany, 2011, pp. 2292–2297
92 T. Dullweber

54. S. Queisser, K. De Keersmaecker, T. Borgers, E. Wefringhaus, D. Nagel, B.-U. Sander, M.

Loehmann, M. Weber, F. Delahaye, Inline single side polishing and junction isolation for rear
side passivated solar cells, in Proceedings of the 24th European Photovoltaic Solar Energy
Conference, Hamburg, Germany, 2009, pp. 1792–1794
55. E. Cornagliotti, M. Ngamo, L. Tous, R. Russell, J. Horzel, D. Hendrickx, B. Douhard, V.
Prajapati, T. Janssens, J. Poortmans, Integration of inline single-side wet emitter etch in
PERC cell manufacturing. Energy Procedia 27, 624–630 (2012)
56. C. Kranz, S. Wyczanowski, S. Dorn, K. Weise, C. Klein, K. Bothe, T. Dullweber, R. Brendel,
Impact of the rear surface roughness on industrial-type PERC solar cells, in Proceedings of the
27th European Photovoltaic Solar Energy Conference, Frankfurt, Germany, 2012, pp. 557–
560
57. C. Comparotto, M. Noebels, L. Popescu, A. Edler, J. Ranzmeyer, T. Klaus, V.D. Mihailetchi,
R. Harney, J. Lossen, T.S. Böscke, D. Schär, H. Nussbaumer, T. Baumann, F.P. Baumgart-
ner, Bifacial n-type solar modules: indoor and outdoor evaluation, in Proceedings of the
29th European Photovoltaic Solar Energy Conference, Amsterdam, The Netherlands, 2014,
pp. 3248–3250
58. G.J.M. Janssen, B.B. Van Aken, A.J. Carr, A.A. Mewe, Outdoor performance of bifacial
modules by measurements and modelling. Energy Procedia 77, 364–373 (2015)
59. H. Mori, Radiation energy transducing device, U.S. Patent 3.278.811, October 1966 (priority
October 1960)
60. I.G. Romijn, B.B. Van Aken, J. Anker, P. Barton, A. Gutjahr, Y. Komatsu, M. Koppes, E.J.
Kossen, M.W.P.E. Lamers, D.S. Saynova, K.C.J.J. Tool, Y. Zhang-Steenwinkel, P.R. Venema,
A.H.G. Vlooswijk, C. Schmitt, H. Kühnlein, N. Bay, M. König, A. Stassen, Industrial cost
effective n-PASHA solar cells with >20% cell efficiency, in Proceedings of the 28th European
Photovoltaic Solar Energy Conference, Paris, France, 2013, pp. 736–740
61. D. Song, J. Xiong, Z. Hu, G. Li, H. Wang, H. An, B. Yu, B. Grenko, K. Borden, K. Sauer, T.
Roessler, J. Cui, H. Wang, J. Bultman, A.H.G. Vlooswijk, P.R. Venema, Progress in n-type
Si solar cell and module technology for high efficiency and low cost, in Proceedings of the
38th IEEE Photovoltaic Specialists Conference, Austin, USA, 2012, pp. 3004–3008
62. V.D. Mihailetchi, J. Jourdan, A. Edler, R. Kopecek, R. Harney, D. Stichtenoth, J. Lossen, T.
Boescke, H.J. Krokoszinski, Screen printed n-type silicon solar cells for industrial applica-
tion, in Proceedings of the 26th European Photovoltaic Solar Energy Conference, Hamburg,
Germany, 2010, pp. 1446–1448
63. M. Taguchi, A. Yano, S. Tohoda, K. Matsuyama, Y. Nakamura, T. Nishiwaki, K. Fujita, E.
Maruyama, 24.7% record efficiency HIT solar cell on thin silicon wafer. IEEE J. Photovolt.
4, 96–99 (2014)
64. B. Strahm, D. Lachenal, D. Bätzner, W. Frammelsberger, B. Legradic, J. Meixenberger, P.
Papet, G. Wahli, E. Vetter, M. Despeisse, A. Faes, A. Lachowicz, C. Allebé, P.J. Alet, M.
Bonnet-Eymard, C. Ballif, Y. Yao, C. Rychen, T. Söderström, J. Heiber, G. Schiltges, S.
Leu, J. Hiller, V. Fakhfouri, The Swiss Inno-HJT project: fully integrated R&D to boost Si-
HJT module performance, in Proceedings of the 29th European Photovoltaic Solar Energy
Conference, Amsterdam, The Netherlands, 2014, pp. 467–471
65. I. Cesar, I.G. Romijn, D.M. Borsa, G. Galbiati, N.J. C.M. van der Borg, E.E. Bende, A.W.
Weeber, Benchmark of open rear side solar cell with improved Al-BSF process at ECN, in
Proceedings 23rd European Photovoltaic Solar Energy Conference, Valencia, Spain, 2008,
pp. 1770–1775
66. T. Dullweber, C. Kranz, R. Peibst, U. Baumann, H. Hannebauer, A. Fülle, S. Steckemetz,
T. Weber, M. Kutzer, M. Müller, G. Fischer, P. Palinginis, D.H. Neuhaus, The PERC+ cell:
a 21%-efficient industrial bifacial PERC solar cell, in Proceedings of the 31st European
Photovoltaic Solar Energy Conference, 2015, pp. 341–350
67. B. Liu, Y. Chen, Y. Yang, J. Dong, H. Shen, Z. Feng, P.J. Verlinden, Rear-colored double glass
module fabricated with rear-colored bifacial p-type PERC cells, in Proceedings of the 31st
European Photovoltaic Solar Energy Conference, 2015, pp. 659–661
5 High-Efficiency Industrial PERC Solar Cells … 93

68. K. Krauß, F. Fertig, J. Greulich, S. Rein, R. Preu, biPERC silicon solar cells enabling bifacial
applications for industrial solar cells with passivated rear sides. Phys. Status Solidi A 213,
68–71 (2016)
69. S.-Y. Chen, H.-C. Liu, H.-Y. Chang, Y.-Y. Chiu, Y.-H. Wang, Y.-H. Lin, S.-H. Yu, W.-J. Lih,
C.-H. Du, Bifacial solar cells fabricated by PERC process for mass production, in Proceedings
of the 32nd European Photovoltaic Solar Energy Conference, 2016, pp. 772–774
70. M. Xiajie, Y. Bin, C. Yongmei, X. Xinxing, F. Jianbin, T. Hongbo, L. Hua, in Proceedings of
the 27th International Photovoltaic Science and Engineering Conference, Otsu, Japan, 2017,
p. 138
71. Communication with JinkoSolar
72. Neo Solar Power data sheet, 2017, https://www.nsp.com/nspsolarcells?lang=en
73. SolarWorld data sheet of Bisun module, 2017, https://www.solarworld.de/en/products/
sunmodule-bisun-protect/
74. Neo Solar Power data sheet of Glory Bifi module, 2017, https://www.nsp.com/
nspsolarmodules?lang=en
75. Trina Solar data sheet of DUOMAX module, 2017, http://www.trinasolar.com/de/product/
duomax4060/duomax-twin-deg5c07ii
76. LONGi Solar data sheet of LR6-60PD module, 2017, http://en.longi-solar.com/Home/
Products/module/id/12_.html
77. https://www.jinkosolar.com/product_detail_311.html?lan=en
78. https://www.canadiansolar.com/upload/c3ee005ae8689897/8c800b00f763e4e9.pdf
79. https://en.longi-solar.com/home/products/Hi_MO4.html
80. SolarWorld press release, 2015, http://www.pv-tech.org/news/intersolar_europe_solarworld_
to_launch_glass_glass_bifacial_modules
81. T. Dullweber, H. Schulte-Huxel, C. Kranz, S. Blankemeyer, U. Baumann, R. Witteck, M.
Köntges, R. Brendel, Bifacial PERC+ solar cells and modules: an overview, in Proceedings
of the 33rd European Photovoltaic Solar Energy Conference, 2017, pp. 649–656
82. A. Faes, M. Despeisse, J. Levrat, J. Champliaud, N. Badel, M. Kiaee, T. Söderström, Y. Yao,
R. Grischke, M. Gragert, J. Ufheil, P. Papet, B. Strahm, B. Cattaneo, J. Cattin, Y. Baumgartner,
A. Hessler-Wyser, C. Ballif, SmartWire solar cell interconnection technology, in Proceedings
of the 29th European Photovoltaic Solar Energy Conference, 2014, pp. 2555–2561
83. H. Neuhaus (SolarWorld), Progress & perspectives in high-volume PERC manufacturing, in
Presented at the PV CellTech Conference, Penang, Malaysia, 2017
84. L.C. Boon (Trina Solar), Bifacial and halfcut technology, in Presented at the PV ModuleTech
Conference, Penang, Malaysia, 2017
85. https://www.power-technology.com/news/longi-bifacial-perc-modules/
86. https://www.pv-tech.org/news/ja-solars-bifacial-perc-modules-used-in-two-pv-projects-in-
south-korea
87. https://www.pv-tech.org/news/trina-solar-supplying-edf-renewables-with-bifacial-modules-
for-french-low-c
88. https://www.pv-tech.org/news/neo-solar-powers-mono-perc-bifacial-modules-selected-for-
100mw-pv-plant-in
89. https://www.pv-tech.org/guest-blog/ultra-low-solar-bid-of-0.01997-kwh-in-the-us-not-
quite-so-sunny
90. http://taiyangnews.info/business/5-42-gw-bifacial-solar-module-capacity-by-2019/
91. https://en.longi-solar.com/Home/Events/press_detail/id/36_World_s_largest_Bifacial_
Solar_Power_Project_Installed_in_China_s_Qinghai_Province.html
92. B. Fischer, Loss analysis of crystalline silicon solar cells using photoconductance and quantum
efficiency measurements. Ph.D. thesis, University of Konstanz, Germany, 2003
93. S. Miyajima, J. Irikawa, A. Yamada, M. Konagai, Hydrogenated aluminum oxide films
deposited by plasma enhanced chemical vapor deposition for passivation of p-type crys-
talline silicon, in Proceedings of the 23rd European Photovoltaic Solar Energy Conference,
Valencia, Spain, 2008, pp. 1029–1032
94 T. Dullweber

94. T.T. Li, A. Cuevas, Effective surface passivation of crystalline silicon by rf sputtered aluminum
oxide. Phys. Status Solidi RRL 3, 160–162 (2009)
95. J. Schmidt, B. Veith, R. Brendel, Effective surface passivation of crystalline silicon using
ultrathin Al2 O3 films and Al2 O3 /SiNx stacks. Phys. Status Solidi RRL 3, 287–289 (2009)
96. G. Dingemans, R. Seguin, P. Engelhart, M.C.M. van de Sanden, W.M.M. Kessels, Silicon
surface passivation by ultrathin Al2O3 films synthesized by thermal and plasma atomic layer
deposition. Phys. Status Solidi RRL 4, 10–12 (2010)
97. L.E. Black, T. Allen, A. Cuevas, K.R. McIntosh, B. Veith, J. Schmidt, Thermal stability of
surface passivation by APCVD Al2 O3 . Sol. En. Mat. Sol. Cells 120, 339–345 (2014)
98. X. Gay, F. Souren, B. Dielissen, M. Bijker, R. Gortzen, D. Pysch, K. Weise, B. Sander, R.
Sastrawan, Post-deposition thermal treatment of ultrafast spatial ALD Al2O3 for the rear side
passivation of p-type PERC solar cells, in Proceedings of the 28th European Photovoltaic
Solar Energy Conference, Paris, France, 2013, pp. 1220–1224
99. J.L. Murray, A.J. McAlister, The Al-Si (aluminum-silicon) system. Bull. Alloy. Phase Diagr.
5, 74–84 (1984)
100. T. Lauermann, A. Zuschlag, S. Scholz, G. Hahn, B. Terheiden, The influence of contact
geometry and sub-contact passivation on the performance of screen-printed Al2O3-passivated
solar cells, in Proceedings of the 26th European Photovoltaic Solar Energy Conference,
Hamburg, Germany, 2011, pp. 1137–1143
101. J. Müller, K. Bothe, S. Gatz, H. Plagwitz, G. Schubert, R. Brendel, Contact formation and
recombination at screen-printed local aluminum-alloyed silicon solar cell base contacts. IEEE
Trans. Electron Devices 58, 3239–3245 (2011)
102. S. Gatz, J. Müller, T. Dullweber, R. Brendel, Analysis and optimization of the bulk and rear
recombination of screen-printed PERC solar cells. Energy Procedia 27, 95–102 (2012)
103. E. Urrejola, K. Peter, H. Plagwitz, G. Schubert, Silicon diffusion in aluminum for rear passi-
vated solar cells. Appl. Phys. Lett. 98, 153508 (2011)
104. J. Müller, K. Bothe, S. Gatz, R. Brendel, Modeling the formation of local highly aluminum-
doped silicon regions by rapid thermal annealing of screen-printed aluminum. Phys. Status
Solidi RRL 6, 111–113 (2012)
105. Y. Gassenbauer, K. Ramspeck, B. Bethmann, K. Dressler, J.D. Moschner, M. Fiedler, E.
Brouwer, R. Drössler, N. Lenck, F. Heyer, M. Feldhaus, A. Seidl, M. Müller, A. Metz, Rear-
surface passivation technology for crystalline silicon solar cells: a versatile process for mass
production. IEEE J. Photovolt. 3, 125–130 (2012)
106. C. Kranz, B. Wolpensinger, R. Brendel, T. Dullweber, Analysis of local aluminum rear contacts
of bifacial PERC+ solar cells. IEEE J. Photovolt. 6(4), 830–836 (2016)
107. C. Kranz, U. Baumann, B. Wolpensinger, F. Lottspeich, M. Müller, P. Palinginis, R. Brendel,
T. Dullweber, Void formation in screen-printed local aluminum contacts modeled by surface
energy minimization. Sol. Energy Mat. Sol. Cells 158, 11–18 (2016)
Chapter 6
High Efficient, Cost-Effective,
and Reliable Silicon Solar Cells
and Modules in Mass Production

J. W. Müller

Abstract Over the last decade, significant improvements in cost reduction as well
as in conversion efficiency increase were achieved in large-scale industrial produc-
tion, making PV cost competitive with other means of electricity generation. In many
regions worldwide, PV achieves the lowest levelized cost of electricity. Several dif-
ferent factors made this tremendous achievement possible-namely economy of scale,
a lean and efficient production process, and high conversion efficiencies. In this work,
some of the key concepts and methods are described based on Hanwha Q CELLS’
experience. The methods and approaches for the fast transfer of cell technologies
from laboratory to production and for accelerated progress in cell efficiency, quality,
and reliability of the cell and module product are described. Over the last decade,
the cell conversion efficiency increased by 0.5%abs per year. Currently, average cell
conversion efficiencies exceeding 20% using boron-doped p-type multicrystalline
(mc-Si) and 22%, using Czochralski-grown silicon (Cz-Si) substrates, are achieved
on a multi-GW scale. Currently (as of Jan 2018) produced Q.ANTUM solar modules
from these cells exhibit output powers of 325 Wp based on 120 half Cz cells and
of 350 Wp based on 72 full multicrystalline cells. Besides just the performance, it
is critical to investigate the reliability of solar modules based on p-type solar cells
with dielectrically passivated rear. In terms of reliability, light-induced degradation
(LID) is examined in detail, with conditions relevant for the activation of both the
boron–oxygen (BO) defect, and “Light and Elevated Temperature Induced Degrada-
tion” (LeTID). While the formation of the BO defect has been considered the most
prominent LID mechanism in boron-doped p-type Cz-Si, recently it was shown that
LeTID is another important issue for passivated emitter and rear cells (PERC) on
both mc-Si and Cz substrates. When not adequately suppressed, LeTID can occur
on p-type PERC with degradation in output power of >6%; critically, it cannot be
suppressed in a straightforward manner by conventional processing steps used to
permanently deactivate the BO defect. In contrast to conventional PERC, Hanwha Q

J. W. Müller (B)
Hanwha Q CELLS GmbH, Sonnenallee 17 – 21, 06766 Bitterfeld-Wolfen OT Thalheim, Germany
e-mail: [email protected]
URL: https://www.q-cells.com

© Springer Nature Switzerland AG 2020 95

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_6
96 J. W. Müller

CELLS’ Q.ANTUM technology is shown to reliably suppress both LID due to BO

defect formation and LeTID, for both p-type mc-Si and Cz-Si substrates.

6.1 Introduction

Hanwha Q CELLS was one of the first companies to start the production of Si
PERC-like cells in 2012. From the first internal PERC cell samples in mid-2009 to
the transfer of the Q.ANTUM [1] process sequence to our production facility in end
2010 and finally to 24/7 cell and module production mode in 2012 took more than
2 years. Reflecting on this and other developments, the timeline and efforts to make
laboratory developments feasible in mass production should not be underestimated.
In the last decade, Hanwha Q CELLS developed concepts for an efficient transfer of
new cell processes from the laboratory to mass production and fast development of
high-quality cell and module products [2]. In this work, these methods are described
and cover the technological progress at Hanwha Q CELLS over the last 10 years of
R&D concerning cell efficiencies, module power, and reliability.
Within the last 4 years, the market share of solar cells with dielectrically passivated
rear side has increased from less than 5% in 2013 [3] to about 15% in 2016 [4] and
is expected to further increase in the upcoming years [4]. The majority of these
cells is represented by the passivated emitter and rear cell (PERC [5]) structure
and is processed on boron-doped p-type silicon substrates. Besides challenges in
process technology, stability issues due to light-induced degradation (LID) have
been a key challenge for high efficiency p-type Cz-Si solar cells. In particular, two
LID mechanisms have hindered a wider adoption of the industrial manufacturing of
PERC-like solar cells on boron-doped p-type silicon substrates: LID due to boron–
oxygen (BO) defect formation [6–8] and “Light and elevated Temperature Induced
Degradation” (LeTID) [9–13]. LID and LeTID are most pronounced for PERC on
Cz- and mc-Si substrates, respectively.
This work reports on the results of mass production of Hanwha Q CELLS’ cell
technology on boron-doped p-type mc-Si and Cz-Si substrates. Over the last decade,
an annual average cell conversion efficiency increase of 0.5%abs. was achieved by
optimizing every process parameter and by the switch from the aluminum back-
surface field (Al-BSF) technology to the Q.ANTUM technology. This efficiency
increase was achieved while keeping a lean process sequence and significantly low-
ering costs, e.g., by reducing the silver paste consumption. Besides device perfor-
mance upon fabrication, LID of Q.ANTUM and PERC is investigated in detail. It
is shown that when not adequately suppressed, LeTID can occur not only in p-type
mc-Si PERC but also in p-type Cz-Si PERC. The suppression of both the BO defect
and LeTID is of utmost importance for both mc-Si and Cz-Si substrates. In contrast to
conventional PERC, Hanwha Q CELLS’ Q.ANTUM technology is shown to reliably
suppress both LID due to BO defect formation and LeTID in modules manufactured
from both p-type mc-Si and Cz-Si substrates [9–13].
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 97

6.2 Approaches for Efficient Cell Development

and Production

Table 6.1 shows an overview of the essential steps for a successful technology transfer
to mass production. HANWHA Q CELLS developed and optimized these approaches
during the last years of cell and module development and transfer activities. Accord-
ing to our findings, most of the work in the development of a new cell technology
suitable for mass production takes place on production equipment. The proof of con-
cept of a new process or process sequence is typically done in the setting of research
institutes. Subsequently, the replacement of standard processes by production pro-
cesses, and the stepwise introduction of new processes into the routine of stable and
well monitored 24/7 manufacturing conditions should be targeted. In this way, it is
also ensured that processes to be introduced are evaluated and optimized directly in
the context of a production process chain. Herewith the advantage of evolutionary cell
development becomes visible. Replacing only one new laboratory process at a time
by a production process enables a systematic benchmarking of both processes and the
investigation of both expected and unexpected interdependencies on cell and module
level. According to our experience, it is thereby essential that research laboratory and
production line facilities are closely connected to each other. Running experiments
on mass production tools furthermore reduce the run-time, thus increases the exper-
imental rate and offers the possibility to improve the statistical relevance by flexibly
increasing the amount of cells. The above-described process, accompanied by the
development of suitable sampling analysis on cell and module level, enables fast and
precise detection of potential targets for optimization. When all new processes are
fully integrated into mass production with a positive business case and the electrical
safety and reliability of the modules are certified for the process sequence, the first
generation of a new cell concept can be produced. It is therefore essential to define
suitable upper and lower process parameter limits (specifications) for each process as
well as action instructions in case of limit violations. This comes along with the effort
to install suitable parameter measurement tools and recipes. Any change in processes
beyond these specifications should be critically examined concerning product and
cost-relevant impacts on cell parameters as well as on production yield, breakage,
throughput, optical criteria, solderability of cells, and module reliability. The high
number of product and cost-relevant criteria in cell and module development leads
to development efforts that can be easily underestimated. The existence of a single
wafer tracking system in combination with a manufacturing execution system (MES)
is another key enabler for fast and efficient development and tight process control.
98 J. W. Müller

Table 6.1 Chronological roadmap of approaches for fast production cell and module development.
Table taken from [2]
Approach Enabled actions and benefits
First sampling in laboratories – First cell technology evaluation and
experience
– Determination of physical parameters →
simulations
R&D pilot production/regular experiment runs – Improved level of process stability
in research environment (proof of concept) – Detection of Ncell improvement potentials
>0.15%abs (single exp. mc-Si)
– Evaluation of new processing tools
– Start of module reliability investigations
– Development of process monitoring and
analysis test structures
Stepwise replacing of laboratory processes by – Benchmarking all processes between R&D
production processes and production tool
Full process sequence in production line – Production relevant throughput and
(experiment level) handling
– Enlarged number of cells per time →
extended module tests possible
– Start calibration of inline measurement
recipes and equipment
– Resolution of Ncell improvement potentials
>0.1%abs (process stability)
Application of tracking systems for evaluation – Detection of process interdependencies and
of the joined data on wafer level deviation root causes
Defining process parameter specifications – Definition of standard operation procedures
(design freeze) to stay within process specifications
(in-spec condition)
Cell and module test productions with – Evaluation of tool uptimes, throughput, and
standard cell process yield in production
– Determination of optical sorting criteria
– More detailed production cost calculation
possible
Module certification (EC/VDE) – External and independent confirmation of
electrical safety and reliability of modules
Start pilot cell and module production – Find process instabilities and tool uptimes
(handover to production) for 24/7 production conditions
In-spec parameter variations (ISV) in raining – Efficiency improvement for this and next
production or test production product generation
– Ncell improvement resolution <0.03%abs
due to large statistics
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 99

6.2.1 Improved Statistical Resolution of Experiments

via Tra.Q

With our unique Tra.Q [14] wafer tracking system, each wafer gets an individual dot
matrix code on the front surface of the wafer at the very beginning of the process
sequence. Using this hard mark, we can accurately track process control data for
each wafer throughout the entire manufacturing sequence from incoming inspection
to final quality inspection at the end of the module line. By combining specific
process data with the corresponding cell tester data via the identification number of
each wafer, we are able to improve the statistical resolution of single experiments
from >0.1%abs (100 cell batch experiment) to <0.03%abs [2]. This can be achieved
by either (1) iteratively switching during the running production between two (or
more) parameter settings of one process tool or (2) loading two parallel running
process tools (with nominally the same or differing parameter settings) thus splitting
the incoming material. To analyze the cell results by the corresponding parameter
setting on cell level, we use Tra.Q data. No additional cell handling is required and
all cells remain in the production chain. These two features enable the use of large
numbers of cells for parameter variations. In order to secure the quality of the cells,
the parameter variation needs to be kept within the process specification, leading to
the term “in-spec variation” (ISV) for this approach. Figure 6.1 illustrates the results
of a 100 cell batch experiment versus an 8000 cell ISV comparing two rear SiNx
deposition recipes (other processes identical). As a measure of statistical significance
of the results for both approaches, the mean of the 95% median confidence interval
(CI) of both recipes is given in the graph. The results state that in the case of the ISV
approach, the statistical resolution for all cell parameters is increased by one order
of magnitude compared to the experiment mode. The improved resolution permits
a very precise optimization of production processes. In our experience, a major part
of continuous cell efficiency improvement is based on such measures of fine-tuning
optimization. Further benefits of the ISV approach are clearly reduced experimental
efforts, costs, and experiment run-times.

6.2.2 Root Cause Finding via Tra.Q

Combining Tra.Q process data together or with cell tester data on the wafer level
furthermore enables the finding of process interactions and hidden root causes for
process deviations or interdependencies. An example of a hidden root cause found
by this approach is the analysis shown in Fig. 6.2 for a fill factor deviation [2]. The
contour plot shows the (interpolated) relative fill factor (FF) distribution of 18,000
Q.ANTUM cells from mass production over the tray position of the AlOx/SiNx
rear passivation deposition tool. By tracking the exact tray position of each cell via
the Tra.Q system, up to 1%rel decreased, FF become visible for three specific tray
positions on the mid-track (blue FF range in plot). Based on this characteristic plot, the
100 J. W. Müller

Fig. 6.1 Box and whisker plot of mc-Si cell results of a rear SiNx recipe comparison via an ISV
(>8000 cells per recipe) contrasted with a corresponding batch experiment (100 cells per recipe).
The 95% confidence interval (CI) for the median of cell parameters is given in the graph. The red
text contains the mean CI value of recipe 1 and 2. Graph taken from [2]

origin of the FF decrease was found in a temporary imprecision in wafer positioning

by one of the two loading stations serving the deposition tool. The positioning was
subsequently corrected and additionally, a tighter control procedure for positioning
was established. This issue would not be easily detectable in cell production lines
without a wafer tracking system. Via Tra.Q, we know the identification number of
each cell affected by such process deviations and if necessary are able to sort them
out before module manufacturing. Thus, our cell and module production stands
out for having the highest quality standards. The benefit of such a single wafer
tracking system becomes more important with the size of a production site. In a 2
GW production line with several hundreds of individual processing tools and more
than a million cells per day, it is almost impossible to ensure tight process control
and efficient failure analysis without such a tracking system. A tracking system also
paves the way for “Industry4.0” and opens the possibility for future self-learning
fabs.

6.3 Cell Efficiency Development of Q.ANTUM Solar Cells

on p-type Silicon Substrates

Figure 6.3 shows sketches of a standard Al-BSF and the Q.ANTUM solar cell struc-
ture. The Q.ANTUM solar cell resembles a PERC-like structure, i.e., it features a
dielectrically passivated rear side with local contacts, front-side junction formation
by phosphorus diffusion and screen-printed metallization.
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 101

Fig. 6.2 Contour plot of the interpolated relative fill factor distribution of 18,000 Q.ANTUM cells
from mass production over the tray position of the rear passivation deposition tool with imprecise
tray loading. Graph taken from [2]

Fig. 6.3 Sketches of a aluminum back-surface field and b Q.ANTUM solar cell structures

Figure 6.4 shows the average cell conversion efficiency as a function of production
date at Hanwha Q CELLS of p-type mc- and Cz-Si Al-BSF and Q.ANTUM solar
cells, respectively. An increase in average efficiency of +0.5%abs/year is observed
over the last decade. This constant average annual learning rate of 0.5%abs is achieved
both on mc- and Cz-Si substrate. In the case of Cz substrates, the learning curve
for Al-BSF cells is disrupted in the beginning of 2013 leading to a significantly
flatter increase compared with the Q.ANTUM solar cell [15, 16]. By switching
to the Q.ANTUM solar cell structure, the increase in conversion efficiency of +
0.5%abs/year could be maintained, and therefore the efficiency curves of both struc-
tures are decoupled with an increasing gap between Q.ANTUM and Al-BSF cells.
A similar but less pronounced effect is shown in case of mc-Si substrates. Within the
next 2 to 3 years, we do not expect a significant slowdown in the yearly efficiency
increase of Q.ANTUM solar cells, leading to average cell conversion efficiencies
exceeding 23% and 21% for Cz-Si and mc-Si, respectively. This steady increase
in cell efficiency is achieved while maintaining a lean manufacturing process and
significantly reducing the manufacturing costs per Wp, e.g., by lowering the silver
paste consumption.
102 J. W. Müller

Fig. 6.4 Average solar cell conversion efficiency of mass production as a function of time at
Hanwha Q CELLS. Over the last decade, an average annual learning rate of 0.5%abs. is reached
both on mc-Si and Cz-Si substrates

Figure 6.5b shows a typical distribution of cell conversion efficiency of more

than 3 million Q.ANTUM solar cells on p-type Cz-Si substrates in mass production
in January 2018. A tight distribution with maximum frequencies between 21.9 and
22.1% can be seen, resulting in an average efficiency of 22.0%. Figure 6.5a shows
a similar distribution of cell conversion efficiency of Q.ANTUM solar cells on cost-
effective p-type mc-Si substrates. The distribution is slightly broader than in case of
Cz-Si with maximum frequencies between 19.8% and 20.2% resulting in an average
efficiency of 20.0%. These cell efficiencies lead to manufactured Q.ANTUM solar
modules based on mc-Si substrates exhibiting power classes of 350 Wp, utilizing 72 6
inch full-square 4-busbar solar cells and standard module interconnection technology.
In other words, no power-enhancing measures such as the use of half-cells, multi-wire
approaches (e.g., soldering of multiple wires [17, 18] or connection of embedded
wires during lamination [19, 20]), or light-capturing ribbons have been applied for
the modules with mc-Si. For Cz-Si substrates, a more advanced module technology
based on half-cells and six wire interconnection is applied. The half-cell technology
halves the electrical current in the copper interconnects, thus the power loss due
to series resistance in the interconnects is reduced by a factor of four, leading to a
relative increase in module power by 3%rel. Moreover, using round wires instead
of rectangular ribbons reduces the effective shaded area by roughly 60% of the wire
diameter due to total internal refection at the glass–air interface of the module. This
leads to an additional relative module power increase of 2%rel. Using cells with six
busbars instead of four busbars reduces the power loss due to shorter finger lengths
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 103

Fig. 6.5 Cell conversion

efficiency distribution of
several million Q.ANTUM
solar cells on p-type silicon
substrates in mass
production in January 2018.
a Q.ANTUM cells on mc-Si
with average efficiency of
20.0%, b Q.ANTUM cells
on Cz-Si with average
efficiency of 22.0%

contributing an additional 0.5%rel module power increase. All three measures sum
up to a gain of 5.5%rel compared to a module technology based on full-cells and
four rectangular ribbon interconnection. Module powers of 320 Wp and 325 Wp
are reached based on 120 half-cut p-type 6 inch pseudo squared Cz-Si substrates
and six wire interconnection. These output powers are at the same level or even
higher than those for modules with industrial PERT-like (passivated emitter, rear
totally diffused [21]) solar cells on n-type silicon substrates. Furthermore, the gap
to the power classes of modules with the highest efficiency n-type Cz-Si solar cell
structures such as heterojunction [22–25] and rear-contact solar cells [26–29] is
104 J. W. Müller

rapidly narrowing. Importantly, compared to the comparably complex and thus higher
cost manufacturing process of those n-type cells, the Q.ANTUM technology has an
extremely competitive manufacturing cost.

6.4 Light-Induced Degradation of PERC

Besides maximizing conversion efficiency and output power, the long-term stability
of solar cells and modules is of crucial importance. Two key mechanisms that are
potential hazards for the reliability of solar cells and modules based on boron-doped
p-type PERC-like technology are the boron–oxygen (BO) defect and “Light and ele-
vated Temperature Induced Degradation” (LeTID). The impact of these mechanisms
on PERC cell and module performance is discussed in this section and it is shown
that Q.ANTUM technology suppresses both LID and LeTID in mc- and Cz-Si.
To separate BO defect formation and the formation of LeTID, different temper-
atures, excess charge carrier densities, and treatment times are applied. The applied
nomenclature is as follows: CID describes current-induced degradation, which means
that the same excess charge carrier densities are established by current injection in
the dark as with LID. Two levels of excess charge carrier density are considered:
OC and MPP mode, which correspond to the excess charge carrier concentrations
at open-circuit and at maximum power point at an illumination intensity of Eill =
1 kWm−2 .

6.4.1 Boron–Oxygen Defect

To investigate the impact of BO defect formation on device performance, Q.ANTUM

solar cells and PERC without treatment to permanently deactivate the BO defect have
been processed on boron-doped p-type Cz-Si substrates from different industrial
suppliers. These samples are then subjected to light soaking with an illumination
intensity of ~1 sun-equivalent for 24 h at a temperature of 25 °C. For typical boron
and oxygen concentrations, the formation of the BO defect has been shown to be
almost independent of illumination intensity in the range of 0.01 kWm−2 < Eill <
1 kWm−2 [6] and to typically be close to saturation after 24 h of light soaking at 25 °C
[26, 27]. Figure 6.6 shows values of relative degradation in maximum output power
upon light soaking. While a significant degradation of 4 to 5%rel is observed for
PERC without BO defect stabilization, LID can be suppressed to less than 0.5%rel
by applying Q.ANTUM technology [12].
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 105

Fig. 6.6 Light-induced 25°C, LID OC mode, 24 h

degradation of maximum
output power of Q.ANTUM 1 Q.ANTUM
solar cells and PERC without PERC w/o BO stabilisation

degradation Δ PMPP (%rel)

Relative light-induced
BO stabilization on Cz-Si 0
substrates from different
suppliers (~1 sun- -1
equivalent/25 °C/24 h).
Graph taken from [12]
-2

-3

-4

-5
A B C
Wafer supplier

6.4.2 Light and Elevated Temperature Induced Degradation

(LeTID)

In contrast to BO defect formation, LeTID has so far mainly been associated with a
potential issue for mc-Si PERC [9–11]. In previous studies by Hanwha Q CELLS [11,
12], solar modules have been fabricated from mc- and Cz-Si PERC with intentionally
manipulated LeTID sensitivity (“high LeTID”/“medium LeTID”) along with LeTID-
suppressing mc- and Cz-Si Q.ANTUM modules, and these have been subjected to
different treatments. Figure 6.7 (right) shows the results for mc-Si: relative module
power degradation as a function of treatment time. Severe degradation of more than
10% is observed for “high LeTID” PERC modules. While the “medium LeTID”
treatment reduces degradation to about 4%, Q.ANTUM technology is shown to
thoroughly suppress LeTID. Note: the applied test conditions are CID in MPP mode
at elevated temperature. Compared with OC mode, the lower excess charge carrier
densities in MPP mode lead to slower formation of LeTID but to a higher extent since
regeneration sets in at a later point of time [11]. Hence, to test worst-case conditions
at a reasonable speed, Hanwha Q CELLS suggests the testing conditions of CID in
MPP mode at 75 °C to evaluate the long-term stability of silicon solar modules in
terms of LeTID susceptibility.
As recently published by Hanwha Q CELLS [12], LeTID also occurs in Cz-Si
cells. Three sorts of solar modules have been manufactured: (i) PERC modules with-
out permanent deactivation of the BO defect, (ii) PERC modules with a commercially
available processing step to permanently deactivate the BO defect, and (iii) LID- and
LeTID-suppressing Q.ANTUM modules. To separate the formation of the BO defect
and LeTID, two different CID treatments have been applied: (i) OC mode at TLID =
25 °C, which corresponds to the conditions applied in Sect. 6.4.1 to test for BO defect
formation and (ii) MPP mode at TLID = 75 °C, which corresponds to the conditions
suggested by Hanwha Q CELLS to test LeTID.
106 J. W. Müller

Fig. 6.7 (left) Relative degradation in maximum output power of modules with Cz-Si solar cells
as a function of treatment time at temperatures of 25 and 75 °C in OC and MPP mode, respectively.
Resulting degradation values for PERC modules with and without BO stabilization are displayed,
along with LID- and LeTID-suppressing Q.ANTUM modules. (right) Relative power degradation
of modules with mc-Si solar cells as a function of treatment time at temperatures of 75 and 85 °C in
MPP mode. Resulting degradation values for PERC modules with maximum and medium LeTID
sensitivity are displayed, along with LeTID-suppressing Q.ANTUM modules. It can be seen that
Hanwha Q CELLS Q.ANTUM technology suppresses LID and LeTID for both mc- and Cz-Si.
Graph taken from [12]

Figure 6.7 (left) shows values of relative module power degradation as a function
of treatment time. In OC mode at TLID = 25 °C, the Cz-Si PERC modules without
BO stabilization show a degradation of around 6%, which is in the same order as
for the PERC discussed in Sect. 6.4.1. By applying a commercially available pro-
cessing step to permanently deactivate the BO defect (PERC w/BO stabilization),
this degradation can be reduced to about 1%. However, when subjecting the PERC
modules w/BO stabilization to CID in MPP mode at TLID = 75 °C, a severe degra-
dation of about 4% is observed. As expected from the discussion in Sect. 6.4.1, LID
due to BO defect formation is close to saturation after 24 h of light soaking at 25 °C
while degradation due to LeTID at 75 °C progresses in the considered time scale.
When comparing Fig. 6.8 (left) and (right), it can be seen that the extent of LeTID of
the Cz-Si PERC modules w/BO stabilization and its formation rate approximately
correspond to the mc-Si PERC modules w/medium LeTID [12]. In conclusion, a
strong LeTID signal is observed also on p-type Cz-Si PERC modules, despite the
suppression of excessive LID due to BO defect formation by commercially avail-
able BO stabilization processing. As illustrated in Fig. 6.7, Hanwha Q CELLS’
Q.ANTUM technology not only suppresses LID and LeTID on mc-Si but also on
Cz-Si substrates [12].
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 107

Fig. 6.8 Relative module Lab Degradation Time tLab [h]

power loss over time due to
LeTID in the laboratory 0 200 400 600 800 1000 1200

Rel. Module Power Loss due to LeTID [%]

(upper x-axis) and from field
operation in Thalheim (DE) 2.0 LeTID-sens. PERC Lab
and Nicosia (CYP) (lower Q.ANTUM Lab
x-axis). Graph taken from
[13] 0.0

-2.0

-4.0 LeTID-sens. PERC Field DE

LeTID-sens. PERC Field CYP
Q.ANTUM Field CYP
-6.0

-8.0

0 1 2 3 4
Field Operation Time tField [a]

6.4.3 LeTID Under Real Field Condition

The impact of LeTID was investigated not only in the laboratory but also under real
outdoor field conditions [13]. Standard industrial mc-Si substrates were neighbor
sorted into two batches, for a direct comparison of PERC vs. Q.ANTUM technology
that excludes material variability. LeTID-sensitive PERC modules were manufac-
tured with one of the batches and Q.ANTUM modules with the other. The module
degradation experiments in the laboratory were performed at a temperature of 75 °C
in a climate chamber with CID (MPP) and 1 sun illumination.
LeTID-sensitive PERC and LeTID-suppressing Q.ANTUM modules were
installed on outdoor test fields in Thalheim, Germany (DE) and Nicosia, Cyprus
(CYP). The test fields represent temperate and Mediterranean climates, respectively.
The irradiance (Emod ) in the module plane and module temperature (Tmod ) as well as
the energy yield of each system was logged regularly. Once every 3 months, the mod-
ules installed in Germany and Cyprus were removed from the system racks and tested
by means of current–voltage measurements under standard test conditions (STC) in
the laboratory. After each testing cycle, the modules were reinstalled outdoors.
For Thalheim, both the temperature and irradiances are significantly lower than
in Nicosia. While in Thalheim, the modules operate for ~30% of all operation time
in comparably lowlight conditions with Emod < 400 W/m2 , in Nicosia, the highest
irradiance was measured to be in the range of 900–1000 W/m2 . These relatively
high irradiances around 1000 W/m2 in Cyprus resulted in comparably high module
temperatures. The weighted average Tmod of 35–40 °C in Nicosia is significantly
higher compared to Thalheim with 5–10 °C. In Nicosia, elevated temperatures with
108 J. W. Müller

Tmod > 50 °C occurred for ~25% of annual field operation time. Hence, these climatic
conditions in Cyprus are well-suited for the characterization of LeTID in comparison
to the temperate climate in Germany with ~70% of annual field operation time having
Tmod < 25 °C [13].
The irradiance in the module plane and the module temperatures at two different
test fields in Cyprus and Germany were measured. Together with a model for the
irradiance- and temperature-dependence of LeTID kinetics, a site-specific time con-
stant corresponding to the proposed accelerated laboratory test performed at 75 °C
is deduced; e.g., 290 h laboratory test time corresponds to 1 year of field installation
time in Cyprus. Figure 6.8 shows as filled symbols the relative module power loss
due to LeTID for LeTID-sensitive PERC and LeTID-suppressing Q.ANTUM mod-
ules vs. field operation time (lower x-axis) in Cyprus and Germany. Furthermore,
the corresponding values for degradation tests in the laboratory are shown as open
squares. The time-dependent relative module power loss under laboratory testing at
75 °C with CID in MPP mode corresponds to the upper x-axis (lab degradation time
tLab ). The field performance can be modeled well by laboratory tests, see Fig. 6.10.
LeTID-sensitive PERC modules showed a significant loss in module power of more
than 7% due to LeTID, both in the field and under accelerated aging in the labora-
tory. Equivalent modules in Germany showed a degradation of ~2.5% for the same
time period. Due to lower irradiance values and module temperature in Germany,
the LeTID rate is less pronounced than in Cyprus. Figure 6.9 shows the relative loss
of specific yield on the left and LCOE on the right y-axis compared with a refer-
ence system during 3 years of field installation in Cyprus. The loss in system output
power due to LeTID is shown in Fig. 6.8 affects one-to-one the energy yield and the
LCOE. It can be seen that the LeTID-suppressing Q.ANTUM modules show a higher

Fig. 6.9 Loss in specific

n.
n.

n.
n.
l.
l.

l.
l.

l.

yield and LCOE relative to

Ja
Ju
Ju
Ja

Ju
Ja
Ja
Ju

Ju
Rel. Loss to Reference in Specific Yield [%]

reference due to LeTID

during field operation in 2.0 2.0
Rel. Loss to Reference in LCOE [%]

Nicosia (CYP). Graph taken

from [13]
0.0 0.0
soiling of system
LeTID-sens. PERC Field CYP
-2.0 Q.ANTUM Field CYP -2.0

-4.0 -4.0

-6.0 -6.0

-8.0 -8.0

0 1 2 3 4
Field Operation Time tField [a]
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 109

Fig. 6.10 Relative module Lab Degradation Time tLab [h]

power degradation over time
0 580 1160 1740 2320 2900 3480 4060
due to LeTID. The lower

Rel. Module Power Loss due to LeTID [%]

x-axis corresponds to field 2.0
operation time in Nicosia
(CYP) and the upper x-axis
0.0
to CID laboratory time in
MPP mode at 75 °C in a
climate chamber. Graph -2.0
taken from [13] LeTID-sens. PERC Field CYP
-4.0 LeTID-sens. PERC Lab
Q.ANTUM Field CYP
Q.ANTUM Lab
-6.0

-8.0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Field Operation Time tField [a]

specific yield and thus higher revenue than the reference system. The lower specific
yield after approximately 2 years field installation time was due to soiling of the
Q.ANTUM system in Cyprus from nearby construction, which was removed at the
end of October. As shown in Figs. 6.8, 6.7, 6.8, and 6.9, by applying the optimized
defect engineering of the Q.ANTUM process, LeTID could be fully suppressed in the
field [13]. Figure 6.10 shows that Q.ANTUM modules do not degrade appreciably
even after extended laboratory testing of several thousand hours. With the model that
was developed and the site-specific time constant, the laboratory degradation time of
4300 h is predicted to correspond to 15 years of field installation in a Mediterranean
climate [13].

6.5 Summary and Conclusion

This work addresses three crucial points for cost-effective PV cell and module pro-
duction: first, lean and efficient production and development methods; second, high
module power based on high efficiency cell and module technology; third, high reli-
ability and energy yield by effective suppression of any light-induced degradation
(LID).
A single wafer tracking system in combination with a manufacturing execution
system was proven to be a powerful tool to detect process deviation and find their
root causes. By using in-spec variation in combination with large amount of cells,
statistically significant differences in conversion efficiency of less than 0.03%abs.
are resolvable. This allows a very efficient fine-tuning of the whole process sequence
and thus improves the average conversion efficiency of a production line. Using a
systematic development process to mature and upscale lab results was found to speed
110 J. W. Müller

up implementation of innovations in a production environment. Thus, over the last

decade, the average cell conversion efficiency of industrial solar cells fabricated on
boron-doped p-type Czochralski-grown silicon (Cz-Si) and multicrystalline silicon
(mc-Si) substrates increased by 0.5%abs. per year. Apart from process optimizations,
one major technology platform change was implemented—going from the Al-BSF
technology to Hanwha Q CELLS’ Q.ANTUM technology, a PERC-based technol-
ogy. Currently (Jan 2018), average solar cell conversion efficiencies of 22% and 20%
are achieved in mass production on p-type Cz- and mc-Si substrates, respectively. We
still see significant efficiency headroom for the Q.ANTUM technology and expect
to approach 24% and 22% average conversion efficiency within the next years while
maintaining a lean and cost-effective manufacturing process.
In combination with advanced interconnection and module technology, namely
half-cell and six wire technology, Q.ANTUM module output powers of >325 Wp are
achieved using 120 Cz-Si half-cells. The reliability of the wire connection technology
is now well-established, with a soldering technique proven by yearlong studies.
Q.ANTUM technology on cost-effective mc-Si substrates leads to solar modules
exhibiting output powers of >350 Wp with 72 full 4-busbar cells and standard module
technology.
One key concern for the long-term stability of Cz-Si and PERC cells in the field
has been LID due to boron–oxygen (BO) defect formation and “Light and elevated
Temperature Induced Degradation” (LeTID). By applying a commercially available
processing step for the permanent deactivation of the BO defect, LID due to BO
defect formation can be reduced from up to 6% to around 1% in conventional Cz-
PERC. However, despite this BO defect stabilization, LeTID can be a severe issue for
the long-term stability of PERC if not adequately suppressed, especially on mc-Si
but also on Cz-Si substrates. The extent of LeTID in mc-Si and Cz-Si PERC can
be manipulated, showing power degradation values of up to 11% on top of LID for
different scenarios, despite “standard” permanent BO defect deactivation. Hence, the
suppression of the BO defect is not sufficient, but instead, additional measures are
necessary to detect and suppress LeTID to ensure long-term stability in the field.
The results presented show performance loss measurements due to LeTID not
only in the lab but also in real field system installations. The observed degradation
behavior in the field can be adequately predicted by using a site-specific time constant,
which is easily adaptable to any climate zone.
In contrast to conventional PERC, Hanwha Q CELLS’ Q.ANTUM technology
is shown to reliably suppress both LID due to BO defect formation and LeTID in
modules manufactured from both p-type mc-Si and Cz-Si substrates. Outdoor field
operation data and long-term laboratory testing in climate chambers, corresponding
to a field installation time of 15 years in a Mediterranean climate, validated the
effective LeTID suppression of Q.ANTUM. This enables the superior long-term
stability and field performance of Q.ANTUM compared with conventional PERC
modules.

Acknowledgements The author acknowledges the staff of the R&D department and production at
Hanwha Q CELLS for their contribution to this work.
6 High Efficient, Cost-Effective, and Reliable Silicon Solar Cells … 111

References

1. P. Engelhart et al., Q.ANTUM—Q-Cells Next Generation High-Power Silicon Cell & Module
Concept, in Proceedings of the 26th European Photovoltaic Solar Energy Conference and
Exhibition, Hamburg, Germany (2011), pp. 821–826
2. S. Engelhart et al., 3 years of high quality mc-Si Q-ANTUM production experience—
approaches for efficient cell and module development, in Proceedings of the 31st European
Photovoltaic Solar Energy Conference and Exhibition, Hamburg, Germany (2015), pp. 273–
278
3. International Technology Roadmap for Photovoltaic (ITRPV): 2013 Results, Fifth Edition,
Revision 1, March 2014. http://www.itrpv.net/
4. International Technology Roadmap for Photovoltaic (ITRPV): 2016 Results, Eighth Edition,
March 2017. http://www.itrpv.net/
5. A.W. Blakers, A. Wang, A.M. Milne, J. Zhao, M.A. Green, 22.8% efficient silicon solar cell.
Appl. Phys. Lett. 55(13), 1363 (1989)
6. S. Rein, T. Rehrl, W. Warta, S.W. Glunz, G. Willeke, Electrical and thermal properties of
the metastable defect in boron-doped Czochralski silicon (Cz-Si), in Proceedings of the 17th
European Photovoltaic Solar Energy Conference, Munich, Germany (2001), pp. 1555–1560
7. J. Schmidt, K. Bothe, Structure and transformation of the metastable boron- and oxygen-related
defect center in crystalline silicon. Phys. Rev. B 69(2), 24107 (2004)
8. V.V. Voronkov, R. Falster, Latent complexes of interstitial boron and oxygen dimers as a reason
for degradation of silicon-based solar cells. J. Appl. Phys. 107(5), 53509 (2010)
9. K. Ramspeck et al., Light induced degradation of rear passivated mc-Si solar cells, in Proceed-
ings of the 27th European Photovoltaic Solar Conference and Exhibition, Frankfurt, Germany
(2012), pp. 861–865
10. F. Fertig, K. Krauß, S. Rein, Light-induced degradation of PECVD aluminium oxide passivated
silicon solar cells. Phys. Stat. Sol. (RRL) 9(1), 41–46 (2015)
11. F. Kersten et al., Degradation of multicrystalline silicon solar cells and modules after illumi-
nation at elevated temperature. Sol. Energy Mater. Sol. Cells 142, 83–86 (2015)
12. Fertig et al., Mass production of p-type Cz silicon solar cells approaching average stable
conversion efficiencies of 22%. Energy Procedia 124, 338–345 (2017)
13. F. Kersten et al., System performance loss due to LeTID. Energy Procedia 124, 540–546 (2017)
14. S. Wanka et al., Tra.Q—laser marking for single wafer identification—production experience
from 100 million wafers, in Proceedings of the 26th European Photovoltaic Solar Energy
Conference and Exhibition, Hamburg, Germany (2011), pp. 1104–1107
15. A. Mohr et al., 20%-efficient rear side passivated solar cells in pilot series designed for con-
ventional module assembly, in Proceedings of the 26th European Photovoltaic Solar Energy
Conference and Exhibition, Hamburg, Germany (2011), pp. 2150–2153
16. M. Schaper et al., The Q.ANTUM technology platform applied to p-Cz to maintain constant
efficiency increase per year, in Proceedings of the 32nd European Photovoltaic Solar Energy
Conference and Exhibition, Munich, Germany (2016)
17. J. Walter, M. Tranitz, M. Volk, C. Ebert, U. Eitner, Multi-wire interconnection of busbar-free
solar cells. Energy Procedia 55, 380–388 (2014)
18. S. Braun, G. Hahn, R. Nissler, C. Pönisch, D. Habermann, Multi-busbar solar cells and modules:
high efficiencies and low silver consumption. Energy Procedia 38, 334–339 (2013)
19. T. Söderström, P. Paper, J. Ufheil, Smart wire connection technology, in Proceedings of the
28th European Photovoltaics Solar Energy Conference and Exhibition, Paris, France (2013),
pp. 495–499
20. A. Schneider, L. Rubin, G. Rubin, Solar cell efficiency improvement by new metallization
techniques—the Day4 electrode concept, in Proceedings of the 4th World Conference on Pho-
tovoltaic Energy Conversion, Waikoloa, HI, USA (2006), pp. 1095–1098
21. J. Zhao et al., High efficiency PERT cells on n-type silicon substrates, in Conference Record of
the 29th IEEE Photovoltaic Specialists Conference, New Orleans, LA, USA (2002), pp. 218–
221
112 J. W. Müller

22. K. Yoshikawa et al., Silicon heterojunction solar cell with interdigitated back contacts for a
photoconversion efficiency over 26%. Nat. Energy 2, 17032 (2017)
23. K. Masuko et al., Achievement of more than 25% conversion efficiency with crystalline silicon
heterojunction solar cell. IEEE J. Photov. 4(6), 1433–1435 (2014)
24. M. Tanaka et al., Development of new a-Si/c-Si heterojunction solar cells: ACJ-HIT (Artificially
Constructed Junction-Heterojunction with Intrinsic Thin-Layer). Jpn. J. Appl. Phys. 31, Part
1, No. 11, 3518–3522 (1992)
25. S. de Wolf, A. Descoeudres, Z.C. Holman, C. Ballif, High-efficiency silicon heterojunction
solar cells: a review. Green 2(1), 7–24 (2012)
26. E. van Kerschaver, G. Beaucarne, Back-contact solar cells: a review. Prog. Photovolt Res. Appl.
14(2), 107–123 (2006)
27. D.D. Smith, Silicon solar cells with total area efficiency over 25%, in Proceedings of the 42nd
IEEE Photovoltaics Specialists Conference, Portland, Oregon, USA, 2016
28. M.D. Lammert, R.J. Schwartz, The interdigitated back contact solar cell: a silicon solar cell
for use in concentrated sunlight. IEEE Trans. Electron Devices 24(4), 337–342 (1977)
29. R.M. Swanson et al., Point-contact silicon solar cells. IEEE Trans. Electron Devices 31(5),
661–664 (1984)
Chapter 7
Silicon Heterojunction Technology:
A Key to High Efficiency Solar Cells
at Low Cost

A. S. Abramov, D. A. Andronikov, S. N. Abolmasov and E. I. Terukov

Abstract Solar cells made of silicon wafers have dominated the terrestrial photo-
voltaic (PV) market since the beginning of industrial commercialization of PV.

7.1 High-Efficiency Silicon PV Technologies Overview

Solar cells made of silicon wafers have dominated the terrestrial photovoltaic (PV)
market since the beginning of industrial commercialization of PV. Today, about 90%
of solar modules are made of silicon solar cells. There are two main reasons for
this. First, silicon is one of the most common elements on the Earth, which is abun-
dant and eco-friendly. Second, silicon is the main semiconductor material in the
microelectronic industry. The presence of a vast market of the equipment for micro-
electronics, the high level of development of individual technological steps and the
possibility of their transfer greatly facilitated and speeded up the growth of silicon
PV. Since the very beginning, Si solar cells have demonstrated high conversion effi-
ciencies among other technologies used in mass production of solar modules for

Dedicated to the 100 Anniversary of Ioffe Institute.

A. S. Abramov (B) · D. A. Andronikov · E. I. Terukov

Ioffe Institute, 26 Polytechnicheskaya Str., St. Petersburg 194021, Russia
e-mail: [email protected]
D. A. Andronikov
e-mail: [email protected]
E. I. Terukov
e-mail: [email protected]
A. S. Abramov · D. A. Andronikov · S. N. Abolmasov · E. I. Terukov
R&D Center of Thin Film Technologies in Energetics, 28 Polytechnicheskaya Str., St. Petersburg
194064, Russia
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 113

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_7
114 A. S. Abramov et al.

terrestrial PV market [1]. Multijunction solar cells fabricated from III to V semicon-
ductor compounds have demonstrated significantly higher efficiencies (~38% for an
InGaP/GaAs/InGaAs cell at one sun [2]) and are currently the mainstream product
for spacecraft solar panels; however, these are still on their way to full-scale terrestrial
applications due to elevated production costs.
High efficiency silicon-based PV technologies require advanced approaches in
the solar cell design to address such issues as effective light absorption and collec-
tion of photo-generated carriers, suppression of carrier recombination losses, reduc-
tion of shadow losses caused by front electrodes and cell assembly in the module.
The specific feature of high efficiency silicon solar cells is the implementation of
surface passivation layers that allow decreasing significantly recombination losses
of photo-generated carriers at the interfaces. Effective surface passivation is usu-
ally achieved by using dielectric layers such as silicon nitride (SiNx ), silicon oxide
(SiO2 ), aluminum oxide (Al2 O3 ) (see Chap. 4 of this book for more details) or intrin-
sic hydrogenated amorphous silicon or their combinations, depending on the type of
technology used for manufacturing of high efficiency silicon solar cells.
Among commercially available silicon solar cells, there are three main types of
cells (see Fig. 7.1), which currently have conversion efficiency of above 22%. They
include: (1) interdigitated back contact (IBC) cells, (2) passivated emitter and rear
contact cells (PERC) and (3) silicon heterojunction (SHJ) cells. Up to now, the IBC
cells, introduced into terrestrial PV market by SunPower Corporation around 2004
[3], have demonstrated the highest cell (>24%) and module (>22%) efficiency in

Fig. 7.1 Schematic cross-sectional view of high efficiency c-Si solar cells: a IBC cell (FSF—front
surface field), b PERC cell, c PERT cell and d SHJ cell (ITO—indium tin oxide, i-a-Si:H—intrinsic
hydrogenated amorphous silicon) ; bifacial, rear emitter design is shown
7 Silicon Heterojunction Technology … 115

mass production [4] among these three designs. The record efficiency of SunPower
R&D IBC cells is 25.2% [4]. The key design features that contribute to very high
efficiency are: (1) a gridless front surface, which permits to avoid shadow losses and
optimize front surface passivation, (2) localized back contacts with reduced contact
recombination losses due to SiO2 passivation, and (3) backside metallization that
provides internal rear surface reflection and very low series resistance. An additional
benefit of IBC cell design is that the cell interconnection is made at one (rear)
side allowing to reduce the distance between cells and hence to increase the module
efficiency. High efficiency and aesthetic appearance of IBC modules make them very
attractive for mobile PV systems that can be used as an energy source in cars, trucks,
planes, boats and so on [5]. However, their large-scale application (e.g., in centralized
power stations or as a part of building integrated PV) is limited by comparatively
complex manufacturing process resulting in high production costs.
On the other hand, the PERC cell has an advantage of being more compatible
with existing production lines used for the standard technology—aluminum back
surface field (Al-BSF) on p-type silicon, which represents a very large proportion
(>90%) of world PV market. For this reason, many cell producers consider the
PERC concept as a natural development of standard technology. The PERC cell
concept was proposed by researchers from University of New South Wales (UNSW)
in the late 1980s [6]. In this device, the back surface of the cell is oxidized and then
p+ regions are formed in localized areas of the device by opening windows in the
oxide layers and then diffusing dopant impurities into these regions, as schematically
shown in Fig. 7.1b. The silicon dioxide passivates the back surface of the cell and
is much more effective at reflecting light back into the cell than the Al-BSF contact
and p+ regions alone. Such cells made of boron-doped, p-type c-Si are already in
production on a large scale in Asia, with an efficiency of about 22% [7], not taking
into account light-induced degradation (LID) effect due to boron–oxygen complex
formation [8]. To avoid the LID effect, there have been suggestions that the industry
needs to move as a whole to phosphorus-doped, n-type c-Si wafers, which is also
a feasible substrate for high-performance PERC cells. Since phosphorus does not
form such detrimental defects with oxygen, this allows higher lifetimes in standard
Czochralski (CZ) grown ingots. For this reason, the passivated emitter rear totally
diffuse (PERT) cell—a close cousin of PERC cell—has attracted much attention
recently [9, 10]. This technology, however, requires additional steps to execute it
(e.g., to grow SiNx and Al2 O3 layers, as can be seen in Fig. 7.1c), which increase the
production costs compared to the PERC technology. A specific feature of PERT cell
is its high (>90%) bifaciality. This means that such cells assembled in a module can
benefit from an albedo of surrounding area and the gain in the energy yield can be
rather impressive. Although the procedure for measuring the gain of bifacial modules
having transparent protective cover at the rear side is not certified yet, recent studies
have demonstrated that the gain could be in the range from 10 to 20%, depending
on the albedo and design of solar station. For example, as shown in [11], an 18.5%
cell mounted over a surface with intermediate albedo can yield a bifacial gain of
20%, corresponding to an equivalent cell efficiency of 22%. Hence, it is reasonable
116 A. S. Abramov et al.

to assume that 21–22% bifacial PERT cells could behave in outdoor conditions like
24–25% IBC cells, which have rather moderate bifaciality.
Another solar cell concept with inherent bifaciality is heterojunction with intrinsic
thin layer (HIT) cell invented by Sanyo in the early 1990s [12]. In this device (as well
as in other SHJ cells), excellent surface passivation is achieved by deposition of very
thin intrinsic hydrogenated amorphous silicon (a-Si:H) layers on both sides of n-type
wafer, as schematically shown in Fig. 7.1d. The a-Si:H passivation layers grown at
low temperature allow to reach much higher open-circuit voltages (up to 750 mV
[13]) and smaller temperature coefficients (below—0.3%/°C [14, 15]) in SHJ cells
in comparison to high temperature (thermal diffusion) approaches employed in IBC,
PERC or PERT cells. The simple structure of SHJ cells makes them also suitable
for the IBC cell design. As recently demonstrated by Kaneka, 6 SHJ cells having
the IBC design can even outperform the standard IBC cells and reach an efficiency
of 26.3% [16]. All these benefits make SHJ solar cells very attractive for the PV
industry. Today many cell producers already use SHJ technology for large-scale
manufacturing of solar cells with an efficiency of 22–23%.
In this chapter, the main features of SHJ technology are considered. As IBC-SHJ
technology is still a subject of R&D and not industrialized yet, we focus on two-side
contacted bifacial technology, depicted in Fig. 7.1d, which is the mainstream in the
mass production of SHJ solar cells.

7.2 Key Features of SHJ Solar Cell Technology

7.2.1 SHJ Cell Technology at a Glance

The concept of a diode using a heterojunction formed on the contact between doped
amorphous silicon layers and c-Si wafer was initially proposed in 1974 [16]. How-
ever, the concept was commercialized only in 1990s with implementation of very
thin intrinsic a-Si:H layers between doped layers and c-Si wafer [12], as mentioned
above. This ultra-thin a-Si:H layers with typical thickness of few nanometers have
crucial effect on the performance of SHJ cells. Their aim is to suppress surface
recombination by chemical passivation of dangling bonds on c-Si wafer surface with
forming of Si–Si and Si–H bonds and to prevent defect generation by dopant atoms
during deposition of doped layers [17].
A cross-sectional view of the conventional SHJ solar cell and main steps of its
manufacturing process sequence are given in Fig. 7.2. Typically, the SHJ cell is
composed of n-type c-Si wafer coated on both sides with thin intrinsic and doped
a-Si:H layers. These layers then fully covered with indium tin oxide (ITO) films
followed by screen-printing of contact metal grids for current collection using of
low-temperature Ag conductive paste.
7 Silicon Heterojunction Technology … 117

Fig. 7.2 a Schematic cross-sectional view of conventional SHJ solar cell and b main steps of its
fabrication process sequence

7.2.2 Wafers for SHJ Cells

Like for all high performance c-Si solar cells, wafer quality is a key to high effi-
ciency SHJ cells. Although record efficiency values reported in the literature have
been obtained using high-purity float zone (FZ) c-Si wafers, the development of
Czochralski process and continuous improvement of polysilicon quality allowed to
reduce impurity concentrations in CZ wafers with reasonable production costs. As
a result, the open-circuit voltage (Voc ) of SHJ cells has recently reached values as
high as 750 mV [13].
Up to now, only monocrystalline CZ wafers are used for large-scale manufacturing
of SHJ solar cells. Despite promising results have been reported by several research
groups for SHJ cells made of epitaxial Si wafers or so-called “monolike wafers”
[18], produced by advanced directional solidification process [19], the use of these
potentially low-cost wafers in mass production requires further developments.
Both n- and p-doped c-Si wafers have been used for manufacturing SHJ solar cells,
however, the highest efficiency was obtained for n-doped wafers [20]. The asymmetry
between conduction and valence band offsets is generally assumed as a reason of
worse performance of p-type SHJ solar cells [21]. It has also been speculated that
in the case of n-type Si strong bending of conductive band at p/n junction prevents
back diffusion of electrons and, therefore reduces their surface recombination, while
hole transport through narrow “spike” barrier is facilitated by comparatively high
density of localized states in the valence band tail of a-Si:H (see Fig. 7.3). Thus, for
118 A. S. Abramov et al.

Fig. 7.3 Schematic of band

diagram of a bifacial SHJ
solar cell based on n-type
silicon wafer

n-type wafers, intrinsic a-Si:H passivation layer along with field effect passivation,
resulted from the large valence band offset, decreases recombination rate at the
p/n interface, which almost approaches the intrinsic recombination limit of the high-
quality c-Si [22]. Additional benefits of n-type c-Si are the absence of LID, associated
with boron–oxygen complexes in p-type material, as explained above, and lower
sensitivity of n-type silicon to metal impurities [23], which are usually present in the
silicon feedstock, therefore less effort has to be made to obtain n-type silicon wafers
with high electronic quality.
The electronic properties of monocrystalline silicon wafers for high efficiency
solar cells are determined by impurities and dopant concentrations. Since measure-
ments of these parameters require special techniques that can hardly be used in mass
production process, the minority carrier lifetime and wafer resistance are usually
measured in practice as the main parameters to trace the quality of wafers or ingots.
These parameters usually vary along and across ingots and their profiles depend on
the details of ingot growth process and post treatment. Therefore, it has been pro-
posed to use the measured lifetime/resistance ratio as a cumulative measure of wafer
and heterojunction qualities [24]. Recent studies have also shown that for SHJ cells
with Voc > 750 mV one has to use passivated wafers with a lifetime-to-resistance
ratio above 4 ms/Ohm [25].
The most significant advantage of SHJ technology in terms of cost reduction is
that all process steps are performed at low temperatures (<250 °C) favoring the use of
thin wafers for SHJ solar cell production. Recent progress in wafer slicing technology
due to implementation of diamond wire technology has resulted in mass production
of low-cost wafers with thickness less than 160 µm. Wafers with as cut thickness of
150 µm have recently been successfully implemented in the SHJ production process
without losses in the power of modules [26]. Although further reduction of SHJ cell
thickness is also possible without significant losses in the efficiency [27], currently
implementation of thinner wafers in mass production is limited by handling issues
resulting in excess wafer brakeage rates.

7.2.3 Wafer Texturing and Cleaning

Similar to other c-Si PV technologies, wet chemical treatment is the first step in
production sequence of SHJ cells. It usually includes the following steps:
7 Silicon Heterojunction Technology … 119

– etching of highly defective surface part of wafer (surface damage etch—SDE);

– forming of special surface morphology (texture) reducing light reflection from
wafer surface (TEX);
– cleaning of wafer surface from organic and metal impurities.
If first two steps based on isotropic (SDE) and anisotropic (TEX) alkali etching
are similar to those used in other silicon PV technologies, the last step significantly
differs. To obtain high-quality surface passivation the wafer surface should be extra
clean. Therefore, enhanced cleaning procedure is necessary for production of high-
efficiency SHJ cells, and includes removal of residual organic, ionic and metallic
contamination (coming from wet etching/texturing solutions) from c-Si wafer sur-
face. In addition, heavy deionized (DI) water rinse steps are used between each
chemical treatment. The wet chemical treatment ends up with a short HF dip, which
removes native oxide and passivates c-Si surface with hydrogen atoms prior to a-Si:H
PECVD processes.
At the beginning of SHJ technology development, the RCA cleaning process,
which is common in microelectronics [28], was adapted for wafer cleaning [29].
The RCA process, however, includes hydrogen peroxide, which has high decompo-
sition rate at process temperatures above 65 °C that are required to reach reason-
able throughput during mass production of SHJ solar cells. On the other hand, a
substitution of H2 O2 with ozone (O3 ) as an oxidizing agent has also been used in
microelectronics for a long time [30]. The O3 -based cleaning process allows reduc-
ing chemical consumption and waste generation, and facilitates the process control.
It has been shown that O3 -based process is effective not only for cleaning of organic
residuals but also in removing metallic contamination [31, 32]. Consequently, an
industrial cleaning procedure usually includes three steps—cleaning from organics
in DI water/O3 mixture (DIO), cleaning from metallic contamination in HCl/HF
solution either with or without O3 and final oxidation in HCl/O3 solution. A recent
study [33] has shown that this process sequence can be reduced down to two process
steps or even to a single cleaning step by fine tuning of concentration of chemicals.

7.2.3.1 Surface Passivation and Junction Formation

High-quality surface passivation is a key to achieve high values of Voc in high-

efficiency silicon-based solar cells. An insertion of thin (<10 nm) hydrogenated
amorphous silicon (a-Si:H) layers between c-Si wafer and doped a-Si:H layers lead
to higher Voc values in comparison to those registered in case when intrinsic a-Si:H
layers are absent [34]. In the last two decades the surface passivation of SHJ cells
has been essentially improved, resulting in Voc values close to 750 mV [13], which
approach the theoretical limit (~760 mV) and underline the particular appeal of this
technology. From a microscopic point of view, surface passivation of c-Si by a-Si:H is
most likely closely linked to hydrogenation of surface states [35], where the hydrogen
is supplied from the passivating film during its growth and post-annealing process.
Generally, low temperature annealing of a-Si:H films in ambient air (which in some
120 A. S. Abramov et al.

extend mimics curing process of low temperature silver paste during manufacturing
of SHJ solar cells) improves the passivation properties. As a rule, a necessary condi-
tion for good surface passivation is that the a-Si:H/c-Si interface should be atomically
sharp, meaning that silicon epitaxial growth is avoided, that is, no crystalline material
is deposited [36]. This can be achieved by a proper tuning of a-Si:H film properties
during deposition process.
In practice, a-Si:H layers are commonly grown by plasma-enhanced chemical
vapor deposition (PECVD) method using parallel-plate capacitively coupled plasma
(CCP) discharge in pure silane or in silane-hydrogen mixtures at temperatures close
to 200 °C. It appears that the most critical process parameters for surface passivation
are the hydrogen-to-silane gas flow ratio and power density, whereas the gas pressure
may affect the film thickness uniformity. It has been reported that plasma regimes
close to the amorphous-to-microcrystalline transition are well suited for deposition of
device-grade passivating a-Si:H layers [37]. These regimes can be produced either
in silane plasmas highly diluted with hydrogen or in highly depleted pure silane
plasmas [38]. The use of hydrogen plasma treatments during the a-Si:H deposition
might also be beneficial [39]. Note that today most industrial CCP reactors employ
RF (13.56 MHz) power for plasma generation. This is mainly due to the fact that
the low excitation frequency allows better control of a-Si:H film thickness over large
areas (>1 m2 ). Nevertheless, the main drawbacks of standard RF PECVD reactors
are: 1) low a-Si:H deposition and etching (during in situ plasma cleaning) rates due to
low plasma density limiting the PECVD system throughput and 2) high-energy ion
bombardment of the a-Si/c-Si interface due to high sheath voltages resulting in higher
defect densities in passivating a-Si:H layers. Owing to these reasons, alternative
deposition techniques such as very high frequency (VHF) PECVD [35, 39], hot wire
(or catalytic) CVD [40, 41], expanding thermal plasmas [42], and so on [43, 44]
have also been investigated in the recent years for c-Si surface passivation both in
academia and industry. Currently, Hevel is the only company, which employs VHF
(40.68 MHz) PECVD reactors for mass production of high efficiency HJT solar cells.
It has been shown [45] that Gen5 KAI PECVD reactors, originally designed by the
former Oerlikon Solar for mass production of micromorph modules, can provide an
excellent uniformity (<5%) of optical and electrical properties of a-Si:H layers across
entire surface of a 110 × 130 cm2 wafer carrier. Since spring of 2017 [46], Gen5 KAI
PECVD reactors have successfully been used for mass production of high-efficiency
SHJ solar cells at the Hevel production facility, as described in Sect. 7.3.
To fabricate SHJ solar cells, the passivating intrinsic a-Si:H layers should be
coated with p- and n-doped a-Si:H layers for the front emitter and back surface
field formation, respectively. However, as shown in [47], their electronic passivation
properties are often found to be inferior in comparison to those of their intrinsic
counterparts despite the fact that they may yield a field effect at the a-Si:H/c-Si
interface. Deterioration of surface passivation is more pronounced for p-doped layers
rather than for n-doped layers. On the other hand, n-doped a-Si:H layers can have
electric conductivity one or two orders in magnitude higher compared to p-doped
a-Si:H layers [48]. Besides, n-doped layers can easily be made nano- or micro-
crystalline, which makes them even more conductive and transparent. As a result,
7 Silicon Heterojunction Technology … 121

the rear emitter (RE) design in which the emitter is located at the rear surface of
c-Si wafer is more suitable for manufacturing of high-efficiency SHJ solar cells. For
RE configuration the front lateral current collection is facilitated by the Si wafer
itself, as electrons abundant in the n+ region near the cells front surface have a very
high diffusion length. Since the electron transport is supported by the wafer, some
limitations are lifted for the front TCO layer as well as the current collection grid
allowing making them more transparent and less dense, respectively. In other words,
this leads to cell with higher fill-factor at higher short-circuit current compared to the
FE (front emitter) counterpart. It should also be mentioned that in recent years there
has been growing interest in developing of amorphous and nano-crystalline silicon
oxide layers [49, 50] acting as passivation and window layers for high-efficiency
SHJ solar cells due to their higher optical band-gap energy, resulting in less parasitic
absorption.
As mentioned above, the c-Si wafer surface should be clean and nonoxidized
in order to obtain good solar cell performance in terms of surface passivation. In
this matter ex situ photoluminescence (PL) imaging is a useful technique for process
optimization, in particular during mass production process in industry. This is because
PL measurements are nondestructive, environment-insensitive and contactless, and
hence can practically be applied at any process step. Although several mechanisms
affect the PL response, it is generally found that intense PL signals correlate with
good interface properties. A strong PL response is widely regarded as an indicator
of a high-quality surface passivation. In situ PL measurements can also be used for
more precise control of a-Si:H passivation properties [51, 52].

7.2.4 Transparent Conductive Oxides

Since thin-doped a-Si:H layers have high resistance, they are commonly coated with
transparent conductive oxides (TCO) layers to collect the photo-generated current.
The total current produced by 6 in size SHJ solar cells usually exceeds 9 A, therefore
additional metallic electrodes are required for efficient current collection, which are
the subject of next section. The sheet resistance of TCO layers must be sufficiently
low (typically below 100 /sq.) to avoid deteriorating of the fill factor. Direct current
(DC) sputtering is commonly used in industry for double-side deposition of TCO
layers simplifying the manufacturing process of bifacial HJT solar cells. In this case,
the wafer carrier acts as the edge shadowing mask at the rear surface resulting in
edge insolation between front and rear ITO layers. The size of edge insolation should
be minimal to obtain maximum short-circuit current. The use of cylindrical rotary
targets allows increasing TCO target utilization factor (up to 80%) and, in turn, to
extend their exchange period, minimizing the production costs [53]. On the contrary,
planar TCO targets that are widely used in R&D studies have a target utilization
factor of only 30%. Ceramic indium tin oxide (ITO) targets of various composition
(In2 O3 :SnO2 = 90:10, 95:5, 97:3, etc.) are often the TCO material of choice in
SHJ cells, in particular for the front TCO layer, which also acts as an antireflection
122 A. S. Abramov et al.

coating (ARC). The refractive index of most TCO layers is close to 2 at a wavelength
of 600 nm, therefore the thickness of the front ITO layer must be fixed at about 75 nm
to minimize reflection losses.
Carrier mobility of ITO µ = 20–40 cm2 /Vs are typical for DC sputtered ITO
layers, and carrier density can be tuned from 1019 to 1021 cm−3 by adjusting the flow
of oxygen in argon atmosphere during sputtering, giving front ITO layers with sheet
resistances below 100 /sq. and optical transmittance above 90% for 80–100 nm
thick layers [54, 55]. Among typical ITO target compositions 97:3 ITO targets yield
thin films with high mobility and optical transparency, making them suitable for the
front TCO. Nevertheless, alternative TCO materials with even higher motilities and
therefore lower absorption for the same sheet resistance have also been extensively
investigated in recent years. These include hydrogen (IO:H) and tungsten (IWO)
doped indium oxide [56, 57] or their combination (IWOH) [58], however, to obtain
such high mobility layers on large scale is challenging. It should also be emphasized
that the high UV content of magnetron sputtering discharges in combination with high
energy neutrals can lead to damage of underlying a-Si:H layers [59]. Therefore, there
has also been interest in the development of “soft” deposition techniques for TCO
layers such as low pressure chemical vapor deposition (LPCVD) (mainly used for
aluminum-doped zinc oxide (AZO) [60]), remote plasma deposition (RPD) [58] and
most recently atomic layer deposition (ALD) [61]. Nevertheless, inline magnetron
sputtering is still the workhorse of the PV industry for depositing of various TCO
layers.

7.2.5 Metallization

The process requirements for manufacturing SHJ solar cells have several advantages
in comparison to those of conventional homojunction c-Si solar cells. First of the
advantages is low thermal budget during the heterojunction formation. The deposi-
tion temperature of a-Si:H and ITO layers is usually less than 250 °C. Second, the
time required to form the a-Si:H/c-Si junctions and contact layers is also shorter for
SHJ cells than for conventional c-Si solar cells based on thermal diffusion processes.
Third, wafer bowing is suppressed due to the low process temperature and symmetric
structure of SHJ solar cells. However, the main disadvantage of using low temper-
ature processes is that standard firing-through metallization techniques (with firing
temperatures in the range of 800–900 °C) cannot be implemented in SHJ cells. This is
because the a-Si/c-Si heterojunction does not withstand process temperatures above
200–250 °C, at which hydrogen effusion from internal surfaces of the heterojunction
leads to detrimental effects on the cell performance. Due to this reason, the so-called
low curing temperature (LCT) silver paste is commonly used for metallization of
SHJ cells via screen-printing, which is currently the state-of-the-art technology for
the metal grid deposition. The LCT Ag paste, however, has a typical bulk resistivity
in the range of 5–6 µ*cm [62], that is, 2–3 times higher than the bulk resistivity of
conventional firing pastes. Unlike firing-through paste printed fingers, which after
7 Silicon Heterojunction Technology … 123

firing step became dense and highly conductive (their conductivity is almost equal
to bulk Ag), fingers made of LCT Ag paste represent a polymeric matrix containing
Ag particles even after curing. Hence, the contact resistance of the particles limits
the conductivity of fingers made of LCT Ag paste. On the other hand, the LCT paste
is in direct contact with thin TCO layer rather than with Si surface, therefore the
contact resistance between ITO and LCT Ag paste is another important parameter
directly affecting the SHJ solar cell performance. Consequently, it is obvious that
five busbar (5BB) or multiwire concepts are favorable for interconnection of SHJ
cells with LCT Ag paste contact grids in order to minimize the current collection
ohmic losses and to increase the cell efficiency. In addition, the resistivity of LCT Ag
paste can be controlled by changing the form and size of Ag particles, ranging from
microscopic flakes to spherical nanoparticles. Several ideas have also been proposed
for decreasing the Ag content of LCT paste to minimize its cost such as replacing
the Ag particles by silver-coated copper or graphite; however, to the best of our
knowledge, none of these alternatives has been implemented in mass production yet.
To obtain the best cell performance, fine-tuning of the LCT Ag paste properties
(such as viscosity, polymeric and silver content) should be performed as well as a
trade-off between its printability, adhesion, aspect ratio and bulk conductivity should
be found. For the state-of-the-art products a finger width of as low as 30–40 µm and
an aspect ratio of 1:3—1:2 can be achieved in combination with busbarless design,
resulting in SHJ cell efficiencies above 23% [63, 64].
It should be mentioned that the cell interconnection is one of the factors that have
an impact on the total cost of SHJ solar modules. Since conventional soldering is, as a
rule, not compatible with LCT Ag paste [65], alternative approaches like SWCT [66]
or ECA [67] ribbon gluing have been implemented. Both technologies have proven
their reliability; however both have higher cost compared to conventional soldering
due to incorporation of expensive materials (such as Ag or In) in the interconnection
elements.
As of today, the only alternative to LCT Ag paste for SHJ cell metallization is
Cu electroplating [68]. This process can boast very small finger line width down to
20–25 µm, very high bulk conductivity of the fingers (bulk resistivity <2 µ*cm),
allows for acceptable peel-force after soldering, good module reliability, lower pro-
duction costs, however suffers from significant cost of the manufacturing equipment.
Nevertheless, future scenarios predict significant decrease of the overall equipment
and production costs placing Cu electroplating as a serious competitor to standard
Ag paste screen-printing for HJT solar cell metallization [69].

7.3 Conversion of “Micromorph” to SHJ Technology

Recently, record SHJ solar cell efficiencies have attracted attention of companies
manufacturing equipment for mass production of c-Si solar cells. In recent years, not
only production tools designed for different process steps of SHJ technology but also
turnkey production lines have appeared on the PV market [70]. Despite these lines
124 A. S. Abramov et al.

allowed production of high-efficiency SHJ cells, there was a question behind—can

PV production tools originally designed for silicon thin film PV be reused in produc-
tion lines for SHJ cells? Indeed, in the late 2000s, many production lines designed
for single or double-junction amorphous silicon PV have been installed worldwide
and became noncompetitive in next few years with drastic drop in polysilicon price.
The core equipment of these lines—PECVD reactors—are similar to those required
for producing of SHJ solar cells. Reuse of these complex and expensive systems
can yield significant cost savings to the companies deciding to enter SHJ solar cell
market and give them a chance for return of initial investments made in Si thin film
PV. Moreover, such approach can enhance positive ecological effect of PV by sup-
pression of CO2 emission required for production of new PV mass production tools
and for recycling old ones.
The answer to the abovementioned question was found at the Ioffe Institute. This
leading Russian research center in semiconductor physics and PV technology, which
celebrates its 100th anniversary in 2018, is alma mater of three Nobel Prize win-
ners. It is remarkable that the last of them, Professor Alferov, got his Noble price
for works on heterojunctions. Despite the main focus of the Ioffe Institute was on
heterojunctions made of III–V semiconductor compounds for CPV and space appli-
cations, there were also significant activities on amorphous silicon technology since
1980s [71]. Research activity on silicon thin film PV at the Ioffe Institute has partic-
ularly been intensified in 2010 with start of Hevel PV project, launched by Renova
group of companies and Rusnano. In frame of the project, Thin Film Technology in
Energetics (TFTE) R&D Center was established at the Ioffe Institute as a research
unit focused on technological support of Hevel’s production line. For that purpose,
the R&D center was equipped with a pilot line having process tools similar to those
used in micromorph production lines, but with lower production capacity, and a set
of laboratory tools needed for solar cell characterization and material studies. Con-
sequently, Hevel project was initially aimed for mass production of micromorph PV
modules. Nevertheless, the sharp drop in polysilicon price and continuous progress
in the development of conventional c-Si PV made micromorph modules noncompet-
itive soon after the beginning of Hevel project. Therefore, a demand for significant
improvements of the micromorph production line has been addressed to the Ioffe
Institute. The proposed solution was to convert Hevel’s production line into SHJ
solar cell line. This proposal was supported by demonstration that KAI PECVD
reactors, originally designed by Oerlikon Solar for micromorph technology, can be
effectively used in SHJ cell production process [45]. Soon after the so-called HJT
(heterojunction technology) project has been started, the most challenging part of
the project was to adapt PECVD reactors, designed for micromorph module pro-
duction, to the requirements of SHJ technology. Besides, all process steps described
in Sect. 7.2 were developed and optimized at TFTE R&D center. As a result, the
efficiency of laboratory cells above 22% has been achieved within 3 years from the
start of the project, as demonstrated in Fig. 7.4. The implementation of this tech-
nology into the mass production allowed to double Hevel’s production capacity and
resulted in an average cell efficiency of 21.5% at the beginning of mass production
in April, 2017 [46]. In the next one and half years, the average SHJ cell efficiency
was enhanced up to 22.8% by continuous technology improvement.
7 Silicon Heterojunction Technology … 125

Fig. 7.4 a Progress in R&D SHJ solar cell efficiency achieved at TFTE R&D Center in collaboration
with the Ioffe Institute and b an example of SHJ cell efficiency distribution for a weekly mass
production at the Hevel production facility

7.4 SHJ Module Technology and Reliability Issues

7.4.1 Cell Requirements

One of the main challenges in assembling of c-Si solar cell module is to minimize the
so-called cell to module losses (CTM) that have both optical and electrical origins. To
tackle the optical losses the front glass with antireflective coating (ARC) is generally
used today for field PV applications. On the other hand, applications such as building
integrated photovoltaics (BIPV) and automotive integrated photovoltaics (AIPV)
tend to avoid ARC due to visual appearance issues, which allows for up to 3%
relative power gain compared to standard solar glass. To minimize electrical losses,
current collecting and interconnecting ribbons with higher cross-section are being
used. In this respect, one should admit an inherent advantage of SHJ cells compared to
traditional Al:BSF or even PERC cells, since the same power can be achieved by SHJ
cells at lower operating current due to significantly higher operating voltage values,
caused by excellent passivation properties of a-Si:H layers introduced at the cell
126 A. S. Abramov et al.

surface [20]. Lower cell operating current leads to lower requirements to electrical
resistance of interconnection elements, hence lowering resistive CTM losses.
Furthermore, to avoid unnecessary CTM losses related to the cell current mismatch
in a module, it is very beneficial to perform sorting by maximum power point current
additionally to the standard efficiency bin cell sorting. At Hevel production facility,
this has proved to result in a power gain of up to 2 W on a 60-cell module compared
to standard efficiency bin sorting strategy and at the same time minimizing the risk
of appearance of hot cells due to the current mismatch.
Typical electroluminescence (EL) images of SHJ modules with current matched
(a) and current mismatched (b) cells are presented in Fig. 7.5. The advantage of
current match sorting is clearly seen by homogeneous EL light intensity distribution
for the current matched case.

7.4.2 Cells Interconnection

Interconnection of SHJ cells is a stumbling stone for the whole process chain [72],
as soldering which is used for interconnection of conventional c-Si cells is not com-
patible with LTC Ag paste, which has to be applied instead of the standard firing
through silver paste due to temperature restrictions for the a-Si/c-Si heterojunction.
This type of pastes has higher bulk resistivity (2–3 times that of the high temperature
pastes) and low adhesion after soldering. Usually, Ag busbars are easily peeling off
the ITO surface with force well below 1 N/mm [73].
To overcome this limitation, new cell interconnection technologies have been
proposed, for example, gluing of ribbon using electrical conductive adhesive (ECA)
[67, 74] or multiwire interconnection using low-temperature attachment of a foil with
embedded InSn coated wires called SmartWire Connection Technology (SWCT® )
[66, 75]. InSn alloy with a low melting temperature of around 120 °C has a good adhe-
sion, both to Ag paste and ITO-layer itself; hence a metallurgic contact between the
wire and the cell surface is established after temperature treatment. SWCT technol-
ogy does not require very precise positioning of the ribbons relative to the metalliza-
tion grid, which is one of the major challenges for multiwire technology. The initial
attachment of the cells to the wires in SWCT does not require precise soldering of the
wires to the solder pads and is commonly done by an adhesive layer-containing foil,
which allows using a large number (up to 24) of relatively thin wires of 200–250 µm
in diameter.
Optimizing the SWCT module assembly bill of materials (BOM)—by tuning the
optical properties of the lamination foils and electrical properties of the wires—we
have achieved at Hevel a gain of 9 W on a 60-cell module compared to standard BOM
without increasing the material costs and changing the cell manufacturing process
and efficiency (see Table 7.1).
7 Silicon Heterojunction Technology … 127

Table 7.1 Best production

Standard BOM Optimized BOM
values for module
IV-characteristics with Pmax (W) 310 326
standard and optimized BOM Vpmax (V) 35.5 36.67
Ipmax (A) 8.75 8.89
Isc (A) 9.37 9.33
Voc (V) 43.34 44.15
FF (%) 76.33 79.14

7.4.3 Module Types

SHJ solar cells possess an inherent property of having bifacial light sensitivity. Unlike
competitors, SHJ cells do not require any special process steps to become bifacial—
only an additional screen printing step for the back cell side. At the same time,
producing bifacial SHJ cells allows to avoid full metallization of the backside, sim-
plifying significantly the PVD equipment that is no more required to deposit metallic

Fig. 7.5 EL images of SHJ

modules composed of
current matched sorted cells
(a) and cells without current
sorting
128 A. S. Abramov et al.

layers and thus allowing significant CAPEX savings [55]. The bifaciality factor mea-
sured for SHJ bifacial solar cells is as high as 95% [76], which is significantly higher
than that of the PERC cells—80% [77]. In combination with white backsheet when
assembled into the module, bifacial cells yield the same power as monofacial cells
of the same quality, since lack of Ag back side reflector is compensated by light
reflection from the white backsheet. At the same time, full benefit of cell bifaciality
can be fulfilled when they are assembled with transparent backsheet or back glass. In
this case, taking into account the albedo of the module surroundings and the amount
of scattered light, the yield of bifacial SHJ modules can surpass that of any monofa-
cial module—even the most efficient IBC cell-based products. The yield of bifacial
modules can exceed the yield of monofacial ones of the same class by 13% even for
moderate albedo (green grass in summer and snow coverage in winter), according
to measurements conducted at Hevel [78].
In this respect, two types of modules are being produced at Hevel’s production
facility:
– Monofacial glass/backsheet modules with bifacial SHJ solar cells that allow reach-
ing maximum module power at STC IV-curve measurements. These modules are
the best option for application in environment where the bifacial effect cannot be
utilized (rooftop applications, BIPV, locations with low albedo levels).
– Bifacial glass/glass modules with bifacial SHJ solar cells (at the moment at devel-
opment stage at Hevel) allow obtaining maximum yield in environment with high
backside reflection. Best applications for them are large area solar farms in snowy
regions or in regions with a large number of overcast days when scattered light
plays a significant role.
A typical daily statistic of the module power distribution (glass/white backsheet,
60 cell modules) at Hevel’s production line is presented in Fig. 7.6.

Fig. 7.6 Typical daily glass/white backsheet 60 cell module production statistics at Hevel
7 Silicon Heterojunction Technology … 129

Table 7.2 The measured cell efficiency values of Hevel SHJ cells before and after laser cutting
Isc Voc FF Eff Rs Rsh
As produced 9.230 0.728 79.43 21.83 0.0052 8795.0
After cut 9.219 0.723 78.92 21.54 0.0050 4803.3

7.4.4 Perspective Products

One of the perspective products emerging at global market is a half-cut cell module.
Designed to incorporate two half-cut cell modules connected in parallel to each other
on one glass (alternative configurations also apply) it allows to keep the output IV-
characteristics the same as for standard full-cell modules, while significantly reducing
resistive losses due to two times lower current generated by the half-cells.
A critical process in producing half-cells is laser cutting of the cell. To avoid
unnecessary modifications to cell production equipment and handling tools, it is
preferable to make the laser cut as the last process step on a finished cell. This brings
certain complications, as very special care has to be taken not to shunt the cell by
melting the ITO layers together on the cell sides in the cut area. The cut region also
suffers from lack of surface passivation, as the border area loses passivation after the
cut. Taking into account these facts, it is natural to expect a certain decrease of the
cell efficiency for half-cut cells. The results of cell efficiency measurement of Hevel
SHJ cells before and after the cut are presented in Table 7.2. Note that the cells have
not been specially selected by efficiency values.
The drop in efficiency comprises from a slight decrease of Voc (5 mV) and FF
(0.51%). The difference in Isc is caused by deletion of cell material during the cutting
process. The total efficiency loss is below 0.3% abs, which is quite acceptable and
is the state-of-the-art for existing laser cutting technology. It should also be noted
that the laser cut does not influence significantly the cell shunt resistance; hence
the reverse current value remains in the acceptable range for the modules to pass
reliability testing.
The observed cell efficiency drop is fully compensated on a module level by the
decrease of CTM due to lower overall cell current. At Hevel, we have achieved a
4–5 W relative increase of module power at STC when comparing 144 half-cell and
72 full-cell modules assembled from the same power class cells. For our 144 half-cell
glass/glass module prototype, we have achieved the power figures shown in Table 7.3.
The module is characterized by a high fill-factor value that evidences decreased ohmic
losses due to lower current generated by each half-cell. The values for 10 and 20%
albedo clearly indicate that bifacial modules when used in the right environment can
ensure unsurpassed energy yield by far exceeding that of the monofacial modules.
130 A. S. Abramov et al.

Table 7.3 Parameters of

Measurement STC BiFi + 10%a BiFi + 20%a
bifacial 144 half-cell
conditions
glass/glass module prototype
Pmax (W) 389 425 460
Impp (A) 8.78 9.55 10.31
Vmpp (V) 44.30 44.47 44.67
Isc (A) 9.33 10.21 11.09
Voc (V) 53.06 54.15 54.81
FF (%) 78.53 76.87 75.68
a Calculated values based on 10 and 20% backside illumination and

module power bifaciality factor of 93%

7.5 Summary

Nowadays, SHJ technology yields the highest conversion efficiencies at R&D level
for both double- and one-side contacted silicon solar cells due to excellent surface
passivation properties of a-Si:H layers. Low temperature processes used for manu-
facturing of SHJ solar cells can result in potential cost savings when wafer thickness
is reduced below 150 µm, which is currently limited by higher breakage rates caused
by the use of standard automation equipment in mass production. In addition, SHJ
technology can be implemented in a-Si thin-film production lines in a cost-effective
way, as has been demonstrated by Hevel company under the leadership of the Ioffe
Institute. Despite great challenges in interconnecting of HJT solar cells, reliable
solutions exist, which enable assembling HJT solar modules with ultra-high output
power. One of the main advantages of HJT technology is an inherent bifaciality of
the cells that allows obtaining record energy yields for solar modules with backside
sensitivity in outdoor environments where reflected or scattered light can be utilized.
In combination with half-cell approach benefiting from lower ohmic losses due to
decreased cell current passing through the conductive elements of the module, an
output power as high as 460 W at STC+20% bifacial gain could be obtained for
glass/glass 72 cell modules at Hevel.

References

1. M.A. Green et al., Prog. Photovolt. 26, 427 (2018)

2. K. Sasaki, et al., in Proceeding of the 9th International Conference on Concentrating Photo-
voltaics Systems, Miyazaki, 2013
3. K.R. McIntosh, et al., in Proceeding of the 3rd World Conference on Photovoltaic Energy
Conversion, Osaka, 2003, p. 971
4. D.D. Smith, et al., in Proceeding of the 43rd IEEE Photovoltaic Specialists Conference, Port-
land, 2016, p. 3351
5. P. Verlinden, Interdigitated back contact solar cells, in Photovoltaic Solar Energy: From Fun-
damentals to Applications, 1st edn., Wiley & Sons, 2017, pp. 92–102
7 Silicon Heterojunction Technology … 131

6. M.A. Green, Sol. Energy Mater. Sol. Cells 143, 190 (2015)
7. Y. Feng et.al., in IEEE 43rd Photovoltaic Specialists Conference (PVSC), Portland, 3360 (2016)
8. S. Wilking, C. Beck, S. Ebert, A. Herguth, G. Hahn, Sol. Energy Mater. Sol. Cells 131, 2 (2014)
9. T. Blevin et al., Sol. Energy Mater. Sol. Cells 131, 24 (2014)
10. W. Cai et al., Energy Proc. 92, 399 (2016)
11. L. Kreinin, N. Bordin, N. Eisenberg, 1st BifiPV Workshop, Konstanz, 2012
12. M. Taguchi, et al., in Proceeding of the 5th International Photovoltaic Science and Engineering
Conference, Kyoto, 1990, p. 689
13. M. Taguchi et al., IEEE J. Photovoltaics 4, 96 (2014)
14. A.V. Sachenko et al., J. Appl. Phys. 119, 225702 (2016)
15. K. Yoshikawa et al., Nature Energy 2, 17032 (2017)
16. W. Fuhs, K. Niemann, J. Sruke, Bull. Am. Phys. Soc. 19, 394 (1974)
17. M. Schaper, J. Schmidt, H. Plagwitz, R. Brendel, Prog. Photovolt., Res. Appl. 13, 381 (2005)
18. E. Kobayashi, Y. Watabe, R. Hao, T.S. Ravi, Prog. Photovolt: Res. Appl. 24, 1295 (2016)
19. F. Jay, D. Muñoz, T. Desrues, E. Pihan, V. Amaralde-Oliveira, N. Enjalbert, A. Jouini, Sol.
Energy Mater. Sol. Cells 130, 690 (2014)
20. S. De Wolf, A. Descoeudres, Z.C. Holman, C. Ballif, Green 2, 7 (2012)
21. W. Fuhs, L. Korte, M. Schmidt, J. Optoelectron. Adv. M. 8, 1989 (2006)
22. J. Melskens, B.W.H. van de Loo, B. Macco, L.E. Black, S. Smit, W.M.M. Kessels, IEEE J.
Photovolt. 8, 373 (2018)
23. D. MacDonald, L.J. Geerligs, Appl. Phys. Lett. 85, 4061 (2004)
24. D. L. Bätzner, et al., in Proceeding of the 31st European Photovoltaic Solar Energy Conference,
Hamburg, 2015, p. 1042
25. A. Danel, S. Harrison, F. Gérenton, R. Varache, J. Veirman, in Proceeding of the 35th European
Photovoltaic Solar Energy Conference, Brussels, 2018, 2DO.1.2
26. D. Andronikov, et al., in Proceeding of the 35th European Photovoltaic Solar Energy Confer-
ence, Brussels, 2018
27. A. Abramov, et al., in Proceeding of the 35th European Photovoltaic Solar Energy Conference,
Brussels, 2018
28. W. Kern, J. Electrochem. Soc. 137, 1887 (1990)
29. Q.Wang, et al., in Proceeding of the 33rd IEEE Photovoltaic Specialist Conference, San Diego,
2008
30. G.Chen, in Proc. Int. Conf. Wafer Rinsing, 1999
31. T.Bearda et al., Solid State Phenom. 187, 341 (2012)
32. F.Sevenig et al., Energy Proc. 8, 288 (2011)
33. V. Breus et al., in Proceedings of the 35th European Photovoltaic Solar Energy Conference,
Brussels, 2018
34. M. Tanaka et al., Jpn. J. Appl. Phys. 31, 3518 (1992)
35. S. De Wolf, S. Olibert, C. Ballif, Appl. Phys. Lett. 93, 032101 (2008)
36. H. Fujiwara, M. Kondo, Appl. Phys. Lett. 90, 013503 (2007)
37. A. Descoeudres, L. Barraud, R. Bartlone, G. Choong, S. De Wolf, F. Zicarelli, C. Ballif, Appl.
Phys. Lett. 97, 183505 (2010)
38. B. Strahm, A.A. Howling, L. Sansonnens, Ch. Hollenstein, Plasma Sourc. Sci. Technol. 16, 80
(2007)
39. A. Descoeudres et al., Appl. Phys. Lett. 99, 123506 (2011)
40. J.J.H. Gielis et al., Phys. Rev. B 77, 205329 (2008)
41. J.W. Schuttauf et al., Appl. Phys. Lett. 98, 153514 (2011)
42. A. Illiberi, K. Sharma, M. Creatore, M.C.M. van de Sanden, Phys. Stat. Sol. RRL 4, 172 (2010)
43. K. Maydell, E. Conrad, M. Schmidt, Prog. Photovolt. 14, 289 (2006)
44. U.K. Das, M.Z. Burrows, M. Lu, S. Bowden, R.W. Birkmire, Appl. Phys. Lett. 92, 063504
(2008)
45. S. Abolmasov, et al., in Proceedings of the 31th European Photovoltaic Solar Energy Confer-
ence, Hamburg, 2015, p. 1046
132 A. S. Abramov et al.

46. D. Andronikov, et al., in Proceedings of the 33rd European Photovoltaic Solar Energy Con-
ference, Amsterdam, 2017, p. 732
47. S. De Wolf, M. Kondo, J. Appl. Phys. 105, 103707 (2009)
48. A. Shah, Thin-film silicon solar cells, 1st edn. (EPFL Press, Lausanne, 2010)
49. T. Mueller, J. Wong, A.G. Aberle, Energy Proc. 15, 97 (2012)
50. K. Ding, U. Aeberhard, F. Finger, U. Rau, J. Appl. Phys. 113, 134501 (2013)
51. S.N. Abolmasov, P. Roca i Cabarrocas, J. Vac. Sci. Technol. A 33, 021201 (2015)
52. F. Lebreton, S.N. Abolmasov, F. Silva, P. Roca I Cabarrocas, Appl. Phys. Lett. 108, 051603
(2016)
53. P. Lippens, M/ Büchel, D. Chiu, C. Szepesi, Thin Solid Films 532, 94 (2013)
54. C. Frei, D. Chiu, P. Lippens, in Proceedings of the 28th European Photovoltaic Solar Energy
Conference, Paris, 2012, p. 2029
55. S. Abolmasov, et al., in Proceedings of the 33rd European Photovoltaic Solar Energy Confer-
ence, Amsterdam, 2017, p. 787
56. L. Barraud et al., Sol. Energy Mater. Sol. Cells 115, 151 (2013)
57. Z. Lu, F. Meng, Y. Cui, J. Shi, Z. Feng, Z. Liu, J. Phys. D Appl. Phys. 46, 075103 (2013)
58. E. Kobayashi, N. Nakamura, Y. Watabe, in Proceedings of the 27th European Photovoltaic
Solar Energy Conference, Frankfurt, 2012, p. 1619
59. B.M. Meiners, S. Holinski, P. Schäfer, S. Hohage, D. Borchert, in Proceedings of the 31st
European Photovoltaic Solar Energy Conference, Hamburg, 2015
60. G. Choong, P. Bole, L. Barraud, F. Zicarelli, A. Descoeudres, S. De Wolf, C. Ballif, in Pro-
ceedings of the 25th European Photovoltaic Solar Energy Conference, Valencia, 2010, p. 2505
61. Y. Kuang et al., Sol. Energy Mater. Sol. Cells 163, 43 (2017)
62. A. Bettinelli, et al., in Proceedings of the Metallization and Interconnection Workshop, 2017
63. A. Descoeudres et al., Solar Energy, in Press (2018)
64. D. Andronikov et al., in Proceedings of the 35th European Photovoltaic Solar Energy Confer-
ence, Brussels, 2018
65. S. Hoffmann, in Proceedings of the 33rd European Photovoltaic Solar Energy Conference,
Amsterdam, 2017
66. T. Söderström et al., in Proceedings of the 42nd IEEE PVSC, New Orleans, 2015
67. T. Geipel, Z. Huq, U. Eitner, Energy Proc. 55, 336 (2014)
68. A. Lachowicz et al., in Proceedings of the 33rd European Photovoltaic Solar Energy Confer-
ence, Amsterdam, 2017
69. A. Faes et al., Photovolt. Int. 41 (2018)
70. T. Soderstrom et al., in IEEE Proceedings of the of 42nd Photovoltaic Specialists Conference
(PVSC), 3703, 2015
71. A. Kosarev et al., Sov. Tech. Phys. Lett. 11, 28 (1985)
72. A. De Rose et al., Photovolt. Int., 40, 2018
73. S. Hoffmann et al., in Proceedings of the 33rd European Photovoltaic Solar Energy Conference,
Amsterdam, 2017
74. A. Schneider et al., Energy Proc. 38, 387 (2013)
75. A. Schneider, L. Rubin, G. Rubin, in Proceedings of the 4th WCPEC, Waikoloa, 1095 (2006)
76. B. Strahm et al., in Proceedings of the of 7th International Conference on Crystalline Silicon
PV, Freiburg, 2017
77. T. Dullweber et al., in Proceedings of the of 31st European Photovoltaic Solar Energy Confer-
ence, Hamburg, 2015
78. K. Emtsev et al, in Proceedings of the 33rd European Photovoltaic Solar Energy Conference,
Amsterdam, 2017
Chapter 8
III–V Solar Cells and Concentrator
Arrays

Zh. I. Alferov, V. M. Andreev and M. Z. Shvarts

Abstract Semiconductor heterostructures allow solving the problems of controlling

the fundamental parameters of semiconductor devices owing to the possibility of
change in the electronic band structure, bandgaps and refractive indices of a material
itself during epitaxial growth, as well as effective masses and mobilities of charge
carriers in it.

8.1 Introduction—From Primary Heterostructures

to III–V Solar Cells

Semiconductor heterostructures allow solving the problems of controlling the funda-

mental parameters of semiconductor devices owing to the possibility of change in the
electronic band structure, bandgaps and refractive indices of a material itself during
epitaxial growth, as well as effective masses and mobilities of charge carriers in it.
The development of the physics and technology of semiconductor heterostructures
has resulted in remarkable changes in our everyday life. Heterostructure electronics
is widely used in many areas of our civilization. It is hardly possible to imagine
our life without dual heterostructure (HS) laser-based telecommunication systems,
heterostructure solar cells (SCs), light-emitting diodes (LEDs), heterostructure bipo-
lar transistors (HEMTs), and low-noise high electron mobility transistors for high-
frequency applications including, for example, satellite television. Now dual HS

This chapter is dedicated to the memory of Valery Rumyantsev (Ioffe Institute).

Zh. I. Alferov (B)

St Petersburg Academic University, 8/3 Khlopina Str, 194021 St.-Petersburg, Russia
e-mail: [email protected]
V. M. Andreev · M. Z. Shvarts
Ioffe Institute, 26 Polytechnicheskaya str, 194021 St.-Petersburg, Russia
e-mail: [email protected]
M. Z. Shvarts
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 133

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_8
134 Zh. I. Alferov et al.

laser has entered practically every house as CD players. Heterostructure solar cells
are widely used in space and terrestrial applications.
The idea of using heterojunctions in semiconductor electronics had already been
put forward at the beginning of the electronics era. Already in his first patent con-
cerned with p-n junction transistors, W. Shockley proposed a wide-gap emitter to
obtain unidirectional injection of charge carriers [1]. A. Gubanov was the first to
analyze theoretically current–voltage characteristics of isotype and anisotype het-
erojunctions [2]. However, the most important theoretical investigations at this early
stage of heterostructure research were done by H. Kroemer, who introduced the
concept of quasi-electric and quasi-magnetic fields in a graded-band heterojunction
and made an assumption that heterojunctions might exhibit extremely high injection
efficiencies in comparison to homojunctions [3]. During the same period, various
concepts were developed regarding application of heterostructures in semiconduc-
tor SCs. The next important step was done several years later when a concept of
double-heterostructure lasers had been formulated independently by Alferov [4] and
Kroemer [5]. In 1966, it was predicted that the density of injected charge carriers
could be by several orders of magnitude higher than the carrier density in the wide-
gap emitter (“superinjection” effect) [6]. In the same year, in the paper [7] the main
advantages of the dual HS concept were summarized for the use in various devices,
especially lasers and high-power rectifiers.
At that time, general skepticism existed with respect to the possibility of creating
an “ideal” heterojunction with a defect-free interface, where theoretically predicted
excellent injection properties would be realized. In fact, the pioneering study of the
first lattice-matched epitaxial single-crystal Ge–GaAs heterojunctions by Anderson
[8] gave no proof of non-equilibrium charge carrier injection in heterostructures.
Mostly owing to this general skepticism, only few groups had been trying to find
an “ideal couple”, which seemed to be a quite difficult problem. Many conditions
of compatibility between thermal, electrical and chemical properties, as well as the
crystalline and band structures of the contacting materials are to be met.
A lucky combination of certain properties of gallium arsenide, that is, small charge
carrier effective mass, wide energy gap, effective radiation recombination, sharp
optical absorption edge due to the “direct” band structure, high electron mobility
at the absolute minimum of the conduction band and its strong reduction at the
nearest minimum at the (100) point ensured for this material a place of “active
region” in different electronic devices, even at that initial stage of research. Since
the maximum effect would be achieved by using it in the heterostructures with
wide-bandgap materials, the most promising systems considered at that time were
GaP-GaAs and AlAs-GaAs. To be “compatible”, materials of the “couple” must
have, as the first and the most important condition, close lattice constants; therefore,
heterojunctions in the AlAs-GaAs system were preferable. However, before starting
the work on preparation and study of these heterojunctions, one had to overcome
a certain psychological barrier. By that time, AlAs had long been synthesized, but
many properties of this compound remained poorly studied, since AlAs was known
to be chemically unstable and it decomposes in moist air. The possibility of preparing
stable and applicable heterojunctions in this system seemed to be not promising.
8 III–V Solar Cells and Concentrator Arrays 135

Initially, attempts to create double heterostructures were related to a lattice-

mismatched GaAsP system. Zh. Alferov et al. succeeded in fabricating first dual
HS lasers in this system by vapor phase epitaxy. However, due to lattice mismatch,
the lasing occurred only at liquid nitrogen temperature, similar to the behavior of the
homojunction lasers. At the same time, it was alighted that small crystals of AlGaAs
solid alloys of different compositions, which had been prepared by growing from
a melt, were stable at least for 2 years. It immediately became clear that AlGaAs
is suitable for preparing durable heterostructures and devices. Studies of the phase
diagrams and growth kinetics in this system and the development of a version of
the liquid phase epitaxial growth method resulted soon in the fabrication of the first
lattice-matched AlGaAs heterostructures [9, 10].
Then the progress in the semiconductor heterostructure field was very rapid. The
following properties and effects had been experimentally proved: unique injection
properties of a wide-gap emitter and effect of superinjection [11]; stimulated emission
at recombination [12]; the band diagram of the AlGaAs–GaAs heterojunction and
carefully studied luminescence properties [13]; effect of carrier diffusion in a graded-
band heterostructure. At the same time, majority of the most important devices with
realization of the main advantages of the heterostructure concept have been created:
low threshold room temperature lasers [14–17]; high-efficient LEDs [10, 18]; SCs
[19]; bipolar transistors [20]; and p-n–p-n switching devices [21].
At that early stage of the heterostructure physics and technology, it became clear
that new lattice-matched structures are needed in order to cover a wider range of
wavelength spectrum. The first important step was done in the works [22, 23], in
which various lattice-matched heterojunctions based on quaternary III–V solid solu-
tions were proposed for independent variation of the lattice constant and the bandgap.
Soon, InGaAsP compositions were recognized among the most important ones for
many different practical applications: InGaAsP/InP for lasers in the infra-red inter-
vals suitable for fiber optics communications [24], and InGaAsP/GaAs lasers in the
visible region [25]. In the early 1970s, ideal lattice-matched heterostructures were
limited by the mentioned materials only. Later this “world map” of the III–V het-
erostructures was drastically expanded.
Since the first solar-powered satellites Vanguard-1 and Sputnik-3 were launched
in 1958, SCs based on Si had become the main sources of electricity on the space-
crafts. The first space arrays were based on single crystal silicon SCs characterized
by efficiency of about 10%. During the 1960s and 1970s, considerable improve-
ments in the Si cell design and technology were introduced that allowed increase
in the efficiency of up to 18%. This was achieved owing to fabrication of “violet”
cells with increased short-wavelength photosensitivity, formation of so-called back
surface field, application of photolithography to ensure optimal front grid pattern;
reduction of optical losses by front surface texturing and improved antireflection
coating deposition. These advanced Si cells are used till now for such space missions
that do not strictly require III–V SCs with their higher efficiency and better radiation
stability [26, 27].
At the beginning of 1960s, it was found that GaAs-based SCs with the Zn-diffused
p-n junction ensured better temperature stability and higher radiation resistance. One
136 Zh. I. Alferov et al.

of the first scaled applications of the temperature-stable GaAs SCs took place on the
Russian spacecrafts Venera-2 and Venera-3 launched in November 1965 to the “hot”
planet Venus. The area of each GaAs solar array fabricated by the Russian Enterprise
KVANT for these spacecrafts was 2 m2 . Then the Russian moon-cars were launched
in 1970 (Lunokhod-1) and in 1972 (Lunokhod-2) with GaAs 4 m2 solar arrays in each.
The operating temperature of these arrays on the illuminated surface of the moon
was about 130 °C. Therefore, silicon-based SCs could not operate effectively in these
conditions. GaAs solar arrays have shown efficiency of 11% and have provided the
energy supply during the life-time of these moon-cars.
The first AlGaAs/GaAs SCs with passivating wide-bandgap window were created
in 1970 [19]. In the following decades, by means of the liquid-phase epitaxy (LPE)
of AlGaAs/GaAs heterostructures [19–36], their AM0 efficiency was increased up
to 18–19% [34–38] owing to the intensive investigations in the fields of physics
and technology of space SCs [39–42]. These investigations were stimulated and
supported by ambitious space programs in the former USSR [33] and in the USA
[26, 27].
High efficiency and improved radiation hardness of the AlGaAs/GaAs SCs stimu-
lated for the high-scale production of AlGaAs/GaAs space arrays for the spacecrafts
launched in 1970–1980s. For example, an AlGaAs/GaAs solar array with a total area
of 70 m2 was installed in the Russian space station MIR launched in 1986 (Fig. 8.1).
During 15 years on the orbit, the array degradation appeared to be lower than 30%
under trying conditions of its operation, of appreciable shadowing, effects of numer-
ous docking and ambient environment of the station. At that time, it was the most
scaliest demonstration of the advantages of AlGaAs/GaAs SC in space applications.

AlGaAs/GaAs array

Fig. 8.1 Command module of “MIR” space station was launched into orbit on March 13, 1986 with
PV array based on AlGaAs/GaAs SCs developed at Ioffe Institute and fabricated in NPO “Kvant”.
The array has been operating during the 15-year space station span-time
8 III–V Solar Cells and Concentrator Arrays 137

Further improvement of the LPE technology allowed obtaining [43, 44] the efficien-
cies of 24.6% (AM0, 100 suns) on the base of the heterostructures with an ultra-thin
window AlGaAs layer and a back surface field layer.
Since the late 1970s, AlGaAs/GaAs heterostructures were produced also by
the metal organic chemical vapor deposition (MOCVD) technique (later known as
MOVPE—metal-organic vapor phase epitaxy) [45, 46]. The advantage of MOVPE
is a possibility to fabricate multi-layer structures in high-yield reactors with layers of
a specified composition, and its precise thickness varied from 1 to 10 nm to several
microns. AlGaAs/GaAs heterostructures with an ultra-thin (~0.03 µm) top window
layer and with a back surface wide-bandgap barrier were fabricated by MOVPE for
space cells. AlGaAs/GaAs 4 cm2 space SCs with efficiencies of 21 [47] and 21.7%
[48] were fabricated on the base of these structures.
Enhanced light absorption was provided in the cells with an internal Bragg reflec-
tor [49–51]. This dielectric mirror increases the effective absorption length of sunlight
within the long-wavelength part of the photoresponse spectrum and allows making
the base layer thinner. In this case, the cell efficiency is more tolerant to reduction of
the carrier diffusion length; as a result, these cells are more radiation-resistant [51].
Owing to the fact that MOVPE is capable of producing single crystal layers on
silicon and germanium substrates, it has a potential for fabrication of low-cost, high-
efficiency III–V SCs on these substrates. The lattice mismatch of 4% between Si and
GaAs does not allow growing GaAs on Si with a sufficient quality. However, there is
a progress in improving the GaAs/Si structure quality by using special structures and
growth techniques: strained superlattice, thermo cyclic growth and cyclic structure
annealing.
Ge is a quite good lattice matched to GaAs material. Therefore epitaxial growth of
GaAs with high quality was realized by MOVPE, which is now the basic technique
for the growth of multi-layer AlGaAs/GaAs/Ge single-junction and GaInP/GaAs/Ge
multi-junction (MJ) SCs. This method provides a good crystal quality of epitaxial
structures on Ge substrates, high productivity and reproducibility.
Among other single-junction SCs, the InP-based cells are rather promising for the
space application owing to the fact that InP has a higher radiation resistance [27, 41,
52] than GaAs. However, there are some obstacles in the scale application of InP-
based cells in space arrays. First, there is no lattice-matched wide-bandgap window
for InP to make stable passivation of the front surface. Second, it is difficult to grow
this material with high quality on the Ge and Si substrates due to lattice mismatches
as high as 8% between InP and Si and 4% between InP and Ge.
Multi-junction (tandem) SCs ensured further increase of III–V SC efficiencies.
Despite a large number of theoretical studies of tandem SCs [53–56], their efficien-
cies remained low for a long time, since the ohmic and optical losses in available
designs were unacceptably large. Monolithic and mechanically stacked tandem cells
with increased efficiencies were developed in the beginning of 1990s. In mechan-
ically stacked tandems with GaAs-GaSb, GaInP/GaAs-GaSb and similar config-
urations, including spectrum splitting concept [57–62], the efficiencies exceeding
30% were achieved under the concentrated sunlight. Monolithic tandem SCs have
138 Zh. I. Alferov et al.

been developed and fabricated by MOVPE on the structures of GaInP/GaAs [63–

65], GaInP/GaInAs [66], GaInP/Ga(In)As/Ge [67–70], GaInP/GaAs/GaInNAsSb
[71] and other heterostructures [70, 72–74]. The latest highly efficient results were
demonstrated on multi-junction SCs fabricated with inverted metamorphic and wafer-
bonded technologies and with the use of modern III–V/Si and III–V/SiGeSn struc-
tures [75–83].

8.1.1 Single-Junction AlGaAs/GaAs Concentrator Solar

Cells

The creation of AlGaAs/GaAs heterostructure SCs opened up the new possibilities

of increasing the efficiency of solar energy conversion. The idea of wide-bandgap
window was realized for SCs, which allowed protecting the photoactive region of
the cells against the influence of surface electronic states. Defect-free heterojunc-
tions between AlGaAs (wide-gap window) and p-n GaAs (photoactive region) were
successfully formed, which provided ideal conditions for the photogeneration of
electron–hole pairs and their collection at the p-n junction. Since SCs with a GaAs
photoactive region turned out to be even more radiation-resistant, they quickly found
an application in space solar arrays, despite their significantly higher cost compared
to silicon cells.
First, the development of AlGaAs/GaAs SCs was based on relatively simple struc-
tures and technologies. The relatively simple LPE technique was applied. Only one
wide-gap p-AlGaAs layer had to be grown, whereas the p-n junction was formed by
the diffusion of a p-type impurity from the melt into the base material of n-GaAs
(Fig. 8.2a, c). From the middle of the 1980s, high-tech methods began to penetrate the
sphere of semiconductor solar photovoltaics. The progress in the field of GaAs-based
SCs was stimulated by the application of new epitaxial techniques for heterostructure
growth. The main achievement here was metal-organic chemical vapor deposition.
New technologies allowed for improving the SC structure parameters. First, the
wide-gap AlGaAs window was optimized, and its thickness became comparable
with that of the nanosized active regions in heterolasers. The AlGaAs layer also
served as the third component in the triple-layered interference antireflection coating
(ARC) of a cell. Prismatic cover was applied to reduce the shadowing losses on the
top grid finger contacts in concentrator SCs (Fig. 8.3a). A heavily doped GaAs
contact layer was grown on the top of the wide-gap AlGaAs window, and it was
removed during the post-growth treatment in the areas between the contact stripes
(Fig. 8.3b). Second, a back (behind the p-n junction) wide-gap layer was introduced,
which ensured, along with the front wide-gap layer, a double-sided confinement of
photogenerated carriers within the region of light absorption (Figs. 8.2b and 8.3).
The recombination losses of carriers before their collection at the p-n junction were
reduced. At this stage of the optimization of single-junction AlGaAs/GaAs photocell
heterostructures, the modified low-temperature LPE technique was still competing
with the newly developed MOVPE technique.
8 III–V Solar Cells and Concentrator Arrays 139

(a) GaAs AlGaAs (b)

(c) (d)

Fig. 8.2 Energy band diagrams of SCs based on AlGaAs/GaAs heterostructures: a structure with a
p-n-junction in GaAs and with a frontal wide-gap “window” of p-AlGaAs; b structure with a back
potential barrier in the n-region; c structure with a frontal p-AlGaAs layer of variable composition;
d structure with a Bragg mirror

(a) (b)

prismatic cover
p+ GaAs
ARC ZnS ARC
p-AlGaAs:Mg p-AlGaAs 0.05 µm
p-GaAs:Mg(Zn) 1.5 µm p-GaAs 0.5 µm

n-GaAs 1.5 µm
n-GaAs:Te 3.0 µm
n-AlAs/GaAs
p-AlGaAs:Te 1.0 µm 12 period BR

n -GaAs:Sn
n+ GaAs substrate
(Substrate) 400 µm

contact

Fig. 8.3 Heterostructure of the AlGaAs/GaAs-based SCs: a prepared by LPE with thick (3 µm)
base n-GaAs layer, b prepared by MOVPE with internal Bragg reflector and thin (1.5 µm) base
n-GaAs layer
140 Zh. I. Alferov et al.

30
29.3%, 49.9 x
28.8%, 1 x
29 2
1
26.2%, 1000x
26 25%, 2000x
Efficiency, %

25 24.6%, 100x

24
3
23 5
23%, 5800x

22
110 100 1000
Concentration, Suns

Fig. 8.4 Highest efficiencies in the AlGaAs/GaAs single-junction SCs under AM0 (curve 3),
AM1.5 g (dot 1) and AM1.5d (data 2, 4, 5) spectrum irradiation [27, 43, 70, 84, 85]

For these structures the record efficiency of 29.3% for illumination with the con-
centrated (49.9 suns, AM1.5d) sunlight was measured in MOVPE-grown SCs [70]. At
the same time, the record efficiency of 24.6% for single-junction cells at illumination
with 100 × concentration of AM0 sunlight (Fig. 8.4) still belongs to LPE-grown SCs
[27, 43]. Also, the highest efficiencies for high concentration ratios in the range of
1000–2000 suns (AM1.5d) were measured in the LPE-grown AlGaAs/GaAs single-
junction SCs (Fig. 8.4): 26.2% (1000×) and 25.0% (2000×) [84]. These cells can
operate under ultra-high sunlight concentration with efficiency as high as 23% at
5800 suns (AM1.5d) [85].
In MOVPE-grown AlGaAs/GaAs SC structures, a single wide-bandgap AlGaAs
layer, which forms the back potential barrier, can be replaced by a system with pairs of
AlAs/GaAs layers making a Bragg mirror (Figs. 8.3b and 8.5) [51]. The wavelength
of the reflection peak for such a mirror is chosen in the vicinity of the absorption
edge of the photoactive spectrum region, so that the long-wavelength light that was
not absorbed in this region can be absorbed during the second passage after reflection
from the mirror (Fig. 8.2d). At the same time, the wide-gap mirror layers continue to
serve as the back potential barrier for photogenerated carriers. In these conditions,
the thickness of the photoactive region can be reduced by half without loss of current
as compared to the thickness of structures without a mirror (Fig. 8.3b). This factor
led to a significant increase in the radiation resistance of such type of photocells,
because the number of lattice defects generated under irradiation by high-energy
particles decreases proportionally to the thickness of the photoactive region.
More complicated structures with several Bragg reflectors [86] tuned for several
IR bands of solar radiation allow increasing the reflection of the sub-bandgap IR
radiation to reduce SC operation temperature and, as a result, to increase the cell
efficiency.
8 III–V Solar Cells and Concentrator Arrays 141

(a) (b)
pAl0.8Ga0.2As p+ GaAs p-AlGaAs
p+GaAs n-GaAlAs
p-GaAs GaAs AlAs/GaAs
p n Bragg
n-GaAs reflector
n+ GaAs
n-GaAs/AlAs
12 periods

1μ m n-GaAs 0.5 μ m

Fig. 8.5 SEM (a) and STM (b) images of the SC heterostructure with internal Bragg reflector

The effect of the start-up processes on re-reflection and recycling of radiation

can be enforced essentially using thin-film GaAs-based SCs. Application of the new
epitaxial lift-off (ELO) process for fabricating thin-film single-junction GaAs SCs
without light concentration has allowed raising the efficiency of up to 28.8% and
simultaneously registering extremely high open-circuit voltage of 1.122 V [70]. This
voltage value is an indicator of the efficiency of the photon recycling process, at which
photons absorbed in the GaAs photoactive layers can be radiatively re-emitted, and
then again re-absorbed in the same GaAs device. Efficient photon recycling ensures
higher carrier density generated by a SC, because, in the ideal case, photons that
are absorbed in the semiconductor either create an electron–hole pairs that are given
an impact into current of the device, or are radiatively re-emitted in order to be re-
absorbed with new electron–hole pairs generation. Photon recycling therefore results
from the combination of minimizing the loss of photons to processes other than carrier
generation, as well as minimizing the non-radiative recombination of carriers. This
leads to increased carrier density within the device, which in turns leads to increasing
the Voc of the device owing to greater quasi-Fermi-level splitting.
The processes of photon recycling can be started-up also in using the above-
mentioned Bragg reflectors, which opens the possibility of thickening photoactive
layers in MJ SC subcells, thus being a contributory factor in raising their efficiency
and radiation tolerance [51, 87, 88].

8.2 Multi-junction Solar Cells

The idea of tandem solar cells began to be discussed in the early 1960s and was
considered to be promising. However, increasing the efficiency seemed a long way
away. The situation started to be changed in the late 1980s, when many research
142 Zh. I. Alferov et al.

groups concentrated their efforts on developing different types of multi-junction

SCs. The optimal bandgaps for dual and triple-junction SCs are shown in Fig. 8.6.
At the first stage, the best results on efficiency were obtained in mechanically
stacked MJ SC concept started in 1989 by Fraas and Avery with the demonstration
of a 32.6% (AM1.5d, 100×) efficient concentrator GaAs/GaSb mechanically stacked
dual-junction cell [89]. However, everyone understood that the real promising cells
would be those with a monolithic structure.
The first high-efficiency monolithic tandem SCs were fabricated by MOVPE on
GaInAs/InP [90], GaAs/AlGaAs [91]. To prepare GaInAs/InP tandem cells [90],
GaInAs n- and p-layers are grown on an InP substrate to form the bottom subcell.
InP n- and p-layers were grown on the top of this structure to form the top subcell.
The contacts were made according to a three-terminal scheme: one back contact
covering the entire surface and two ridge-like contacts to the illuminated surface.
One of them served to connect the top of the bottom cell and the back of the top cell.
The cell efficiencies were 8.9% for the bottom GaInAs cell and 22.9% for the top
InP cell, the total efficiency being as high as 31.8% AM 1.5 at concentration ratio of
50×.
High-efficiency AlGaAs/GaAs monolithic tandem cells have been made with
bottom GaAs and top Al0.37 Ga0.63 As subcells [91]. An efficiency of 27.6% measured
under AM1.5g spectrum irradiation was achieved in this tandem (top cell: Eg =
1.93 eV). The component cells were electrically connected by a metallic contact
fabricated during the post-growth processing.
Two-terminal monolithic tandem SCs based on AlGaAs/GaAs heterostructures
were fabricated by low-temperature LPE [92]. The top and bottom cells were con-
nected by a tunnel junction formed in GaAs. The heterostructure was grown by a
two-stage LPE procedure. The bottom GaAs-based cell structure was grown at the

2.5 AlP Dual- Triple-

Junction Junction
GaP AlAs SC SC
2.0
Band gap energy, eV

1.5 AlSb
GaAs InP
1.0
Si
GaSb
Ge
0.5
InAs
InSb
0.0
5.4 5.6 5.8 6.0 6.2 6.4 6.6
Lattice constant, A

Fig. 8.6 Band gaps of III–V compounds and their solid solutions versus the lattice constants of
these materials. The optimum E g is shown for the dual- and triple-junction SCs
8 III–V Solar Cells and Concentrator Arrays 143

first stage. Heavily doped p+ -GaAs and n+ -GaAs layers were grown on this struc-
ture to prepare a tunnel junction. Their thickness was chosen as thin as possible
(8–10 nm) to minimize the sunlight absorption losses within a GaAs-based subcell.
The doping levels were as high as 1020 cm−3 when Ge or Te is added to the melt.
This is quite sufficient to form a tunnel junction. The top AlGaAs layer was grown
to protect tunnel diode during the second LPE growth. Reference samples of such
GaAs subcells (without p+ and n+ layers and with thin pAl0.9 Ga0.1 As window layer)
have demonstrated outdoor efficiency exceeding 27% (AM1.5, 100–300 suns). At
the second stage, the top AlGaAs subcell was grown with the photoactive region
made of Alx Ga1−x As (x > 0.3). These cells demonstrated rather high external quan-
tum yield of 80–90% in the spectral range of 650–450 nm. An open-circuit voltage
of 2.53 eV and fill factor of 0.8 at 50 suns were obtained using these tandem cells.
Further developments of monolithic tandem SCs by LPE had not been continued
because the reproducibility of two-stage LPE progress was not ensured.
Researchers from the NREL (National Renewable Energy Laboratory, USA) were
the first to obtain the sufficient efficiency increase in the monolithic dual-junction
SCs [93]. Using the MOVPE technique, they grew GaInP/GaAs structures matched
by their lattice constants, in which the top photocell had a p-n junction in In0.5 Ga0.5 P
and the bottom one was in GaAs. The cells were electrically connected in series by
means of a tunnel p-n junction specially formed between the cascades. Efficiency of
30.2% (AM1.5d, 180×) was obtained in these dual-junction SCs.
At the same time, the interest on triple-junction cells was growing. As a conse-
quence of well-directed efforts, efficiency as high as 35–42% has been demonstrated
in monolithic GaInP/Ga(In)As/Ge SCs [69, 94–97].

8.3 Lattice-Matched GaInP/Ga(In)As/Ge Triple-Junction

Solar Cells

The vast structures for modern high-efficient MJ SCs are fabricated by the metal-
organic vapor phase epitaxy (MOVPE technique) on highly productive commercial
reactors. Ge of p-type is used as an active substrate, on which a lattice-matched (LM)
triple-junction (3J) heterostructure—Ga0.50 In0.50 P/Ga0.99 In0.01 As/Ge (widely used
abbreviation is GaInP/GaAs/Ge)—of high crystalline perfectness is formed. Emitters
of n-type Ge are obtained with the help of diffusion processes. Series connections of
junctions are provided by means of low-resistance and optically transparent tunnel
diodes. Each layer in the 3J GaInP/GaAs/Ge LM heterostructure brought to perfection
(see Fig. 8.7) is optimized by composition, doping level and thickness [98].
Top and middle subcells include the following layers: back-surface field (BSF)
layer, base, spacer, emitter and window. The Ge subcell consists of a base (substrate),
a diffused emitter and a window. Subcells are connected in series by tunnel diodes,
which in turn include highly doped thin (10–20 nm) layers.
144 Zh. I. Alferov et al.

Fig. 8.7 Typical Top contact

GaInP/Ga(In)As/Ge n+ -
structure of a monolithic Cap layer Ga(In)As ARC
triple-junction SC on p-Ge window n-AlGaInP
emitter n-GaInP
substrate i-GaInP
Top cell spacer
base p-GaInP
BSF p-AlGaInP
p++ - AlGaAs
Tunnel diode
n++ - GaInP
window n-GaInP
emitter n-Ga(In)As
Middle cell spacer i-Ga(In)As
base p-Ga(In)As
BSF p-GaInP
p++ - GaAs
Tunnel diode
n++ - GaAs
window n-GaInP
emitter n-Ge
Bottom cell base
p-Ge substrate

Bottom contact

However, the combination of forbidden gaps for subcells in the GaInP/GaInAs/Ge

LM heterostructure is not the optimal one: the bottom narrowband Ge p-n junction
generates substantially greater current than GaInP and GaAs subcells. A part of
sunlight energy corresponding to the “excess” photons absorbed in Ge is lost without
the use, thus limiting the MJ SC efficiency. For this reason, a substantial increase of
the 3J SC with a Ge substrate should not be expected. The investigations being carried
out on raising the MJ SC efficiency are associated with the increase in the number
of p-n junctions based on materials with optimum combinations of Eg values.
Estimations show that the 3J SC efficiency can be raised substantially by replacing
the narrowband Ge subcell with wider band one based on the GaInAs solid solution
isolattice to GaAs with Eg in the range of 0.7–0.5 eV with a prospect to approach
to Eg ~1 eV [71]. Further increase in the SC efficiency is ensured by introducing
additional narrowband junctions isolattice to GaAs with Eg in the range of 0.7–0.5 eV
to the structure.
Since the technology of the wideband tandem GaInP/GaInAs (1.86/1.41 eV) is
well developed, it is, as a rule, the basis for designing 3–4J SCs. Efforts are mainly
directed to develop narrowband subcells, that is, to search for materials with the
forbidden gap smaller than that of GaAs (Eg = 1.41 eV).
For this reason, researchers try to replace active Ge (at a p-n junction) with more
effective, strong and light Si [99–101]. However, owing to great difference between
lattice constants and between coefficients of thermal expansion (CTE), it was not
possible up to now to obtain good quality GaAs/Si layers by direct epitaxial growth
[101]. Good results have been demonstrated for the technology of direct bonding of
III–V heterostructures with Si plates [100–103]. Impressive achievements have been
8 III–V Solar Cells and Concentrator Arrays 145

obtained by inverted growth of a monolithic 4J architecture with two metamorphic

layers of GaInAs with Eg equal to 1 and 0.7 eV [104]. Break-through results on
obtaining device quality of a new class of GaInAsNSb were obtained by the MBE
technique [71, 105]. In these compounds, the forbidden gap width varied depending
on the N portion from 1.3 to 0.7 eV at complete matching with the GaAs or Ge
lattice. This opens up for the possibilities of fabricating 4, 5 and even 6J SCs with
a record efficiency. Besides, at present, for narrowband subcells, 3J Si-Ge-Sn solid
solutions of the IV group are developed. Combination of these three elements allows
obtaining compounds lattice matched with GaAs and Ge with a forbidden gap from
0.5 to 1.2 eV [82].

8.4 Lattice-Mismatched (Metamorphic) Heterostructures

for Multi-junction Solar Cells

In developing modern high-effective MJ SCs, the following main methods and tech-
nologies are applied:
Upright metamorphic growth (UMM). It is monolithic growth of several crys-
tals with different lattices, as a rule, in the direction from narrowband subcells to
wideband ones. For acceptable crystallographic quality of photoactive layers, inter-
mediate compensating (buffer) layers with gradually varying lattice constant are
usually applied [67, 68].
Inverted metamorphic growth (IMM). At IMM, the MJ SC structure is formed
starting from the wideband, isolattice subcell to the substrate of a MJ SC with fol-
lowing going to the layers of narrowband metamorphic materials [80, 104, 106–
108]. Advantage of the mentioned approach compared to the “direct” structure is
the transfer of the growth of metamorphic (not lattice matched) layers to later stages
of epitaxial process, what allows creating perfect wideband subcells. One can select
an optimal forbidden gap for narrowband subcells, since they are not associated
with grown substrate. Then the structure is separated from the substrate (technology
ELO—epitaxial lift-off) and transferred to a new carrier.
Separation of MJ SC thin layers from a massive isolattice substrate is performed
by selective etching in hydrofluoric acid along the “sacrificial” layer (usually the
AlGaAs layer) specially formed during structure growth. Selectivity of etching
by “sacrificial” layer with respect to the adjacent GaAs layers reaches 106 . After
separation, the epitaxial film is transferred to a new supporting substrate, which can
be thin, flexible, superlight and with better heat conduction than the growth one [109].
The initial substrates are re-usable, which reduces the structure cost substantially.
Owing to super-small structure thickness, the ratio of power/mass of standard and
flexible solar arrays has improved significantly.
Wafer bonding technology for MJ SCs is connection of heterostructures grown
separately on different substrates. This opens up the wide possibilities in choosing
146 Zh. I. Alferov et al.

the substrates for growing and avoids the use of metamorphic buffer layers that com-
pensate variation in the lattice constant along the direction of growth of buffer layers.
Usually one of the structures is obtained by the IMM growth. Then the plates are
pressed followed by removal of one of the substrates by the ELO technique. The given
process of creating MJ SCs is rather complicated and requires high technological cul-
ture [102, 103, 110–113]. The highest efficiencies for MJ SCs fabricated by the com-
bined technology “IMM-bolding” are: 46.1% for 4J GaInP/GaAs//GaInAsP/GaInAs
SC (C = 312, AM1.5d) [76]; 30.2% for 3J GaInP/AlGaAs//Si SC (C = 1, AM1.5g)
[78]; 35.8% (C = 1, AM0) and 38.8% (C = 1, AM1.5d) for 5J SC [75].
The problem in the crystal quality arises during the development of growth pro-
cedure for such metamorphic SCs. It should be noted that GaInAs with near to ideal
bandgap of 1.1 eV has the lattice mismatch of 1.6% with the Ge substrate. In spite
of the higher theoretical efficiency of such metamorphic cells, the achieved cell effi-
ciencies are approximately equal to those for lattice-matched cells. Nevertheless, the
highest efficiency (>40%) has been reached by Spectrolab, Inc. in the metamorphic
Ga0.44 In0.56 P/Ga0.92 In0.08 As/Ge 3-junction concentrator SCs [69].
The following statement may be formulated on the basis of the presented data: het-
erostructures of high-efficiency multi-junction SCs are the most complicated struc-
tures among all semiconductor devices. The confirmation of this declaration is as
follows:
– Full range of the III–V materials in view of the binaries and solid alloys is involved
in the formation of the MJ SC structure (including nitrides for advanced cells). A
very wide range of the layer bandgaps covering photon spectrum from UV to IR
should be ensured. All the layers are lattice matched, or moderately metamorphic.
– Triple-junction SC consists of about 20 layers and the quantity of layers will be
higher or more than in the advanced 4÷6-junction cells, or in the cells with Bragg
reflectors, as well as in the cells incorporating superlattices and quantum dots.
Layer thickness varied in a very wide range from 10 nm in the tunnel diodes up
to a few micrometers in the photoactive regions.
– Doping level varied from 1015 to 1016 cm−3 in the spacer layers up to 1019 –1020
cm−3 in commutating tunnel diodes. A drastic change in the doping level should
be ensured in the tunnel diodes and this sharpness should be conserved in the
structure during further crystallization at enough high growth temperatures. Also,
the operating capacity of the tunnel junctions should be ensured during all the
life-time of outdoor SC operation at various temperatures and high photocurrent
densities.
– III–V SC structures are grown on Ge substrate, being the foreign material in relation
to the cell one. A number of additional technological problems are expected, if Ge
substrate is substituted to Si one, which is more promising from economical point
of view.
– Individual cell dimensions varied from 1 mm2 in the terrestrial concentrator mod-
ules with mini-lenses up to about 30 cm2 in arrays without concentration for space
application. It means that extremely high quality of growth and post-growth tech-
nologies must be ensured.
8 III–V Solar Cells and Concentrator Arrays 147

– SCs are among such semiconductor devices that operate under inclement climatic
conditions. A thin glass sheet or dielectric layer is the only barrier between envi-
ronment and the cell structure.
– In concentrator SCs, optically transparent protectors and front layers are subjected
to action in a highly intensive light (up to 1000 suns and higher in the center of
the focal spot), so that a radiation resistance of corresponding materials should be
ensured.
– Conflicting demands to the cell structure are accompanied by a requirement to
provide the lowest cell cost as an essential issue in the balance-of-system cost of
solar electricity.

8.5 Multi-junction Solar Cells: Current Status

of High-Efficiency Data

In spite of the complexity of MJ SC structures and technology, a number of insti-

tutes and companies are continuing their MJ cells development for PV applications,
including concentrator approach, because it opens the perspectives for significant
increase in efficiency and reduction in the solar electricity cost.
It is clear that SCs are the key elements in the concentrator PV installations.
Table 8.1 summarizes the efficiencies of terrestrial single- and multi-junction (from
2 to 4 junctions) III–V concentrator SCs demonstrated in this century (after the year
2000). Cell producers, sunlight concentration ratios (suns), conditions at measure-
ments (air mass) and the cell structure (type of SCs) are shown in the table. The cells in
the upper part of the table are characterized by a rather high efficiency. Involving the
high-power companies (Spectrolab, AZUR Space, EMCORE Photovoltaics, Sharp,
SolAero Technologies Corp., etc.) into “terrestrial” concentrator activity allowed
organizing a high-scale production of MJ concentrator SCs for concentrator mod-
ules [114–116].
MJ SCs are characterized by a complicated structure. In a triple-junction SC, three
photoactive p-n junctions are connected in series by two intermediate n+ -p+ tunnel
junctions, biased by voltage in the forward direction. Very high doping concentra-
tions of the tunnel junction regions require small thicknesses of the corresponding
layers in the structure to reduce the non-photoactive light absorption. In addition, two
enclosing barrier layers with the purpose to minimize dopant diffusion from the n+
and p+ layers in both tunnel junctions should be introduced in a multi-junction SC
structure. Often, technological difficulties associated with the formation of tunnel
junctions are the reason for lower conversion efficiency of a cell. This is especially
true in the case of cells intended for operation under high sun concentration condi-
tions.
If the generated photocurrent density in a SC exceeds the density of the peak
tunnel current (J p ) of one or both tunnel junctions, in this case there exists a part of
the I–V curve near the open circuit voltage point, where a trace from the action of
Table 8.1 Selected efficiencies of the concentrator solar cells based on III–V compounds
148

Year Producer (test lab) Eff (%) Suns Air mass Type of solar cell Ref.
2001 FhG-ISE 30.2 300 AM1.5d GaInP/GaInAs 2J monolithic [117]
2001 IOFFE/IES-UPM 26.2 1000 AM1.5d AlGaAs/GaAs single-junction [84]
25 2000
2001 JX Crystals 34 15 AM0 GaInP/GaAs-GaSb 3J hybrid circuit [118]
2005 Fraunhofer ISE 35.2 600 AM1.5d GaInP/GaInAs/Ge, monolithic [119]
RWE-SSP
2006 IOFFE 35 50 AM1.5 low-AOD GaInP/GaAs/GaSb 3J hybrid [59]
monolithic/mech.stacked
2006 Toyota TI 38.9 489 AM1.5g GaInP/GaInAs/Ge, 3J monolithic [97]
Sharp Corp.
2006 Spectrolab 39.3 179 AM1.5 low-AOD GaInP/GaInAs/Ge, 3J monolithic [117]
2006 Spectrolab 40.7 240 AM1.5 low-AOD GaInP/GaInAs/Ge, 3J monolithic [69]
2009 Spectrolab (NREL) 41.6 364 AM1.5d GaInP-Ga(In)As-Ge 3J, lattice matched, [69]
commercially available
2009 FhG-ISE (FhG-ISE) 41.1 454 AM1.5d GaInP-GaInAs-Ge 3J, upright metamorphic; [68]
commercially available from AZUR SPACE,
Spectrolab
2011 Spire (NREL) 42.3 406 AM1.5d GaInP-GaAs-wafer-GaInAs 3J, epi growth lattice [72]
matched on front and inverted metamorphic on
back of GaAs wafer
2012 Alta Devices (NREL) 28.8 1 AM1.5 g GaAs single-junction [120]
2012 Solar Junction (NREL) 43.5 925 AM1.5d GaInP/GaAs/GaInNAs 3J, MBE, lattice matched, [71]
dilute nitrides, grown on GaAs
(continued)
Zh. I. Alferov et al.
Table 8.1 (continued)
Year Producer (test lab) Eff (%) Suns Air mass Type of solar cell Ref.
2013 Sharp 43.5 306 AM1.5d GaInP/GaAs/GaInAs 3 J, inverted metamorphic [121]
(FhG-ISE)
2013 NREL (NREL) 42.6 327 AM1.5d GaInP/Ga(In)As/GaInAs 3 J, inverted [122]
Emcore (NREL) 42.4 325 metamorphic [80]
2014 Fraunhofer ISE/Soitec/CEA (AIST) 46 508 AM1.5d GaInP/GaAs//GaInAsP/GaInAs 4 J, wafer [123]
bonding, lattice matched grown on GaAs and InP
2014 NREL (NREL) 45.7 234 AM1.5d GaInP/GaAs/GaInAs/GaInAs 4J, inverted [124]
metamorphic
2017 NREL (NREL) 35.9 1 AM1.5 g GaInP/GaAs//Si mechanically stacked [125]
8 III–V Solar Cells and Concentrator Arrays
149
150 Zh. I. Alferov et al.

Fig. 8.8 A family of the 5

illuminated I–V curves at 14153x
T=23°C
different light intensities
from the high power flash 4
solar simulator for 3J SC
9861x
(designated illumination area

Current, A
3
is 2 mm in diameter)

6217x
2

C=2831x
1

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Voltage, V

the internal tunnel diode(s) can be revealed. The J p value is an important parameter
of a SC, establishing the limit in the photocurrent density for efficient operation. In
Fig. 8.8, a family of the illuminated I–V curves for a 3J SC is shown. It was measured
at different light intensities from a high power flash solar simulator [126].
A tunnel peak is revealed at sun concentration beginning from ~9800×. Resolution
of the tunnel peaks is difficult, when I–V curves are strongly deformed due to obvious
action of the cell internal resistance. The value of J p as high as 65 A·cm−2 has been
calculated from I–V curve for C = 9861×, when considering only the designated
illumination area.
The main result of Fig. 8.8 is that 3J SCs can operate at illumination intensities
as high as several thousand suns. Tunnel peaks have a tendency to move toward
the open-circuit voltage point, when illumination increases. At the same time, peak
current decreases. Probably, it is caused by overlapping of the tunnel curves with a
part of the curve related to the photoactive junctions being more vertical near the
open-circuit point at higher photocurrents. Other reasons may be regarded as well.
Among them is a parasitic photoactivity of the tunnel junctions, where the generated
photocurrent reduces peak current. Another reason can be associated with the carrier
transport through the cell structure, where the thickness of all the layers is smaller
than the diffusion length of the photogenerated carriers. Further investigations could
make clear this phenomenon.
Ohmic losses reduction obtained by lowering both series and sheet resistances
ensures a high fill factor value exceeding 0.87 at concentration ratios up to 1000 suns
and FF = 0.8 at 4000 suns (Fig. 8.9). Efficiencies of about 37% at 1000×, 36% at
2500 × and 34.5% at 4000 × have been measured in one of the Spectrolab’s cells
available in the market.
8 III–V Solar Cells and Concentrator Arrays 151

95
35

Efficiency, %
90 Eff
30
FF, % 85 FF
25

80 20

75 15

3.0
Voc, V

2.8 Voc
2.6
2.4
2.2
1 10 100 1000
Concentration, X

Fig. 8.9 Efficiency (Eff), fill factor (FF) and open-circuit voltage (V oc ) versus sunlight concen-
tration ratio (AM1.5d low AOD) for a GaInP/GaInAs/Ge triple-junction SC developed in IOFFE
Institute (total area is 4 mm2 and designated illumination area is 1.7 mm in diameter)

8.6 Concentrator PV Modules and Installations with III–V

Solar Cells

The concentrator approach is the only way for the large-scale use of high-efficiency
III–V SCs for terrestrial applications. Indeed, optical elements made of relatively
cheap materials can focus the sunlight on small-in-area cells, which allows reducing
drastically the consumption of semiconductor materials in production of solar arrays.
The costs for optical elements and mechanical sun tracking systems are covered at
the expense of larger amount of “solar” electricity produced in higher-in-efficiency
cells and owing to normal positioning of the PV modules with respect to sunrays
during the daytime. For more than 30 years, many research groups were engaged in
the development of concentrator PV systems. Under the development were the SC
structure and fabrication technology, effective concentrator optical systems, module
design and sun trackers [28, 40, 127, 128]. Interest in concentrator PV systems
grew substantially after promoting in practice the higher-in-efficiency MJ SCs which
proved promising in achieving photovoltaic conversion efficiencies of nearly 35–40%
at high-level sunlight concentration.
In the first concentrator modules and installations, large-area mirrors (0.5–1 m
in diameter) focused the sunlight on cells of several square centimeters in area.
Cooling by water or by means of thermal pipes was necessary [127–130] (Fig. 8.10a).
Appearance of the technology accessible for Fresnel lens (FL) fabrication determined
the revision of the photovoltaic module design. SCs could be placed behind the
concentrators in this case. The module housing could serve as a protector from the
environment (Fig. 8.10b). Since the FLs had smaller dimensions (25 × 25 cm2 ), the
152 Zh. I. Alferov et al.

Fig. 8.10 Solar PV installations with AlGaAs/GaAs SCs and with: a parabolic mirrors, 1981 [130].
In photo: Zh. Alferov (on the right), V. Rumyantsev (in the center) and V.Tuchkevich (on the right);
b Fresnel lenses (25 × 25 cm2 each) 1986 [131]

photocell dimensions were also decreased down to less than 1 cm2 . Characteristics of
such photocells were improved owing to lower internal ohmic losses and simplified
assembly. For cooling the cells, it was suffice to use heat conductivity of the module
metallic housing with walls of reasonable thickness [131].
In the late 1980s, the concept of radical decrease in the concentrator dimensions
on retention of a high sunlight concentration ratio has been proposed [132, 133].
The first experimental modules of such type consisted of a panel of lenses, each
of 10 × 10 or 25 × 25 mm2 , focusing radiation on the AlGaAs/GaAs cells of 1–
3 mm in size (see Fig. 8.11). At that time, the main advantages of a module with
small-aperture area concentrators were formulated: the requirements are essentially
relieved imposed on the capability of heat sinking material to conduct heat, on its
thermal expansion coefficient and on its thickness. The focal distance of such lenses
appears to be comparable with the structural thickness of the conventional modules
without concentrators.

(a) (b)

1 cm

Fig. 8.11 PV modules employing small-size aperture concentrators and AlGaAs/GaAs cells [132,
133]: a glass smooth-surface lenses; b plastic Fresnel lenses (Flight experiment on communication
satellite Molniya-3K in 2001)
8 III–V Solar Cells and Concentrator Arrays 153

The advantages of small-size concentrator cells are the following: low ohmic
losses at collection of low current in the conditions of non-uniform light intensity
distribution at high local photocurrent density; high cell chip throughput from a
wafer; possibility to apply the high-productive mounting methods. Somewhat later,
this approach resulted in creation of the “all-glass” photovoltaic modules with III–V
cells and panels of small-aperture area FLs (each lens is 4 × 4 cm2 ). The lens panels
had a “glass-silicone” composite structure. This work was carried out owing to close
cooperation between the research teams at the Ioffe Institute (St. Petersburg, Russia)
and Fraunhofer Institute for Solar Energy Systems (Freiburg, Germany) [134–137].
In recent years, the team of the Ioffe Institute has developed the concentrator
modules in which FLs are arranged on a common superstrate with a panel of 12
× 12 lenses. The cells as small as 2 × 2 mm2 with designated area of 1.7 mm in
diameter operating at mean concentration ratio of 500–700 × are used (see Fig. 8.12)
[138–147].
In the case of the use of the monolithic MJ cells, the following fundamental
advantage should be noted. They operate at lower photocurrent, so that ohmic losses,
say, in a 5-junction cell, would be a factor of 5 low of the losses in a one-junction
cell operating within the same spectral range. Non-uniformity in the illumination
distribution along the cell surface is a negative feature of concentration. For MJ
cells, an important factor is chromatic aberrations in concentrators of the refractive
type. Negative influence of this type of non-uniformity in the illumination cannot be
compensated by the use of a more dense contact grid, because lateral currents arise
between subcells inside the cell structure. Small-size cells have certain advantages

Fig. 8.12 PV tracker equipped with Fresnel lens concentrator modules

154 Zh. I. Alferov et al.

in this respect. On the other hand, parameters of a concentrator as an element of a

PV system should be optimized in the required way.

8.6.1 Design of Fresnel Lens Sunlight Concentrators

Different lens as sunlight concentrators finds its expanding application in photo-

voltaic installations. The main disadvantage of lenses as concentrators is conditioned
with the chromatic aberrations (CA), which, at the wide range of the sunlight spec-
trum and finite angular dimension of the sun, leads to “blurring” of the concentrated
radiation in the plane of the photoreceiver location and to the considerable decrease in
the average level of the irradiance due to the spatial and spectral energy redistribution.
This negative effect can be essentially diminished by a correct choice of the
optimum combination of the lens dimension, its focal distance and optical parameters
determining the refracting surface profile. The optimization criterion is the maximum
of the average sunlight concentration corresponding to the focal spot minimum size.
The optical parameter controlling the profile of the lens at its pre-assigned dimensions
and focal distance is the refractive index, the values of which are taken from its
precisely determined dependence on wavelength for a chosen lens material. Any
optimization procedure may be used for solving the problem.
Such is suggested general approach for designing lens concentrators, which will
be illustrated by an example of choosing design parameters of flat-plane FLs, being
among the most promising ones for solar photovoltaic installations [148–152]. All
calculations could be performed using two concentration process models based on
the geometrical optics and photometry (power optics) concepts [153, 154].
To account for the physical essence of the proposed approach, it is appropriate
to consider the problem in the geometrical optics approximation. It is known [155]
that, when there is no CA, the concentration ratio of an ideal lens C max
g is determined
by its opening angle (or by the ratio of the diameter 2r l to the focal distance f ) and
by the angular diameter of the sun 2ϕ s = 32 . At the optimum ratio r l /f = 1, the
C max
g value is 11,500 and the focal spot radius is approximately 107 times smaller
than the lens radius. The presence of CA, which stems from the dependence of the
refractive index of the lens material on the radiation wavelength, n = n(λ), leads to
a significant increase in the focal spot radius r s and, as a result, to the considerable
decrease of the geometrical concentration C g = (r l /r s )2 .
Figure 8.13 shows schematically how a sunlight beam passing through the terminal
tooth of a lens decomposes due to light dispersion into several monochromatic beams
(each with its own wavelength) with an angular divergence of 2ϕ s . The central ray
of a beam with the wavelength corresponding to the refractive index ncalc chosen
for calculations of the FL profile strikes the center F 0 of the focal spot. Beams with
wavelengths other than that corresponding to ncalc will cross the optical axis above
(UV radiation) or below (IR radiation) the focal plane, thus forming additional foci
(F UV and F IR , respectively) on the optical axis. In this case, the spot radius in the
focal plane will be determined by its intersection with the extreme ray of the beam
8 III–V Solar Cells and Concentrator Arrays 155

Fig. 8.13 Schematic diagram of sunlight beam dispersion on the terminal tooth of a FL

with the wavelength at the corresponding end of the radiation spectrum in the UV or
IR region.
Following the geometrical concept presented in [156], the dependences of max-
imal values of geometrical concentration ratio and minimal focal spot radii on ncalc
could be found. It can be seen (Fig. 8.14) to what extent the maximum concentration
is sensitive to the accuracy of ncalc selection. Thus, the influence of CA can essentially
opt
be lowered down by choosing the correct combination of n calc and f opt and the lens
profile corresponding to these parameters and hence the geometrical concentration
can be increased by 1.5–2 times.
The main lens characteristic is the distribution of the concentrated radiation den-
sity in the focal plane. This so-called optical-power characteristic (OPC) allows
estimating comprehensively the lens quality—its concentrating capability and opti-
cal efficiency. From the calculation results or OPC measurements, one can determine
the average radiation concentration in a spot of any diameter (i.e., for a SC of any
size) and the corresponding optical efficiency of the concentrator–receiver system.
Besides, the OPC gives a possibility to estimate the character of the irradiance dis-
tribution in the focal spot and its effect on the SC I–V curve.
In the photovoltaic model of the concentration process developed by us before
[154, 156], the parameter controlling profile of the lens at its pre-assigned dimen-
sions and focal distance is the refraction index, values of which are taken from its
precisely determined dependence on the wavelength for a chosen lens material, and
156 Zh. I. Alferov et al.

240 2.2

max
220

, mm
Maximum geometric concentration Cg
2.0
200

min
Minimum focal spot radius rs
180
1.8

160

1.6
140

120
1.4
100
opt
n calc
80 1.2
1.496 1.500 1.504 1.508 1.512 1.516 1.520 1.524 1.528
Refractive index ncalc

Fig. 8.14 Maximum geometrical concentration C max

g and minimum focal spot radius r min
s versus
refractive index ncalc

the optimization parameter is the maximum of the average sunlight concentration

corresponding to the focal spot minimum size. The basis for calculation is the OPC
of a lens, which characterize the concentrated radiation density distribution in its
focal plane. In determining the average values of the concentration ratio and optical
efficiency, not the entire focal spot area is considered, but the circle area, in which
95–98% of the concentrated radiation power is collected. This allows obtaining a
higher average concentration on a receiver without essential losses of the power of
the radiation passed through a lens.
A preliminary analysis of different type of optical materials for manufacturing
FLs has shown that an increase in the refraction index affects positively the lens
concentrating capability. For the 40 × 40 mm2 lenses, selection of the optimum
profile parameters for the terrestrial solar spectrum in the range of 320–920 nm was
done. FL profile designing was carried out for two practically available materials:
silicon compound (Wacker 604) and urethane polymer (Fig. 8.15). It has been found
that by choosing the material with greater refraction index, the average radiation
concentration in the spot increased by 1.5 times, and, in the ideal case, reaches 780×
(Fig. 8.16). In this case, 95% of the power is concentrated within the spot of 1.5 mm in
diameter, whereas for a FL of silicon compound the spot diameter was approximately
1.9 mm.
8 III–V Solar Cells and Concentrator Arrays 157

Fig. 8.15 Optical materials 1.56

refraction index curves
1.54

1.52

Refractive index
1.50
Urethane
1.48

1.46

1.44
Wacker
1.42

1.40

400 600 800 1000 1200

Wavelength, nm

850 2.1

800 2.0
Average concentration ratio, X

780 X
dmin=1.88 mm
Focal spot diameter (d), mm

750 1.9
Urethane
Wacker
700 1.8

650 1.7

600 1.6
dmin=1.5 mm
550 1.5 Urethane
510 X Wacker
500 1.4

450 1.3

400 1.2
60 65 70 75 80 85 90 95 100 65 70 75 80 85 90 95 100
Focal distance, mm Focal distance, mm

Fig. 8.16 Maximum average concentration ratio (left plot) and minimum focal spot diameter (right
plot) versus focal distance for 40 mm × 40 mm Fresnel lenses made of Wacker 604 and urethane

8.7 Module Efficiency Improvement

8.7.1 Compensation of Chromatic Aberration Negative Effect

As discussed earlier, the main disadvantage of FL as a concentrator is conditioned

by the chromatic aberration. It is obvious that the actual problem is of lowering the
negative effect of the CA on the power efficiency of the system “FL-SC” [157–160].
This problem can be solved, both by a purposeful change of the concentrated light
characteristics by means of varying the FL profile [153, 154, 156] and matching these
158 Zh. I. Alferov et al.

characteristics with those of a SC [160, 161] and by correct choice of the mutual
location of a FL and a SC [151, 162, 163].
In previous section, a procedure for diminishing the negative effect of the CA by
optimizing the FL profile has been proposed with the aim of achieving the maximum
level of average sunlight concentration on a SC at a sufficiently high value of the
“FL-SC” system optical efficiency [162]. However, it was obvious that the proposed
procedure is best suited for the systems based on non-selective photoreceivers, which
are insensitive to spectrum content variations on their photosensitive surface. In the
case of systems based on MJ SCs, an additional study is required to analyze the
feasibility of the procedure application, since the effect of CA leads to differences in
spatial and spectral irradiance distributions on the separate p-n junctions of a MJ SC,
and if the lens is preliminary assigned or designed by allowing for the requirements
pointed out above, it is not clear at what distance between FL and SC the maximum
power efficiency can be achieved.
In [151, 162, 163] on investigating the behavior of the 3J SC current–voltage
characteristic parameters, it was experimentally shown that there exists for a SC such
a position on the FL optical axis at which the “FL-SC” pair maximum efficiency can
be achieved.
However, in some practical cases the optimum position of a SC with respect to
a lens could be found on the basis of preliminary investigations of lens optical-
power characteristics and their analysis accounted for the 3J SC spectral response.
Herein, the optimal position of the SC with respect to the lens has been found by
direct laboratory experiment, leaving open the question of the correspondence of the
detected lens-to-cell distance to the designed focal one. The theoretical study results
have been verified experimentally under natural sun [164].
Consider, as an example, a silicone-on-glass lens (Wacker 604) of size 40 mm ×
40 mm. The refractive profile parameters have been chosen for the terrestrial solar
spectrum of 320–920 nm, which corresponds to the spectral response range for the
top and middle junctions of the GaInP/GaAs/Ge 3J SC, since just these junctions
are mainly contributing to the photocurrent of a SC and determine its efficiency.
The FL focal distance, at which the minimum focal size and the maximum average
concentration were ensured, was of 85 mm (see Fig. 8.16). Values of the tilt angles
and tooth highs of the designed optimal profile were used in the manufacture of
negatively profiled mold by means of the diamond cutting.
For designed lens an optical-power characteristic (OPC) has been rated within
the three wavelength ranges: 340–680 nm (blue light), 600–920 nm (red light) and
880–1840 nm (infrared light) corresponding to the spectral responses of separate
junctions of the 3J SC at different distances from the lens (Fig. 8.17a).
However, on the basis of these characteristics, it is impossible to estimate the
efficiency of converting the radiation concentrated by a lens into the photocurrent
by the separate junctions and to determine the cell optimal position on the lens
optical axis. To do this, it is necessary to normalize OPC with consideration for the
spectral response of individual subcells of 3J SC. Figure 8.17b presents distributions
of densities of photocurrents generated by p-n junctions at different distances of a
SC from a lens.
8 III–V Solar Cells and Concentrator Arrays 159

(a) 700
-1 mm -0.5 mm 0 mm +1 mm +2 mm
600
2
Power density, mW/mm
500

400

300

200

100

0
-1 0 1 -1 0 1 -1 0 1 -1 0 1 -1 0 1
Spot radius, mm

(b)
300 -1 mm -0.5 mm 0 mm +1 mm +2 mm
2
Photocurrent density, mA/mm

250

200

150

100

50

0
-1 0 1 -1 0 1 -1 0 1 -1 0 1 -1 0 1
Spot radius, mm

Fig. 8.17 Rated optical-power characteristics a for three wavelength ranges (340–680 nm—blue
solid lines, 600–920 nm—red dash lines and 880–1840 nm—wine short dash lines) at different
distances from a lens (designing focal distance is 85 mm, defocusing is equal to 0) and corresponding
distributions of photocurrent densities generated by p-n junctions of 3J SC (b) (GaInP—blue solid
lines, GaAs—red dash lines, Ge—wine short dash lines)

Recalculation of the power distributions into the “current” ones using data on
the cell spectral response shows that, going from the top p-n junction to the bottom
one, the areas under the photocurrent density distribution curves and also the values
of the maximum density of the photocurrent generated in the SC center rise due
to differences in the spectral responses and, hence, in the efficiencies of converting
solar radiation into current.
160 Zh. I. Alferov et al.

Since, in converting the concentrated radiation, an essential part of power losses

in a 3J SC is conditioned by the non-uniform distribution of the photocurrent density
along photoactive layers of separate junctions and by the difference in the values
of the photocurrent generated by the junctions (the latter is due to smearing out the
light spots caused by the CA), in determining the 3J SC optimum position on the lens
optical axis, the distribution of the generated photocurrent density in mA/mm2 should
be considered instead of the distribution of the radiation flux density in mW/mm2 .
Analysis of the obtained dependences from the point of view of their effect on the
“FL-SC” system power characteristics indicates the following. A drastic luminance
differential from the periphery to the center of the focal spot leads to a high local
density of the photocurrent generated by a p-n junctions in the cell center, which,
depending on the position of the receiver with respect to the lens, results in rising
power losses on the internal resistance of the top junction, when “blue” light is
strongly focused; the same for the middle junction and “red” light and the bottom
junction and “infrared” light.
Figure 8.18 presents the dependencies of densities of the local photocurrents
generated in the cell centers on the cell–lens distance. The dependencies show that
the strongest negative effect on the SC fill factor is from the GaInP cell side, when
the 3J SC is closer to the lens by 0.5 mm than the rated focal distance (85 mm),
but from the GaAs and Ge cells, when the SCs are farther by 1.1 mm and 1.5 mm,
respectively, since the local photocurrent generation in the center of each cell at these
3J SC positions appears to be maximum.
However, for the 3J SC, not only the high level of the locally generated photocur-
rent can have a negative effect on the FF. Integrating the dependencies presented in
Fig. 8.17b allows obtaining values of the photocurrents generated by the junctions
at different distances of a SC from a FL, which are shown in Fig. 8.19 for a SC with
the photosensitive area of 2.3 mm in diameter. Practically for the whole range of the
defocusing (from −1 to +2 mm), the “FL-SC” system will operate in the mode of
current limiting by the top GaInP junction. Only in the case of very strong shift of

Fig. 8.18 Densities of local

2

300
Local photocurrent density, mA/mm

photocurrents generated in
the center of subcells as a 270
function of the distance 240
between 3J cell and FL
210
Ge
180

150 GaAs
GaInP
120

90

60
-3 -2 -1 0 1 2 3 4 5 6
Defocused position, mm
8 III–V Solar Cells and Concentrator Arrays 161

Fig. 8.19 Photocurrents of 300

subcells versus cell–lens Ge
distance 280

Photocurrent, mA
260

240

220

200 GaAs

180 GaInP

160
-2 -1 0 1 2 3 4
Defocused position, mm

a SC toward a lens (defocusing is – 2 mm), which, apparently, is not optimal one

from the point of view of red and infrared light focusing, the current limitation by the
middle GaAs junction arises. In the –0.8 mm position, the currents of the GaInP and
GaAs junctions coincide, and, hence, the SC FF should be minimal at this position.
Besides, the Ge junction current is also very close to the GaInP and GaAs cells’
currents, which will also increase the FF drop in the defocusing range from –1.5 to
–0.5 mm.
But, as shown earlier (see Fig. 8.18), when the 3J SC is by 0.5 mm closer to the
lens, the negative effect on the SC I–V characteristic will take place in the GaInP cell
side due to the local generation photocurrent density being maximum. Therefore,
when 3J SC is located at a distance of 84–85.5 mm from the lens, one could expect
a maximum drop in the FF resulting in the considered “FL-SC” system efficiency
reduction.
The second minimum of the FF should be expected, when a SC is placed within the
defocusing range from +1.1 mm to +1.4 mm, due to the maximum of local density
of the photocurrent generated by GaAs junction (see Fig. 8.18) and the absolute
maximum of its integrated current (see Fig. 8.19).
The following minimum of FF may be caused by the Ge junction in the defocusing
range from +1.5 mm to +2.5 mm (see Figs. 8.18 and 8.19). However, since at the
+2.5 mm defocusing the top junction current becomes too small, only the +1.5 mm
position is worthy of consideration. Power losses on the internal resistance of the Ge
junction will lead to broadening of the second FF minimum due to the high local
photocurrent density in the cell center.
Between two minima of the I–V characteristic fill factor, when a SC is placed
from a lens at distances of 84–84.5 and 86.1–86.5 mm, respectively, there must be a
FF maximum, which resulted from decrease in the load on the top GaInP cell from
generation of the local photocurrent and its integral value and from the limitation
of the current from the top junction side at simultaneous increase in the difference
162 Zh. I. Alferov et al.

between the photocurrents of all three cells. This maximum should be expected in
the defocusing range close to +0.5 mm.
Thus, the results of the analysis presented have allowed one to determine pre-
liminarily the optimum distance between a SC and a lens, at which there must be a
maximum of the “FL-SC” system efficiency.
The verification of theoretical analysis data and the study of the effect of 3J
SC position on the “FL-SC” system efficiency were carried out outdoors at natural
sunlight. Controlled positioning of the 3J SC on the lens optical axis and variation
in the distance between the lens and the cell was performed by means of X-Y-Z
translation stage. The lens position required with respect to the cell location plane
was ensured by a rotation stage for varying its fixing angle. A 3J SC with a designated
illumination area of 2.3 mm in diameter was mounted on a heat-sink base with a
vacuum chuck and connected electrically with the I–V tracer.
The whole system, the picture of which is presented in Fig. 8.20, was established
on the sun two-axis tracker at the Ioffe Institute and adjusted normally to the incident
sunlight by an optical sighting device. The I–V curves outdoor measurements for
“FL-SC” pair were made on midday of 22 June in Saint-Petersburg. The direct
sunlight irradiation was checked concurrently using pyrheliometer CH 1 mounted
on the tracker. The average value of irradiation during the tests was 750 ± 10 W/m2 .
Dependencies of the I–V characteristic parameters (I SC , V OC, FF) and efficiency of
the cell-to-lens distance are presented in Fig. 8.21.

Fig. 8.20 An experimental “FL-SC” pair at outdoor testing

8 III–V Solar Cells and Concentrator Arrays 163

Fig. 8.21 Dependencies of Defocusing position, mm

the ISC , VOC, FF and -5 -4 -3 -2 -1 0 1 2 3 4 5

Photocurrent, mA
efficiency of the cell–lens
2.9
distance 140
VOC

Voc , V
ISC
2.8 120

2.7 100
Eff 30
25
20

Efficiency, %
0.86

FF
0.84 15
0.82 FF 10
5
0.80
0
80 81 82 83 84 85 86 87 88 89 90
Cell - lens distance, mm

The shape of the dependence of the FF on the “cell–lens” distance containing two
minima and one maximum confirms completely the theoretical estimations made
before. Actually, in the defocusing range of interest, at which the level of the pho-
tocurrent generated by a cell changes a little (i.e. the whole light, practically, is
focused on the SC photosensitive area), there is an absolute maximum of the FF
equal to 86% and a corresponding maximum efficiency to it. These maxima are
achieved at a distance of the SC from the lens equal to 85.5 mm (defocusing is
+0.5 mm), which corresponds exactly to the value obtained from the preliminary
theoretical analysis.
The first FF minimum position (FF = 80%) is found at the SC location of 83.5 mm
from the lens, which is by 0.5 mm closer to the lens, than shown by theoretical
analysis. The second broad minimum is achieved at the SC location of 87–88 mm
from the lens, which is almost by 1 mm farther. Such a shift in the minima positions
may be explained by imperfection of the lens refracting surfaces, both at the step of
fabricating the negative mold and also in forming the Fresnel profile on glass by the
direct copying technique.
By and large, it can be concluded that the data obtained at the experimental
investigation step on the SC position with respect to the lens coincide quite precisely
with the theoretical analysis results concerning primarily the determination of the
real SC position, at which the “FL-SC” system maximum efficiency is achieved.
It has been shown that, for the “FL-SC” pair considered, the optimum distance, at
which a SC should be placed with respect to a lens, is +0.5 mm to the designing
focal distance.
Allowing for the gently sloping efficiency curve in the maximum region, one may
conclude that a weakened effect of the CA on the “FL-SC” system characteristics can
be achieved in using the proposed procedure for designing lenses. In this case, a SC
can be placed in the focal plane of such a lens or somewhat farther (+0.5 mm) from the
designing distance. It might be well pointed out that, in the real system, the efficiency
164 Zh. I. Alferov et al.

maximum is achieved at the FF maximum, which allows using the dependence of

the latter on the “FL-SC” distance as a probe for funding the optimum SC location.

8.8 Conclusion

At present, III–V heterostructure solar cells are widely used for space applications.
Progress in terrestrial applications of III–V solar cells is associated with the devel-
opment of cells with efficiencies exceeding 45% at the concentrated sunlight. These
devices can form a technical basis for large-scale solar power engineering in the
future. In this case, considerable amount of electrical energy supply to our homes
will be generated by heterostructure solar cells illuminated by the sun through the
concentrators.
With the development of solar power engineering and the increase in capacity of
solar PV installations, it becomes increasingly necessary to satisfy two conflicting
requirements—increase of the installation efficiency and reduction of the generated
electricity cost. To raise the efficiency, it is necessary to use expensive III–V semicon-
ductor materials, such as GaAs, and complex multilayer semiconductor structures
including those for SCs with several p-n junctions. However, the high cost of the
materials and technologies for fabricating MJ SCs does not allow achieving a low
cost of installations with plane cells of such type, which makes their wide application
difficult. The problem can be solved by successively creating and applying SPVIs
with sunlight concentrators. In this case, the size of a SC, which is located in the con-
centrator focus, is decreased proportionally to the radiation concentration ratio—in
tens and hundreds of times. Hence, consumption of expensive semiconductor mate-
rials and structures and their contribution into the cost of the generated power unity
are reduced radically. Moreover, due to the SC efficiency rise, the installation area
required for obtaining the electric power unity lowers. This gives an additional econ-
omy of means expended for designing, construction and mounting works in creating
terrestrial solar PV installations, or for launching space solar arrays, for maintaining
parameters of a spacecraft orbit and so on. As a result, the cost of the unit electric
power of installations with radiation concentrators will be substantially lower than
that in the case of conventional plane space arrays. Thus, it is obvious that SPVIs
with radiation concentrators have wide prospects to be applied in developing solar
power engineering.
The main functional element of PV installations with radiation concentrators is a
PV module, which is a set of definite amount of structurally integrated “concentrator-
SC” systems, each of which contains one initial concentrator (e.g., Fresnel lens) and
a SC on a heat sink in its focus. In a number of cases, a secondary concentrating
element is installed above the SC.
Combination of optically effective concentrators with modern heterostructure
SCs ensures the maximum efficiency of sunlight conversion by PV modules and
is superior to all other types of radiation power converters. By 2020, the photovoltaic
community predicts overcoming 50% of MJ SC efficiencies (AM1.5d, concentrated
8 III–V Solar Cells and Concentrator Arrays 165

radiation). To achieve these indices, at present, transfer from a standard three-junction

SC to more efficient four-, five- and six-junction SCs architectures is observed; new
materials with optimal forbidden gaps are being improved and created; combined
technologies are being developed with the aim to use lattice-mismatched semicon-
ductor materials.
Practice in developing the sunlight PV conversion technique shows that high
values of SC and PV system efficiencies demonstrated in indoor conditions, as a
rule, appear on the large-scale market of PV production in 2–3 years, promoting
humanity to an energetically secure “green” future.

References

1. W. Shockley, Circuit Element Utilizing Semiconductor Material. U.S. Patent 2269347,

September 25, 1951
2. A.I. Gubanov, Theory of the contact between two semiconductors with different types of
conduction. Zh. Tekh. Fiz. 20, 1287 (1950)
3. H. Kroemer, Theory of a wide-gap emitter for transistors. Proc. IRE 45(11), 1535–1537 (1957)
4. Zh.I. Alferov, The double heterostructure: concept and its applications in physics, electronics
and technology (Norstedts Tryckeri, Stockholm, Les prix Nobel, 2001), pp. 65–93
5. H. Kroemer, A proposed class of heterojunction injection lasers. Proc. IEEE 51(12), 1782–
1783 (1963)
6. Zh. I. Alferov, V.B. Khalfin, R.F. Kazarinov, A characteristic feature of injection into hetero-
junctions. Fiz. Tverd. Tela 8, 3102–3105 (1966) [Sov. Phys. Solid State 8, 2480 (1967)]
7. Z.I. Alferov, Possible development of a rectifier for very high current densities on the bases
of a p-i-n (p-n-n, n-p-p) structure with heterojunctions. Fiz. Tekh. Poluprovodn. 1, 436–438
(1966) [Sov. Phys. Semicond. 1, 358–361 (1967)]
8. R.L. Anderson, Germanium-gallium arsenide heterojunctions. IBM J. Res. Dev. 4(3), 283–
287 (1960)
9. Z.I. Alferov, V.M. Andreev, V.I. Korol’kov, D.N. Tret’yakov, V.M. Tuchkevich, High-voltage
p-n junctions in Gax Al1−x crystals. Fiz. Tekh. Poluprovodn. 1, 1579–1581(1967) [Sov. Phys.
Semicond. 1, 1313–1314 (1968)]
10. H.S. Rupprecht, J.M. Woodall, G.D. Pettit, Efficient visible electroluminescence at 300 K
from Ga1−x Alx As p-n junctions grown by liquid-phase epitaxy. Appl. Phys. Lett. 11(3), 81–
83 (1967)
11. Z.I. Alferov, V.M. Andreev, V.I. Korol’kov, E.L. Portnoi, D.N. Tret’yakov, Injection properties
of n–Alx Ga1−x As-p-GaAs heterojunctions. Fiz. Tekh. Poluprovodn. 2, 1016–1017 (1968)
[Sov. Phys. Semicond. 2, 843–844 (1969)]
12. Z.I. Alferov, V.M. Andreev, V.I. Korol’kov, E.L. Portnoy, D.N. Tret’yakov, Coherent radiation
of epitaxial heterojunction structures in the AlAs–GaAs system. Fiz. Tekh. Poluprovodn. 2,
1545–1547 (1968) [Sov. Phys. Semicond. 2, 1289–1291 (1969)]
13. Z.I. Alferov, V.M. Andreev, V.I. Korol’kov, E.L. Portnoi, D.N. Tret’yakov, Recombination
radiation in epitaxial structures in the AlAs–GaAs system, in Proceedings of IX International
Conference on the Physics of Semiconductors, Moscow, vol. 1 (Nauka, Leningrad, 1968),
pp. 504–510
14. Z.I. Alferov, V.M. Andreev, E.L. Portnoy, M.K. Trukan, AlAs–GaAs heterojunctions injection
lasers with a low room-temperature threshold. Fiz. Tekh. Poluprovodn. 3, 1328–1332 (1969)
[Sov. Phys. Semicond. 3, 1107–1110 (1970)]
15. I. Hayashi, Heterostructure lasers. IEEE Trans. Electr. Devices 31(11), 1630–1642 (1984)
166 Zh. I. Alferov et al.

16. Z.I. Alferov, V.M. Andreev, D.Z. Garbuzov, Y.V. Zhilyaev, E.P. Morozov, E.L. Portnoi, V.G.
Trofim, Investigation of the influence of the AlAs–GaAs heterostructure parameters on the
laser threshold current and the realization of continuous emission at the room temperature.
Fiz. Tekh. Poluprovodn. 4, 1826–1829 (1970) [Sov. Phys. Semicond. 4, 1573–1575 (1971)]
17. I. Hayashi, M.B. Panish, P.W. Foy, S. Sumski, Junction lasers which operate continuously at
room temperature. Appl. Phys. Lett. 17, 109–111 (1970)
18. Z.I. Alferov, V.M. Andreev, V.I. Korol’kov, E.L. Portnoi, A.A. Yakovenko, Spontaneous radi-
ation sources based on structures with AlAs–GaAs heterojunctions. Fiz. Tekh. Poluprovodn.
3, 930–933 (1969) [Sov. Phys. Semicond. 3, 785–787 (1970)]
19. Z.I. Alferov, V.M. Andreev, M.B. Kagan, I.I. Protasov, V.G. Trofim, Solar-energy converters
based on p-n Alx Ga1−x As-GaAs heterojunctions. Fiz. Tekh. Poluprovodn. 4, 2378–2379
(1970) [Sov. Phys. Semicond. 4, 2047–2048 (1971)]
20. Z.I. Alferov, F.A. Ahmedov, V.I. Korol’kov, V.G. Nikitin, Phototransistor utilizing a GaAs–
AlAs heterojunction. Fiz. Tekn. Poluprovodn. 7, 1159–1163 (1973) [Sov. Phys. Semicond. 7,
780–782 (1973)]
21. Z.I. Alferov, V.M. Andreev, V.I. Korol’kov, V.G. Nikitin, A.A. Yakovenko, p-n-p-n structures
based on GaAs and on Alx Ga1−x As solid solutions. Fiz. Tekn. Poluprovodn. 4, 578–581
(1970) [Sov. Phys. Semicond. 4, 481–483 (1971)]
22. Z.I. Alferov, V.M. Andreev, S.G. Konnikov, V.G. Nikitin, D.N. Tret’yakov, Heterojunctions
on the base of III–V semiconducting and of their solid solutions, in Proceedings of Interna-
tional Conference Physical Chemical Semiconductors Heterojunctions and Layer Structures,
Budapest, 1970, vol. 1, ed. by G. Szigeti (Academiai Kiado, Budapest, 1971), pp. 93–106
23. G.A. Antipas, R.L. Moon, L.W. James, J. Edgecumbe, R.L. Bell, in Gallium Arsenide and
Related Compounds, vol. 17. 1972 Conference Series IOP, London: Institute of Physics, 1973,
p. 48
24. A.P. Bogatov, L.M. Dolginov, L.V. Druzhinina, P.G. Eliseev, L.N. Sverdlova, E.G.
Shevchenko, Heterolasers on the base of solid solutions Gax In1−x Asy P1−y and
Alx Ga1−x Sby As1−y . Kvantovaya Electron. 1, 2294 (1974) [Sov. J. Quant. Electr. 1, 1281,
1974]
25. Zh. I. Alferov, I.N. Arsent’ev, D.Z. Garbuzov, S.G. Konnikov, V.D. Rumyantsev. Generation
of coherent radiation in nGa0.5 In0.5 P-pGax~0.55 In1−x Asy~0.10 P1−y —nGa0.5 In0.5 P. Pisma Zh.
Tech. Fiz. 1, 305–310, 1975 [Sov. Phys.-Tech. Phys. Lett. 1, 147–148, 1975]
26. D. Flood, H. Brandhorst, Space solar cells, in Current Topics in Photovoltaics, vol. 2, ed. by
T.J. Coutts, J.D. Meakin (Academic Press, New York, 1987, London), pp. 143–202
27. S.G. Bailey, D.J. Flood, Space photovoltaics. Prog. Photovolt. Res. Appl. 6(1), 1–14 (1998)
28. V.M. Andreev, V.A. Grilikhes, V.D. Rumyantsev, Photovoltaic conversion of concentrated
sunlight (Wiley & Sons Ltd., New York, 1997)
29. H.J. Hovel, J.M. Woodall, High-efficiency AlGaAs-GaAs solar cells. Appl. Phys. Lett. 21,
379–381 (1972)
30. V.M. Andreev, T.M. Golovner, M.B. Kagan, N.S. Koroleva, T.A. Lubochevskaya, T.A. Nuller,
D.N. Tret’yakov, Investigation of high efficiency AlGaAs-GaAs solar cells. Sov. Phys. Semi-
cond. 7(12) (1973)
31. Zh.I. Alferov, V.M. Andreev, G.S. Daletskii, M.B. Kagan, N.S. Lidorenko, V.M. Tuchke-
vich. Investigation of high efficiency AlAs-GaAs heteroconverters, in Proceedings of World
Electrotechnology, Congress, Moscow, 1977, Section 5A, report 04
32. H.J. Hovel, Solar cells, in Semiconductors and Semimetals, ed. by R.K. Willardson, A.C.
Beer, vol. 11 (New York, London: Acad. Press, 1975)
33. V.M. Andreev, III–V heterostructure photovoltaics in Russia, in Proceedings of the 17th
European Photovoltaic Solar Energy Conference, Munich, 2001, pp. xxxi–xxxii (2001)
34. J.M. Woodall, H.J. Hovel, An isothermal etchback-regrowth method for high efficiency
Ga1-x Alx As-GaAs solar cells. Appl. Phys. Lett. 30, 492–493 (1977)
35. V.M. Andreev, V.R. Larionov, V.D. Rumyantsev, O.M. Fedorova, Sh.Sh. Shamukhamedov, P
AlGaAs-pGaAs-nGaAs solar cells with efficiencies of 19% at AM0 and 24% at AM1.5. Sov.
Tech. Phys. Lett. 9(10), 537–538 (1983)
8 III–V Solar Cells and Concentrator Arrays 167

36. H.J. Hovel, Novel materials and devices for sunlight concentrating systems. IBM J. Res.
Develop. 22, 112–121 (1978)
37. E. Fanetti, C. Flores, G. Guarini, F. Paletta, D. Passoni, High efficiency 1.43 and 1.69 eV band
gap Ga1-x Alx As-GaAs solar cells for multicolor applications. Solar cells 3, 187–194 (1981)
38. R.C. Knechtly, R.Y. Loo, G.S. Kamath, High-efficiency GaAs solar cells. IEEE Trans. Electron
Devices 31(5), 577–588 (1984)
39. H.S. Rauschenbach, Solar cell array design handbook (Litton Educational Publishing Inc,
The principles and technology of photovoltaic energy conversion. New York, 1980)
40. A. Luque, G. LAraujo. Solar cells and optics for photovoltaic concentration. Bristol-
Philadelphia: Adam Hilger, 1989, 531 p
41. L.D. Partain (ed.), Solar Cells and Their Application (New York: Wiley, 1995)
42. P.A. Iles, Future of Photovoltaic for Space Applications. Prog. Photovoltaics Res. Appl. 8,
39–51 (2000)
43. V.M. Andreev, A.B. Kazantsev, V.P. Khvostikov, E.V. Paleeva, V.D. Rumyantsev, M.Z.
Shvarts. High-efficiency (24.6%, AM0) LPE Grown AlGaAs/GaAs concentrator solar cells
and modules. In Conference Record of the First World Conference on Photovoltaic Energy
Conversion (WCPEC) (Waikoloa, Hawaii, 1994), pp. 2096–2099
44. V.M. Andreev, V.D. Rumyantsev, A3 B5 based solar cells and concentrating optical elements
for space PV modules. Sol. Energy Mater. Sol. Cells 44, 319–332 (1996)
45. R.D. Dupuis, P.D. Dapkus, R.D. Vingling, L.A. Moundy, High-efficiency GaAlAs/GaAs het-
erostructure solar cells grown by metalorganic chemical vapor deposition. Appl. Phys. Lett.
31, 201–203 (1977)
46. N.J. Nelson, K.K. Jonson, R.L. Moon, H.A. Vander Plas, L.W. James, Organometallic-sourced
VPE AlGaAs/GaAs concentrator solar cells having conversion efficiencies of 19%. Appl.
Phys. Lett. 33. 26–27 (1978)
47. J.G. Werthen, G.F. Virshup, C.W. Ford, C.R. Lewis, H.C. Hamaker, 21% (one sun, air mass
zero) 4 cm2 GaAs space solar cells. Appl. Phys. Lett. 48, 74–75 (1986)
48. S.P. Tobin, S.M. Vernon, S.J. Woitczuk, C. Baigar, M.M. Sanfacon, T.M. Dixon. Advanced
in high-efficiency GaAs solar cells, in Conference Record of the 21st IEEE Photovoltaic
Specialists Conference (PVSC), Kissimimee, pp. 158–162 (1990)
49. S.P. Tobin, S.M. Vernon, M.M. Sanfacon, A. Mastrovito, Enhanced light absorption in GaAs
solar cells with internal Bragg reflector, in Conference Record of the 22nd IEEE Photovoltaic
Specialists Conference, pp. 147–152 (1991)
50. V.M. Andreev, V.V. Komin, I.V. Kochnev, V.M. Lantratov, M.Z. Shvarts, High-efficiency
AlGaAs-GaAs solar cells with internal Bragg reflector, in Conference Record of the First
World Conference on Photovoltaic Energy Conversion (WCPEC) (Waikoloa, Hawaii, 1994),
pp. 1894–1897
51. M.Z. Shvarts, O.I. Chosta, I.V. Kochnev, V.M. Lantratov, V.M. Andreev, Radiation resistant
AlGaAs/GaAs concentrator solar cells with internal Bragg reflector. Sol. Energy Mater. Sol.
Cells 68, 105–122 (2001)
52. M. Yamaguchi, Space solar cell R&D activities in Japan, in Proceedings of the 15th Space
Photovoltaic Research and Technology Conference, pp. 1–10 (1997)
53. C.C. Fan, B.-Y. Tsaur, B.J. Palm, Optimal design of high-efficiency tandem cells, in Confer-
ence Record of the 16th IEEE Photovoltaic Specialists Conf. (PVSC), pp. 692–698 (1982)
54. M.A. Green, Solar Cells (Prentice-Hall Inc., New Jersey, 1982)
55. M.F. Lamorte, D.H. Abbott, Computer modeling of a two-junction, monolithic cascade solar
cell. IEEE Trans. Electron Devices 27, 231–249 (1980)
56. M.B. Spitzer, C.C. Fan, Multijunction cells for space applications. Solar Cells 29, 183–203
(1990)
57. L. Fraas, B. Daniels, Han-Xiang Huang, J. Avery, C. Chu, P. Iles, M. Piszczor, Over 30%
efficient InGaP/GaAs/GaSb cell-interconnected-circuits for line-focus concentrator arrays,
in Proceedings of the 28th IEEE Photovoltaic Specialists Conference (PVSC) (Anhorage,
Alaska, 2000), pp. 1150–1153
168 Zh. I. Alferov et al.

58. L. Fraas, J. Avery, D. Scheiman, AM0 calibration of 34% efficient mechanically stacked
GaInP/GaAsGaSb circuits, in Proceedings of the 29th IEEE Photovoltaic Specialists Confer-
ence (PVSC) (New Orleans, 2002), pp. 912–915
59. M.Z. Shvarts, P.Y. Gazaryan, V.P. Khvostikov, V.M. Lantratov, N.K. Timoshina, InGaP/GaAs-
GaSb and InGaP/GaAs/Ge-InGaAsSb hybrid monolithic/stacked tandem concentrator solar
cells, in Proceedings of the 21st European Photovoltaic Solar Energy Conference (Dresden,
2006), pp. 133–136
60. I. Mathews, D. O’Mahony, K. Thomas, B. Corbett, A.P. Morrison, Adhesive bonding for
mechanically stacked solar cells. Prog. Photovolt. Res. Appl. 23(9), 1080–1090 (2015)
61. J.D. McCambridge, M.A. Steiner, B.A. Unger, K.A. Emery, E.L. Christensen, M.W. Wan-
lass, A.L. Gray, L. Takacs, R. Buelow, T.A. McCollum, J.W. Ashmead, G.R. Schmidt, A.W.
Haas, J.R. Wilcox, J.V. Meter, J.L. Gray, D.T. Moore, A.M. Barnett, R.J. Schwartz, Compact
spectrum splitting photovoltaic module with high efficiency. Prog. Photovoltaics Res. Appl.
19, 352–360 (2011)
62. M.A. Green, M.J. Keevers, I. Thomas, J.B. Lasich, K. Emery, R.R. King, 40% efficient sunlight
to electricity conversion. Prog. Photovoltaics Res. Appl. 23(6), 685–691 (2015)
63. M.A. Green, K. Emery, D.L. King, Y. Nishikawa, W. Warta, E.D. Dunlop, D.H. Levi, A.W.Y.
Ho-Baillie, Solar cell efficiency tables (version 49). Prog. Photovoltaics Res. Appl. 25, 3–13
(2017)
64. B.M. Kayes, L. Zhang, R. Twist, I.K. Ding, G.S. Higashi, Flexible thin-film tandem solar
cells with >30% efficiency. IEEE Journal of Photovoltaics 4, 729–733 (2014)
65. I. Garcia, I. Rey-Stolle, B. Gallana, C. Algora, A 32.6% efficient lattice-matched dual-junction
solar cell working at 1000 suns. Appl. Phys. Lett. 94(5), 053509 (2009)
66. J. Ohlmann, D. Lackner, J.F. Sanchez, M. Zedda, A. Wekkeli, M. Steiner, A. Fallisch, F.
Dimroth, 35.1% efficient dual-junction solar cells optimized for direct hydrogen generation,
in Proceedings of the 44th IEEE Photovoltaic Specialists Conference (PVSC), Washington
(2017)
67. R.R. King, D.C. Law, K.M. Edmondson, C.M. Fetzer, G.S. Kinsey, H. Yoon, R.A. Sherif,
N.H. Karam, 40% efficient metamorphic GaInP/GaInAs/Ge multijunction solar cells. Appl.
Phys. Lett. 90, 183516 (2007)
68. W. Guter, J. Schöne, S. P. Philipps, M. Steiner, G. Siefer, A. Wekkeli, E. Welser, E. Oliva,
A.W. Bett, F. Dimroth, Current-matched triple-junction solar cell reaching 41.1% conversion
efficiency under concentrated sunlight. Appl. Phys. Lett. 94(22), 223504–223506 (2009)
69. R.R. King, A. Boca, W. Hong, X.-Q. Liu, D. Bhusari, D. Larrabee, K.M. Edmondson, D.C.
Law, C.M. Fetzer, S. Mesropian, N.H. Karam, Band-Gap engineered architectures for high-
efficiency multijunction concentrator solar cells, in Proceedings of the 24th European Pho-
tovoltaic Solar Energy Conference, Hamburg (2009), pp 55–61
70. M.A. Green, Y. Hishikawa, E.D. Dunlop, D.H. Levi, J. Hohl-Ebinger, Anita W.Y. Ho-Baillie,
Solar cell efficiency tables (version 51). Prog. Photovoltaics Res. Appl. 26, 3–12 (2018)
71. V. Sabnis, H. Yuen, M. Wiemer, High-efficiency multijunction solar cells employing dilute
nitrides. AIP Conf. Proc. 1477, 14–19 (2012)
72. P. Chiu, S. Wojtczuk, X. Zhang, C. Harris, D. Pulver, M. Timmons, 42.3% efficient
InGaP/GaAs/InGaAs concentrators using bifacial epigrowth, in Proceedings of the 37th IEEE
Photovoltaic Specialists Conference (PVSC), Seattle (2011), pp. 771–774
73. Press release, Sharp Corporation, 31 May 2012. Accessed at http://www.sharp-world.com/
corporate/news/120531.html
74. A. Yoshida, T. Agui, N. Katsuya, K. Murasawa, H. Juso, K. Sasaki, T. Takamoto, Develop-
ment of InGaP/GaAs/InGaAs inverted triple junction solar cells for concentrator application,
in Proceedings of the 21st International Photovoltaic Science and Engineering Conference
(PVSEC-21), Fukuoka (2011)
75. P.T. Chiu, D.L. Law, R.L. Woo, S. Singer, D. Bhusari, W.D. Hong, A. Zakaria, J.C. Boisvert,
S. Mesropian, R.R. King, N.H. Karam, 35.8% space and 38.8% terrestrial 5J direct bonded
cells, in Proceedings of the 40th IEEE Photovoltaic Specialists Conference (PVSC), Denver,
Colorado, pp. 11–13
8 III–V Solar Cells and Concentrator Arrays 169

76. F. Dimroth, T.N.D. Tibbits, M. Niemeyer et al., Four-junction wafer-bonded concentrator

solar cells. IEEE Journal of Photovoltaics 6, 343–349 (2016)
77. https://www.ise.fraunhofer.de/en/press-media/news/2017/31–3-percent-efficiency-for-
silicon-based-multi-junction-solar-cell.html
78. R. Cariou, J. Benick, P. Beutel et al., Monolithic two-terminal III–V//Si triple-junction solar
cells with 30.2% efficiency under 1-Sun AM1.5g. IEEE J. Photovoltaics 7(1), 367–373 (2017)
79. A.P. Kirk, High efficacy thinned four-junction solar cell. Semicond. Sci. Technol. 26(12),
125013 (2011)
80. D. Aiken, E. Dons, S.-S. Je, N. Miller, F. Newman, P. Patel, J. Spann, Lattice matched solar cells
with 40% average efficiency in pilot production and a roadmap to 50%. IEEE J. Photovoltaics
3(1), 542–547 (2013)
81. M. Ochoa, I. García, I. Lombardero et al., Advances towards 4J lattice-matched including
dilute nitride subcell for terrestrial and space applications, in Proceedings of the IEEE 43rd
Photovoltaic Specialists Conference (PVSC), Portland, 2016, pp. 0052–0057
82. R. Roucka, A. Clark, B. Landini, Si-Ge-Sn Alloys with 1.0 eV Gap for CPV Multijunction
Solar Cells. AIP Conf. Proc. 1679, 040008–040008 (2015)
83. T. Wilson, T. Thomas, M. Führer, N.J. Ekins-Daukes, R. Roucka, A. Clark, A. Johnson,
R. Hoffman, D. Begarney, Single and multi-junction solar cells utilizing a 1.0 eV SiGeSn
junction, in AIP Conference Proceedings, vol. 1766 (2016), pp. 060006
84. C. Algora, E. Ortiz, I. Rey-Stolle, V. Diaz, P. Pena, V.M. Andreev, V.P. Khvostikov, V.D.
Rumyantsev, A GaAs solar cell with efficiency of 26.2% at 1000 suns and 25.0% at 2000
suns. IEEE Trans. Electron Devices 48(5), 840–844 (2001)
85. V.M. Andreev, V.P. Khvostikov, V.R. Larionov, V.D. Rumyantsev, E.V. Paleeva, M.Z. Shvarts,
C. Algora, 5800 Suns AlGaAs/GaAs concentrator solar cells, in Proceedings of Techni-
cal Digest of the International Photovoltaic Science and Engineering Conference, Sapporo
(1999), pp. 147–148
86. V.M. Andreev, I.V. Kochnev, V.M. Lantratov, S.A. Mintairov, V.D. Rumyantsev, M.Z. Shvarts,
Ultra-violet sensitive infra-red reflective AlGaAs/GaAs solar cells with two Bragg reflectors,
in Proceedings of the 16th European Photovoltaic Solar Energy Conference, Glasgow (2000),
pp. 1019–1021
87. I. García, J. Geisz, M. Steiner, J. Olson, D. Friedman, S. Kurtz, Design of semiconductor-
based back reflectors for high voc monolithic multijunction solar cells, in Preprint Presented
at the IEEE Photovoltaic Specialists Conference, Austin, 2012
88. R. Campesato, G. Gori, M. Casale, G. Gabetta, G. Muthusamy, Sankaran, P. Suresh, E.R. Uma,
Radiation effects on advanced multi junction solar cells for space missions, in Proceedings
of the 32nd European Photovoltaic Solar Energy Conference and Exhibition, Munich (2016),
pp. 1411–1414
89. L.M. Fraas, J.E. Avery, J. Martin, V.S. Sundaram, G. Giard, V.T. Dinh, T.M. Davenport, J.W.
Yerkes, M.J. O’Neil, Over 35-percent efficient GaAs/GaSb tandem solar cells. IEEE Trans.
Electron Devices 37(2), 443–449 (1990)
90. M.W. Wanlass, J.S. Ward, K.A. Emery, T.A. Gessert, C.R. Osterwald, T.J. Coutts, High
performance concentrator tandem solar cells based on IR-sensitive bottom cells. Solar cells
30(1–4), 363–371 (1991)
91. B.-C. Chung, G.F. Virshup, S. Hikido, N.R. Kaminar, 27.6% efficiency (1 sun, air mass 1.5)
monolithic Al0.37 Ga0.63 As/GaAs two junction cascade solar cell with prismatic cover glass.
Appl. Phys. Lett. 55, 1741–1743 (1989)
92. V.M. Andreev, V.P. Khvostikov, E.V. Paleeva, V.D. Rumyantsev, S.V. Sorokina, M.Z. Shvarts,
V.I. Vasilev, Tandem solar cells based on AlGaAs/GaAs and GaSb structures, in Proceedings
of the 23rd International Symposium on Compound Semiconductors (St. Petersburg, Russia,
1996), pp. 425–428
93. D.J. Friedman, S.R. Kurtz, K.A. Bertness, A.E. Kibbler, C. Kramer, J.M. Olson, D.L. King,
B.R. Hansen, J.K. Snyder, GaInP/GaAs monolithic tandem concentrator cells, in Proceedings
of the 1st World Conference on Photovoltaic Energy Conversion (WCPEC) (Waikoloa, Hawaii,
1994), pp. 1829–1832
170 Zh. I. Alferov et al.

94. R.R. King, D.C. Law, C.M. Fetzer, R.A. Sherif, K.M. Edmondson, S. Kurtz, G.S. Kinsey,
H.L. Cotal, D.D. Krut, J.H. Ermer, N.H. Karam, Pathways to 40%-efficient concentration
photovoltaics, in Proceedings of the 20th European International Photovoltaic Science and
Engineering Conference (PVSEC-20), Barcelona (2005), pp. 6–10
95. F. Dimroth, R. Beckert, M. Meusel, U. Schubert, A.W. Bett, Metamorphic
Gay In1−y P/Ga1−x Inx As tandem solar cells for space and for terrestrial concentrator
applications at C > 1000 suns. Prog. Photovoltaics Res. Appl. 9(3), 165–178 (2001)
96. D.J. Aiken, M.A. Stan, S.P. Endicter, G. Girard, P.R. Sharps, A loss analysis for a 28% efficient
520× concentrator module, in Proceedings of the IEEE 4th World Conference on Photovoltaic
Energy Conversion(WCPEC), Hawaii (2006), pp. 686–689
97. M. Yamaguchi, Y. Okada, A. Yamamoto, T. Takamoto, K. Araki, Y. Ohshita, Novel mate-
rials and structures for high efficiency multi-junction solar cells, in Proceedings of the 21st
European Photovoltaic Solar Energy Conference, Dresden (2006), pp. 53–56
98. A.W. Bett, S.P. Philipps, S. Essig, S. Heckelmann, R. Kellenbenz, V. Klinger, M. Niemeyer,
D. Lackner, F. Dimroth, Overview about technology perspectives for high efficiency solar
cells for space and terrestrial applications, in Proceedings of the 28th European International
Photovoltaic Science and Engineering Conference (PVSEC-28) (2013), pp. 1–6
99. S.A. Ringel, J. Carlin, C. Andre, M. Hudait, M. Gonzalez, D. Wilt, E.B. Clark, P. Jenkins,
D. Scheiman, A. Allerman, E.A. Fitzgerald, C.W. Leitz, Single-junction InGaP/GaAs solar
cells grown on Si substrates with SiGe buffer layers. Prog. Photovoltaics Res. Appl. 10(6),
417–426 (2002)
100. K. Derendorf, S. Essig, E. Oliva, V. Klinger, T. Roesener, S.P. Philipps, J. Benick, M. Hermle,
M. Schachtner, G. Siefer, W. Jäger, F. Dimroth, Fabrication of GaInP/GaAs//Si solar cells by
surface activated direct wafer bonding. IEEE J. Photovoltaics 3(4), 1423–1428 (2013)
101. F. Dimroth, T. Roesener, S. Essig, C. Weuffen, A. Wekkeli, E. Oliva, G. Siefer, K. Volz,
T. Hannappel, D. Häussler, W. Jäger, A.W. Bett, Comparison of direct growth and wafer
bonding for the fabrication of beam GaInP/GaAs dual-junction solar cells on silicon. IEEE J.
Photovoltaics 4(2), 620–625 (2014)
102. J.M. Zahler, K. Tanabe, C. Ladous, T. Pinnington, F.D. Newman, H. Atwater, High efficiency
InGaAs solar cells on Si by InP layer transfer. Appl. Phys. Lett. 91(1), 012108 (2007)
103. M.J. Archer, D.C. Law, S. Mesropian, M. Haddad, C.M. Fetzer, A.C. Ackerman, C. Ladous,
R.R. King, H.A. Atwater, GaInP/GaAs dual junction solar cells on Ge/Si epitaxial templates.
Appl. Phys. Lett. 92, 103503–103503 (2008)
104. P. Patel, D. Aiken, A. Boca, B. Cho, D. Chumney, M.B. Clevenger, A. Cornfeld, N. Fatemi,
Y. Lin, J. McCarty, F. Newman, P. Sharps, J. Spann, M. Stan, J. Steinfeldt, C. Strautin, T.
Varghese, Experimental results from performance improvement and radiation hardening of
inverted metamorphic multijunction solar cells. IEEE J. Photovoltaics 2, 377–381 (2012)
105. D. Jackrel, S. Bank, H. Yuen, M. Wistey, J. Harris, A. Ptak, S. Johnston, D. Friedman, S.
Kurtz, Dilute nitride GaInNAs and GaInNAsSb solar cells by molecular beam epitaxy. J.
Appl. Phys. 101, 114916 (2007)
106. A.B. Cornfeld, M. Stan, T. Varghese et al. Development of a large area inverted metamor-
phic multi-junction (IMM) highly efficient AM0 solar cell, in Proceedings of the 33rd IEEE
Photovoltaic Specialists Conference (PVSC), San Diego, CA (2008), pp 88–92
107. J.F. Geisz, D.J. Friedman, J.S. Ward et al., 40.8% efficient inverted triple-junction solar cell
with two independently metamorphic junctions. Appl. Phys. Lett. 93, 123505 (2008)
108. R.M. France, J.F. Geisz, M.A. Steiner, D.J. Friedman, J.S. Ward, J.M. Olson, W. Olavarria,
M. Young, A. Duda, Pushing inverted metamorphic multijunction solar cells toward higher
efficiency at realistic operating conditions. IEEE J. Photovoltaics 3, 893–898 (2013)
109. C. Youtsey, J. Adams, R. Chan, V. Elarde, G. Hillier, M. Osowski, D. McCallum, H. Miyamoto,
N. Pan, C. Stender, R. Tatavarti, F. Tuminello, A. Wibowo, Epitaxial lift-off of large-area GaAs
thin-film multi-junction solar cells, in Proceedings of the CS MANTECH Conference, Boston,
Massachusetts (2012), on CD
110. H. Moriceau, F. Rieutord, F. Fournel, Y. Le Tice, L. Di Cioccio, C. Morales, A.M. Charvet,
C. Deguet, Overview of recent direct wafer bonding advances and applications, in Advances
in Natural Sciences: Nanoscience and Nanotechnology, vol. 1 (2010), p. 043004
8 III–V Solar Cells and Concentrator Arrays 171

111. Q.-Y. Tong, U. Gosele, Semiconductor Wafer Bonding: Science and Technology (Electro-
chemical Society, N.Y., 1999)
112. D.C. Law, D.M. Bhusari, S. Mesropian, J.C. Boisvert, W.D. Hong, A. Boca, D.C. Larrabee,
C.M. Fetzer, R. R. King, N.H. Karam, Semiconductor-bonded III–V multijunction space
solar cells, in Proceedings of the 34th IEEE Photovoltaic Specialists Conference (PVSC),
Philadelphia, PA (2009), pp 2237–2239
113. J. Boisvert, D. Law, R. King, D. Bhusari, X. Liu, A. Zakaria, W. Hong, S. Mesropian, D.
Larrabee, R. Woo, A. Boca, K. Edmondson, D. Krut, D. Peterson, K. Rouhani, B. Benedikt,
N. H. Karam, Development of advanced space solar cells at Spectrolab, in Proceedings of the
35th IEEE Photovoltaic Specialists Conference (PVSC), Honolulu, HI (2010), pp 123–127
114. Handbook of Photovoltaic Science and Engineering. 2nd edn. A. Luque, S. Hegedus ed. by
John Wiley & Sons, Ltd. (2011)
115. Terrestrial concentrator solar cell systems, in Solar Cells and Their Applications, 2nd edn.,
Part III, ed. by L.M. Fraas, L.D. Partain (Hoboken, New Jersey: John Wiley & Sons, Inc.,
2010)
116. Handbook on Concentrator Photovoltaic Technology, ed. by A. Carlos, Rey-Stolle Ignacio.
(John Wiley & Sons, Ltd. 2016)
117. M.A. Green, K. Emery, D. L. King, Y. Nishikawa, W. Warta, Solar cell efficiency tables
(version 29), in Progress in Photovoltaics: Research and Applications, vol. 15, pp. 35–40
(2007)
118. L.W. Fraas, W.E. Daniels, H.X. Huang, L.E. Minkin, J.E. Avery, M.J. O’Neill, A.J. McDanal,
M.F. Piszczor, 34% efficient InGaP/GaAs/GaSb cell-interconnected-circuits for line-focus
concentrator arrays, in Proceedings of the 17th European Photovoltaic Solar Energy Confer-
ence, Munich (2001), pp. 2300–2303
119. Zh.I. Alferov, V.M. Andreev, V.D. Rumyantsev, III–V heterostructures in photovoltaics, in
Concentrator Photovoltaics, vol. 130, ed. by A. Luque, V. Andreev (New York: Springer
Series in Optical Sciences, 2007), pp. 25–50
120. B.M. Kayes, H. Nie, R. Twist, S.G. Spruytte, F. Reinhardt, I.C. Kizilyalli, S. Gregg, Higashi,
27.6% conversion efficiency, a new record for single-junction solar cells under 1 sun illumi-
nation, in Proceedings of the 37th IEEE Photovoltaic Specialists Conference (PVSC), Seattle
(2011), pp. 4–8
121. K. Sasaki, T. Agui, K. Nakaido, N. Takahashi, R. Onitsuka, T. Takamoto, “Development of
InGaP/GaAs/InGaAs inverted triple junction concentrator solar cells». AIP Conf. Proc. 1556,
22–25 (2013)
122. J.F. Geisz, A. Duda, R.M. France, D.J. Friedman, I. Garcia, W. Olavarria, J.M. Olson, M.A.
Steiner, J.S. Ward, M. Young, Optimization of 3-junction inverted metamorphic solar cells
for high-temperature and high-concentration operation, in 9th International Conference on
Concentrator Photovoltaic Systems, ed. by J. Mendiguren Olaeta, B. Rolfe, E. Atzema, D.
Hanselman, L. Galdos Errasti, P. Hodgson, M. Weiss, AIP Conference Proceedings, vol. 1477
(2013), pp. 44–48
123. F. Dimroth, M. Grave, P. Beutel, U. Fiedeler, C. Karcher, T.N.D. Tibbits, E. Oliva, G. Siefer,
M. Schachtner, A. Wekkeli, A.W. Bett, R. Krause, M. Piccin, N. Blanc, C. Drazek, E.
Guiot, B. Ghyselen, T. Salvetat, A. Tauzin, T. Signamarcheix, A. Dobrich, T. Hannappel,
K. Schwarzburg, Wafer bonded four-junction GaInP/GaAs//GaInAsP/GaInAs concentrator
solar cells with 44.7% efficiency. Prog. Photovoltaics Res. Appl. 22(3), 277–282 (2014)
124. R.M. France, J.F. Geisz, I. Garcia, M.A. Steiner, W.E. McMahon, D.J. Friedman, T.E. Mori-
arty, C. Osterwald, J. Scott Ward, A. Duda, M. Young, W.J. Olavarria, Quadruple-junction
inverted metamorphic concentrator devices. IEEE J. Photovoltaics 5(1), 432–437 (2015)
125. Essig, S. et al., Raising the one-sun conversion efficiency of III–V/Si solar cells to 32.8% for
two junctions and 35.9% for three junctions. Nat. Energy, 2, article number 17144 (2017)
126. V.M. Andreev, E.A. Ionova, V.R. Larionov, V.D. Rumyantsev, M.Z. Shvarts, G. Glenn, Tunnel
diode revealing peculiarities at I–V measurements in multijunction III–V solar cells, in Pro-
ceedings of the IEEE 4th World Conference on Photovoltaic Energy Conversion (WCPEC),
Hawaii (2006), pp. 799–802
172 Zh. I. Alferov et al.

127. G. Sala, A. Luque, Past experiences and new challenges of PV concentrators, in Concentrator
Photovoltaics, vol. 130, ed. by A. Luque, V. Andreev. New York: Springer Series in Optical
Sciences (2007), pp. 1–24
128. V.D. Rumyantsev, Terrestrial concentrator PV systems in Concentrator Photovoltaics, vol.
130, ed. by A. Luque and V. Andreev. New York: Springer Series in Optical Sciences (2007),
pp. 151–174
129. Zh.I. Alferov, V.M. Andreev, Kh.K. Aripov, V.R. Larionov, V.D. Rumyantsev, Pattern of
autonomal solar installation with heterostructure solar cells and concentrators. Geliotechnica,
2, 3–6 (1981) [Appl. Solar Energy 2, 1981]
130. Zh.I. Alferov, V.M. Andreev, Kh.K. Aripov, V.R. Larionov, V.D. Rumyantsev, Solar photo-
voltaic installation with 200 Watt output based on AlGaAs-heterophotocells and reflective
concentrators, Geliotechnika, 6, 3–6 (1981) [Appl. Solar Energy 6, 1981]
131. A.A. Vodnev, A.V. Maslov, V.D. Rumyantsev, Sh.Sh. Shamukhamedov, Experience on cre-
ation of the solar installations based on AlGaAs/GaAs-photocells with concentrators, in Sun-
light Concentrators for Photovoltaic Power Installations, ed. by V.A. Grilikhes, Leningrad:
Energoatomizdat (1986), pp. 25–29 (in Russian)
132. V.M. Andreev, A.A. Alaev, A,B. Guchmazov, V.S. Kalinovsky, V.R. Larionov, K.Ya. Rasulov,
V. D. Rumyantsev, High-efficiency AlGaAs-heterophotocells operating with lens panels as
the solar energy concentrators, in Proceedings of the all-Union Conference “Photovoltaic
phenomena in semiconductors”, Tashkent, 1989, pp. 305–306 (in Russian)
133. M.Z. Shvarts, V.D. Rumyantsev, V.A. Grilikhes, A.A. Soluyanov, Investigation of a Fresnel
lens concentrator photovoltaic module. Appl. Solar Energy 36(4), 19–28 (2000)
134. Project: INTAS96–1887, 1997-2000 years, “Photovoltaic installation with sunlight concen-
trators”, Final Report (2000)
135. V.D. Rumyantsev, M. Hein, V.M. Andreev, A.W. Bett, F. Dimroth, G. Lange, G. Letay, M.Z.
Shvarts, O.V. Sulima, Concentrator array based on GaAs cells and Fresnel lens concentrators,
in Proceedings of the 16th European Photovoltaic Solar Energy Conference and Exhibition,
Glasgow (2000), pp. 2312–2315
136. V.D. Rumyantsev, V.M. Andreev, A.W. Bett, F. Dimroth, M. Hein, G. Lange, M.Z. Shvarts,
O.V. Sulima, Progress in development of all-glass terrestrial concentrator modules based on
composite Fresnel lenses and III–V solar cells, in Proceedings of the 28th IEEE Photovoltaic
Specialists Conference (PVSC), Anhorage, Alaska (2000), pp. 1169–1172
137. A.W. Bett, C. Baur, F. Dimroth, G. Lange, M. Meusel, S. van Riesen, G. Siefer, V.M. Andreev,
V.D. Rumyantsev, N.A. Sadchikov, FLATCONTM –modules: technology and characterization,
in Proceedings of the 3rd World Conference on Photovoltaic Energy Conversion (WCPEC),
Osaka (2003), pp. 634–637
138. V.D. Rumyantsev, Solar concentrator modules with silicone-on-glass Fresnel lens panels and
multijunction cells. Optics Express 18(S1), 17–A24 (2010)
139. V.M. Andreev, N.Y. Davidyuk, E.A. Ionova, P.V. Pokrovskii, V.D. Rumyantsev, N.A. Sad-
chikov, Parameter optimization of solar modules based on lens concentrators of radiation and
cascade photovoltaic converters. Tech. Phys. 55(2), 277–284 (2010)
140. NYu. Davidyuk, E.A. Ionova, D.A. Malevskii, V.D. Rumyantsev, N.A. Sadchikov, Effect
of secondary lens concentrators on the output parameters of solar modules with cascade
photovoltaic converters. Tech. Phys. 55(7), 1003–1008 (2010)
141. V.D. Rumyantsev, Yu.V. Ashcheulov, N.Yu. Davidyuk, E.A. Ionova, P.V. Pokrovskiy, N.A.
Sadchikov, V.M. Andreev, CPV modules based on lens panels, in Proceedings of the 5th
Forum on New Materials, CIMTEC 2010–12 International Ceramics Congress and 5th Forum
on New Materials, Montecatini Terme, Italy. Advances Science Technology, vol. 74 (2010),
pp. 211–218
142. V.D. Rumyantsev, N.Yu. Davidyuk, E.A. Ionova, D.A. Malevskiy, P.V. Pokrovskiy, N.A. Sad-
chikov, M. Sturm, HCPV modules with primary and secondary minilens panels, in Proceed-
ings of the 6th International Conference on Concentrating Photovoltaic Systems, Freiburg,
2010, AIP Conference Proceedings, vol. 1277 (2010), pp. 97–100
8 III–V Solar Cells and Concentrator Arrays 173

143. V.M. Andreev, V.D. Rumyantsev, N.Yu. Davidyuk, E.A. Ionova, V.R. Larionov, D.A.
Malevskiy, A.O. Monastyrenko, P.V. Pokrovsky, N.A. Sadchikov, Concentrator PV installa-
tions based on modules with Fresnel minilens parquets, in Proceedings of the 25th European
Photovoltaic Solar Energy Conference and Exhibition/5th World Conference on Photovoltaic
Energy Conversion(WCPEC), Valencia (2010), pp. 102–107
144. V.D. Rumyantsev, V.M. Andreev, A.V. Chekalin, NYu. Davidyuk, O.A. Im, E.V. Khazova,
N.A. Sadchikov, Progress in developing HCPV modules of SMALFOC-design. AIP Conf.
Proc. 1556, 185–188 (2013)
145. V.M. Andreev, N.Y. Davidyuk, E.A. Ionova, V.D. Rumyantsev, Photovoltaic modules with
cylindrical waveguides in a system for the secondary concentration of solar radiation. Tech.
Phys. 58(9), 1323–1328 (2013)
146. V.M. Andreev, N.Yu. Davidyuk, D.A. Malevski, A.N. Pan’chak, V.D. Rumyantsev, N.A.
Sadchikov, A.V. Chekalin, A. Luque, New-generation concentrator modules based on cascade
solar cells: design and optical and thermal properties. Tech. Phys. 59(11), 1650–1657 (2014)
147. V.D. Rumyantsev, V.M. Andreev, A.V. Chekalin, N.Yu. Davidyuk, N.A. Sadchikov, HCPV
modules of SMALFOC Design in Versions for PV and PV/T Operation, in Proceedings of the
40th IEEE Photovoltaic Specialists Conference (PVSC), Denver, Colorado (2014), pp. 2720–
2723
148. V. Grilikhes, V. Rumyantsev, M. Shvarts, Indoor and outdoor testing of space concentrator
AlGaAs/GaAs photovoltaic modules with Fresnel lenses, in Proceedings of the 25th IEEE
Photovoltaic Specialists Conference (PVSC), Washington (1996), pp 345–348
149. R. Leutz, A. Suzuki, A. Akisawa, T. Kashiwagi, Developments and designs of solar engineer-
ing Fresnel lenses, in Proceedings of Symposium on Energy Engineering in the 21st Century
(SEE2000), vol.2, 2000, pp. 759–765
150. V.D.Rumyantsev, O.I.Chosta, V.A.Grilikhes, N.A.Sadchikov, A.A.Soluyanov, M.Z.Shvarts,
V.M.Andreev. “Terrestrial and space concentrator PV modules with composite (glass-silicone)
Fresnel lenses” in Proc. of the 29th IEEE Photovoltaic Specialists Conference (PVSC), New
Orleans, 2002, pp. 1596–1599
151. H. Cotal, R. Sherif, The effects of chromatic aberration on the performance of GaInP/GaAs/Ge
concentrator solar cells from fresnel optics, in Proceedings of the 31 st IEEE PV Specialists
Conference(PVSC), Lake Buena Vista, USA (2005), pp. 747–750
152. V.D. Rumyantsev, N.A. Sadchikov, A.E. Chalov, E.A. Ionova, D.J. Friedman, G. Glenn,
Terrestrial concentrator PV modules based on GaInP/GaAs/Ge TJ cells and minilens panels,
in Proceedings of the 4th World Conference on Photovoltaic Energy Conversion (WCPEC),
Hawaii (2006), pp. 632–635
153. E.V. Bobkova, V.A. Grilikhes, A.A. Soluyanov, M.Z. Shvarts, Effect of chromatic aberration
on the concentration of solar radiation by fresnel lenses. Tech. Phys. Lett. 32(12), 1039–1042
(2006)
154. E. Bobkova, V. Grilikhes, A. Soluyanov, M. Shvarts, The method for choosing the optimal
parameters of flat-plane Fresnel lenses, intended for sunlight concentration. Geliotekhnica,
vol. 3, pp. 50–57 (2006) [Translated into English in Applied Solar Energy]
155. V. Weinberg, Optical in the solar power plants (Oborongiz, Leningrad, 1959), p. 256
156. V.A. Grilikhes, M.Z. Shvarts, A.A. Soluyanov, E.V. Vlasova, V.M. Andreev, The new approach
to design of Fresnel lens sunlight concentrator, in Proceedings of the Forth International
Conference on Solar Concentrators for the Generation of Electricity or Hydrogen, El Escorial,
Spain (2007), pp. 49–52
157. L. James, Effects of concentrator chromatic aberrations on multi-junction cells, in Confer-
ence Record of the First World Conference on Photovoltaic Energy Conversion (WCPEC),
Waikoloa, Hawaii (1994), pp. 1799–1802
158. S. Kurtz, D. Friedman, J. Olson. The effect of chromatic aberrations on two-junction, two-
terminal devices in a concentrator system, in Conference Record of the First World Conference
on Photovoltaic Energy Conversion (WCPEC), Waikoloa, Hawaii (1994), pp. 1791–1794
159. S. Kurtz, M. O’Neill, Estimating and controlling chromatic aberration losses for two-junction,
two-terminal devices in refractive concentration systems, in Proceedings of the 25th IEEE PV
Specialists Conference(PVSC), Washingon, DC (1996), pp. 361–364
174 Zh. I. Alferov et al.

160. K. Araki, M. Yamaguchi, Improvement of mismatching in concentrator modules using III–V

cells, in Proceedings of the 17th European Photovoltaic Solar Energy Conference, Munich,
2001, pp. 2187–2190
161. M.J. O’Neill, Color-mixing lens for solar concentrator system and methods of manufacture
and opreation thereof, U.S. Patent 6031179 (2000)
162. K. Nishioka, T. Takamoto, W. Nakajima, T. Agui, M. Kaneiwa, Y. Uraoka, T. Fuyuki, Analysis
of triple-junction solar cell under concentration by spice, in Proceedings of the 3rd World
Conference on Photovoltaic Energy Conversion(WCPEC), Osaka (2003), pp. 869–872
163. A. Yoshida, H. Juso, T. Agui, K. Nakamura, K. Sasaki, T. Takamoto, M. Tanaka, M.
Kaneiwa, K. Okamoto, Characteristics of concentrator triple junction cell optimized for cur-
rent matching, in Proceedings of the 4th World Conference on Photovoltaic Energy Conversion
(WCPEC), Hawaii (2006), pp. 757–759
164. M.Z. Shvarts, V.A. Grilikhes, N.H. Timoshina, A.A. Soluyanov, E.V. Vlasova, Weakening of
the chromatic aberration negative effect on the performance of concentrator multi-junction
solar cells, in Proceedings of the 22nd European Photovoltaic Solar Energy Conference,
Milan (2007), pp. 126–131
Chapter 9
CIGS Thin Film
Photovoltaic—Approaches
and Challenges

F. Kessler, D. Hariskos, S. Spiering, E. Lotter, H. P. Huber and R. Wuerz

Abstract After a short overview of the historical development of the Cu(In, Ga)Se2
(CIGS) thin film solar cell and its special features, we give an overview of the
deposition and optimization of the p-type CIGS absorber as well as the subsequent
n-type buffer layer and the molybdenum back contact. Developments to increase
efficiency by optimizing the implemented bandgap grading via the local gallium
content as well as the specific doping of the CIGS absorber by means of various
alkali elements deepen the understanding. Finally, modern approaches to monolithic
cell interconnection are discussed, e.g. via the use of modern ultra-short pulse lasers.
Numerous literature references enable the reader to gain an even deeper insight if
required.

F. Kessler (B) · D. Hariskos · S. Spiering · E. Lotter · R. Wuerz

ZSW – Zentrum für Sonnenenergie- und Wasserstoff-Forschung Baden-Württemberg,
Meitnerstr. 1, 70563 Stuttgart, Germany
e-mail: [email protected]
D. Hariskos
e-mail: [email protected]
S. Spiering
e-mail: [email protected]
E. Lotter
e-mail: [email protected]
R. Wuerz
e-mail: [email protected]
H. P. Huber
Laserzentrum Hochschule München, Fakultät für angewandte Wissenschaften und Mechatronik,
Hochschule für angewandte Wissenschaften München, Lothstr. 34, 80335 Munich, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 175

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_9
176 F. Kessler et al.

9.1 Introduction

9.1.1 Brief History

Harry Hahn et al. were the first to synthesize the ternary chalcopyrite CuInSe2 (CIS)
and some others in the laboratory [1]. This first CIS (CuInSe2 ) material was grown
by mixing two binary and stoichiometric educts at a temperature of 900 °C for 12 h.
They determined the lattice constants by the Debye-Scherrer method and focused
on structural analysis of the new material. It took about 20 years (1974) for a team
from Bell Laboratories Holmdel to investigate the electronic and optical properties
in more detail. The Wagner, Shay, Migliorato and Kasper group produced the first
heterostructures consisting of a p-type CuInSe2 absorber and an n-type CdS layer [2].
They were looking for a broad-band detector in the visible and NIR as an alternative
to Si or CdS/Cu2 S junctions. The CuInSe2 single crystals were produced by melt
growth from stoichiometric mixtures, which were subsequently annealed under Se
atmosphere at 600 °C for 24 h to achieve p-type conductivity. The n-type CdS layer
was evaporated to a thickness of 5–10 μm. Indium on CdS and gold on CuInSe2
were used as ohmic contacts. After all the processes, the solar cell efficiency was
around 5%.
Just one year later, the Wagner et al. group stated that they had achieved a
PV conversion efficiency of 12% with the same heterostructure [3]. An important
leap forward has been taken by leaving the approach of manufacturing PV devices
from thick and single-crystalline CuInSe2 absorbers. Instead, some scientists dis-
covered the real potential of this direct bandgap CIS absorber which should require
only a few micrometers to absorb the entire solar spectrum above its bandgap of
E g = 1.01–1.04 eV. Kazmerski et al. [4] were the first to fabricate thin film CIS/CdS
devices. The CdS layer (d = 6 μm) and the CIS absorber (5–6 μm) were still quite
thick and reached a conversion efficiency of 4–5%. But Mickelsen and Chen [5]
(Boeing Aerospace, 1981) achieved a breakthrough. They attained a CIS/CdS effi-
ciency of 9.4% by using thin film vacuum deposition processes for all functional
layers and a total stack thickness of only 5 μm.
Many other scientific discoveries such as the introduction of Ga to increase the
bandgap from about 1 eV toward the optimum value of 1.3–1.5 eV (depending on
the sun spectrum), the installation of a bandgap grading and the defined admixture of
alkali elements (“dopants”) could boost the cell efficiency of Cu(In,Ga)Se2 (CIGS)-
based solar cells up to values around 23%. The CIGS-based solar cells have not
only survived the ups and downs of PV from 2000 to 2012, that is, the unrealistic
expectation of market growth, but also established competition in Si technology
and other thin film PV technologies such as CdTe and a-Si:H-based approaches.
Nevertheless, the production volume of CIGS modules is still around 2%, whereas
Si dominates the scenario with a share of >90%. The optimistic assumptions of
Chopra et al. [6] and Mitchell et al. [7], according to which the manufacturing costs
of thin film devices from 2004 would be lower than those of Si thick film technology,
could not be verified at least until 2004.
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 177

Advantageous features of CIGS

There are numerous reasons explaining why CIGS is currently one of the most
promising PV technologies:
– CI(G)S is a semiconductor with direct bandgap and high optical absorption coef-
ficient in the range of 105 cm−1 , allowing a low absorber thickness of 2–3 μm or
even less if multi-reflection technologies are applied.
– Adjustable graded bandgap between 1.04 and 1.14 eV without degradation of
optoelectronic properties.
– Dopability via alkali metals, especially Na and K.
– Short energy payback time of about 1 year.
– Microcrystalline absorbers with saturated defects and electrical properties that can
be even better than a single crystal.
– Large variety of deposition methods, both in vacuum and vacuum-free.
– The typical substrate structure allows for the free choice of substrates from glass
(typically float glass) to polymer or metal foils. Cost-effective coating approaches
such as roll-to-roll in vacuum or vacuum-free can be used; energy savings can be
done through low heat capacities of the thin functional layers and/or the use of
thin substrates.
– The manufacture of highly flexible, lightweight and material-saving devices
enables optimum integration into facades (BIPV), the automotive sector, aerospace
applications and leisure applications.
– The possibility to fabricate heterojunction structures and tandems, to modify
device properties by interface modifications and intermediate thin functional lay-
ers.
– Cell connections via monolithic connections enable the production of customized
modules with minimal effort (only possible on insulating substrates/foils); how-
ever, if cell sorting is desired, other techniques can also be used; for example,
shingling of cells (metallic substrates), wiring and screening technology.
– Very high stability in space against high-energy electrons, making CIGS generators
a suitable candidate for special space missions in highly radiant regions [8, 9] (e.g.
the Van Allen belt).

9.1.2 Structure of a CIGS Solar Cell

CI(G)S crystallizes in a tetragonal chalcopyrite lattice (Fig. 9.1). The ratio of the
lattice constants c/a is close to 2. The unit cell can be deduced from the face-centered
cubic lattice of ZnSe (II–VI group), where Zn is substituted by Cu (group I) and In
(resp. Ga) (group III) ordered according to the Grimm-Sommerfeld rule. In this way,
the CIS or CIGS unit cell can be constructed by two sphalerite unit cells. Each Se
atom is surrounded by four metal atoms (Cu, In/Ga), which form a tetrahedron [10].
178 F. Kessler et al.

Fig. 9.1 Tetragonal

chalcopyrite unit cell of
CI(G)S

CI(G)S thin films with a thickness of typically 1–3 μm are multicrystalline and
range from very small crystallites 1 μm to crystallites of a few micrometers diam-
eter. Attempts and expectations to improve the quality of optoelectronic materials
by using single crystals failed and no longer make sense, as the grain boundaries
can be passivated very successfully. It is even not straightforward to give substan-
tial evidence that absorbers with smaller grain size must result in devices with worse
optoelectronic properties, at least if the grains are not too small and the passivation of
grain boundary states (e.g. by alkali doping) was successful. Multicrystalline CIGS
thin film absorbers are inherently p-type which is mainly caused by Cu vacancies.
However, early investigations on CIS report the existence of both n- and p-type single
crystals, depending on growth conditions [11].
The structure of a typical CIGS-based solar cell is shown in Fig. 9.2. Owing to
the relatively low total thickness of the active layers of approx. 4 μm, the solar cell
or module requires a suitable carrier substrate that meets all technical and economic

Fig. 9.2 Sketch of the cell

structure (upper picture) and
corresponding band diagram
of a “standard” CIGS type
PV device (lower picture)
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 179

requirements such as high temperature stability of 400–700 °C without degassing or

decomposition, low roughness with regard to the layer thicknesses and commercial
availability at reasonable costs. Standard float glass, a mass product widely used in
the window industry, is a pretty good compromise, especially when it is not about high
weight and brittleness. In addition, float glass or soda-lime glass has a reproducible
and extremely low average roughness (Ra ) of a few nanometers. In combination with
its relatively hard surface and high optical transparency, glass is also an ideal material
for monolithic cell connection by laser and mechanical scribing.
According to Fig. 9.2, the active stack deposition normally starts with the molyb-
denum (Mo) back contact, followed by a p-type absorber layer (CIGS) , an n-type
buffer layer (e.g. CdS, Zn(O,S), In2 S3 or others), a thin high-resistance undoped zinc
oxide layer (i-ZnO) and a highly n-type Al-doped ZnO (ZnO:Al or AZO) layer as
transparent conductive oxide (TCO) front contact. The high AZO thickness of up to
1 μm in Fig. 9.2 is used to achieve a sufficient lateral conductivity for the fabrica-
tion of monolithically integrated modules (normally) without a metal grid on top.
However, to verify the highest efficiency at small cell level, the ZnO:Al thickness
is reduced to approx. 200 nm and a fine Ni–Al–Ni grid is applied for the transport
of the charge carriers. An anti-reflective layer of MgF2 with a typical thickness of
105 nm reduces the reflection losses.
CIGS-based thin film devices are preferably manufactured in “substrate configu-
ration” (see Fig. 9.2) to protect the buffer and window layers from damage from the
CIGS process temperature (400–700 °C). In substrate structures, light enters through
the finally deposited layer (here: ZnO:Al), in contrast to superstrate configurations
such as CdTe-based devices, in which the photons first penetrate into the PV device
via the transparent substrate. The advantage of the superstrate approach is that opaque
films/carriers can also be used as carrier material. The beneficial consequences are
manifold and concern both the fabrication procedure (e.g. roll-to-roll manufacturing
on stainless steel foil or polyimide) and the end product (weight, brittleness, flexibil-
ity). A substrate configuration also offers a wide range of possibilities when looking
at hetero tandem structures with (e.g.) CIGS as back cell and perovskite as front cell.
The lower picture of Fig. 9.2 indicates the direction of the electron minority
carriers in the conduction band and the hole majority carriers in the valence band, as
well as the space charge region (SCR). The little spike in the conduction band is a
result of the CdS buffer layer. However, the potential electron-blocking behavior
is only observed at low temperatures. Igalson et al. [12] showed by admittance
spectroscopy, dark and light I–V–T analysis and C–V profiling that the electron
barrier at the front is significantly more pronounced for CdS buffers compared to
In2 S3 buffered samples.
For instance, the I–V curves at 120 K showed a pronounced kink in the first
quadrant (roll-over effect), which was not observed with In2 S3 buffers. The small
hump near the Mo back contact is due to the formation of a thin MoSe2 layer [13, 14]
which is responsible for the back-surface field formed for the electrons and holes. A
further discussion can be found in Sect. 9.2.2.1.
180 F. Kessler et al.

9.2 Device Fabrication

Some advantageous features of CIGS-based PV modules are:

• The possibility to grow the absorber by a comparably large number of thin film
deposition methods, including both non-vacuum and vacuum approaches.
• The possibility of choosing very different carrier materials and thus being able
to supply PV generators for almost all conceivable areas of application such as
conventional applications (on-roof and free-field), well-coordinated building inte-
gration (BIPV), automotive applications (cars, truck tarpaulins), leisure activities,
aviation and space applications.
• Material and energy savings in production by reducing the absorber thickness and
optimizing multiple reflection effects.
• The continuing increase in generator efficiencies.
• The potential to fabricate tandem devices, for example, in a perovskite/CIGS struc-
ture with CI(G)S as the bottom cell; even CIGS/CI(G)S tandem structures may
be a future option consisting of a high and low bandgap CIGS device. However,
some obstacles still need to be overcome for both approaches.
The multitude of implementation options and diversification, on the other hand,
carries the risk that optimization processes for a certain product would take con-
siderably longer than with a PV concept with only one meaningful manufacturing
process, which is developed and applied in parallel by all users.
The core process bearing the most profound know-how within a CIGS-based
PV device is the deposition of the p-type CIGS absorber. To produce a highly effi-
cient solar cell, however, a deep understanding of all other layers is also necessary,
especially the adjacent layers such as buffer and back contact. In addition to the bulk-
correlated layer properties, particular attention must be paid to the CIGS interfaces,
which can significantly hamper the carrier transport, for example, by implementing
increased recombination states in the gap or the formation of barriers in the con-
duction or valence band, which impede the transport of charge carriers (see also
Fig. 9.2). Especially the optimization of the pn-junction, that is the p-CIGS/n-buffer
interface, is complex, especially if a Cd-free alternative buffer has to be applied due
to environmental reasons. The following discussions are only intended to show how
small changes can have a big impact.

9.2.1 CIGS Layer

9.2.1.1 Phase Diagram, Copper and Gallium Content

Figure 9.3 shows the phase diagram along the pseudobinary Cu/In/Se system
Cu2 Se–In2 Se. A detailed discussion of the ternary Cu/In/Se phase diagram including
the history on the way to calculation and verification of this system was provided by
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 181

Fig. 9.3 Pseudobinary

phase diagram of the ternary
Cu/In/Se system along the
Cu2 Se and In2 Se3 axis
(original graphic by Chang
et al. [20], later revised by
Stanbury et al. [15] and in
this publication)

Stanbery [15]. The Cu/In/Ga/Se diagram is even more complicated and less under-
standable, but usually the ternary Cu/In/Se system is also useful to explain the quater-
nary system. The desired crystalline chalcopyrite region suitable for the production
of efficient thin film solar cells is the p-type phase (CuInSe2 ). The phase is stable from
room temperature to approx. 810 °C and is only present at a substoichiometric Cu
content of 24–24.5% [16]. The copper deficient n-type β-phase CuIn3 Se5 should be
avoided or reduced to a minimum in the bulk, in order to achieve a homogeneous and
strongly p-type semiconductor. At the interface to the n-type buffer/TCO interface,
however, the n-type and β-phase CuIn3 Se5 (resp. Cu(In,Ga)3 Se5 ) or the so-called
ordered vacancy compound (OVC or ordered defect compound, ODC) is intended
and very helpful to provide a proper hetero interface. The OVC layer is automatically
formed at high temperatures in the direction of the interface to the n-type CdS buffer.
This buried pn junction in the absorber reduces recombination at the heterogeneous
CI(G)S/CdS interface [17, 18]. Another beneficial effect of Cu poor grown absorbers
as postulated by Schock, Rau and Turcu [14, 19] is the Fermi level pinning close to
the conduction band and the bandgap widening at the absorber surface. In summary,
as a consequence of type conversion, bandgap widening and Fermi level pinning at
the interface, the heterojunction losses of Cu-poor CIGS absorbers are low and the
recombination essentially takes place in the space charge region (SCR).
The parameter range for the fabrication of an efficient chalcopyrite absorber is
rather limited, as demonstrated in Figs. 9.3 and 9.4 [21] by the PV-active α-phase.
For the fabrication of suitable PV absorbers the Cu(In,Ga)Se2 layers should
be slightly Cu deficient as the Cu vacancies contribute to the p-type conductiv-
ity. Devices with the potential of high efficiencies should have a Cu content in the
range of 0.88 < Cu/(In + Ga) < 0.95 [23]. However, if all Cu vacancies corresponded
to a doping state, the defect or dopant concentration would be >1020 /cm3 and thus
more than three orders of magnitude higher than any reasonable doping level. Addi-
tionally, only a part of the excess vacancy states can be compensated by the higher
182 F. Kessler et al.

Fig. 9.4 Simplified ternary Cu/In/Se phase diagram indicating the α-phase region for the fabrication
of PV-relevant CIS-material by a solid bold line (original diagram by Gödecke et al. [22], redrawn
in this publication). The dashed axis corresponds to the phase diagram of Fig. 9.3

Cu deficiency of the Cu(In,Ga)3 Se5 OVC layer. Its thickness increases with the Cu
deficiency of the bulk material and varies between 5 and 60 nm [24]. The main
reason, however, describing the electronically extremely tolerant behavior of CIGS
is not the OVC layer, but the formation of inactive and neutral defect complexes.
Defect complexes consisting of two Cu vacancies (2 VCu ) and an In-Cu anti-site
(InCu ) are particularly suitable candidates that allow a Cu deficiency as low as 22%,
as it results from measurements [14] and calculations [25]. The defect chemistry
of CIGS and its derivatives is complex. However, understanding the interactions
between charged and neutral defects under illumination and temperature activation
is necessary to describe all experimentally observed effects and to further improve
efficiency and reduce metastability. A detailed description of the defect models can
be found elsewhere [26–30].
According to the early detailed balance calculations of Shockley and Queisser
[31] (Shockley-Queisser limit) the ideal bandgap of a single-junction cell should be
in the range of 1.3 eV. Later, their estimates of black-body solar radiation with a
surface temperature of 6000 °C and the assumption of pure radiation recombination
were refined by the introduction of the AM1.5g spectrum, which better describes
the solar spectrum on Earth, corresponding to a lower integrated light intensity/m2
by slightly more than a factor of 1.5 [32]. If small fluctuations of the maximum
efficiency are neglected by different input parameters (e.g. assumed spectrum), it is
obvious that the CuInSe2 band spacing should be increased from 1.04 eV to slightly
above E g = 1.3 eV. Hanna et al. [33] investigated the bandgap E g as a function of
the Ga content x = GGI = Ga/(Ga + In) from x = 0 (CuInSe2 )) to x = 1 (CuGaSe2 ).
Up to a Ga content of x = 0.3, the bandgap was not only increased to E g 1.2 eV,
but the bulk defect concentration was also reduced to a minimum. The number of
defect states depends on the CIGS deposition method but their minimum is always
at x ≈ 0.3. The evaluation of the losses of the open-circuit voltage V oc relative to
E g by applying the difference E g − qV oc versus E g could show an analog minimum
of E g − qV oc at E g ≈ 1.2 eV [33]. Up to a Ga concentration Ga/(Ga + In) of x <
0.3, the open-circuit voltage V oc increases superlinear but sublinear with x > 0.3.
With x > 0.3 [34] the defect concentration in the bulk rises. Thus, the admixture
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 183

of Ga is limited. Nevertheless, Jackson et al. [35] showed that high IV conversion

efficiencies of η ≥ 19% can be achieved with Cu(In(1−x) ,Gax )Se2 based devices
up to a Ga content of 0.21 ≤ Ga/(Ga + In) ≤ 0.38 corresponding to a bandgap of
1.11–1.23 eV, which demonstrates that the materials tolerance is higher than expected
due to the defect concentrations [33]. Other authors [36] could even slightly expand
the useful bandgap region up to 1.35 eV.
Cu(In,Ga)Se2 based thin film solar cells have already achieved efficiencies close
to 23% with the two main and currently most industry-relevant deposition meth-
ods, such as co-evaporation [37] and sequential sputtering plus annealing [38]. The
multicrystalline CIGS semiconductor shows a rather good-natured behavior and
high-efficiency cells >20% can be achieved not only by large grains over the whole
absorber thickness but also by a structure with many small grains. Fortunately, the
grain boundaries can be easily neutralized and passivated (e.g. by alkali supply) and
do not hamper the device. Some mechanisms and details on how to optimize the
CIGS absorber by treatments with alkali metals are described in Sect. 9.2.1.5.
Many attempts have been made to increase the bandgap toward 1.4 eV and above
in order to improve the efficiency or to develop a wide-bandgap material to be applied
as front cell of a tandem device. Experiments are still ongoing and physically chal-
lenging. After all, a sulfur-based Cu(In,Ga)S2 solar cell with a relatively high V oc
of 920 meV has achieved a maximum efficiency of 15.5% [39]. Best efficiencies
with absorbers of CuInS2 (E g = 1.53 eV) reached an efficiency of 11.4% [40]; with
CuGaSe2 (E g = 1.68 eV) the record efficiency is 11.9% [41]. Goto et al. [42] reported
on the improvement of a CuInS2 absorber by adding a CuGaS2 seed layer prior to the
CuInS2 . With the absorber formed after some sulfurization and KCN etching steps,
a cell efficiency of 13.0% could be demonstrated (V oc = 690 meV).

9.2.1.2 CIGS Co-evaporation

The CIGS co-evaporation of the pure elements copper, indium, gallium and selenium
at an elevated substrate temperature of 400–700 °C is one of the first and still most
successful deposition methods. The layer composition can be adjusted relatively
precisely by the quantity and time of the evaporated elements Cu, In, Ga and Se. The
contamination of the absorber depends essentially on the purity of the evaporated
elements. Excess selenium is provided over the entire coating time. The so-called
three-stage process introduced by Contreras et al. (NREL) [43] at laboratory level
gained great importance in the CIGS community, as reproducible results with high
efficiency could be achieved using this formulation. The basic idea was to grow the
absorber via a Cu-rich phase and to finish the film with a Cu poor phase, so that the
final absorber is slightly Cu-deficient. The original flux and temperature profile of
the metals over time is shown in Fig. 9.5. The evaporation begins with a Cu-free first
deposition phase at a moderate temperature of 300–350 °C. During the second phase
the temperature increased to 500–560 °C, while the metal flux is dominated by Cu.
After the Cu-rich second phase the deposition is terminated by In and Ga evaporation
184 F. Kessler et al.

Fig. 9.5 Flux profiles of a

three-stage CIGS process
versus deposition time
(original diagram by
Contreras et al., redrawn in
this publication [43])

converting the absorber back to a Cu-poor total composition. Se is present in excess

in all phases, including the cooling phase.
Another finding of NREL in connection with the three-stage CIGS processing was
the discovery of the beneficial effect of a double profiling structure of the absorber
composition, that is, variation of the bandgap (via the Ga content) over the thickness.
A simplified and schematic notch structure is shown in Fig. 9.6. The band is bent
essentially within the conduction band (CB) and corresponds to a bandgap minimum
in the bulk toward the front interface. The highest bandgap is implemented at the back
contact and a slightly enhanced E g at the front interface in order to reduce interface
recombination of minority carriers. The minimum bandgap at the notch (E gn ) is
necessary to absorb low energy photons and corresponds with the photogeneration
gap and thus the short circuit current density (J sc ) [44]. It is advantageous that the
enlarged gap in the space charge region (E gSCR ) toward the front side is correlated with
the open-circuit voltage (V oc ). The difference between the photogeneration distance
and the electrical distance (E gSCR − E gn ∼ = 100 meV) is therefore the performance
gain resulting from the double grading.
Based on the double grading and in conjunction with the three-stage processing,
many institutes [45–48] were able to produce CIGS solar cells with high efficiency.
By varying and fine-tuning the bandgap profile (or Ga/(Ga + In)) highest efficiencies
>20% could be demonstrated [49, 50], especially at low (<500 °C) temperature on
polyimide substrates [51].
The bandgap formation (resp. Ga/(Ga + In)) and thus the multistage deposition
formulations with Cu-rich phase were modified in various ways, for example, by
introducing a double-notch [48], widening of the notch width, or shifting the notch
forward or backward and so on. Tests with differently shaped Ga double profiles show
that different shapes can be selected to achieve high efficiencies ≥20.0% (Fig. 9.7).
Moreover, the achieved efficiency is not only dependent on the Ga gradient. For
example, the right alkali elements must be added to the absorber in the right quantity

Fig. 9.6 Simplified band

diagram with notch profile as
introduced by NREL first
(adopted from Contreras
et al. [43], redrawn in this
publication)
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 185

Fig. 9.7 CIGS solar cells

with different double-graded
Ga/(Ga + In) profiles and
efficiencies ≥20% [50, 52].
A deeper notch, which is
shifted further forward,
seems to be advantageous
(dotted curve)

and at the right time (see Sect. 9.2.1.5). Further points to consider are: the interfaces,
especially the band alignment to the buffer layer, spikes or cliffs in the CB or defect
states impeding the carrier transport.
Owing to the complexity of the three-stage process with regard to production,
alternative multistage or single-stage evaporation processes, which are more com-
patible with inline processing, were pushed forward. Salomé et al. [53] compared
single-stage and three-stage CIGS depositions with the same total Ga content and
confirmed the typical expectations: The three-stage sample consisted of larger grains,
had a 28 meV higher V oc , a 0.1 mA/cm2 higher J sc and an efficiency of 17.0%. The
efficient three-stage sample was 0.7% higher than that of the single-stage sample. As
a result, however, the differences were surprisingly low. In addition, the notch of the
three-stage sample was quite deep (too deep) in the absorber so that its full benefit
could not be unfold. The single-stage device showed a quasi-linear increase in Ga
content from front to back and crossed the Ga/(Ga + In) curve at a depth of 0.7 μm.
The co-evaporation process offers the possibility of a precise determination of the
absorber composition and the local Ga content already during the film growth. Doping
elements can also be added during the desired growth phase. Another important
parameter is the temperature profile of the substrate which controls the interdiffusion
of the elements. The main challenges in production are process control, acceleration
of deposition rates, large-area efficiencies and process yield to reduce costs. In most
cases, therefore, no real three-stage process is preferred for production, but rather
various modifications from single-stage inline to multistage processes of all kinds.
Development is not yet complete.
In an early development stage, the focus was not only on controlling material
flows and film growth, but also on the commercial availability of suitable evaporation
sources for large-area inline deposition. In the meantime, the deposition width is no
longer limited by the sources, and various tools such as in situ microbalances and
AAS measurements have proven themselves for flow control [54]. In continuous
inline production, XRF can be used indirectly to control film composition through a
loop-back mechanism in conjunction with the evaporation sources. Other advanced
process control tools such as laser light scattering (LLS), Raman or time-resolved
photoluminescence are discussed by Scheer et al. [55] and can also be used for all
CIGS coating processes.
186 F. Kessler et al.

9.2.1.3 CIGS by Sequential Deposition and Subsequent Annealing

The sequential precursor separation and the subsequent selenization and annealing is
the second main approach for the separation of CIGS absorbers on an industrial level.
The driving force behind these developments was the search for a large area, com-
patible CIGS coating method that enables the reproducible and easily controllable
growth of homogeneous layers. The sequential approach can be traced back to the
early 1980s and is based on the developments of ARCO Solar, Inc. (ASI) [56], where
the metallic precursors (Cu,In) have been deposited by electrodeposition. After the
evaporation of selenium, the stack was annealed. Further development steps were the
introduction of gallium, the conversion from galvanic deposition to sputtered precur-
sors and the improvement of selenization through the use of gaseous H2 Se as a Se
source. In the early 1990s, Siemens Central R&D group (next owner and successor
of ASI) developed a rapid thermal processing (RTP) with a vaporized Se precursor
and finally Siemens Solar Industries introduced an additional “sulfurization after
selenization” process (SAS) [57].
This SAS processing by H2 Se and H2 S annealing of a sputtered Cu–In–Ga pre-
cursor is currently one of the most advanced deposition methods which has been
continuously and very successfully optimized within the last few years by Solar
Frontier K.K. (established 2006). In 2017 Solar Frontier K.K. had reached a produc-
tion capacity of >1 GW on glass substrates.
Physically and with a view on process control, the selenization and sulfurization
via H2 Se and H2 S is sufficient and works well [58]. The record efficiency of 22.9%
[38] (cell area: 1 cm2 ) achieved in 2017 by Solar Frontier is based on this technology.
However, since H2 Se is highly toxic (much more than H2 S) and highly flammable,
other supply methods such as Se vapor or the deposition of a Se layer on the top
side have been extensively tested [59–63]. Avancis, another CIGS manufacturer on
glass, starts the absorber deposition with a low temperature DC magnetron sputtered
Cu–In–Ga stack with an evaporated Se layer on the top [57, 63]. The second step is
a high temperature RTP treatment (≤550 °C) in sulfur-containing gas, which forms
the Cu(In,Ga)(S,Se)2 absorber, which is very similar to the Solar Frontier process
but less dangerous due to the Se supply by elemental selenium. The disadvantage of
using elemental Se vapor is that Ga accumulates at the back contact and thus forms
a Ga-free low bandgap absorber at the front after the RTP process. A method for
overcoming this challenge was described by Schmidt et al. [64].
Both the admixture of gallium and sulfur to the CuInSe2 matrix increases the
bandgap. In the gaseous sulfur treatment by the RTP process, the S atoms are replaced
[65] by the Se atoms near the front surface and form a kind of double or multi-stage
bandgap over the absorber thickness, similar to the double sorting described above for
the highly efficient co-evaporated CIGS. While the substitution of In by Ga essentially
affects the conduction band (CB), the addition of sulfur mainly influences the valence
band (VB) [65]. Owing to the S-conditioned widening of the bandgap to the front,
the VB is pulled downwards, whereby the photogenerated holes are rejected from the
front interface and thus a carrier recombination at the interface is prevented. Another
beneficial effect is the increased V oc compared to samples without sulfurization.
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 187

9.2.1.4 Alternatives

CIGS absorbers were produced not only by co-evaporation of the elements, but
also by evaporation from binary, ternary or quaternary sources. Flash evaporation
and pulsed laser deposition were investigated with the aim of improving deposition
control. Some other alternative methods have also been investigated and are described
in the overview of Kemell et al. [24] but will not be described here. Perhaps the third
most important deposition method is the galvanic deposition of the elements or binary
compounds, followed by a selenization and annealing step [66] (two-stage process).
Although the single-step deposition of CuInSe2 has been investigated over a long
period of time [67–69], two-stage processing has finally proven superior by better
control of composition and performance [70, 71]. Before the closure of NEXCIS
(France) in 2015 the company had achieved an efficiency of 14.00% on 60 cm ×
120 cm [70] and 17.3% [70] on small area cell level with a two-stage process. The US
company Solopower [71] is able to fabricate flexible electrodeposited CIGS modules
at a length of 1.5 m on metal foil with efficiencies >13%.
Other non-vacuum and solution [72] or nanoparticle-based deposition processes
[73] may also have reasonable efficiencies, but the process control and absorber com-
position is still less mature to compete with electrodeposition, sequential sputtering
and co-evaporation. Compared to vacuum-based processes, however, the material
utilization in solution or galvanic processes is significantly higher. In addition, the
non-vacuum investments for production are significantly lower. Since module effi-
ciency has a direct and linear effect on production costs, its effect is often much
more serious than reducing the material costs of a single layer by a few percent, as it
accounts for only a part of the total module costs. The share of each layer in the total
costs depends on the applied CIGS approach, the country of production, abundance
of materials and many other things. For instance, according to the estimations of
ISET [74] for its ink approach, the expenses for the glass substrate plus cover glass
amounts to 32% of the total materials costs. A discussion of manufacturing costs of
glass–glass CIGS devices is given (e.g.) by Horowitz et al. [75].

9.2.1.5 Alkali Doping

In contrast to silicon-based solar cells, alkali metals show a positive impact on the
performance of CIGS solar cells. The positive effect of sodium (Na) on cell efficiency
was first demonstrated in 1993 when CIGS solar cells were deposited on sodium oxide
containing glass substrates [76]. The main effect of Na is a strong increase in the
open-circuit voltage V oc and the fill factor FF [77, 78] caused by an increase in the
net acceptor concentration [79]. For flexible or Na-free substrates, an external supply
of Na is necessary for high cell efficiencies. Here Na can be supplied before CIGS
growth by a Na-doped Mo layer [80], by a NaF precursor layer [77], during the CIGS
growth by NaF co-evaporation [81] or after the CIGS growth by NaF post-deposition
treatment (PDT) [78]. The time at which Na is applied has a strong impact on the
growth of the CIGS layer: Since Na hinders the interdiffusion of Ga and In, steeper
188 F. Kessler et al.

Ga gradients and smaller grains are obtained if Na is present during growth [81].
In addition, Na impedes the annihilation of stacking faults [82] so that in case of
temperature-sensitive substrates such as polyimide, where the CIGS layer must be
grown at low temperature, the CIGS should be first grown alkali-free and then Na
should be added by a NaF-PDT process [78]. It is not yet clear whether the positive
effect of Na is a grain boundary-related [83, 84] or a bulk-related [85, 86] effect.
Atom probe tomography measurements (APT) revealed a strong segregation of Na
at grain boundaries up to a concentration of 1.7 at.% [87]. On the other hand, Na was
also found in the bulk of grains up to a concentration of 60 ppm or 2.6 × 1018 cm−3
by APT [87] which is much higher than the typical doping concentration of 1 × 1015
− 1 × 1016 cm−3 in Na-doped CIGS layers. Na was thus able to passivate the grain
boundaries and dope the bulk of CIGS layers. The influence of the heavier alkali
elements potassium (K) and cesium (Cs) on the cell performance was investigated
as early as 1997, but was small compared to Na [88] and not the subject of research.
In 2012, potassium was re-discovered as an efficiency booster.
Within the framework of an EU project (hipoCIGS), R. Wuerz (ZSW) observed
that CIGS solar cells deposited on enamelled steel showed higher efficiencies than on
soda-lime glass (SLG), whereas SLG was previously considered the substrate with
the highest efficiency potential. Potassium was found as the reason for the efficiency
increase, as the enamel layer provided much more potassium than the SLG substrate,
while the supply of Na was comparable for both substrates. A detailed analysis of the
subsequent incorporation of potassium via a KF-PDT treatment was first performed
by Chirilă et al. (EMPA) [51] in 2013. At the same time and independently, the
influence of KF-PDT on the absorber properties was investigated at ZSW [89].
KF-PDT increased cell efficiency in a similar way to NaF-PDT [89]. The appli-
cation of KF-PDT on a Na-doped layer resulted in a record efficiency of 20.4% on
flexible polyimide film [51], which is still (2018) the highest efficiency for a thin film
solar cell on a flexible substrate. When applying the KF-PDT on Na-doped CIGS
layers on SLG, the cell efficiency was further ramped up to 20.8% [90], 21.0% [91]
and 22.3% [92]. The KF-PDT is also effective for sulfur-containing absorbers based
on Cu(In,Ga)(S,Se)2 [92]. It has further improved the efficiency of the modules in a
full-scale industrial production process [93]. The main effect of the KF-PDT is an
increase in the open-circuit voltage V oc and fill factor. This effect can be explained
by various experimental findings.
The potassium treatment:
(1) leads to a strong depletion of copper (Cu) at the surface of the CIGS layer [51]
and to the formation of K–In–Se surface species [94].
(2) leads to a widening of the surface bandgap of the CIGS layer to 2.52 eV [95],
which corresponds well to the bandgap of KInSe2 with E g = 2.68 eV [96].
Hence, formation of KInSe2 at the surface of the CIGS layer due to the KF-
PDT seems to be likely.
(3) reduces surface recombination [97, 98], due to the changed surface composition
(4) allows a reduction of the CdS layer thickness without loss at fill factor [51] and
thus increases quantum efficiency in the UV range
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 189

(5) enhances the charge carrier density of the CIGS layer [99]
(6) increases carrier lifetime [99, 100]
(7) increases the photoluminescence intensity [99], indicating reduced non-
radiative recombination
(8) reduces band fluctuations [100].
A combined NaF- and KF-PDT proved to be more effective in increasing cell
efficiency than a single NaF or KF-PDT [97].
In the meantime, PDT has also been successfully extended to the heavier alkali
elements Rb and Cs. Efficiency was even increased to 22.6% [37] and 22.9% [38],
respectively, through RbF and CsF PDT. APT measurements showed that Rb strongly
segregates at grain boundaries (GB) and that Rb pushes Na and K out of GBs [101].
Further, nanoXRF (X-ray fluorescence) measurements combined with electron back
scatter diffraction (EBSD) have demonstrated, that Rb only segregates at random
GBs, but not at 3 GBs [102]. Time-resolved photoluminescence measurements
revealed a strong increase of carrier lifetime due to the RbF PDT [103]. Furthermore,
the formation of an RbInSe2 interfacial layer at the CdS/CIGS interface was supposed
to be the reason for a strong increase in V oc by the RbF-PDT [104]. A combination
of light (Na) and heavy (Rb, Cs) alkali metals is therefore necessary to achieve
maximum cell efficiencies.

9.2.2 Back and Front Contact

9.2.2.1 Mo Back Contact

Molybdenum has been identified as one of the most suitable back contact materials
since the beginning of the development of CI(G)S solar cells. It meets all the required
specifications—such as sufficient stability in selenium and sulfur atmosphere, rela-
tively stable and inert at high temperatures, formation of a low-resistance contact to
the absorber [105], virtually no reactions with Cu, In, Ga at CIGS growth conditions
and no diffusion of Mo into the absorber. Furthermore, the high material hardness
enables mechanical scribing (second and third scribe) on glass substrates for mono-
lithic interconnection without impairing the glass matrix. On the other hand, the Mo
layer is permeable for diffusion of sodium from the soda-lime glass (SLG) into the
growing CI(G)S. Thus, Na doping of the absorber can be controlled at least partially
by the substrate temperature and the Mo properties.
Molybdenum has a relatively low coefficient of linear thermal expansion (CTEMo :
4.8–5.9 × 10−6 K−1 ) compared to the soda-lime glass substrate (CTESLG : ≈9 × 10−6
K−1 ) or the absorber. The experimentally determined linear CTE of CuInSe2 [106]
at T = 300 K according to the work of Paszkowicz et al. is CTEL = 8.62 × 10−6 K−1 .
Along the a-axis of the chalcopyrite crystal the thermal expansion coefficient is 9.15
× 10−6 K−1 , and along the c-axis it is 7.61 × 10−6 K−1 . The strong difference in CTE
can lead to adhesion problems during the high-temperature deposition process of the
190 F. Kessler et al.

CIGS layer or bending of the glass substrate due to stresses. The Mo back contact is
normally deposited by e-gun evaporation or sputtering. Especially the DC magnetron
sputtering from planar or rotatable targets is the method of choice today due to its
excellent suitability for homogeneous industrial large-area deposition. The intrinsic
properties of the Mo layer can be modified via the sputtering parameters, especially
the gas pressure [107]. Low-pressure sputtering leads to denser material with high
conductivity and compressive stress, while comparatively high-pressure sputtering
leads to tensile stress, lower density and low conductivity. These properties can be
used to adapt the Mo layer to the respective substrate, for example, to ensure sufficient
film adhesion without cracks or delamination or to ensure the correct amount of Na
diffusion through back contact, that is, to adjust the barrier properties. Often a Mo
bilayer is deposited, starting with a more flexible and less dense layer to compensate
for CTE mismatch between the substrate and the back contact, improving adhesion
to the substrate, followed by a denser, that is at lower pressure deposited layer of high
conductivity. In principle, the back contact can be adapted to all common substrates
such as soda-lime glass, stainless steel and polyimide. If required, an additional thin
adhesive layer can be sputtered under the molybdenum.
Another important property of molybdenum is the formation of a thin (a few
nanometers up to some tens of nanometers) MoSe2 layer depending on the sele-
nium exposure time and the substrate temperature [108, 109]. The MoSe2 layer is
responsible for the formation of an ohmic or quasi-ohmic back contact [110]. Without
MoSe2 layer a Schottky contact with a barrier height of about 0.87 eV [111] is created
which impedes the CIGS/Mo hole flux [112]. It has a direct bandgap of 1.35–1.55 eV
[110, 113] and an indirect bandgap of 1.06–1.19 eV [114–116]. The details of the
CIGS/MoSe2 /Mo band alignment are still the subject of discussions. We assume that
the CIGS/MoSe2 contact is an ohmic and the MoSe2 /Mo transition is a Schottky
contact to holes. Since the MoSe2 layer thickness is very thin, however, tunneling
of holes is possible and the measured current–voltage behavior is ohmic [112, 117]
(see also Fig. 9.2). Therefore, a thin MoSe2 layer is very important to achieve highly
efficient CIGS devices that have been proven not only for co-evaporated absorbers
but also for the industrialized sequential sputtering plus selenization process (SAS)
[117]. Even with conductive alternative ZnO-based back contacts, a thin MoSe2 layer
improves device performance and forms a quasi-ohmic contact [118].
A molybdenum back contact works at both low and high CIGS deposition tem-
peratures (400 °C < T < 750 °C).
Orgassa et al. [119] investigated a variety of electron beam evaporated alterna-
tive back contact metals such as tungsten (W), molybdenum (Mo), chromium (Cr),
tantalum (Ta), niobium (Nb), vanadium (V), titanium (Ti) and manganese (Mn) for
CIGS solar cells on soda-lime glass. Best cell efficiencies obtained with evaporated
single-stage CIGS and substrate temperature of 550 °C were 14.2% on W, 13.8%
on Mo, 13.3% on Ta, 10% on Nb, 5.9% on Cr and 3.4% on V. While almost no Se
reactions were observed for W, Mo, Ta and Nb, titanium and manganese layers were
completely consumed by chemical reactions. The authors also had in mind to find a
back contact of higher reflectivity than Mo, which is one of the main drawbacks of
a Mo back contact layer. A highly reflective back contact would allow the absorber
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 191

thickness to be reduced to 0.5–1 μm by multiple reflections. From that perspective

Ta and Nb were superior to Mo or W, at least at wavelengths λ > 700 nm. The search
for suitable highly reflective back contacts is still ongoing and various approaches
have been described in the literature based on the sputtering of a thin ZrN [120,
121] or an Al/InZnO (IZO) layer onto a standard Mo contact. Bissig et al. [122]
achieved with such a Mo(500 nm)/Al(100 nm)/InZnO(260 nm)/Mo(2.6 nm) back
contact with improved reflectivity and high-quality three-stage CIGS an improve-
ment in efficiency from 19.5% (ref.) to 19.9%.

9.2.2.2 Transparent Front Contact

The transparent, highly conductive front contact is probably the least CIGS-specific
layer. Therefore, this topic is only briefly addressed and the reader is referred to the
extensive literature.
Usually aluminum-doped ZnO [50, 51, 63] (ZnO:Al or AZO) is deposited by dc
magnetron sputtering as a wide bandgap window layer (undoped ZnO: E g = 3.3 eV
[123], E g increases with doping), after a highly resistive undoped ZnO (i-ZnO) layer,
which is deposited by rf sputtering. We prefer here to discuss the thin (approximately
50 nm) i-ZnO layer in conjunction with the buffer (see Sect. 9.2.3.2).
Owing to its low carrier mobility of ≤30 cm2 /(Vs) AZO has to be heavily doped so
that the Fermi level is raised up to or above the conduction band edge to achieve suf-
ficiently high conductivity. However, due to optical losses with increased wavelength
in the NIR/IR region resulting from undesired free carrier absorption, a compromise
between conductivity and transparency is sought for monolithically connected mod-
ules. The thickness to be selected depends on the width of the cell stripes (a few mm)
and is about 300–1000 nm for modules without an additional metal grid. The use
of boron-doped ZnO (ZnO:B or BZO), which is characterized by better optoelec-
tronic properties with a lower degree of doping, has also proven to be an industrial
alternative. This transparent conductive oxide is widely used by [58] Solar Frontier
applying the soft MOCVD method for deposition.
In the search for more transparent, conductive and, in terms of degradation by
moisture, more stable n-conducting window layers, numerous other potentially suit-
able TCOs were investigated, such as ITO [124] (In2 O3 :Sn), IOH [125] (In2 O3 :H),
Al-doped ZMO [126] (Zn1−x Mgx O:Al, x ≤ 0.1), FTO (F-doped SnO2 [127]) or a-
IZO [128] (amorphous InZnO) . For example, ITO or IZO have proven to be much
more stable than ZnO in the damp-heat test (1000 h at 85 °C and 85% rel. hum.)
[127, 128]. The significantly higher costs can be partially compensated by the higher
conductivity and thus lower layer thickness. Another promising TCO is IOH which
is characterized by high electron mobility and—like ITO and IZO—can be produced
with an amorphous structure [125].
Especially with regard to further efficiency increases, the full theoretical potential
of the TCO must be exploited as well. Interesting additional options result, for exam-
ple, from the combination of monolithic cell connection and the use of a metal grid
192 F. Kessler et al.

[93]. For instance, the TCO thickness can be reduced while simultaneously increas-
ing transmittance and lateral conductivity. With the help of simulations, the optimum
compromise between dead surface, light incidence and electrical conductivity can
be determined and thus the module efficiency is optimized [93].

9.2.3 Buffer

9.2.3.1 Main Buffer Approaches

Based on the research in CdS solar cells in the 1960s [129], this n-type material was
already well known and a potential and within a short time successfully working
counterpart for the chalcopyrite absorber to build the p-n junction [3]. In the course
of prospecting new ways of efficiency improvement and for environmental aspects,
alternative buffer materials gained more interest toward turn of the century. CIGS
solar modules with buffer layers based on zinc or indium chalcogenides entered the
industrial production in recent years.
The electronic properties of the heterojunction are essential for the device perfor-
mance. The exact role of the buffer layer has been studied for decades but not yet
fully understood. However, thanks to the intensive investigations on various buffer
materials, the progress in understanding the mechanisms at the heterojunction took
many steps forward.
(i) CdS buffer
At the initial stage of development of CuInSe2 solar cells, evaporated CdS, doped
with indium (CdS:In), with a thickness of 1–6 μm was used as a negatively conducting
front electrode first on CuInSe2 single crystals [2] and later on CuInSe2 thin films
[4]. Optionally, ZnS was admixed forming CdZnS to increase the bandgap of the
front electrode [130].
This first device structure was successively modified by replacing the thick CdS
front contact with a significantly more transparent window layer consisting of a
thin CdS buffer and a zinc oxide double layer structure, which in turn consists of a
thin undoped high-resistance layer and a thicker doped conductive layer [131, 132].
Optionally, CdZnS was applied instead of the pure CdS buffer [131]. The choice of
doping element for ZnO was influenced to some extent on the deposition method,
that is, B doping for chemical vapor deposition (CVD) and Al doping for magnetron
sputtering.
A solution growth process, known as chemical bath deposition (CBD), was estab-
lished for the deposition of the CdS buffer layer [133, 134]. The major advantage of
the CBD process is the possibility to grow very thin and dense layers with excellent
step coverage even on very rough CIGS surfaces (Fig. 9.8).
Furthermore, the interaction of the solution chemistry with the CIGS surface in
the initial stage of CdS growth also leads to cleaning and mixing effects on the CIGS
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 193

Fig. 9.8 SEM image of a

CdS(top layer)/CIGS(bottom
layer) cross-section
demonstrating the excellent
covering capability of CBD
CdS

surface, which are assumed to have a positive effect on the p/n junction formation
[135, 136].
Over the last three decades, CdS/i-ZnO/ZnO:Al has been one of the most promi-
nent window layer structures in CIGS solar cells and continues to dominate parts
of the CIGS technology enabling the highest conversion efficiencies compared to
other alternative structures. Important milestones in the development of the CIGS
technology such as the incorporation of Ga in CuInSe2 , process development (three-
stage co-evaporation, sequential processing), and doping with alkali metals were
accompanied with CBD CdS as buffer material [137, 138].
New trends in the CIGS technology, which are mainly influenced by the alkali
post-deposition treatment (PDT), are a further reduction of CBD CdS thickness [51]
and the replacement of the i-ZnO layer by other more transparent materials such
as i-ZnMgO [139] or TiO2 [140]. However, a reduction of the CdS thickness well
below 10 nm can be accompanied by incomplete coverage of the rough CIGS surface.
Sparse CdS growth on the CIGS surface at the very first stage of the CBD process
is influenced by the different CIGS grain orientation [141]. An improvement of the
initial CdS growth could be demonstrated after modification of the CIGS surface
composition with alkali metals [142] or after oxide formation by annealing the CIGS
in air [143]. All these developments are aimed at reducing parasitic absorption in
the blue wavelength region without any V OC loss and thus increasing the power
conversion efficiency.
A schematic representation of the device structure evolution correlated with the
quantum efficiency improvement is shown in Figs. 9.9 and 9.10.
(ii) Alternative buffer layers
The motivation to replace CdS with other materials was driven by two aspects:
(a) The potential for increasing the short current density by avoiding parasitic
absorption in the blue wavelength range between 300 and 550 nm.
(b) Compliance with the RoHS1 directives for the reduction of Cd in electronic
devices.

1 RoHS: Acronym for Restriction on the use of certain Hazardous Substances.

194 F. Kessler et al.

Fig. 9.9 Optimization steps of the interface of CIGS-based solar cells including a CdS buffer

Fig. 9.10 The effect of reduction of parasitic light absorption in the blue wavelength region by
reducing the CdS thickness as demonstrated by external quantum efficiencies in the wavelength
region between 300 and 550 nm of different CIGS-based cells [131, 144, 145] according to the
structures in Fig. 9.9. The potential for maximum gain in spectral response is demonstrated with an
internal quantum efficiency of a CdS-free cell with Zn(O,S)/(Zn,Mg)O buffer
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 195

Over the last two decades, various more transparent and environmentally friendly
materials have been tested than CdS. Most of the works dealt with chalcogenides
based on indium or zinc, such as Zn(S,O), ZnSe, ZnInSex , (Zn,Mg)O, In2 S3 and
In2 Se3 . An elaborate review of the buffer layer development is provided by Naghavi
et al. [146].
In addition to the CBD process used for CdS, a large number of other deposition
methods for the various Cd-free buffer materials, such as atomic layer deposition
(ALD), metal organic vapor deposition (MOCVD), sputtering, ion layer gas reac-
tion (ILGAR), thermal evaporation, electrodeposition (ED), and so on, were inves-
tigated. From an industrial perspective, gas phase processes attract more attention
because they offer the advantage of improved process integrity in the almost vacuum-
dominated CIGS module production lines [147]. However, the characteristics of a
material deposited by different methods can have a wide variety [148] and must be
carefully adapted to the specific absorber material applied. Pre- or post-deposition
treatments such as annealing or light-soaking may be necessary [149, 150]. Mughal
et al. [151] show an overview about champion cell efficiencies for different materials
and deposition methods.
As shown in Fig. 9.10 using a CIGS cell with CBD-Zn(S,O) buffer as an example,
a lower parasitic absorption in the blue wavelength range can also offer a gain in
quantum efficiency even in comparison with improved devices with a thinner CdS
buffer layer. Nevertheless, it can be observed that a gain in short-circuit current is
often accompanied by losses in the open-circuit voltage [152].
The performance of Cd-free cells is significantly influenced by the absorber
material used and matters related thereto such as band-alignment, surface com-
position and interface chemistry. As described in Sects. 9.2.1.2 and 9.2.1.3 co-
evaporated absorbers have a similar bandgap grading compared to sequentially pre-
pared absorbers by selenization/sulfurization of metallic stacks but they are realized
by different composition gradients resulting in different chemical surface composi-
tions. The surface bandgap of co-evaporated CIGS depends on the Ga concentration
while the surface bandgap of sequentially prepared absorbers depend on the sulfur
concentration. Thus, the band alignment between absorber and buffer is affected in
different ways. An increased Ga surface concentration results in an increased con-
duction band-offset (CBO) but an increased S surface concentration results in an
increased valence band-offset (VBO) [150].
So far, Cd-free buffers have not been industrially established in co-evaporated
CIGS-based modules. Realized cell peak efficiencies are comparable, but statisti-
cally below those of the reference devices with CdS. The reason for this is losses
in the open-circuit voltage, which reverse short-circuit current gains, combined with
a reduced reproducibility of the device properties. Progress in reducing or compen-
sating V oc loss for cells with Zn(O,S) buffer has been demonstrated by CIGS post-
deposition treatment with heavier alkali metals. A certified small-area cell efficiency
of 21% could be realized [153].
For sequentially processed sulfur-containing CuIn1−x Gax Se2−y Sy (CIGSS) mod-
ules, Cd-free materials such as CBD-Zn(S,O) and thermally evaporated In2 S3
are already well-established in industrial production. The company Solar Frontier
196 F. Kessler et al.

(a) (b) (c)

Fig. 9.11 Scheme of CIGS cell stacks on glass/Mo substrates established in the industry with
CdS/i-ZnO buffer system and alternative buffer layers: a co-evaporation—Solibro GmbH [158],
b sequential—Avancis GmbH [147], c sequential—Solar Frontier [58]

demonstrated a 19.2% efficient 30 × 30 cm2 record mini-module and a small-area

cell efficiency of 22.0% with CBD-Zn(S,O) buffer layer [58, 154]. Using Inx Sy as
buffer the company AVANCIS GmbH realized a 17.9% 30 × 30 cm2 module [155].
For the In2 S3 buffer aspects such as the elementary interdiffusion between
absorber and buffer layer during buffer deposition or in the following process steps
significantly influence the chemical and electrical properties of the interface, instanc-
ing copper and sodium diffusion. Barreau [156] gives an overview of earlier inves-
tigations on the indium sulfide buffer layer and the influence of the incorporation of
a third atom on its behavior on co-evaporated CIGS devices. Avancis GmbH con-
trols the sodium content and the resulting electronic properties of their sequentially
processed solar modules by direct deposition of a PVD-In2 S3 :Na buffer layer [157].
Well-established device structures for the production of commercial CIGS mod-
ules are shown in Fig. 9.11.

9.2.3.2 The i-ZnO Layer

Historically, the undoped ZnO layer is correlated with the window layer, especially
since both layers, i-ZnO and ZnO:Al, are usually deposited by sputtering processes.
The thin i-ZnO layer has no direct influence on the electrical performance of solar
cells, but has a positive effect on module stability and shunt reduction [159].
Within the framework of buffer development, the n-junction of the buffer system
became interesting and important again, so that both layers can also be considered
as a double buffer layer. Alternative materials such as ZnMgO were explored [160].
For Zn(S,O) buffer layers the replacement of i-ZnO by ZnMgO shows a much more
favorable electronic structure [161].
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 197

9.2.4 Substrate-Related Issues

CIGS thin film solar cells are only a few micrometer thick, and deposited as standard
to several millimeter thick float glass substrates. Since float glass is produced in
large quantities for the window industry, it can be produced very cost-effectively. A
theoretically possible thin, light and flexible PV generator is thus transformed into a
relatively heavy and fragile module, especially as the front encapsulation usually also
consists of a glass plate. Hence, to exploit the full potential of CIGS technology, roll-
to-roll production (R2R) of light and flexible devices should be aimed for. Numerous
metallic and non-metallic foils have been successfully tested [162], but the variety
of possible carrier foils is rather limited with regard to an industrial application in
which economic aspects play an important role.
The most important requirements for a flexible substrate are:
– High temperature resistance in vacuum and selenium atmosphere: wider range of
deposition parameters
– no outgassing at high temperatures in vacuum and no water absorption: no con-
tamination of the absorber
– suitable coefficient of thermal expansion CTE in comparison to CIGS: higher film
adhesion
– smooth, hard and reproducible surface: simplified scribing via laser or mechanical
tool, less shunts, higher yield
– electrically insulating (advantageous): monolithic interconnection possible
– lightweight, flexible and low cost foils available: improved competitiveness.
Since none of the commercially available films meet all criteria, compromises
still have to be made, depending on the application. Owing to the low thickness
of the active layers and the large surface-to-volume ratio, surface roughness and
diffusion processes play a large and often essential role. For example, a change
of carrier material can require a considerable re-optimization of the individual PV
layers. Especially important is the back contact, which often has to be readapted in
order to ensure good layer adhesion without cracking.
The thermal conductivity and heat capacity of the film also influences layer growth
and layer quality, not only of the absorber.
The optimal parameters for CIGS deposition and the type of doping depend
strongly on the substrate temperature applied, so that an absorber optimized at high
substrate temperatures cannot be deposited with the same parameters at low temper-
atures. High CIGS substrate temperatures (T ≥ 600 °C) have the advantage that the
desired CIGS crystallites are formed faster, which has an effect on maximum web
speed and thus on production costs. Lower substrate temperatures (T ≤ 450 °C),
on the other hand, have the advantage that undesired diffusion processes are less
pronounced and the CIGS composition can be better controlled.
Other aspects that are strongly substrate-correlated are the alkali doping of the
CIGS absorber and the monolithic or non-monolithic series connection of cells. If the
end product is to be flexible or lightweight, glass encapsulation must be dispensed
198 F. Kessler et al.

with, which is less a technical than a cost issue, especially as both the Mo back
contact and the ZnO front contact are very sensitive to moisture and must be well
protected.
Metal foils
Metal foils are ideal for R2R coating in vacuum. They do not deform or tear even
under relatively high tensile stress and temperatures. Due to the production-related
rolling process, they are relatively rough compared to glass or polyimide. The dif-
ferent roughness of each individual batch, which depends on the rollers used, can
pose a certain challenge. An even greater challenge is the cost-effective electrical
insulation of the metal foil. Technically, however, this is now possible and a cost-
effective solution can be expected in the future. The most important metal substrate is
currently ferritic stainless steel foil. Normally, this must be provided with a diffusion
barrier [163–165] against iron to prevent contamination of the absorber. Within the
framework of technical progress, the use of low-cost structural steel would also be
conceivable in the foreseeable future.
All commercial CIGS companies that work with metal foils currently use non-
insulated stainless steel foils such as Miasolé (CIGS sputtering and selenization),
Global Solar Energy (CIGS co-evaporation) or Solopower (electrodeposition).
Electrically insulating films
Thin glass is currently no alternative due to its high costs/m2 , although it is avail-
able in a thickness <50 μm. There is not much choice of polymers either: The
only material currently on the market that can withstand temperatures up to 450 °C
is polyimide. Polyimide is available from a thickness of 7.5 μm. The reasons for
choosing polyimide film for R2R production are that PI is electrically insulating
and lightweight and has a well reproducible and relatively smooth surface. At high
temperatures, polyimide films do not melt but begin to decompose, making them
increasingly brittle. The films can be heated up to 600 °C for a short time without
tearing. Depending on the manufacturer and storage prior to coating, polyimide films
can absorb a few percent water, which can then degas during the CIGS process at
400–450 °C and affect the growing CIGS layer. To prevent this, the film is usually
degassed and pretreated prior to the coating process.
The two companies Ascent Solar and Flisom deposit CIGS on polyimide film by
low temperature co-evaporation using the R2R process.
A detailed discussion of substrate-related issues and the fabrication of flexible
CIGS devices can be found elsewhere [166–168].

9.2.5 Series Interconnection of Cells

9.2.5.1 Monolithic Interconnection

The monolithic interconnection of solar cells, that is the series interconnection of PV

cells via at least three pattering steps between the deposition of functional layers, is a
special unique feature of thin film solar cells [169]. As a minimum requirement, the
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 199

substrate has to be electrically insulating, such as glass, polyimide or insulated metal

foil. Monolithic interconnection means that on a single sheet of substrate several
solar cells are serially connected by reasonable overlapping of parts of the front and
back contact of neighbored cells. The classic and simplest way to do this is via three
patterning steps (P1, P2 and P3) performed after the back contact, semiconductors
and conductive window deposition.
The sequential deposition of the individual layers of thin film solar cells allows
dividing the module area into individual cells and monolithically interconnect these
cells in series without any additional interconnection elements like wires, metal grids
or ribbons, for instance. Thus, the high current density parallel to the surface, which
results from the large area of the thin film modules can be significantly reduced by
a factor which correlates directly with the number of subcells. Thus, the application
of a metallic grid, which collects the current on, for example, silicon wafer cells, is
dispensable. In parallel, the voltage of a given PV active area can be increased from
typically less than one volt (one cell) to some tens or even 100 V depending on the
number of solar cells [170].
The basic sequence of the patterning steps for the interconnection outlined in
Fig. 9.12 is initiated with the separation of the back contact layer into galvanic
isolated stripes a few mm wide, typically around 5 mm, which define the individual
cells and the width that the cell current must bridge (P1). After deposition of the
absorber and possibly further layers such as the CdS buffer and the i-ZnO, a second
scribe (P2) next to the P1 groove removes all layers except the back contact and thus
creates contact opening for the subsequent conductive connection between the back
contact and the AZO front contact of adjacent solar cells.
Finally, a third line (P3) is necessary to galvanically separate the cell stripes which
have been short-circuited so far by the AZO layer (Fig. 9.12).
The scribing processes P1 and P3 have to be optimized for maximum isolation
resistance. The required values for P1 and P3 are a minimum of about 0.1 M cm.
The P2 scribe must be optimized for minimum contact resistance. Here in literature,
specific interconnect resistances in the range of 0.3–0.8  cm for mechanically
scribed contacts and 0.1–0.2  cm for laser structured contacts have been reported
[171].
The distance from P1 to P3 in lateral direction forms a photovoltaic inactive zone
and has to be minimized to minimize aperture losses [172]. Mechanical scribing
leads to an about 300 μm wide dead area. The first reported all-laser structuring

Fig. 9.12 Schematic

drawing illustrating the
monolithic series
interconnection
200 F. Kessler et al.

could reduce the dead area to 190 μm, which seems to be a good value for a reliable
process in production. The reported minimum value lies at 70 μm, which is difficult
to maintain in production [173].
Completely different methods can be used to implement the individual patterning
steps. Some possible techniques for performing P1–P3 are:
– chemical etching by means of lithographic masks
– shingling: cutting of CIGS material on flexible substrates and connecting back and
front contact with a metal tape [174, 175]
– mechanical scribing with needles
– selective structuring or lift-off by localized energy input, in particular pulsed laser
radiation, but also more exotic other means such as electron beams.
There are also other methods such as the local application of an anti-sticking layer
below a functional layer, which supports the detachment or even totally prevents the
layer deposition on this area.
Chemical etching is suitable for all three structuring steps of CIGS modules.
However, the need to apply and remove masks for photolithography increases the
number of process steps and thus the effort and costs, as does the alternative use of
printing technology.
Nowadays, industry standard is using selective picosecond laser structuring for
P1 and P2, and mechanical [176] or laser scribing for P3 [177, 178].
Laser scribing
P1: Laser ablation is the preferred method for the patterning of the back contact
(P1). In CIGS modules this layer usually is molybdenum with a thickness of 200–
1000 nm. Since Mo is normally deposited on a glass substrate which has a much
lower melting point, the laser power has to be applied via short or ultrashort pulses
in the nanosecond or picosecond range in order to evaporate the highly absorbing
metal layer before too much heat has flowed into the substrate [179] (Fig. 9.13).
The absorption length of Mo at commercial available laser green (523 nm) and
infrared (1064 nm) lies in a range of 10–20 nm [180]. Heat diffusion lengths for
pulse durations of 10 ps and 10 ns lie at about 50 nm and 1.5 μm, respectively [181],
leading to a heat-affected zone in the same range. Selective laser structuring can be
performed from the film side and the substrate side in a single pulse process [182].
Owing to the long heat diffusion length, short pulses in the nanosecond range heat
the Mo film homogenously, remove the film mainly in the liquid phase and support
the removal process through the formation of thermomechanical forces that lift off
the layer from the substrate. The removal fluence threshold lies at about 1–2 J/cm2
for about 500 nm thick Mo films [181, 183]. The nanosecond process always leads
to unwanted cracks and burr in the film and substrate or barrier layer (e.g. SiNx)
melting or damage (see Fig. 9.13, left side).
Ultrashort pulses in the picosecond or femtosecond range lead to a confined energy
situation, where a strong pressure wave is generated in the laser-irradiated interface
[184]. The laser pulse is heating a relatively thin contact layer on a thick, stiff substrate
(e.g. 3 mm thick glass). If the laser is irradiated from the film side, the ablation
threshold is at about 1.7 J/cm2 , the fluence needed for a removal of the complete film
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 201

Fig. 9.13 P1 groove performed by use of a nanosecond laser (left) and a picosecond laser with
10.2 ps pulse duration from the substrate side (right). The nanosecond laser leads to rims at the edge
of the ablated Mo-film, which are comparable to the thickness of the Mo layer of about 500 nm.
Additionally, microcracks and partial material lift-offs are produced in a range of 5 μm to the sides
of the groove. The substrate glass shows faint melting zones in the pulse overlapping areas. The
edges of the scribe have a molten appearance and are paralleled by darker zones which are covered
by condensed glass vapor. The picosecond laser process is free from thermal damages

lies between 5 and 10 J/cm2 [170, 182]. The film removal is accomplished by direct
laser ablation, evaporation of a surface layer, thermal expansion and pressure wave
assisted, photo-mechanical ablation of the Mo film. In this way melting and damage
of the film and substrate can be almost completely avoided [185].
The confined energy situation can be further assisted by passing the laser beam
through the glass substrate onto the backside of the Mo layer and thus directly
heating the interface. In a so-called direct-induced ablation, the Mo film is lifted-off
by an ultrafast melting of an about 50 nm thick interface layer, heat expansion and
subsequent pressure wave generation, which causes the layer to be directed away
from the substrate after about 100 ps. The conservation of this momentum pulls on
the layer with a force which can become high enough to overcome the adhesion and
the layer is tearing off after about 10 ns. The film removal is performed by punching
out intact film disks leaving behind very well-defined circular edges without thermal
damage, melting and debris [182, 185, 186] (see Fig. 9.13, right side).
For these reasons, the application of picosecond laser pulses via the glass side
(i.e. the back side) is considered to be the preferred P1 method with the best results.
Additionally, according to Heise et al., the threshold fluence of the laser is 0.1 J/cm2 ,
which is more than one order of magnitude smaller compared to P1 laser patterning
from the front side [178]. The process window lies between about 0.5 and 1 J/cm2 .
Nevertheless, in case of opaque substrates, P1 patterning from the back is no
option. P1 laser scribing has been carried out with nanosecond (ns) and picosecond
(ps) lasers of green (532 nm) or infrared (1064 nm) wavelength and a pulse overlap
of ≥30%. Thus, ultrashort pulse lasers (femtosecond or picosecond range) are very
advantageous, but not mandatory. The wavelength plays only a minor role [187]. In
order to minimize damage to the substrate and to obtain the smoothest possible, the
pulse duration should lie in the ultrafast domain and the energy should be kept as low
202 F. Kessler et al.

Fig. 9.14 Left: Removal of CIGS by a pulsed nanosecond laser (P2). Owing to excess energy in the
centre of the trace, the Mo back contact shows cracks. Right: The picosecond laser process avoids
microcracks in the Mo and leads to a lower contact resistance

as possible. With average powers of 5–10 W line scribing speeds of 10 m/s and more
can be realized from the substrate side, while powers of 50–100 W are necessary to
reach the same scribing speeds from the film side.
P2: The removal of the absorber layer for the contact opening step P2 can be
done with multiple highly overlapping pulses with direct laser ablation (Fig. 9.14).
As with P1, thin parts of the absorber material can be vaporized with short and
ultrashort pulses [170]. However, the energy input has to be carefully controlled. If
the absorber is incompletely removed, an increased contact resistance results. On the
contrary, if the fluence is too high, the back contact will be damaged. Since the back
contact is usually significantly thinner than the absorber, fluctuations in the absorber
thickness can quickly lead to excess energy being capable of melting or cracking the
back contact (Fig. 9.14, left).
The ablation threshold of a single 10 ps laser pulse for the absorber lies at fluence
of 0.1–0.2 J/cm2 . Only a few 10 nm thick part of the absorber is removed at the
threshold. For a reliable structuring process of 2–3 μm thick absorber, the fluence
and the average number of pulses per position are increased to 1–2 J/cm2 and about
10–30 pulses per position (Fig. 9.14, right). The short thermal diffusion length of the
picosecond pulse limits the thermal damage and avoids microcracks. In addition, the
precise depth control of the picosecond laser process enables a removal of a weakly
conducting MoSe or MoS layer (remaining from the absorber formation) without
damaging the Mo film, thus the contact resistance of the later deposited ZnO film
can be reduced to about 50% compared to mechanical scribing [170]. A formation of
ripples, so-called laser-induced periodic surface structures (LIPSS [188]) at the Mo
surface is observed and might have also a positive impact on lowering the contact
resistance.
Another variation of the P2 laser processing is the CIGS decomposition due to
selective evaporation of the selenium. The metallized absorber does not need to
be completely removed due to its increased conductivity compared to the original
composition. However, a noticeable additional contact resistance remains [171, 189].
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 203

P2 can even be performed after deposition of the AZO layer by de-selenization

of the CIGS absorber with a so-called P2 microwelding; however, it is challenging
to achieve a sufficiently low contact resistance [190]. The potential advantage of
this method is that P2 and P3 can be done simultaneously, for example, by two
different lasers or P2 by laser and P3 mechanically. Hence, one alignment and process
interruption step could be avoided in production [191].
To remove CIGS by lift-off instead of CIGS direct ablation by heating and evap-
oration, Moser et al. [192] and Burn et al. [187] suggested using a long wavelength
laser of 1550 nm that is not absorbed by the absorber but by the underlying Mo.
P2 can be performed by a wide range of methods and lasers, such as continuous
wave, pulsed, nanosecond, picosecond, long and short wavelengths. Nevertheless,
the most reliable and CIGS process-tolerant laser parameters that can be used in
production seems to be the multipulse direct ablation with picosecond lasers due to
avoiding of unwanted microcracks and superior contact properties [176–178].
P3: To achieve a solar cell separation, the transparent conductive front contact
has to be galvanically separated with a P3 scribe by either removing just the ZnO
layer on top of the CIS or both, ZnO and CIS on top or the Mo back electrode. Both
processes can be performed with picosecond lasers [170, 177, 178].
However, the direct approach to couple the laser energy into the contact layer
material by interband absorption, which corresponds to UV radiation, requires a
high fluence input when using nanosecond pulses. As a consequence, the edge zones
of the P3 groove are thermally damaged. An improvement for the selective structuring
of transparent thin films is to apply UV picosecond pulses with reduced absorption
length and thermal diffusion length [193].
Indirect ablation with picosecond lasers has proven to be much more effective
[194]. For this, a wavelength is needed for which the window layer is transparent
and which is absorbed at the interface in the first layers of the absorber. Additionally,
the pulse duration has to be short enough to create a confined energy situation [184].
Indirect ablation is not yet fully understood, but the mechanism should be similar to
the pressure wave, that is, mainly photomechanical lift-off as described for the Mo
film removal (P1) over the substrate side. The heat expansion of a few 10 nm thin
absorber material interface, combined with melting and eventually evaporating, is
creating a pressure wave, which lifts-off the front contact mainly photo-mechanically.
The amount of energy required for this is very small and typically in the range of
only one-tenth of the energy required to evaporate the window layer [170]. But the
CIGS surface has undergone substantial heating and results in some thermal damage,
leading to shunts at the P3 edge, which can be detected in thermography. In summary,
the picosecond laser P3 achieves the efficiency of the mechanical P3, but does not
improve it. However, the advantages of the P3 laser scribing are the reduction of the
dead zone and a non-contact, wear-free process (Fig. 9.15).
P3 laser scribing is also possible by removing the entire layer stack down to the
back contact [170, 187]. However, this is not only unnecessary but even unfavorable
[195] as the increased energy input (fluence) required to remove the CIGS also
increases the thermal damage of the P3 edges.
204 F. Kessler et al.

Fig. 9.15 SEM images of a P3 scribe performed with a 10 picosecond laser at 532 nm (top row)
and 1064 nm (bottom). The AZO and the buffer layers are removed from the CIGS. The CIGS
surface has been thermally damaged within the scribe and shows some microcracks in the REM
image (bottom)

In literature all fiber laser scribing [196] of CIGS modules has been investigated
and discussed, as well as the possibility of high-speed scribing [172, 187]. Owing to
the progress in laser technology, development in the direction of robust, powerful and
industrial ultrashort-pulse lasers introduces a 10–15% solar efficiency increase by
reducing the dead area and lowering the contact resistance of front and back contact
[172, 177, 195].
Monolithic solar cell connection on sensitive flexible substrates
All laser scribing is mandatory if the monolithic cell interconnection has to be per-
formed on thin (25–50 μm) temperature-sensitive polyimide substrates or rough or
soft substrates (in comparison to glass). Polyimide (PI) films, for instance, absorb
water, start to decompose at rather low temperatures (T > 450 °C), are highly flexible
and normally have a linear coefficient of thermal expansion (CTE) which is clearly
higher than soda lime glass or molybdenum.
P1: To prevent thermally induced irreversible dilatation of the plastic film at the
P1 grooves and to avoid thermal damage of the substrate by the laser, the energy input
via the laser should be as low as possible. Thus, picosecond or even femtosecond
short pulse visible (532 nm) or near IR lasers (1064 nm) combined with the lowest
possible fluence should be preferred. Stress-related Mo detachment is preferred to
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 205

thermally induced removal and should dominate the P1 process. Although the process
can be carried out in such a way that the plastic substrate remains almost undamaged,
some excess energy is advantageous in order to overcome the inhomogeneities and
height fluctuations that could otherwise locally lead to residual material bridges in
the scribe [197]. The Mo-on-PI adhesion is generally worse than the Mo-on-glass
adhesion, which would support P1 laser scribing. However, in view of the entire
solar cell and the other patterning steps, Mo adhesion should be as good as possible.
Depending on the back-contact properties, the thickness of the polymer film and the
applied wavelength, P1 can also be performed with nanosecond lasers and even from
the rear.
P2: P2 performed on polyimide substrates seems to be the most challenging
scribe especially with view on high-speed laser patterning. Moderate Mo adhesion
on polyimide is advantageous for P1, but for P2 the adhesion on the substrate should
be as high as possible in order not to remove the Mo layer. The P2 laser processing
on polyimide (PI) is much more delicate compared to soda-lime glass substrates
as the layer stack can absorb much less energy without harming the PI film. Thus,
short pulse lasers with low pulse energy and high overlap would be a good choice
to remove the absorber step by step without damaging the layers underneath [172,
197]. In this process, the CIGS layer is gradually evaporated pulse by pulse. High
repetition rates of up to 1 MHz are therefore the means of choice for implementing
the high-speed patterning required for successful industrialization.
P3: The P3 scribe is independent of the substrate type, at least if only the TCO
window is removed as recommended also on glass (see previous chapter).
There is currently no uniform or universally valid concept for laser patterning on
polymer film. However, the method of choice should be flexible and not depend on
the individual layer properties such as adhesion or layer thickness.
Mechanical scribing
Owing to the relatively low adhesion of CIGS on the molybdenum back contact and
the much higher hardness of Mo, the absorber can be brushed off the back contact
with mechanical scribing tools such as needles. Even brushes may be used if larger
areas need to be decoated. A common tool used for P2 and P3 is a cone-shaped pen
with a flat tip. The tip diameter determines the minimum groove width. It is typically
in the range of a few tens of microns.
Owing to the stiffness of the TCO layer, the P3 groove is considerably more
chipped than P2 (Fig. 9.16). In order to achieve a long service life, a very hard
material such as carbide is preferred. But even though there is only very little abrasion
with such hardened tools, they need to be replaced quite frequently, because they
break from time to time due to accidental bumps on the substrate surface. Another
problem that may arise from this mechanical scribing technique is related to the
aggregation of rubbed layer material under the tip of the tool. In case of P3 this is a
mixture of CIGS and ZnO particles, which are hard enough to scratch the underlying
molybdenum layer. Such scratches can substantially increase the contact resistance
between the cells, especially if corrosion due to humidity additionally attacks this
contact (Fig. 9.17).
206 F. Kessler et al.

Fig. 9.16 P2 (left) and P3

(right) by mechanical
scribing

Fig. 9.17 SEM image of P3

scribe after storage in humid
air for some days. Moisture
penetrated the Mo layer
through barely visible cracks
and led to chemical reactions
between glass, Mo und H2 O.
The deposits on the surface
consist of carbonates and
other compounds of sodium
and molybdenum

9.2.5.2 Alternative Concepts

Alternative concepts for the integrated serial interconnection shift one or more pat-
terning steps toward the end of the process chain [190, 198–200]. This is partic-
ularly important if several processes have to be carried out without interruption
of vacuum processing, such as in the roll-to-roll coating. In this quasi-monolithic
or post-deposition approach P1, P2 and P3 are performed simultaneously after all
active layers have been deposited. This, however, eliminates the functional contribu-
tion of the stacked layers to the serial interconnection making two additional work
steps necessary: The connection of neighbored cells across P1 is no longer achieved
“automatically” by the deposition of the AZO layer, but has to be provided by an addi-
tionally applied contact material such as a conductive Ag paste. As the P1 grooves
are no more filled by isolating CIGS material, the uncovered P1 groove now has to
be filled with an additional insulator (paste, ink) to prevent short circuits [169, 198,
200] (Fig. 9.18).
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 207

Fig. 9.18 Modified interconnection scheme: P1 and P2 are performed after all layers have been
deposited. As a consequence, P1 must be filled with an insulator (drawn in orange) and an additional
contact material is required to connect adjacent cells by bridging P1 and connecting the TCO of the
left cell with the Mo back contact of the right cell via P2 (drawn in blue)

A compromise between the initial three-step and the five-step approach is the
four-step method [200] in which P1 is postponed only and applied together with
P2 before the TCO top contact layer deposition. This method only requires the
additional printing of an isolation line while the AZO layer continues to establish
the cell connection.
The discussion of a non-monolithic connection, for example, of large solar cells on
metal substrates separated from each other, is to be dispensed with here. In principle,
however, individual cells can be connected in series via contact strips and grids such
as Si wafers [201].

9.3 Future Prospects

Although CIGS technology has long since matured, new findings have enabled a
continuous and gradual increase in efficiency, as has been demonstrated in recent
years, for example, by the introduction of a defined gallium gradient and precisely
performed alkali doping. It can be assumed that further insights to increase the quan-
tum yield and reduce the radiation-free recombination will bring the efficiencies
closer to the theoretical limit. Another possibility to increase the efficiency already
with the existing CIGS quality is the possibility to integrate CIGS as a back cell
into a tandem structure with perovskite as front solar cell. The efficiency of a per-
ovskite CIGS tandem structure is currently 23.9% [202], achieved by a four-terminal
configuration.
With the help of CIGS thin film solar cell technology, the most efficient flexible
solar generators currently available can be produced. The most commonly used sub-
strate films are made of stainless steel or polyimide. Advances in materials science,
for example in the deposition of efficient and cost-effective functional layers for sub-
strate isolation, substrate modification or as a diffusion barrier, can further reduce
the influence of the substrate on the properties of the solar cell.
By producing cost-effective, lightweight, flexible and efficient PV modules on
mild steel, for example in mobile systems or as a roll-out for shading elements,
CIGS technology can reach people at all levels of everyday life and thus contribute
to reducing CO2 emission and achieving climate protection targets.
208 F. Kessler et al.

References

1. H. Hahn, G. Frank, W. Klingler, A.-D. Meyer, G. Störger, Untersuchungen über ternäre

Chalkogenide. Über einige ternäre Chalkogenide mit Chalkopyritstruktur. Zeitschrift für anor-
ganische und allgemeine Chemie, Band 271, 153–170 (1953)
2. S. Wagner, J.L. Shay, P. Migliorato, CulnSe2/CdS heterojunction photovoltaic detectors. Appl.
Phys. Lett. 25, 434 (1974). https://doi.org/10.1063/1.1655537
3. S. Wagner, J. Shay, H. Kasper, The p-CuInSe2/n-CdS Heterodiode: photovoltaic detector,
solar cell and light emitting diode. J. Phys. Colloq. 36(C3), C3-101–C3-104 (1975). https://
doi.org/10.1051/jphyscol:1975319. jpa-00216289
4. L.L. Kazmerski, F.R. White, G.K. Morgan, Thin-film CuInSe2/CdS heterojunction solar cells.
Appl. Phys. Lett. 29, 268 (1976). https://doi.org/10.1063/1.89041
5. R.A. Mickelsen, W.S. Chen, Development of a 9.4% efficient thin-film CuInSe2/CdS solar
cell, in Conference Record of the IEEE Photovoltaic Specialists Conference, Proceedings of
the Fifteenth IEEE Photovoltaic Specialists Conference (Kissimmee, FL, 1981), pp. 800–804
6. K.L. Chopra, P.D. Paulson, V. Dutta, Thin-film solar cells: an overview. Prog. Photovolt. Res.
Appl. 12, 69–92 (2004). https://doi.org/10.1002/pip.541
7. R.L. Mitchell, C.E. Witt, R. King, D. Ruby, PVMaT advances in the photovoltaic industry
and the focus of future PV manufacturing R&D. in Conference Record of the 29th IEEE
Photovoltaic Specialists Conference (IEEE, New Orleans, Louisiana, 2002), pp. 1444–1447
8. A. Jasenek, U. Rau, T. Hahn, G. Hanna, M. Schmidt, M. Hartmann, H.W. Schock, J.H.
Werner, B. Schattat, S. Kraft, K.-H. Schmid, W. Bolse, Defect generation in polycrystalline
Cu(In,Ga)Se2 by high-energy electron irradiation. Appl. Phys. A 70, 677–680 (2000). https://
doi.org/10.1007/s003390000512
9. S. Dabbabi, T.B. Nasr, N.T. Kamoun, CIGS solar cells for space applications: numerical
simulation of the effect of traps created by high-energy electron and proton irradiation on the
performance of solar cells. JOM, Miner. Metals Mater. Soc. (2018). https://doi.org/10.1007/
s11837-018-2748-9
10. J. Ramanujam, U.P. Singh, Copper indium gallium selenide based solar cells—a review.
Energy Environ. Sci. 10, 1306–1319 (2017). https://doi.org/10.1039/c7ee00826k
11. H. Neumann, N. Van Nam, H.-J. Hobler, G. Kuhn, Electrical properties of n-TYPE CuInSe2
single crystals. Solid State Commun. 25, pp. 899–902 (1978)
12. M. Igalson, A. Urbaniak, K. Macielak, M. Tomassini, N. Barreau, S. Spiering, Barriers for cur-
rent transport in CIGS structures, in Conference Record of the IEEE Photovoltaic Specialists
Conference (2011). https://doi.org/10.1109/pvsc.2011.6186511
13. T. Wada, N. Kohara, T. Negami, M. Nishitani, Chemical and structural characterization
Cu(In,Ga)Se2/Mo in Cu(In,Ga)Se2 solar cells. Jpn. J. Appl. Phys. 35, 1253 (1996)
14. H.-W. Schock, U. Rau, The role of structural properties and defects for the performance of
Cu-chalcopyrite-based thin-film solar cells. Phys. B 308–310, 1081–1085 (2001)
15. B.J. Stanbery, Copper indium selenides and related materials for photovoltaic devices. Crit.
Rev. Solid State Mater. Sci. 27(2), 73–117 (2002). https://doi.org/10.1080/20014091104215
16. N.G. Dhere, Present status and future prospects of CIGSS thin film solar cells. Sol. Energy
Mater. Sol. Cells 90, 2181–2190 (2006)
17. H.-W. Schock, R. Noufi, Prog. Photovolt.: Res. Appl. 8, 151 (2000)
18. R. Klenk, Thin Solid Films 387, 135–140 (2001)
19. M. Turcu, U. Rau, Fermi level pinning at CdS/Cu(In, Ga)(Se, S)2 interfaces: effect of chal-
copyrite alloy composition. J. Phys. Chem. Solids 64, 1591–1595 (2003)
20. C.H. Chang, A. Davydov, B.J. Stanbery, T.J. Anderson, Thermodynamic assessment of the
Cu-In-Se system and application to thin film photovoltaics, in The Conference Record of the
25th IEEE Photovoltaic Specialists Conference (IEEE, Piscataway, 1996), p. 849
21. H.-W. Schock, Properties of chalcopyrite-based materials thin-film solar cells, in Thin-
Film Solar Cells, Next Generation Photovoltaics and Its Applications, ed. by Y. Hamakawa
(Springer, Berlin, Heidelberg, New York, 2004)
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 209

22. T. Gödecke, T. Haalboom, F. Ernst, Phase Equilibria of Cu-In-Se. Z. Metallkd. 91, 622–662
(2000)
23. M.A. Contreras, K. Ramanathan, J. AbuShama, F. Hasoon, D.L. Young, B. Egaas, R. Noufi,
Diode characteristics in state-of-the-art ZnO/CdS/Cu(In1-x Gax)Se2 solar cells. Prog. Photo-
volt: Res. Appl. 13, 209–216 (2005). https://doi.org/10.1002/pip.626
24. M. Kemell, M. Ritala, M. Leslkelä, Thin film deposition methods for CuInSe2 solar cells. Crit.
Rev. Solid State Mater. Sci. 30(1), 1–31 (2005). https://doi.org/10.1080/10408430590918341
25. S.B. Zhang, S.H. Wei, A. Zunger, H. Katayama-Yoshida, Defect physics of the CuInSe2
chalcopyrite semiconductor. Phys. Rev. B 57, 9642 (1998)
26. M. Malitckaya, H.-P. Komsa, V. Havu, M.J. Puska, First-principles modeling of point defects
and complexes in thin-film solar-cell absorber CuInSe2. Adv. Electron. Mater. 3, 1600353
(2017)
27. J. Pohl, K. Albe, Intrinsic point defects in CuInSe2 and CuGaSe2 as seen via screened-
exchange hybrid density functional theory. Phys. Rev. B 87, 245203 (2013)
28. C. Persson, Y.-J. Zhao, S. Lany, A. Zunger, n-type doping of CuInSe2 and CuGaSe2. Phys.
Rev. B 72, 035211 (2005)
29. S. Siebentritt, M. Igalson, C. Persson, S. Lany, The electronic structure of chalcopyrites—
bands, point defects and grain boundaries. Prog. Photovolt: Res. Appl. 18, 390–410 (2010).
https://doi.org/10.1002/pip.936
30. S. Siebentritt, Chalcopyrite compound semiconductors for thin film solar cells. Curr. Opin.
Green Sustain. Chem. 4, 1–7 (2017). https://doi.org/10.1016/j.cogsc.2017.02.001
31. W. Shockley, H.J. Queisser, Detailed balance limit of efficiency of p-n junction solar cells. J.
Appl. Phys. 32(3), 510–519 (1961). https://doi.org/10.1063/1.1736034
32. S. Rühle, Sol. Energy 130, 139–147 (2016). https://doi.org/10.1016/j.solener.2016.02.015
33. G. Hanna, A. Jasenek, U. Rau, H.W. Schock, Influence of Ga-content on the bulk defect
densities of Cu(In, Ga)Se2. Thin Solid Films 387, 71–73 (2001)
34. U. Rau, M. Schmidt, A. Jasenek, G. Hanna, H.W. Schock, Electrical characterization of Cu(In,
Ga)Se2 thin-film solar cells and the role of defects for the device performance. Sol. Energy
Mater. Sol. Cells 67, 137–143 (2001)
35. P. Jackson, R. Würz, U. Rau, J. Mattheis, M. Kurth, T. Schlötzer, G. Bilger, J.H. Werner, High
quality baseline for high efficiency, Cu(In1-x, Gax)Se2 solar cells. Prog. Photovolt. Res. Appl.
15, 507–519 (2007). https://doi.org/10.1002/pip.757
36. M.A. Contreras, L.M. Mansfield, B. Egaas, J. Li, M. Romero, R. Noufi, E. Rudiger-Voigt,
W. Mannstadt, Wide bandgap Cu(In, Ga)Se2 solar cells with improved energy conversion
efficiency. Prog. Photovolt: Res. Appl. 20, 843–850 (2012). https://doi.org/10.1002/pip.2244
37. P. Jackson, R. Wuerz, D. Hariskos, E. Lotter, W. Witte, M. Powalla, Effects of heavy alkali
elements in Cu(In,Ga)Se2 solar cells with efficiencies up to 22.6%. Phys. Status Solidi RRL
10(8), 583–586 (2016). https://doi.org/10.1002/pssr.201600199
38. V. Bermudez, A. Perez-Rodriguez, Understanding the cell-to-module efficiency gap in Cu(In,
Ga)(S, Se)2 photovoltaics scale-up. Nat. Energy 3, 466–475 (2018)
39. H. Hiroi, Y. Iwata, S. Adachi, H. Sugimoto, A. Yamada, New world-record efficiency for pure-
sulfide Cu(In, Ga)S2 thin-film solar cell with Cd-free buffer layer via KCN-free process. IEEE
J. Photovolt. 6(3), 760–763 (2016)
40. K. Siemer, J. Klaer, I. Luck, J. Bruns, R. Klenk, D. Bräunig, Efficient CuInS2 solar cells from
a rapid thermal process (RTP). Sol. Energy Mater. Sol. Cells 67, 159–166 (2001)
41. F. Larsson, N.S. Nilsson, J. Keller, C. Frisk, V. Kosyak, M. Edoff, T. Törndahl, Record 1.0 V
open-circuit voltage in wide band gap chalcopyrite solar cells. Prog. Photovolt. Res. Appl.
25, 755–763 (2017). https://doi.org/10.1002/pip.2914
42. H. Goto, Y. Hashimoto, K. Ito, Efficient thin film solar cell consisting of
TCO/CdS/CuInS2/CuGaS2 structure. Thin Solid Films 451–452, 552–555 (2004). https://
doi.org/10.1016/j.tsf.2003.11.045
43. M.A. Contreras, J. Tuttle, A. Gabor, A. Tennant, K. Ramanathan, S. Asher, A. Franz, J. Keane,
L. Wang, J. Scofield, R. Noufi, High efficiency Cu(ln,Ga)Se2-based solar cells: processing
of novel absorber structures, in Proceedings 1st World Conference on Photovoltaic Energy
Conversion (IEEE, Piscataway, NJ, 1994), p. 68
210 F. Kessler et al.

44. T. Dullweber, G. Hanna, U. Rau, H.W. Schock, A new approach to high-efficiency solar cells
by band gap grading in Cu(In, Ga)Se2 chalcopyrite semiconductors. Sol. Energy Mater. Sol.
Cells 67, 145–150 (2001)
45. Y. Hahsimoto, N. Kohara, T. Negami, M. Nishitani, T. Wada, Surface characterization of
chemically treated Cu(In, Ga)Se2 thin films. Jpn. J. Appl. Phys. 35, 4760–4764 (1996)
46. K. Ramanathan, M.A. Contreras, C.L. Perkins, S. Asher, F.S. Hasoon, J. Keane, D. Young,
M. Romero, W. Metzger, R. Noufi, J. Ward, A. Duda, Properties of 19.2% efficiency
ZnO/CdS/CuInGaSe2 thin-film solar cells. Prog. Photovolt: Res. Appl. 11, 225–230 (2003).
https://doi.org/10.1002/pip.494
47. E. Avancini, R. Carron, B. Bissig, P. Reinhard, R. Menozzi, G. Sozzi, S. Di Napoli, T. Feurer,
S. Nishiwaki, S. Buecheler, A.N. Tiwari, Impact of compositional grading and overall Cu
deficiency on the near-infrared response in Cu(In, Ga)Se2 solar cells. Prog. Photovolt: Res.
Appl. (2016). https://doi.org/10.1002/pip.2850
48. A. Chirilă, S. Buecheler, F. Pianezzi, P. Bloesch, C. Gretener, A.R. Uhl, C. Fella, L. Kranz,
J. Perrenoud, S. Seyrling, R. Verma, S. Nishiwaki, Y.E. Romanyuk, G. Bilger, A.N. Tiwari,
Highly efficient Cu(In, Ga)Se2 solar cells grown on flexible polymer films. Nat. Mater. 10,
857–861 (2011)
49. S. Niki, M. Contreras, I. Repins, M. Powalla, K. Kushiya, S. Ishizuka, K. Matsubara, CIGS
absorbers and processes. Prog. Photovolt: Res. Appl. 18, 453–466 (2010). https://doi.org/10.
1002/pip.969
50. P. Jackson, D. Hariskos, R. Wuerz, O. Kiowski, A. Bauer, T.M. Friedlmeier, M. Powalla,
Properties of Cu(In,Ga)Se2 solar cells with new record efficiencies up to 21.7%. Phys. Status
Solidi RRL 9(1), 28–31 (2015). https://doi.org/10.1002/pssr.201409520
51. A. Chirilă, P. Reinhard, F. Pianezzi, P. Bloesch, A.R. Uhl, C. Fella, L. Kranz, D. Keller,
C. Gretener, H. Hagendorfer, D. Jaeger, R. Erni, S. Nishiwaki, S. Buecheler, A.N. Tiwari,
Potassium-induced surface modification of Cu(In, Ga)Se2 thin films for high-efficiency solar
cells. Nat. Mater. 12, 1107–1111 (2013). https://doi.org/10.1038/NMAT3789
52. M. Powalla, P. Jackson, W. Witte, D. Hariskos, S. Paetel, C. Tschamber, W. Wischmann,
High-efficiency Cu(In, Ga)Se2 cells and modules. Sol. Energy Mater. Sol. Cells 119, 51–58
(2013)
53. P.M.P. Salomé, V. Fjällström, P. Szaniawski, J.P. Leitão, A. Hultqvist, P.A. Fernandes, J.P.
Teixeira, B.P. Falcão, U. Zimmermann, A.F. da Cunha, M. Edoff, A comparison between thin
film solar cells made from co-evaporated CuIn1-x Gax Se2 using a one-stage process versus a
three-stage process. Prog. Photovolt: Res. Appl. 23, 470–478 (2015). https://doi.org/10.1002/
pip.2453
54. M. Powalla, B. Dimmler, Scaling up issues of CIGS solar cells. Thin Solid Films 361–362,
540–546 (2000)
55. R. Scheer, A. Pérez-Rodríguez, W.K. Metzger, Advanced diagnostic and control methods
of processes and layers in CIGS solar cells and modules. Prog. Photovolt: Res. Appl. 18,
467–480 (2010)
56. V.K. Kapur, U.V. Choudary, A.K. P. Chu, US patent no. 4,581,108 (1986)
57. V. Probst, W. Stetter, W. Riedl, H. Vogt, M. Wendl, H. Calwer, S. Zweigart, K.-D. Ufert,
B. Freienstein, H. Cerva, F.H. Karg, Rapid CIS-process for high efficiency PV-modules:
development towards large area processing. Thin Solid Films 387, 262–267 (2001)
58. T. Kato, Cu(In,Ga)(Se,S)2 solar cell research in solar Frontier: progress and current status.
Jpn. J. Appl. Phys. 56, 04CA02 (2017). https://doi.org/10.7567/jjap.56.04ca02
59. G.D. Mooney, A.M. Hermann, J.R. Tuttle, D.S. Albin, R. Noufi, Formation of CuInSe2 thin
films by rapid thermal recrystallization. Appl. Phys. Lett. 58, 2678 (1991). https://doi.org/10.
1063/1.104805
60. J.A. Frantz, J.D. Myers, R.Y. Bekele, V.Q. Nguyen, B.M. Sadowski, S.I. Maximenko,
R.J. Walters, J.S. Sanghera, Recent progress in sputtered Cu(In,Ga)Se2 absorbers for
photovoltaics. https://www.researchgate.net/publication/282960353, https://doi.org/10.1364/
noma.2015.ns3b.2
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 211

61. C.-H. Chen, W.-C. Shih, C.-Y. Chien, C.-H. Hsu, Y.-H. Wua, C.-H. Lai, A promising sputtering
route for one-step fabrication of chalcopyrite phase Cu(In, Ga)Se2 absorbers without extra
Se supply. Sol. Energy Mater. Sol. Cells 103, 25–29 (2012)
62. Y.-C. Wang, H.-P.D. Shieh, Double-graded bandgap in Cu(In, Ga)Se2 thin film solar cells by
low toxicity selenization process. Appl. Phys. Lett. 105, 073901 (2014). https://doi.org/10.
1063/1.4893713
63. F. Karg, High efficiency CIGS solar modules. Energy Procedia 15, 275–282 (2012)
64. S.S. Schmidt, C. Wolf, H. Rodriguez-Alvarez, J.-P. Bäcker, C.A. Kaufmann, S. Merdes, F.
Ziem, M. Hartig, S. Cinque, I. Dorbandt, C. Köble, D. Abou-Ras, R. Mainz, R. Schlatmann,
Adjusting the Ga grading during fast atmospheric processing of Cu(In, Ga)Se2 solar cell
absorber layers using elemental selenium vapor. Prog. Photovolt: Res. Appl. 25, 341–357
(2017). https://doi.org/10.1002/pip.2865
65. T. Kobayashi, H. Yamaguchi, Z.J.L. Kao, H. Sugimoto, T. Kato, H. Hakuma, T. Nakada,
Impacts of surface sulfurization on Cu(In1_x, Gax)Se2 thin-film solar cells. Prog. Photovolt:
Res. Appl. 23, 1367–1374 (2015). https://doi.org/10.1002/pip.2554
66. N.B. Chaure, Electrodeposited CuIn1-x Gax Se2 thin films from non-aqueous medium for solar
cell Applications. J. Renew. Sustain. Energy 5, 031604 (2013). https://doi.org/10.1063/1.
4808023
67. R.N. Bhattocharya, Solution growth and electrodeposited CulnSe2 thin films. J. Electrochem.
Soc. 130(10), 2040 (1983)
68. A. Kampmann, V. Sittinger, J. Rechid, R. Reineke-Koch, Large area electrodeposition of
Cu(In, Ga)Se2. Thin Solid Films 361–362, 309–313 (2000)
69. S.N. Qiu, L. Li, C.X. Qiu, I. Shih, C.H. Champness, Study of CulnSe2 thin films prepared by
electrodeposition. Sol. Energy Mater. Sol. Cells 37, 389–393 (1995)
70. C. Broussillou, C. Viscogliosi, A. Rogee, S. Angle, P.P. Grand, S. Bodnar, C. Debauche,
J.L. Allary, B. Bertrand, C. Guillou, L. Parissi, S. Coletti, Statistical process control for
Cu(In,Ga)(S,Se)2 electrodeposition-based manufacturing process of 60 × 120 cm2 modules
up to 14,0% efficiency, in Conference Record of the 42nd IEEE Photovoltaic Specialist Con-
ference (2015). https://doi.org/10.1109/pvsc.2015.7356224
71. S. Aksu, S. Pethe, A. Kleiman-Shwarsctein, S. Kundu, M. Pinarbasi, Recent advances in
electroplating based CIGS solar cell fabrication, in Conference Record of the 38th IEEE
Photovoltaic Specialist Conference (2012). https://doi.org/10.1109/pvsc.2012.6318235
72. A.R. Uhl, C. Fella, A. Chirilă, M.R. Kaelin, L. Karvonen, A. Weidenkaff, C.N. Borca, D.
Grolimund, Y.E. Romanyuk, A.N. Tiwari, Non-vacuum deposition of Cu(In, Ga)Se2 absorber
layers from binder free, alcohol solutions. Prog. Photovolt: Res. Appl. 20, 526–533 (2012).
https://doi.org/10.1002/pip.1246
73. V. Kapur, R. Kemmerle, A. Bansal, J. Haber, J. Schmitzberger, P. Le, D. Guevarra, V. Kapur, T.
Stempien, Manufacturing of “ink based” CIGS SOLAR cells/modules, in Conference Record
of the 33rd IEEE Photovoltaic Specialist Conference (2008), pp. 1–5. https://doi.org/10.1109/
pvsc.2008.4922496
74. V.K. Kapur, A. Bansal, S. Roth, Roadmap for manufacturing cost competitive CIGS modules,
in Conference Record of the 38th IEEE Photovoltaic Specialist Conference (2012), pp. 3343–
3348. https://doi.org/10.1109/pvsc.2012.6318289
75. K.A.W. Horowitz, R. Fu, M. Woodhouse, An analysis of glass-glass CIGS manufacturing
costs. Sol. Energy Mater. Sol. Cells 154, 1–10 (2016)
76. J. Hedström, H. Ohlsén, M. Bodegård, A. Kylner, L. Stolt, D. Hariskos, M. Ruckh, H. W.
Schock, ZnO/CdS/Cu(In,Ga)Se2 thin film solar cells with improved performance, in Confer-
ence Record of the 23rd IEEE Photovoltaic Specialists Conference (IEEE, Piscataway, NJ,
1993), pp. 364–371
77. K. Granath, M. Bodegård, L. Stolt, The effect of NaF on Cu(In, Ga)Se2 thin film solar cells.
Sol. Energy Mater. Sol. Cells 60, 279–293 (2000)
78. D. Rudmann, D. Brémaud, H. Zogg, A.N. Tiwari, Na incorporation into Cu(In, Ga)Se2 for
high-efficiency flexible solar cells on polymer foils. J. Appl. Phys. 97, 084903 (2005)
212 F. Kessler et al.

79. M. Ruckh, D. Schmid, M. Kaiser, R. Schäffler, T. Walter, H.W. Schock, Conference Record of
the 24th IEEE Photovoltaic Specialists Conference (Waikoloa, Hawaii, 1994), pp. 156–159
80. R. Wuerz, A. Eicke, F. Kessler, P. Rogin, O. Yazdani-Assl, Alternative sodium sources for
Cu(In, Ga)Se2 thin-film solar cells on flexible substrates. Thin Solid Films 519, 7268–7271
(2011)
81. W. Thongkham, A. Pankiew, K. Yoodee, S. Chatraphorn, Enhancing efficiency of Cu(In,
Ga)Se2 solar cells on flexible stainless steel foils using NaF co-evaporation. Sol. Energy 92,
189–195 (2013)
82. H. Stange, S. Brunken, H. Hempel, H. Rodriguez-Alvarez, N. Schäfer, D. Greiner, A. Scheu,
J. Lauche, C.A. Kaufmann, T. Unold, D. Abou-Ras, R. Mainz, Effect of Na presence during
CuInSe2 growth on stacking fault annihilation and electronic properties. Appl. Phys. Lett.
107, 152103 (2015)
83. A. Urbaniak, M. Igalson, F. Pianezzi, S. Buecheler, A. Chirilă, P. Reinhard, A.N. Tiwari,
Effects of Na incorporation on electrical properties of Cu(In, Ga)Se2-based photovoltaic
devices on polyimide substrates Sol. Energy Mater. Sol. Cells 128, 52–55 (2014)
84. C. Persson, A. Zunger, Anomalous grain boundary physics in polycrystalline CuInSe2: the
existence of a hole barrier. Phys. Rev. Lett. 91, 266401 (2003)
85. S.H. Wei, S.B. Zhang, A. Zunger, Effects of Na on the electrical and structural properties of
CuInSe2 . J. Appl. Phys. 85, 7214–7218 (1999)
86. Z.K. Yuan, S. Chen, Y. Xie, J.S. Park, H. Xiang, X.G. Gong, S.H. Wei, Na-diffusion enhanced
p-type conductivity in Cu(In, Ga)Se2: a new mechanism for efficient doping in semiconduc-
tors. Adv. Energy Mat. 6, 1601191 (2016)
87. O. Cojocaru-Mirédin, T. Schwarz, P.P. Choi, M. Herbig, R. Wuerz, D. Raabe, Atom probe
tomography studies on the Cu(In,Ga)Se2 grain boundaries. J. Vis. Exp. (74), e50376. https://
doi.org/10.3791/50376 (2013)
88. M.A. Contreras, B. Egaas, P. Dippo, J. Webb, J. Granata, K. Ramanathan, S. Asher, A. Swart-
zlander, R. Noufi, On the role of Na and modifications to Cu(In,Ga)Se2 absorber materials
using thin MF (M = Na, K, Cs) precursor layers, in Conference Record of the 26th IEEE
Photovoltaic Specialists Conference (Anaheim, California, 30 September–3 October 1997),
pp. 359–362
89. A. Laemmle, R. Wuerz, M. Powalla, Efficiency enhancement of Cu(In, Ga)Se2 thin-film
solar cells by a post-deposition treatment with potassium fluoride. Phys. Status Solidi RRL
7, 631–634 (2013). https://doi.org/10.1002/pssr.201307238
90. P. Jackson, D. Hariskos, R. Wuerz, W. Wischmann, M. Powalla, Compositional investigation
of potassium doped Cu(In,Ga)Se2 solar cells with efficiencies up to 20.8%. Phys. Status Solidi
RRL 8, 219–222 (2014)
91. D. Herrmann, P. Kratzert, S. Weeke, M. Zimmer, J. Djordjevic-Reiss, R. Hunger, P. Lindberg,
E. Wallin, O. Lundberg, L. Stolt, CIGS module manufacturing with high deposition rates and
efficiencies, presented at the IEEE 40th Photovoltaic Specification Conference (Denver, CO,
USA, 8–13 June 2014)
92. K.F. Tai, R. Kamada, T. Yagioka, T. Kato, H. Sugimoto, From 20.9 to 22.3% Cu(In,Ga)(S,Se)2
solar cell: reduced recombination rate at the heterojunction and the depletion region due to
K-treatment. Jpn. J. Appl. Phys. 56, 08MC03 (2017)
93. O. Lundberg, E. Wallin, V. Gusak, S. Södergren, S. Chen, S. Lotfi, F. Chalvet, U. Malm, N.
Kaihovirta, P. Mende, G. Jaschke, P. Kratzert, J. Joel, M. Skupinski, P. Lindberg, T. Jarmar, J.
Lundberg, J. Mathiasson, L. Stolt, Improved CIGS modules by KF post deposition treatment
and reduced cell-to-module losses, in 43rd Photovoltaic Specialist Conference (Denver, 2016),
pp. 1293–1296
94. E. Handick, P. Reinhard, R.G. Wilks, F. Pianezzi, T. Kunze, D. Kreikemeyer-Lorenzo, L.
Weinhardt, M. Blum, W. Yang, M. Gorgoi, E. Ikenaga, D. Gerlach, S. Ueda, Y. Yamashita,
T. Chikyow, C. Heske, S. Buecheler, A.N. Tiwari, M. Bär, Formation of a K-In-Se surface
species by NaF/KF postdeposition treatment of Cu(In, Ga)Se2 thin-film solar cell absorbers.
ACS Appl. Mater. Interfaces. 9, 3581–3589 (2017)
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 213

95. E. Handick, P. Reinhard, J.-H. Alsmeier, L. Köhler, F. Pianezzi, S. Krause, M. Gorgoi, E.

Ikenaga, N. Koch, R.G. Wilks, S. Buecheler, A.N. Tiwari, M. Bär, Potassium postdeposition
treatment-induced band gap widening at Cu(In, Ga)Se2 surfaces—reason for performance
leap? ACS Appl. Mater. Interfaces. 7, 27414–27420 (2015)
96. Z.Z. Kish, V.B. Lazarev, E.Y. Peresh, E.E. Semrad, Compounds in In2 Se3 -K2 Se. Neorg. Mater.
24, 1602–1605 (1988)
97. F. Pianezzi, P. Reinhard, A. Chirilă, B. Bissig, S. Nishiwaki, S. Buecheler, A.N. Tiwari,
Unveiling the effects of post-deposition treatment with different alkaline elements on the
electronic properties of CIGS thin film solar cells. Phys. Chem. Chem. Phys. 16, 8843–8851
(2014)
98. H. Elanzeery, F. Babbe, M. Melchiorre, E. Robert, A. Zelenina, S. Siebentritt, Potassium
fluoride ex-situ treatment on both Curich and Cu-poor CuInSe2 thin film solar cells. IEEE J.
Photovolt. 7, 684–689 (2017)
99. I. Khatri, H. Fukai, H. Yamaguchi, M. Sugiyama, T. Nakada, Effect of potassium fluoride post-
deposition treatment on Cu(In, Ga)Se2 thin films and solar cells fabricated onto sodalime glass
substrates. Sol. Energy Mater. Sol. Cells 155, 280–287 (2016)
100. S.A. Jensen, S. Glynn, A. Kanevce, P. Dippo, J.V. Li, D.H. Levi, D. Kuciauskas, Benefi-
cial effect of post-deposition treatment in high-efficiency Cu(In, Ga)Se2 solar cells through
reduced potential fluctuations. J. Appl. Phys. 120, 063106 (2016)
101. A. Vilalta-Clemente, M. Raghuwanshi, S. Duguay, C. Castro, E. Cadel, P. Pareige, P. Jackson,
R. Wuerz, D. Hariskos, W. Witte, Rubidium distribution at atomic scale in high efficient Cu(In,
Ga)Se2 thin-film solar cells. Appl. Phys. Lett. 12, 103105 (2018)
102. P. Schöppe, S. Schönherr, R. Wuerz, W. Wisniewski, G. Martínez-Criado, M. Ritzer, K. Ritter,
C. Ronning, C.S. Schnohr, Rubidium segregation at random grain boundaries in Cu(In, Ga)Se2
absorbers. Nano Energy 42, 307–313 (2017)
103. T. Unold, Talk at EMRS (2018)
104. S. Ishizuka, N. Taguchi, J. Nishinaga, Y. Kamikawa, S. Tanaka, H. Shibata, Group III elemental
composition dependence of RbF post deposition treatment effects on Cu(In, Ga)Se2 thin films
and solar cells. J. Phys. Chem. C 122, 3809–3817 (2018)
105. W.N. Shafarman, J.E. Phillips, Direct current-voltage measurements of the Mo/CuInSe2 con-
tact on operating solar cells, in Conference Record of the 25th IEEE Photovoltaic Specialist
Conference 1996 (Washington DC, USA, 1996), pp. 917–919
106. W. Paszkowicz, R. Minikayev, P. Piszora, D. Trots, M. Knapp, T. Wojciechowski, R. Bacewicz,
Thermal expansion of CuInSe2 in the 11–1,073 K range: an X-ray diffraction study. Appl.
Phys. A 116, 767–780 (2014). https://doi.org/10.1007/s00339-013-8146-9
107. G. Gordillo, F. Mesa, C. Calderon, Electrical and morphological properties of low resistivity
Mo thin films prepared by magnetron sputtering. Braz. J. Phys. 36(3B), 982–985 (2006)
108. D. Abou-Ras, G. Kostorz, D. Bremaud, M. Kälin, F.V. Kurdesau, A.N. Tiwari, M. Döbeli,
Formation and characterisation of MoSe2 for Cu(In, Ga)Se2 based solar cells. Thin Solid
Films 480–481, 433–438 (2005). https://doi.org/10.1016/j.tsf.2004.11.098
109. T. Klinkert, B. Theys, G. Patriarche, M. Jubault, F. Donsanti, J.-F. Guillemoles, D. Lincot,
New insights into the Mo/Cu(In, Ga)Se2 interface in thin film solar cells: Formation and
properties of the MoSe2 interfacial layer. J. Chem. Phys. 145, 154702 (2016). https://doi.org/
10.1063/1.4964677
110. T. Wada, N. Kohara, S. Nishiwaki, T. Negami, Characterization of the Cu(In, Ga)Se2/Mo
interface in CIGS solar cells. Thin Solid Films 387, 118–122 (2001)
111. J.A. Spies, R. Schafer, J.F. Wager, P. Hersh, H.A.S. Platt, D.A. Keszler, G. Schneider, R.
Kykyneshi, J. Tate, X. Liu, A.D. Compaan, W.N. Shafarman, p in double-heterojunction
thin-film solar cell p-layer assessment. Sol. Energy Mater. Sol. Cells 93, 1296–1308 (2009)
112. X. Zhang, M. Kobayashi, A. Yamada, Comparison of Ag(In, Ga)Se2/Mo and Cu(In,
Ga)Se2/Mo interfaces in solar cells. ACS Appl. Mater. Interfaces. 9(19), 16215–16220 (2017).
https://doi.org/10.1021/acsami.7b02548
113. B.K. Choi, M. Kim, K.-H. Jung, J. Kim, K.-S. Yu, Y.J. Chang, Temperature dependence of
band gap in MoSe2 grown by molecular beam epitaxy. Nanoscale Res. Lett. 12, 492 (2017).
https://doi.org/10.1186/s11671-017-2266-7
214 F. Kessler et al.

114. W. Kautek, H. Gerischer, H. Tributsch, The role of carrier diffusion and indirect optical
transitions in the photoelectrochemical behavior of layer type d-band semiconductors. J.
Electrochem. Soc. 127, 2471–2478 (1980)
115. S.M. Delphinec, M. Jayachandranb, C. Sanjeeviraja, Pulsed electrodeposition and character-
ization of molybdenum diselenide thin film. Mater. Res. Bull. 40, 135–147 (2005). https://
doi.org/10.1016/j.materresbull.2004.09.008
116. A.M. Vora, Effect of indium intercalation on various properties of MoSe2 single crystals.
Cryst. Res. Technol. 42(3), 286–289 (2007). https://doi.org/10.1002/crat.200610814
117. K.-J. Hsiao, J.-D. Liu, H.-H. Hsieh, T.-S. Jiang, Electrical impact of MoSe2 on CIGS thin-film
solar cells. Phys. Chem. Chem. Phys. 15, 18174 (2013)
118. J.B. Pang, Y.A. Cai, Q. He, H. Wang, W.L. Jiang, J.J. He, T. Yu, W. Liu, Y. Zhang, Y. Sun,
Preparation and characteristics of MoSe2 interlayer in bifacial Cu(In, Ga)Se2 solar cells.
Phys. Procedia 32, 372–378 (2012)
119. K. Orgassa, H.W. Schock, J.H. Werner, Alternative back contact materials for thin film Cu(In,
Ga)Se2 solar cells. Thin Solid Films 431–432, 387–391 (2003)
120. J. Malmström, S. Schleussner, L. Stolt, Enhanced back reflectance and quantum efficiency in
thin film solar cells with a ZrN back reflector. Appl. Phys. Lett. 85, 2634 (2004). https://doi.
org/10.1063/1.1794860
121. S. Schleussner, T. Kubart, T. Törndahl, M. Edoff, Reactively sputtered ZrN for application as
reflecting back contact in Cu(In, Ga)Se2 solar cells. Thin Solid Films 517, 5548–5552 (2009)
122. B. Bissig, R. Carron, L. Greuter, S. Nishiwaki, E. Avancini, C. Andres, T. Feurer, S. Buecheler,
A.N. Tiwari, Novel back contact reflector for high efficiency and double-graded Cu(In,Ga)Se2
thin-film solar cells. Prog. Photovolt. Res. Appl. 1–7 (2018). https://doi.org/10.1002/pip.3029
123. V. Srikant, D.R. Clarke, On the optical band gap of zinc oxide. J. Appl. Phys. 83(10), 5447–
5451 (1998)
124. C.P. Thompson, S. Hegedus, P. Carcia, R.S. McLean, The effects of device geometry and
TCO/Buffer layers on damp heat accelerated lifetime testing of Cu(In, Ga)Se2 solar cells.
IEEE J. Photovolt. 3, 494 (2013)
125. T. Jäger, Y.E. Romanyuk, S. Nishiwaki, B. Bissig, F. Pianezzi, P. Fuchs, C. Gretener, M.
Döbeli, A.N. Tiwari, Hydrogenated indium oxide window layers for high-efficiency Cu(In,
Ga)Se2 solar cells. J. Appl. Phys. 117, 205301 (2015)
126. Y. Kuwahata, T. Minemoto, Impact of Zn1-xMgxO: Al transparent electrode for buffer-less
Cu(In, Ga)Se2 solar cells. Renew. Energy 65, 113–116 (2014)
127. F.J. Pern, R. Noufi, X. Li, C. DeHart, B. To, Damp-heat induced degradation of transparent
conducting oxides for thin-film solar cells, in Conference Record of the IEEE PVSC (2008),
pp. 43–48
128. R. Menner, S. Paetel, W. Wischmann, M. Powalla, Indium zinc oxide window layer for high-
efficiency Cu(In, Ga)Se2 solar cells. Thin Solid Films 634, 160–164 (2017)
129. T. Shitaya, H. Sato, Single crystal CdS solar cells. J. Appl. Phys. 7, 1348–1353 (1968)
130. R.A. Mickelsen, W. Chen, Polycrystalline thin-film CuInSe2 solar cells, in Records of IEEE
PVSC (1982), pp. 781–785
131. R.R. Potter, C. Eberspacber, L.B. Fabick, Device analysis of CuInSe2/CdZnS/ZnO Solar cells,
in Records of IEEE PVSC (1985), pp. 1659–1664
132. K. Mitchell, C. Eberspacher, J. Ermer, D. Pier, Single and tandem junction CuInSe2 cell and
module technology, in Records IEEE PVSC (1988), pp. 1384–1389
133. J. Kessler, K.O. Velthaus, M. Ruckh, R. Laichinger, H.-W.. Schock, D. Lincot, R. Ortega,
J. Vedel, Chemical bath deposition of CdS on CuInSe2, etching effects and growth kinetics,
in 6th International Photovoltaic Science and Engineering Conference (PVSEC-6) (1992),
pp. 1005–1010
134. M.A. Contreras, M.J. Romer, B. To, F. Hasoon, R. Noufi, S. Ward, K. Ramanathan, Opti-
mization of CBD CdS process in high-efficiency Cu(In, Ga)Se2-based solar cells. Thin Solid
Films 403–404, 204–211 (2002)
135. A. Kylner, The chemical bath deposited CdS/Cu(In, Ga)Se2 interface as revealed by X-ray
photoelectron spectroscopy. J. Electrochem. Soc. 146(5), 1816–1823 (1999)
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 215

136. C. Heske, D. Eich, R. Fink, E. Umbach, T. van Buuren, C. Bostedt, L.J. Terminello, S. Kakar,
M.M. Grush, T.A. Callcott, F.J. Himpsel, D.L. Ederer, R.C.C. Perera, W. Riedl, F. Karg,
Observation of intermixing at the buried CdS/Cu(In, Ga)Se2 thin film solar cell heterojunction.
Appl. Phys. Lett. 74, 1451–1453 (1999)
137. D. Abou-Ras, S. Wagner, B.J. Stanbery, H.-W. Schock, R. Scheer, L. Stolt, S. Siebentritt, D.
Lincot, C. Eberspacher, K. Kushiya, A.N. Tiwari, Innovation highway: breakthrough mile-
stones and key developments in chalcopyrite photovoltaics from a retrospective viewpoint.
Thin Solid Films 633, 2–12 (2017)
138. S. Kurtz, I. Repins, W.K. Metzger, P.J. Verlinden, S. Huang, S. Bowden, I. Tappan, K. Emery,
L.L. Kazmerski, D. Levi, Historical analysis of champion photovoltaic module efficiencies.
IEEE J. Photovolt. 8, 363 (2018)
139. P. Jackson, R. Wuerz, D. Hariskos, E. Lotter, W. Witte, M. Powalla, Effects of heavy alkali
elements in Cu(In,Ga)Se2 solar cells with efficiencies up to 22.6%. Phys. Status Solidi RRL
10, 583–586 (2016)
140. J. Löckinger, S. Nishiwaki, T.P. Weiss, B. Bissig, Y.E. Romanyuk, S. Buecheler, A.N. Tiwari,
TiO2 as intermediate buffer layer in Cu(In, Ga)Se2 solar cells. Sol. Energy Mater. Sol. Cells
174, 397–404 (2018)
141. W. Witte, D. Abou-Ras, D. Hariskos, Chemical bath deposition of Zn(O, S) and CdS buffers:
influence of Cu(In, Ga)Se2 grain orientation. Appl. Phys. Lett. 102, 051607 (2013)
142. T.M. Friedlmeier, P. Jackson, D. Kreikemeyer-Lorenzo, D. Hauschild, O. Kiowski, D.
Hariskos, L. Weinhardt, C. Heske, M. Powalla, A closer look at Initial CdS growth on high-
efficiency Cu(In,Ga)Se2 absorbers using surface-sensitive methods, in Proceedings IEEE
PVSC (2016), pp. 457–461
143. W. Witte, D. Abou-Ras, D. Hariskos, Improved growth of solution-deposited thin films on
polycrystalline Cu(In, Ga)Se2. Phys. Status Solidi RRL 10, 300–304 (2016)
144. D. Hariskos, R. Menner, P. Jackson, S. Paetel, W. Witte, W. Wischmann, M. Powalla, L.
Bürkert, T. Kolb, M. Oertel, B. Dimmler, B. Fuchs, New reaction kinetics for a high-rate
chemical bath deposition of the Zn(S, O) buffer layer for Cu(In, Ga)Se2-based solar cells.
Prog. Photovolt: Res. Appl. 20, 534–542 (2012)
145. T.M. Friedlmeier, P. Jackson, A. Bauer, D. Hariskos, O. Kiowski, R. Wuerz, M. Powalla,
Improved photocurrent in Cu(In,Ga)Se2 solar cells: from 20.8% to 21.7% efficiency with
CdS buffer, and 21.0% Cd-free. IEEE J. Photovolt. 5, 1487–1491 (2015)
146. N. Naghavi, D. Abou-Ras, N. Allsop, N. Barreau, S. Bücheler, A. Ennaoui, C.-H. Fischer,
C. Guillen, D. Hariskos, J. Herrero, R. Klenk, K. Kushiya, D. Lincot, R. Menner, T. Nakada,
C. Platzer-Björkman, S. Spiering, A.N. Tiwari, T. Törndahl, Buffer layers and transparent
conducting oxides for chalcopyrite Cu(In, Ga)(S, Se)2 based thin film photovoltaics: present
status and current developments. Prog. Photovoltaics Res. Appl. 18, 411–433 (2010)
147. J. Palm, T. Dalibor, R. Lechner, S. Pohlner, R. Verma, R. Dietmüller, A. Heiß, H. Vogt, F. Karg,
Cd-free CIS thin film solar modules at 17% efficiency, in Proceedings of the 29th European
Photovoltaic Solar Energy Conference (2014), pp. 1433–1438
148. K. Ramanathan, J. Mann, S. Glynn, S. Christensen, J. Pankow, J. Li, J. Scharf, L. Mansfield,
M. Contreras, R. Noufi, A comparative study of Zn(O,S) buffer layers and CIGS solar cells
fabricated by CBD, ALD, and sputtering, in Proceedings 38th IEEE Photovoltaic Specialists
Conference (2012), pp. 1677–1680
149. W. Witte, D. Hariskos, A. Eicke, R. Menner, O. Kiowski, M. Powalla, Impact of annealing on
Cu(In, Ga)Se2 solar cells with Zn(O, S)/(Zn, Mg)O buffers. Thin Solid Films 535, 180–183
(2013)
150. A. Wachau, J. Schulte, P. Agoston, F. Hübler, A. Steigert, R. Klenk, F. Hergert, H. Eschrich, V.
Probst, Sputtered Zn(O, S) buffer layers for CIGS solar modules—from lab to pilot production.
Prog. Photovoltaics Res. Appl. 25, 696–705 (2017)
151. M.A. Mughal, R. Engelken, R. Sharma, Progress in indium (III) sulfide (In2 S3 ) buffer layer
deposition techniques for CIS, CIGS, and CdTe-based thin film solar cells. Solar Energy 120,
131–146 (2015)
216 F. Kessler et al.

152. W. Witte, D. Hariskos, M. Powalla, Comparison of charge distributions in CIGS thin-film

solar cells with ZnS/(Zn, Mg)O and CdS/i-ZnO buffers. Thin Solid Films 519, 7549–7552
(2011)
153. D. Hariskos, P. Jackson, W. Hempel, S. Paetel, S. Spiering, R. Menner, W. Wischmann, M.
Powalla, Method for a high-rate solution deposition of Zn(O, S) buffer layer for high-efficiency
Cu(In, Ga)Se2-based solar cells. IEEE J. Photovolt. 6, 1321–1326 (2016)
154. T. Kato, A. Handa, T. Yagioka, T. Matsuura, K. Yamamoto, S. Higashi, J.-L. Wu, K.F. Tai, H.
Hiroi, T. Yoshiyama, T. Sakai, H. Sugimoto, Enhanced efficiency of Cd-Free Cu(In, Ga)(Se,
S)2 minimodule via (Zn, Mg)O second buffer layer and alkali metal post-treatment. IEEE J.
Photovolt. 7, 1773–1780 (2017)
155. Module with record efficiency from AVANCIS: Fraunhofer ISE certifies CIGS solar
module with an efficiency of 17.9% (2016). http://www.avancis.de/en/press/news/article/
avancis-erzielt-erneuten-wirkungsgradrekord-fraunhofer-ise-zertifiziert-cigs-solarmodul-
mit-wirkung/
156. N. Barreau, Indium sulfide and relatives in the world of photovoltaics. Sol. Energy 83, 363–371
(2009)
157. D. Hauschild, F. Meyer, A. Benkert, D. Kreikemeyer-Lorenzo, T. Dalibor, J. Palm, M. Blum,
W. Yang, R.G. Wilks, M. Bär, F. Reinert, C. Heske, L. Weinhardt, Improving performance by
Na doping of a buffer layer—chemical and electronic structure of the InxSy:Na/CuIn(S, Se)2
thin-film solar cell interface. Prog. Photovolt. Res. Appl. 26, 359–366 (2018)
158. E. Wallin, U. Malm, T. Jamar, O. Lundberg, M. Edoff, L. Stolt, World-record Cu(In,Ga)Se2-
based thin-film sub-module with 17.4.% efficiency. Prog. Photovolt. 20, 851–854 (2012)
159. K. Ellmer, A. Klein, B. Rech (eds.), Transparent Conductive Zinc Oxide: Basics and Appli-
cations in Thin Film Solar Cells, Chapter 9 (Springer, Berlin/Heidelberg, 2008), pp. 418–419
160. J.V. Li, X. Li, Y. Yan, C.-S. Jiang, W.K. Metzger, I.L. Repins, M.A. Contreras, D.H. Levi,
Influence of sputtering a ZnMgO window layer on the interface and bulk properties of Cu(In,
Ga)Se2 solar cells. J. Vac. Sci. Technol. 27, 2384–2389 (2009)
161. D. Hariskos, B. Fuchs, R. Menner, N. Naghavi, C. Hubert, D. Lincot, M. Powalla, The Zn(S,
O, OH)/ZnMgO buffer in thin-film Cu(In, Ga)(Se, S)2-based solar cells part II: magnetron
sputtering of the ZnMgO buffer layer for in-line co-evaporated Cu(In, Ga)Se2 solar cells.
Prog. Photovolt. Res. Appl. 17, 479–488 (2009)
162. F. Kessler, D. Rudmann, Technological aspects of flexible CIGS solar cells and modules. Sol.
Energy 77, 685–695 (2004). https://doi.org/10.1016/j.solener.2004.04.010
163. K. Herz, A. Eicke, F. Kessler, R. Wächter, M. Powalla, Diffusion barriers for CIGS solar cells
on metallic substrates. Thin Solid Films 43–432, 392–397 (2003)
164. R. Wuerz, A. Eicke, M. Frankenfeld, F. Kessler, M. Powalla, P. Rogin, O. Yazdani-Assl, CIGS
thin-film solar cells on steel substrates. Thin Solid Films 517, 2415–2418 (2009)
165. K.-B. Kim, Effect of metal barrier layer for flexible solar cell devices on stainless steel
substrates. Appl. Sci. Converg. Technol. 26(1), 16–19 (2016). https://doi.org/10.5757/ASCT.
2017.26.1.16
166. P. Reinhard, A. Chirilă, P. Blasch, F. Pianezzi, S. Nishiwaki, S. Buecheler, A.N. Tiwari, Review
of progress toward 20% efficiency flexible CIGS solar cells and manufacturing issues of solar
modules. IEEE J. Photovolt. 3(1), 572–580 (2013). https://doi.org/10.1109/JPHOTOV.2012.
2226869
167. R. Zhang, D.R. Hollars, J. Kanicki, High efficiency Cu(In, Ga)Se2 flexible solar cells fabri-
cated by roll-to-roll metallic precursor co-sputtering method. Jpn. J. Appl. Phys. 52, 092302
(2013). https://doi.org/10.7567/JJAP.52.092302
168. R. Birkmire, E. Eser, S. Fields, W. Shafarman, Cu(InGa)Se2 solar cells on a flexible polymer
web. Prog. Photovolt: Res. Appl. 13, 141–148 (2005). https://doi.org/10.1002/pip.605
169. S. Nakano, T. Matsuoka, S. Kiyama, H. Kawata, N. Nakamura, Y. Nakashima, S. Tsuda, H.
Nishiwaki, M. Ohnishi, I. Nagaoka, Y. Kuwano, Laser patterning method for integrated type
a-Si solar cell submodules. Jpn. J. Appl. Phys. 25, 1936 (1986)
170. G. Heise, A. Börner, M. Dickmann, M. Englmaier, A. Heiss, M. Kemnitzer, J. Konrad, R.
Moser, J. Palm, H. Vogt, H.P. Huber, Demonstration of the monolithic interconnection on CIS
9 CIGS Thin Film Photovoltaic—Approaches and Challenges 217

solar cells by picosecond laser structuring on 30 by 30 cm2 modules. Prog. Photovolt: Res.
Appl. (2014). https://doi.org/10.1002/pip.2552
171. P.-O. Westin, U. Zimmermann, M. Edoff, Laser patterning of P2 interconnect via in thin-film
CIGS PV modules. Sol. Energy Mater. Sol. Cells 09, 1230 (2008). https://doi.org/10.1016/j.
solmat.2008.04.015
172. G. Heise, A. Heiss, H. Vogt, H.P. Huber, Ultrafast lasers improve the efficiency of CIS thin
film solar cells. Phys. Procedia 39, 702–708 (2012)
173. S. Nishiwaki, A. Burn, S. Buecheler, M. Muralt, S. Pilz, V. Romano, R. Witte, L. Krainer,
G.J. Spühler, A.N. Tiwari, A monolithically integrated high-efficiency Cu(In, Ga)Se2 mini-
module structured solely by laser. Prog. Photovolt Res. Appl. 23, 1908–1915 (2015). https://
doi.org/10.1002/pip.2583
174. O. Tober, J. Wienke, M. Winkler, J. Penndorf, J. Griesche, Mater. Res. Soc. Symp. Proc. 763,
371 (2003)
175. F. Kessler, D. Hermann, M. Powalla, Approaches to flexible CIGS thin-film solar cells. Thin
Solid Films 480–481, 491 (2005)
176. H. Vogt, A. Heiss, J. Palm, F. Karg, H. P. Huber, G. Heise, All laser patterning serial intercon-
nection for highly efficient CIGSSe modules, in Proceedings of 26th EUPVSEC (Hamburg,
3DV.2.9, 2011)
177. G. Heise, M. Domke, J. Konrad, F. Pavic, M. Schmidt, H. Vogt, A. Heiss, J. Palm, H.P. Huber,
Monolithical serial interconnects of large CIS solar cells with picosecond laser pulses. Phys.
Procedia 12, 149–155 (2011)
178. G. Heise, A. Heiss, C. Hellwig, T. Kuznicki, H. Vogt, J. Palm, H.P. Huber, Optimization of
picosecond laser structuring for the monolithic serial interconnection of CIS solar cells. Prog.
Photovolt: Res. Appl. 21, 681–692 (2013). https://doi.org/10.1002/pip.1261
179. S. Zoppel, H. Huber, G.A. Reider, Selective ablation of thin Mo and TCO films with fem-
tosecond laser pulses for structuring thin film solar cells. Appl. Phys. A 89, 161–163 (2007)
180. M.M. Kirillova, L.V. Nomerovannaya, M.M. Noskov, Optical properties of molybdenum
single crystals. Sov. Phys. JETP 33, 1210–1214 (1971)
181. M. Domke, L. Nobile, S. Rapp, S. Eiselen, J. Sotrop, H.P. Huber, M. Schmidt, Understanding
thin film laser ablation: the role of the effective penetration depth and the film thickness. Phys.
Proc. 56, 1007–1014 (2014)
182. G. Heise, M. Englmaier, C. Hellwig, T. Kuznicki, S. Sarrach, H. Huber, Laser ablation of thin
molybdenum films on transparent substrates at low fluencies. Appl. Phys. A 102, 173–178
(2011)
183. J. Bovatsek, A. Tamhankar, R.S. Patel, N.M. Bulgakova, J. Bonse, Thin film removal mech-
anisms in ns-laser processing of photovoltaic materials. Thin Solid Films 518, 2897 (2010)
184. L. Zhigilei, Z. Lin, D. Ivanov, Atomistic modeling of short pulse laser ablation of metals:
connections between melting, spallation, and phase explosion. J. Phys. Chem. C 113, 11892
(2009)
185. M. Domke, S. Rapp, M. Schmidt, H. Huber, Ultrafast movies of thin film laser ablation. Appl.
Phys. A 109, 409–420 (2012)
186. J. Sotrop, A. Kersch, M. Domke, G. Heise, H.P. Huber, Numerical simulation of ultrafast
expansion as the driving mechanism for confined laser ablation with ultra-short laser pulses.
Appl. Phys. A 113, 397 (2013)
187. A. Burn, V. Romano, M. Muralt, R. Witte, B. Frei, S. Buecheler, S. Nishiwaki, Selective
ablation of thin films in latest generation CIGS solar cells with picosecond pulses, in Proceed-
ings of SPIE 8243, Laser Applications in Microelectronic and Optoelectronic Manufacturing
(LAMOM) XVII (2012), p. 824318. https://doi.org/10.1117/12.906919
188. J. Bonse, S. Hoehm, S. Kirner, A. Rosenfeld, J. Krueger, Laser-induced periodic surface
structures a scientific evergreen. IEEE J. Sel. Top. Quant. Electron. 23(3) (2017)
189. E. Lotter, B. Freilinger, R. Wächter, M. Powalla, Laser patterning of the semiconductor layer of
CIGS modules under industrial conditions, in Proceedings of the 16th European Photovoltaic
Solar Energy Conference (Glasgow, UK, 1–5 May 2000), pp. 779–782
218 F. Kessler et al.

190. P.-O. Westin, U. Zimmermann, M. Ruth, M. Edoff, Next generation interconnective laser
patterning of CIGS thin film modules. Sol. Energy Mater. Sol. Cells 95, 1062–1068 (2011)
191. P. Gecys, E. Markauskas, A. Zemaitis, G. Raciukaitis, Variation of P2 series interconnects
electrical conductivity in the CIGS solar cells by picosecond laser-induced modification. Sol.
Energy 132, 493–502 (2016)
192. R. Moser, M. Domke, G. Heise, G. Marowsky, H.P. Huber, Nanosecond laser lift-off of a
copper-Indium-diselenide thin film at a wavelength of 1342 nm. Phys. Procedia 41, 739–744
(2013)
193. S. Rapp, M. Schmidt, H.P. Huber, Selective femtosecond laser structuring of dielectric thin
films with different band gaps: a time-resolved study of ablation mechanisms. Appl. Phys. A
122 (2016)
194. G. Heise, M. Dickmann, M. Domke, A. Heiss, T. Kuznicki, J. Palm, I. Richter, H. Vogt, H.
Huber, Investigation of the ablation of zinc oxide thin films on copper–indium-selenide layers
by ps laser pulses. Appl. Phys. A 104, 387–393 (2011)
195. P. Gečys, E. Markauskas, S. Nishiwaki, S. Buecheler, R. De Loor, A. Burn, V. Romano, G.
Račiukaitis, CIGS thin-film solar module processing: case of high-speed laser scribing. Sci.
Rep. 7, 40502 (2017). https://doi.org/10.1038/srep40502
196. A. Burn, M. Muralt, S. Pilz, V. Romano, R. Witte, B. Frei, S. Buecheler, S. Nishiwaki, L.
Krainer, All fiber laser scribing of Cu(In, Ga)Se2 thin-film solar modules. Phys. Procedia 41,
713–722 (2013)
197. D.J. Hwang, S. Kuk, Z. Wang, S. Fu, T. Zhang, G. Kim, W.M. Kim, J. Jeong, Laser scribing
of CIGS thin-film solar cell on flexible substrate. Appl. Phys. A 123, 55 (2017)
198. V. Probst, A. Jasenek, C. Sandfort, A. Letsch, I. Koetschau, T. Hahn, J. Feichtinger, H.
Eschrich, Innovative front end processing for next generation CIS module production. Jpn. J.
Appl. Phys. 54, 08KC12 (2015). http://dx.doi.org/10.7567/JJAP.54.08KC12
199. M.L. Crozier, A. Brunton, S.J. Henley, J.D. Shephard, A. Abbas, J.W. Bowers, P.M.
Kaminski, J.M. Walls, Inkjet and laser hybrid processing for series interconnection of thin
film photovoltaics. Mater. Res. Innovations 18(7), 509–514 (2014). https://doi.org/10.1179/
1433075X14Y.0000000260
200. J. Britt, S. Wiedeman, S. Albright, Process development for CIGS-based thin-film photovoltaic
modules. Phase II technical report. NREL/SR-520–29227 (2000)
201. E. Niemi, J. Sterner, P. Carlsson, J. Oliv, E. Jaremalm, S. Lindström, All-sputtered flexible
CIGS cells at high speed, in Proceedings of 31st EU PVSEC 2015 (Hamburg, 2015), pp. 1010–
1013
202. H. Shen, T. Duong, J. Peng, D. Jacobs, N. Wu, J. Gong, Y. Wu, S.K. Karuturi, X. Fu, K.
Weber, X. Xiao, T.P. White, K. Catchpole, Energy Environ. Sci. 11, 394 (2018). https://doi.
org/10.1039/c7ee02627g
Chapter 10
Perovskite Photovoltaics: From
Laboratory to Industry

D. Forgacs, K. Wojciechowski and O. Malinkiewicz

Abstract The following pages of this book will discuss the current state-of-the-art of
‘perovskite’ photovoltaic technology, that has been awe-inspiring in the last decade.
Since the first edition of HELC-PV, perovskite solar cells appeared and have been
the dominating research topic in photovoltaics, attracting tremendous scientific and
public interest. The technology advanced rapidly, and now in 2018, there are already
several companies existing with the aim of bringing this technology to the market. As
such, it has been a pleasure and an honour for Saule Technologies to be approached
by the editors to review the advancement of this swiftly developing field. While
the fabrication of P1 R10 high record efficiencies with record high efficiencies has
been achieved already, there are still open questions regarding fundamental physical
and chemical processes taking place in such a device. There are a vast number of
publications available that scrutinize the accumulated knowledge and observations
regarding this unique class of materials. As such, it is a challenging—yet rewarding
task to write a chapter that gives an objective perspective on the topic. Our aim
was to create something that provides the readers with a valuable overview that will
stay relevant even when the current state-of-the art will be outdated. We intended to
describe the basic principles that guide the characteristics of perovskites, and then
connect them with their function in solar cells and the related scientific discoveries.
We hope that this chapter will provide the reader with a basic framework of knowledge
that can be referred to when reading future works dealing with perovskites. We hope

D. Forgacs (B)
Director of Knowledge Management, Saule Technologies,
Dunska 11. (WPT, Sigma building 1st floor), 54-427 Wroclaw, Poland
e-mail: [email protected]
K. Wojciechowski
Scientific Director, Saule Technologies,
Dunska 11. (WPT, Sigma building 1st floor), 54-427 Wroclaw, Poland
e-mail: [email protected]
O. Malinkiewicz
Chief Technology Officer, Saule Technologies,
Dunska 11. (WPT, Sigma building 1st floor), 54-427 Wroclaw, Poland
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 219

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_10
220 D. Forgacs et al.

that you will find this work helpful, and in case you do so—please do not hesitate to
recommend it to your colleagues, friends and acquaintances.

10.1 Introduction

Dear reader,
The following pages of this book will discuss the current state-of-the-art of ‘per-
ovskite’ photovoltaic technology, that has been awe-inspiring in the last decade.
Since the first edition of HELC-PV, perovskite solar cells appeared and have been
the dominating research topic in photovoltaics, attracting tremendous scientific and
public interest. The technology advanced rapidly, and now in 2018, there are already
several companies existing with the aim of bringing this technology to the market. As
such, it has been a pleasure and an honour for Saule Technologies to be approached
by the editors to review the advancement of this swiftly developing field. While the
fabrication of P1 R10 high record efficiencies has been achieved already, there are
still open questions regarding fundamental physical and chemical processes taking
place in such a device. There are a vast number of publications available that scru-
tinize the accumulated knowledge and observations regarding this unique class of
materials. As such, it is a challenging—yet rewarding task to write a chapter that
gives an objective perspective on the topic. Our aim was to create something that
provides the readers with a valuable overview that will stay relevant even when the
current state-of-the art will be outdated. We intended to describe the basic principles
that guide the characteristics of perovskites, and then connect them with their func-
tion in solar cells and the related scientific discoveries. We hope that this chapter
will provide the reader with a basic framework of knowledge that can be referred to
when reading future works dealing with perovskites. We hope that you will find this
work helpful, and in case you do so—please do not hesitate to recommend it to your
colleagues, friends and acquaintances.
Solar cells based on halide perovskites have been a scientific sensation for the last
5 years preceding the writing of this book. One reason why perovskite solar cells
attracted tremendous interest is that these class of materials possess intrinsically ben-
eficial qualities, as well as offer the possibility to engineer their properties to match
the requirements and mitigate the effect of their less desirable traits. This peculiar
quality stems from perovskites’ unique characteristics among semiconductors. While
in the case of solar cell technologies that utilize inorganic semiconductors as light
harvesting materials, such as silicon (Si), cadmium-telluride (CdTe), copper-indium-
gallium-diselenide (CIGS), gallium-arsenide (GaAs) and Copper-tin-zinc-selenide
(CZTSe), there are covalent bonds between the atoms in the crystal lattice, the case
of perovskite solar cells is different. The perovskite lattice is comprised of ions and
held together by ionic interactions—although it has been suggested that there is some
covalent character present as well, due to the low electronegativity of heavier halides
[1].
10 Perovskite Photovoltaics: From Laboratory to Industry 221

While halide perovskites are heavily investigated as solar cell materials only
recently, this class of compounds is not new in the field of optoelectronics. Their his-
tory started back in the early nineties, when they were extensively studied by David
Mitzi in the laboratories of IBM [2–4]. He used perovskites as an active material for
diodes and transistors. However, his concerns regarding their sensitivity to moisture
delayed the pursuit of perovskites as viable photovoltaic (PV) materials. The change
came in 2009, when a Japanese research group introduced a perovskite layer into a
liquid dye-sensitized solar cell (DSSC) architecture for the first time [5]. The team
led by prof. Miyasaka used methylammonium lead iodide (CH3 NH3 PbI3 ) perovskite
as a sensitizer on mesostructured titanium dioxide (TiO2 ) achieving a power conver-
sion efficiency (PCE) of 3.8%. In 2011, a 6.5% efficient device was reported from
the research group of Nam-Gyu Park [6]. Unfortunately, the traditional iodide/iodine
liquid electrolyte used in these architectures was dissolving the photoactive mate-
rial, leading to the rapid degradation of fabricated solar cells. Three years later, the
group of prof. Kanatzidis announced a dramatic improvement of the efficiency of
DSSCs by replacing the liquid electrolyte with a tin-based (CsSnI3 ) perovskite as a
hole transporting material (HTM). In doing so, they have created the first perovskite-
containing solid-state DSSC [7]. Power conversion efficiency exceeding 10% was
achieved, significantly overpassing state-of-the-art ss-DSSC values. Another break-
through came soon after, when 2 research groups reported similar results in a very
close succession. Henry Snaith from the University of Oxford, and Nam-Gyu Park
together with the team of École polytechnique fédérale de Lausanne (EPFL) have
replaced the liquid electrolyte with the solid state spiro-OMeTAD to be used as the
HTM, eliminating the cause of perovskite’s fast degradation observed in Miyasaka’s
cells [8, 9]. Furthermore, the Oxford group reported an additional ground-breaking
discovery. By replacing the mesoporous TiO2 film with an insulating Al2 O3 scaf-
fold, their devices reached even higher performances than the reference ones. This
highlighted that perovskites are capable of ambipolar charge transport over large dis-
tances (exceeding the material’s absorption lengths), and there is no explicit need for
a mesoporous electron accepting material to be present for efficient charge collection
[10]. Consequently, this paved the way for the realization of perovskite-based tradi-
tional thin-film architectures. These seminal works attracted the attention of many
scientists working on DSSCs and organic photovoltaics (OPVs). Due to architec-
tural similarities, years of OPV and DSSC research expertise turned out to be very
relevant for the emerging perovskite PV technology. The joint effort and the accu-
mulated knowledge facilitated a rapid development, and in a few years’ time many
breakthrough achievements have been reached. The following sections will attempt
to give a straightforward explanation on why these group of materials have been so
successful, as well as review the challenges associated with their industrialization.
222 D. Forgacs et al.

10.2 Perovskite Properties

Since the first observation of the photoelectric effect by Becquerel in 1839, scien-
tists around the globe were seeking for the perfect material that could be used to
produce power from the most ubiquitous power source in our vicinity—the Sun
[11]. The ultimate goal is to make a great solar cell—efficient, cheap, long lasting,
and environmentally friendly. To reach this goal, a special photoactive material has
to be developed that fulfils several unique requirements. The most important fea-
tures are excellent absorption and transport properties, combined with an easy and
low-cost fabrication. While different semiconducting materials offer some of these
features, the peculiarity of halide perovskites is that they fulfil most of the above
mentioned requirements. This originates from their special crystal structure which
will be discussed first, examining how the components contribute to the properties
of the material.
Methylammonium lead iodide (MAPbI3 ) is up to date the most heavily researched,
archetypical halide perovskite material, and can therefore be seen as a standard
perovskite and a model compound [12, 13]. Like most of the perovskites used in
photovoltaics it consists of three main components: a monovalent cation (MA+ ),
a divalent metallic cation (Pb2+ ), and halogenic anions (I− ). Before we dive into
the physics that govern the exact role of each of these components, it is worth to
take a closer look at the whole picture. Organic-inorganic metal halides are actually
only a small part of a much bigger perovskite family, that adopt similar crystal
structures. Figure 10.1. represents the crystal structure of perovskites, and highlights
the different types of materials in this family.

10.2.1 General Structure

The ionic structure can be described by the general formula ABX3 with a specific
spatial distribution of the components. Here, A is a mono/divalent cation, B is a
di/tetravalent cation, and X is a mono/divalent anion. The first classification of such
a crystal lattice arrangement was made by Gustav Rose, who had found CaTiO3 in
the Ural mountains in 1839, and it has been named after the Russian mineralogist and
nobleman, Lev Perovski. Over time, it has been discovered that there are hundreds
of compositions that adopt a perovskite structure. It was first described by Victor
Goldschmidt [14]. Initially discovered materials were minerals of natural origin,
however a scientific interest stimulated appearance of synthetic perovskites as well.
Researchers demonstrated piezoelectric and ferroelectric properties, and even high
temperature superconductivity in perovskite material, all due to their structure.
In order to form a perovskite crystal, the constituting ions must fulfil require-
ments regarding their size and charge. These parameters can be described by the
tolerance factor, which was studied in different works addressing both oxide and
halide perovskites [15].
10 Perovskite Photovoltaics: From Laboratory to Industry 223

Fig. 10.1 The crystal structure of perovskites, and the different types of materials in this family.
A is a mono/divalent cation, B is a di/tetravalent cation, and X is a mono/divalent anion. Courtesy
of saule technologies. https://onlinelibrary.wiley.com/doi/abs/10.1002/solr.201900144 [144]

In an ideal case, this structure has a cubic symmetry, corresponding to the tolerance
factor values that are close to 1. To fulfil this, the A-site ion must be much larger
than the B-site one. Orthorhombic, rhombohedral or tetragonal structures can also
form, based on the size of the constituting ions. Importantly, the type of crystalline
phase affects the photovoltaic performance. As the structure is dependent on the
temperature as well, the same material can exist in different symmetries based on
thermal conditions [16]. As such, the phase transition points have to be taken into
account when designing a solar cell that has to withstand thermal cycling.
Now, having an insight into the general structure of perovskites, we can take a
closer look at how its individual components contribute to their properties.

10.2.2 The Role of the Metallic Cation

First of all, in order to harvest energy from the light that reaches the surface of a solar
cell, the photoactive material has to be a semiconductor with desirable absorption
224 D. Forgacs et al.

properties. In an ideal case, it has a direct band gap that allows the harvesting of
most of the incident photons with the least possible losses. The state-of-the-art high
efficiency semiconductor for photovoltaics is GaAs [17]. Can perovskites compete
with them? To answer this question, let’s take a look at the electronic structure of
these materials.
Each semiconductor material can be classified as having either an indirect or a
direct band gap, and it has a strong implication on how the lattice can interact with
photons [18]. In the case of indirect band gap materials, the electromagnetic radiation
has to interact with phonons—the vibrational motion of the atoms in the condensed
matter—in order to be absorbed. Due to this requirement of the change in crystal
momentum, the chance that a photon can be harvested in a solar cell that is based on
an indirect band gap material are orders of magnitude less than in the case of a direct
one. Consequently, the thickness of the active layer in such a solar cell has to be
orders of magnitude larger, to have a comparable light harvesting capability. Indirect
bandgap semiconductors, like silicon, need to be much thicker in order to efficiently
harvest all the usable energy of sunlight. This was partially overcome by structuring
their surfaces to increase the optical path length, nevertheless their optimal thickness
remains in the range of around 100 μm [19, 20]. This value is more than 2 orders
of magnitude larger than the ones reported for optimized perovskite photovoltaics,
ranging from 300 to 1000 nm.
DFT calculations have shown that in the case of MAPbI3 the [PbI6 ]4− octahedra
framework is dominating the energetic levels near the band edge [21]. There are
controversies remaining about the electronic contribution of the A-site MA+ cations.
Some studies claim them to be electronically inactive, only influencing the tolerance
factor and the structural stability of the material [22]. However, a recent work has
demonstrated an experimental evidence that the nitrogen atom in the MA+ cation
actually has an effect on the electronic structure of MAPbI3 [23]. Nevertheless,
the valence band maximum (VBM) has strong Pb s and I p antibonding character,
whereas the conduction band minimum (CBM) is mostly contributed from the Pb
p state [24]. In the case of GaAs, the VBM has a p character, and the CBM has
an s one [21]. The main difference comes from the states near the conduction band
minimum, with the s-type displaying more dispersity and p-type ones being more
dense. Simply put, the presence of the p-character in the CBM of perovskites leads
to a very high density of states near the band edge, leading to substantial absorption
characteristics that outclass Si and even GaAs especially in the visible part of the
spectrum (Fig. 10.2).
With the advancement of the perovskite field, an interest arose in developing
perovskites with altered B-site cations. One reason behind this intention is to replace
the most commonly used lead in halide perovskite absorbers, which is frequently
considered as a limitation in their large-scale application.
10 Perovskite Photovoltaics: From Laboratory to Industry 225

Fig. 10.2 (left) The comparison of transitions in a Silicon, b GaAs and c Halide perovskites,
and (right) the absorption coefficient of the most common semiconductor materials for solar cell
applications as the function of photon energy. Reprinted from [21, 140]. Copyright 2015 Royal
Society of Chemistry and 2014 American Chemical Society

10.2.3 The Role of the Halide Anion

As mentioned earlier, having a direct band gap has beneficial effects in itself for the
application of the material in photovoltaics. However, due to the unique ionic nature
of perovskite materials, they offer an additional advantage that the bandgap can be
engineered over the whole visible spectrum, extending into the ultraviolet and near
infrared as well [25]. This is again due to the [PbX6 ]4− octahedra, and its strong
impact on the electronic characteristics. An exchange of a halide ion X alters the size
of the unit cell, resulting in a change of the band gap. It was observed that if phase
stability requirements are met, the absorption onset can be shifted easily [26, 27].
This feature motivated researchers to investigate alternative perovskite composi-
tions displaying different band gap levels than the archetypical MAPbI3 to fit to the
requirements of the particular applications. The initial investigation of this topic was
carried out by Noh et al. [28]. With a simple solution mixing approach, the successful
alloying of MAPbI3 and MAPbBr3 was achieved. In this work, it was highlighted that
increased bromide inclusion leads to a larger extent of unit cell contraction, which in
turn leads to an increase in the material’s band gap, which is in line with Vegard’s law
[29]. Figure 10.3 represents the change in absorption coefficient in MAPb(Ix Br1-x )3
systems as a function of composition.
With the decreasing size of the halides, the band gap increases. In the case of
MAPbX3 the values shift from 1.55 eV for X = I to 2.3 eV and 3.4 eV for X = Br
and X = Cl, respectively, making this class of materials interesting for both single-
and multi-junction photovoltaics, as well as light emitting applications. Besides the
versatility in applications, this also enables the design of solar cells with different
colours. Mixed halide compositions however exhibit a photo-instability, leading to
phase segregation under illumination [30]. When aiming to tune the band gap in
the range that is interesting for tandem applications, researchers have encountered
that solar cells exhibit a subpar Voc with respect to their band gap energy. Photo-
luminescence measurements have indicated that this originates from the formation
226 D. Forgacs et al.

Fig. 10.3 The change in

absorption coefficient in
MAPb(Ix Br1-x )3 systems as
a function of composition.
Reprinted from [102]

of iodine-rich phases with below band gap energy, acting as recombination centres
[30].

10.2.4 The Role of Monovalent Cations

As mentioned before, the role of cations in the electronic structure of perovskites is

still under debate. They however contribute significantly to the solution processabil-
ity of these semiconductors. The small monovalent cations are easily solubilized by
strong polar solvents, while the halide counterions in their salts help in the complex-
ation of plumbate ions [31]. Besides, they have a strong implication on the geometry
of the lattice, which enables the rational design and engineering of the perovskite
material, giving rise to special features when done properly. Let’s take a look at them.
The exchange of A-site cations was first explored when researchers have reported
perovskite solar cells that incorporated formamidinium (FA) A-site cations instead
of the traditional MA ones [32]. The slightly larger size of the FA cation changes
the lattice constant, resulting in a redshift of the absorption onset. FAPbI3 exhibits
a band gap of 1.48 eV, leading to an increment of the maximum achievable short-
circuit current. The tunability of the material was also demonstrated, with a transition
between a tetragonal and cubic phase [27].
Inspired by the tunability of the band gap and the improved thermal stability
of perovskites with cations other than MA, the rational engineering of perovskite
alloys emerged as a robust approach to fabricate materials with desired qualities.
The first work was carried out by Jeon et al. [33]. In this work, MAPbBr3 was
utilized to stabilize the perovskite phase of FAPbI3 , enabling the thermal treatment
to be lowered to 100 °C, and leading to highly efficient solar cells, setting the record
performance at the time with 17.9%.
10 Perovskite Photovoltaics: From Laboratory to Industry 227

Fig. 10.4 The structure of 2D/3D perovskites with two different stoichiometries, as simulated by
Tsai et al. [44]. Reprinted from [44]

This initiated an important research direction, and various compositions were then
explored with the aim of improving the performance of the materials. It was found
that besides the organic MA+ and FA+ cations, inorganic cations, such as Cs+ , Rb+
or K+ can be successfully mixed into a composition of halide perovskites [34–38].
Besides the mixed compositions, solar cells based on fully inorganic perovskites
have also been demonstrared [39, 40]. One of the advantages of these compounds
is high thermal stability, which is in contrast with MAPbI3 that has been shown to
degrade even when kept at moderate temperatures of around 60 °C for a prolonged
time [41, 42] (Fig. 10.4).
Besides the combination of inorganic cations, alternative organic cations have also
been investigated, with longer chain amines being one of specific interest. Their larger
size results in the formation of a 2D, Ruddlesden-Popper layered perovskite structure
[43]. By tuning the stoichiometry of the material, hybrid 2D/3D compositions can
be achieved, enabling the tuning of perovskite properties, such as the band gap,
conductivity, and stability [44, 45]. When an appropriate composition is chosen, this
provides an interesting avenue to reach higher stability without compromising charge
transport. In 2017, Grancini et al. have demonstrated a 2D/3D perovskite solar cell
with a stability over 10000 h under constant illumination [46].

10.3 Processing Methods

So now we understand the factors that contribute to the attractiveness of perovskite

absorbers. Let’s take a look at the possibilities of creating the photoactive layer in
this PV technology.
228 D. Forgacs et al.

Perovskites are unique in this aspect as well. In their case, the deposition is fol-
lowed by a crystallization step, which will have a great impact on the resulting
performance of the solar cell. This is a facile process, as it can happen spontaneously
in the presence of the required precursors [47]. Achieving a material with the desired
properties however, requires the complex understanding of the system and the mul-
titude of contributing factors guiding its evolution [38].
Polycrystalline perovskite thin films can be processed from solution, by evapora-
tion, or by hybrid/alternative methods that combine these approaches [48]. Processing
methods can be further categorized based on the number of deposition steps used to
form perovskite layers, and in most cases films are formed in a one-step or a two-step
method. The processing conditions are closely related to the material’s composition
that is aimed to be formed, but in this chapter we will discuss these two factors
separately, in an attempt to give a more structured overview (Fig. 10.5).
We will start with the discussion on a solution-based processing, as it has been by
far the most common way of producing perovskite films. Also, it holds the highest
prospect of delivering ultimately low manufacturing costs [49].

Fig. 10.5 Processing methods of perovskite thin films. a co-evaporation [76], b 2-step solution
processing via in situ dipping [141], c one- and two-step solution processing via sequential coating
[142], d vapor assisted solution processing [79], e antisolvent method [66]. Reprinted from ref [66,
76, 79, 141, 142]. Copyright 2013–2014 Nature publishing, 2014 American Chemical Society, 2014
AIP publishing
10 Perovskite Photovoltaics: From Laboratory to Industry 229

10.3.1 Solution Processing

The chemical quality, the age, the concentration and the stoichiometry of the pre-
cursors, the coordinating strength and boiling point of the solvents, the atmospheric
composition of the processing environment, the type and duration of post treatments
as well as the surface onto which perovskite is grown—everything will have an effect
on the thin-film morphology and the optoelectronic properties [50, 51]. This will in
turn define its crystallographic properties and greatly affect the device performance.
Separating these parameters is a difficult task, and even though several research
works have addressed these topics, drawing definitive conclusions from the results
requires a careful approach.
The most straightforward approach is the one-step solution processing. In this
case, the precursors (halide salts of the metal B and of the cation A) are dissolved in
a polar aprotic solvent and deposited onto a substrate, using a spin-coating method
for instance. Subsequently, the samples are annealed, leaving a thin film of per-
ovskite on the substrate (Fig. 10.6c). Dimethylformamide (DMF) or dimethyl sul-
foxide (DMSO) are the most widely used solvents, however alternatives such as
γ-butyrolactone (GBL), N-methyl-2-pyrrolidone (NMP), 2-methoxyethanol and var-
ious solvent mixtures have also been investigated [52, 53]. While the implementation
of this approach is easy, gaining control over the morphology of the resulting film
requires profound understanding of the complex processes involved in the perovskite
formation.
In the previous sections, it has been described that the A-site cations affect the
geometry of the perovskite structure. It also has to be noted that their contribution
to processability is of significant importance. It was observed that perovskite pre-
cursor formulations do not contain only solvated ions, but also various colloidal
networks, comprising metal-halide based coordination complexes [31]. It was also

Fig. 10.6 SEM images of perovskite layers deposited from precursor solutions with different
stoichiometries. The ratio of the components is represented in the top-right corner of the images.
Reprinted from [31]
230 D. Forgacs et al.

shown that the size and density of these colloids can have a defining effect on the
nucleation and growth pathways of perovskite crystals, and as such, gaining control
over these processes is of high importance. The colloidal framework is associated
with lead polyhalides, which is accompanied by coordinating solvent molecules,
forming Lewis acid-base adducts [13, 52]. It was shown that by engineering the
strength of the bonds between the solvent molecules and the lead halide species the
morphology of the perovskite films can be improved [54].
Initial works that utilized stoichiometric PbI2 : MAI precursor ratios have reported
poor substrate coverage, with needle-like morphologies of the perovskite [55, 56].
This was attributed to the existence of lead-polyhalide complexes with a preferred
rapid axial growth orientation. It led to poor device performance with a high varia-
tion depending on the measurement area. For high efficiency and reproducible solar
cells, uniform, pinhole-free perovskite layers are needed with large grain sizes. As
such, various approaches and materials have been investigated to provide a greater
extent of control on perovskite formation by enlarging the time window available
for processing [57, 58]. It was proposed that highly coordinative species form slow
down the nucleation and crystallization of perovskites [59]. Additives (e.g.: DMSO,
1,8-diiodooctane (DIO)) aiming to reach this goal have been utilized with a success
to various extents [56, 60, 61].
It was observed that the presence of Cl− ions can also be beneficial for the mor-
phology of the perovskite layer [62]. Thin films prepared from a mixture of PbCl2
and MAI were initially considered as a ‘mixed halide’ perovskite, with chloride ions
partially replacing iodide in a material’s composition [9]. Later reports, and more
in-depth analyses have shown that chloride ions are present in the film in very little
amounts (less than 3–4%), and due to their size being too small to fit in, they are
unlikely to be incorporated into the crystal lattice [63]. The improved performance
is more likely a combination of improved growth with electronic effects that arise
from the presence of chloride ions.
The observation that the stoichiometry of the perovskite thin film can be very
different from the precursor composition opened up new avenues for processing
these layers. It was found that there are various ionic species that can be considered
as ‘spectating’, forming volatile compounds and leaving the system during the growth
of the film [58, 62]. The examination of the processing window for different precursor
materials and compositions was first carried out by Moore et al. [58]. Utilizing in situ
wide angle X-ray scattering WAXS, they analysed the transformation kinetics of
different lead salts to perovskite. They found especially attractive characteristics in
the case of lead acetate.
The nucleation and the growth of perovskite polycrystalline films strongly depend
on the interactions of a solvent with the precursors and the substrate. The solvent
vapour’s partial pressure determines the speed of liquid evaporation, which has an
impact on formation kinetics. As the solvent evaporates, the concentration of the
precursor is increasing, eventually reaching supersaturation [64]. This leads to nucle-
ation, which can be either homogeneous or heterogeneous. The surface tension of
the precursor dispersion and the surface energy and chemistry of the substrate will
have a strong impact on the quality of the grown perovskite layer [65]. As such,
10 Perovskite Photovoltaics: From Laboratory to Industry 231

engineering these parameters became an important part of the solution-processed

perovskite solar cell field.
A breakthrough came in 2014, when Jeon et al. and Xiao et al. reported a method
in which they treated the sample with a non-polar solvent during spinning [66, 67].
Toluene and chlorobenzene were utilized for this purpose, as they are miscible with
the commonly used GBL, DMF and DMSO, however they are poor solvents for the
perovskite precursors. If the wet film is treated with the antisolvent in the proper time
window, the residual solvents will be washed out immediately, and a uniform and
smooth solid layer will be left on the substrate with high coverage. This approach
gave researchers the possibility to lock-in and freeze a perovskite structure dur-
ing its evolution, quenching its further progression. After optimizing the precursor
composition and processing conditions—with special attention to the time window
when the antisolvent is applied—this method can provide a reproducible approach
for deposition of uniform layers that yield high performance devices [68]. As such,
most efficient solar cells to date are fabricated with an anti-solvent strategy [34, 69,
70]. The main limitation of this approach is its scalability to industrial formats.
In a two-step processing, the lead compound is deposited first, and subsequently it
is converted to perovskite by interdiffusion with the organic salt by a second coating
step [71, 72]. The advantage of this method is that it is simpler to control the deposition
of a single compound, and it allows the formation of homogeneous and continuous
films. A crucial requirement of this approach is to enable the penetration of the
organic cation deep into the inorganic layer, as its maximum depth will determine the
obtainable perovskite thickness limiting the maximum achievable photocurrent and
hence the maximum efficiency of the solar cell. This represented a challenge for initial
experiments utilizing the 2-step approach. With the advancement of this technique,
it was found that the infiltration of the cations can be facilitated by improving the
deposition of the first inorganic layer [73]. The reported methods include varying
the solvents used for deposition, utilizing complexing additives, and even structuring
the metal halide [74]. Using these approaches, the best performing devices reached
PCEs higher than 20% [75].

10.3.2 Evaporation

Evaporated perovskites can also be fabricated by one and two-step methods. The
deposition is done by placing the precursors into separate crucibles in an enclosed
chamber, and then heating them up to their sublimation temperature in high vacuum.
The vapours of the materials condense and react on the substrates situated at a given
distance from the sources, forming compact perovskite films at low temperature.
One major advantage of this approach is its intrinsic additivity, as any combination
of materials can be deposited in a stack, without the necessity of fulfilling solvent
orthogonality requirement as it is in a solution-based process. Vacuum-grown per-
ovskites reported in literature were shown to have highly smooth and uniform layers,
providing a reliable processing approach [76]. On the other hand, being a vacuum
232 D. Forgacs et al.

based process, it requires a sophisticated equipment and higher maintenance cost—

which increases the cost of production.
The first vacuum deposited perovskite solar cells were reported in 2013 by Liu
et al. [76]. This work highlighted the first perovskite solar cell in a planar n-i-p
architecture, utilising TiO2 and spiro-MeOTAD as electron and hole selective layers,
respectively. The photoactive layer was based on co-deposited PbCl2 and MAI. In
order to form a stoichiometric composition, these layers had to be exposed to ther-
mal annealing to remove the excess materials. A breakthrough in vacuum deposited
perovskites was achieved by the group of Bolink only a few months later [77]. In that
report, a co-evaporation process was carried out from stoichiometric MAI and PbI2 ,
and no further thermal treatment was needed to achieve a high quality perovskite
layer, and in turn highly efficient solar cell performance. This in combination with
utilizing organic interfacial layers resulted in a great reduction of processing tem-
peratures, making it a versatile method compatible with flexible substrates.
The upcoming years of development in this processing method led to fully vacuum
deposited perovskite solar cells, harnessing the full potential of this approach [78]. By
finding organic interfacial layers with suitable energy levels and appropriate dopants,
Momblona et al. have prepared highly efficient p-i-n and n-i-p perovskite solar cells
by a sequential evaporation of the complete device architecture for the first time.
A significant aspect of the above mentioned work is that it provides an approach
towards perovskite/perovskite tandem solar cells.

10.3.3 Hybrid Processes

Vapor-Assisted Solution Processing (VASP) is a hybrid technique aimed to combine

the high quality layers and control on growth provided by vapor methods, meanwhile
maintaining the simplicity of solution processing [79]. First demonstrated by Chen
et al. in 2013, this approach led to the highest performance in a planar architecture at
the time. The method consists of depositing a metal-halide layer from solution, then
exposing it to the vapours of the organic halide material. Several alternative hybrid
techniques have been developed and reported since then, with various performances
[80].

10.4 Device Architectures

In most perovskite solar cells, the photoactive layer is contacted with charge transport
layers to selectively allow only one type of charge carrier to be extracted [81]. Doing
so will have beneficial effect on the performance of a solar cell, as it will prevent the
recombination of carriers in the electrode layer The transport material has to meet
several requirements to fulfil this role.
10 Perovskite Photovoltaics: From Laboratory to Industry 233

Creating a contact with pertinent selectivity requires the energy levels of the
valence and conduction band, the Fermi-level and the band gap of the transport
layer to be of appropriate values [82]. This facilitates the extraction of one type
of charge carrier, while blocking the other. Fulfilling these requirements will reduce
fundamental electrical losses. A good selective layer will also have negligible optical
losses, and as such it is important to minimize its absorption, as well as detrimental
contributions from interference. An additional basic prerequisite is processability.
In the case of solution processed perovskite solar cells, it means that the layer has
to provide good wetting properties and withstand the aggressive polar solvents used
for perovskite precursors, in case it is processed below the photoactive layer. Solvent
orthogonality has to be met also when the selective layer is deposited on top of the
perovskite. This limits the possible formulations to the ones based on non- and weak
polar solvents, as polar solvents damage and dissolve halide perovskites.
Evolving from dye-sensitized solar cells (DSSCs), the first perovskite solar cells
were fabricated using a mesoscopic TiO2 film to harvest electrons from the pho-
toactive material [5]. Its initial purpose was to provide an extraction interface in
the bulk of the absorber to prevent losses from transport limitations. After discov-
ering the excellent ambipolar transport properties of halide perovskites, this layout
has developed over time to thinner layers, as the role of mesoscopic TiO2 shifted
towards facilitating crystallization [10]. However, it is still the most commonly used
for perovskite cells, hence it is often referred to as the standard architecture. With the
rapid development of perovskite devices, a wide range of selective charge transport
layers and device architectures have been investigated [83].
A conventional method to categorize these solar cells is differentiating them based
on the polarity of the front contact, the side on which light enters the device [84].
When the electron transport layer (ETL) is deposited on the front contact, the device
has an n-i-p structure, whereas if the polarity is inverted (holes are extracted at the
substrate side), the architecture is called p-i-n. Figure 10.7. represents the examples
of common materials and architectures used for the fabrication of perovskite solar
cells.

10.4.1 N-I-P

An archetypical n-i-p perovskite solar cell uses TiO2 as the ETL, and spiro-OMeTAD
as the hole transport layer (HTL) [8]. Notably, the highest reported power conversion
efficiencies were achieved with devices in an n-i-p architecture [85]. Initially, a high
temperature processing was required to achieve high quality TiO2 layers, but with the
advancements of the perovskite field several routes have been explored to overcome
this limitation [86].
Alternative synthesis routes for metal oxides have been proposed, with one promi-
nent method being atomic layer deposition (ALD) [87]. This allows the formation of
compact and high quality thin films, providing additional barrier properties. It was
used for depositing various materials, such as TiO2 , SnO2 , ZnO, NiO and Al2 O3 .
234 D. Forgacs et al.

Fig. 10.7 Examples of common materials and architectures used for the fabrication of perovskite
solar cells. Courtesy of Saule Technologies

Another approach is to use metal foils such as titanium as substrates, that enable
higher temperature fabrication. In this case, finding a viable transparent top elec-
trode is imperative [88]. Besides, module fabrication poses another challenge, as
a solution has to be found to form the series connections without interrupting the
conductive substrate. Different organic materials have been explored as well, both
as stand-alone selective layers and in conjunction with metal oxides to improve their
performance [83].
The most commonly used HTM in n-i-p perovskite solar cells is spiro-OMeTAD
[89]. The deposition of this material has been very well optimized over the years to
yield high efficiency devices, however there are a couple of shortcomings that has
to be overcome. First of all, the fabrication of this material requires a complex syn-
thesis, leading to a high price. Besides, there are concerns regarding the instability
of solar cells based on spiro-OMeTAD induced by doping agents—a typical exam-
ple being a highly hygroscopic Li-TSFI [90]. These additives are used to improve
conductivity. Their detrimental effect point towards the necessity to develop novel
hole transporters. This is an important research direction, and a plethora of novel
materials have been proposed and reported already [89, 91, 92]. Many of them offer
attractive properties with price, stability and efficiency outperforming the established
spiro-OMeTAD.

10.4.2 P-I-N

The traditional p-i-n layout—also referred to as the inverted one—originates from

OPV [93]. Utilization of organic semiconductors as transport layers provides the
inherent advantage of low processing temperatures. Various PEDOT:PSS-type mate-
rials have been reported as viable HTMs, however usually a lower Voc value is
10 Perovskite Photovoltaics: From Laboratory to Industry 235

observed in devices based on them [94]. This could be associated with interfacial
defects leading to SRH recombination. One explanation is a reaction between the
intrinsically p-doped PEDOT:PSS and the perovskite, that could reduce the work
function of the transport layer. Alternative organic HTMs, as well as p-type metal
oxides (such as NiOx , Cu2 O, etc.) were utilized successfully to replace PEDOT,
leading to improved performance and longevity [92].
The most commonly used ETL materials in a p-i-n architecture are fullerene
derivatives, with PCBM being one prominent example [95]. To improve its perfor-
mance, alternative functionalization methods have been reported, that enhance its
solubility and/or promote a uniform arrangement on the surface of the active layer,
leading to a more favourable morphology. Alternative organic materials and metal
oxides have also been investigated [83].
As a general conclusion, metal oxides are an attractive group of materials as charge
selective layers, due to their robustness and moisture barrier properties combined
with a low cost. With appropriate interfacial engineering and processing, they are an
interesting class of materials for transport layers in perovskite solar cells. Organic
materials on the other hand offer tailoring of energy levels and simple low temperature
processing routes.

10.4.3 Tandem Solar Cells

The band gap of MAPbI3 is measured to be around 1.55 eV in a thin film, which
is close to 1.34 eV—a value yielding the theoretical maximum efficiency (33.7%)
for a single junction solar cell, as derived by Shockley and Queisser (SQ) [96].
This limitation can be overcome by the construction of tandem architectures, where
several absorbers are connected electrically, reducing spectral losses from carrier
thermalization. In this case, the performance will be dependent on the band gap
energy of the complementary absorbers [97].
The preparation of a double junction solar cell incorporating commercially avail-
able mature technologies, such as crystalline silicon (c-Si) or CIGS solar cells (with
band gaps between 1.1–1.3 eV), requires a complementary absorber with a band gap
between 1.7–1.9 eV [98]. While this approach can improve device efficiency, it also
makes the fabrication more complex, which is a crucial subject to consider in the
perspective of a viable commercialization (Fig. 10.8).
Due to the above mentioned reasons, halide perovskite solar cells are particularly
interesting for tandem applications, as their band gap can be engineered by an appro-
priate choice of the compound stoichiometry, and they also offer low cost fabrication
methods combined with high performance. Their exceptional tunability makes them
promising candidates to couple with c-Si or CIGS solar cells, and the realization of
an efficient perovskite/perovskite tandem holds an attractive commercial potential
[99–101].
Several works reported solar cells utilizing perovskite absorbers with an opti-
mized band-gap for tandem applications [101]. A commonly observed challenge of
236 D. Forgacs et al.

Fig. 10.8 The

Shockley-Queisser efficiency
limit as the function of the
band gap. The current
efficiency records of
different technologies are
highlighted. Courtesy of
Saule Technologies

such systems is the segregation of halide species under illumination, forming recom-
bination centres that limit performance [102]. This effect is especially pronounced in
mixed I− /Br− compositions that exhibit a band gap in the above mentioned optimal
range.
It was reported, that when alloying lead and tin based perovskites, the band gap
changes in an anomalous way, reaching lower levels than either of the pure materi-
als [103]. This change is accompanied with an improved material stability, making
these compositions interesting for single junction and perovskite/perovskite tandem
applications.
Tandem architectures can be categorized based on the arrangement between the
sub-cells [104]. Two-terminal monolithic (2T), and four-terminal (4T) devices have
been the most extensively investigated in case of perovskites. 4T tandem solar cells
have so far yielded higher performances, however it requires 2 complete devices
with front and back electrodes to be prepared. The monolithic connection is the most
challenging and rewarding layout, as it enables a simple series electronic connection,
and requires only 2 electrode layers. There are however several challenges that have
to be met in order to fabricate highly efficient 2T solar cells. These include process-
ability, current matching between the sub-cells, an efficient charge recombination
layer between the sub-cells, and optical management.
The combination of these demands have so far limited the performance of
perovskite-based monolithic tandem cells to be below that of the single junction
record of the respective sub-cells [104]. Nevertheless, there is a rapid development,
and it is expected that perovskite tandems will soon outperform their single junction
counterparts (Fig. 10.9).
10 Perovskite Photovoltaics: From Laboratory to Industry 237

Fig. 10.9 The most studied perovskite tandem cell architectures (left) 2-terminal (right) 4-terminal.
Courtesy of Saule Technologies

10.5 Device Operation

10.5.1 Charge Generation and Transport

Once an incoming photon interacts with an electron near the VBM, charge carriers
are generated [105]. An electron will be excited to a higher energy state, leaving
behind a hole with a positive charge. As these are carriers with an opposite charge,
an electrostatic Coulomb interaction will bind them together, creating a quasiparticle
called an exciton [106]. In order to generate electricity, these charge carriers have
to be separated and transported to their respective electrodes. Let’s take a look, how
perovskites perform at these tasks!
Materials with a high dielectric constant (like Si or GaAs) allow the formation of
Wannier-Mott excitons. These excitons have binding energies on the order of 10 meV
[107]. It is comparable with the thermal energy at room temperature, which amounts
to 25.7 meV. Consequently, the generation of free charge carriers is facilitated as
excitons easily dissociate, which is beneficial for high performance solar cells. There
are several reports examining the dielectric constant and the exciton binding energies
of halide perovskites, in particular MAPbI3 [108]. The findings are that perovskites
exhibit a high dielectric constant, which is increased further by orders of magnitude
upon illumination. The origin of this behaviour is under scrutiny, with one possible
explanation it being correlated with mobile ions in the lattice. Nevertheless, as the
238 D. Forgacs et al.

high dielectric constant implies, the exciton binding energy in this material is low.
Values between 2 and 55 meV have been reported using different approaches, with
a direct measurement method that showed it to amount to 16 meV [107, 109]. The
preferred generation of free charge carriers is supported by the lack of excitonic peaks
in the absorption spectrum of MAPbI3 at room temperatures. Temperature-dependent
PL measurements have highlighted that the excitonic peak is present in MAPbI3 at
low temperatures, below 100 K [16]. As the temperature increases, the excitonic
peak diminishes and completely disappears at 160 K. At the same time, a new,
higher energy peak appears, indicating the radiative band-to-band recombination of
free charge carriers (Fig. 10.10).
Once the charge carriers are generated, they have to be transported through the
photoactive material, to reach their respective electrodes [84]. This happens via the
combination of two effects: drift—driven by an electric field, and diffusion—driven
by a gradient in concentration. A good contact will extract charges at a rate that is
higher than the rate of diffusion in the bulk of the absorber, and thus a concentration
gradient will be created. However, when the extraction is slow, charge will build up
at the vicinity of the interface retarding the extraction of carriers even further.

Fig. 10.10 Charge generation, transport, and recombination in perovskite solar cells. (1) A photon
is absorbed, electron-hole pairs are generated. (2) Radiative recombination – photons are emitted.
(3) Shallow trap states. (4) Deep trap states leading to non-radiative recombination. Courtesy of
Saule Technologies
10 Perovskite Photovoltaics: From Laboratory to Industry 239

If the carriers recombine before being extracted, the energy of the absorbed pho-
ton will be lost, leading to a decreased power output [24]. Hence, it is important to
understand what the diffusion length of the charge carriers is in a given material, as
it will determine the maximal layer thickness that can be used for efficient transport
properties. The examination of these characteristics were first carried out in 2013
[10]. Using transient absorption and time-resolved photoluminescence decay mea-
surements, Stranks et al. found diffusion lengths that can exceed one micron in a
perovskite film. These high values were surprising, but it corroborated well with the
good performance of solar cells without a mesoporous TiO2 which was previously
thought to be necessary for efficient charge extraction. The observation of long PL-
lifetimes, diffusion lengths and ambipolar transport led to the realization that there
is no explicit need for this and perovskite solar cells can be efficient in a conven-
tional planar thin-film structure [110]. This is in contrast with organic photovoltaics,
where a bulk heterojunction is needed to provide a quick charge separation to prevent
recombination. These findings provided a major breakthrough in understanding the
basic principles of photophysical processes occurring in such systems.

10.5.2 Loss Mechanisms

Charge carriers that are not extracted will recombine. It is an inevitable process, and
it is desirable for all the recombination to be of radiative type, like it would take place
in an ideal solar cell material [111]. This way, under 1-sun illumination, it can enable
a build-up of high internal photon densities resulting in higher operating voltages.
Non-radiative recombination present in the material will act as a loss mechanism,
restraining the maximum efficiency to lower values than that of the SQ limit. At the
SQ limit, the external quantum efficiency (EQE) of the electroluminescence (EL)
is 1, which means that there are no non-radiative losses. The EL can be recorded
by operating the solar cell as an LED, applying forward bias that exceeds the Voc
value. It is an interesting tool for getting information on non-radiative losses, and
optimizing record-breaking cells.
Non-radiative recombination typically originates from the presence of trap states
[112]. A charge carrier can fall into this state, and subsequently recombine with a car-
rier of opposing charge. These trap states are formed due to defects in the crystalline
material, and they can be categorized based on their depth. Deeper are more detri-
mental, providing an easier path for recombination. As such, it is crucial to analyse
the defects in absorber materials, and their contribution to the loss mechanisms.
Crystallographic defects natively present in semiconductors are the product of
the crystal growth, and their appearance cannot be completely avoided. The defect
density is given as their number per cubic centimetre of the material. Semiconductor
materials utilized for solar cell applications exhibit low to moderate densities, with
values reported being sc-Si: 108 cm−3 , poly-Si: 1013 –1014 cm−3 , CIGS: 1013 cm−3 ,
CdTe: 1013 –1015 cm−3 [113]. MAPbI3 single crystals have also been examined, and
low values of around 1011 cm−3 have been found. Surface and interfacial defects are
240 D. Forgacs et al.

considered to be the source of most non-radiative recombination present in perovskite

solar cells.
In a polycrystalline film, the homogeneity of the semiconductor is interrupted at
the grain boundary, and these areas are naturally prone to be more defective [114].
As such, the role of grain boundaries in perovskite films had been under intense
investigation. The reported defect densities for polycrystalline MAPbI3 films are
several orders of magnitude larger than that of single crystals, with relatively high
values of 1015 –1016 cm−3 [113]. While various works correlated the grain boundaries
with the high density of defects and the primary source of non-radiative losses, other
findings demonstrated them with a benign nature. It is generally accepted that they
serve as a platform where moisture-induced degradation can initiate, and that larger
grain sizes are generally beneficial for device performance and stability.
Despite the relatively high level of defect density present, solar cells based on
perovskite thin films show excellent efficiencies with Voc values that are quite high
compared to their band gap energy, and deliver long carrier diffusion lengths [115].
These seemingly contradictory findings point towards an unusual defect tolerance.
Various studies examined the origin of this phenomenon. [23, 24, 116] The findings
indicate that the most likely defects with low formation energies result in shallow
traps that are less detrimental to device performance. This behaviour can be explained
by the characteristics of the valence band constituted by antibonding orbitals, which
probably leads to states in the band and not the band gap in the presence of a broken
bond [24]. Besides, dangling bonds typically present in covalent semiconductors will
not be formed in a strongly ionic lattice, as the additional charge will be screened due
to the high dielectric constant of the perovskite lattice, leading to shallow traps. It was
also found that the less abundant iodine interstitials actually lead to deep level electron
and hole traps in MAPbI3 [117]. Nonetheless, due to the distinctive amphoteric
nature of iodine, the electron traps are kinetically deactivated. This leaves only hole
traps as harmful defects, and they can be passivated via a redox transformation.
Interstitial iodine was also shown to have the lowest migration activation energy of
all ionic species present in a perovskite material [118]. As such, it is an important
research direction to find a robust strategy for the passivation of defects [115]. Various
materials and post treatment methods were explored to achieve this, with promising
improvements on the Voc and stability of the solar cells [35, 119].

10.5.3 Device Measurement and Hysteresis

Solar cells are evaluated and characterized with a current-voltage measurement. This
is a standardized method, that consists of illuminating a solar cell with the AM 1.5G
spectrum with an intensity of 1000 W/m2 , and recording current values as a function
of applied voltage. Three parameters are extracted from the measurement: the short-
circuit current (Jsc ), the open-circuit voltage (Voc ) and the fill factor (FF). These
together are used to determine the solar to power conversion efficiency of a device. A
voltage sweep is applied in two directions; forward and reverse direction. This method
10 Perovskite Photovoltaics: From Laboratory to Industry 241

Fig. 10.11 Current-voltage characteristics of the same perovskite solar cell recorded under. a
Various scanning speeds and b illumination intensities. Reprinted from [143]

is supposed to represent the steady state power output of a solar cell in operation
under the given bias. The shape of the recorded curve and the corresponding figures
of merit should be independent of the sweep direction, speed or biasing history—
excluding minor deviations. It is however not the case when measuring perovskite
solar cells.
Since the early days of perovskite solar cell research, the I-V characterization of
the devices carried ambiguity. It has been observed that the current-voltage curves of
perovskite solar cells display a loop, and this phenomenon is referred to as hysteresis
[120] (Fig. 10.11).
The presence of this characteristic has been reported in many works, and while its
extent have been different, it is pertained regardless of the deposition method of the
perovskite as well as the interfacial layers utilized. It has been also noted, that these
effects change based on the biasing history of the solar cell, as well as the speed at
which the current-voltage sweep has been carried out [121]. This effect will create
uncertainty in the evaluation of the photovoltaic characteristics of devices and the
comparison of the performance of different processes and materials utilized for the
fabrication of the solar cells.
Subsequently, many works addressed this phenomenon, reaching for different
explanations, such as ferroelectricity, charge carrier trapping, or interfacial effects
[122]. There is a rate-independent, slow response, and a rate-dependent, fast response.
Currently there is a fairly broad consensus on the origin of hysteretic effects as
majority of research findings suggest that migration of ionic species present in the
perovskite material play a primary role [118]. In the presence of an electric field,
mobile ions accumulate in the vicinity of the electrodes. This will build up a space
charge adjacent to the collection layers, which in turn will have an effect on charge
transport, extraction and recombination. It results in an enhancement or inhibition
of carrier collection based on the pre-biasing history, and consequently, a hysteretic
current-voltage response (Fig. 10.12).
242 D. Forgacs et al.

Fig. 10.12 The effect of light soaking on the I-V characteristics of perovskite solar cells. Reprinted
from ref [121]

Besides the interaction with the electric field, mobile ionic species in perovskites
were shown to interact with electromagnetic radiation and heat as well [123]. Light-
soaking changes the photovoltage of solar cells, and corresponding PL-measurements
have highlighted an increased lifetime and intensity under illumination.
The problems that the presence of mobile ions generate regarding the estimation
of the actual performance of the solar cell can be overcome by slow I-V sweeps
and maximum power point (MPP) tracking, which are now recommended by the
perovskite community as measurement standards. However, there are concerns that
it raises regarding the internal stability of perovskites, which are one of the major
topics that are under scrutiny today [124].

10.6 Scale up

10.6.1 Commercialization Bottlenecks

Initially thought of as a sensitizer in DSSCs, perovskites were soon recognized as

new thin-film absorbers [83]. The pioneering works initiated an extensive research
on hybrid halide perovskites as a stand-alone PV technology. In the following years,
various methods were explored to deposit high quality perovskite materials [66,
73, 76, 81]. The focus was to understand how the chemistry of diverse precursors is
affected by a multitude of processing techniques and conditions, and how it influences
the optoelectronic properties of the deposited perovskite thin film.
After several years of development, perovskite PV technology has grown into
a viable solution with tangible prospects in the near future [125]. The next step
is to develop appropriate production methodologies, and scrutinize the two major
challenges regarding the material: its stability, and its environmental impact.
10 Perovskite Photovoltaics: From Laboratory to Industry 243

The origin of instability in perovskites can be categorized as either external or

internal, based on the source that introduces it.
The intrinsic instability is related to the low heat of formation of perovskite from
its precursors, PbI2 and MAI in the case of MAPbI3 [41]. It results in weak thermo-
dynamic stabilization, and the material can easily decompose to its constituents at
elevated temperatures. Overcoming these limitations can be achieved by the usage
of alternative, thermodynamically more stable perovskite materials. The inclusion of
FA and Cs ions were shown to improve the thermal stability, as well as an alloying
with lower dimensionality perovskites [37].
The sensitivity towards environmental agents refers to decomposition that is
induced by moisture, oxygen and UV-light [52]. The most straightforward way
to overcome this issue is impervious encapsulation of solar cells, to prevent them
from being exposed to harmful effects. Existing methods for alternative thin-film
photovoltaic technologies as well as OLEDs might be utilized as a guidance here
[126]. Besides, engineering the perovskite material to be intrinsically more endurant
towards these effects will also be beneficial.
The degradation of interfaces arises from photoelectrochemical interactions
between the perovskite and the transport materials and electrodes [127]. There is
a well-known concern regarding the archetypical spiro-OMeTAD. Besides its high
price, this material was shown to deteriorate quickly, especially in the presence of
its common dopants that are used to oxidize spiro-OMeTAD and in turn increase
its conductivity. It was also observed that metal ions from the electrode, and iodide
ions from perovskite can migrate through organic transport layers severely affecting
device performance. The usage of metal oxides and carbon-based electrodes were
shown to reduce these effects [128]. The development of new, robust interfacial layers
will also contribute to overcome this issue.
The migration of ionic species under illumination or via an electric field was
shown to alter the output power of the solar cell [112]. These changes have shown
to be reversible after a few cycles, however, concerns are present whether this has
long-term implications, and intrinsic limitations on the lifetime of perovskite solar
cells.
Besides the stability of the material, researchers are generally concerned about the
lead content of the most efficient perovskite absorbers [129]. Being a heavy metal,
there are obvious questions that arise regarding the environmental and health-related
impacts of a large-scale production that involves lead. As such, it will be necessary
to carry out thorough examinations of these topics, to understand and minimize the
risks associated with them. One solution is to increase device stability and combine it
with proper, fail-proof encapsulation. From a legislative point of view, there might be
also a few limitations to overcome in order to enter the market, as there are precise
indicatives on what levels of heavy metal contents are allowed in a commercial
product.
An alternative approach is the development of lead-free perovskites [130]. How-
ever, due to the previously described electronic structure of these materials, the p-p
transition in lead halide perovskites plays a crucial role in the excellent absorbing
244 D. Forgacs et al.

properties of the material. As such, it is challenging to find an alternative formulation

that allows the formation of a perovskite material with the same excellent properties.
Typically considered alternatives are tin-based perovskites [131]. MASnI3 was
shown to have a narrower band gap than MAPbI3 , which is closer to the optimum
for single junction solar cells. However, due to the amphoteric nature of tin, there are
difficulties that arise regarding the processing and the stability of these absorbers, as
Sn2+ can easily oxidize to Sn4+ which does not form a perovskite structure [132]. The
amphoteric nature of tin also contributes to the toxicological effect of perovskites
based on it, as reported by Babaygit et al. [129]. When exposed to water, tin drastically
lowers its pH value, and eventually it leads to the demise of living organisms at
lower concentrations than lead. Nevertheless, perovskite solar cells based on tin are
investigated, but their efficiencies are much below the state-of-the art devices based
on lead.
Alternative materials, such as germanium- bismuth- or titanium-based perovskites
have also been investigated, however solar cells based on them have so far always
displayed a poor performance [130, 133, 134]. As perovskites cover a huge amount of
formulations, computational methods might provide a viable, quick and efficient way
to discover new photoactive materials. This has been demonstrated by Zhao et al.,
who identified 64 possible new formulations of double perovskites [135]. In these
compositions, a mono- and a trivalent metal cation replaces two divalent ones. Due
to the large number of possible compositions, the development of double perovskite
materials might be a successful way to create efficient and lead-free perovskite solar
cells [130, 134, 136].

10.6.2 Target Markets

A successful market entrance of perovskite solar cells will largely depend on how all
the performance parameters will be related to their price. A commonly used figure of
merit for energy generating assets is the levelized cost of electricity (LCOE) [137].
This provides an assessment of the minimal cost at which the energy has to be sold
to reach break-even over the entire lifetime of the generating asset. It is calculated by
dividing the total average cost of operating electricity source by the total generated
energy. Cai et al. have performed the first detailed LCOE analysis on perovskite solar
cells, taking realistic scenarios into account [138] (Fig. 10.13).
This work pointed out, that the LCOE values that perovskite solar cells can reach
are much below all other currently available energy sources, provided that an effi-
ciency of 12% can be reached with a lifetime of at least 15 years. These values were
estimated for glass-sealed n-i-p modules deposited on a glass/FTO substrate. Several
important conclusions were drawn from the performed analysis. First, the cost of the
photoactive and transport layers, and the solvents used to deposit them contribute
only a small amount to the final cost. More than half of the total module costs are
constituted by the transparent conductive oxide. According to their calculations, the
10 Perovskite Photovoltaics: From Laboratory to Industry 245

Fig. 10.13 Representation of an LCOE analysis of different energy. Sources [138]. Reprinted with
permission from ref [135]

module efficiency plays one of the most important role in determination of the final
price.
While these calculations are pointing towards the great potential that lies in the
commercial entry of perovskites, there are few trends that should be taken into
account. The price of silicon solar cells have been falling rapidly in the last decade,
and continues to do so [139]. There is a learning curve which correlates the increased
amount of cumulative production with the price of the product. However, since
2011 the costs fall under the value predicted by this curve. It has already led to
the bankruptcy of many solar manufacturing companies, and it does not create a
favourable landscape for a new technology to enter the market. As such, it is impera-
tive for entrepreneurs that are planning to bring a new technology into the market to
realize the implications of the above. As a first principle, targeting applications where
crystalline silicon is not viable will be probably the easiest direction for stand-alone
perovskite solar cells.
Considering the above mentioned limiting factors and the advantageous properties
of perovskites, there are several target markets that could be accessed by such solar
cells [137]. Building integrated photovoltaics is one segment, where perovskites
are forecast to “shine”. This is mainly due to their high efficiency and excellent
absorption properties, which enables them to generate power efficiently, even from
indirect or scattered light. As such, perovskite solar modules could be implemented
onto facades, enlarging the area available for photovoltaic energy generation and
contributing to the development of zero-energy buildings. The favourable low-light
performance also enables indoor applications, such as the powering smart home
systems or various low-power electronics. There could be alternative applications,
for instance charging portable electronic devices and electric vehicles.
An alternative approach is to create tandem structures, and doing so on silicon
that already dominates the market seems to be an attractive option [99]. Capitalising
246 D. Forgacs et al.

on the existing large-scale value chain of the dominant PV technology, this approach
could enable a rapid development. While this can improve the performance of sili-
con solar cells, and potentially further reduce the cost of electricity, it also sets high
requirements for perovskites. For this to be viable, the development of a high per-
formance, wide bandgap perovskite is needed, that allows current matching of the
sub-cells [104]. The material should also exhibit a stability comparable with silicon
solar cells, and reaching this point might substantially postpone the realization of a
viable product.
Due to their high efficiency and defect tolerance, perovskites are also promising
candidates for space applications [138]. For the greatest impact, this requires the
development of a light-weight substrate and encapsulation, that will maximize their
power-to-weight ratio. It was shown already that perovskites can outperform the state
of the art GaAs in this figure of merit as well [139].
The ultimate goal is to live on a planet fuelled by clean, safe and reliable energy.
This should be the clear guidance to all organizations that work on perovskites, pro-
viding the paradigm for interinstitutional collaborations, and the design and devel-
opment of future products. It is our responsibility to carefully evaluate all the risks
associated with perovskites, and to find the way to eliminate any potentially harm-
ful aspects. Maximal effort should be made with the aim of collaborating in the
development of a technology that is beneficial for all humankind.

10.7 Advancements by Saule Technologies

Saule Technologies was established in 2014, shortly after the first breakthrough
findings about perovskites were published. The sole purpose of the company is the
commercialization of photovoltaic cells based on this new technology. Saule has
pioneered the application of ink-jet printing for the fabrication of flexible, light-
weight, and efficient perovskite solar modules.
As spin-coating—the most commonly used deposition method is not suitable for
industrial size scaling, an alternative approach has to be utilized. Ink-jet printing has
been chosen as a production method of perovskite solar cells, as it offers several
favourable properties. In this approach, the ink-jet printer nozzle is applying the ink
exactly at a pre-defined location with high resolution. Using this technique enables
the creation of custom patterns, as the shapes and areas covered by each layer can be
all tailored according to the requirements. This feature will allow architects to utilize
BIPV with a great degree of design freedom.
The processing is carried out in ambient conditions with a negligible waste of
materials. This is facilitated by the drop on demand approach, which is accomplished
by driving the nozzles with a piezoelectric material. When the ink-jet head reaches
the targeted location, a voltage signal is used to alter the shape of the transducer,
forcing the drops out of the nozzles. Figure 10.14 (left) represents the operation of
a nozzle. An industrial-scale ink-jet head has several hundreds of them, enabling a
rapid and high-resolution printing at a low cost.
10 Perovskite Photovoltaics: From Laboratory to Industry 247

Utilizing ink-jet printing for large scale production requires a number of param-
eters to be engineered. For example, the ink used for printing has to fulfil viscosity
and surface tension requirements set by the printhead, to enable a stable jetting.
Besides, the solvent system has to provide the necessary solubility for the compo-
nent to be deposited, as well as solvent orthogonality with the underlying layer. A
common strategy to overcome these issues is to utilize solvent mixtures and additives
to meet prerequisites for printability. The wetting properties of the ink also have to
be optimized. If it is “too good”, the deposited droplets will spread rapidly, and it
will prevent a high resolution printing. Poor wetting on the other hand will lead to
non-uniform coverage. Various surface treatment methods can be utilized to optimize
the interaction between the ink and the surface onto which it is deposited.
Saule Technologies was the first company to present a working demonstrator in
2016. The device was a module that contained 24 solar cells with an active area of
1 cm2 , capable of charging a phone under intense illumination. During 2017, Saule
has increased the scale of this technology from 1 cm2 size solar cells deposited on a
laboratory scale into an automated method that covers the size of an A4 paper (210 ×
297 mm). In the same year, the company has also developed a robust encapsulation
method compatible with perovskite solar cells, which prevents moisture and oxygen
from penetrating into the solar cell. This technique has been shown to enable the
undisturbed underwater operation of such modules, highlighting the viability of the
approach.
The developments described above highlight that by 2019–10 years after their first
demonstration—perovskite solar cells have matured to a technology that is close to
its market entrance. It is the fastest developing field of photovoltaics to date, and
while there are still some open questions about the peculiarities of the photoactive
material, we at Saule are sure about its eventual success. We are striving to utilize
the potential of this technology to its utmost extent, with the aim of contributing to

Fig. 10.14 (left) The schematic of an ink-jet printing nozzle, depositing materials on a substrate.
(right) The crystallization of perovskite as seen by the fiducial camera. Courtesy of Saule Tech-
nologies
248 D. Forgacs et al.

Fig. 10.15 The evolution of demonstrators by Saule technologies: (2015) An ink-jet printed per-
ovskite solar cell. (2016) A perovskite solar module made of single cells externally connected in
series. (2017, top) Ink-jet printed flexible perovskite solar module (2017, bottom) An A4 paper-
sized, ink-jet printed perovskite solar module, capable of charging portable electronic devices.
Courtesy of Saule Technologies

a global change leading to clean, safe, sustainable and low-cost energy production
(Fig. 10.15).
Currently, Saule is collaborating with Skanska to develop offices with a BIPV
solution, where semi-transparent and opaque perovskite solar modules will be uti-
lized on the windows and facades of office buildings respectively. The possibilities in
using perovskites for space applications are also being explored in collaboration with
JAXA, the Japanese space agency. The high specific power of light-weight perovskite
modules provides the technology with great potential, and the initial tests regarding
their tolerance to radiation present in such conditions supports this research direction
(Fig. 10.16).
By the end of 2018, the technology development has enabled Saule to create two
major demonstrators, initiating the real-life test of printed perovskite photovoltaics.
Skanska has installed a façade element into their Spark office situated in Warsaw,
Poland consisting of 54 ink-jet printed perovskite solar modules. Henn-na Hotel in
Nagasaki, Japan has erected a curved glass structure containing 72 perovskite mod-
ules. These modules are used to power the illumination of the logos of the respective
companies, and the performance of the solar cells are monitored. The greatest impor-
tance of these demonstrators is to highlight that the technology has matured to the
point that enables ambient processing of hundreds of functional perovskite mod-
ules in a semi-automated manner, and prospects good views towards the upcoming
commercial entrance of such devices (Fig. 10.17).
10 Perovskite Photovoltaics: From Laboratory to Industry 249

Fig. 10.16 (left) Perovskite solar module with a hexagonal pattern (middle) A perovskite solar
module operating underwater (right) An autonomous, battery-free Bluetooth beacon operating in
indoor lighting conditions. Courtesy of Saule Technologies

Fig. 10.17 (left) Curved glass façade element in Nagasaki, Japan with 72 printed perovskite mod-
ules (right) Visualization of a BIPV application: perovskite modules integrated into windows of a
skyscraper. Courtesy of Saule Technologies
250 D. Forgacs et al.

References

1. Q. Chen et al., Under the spotlight: The organic–inorganic hybrid halide perovskite for opto-
electronic applications. Nano Today 10, 355–396 (2015)
2. D.B. Mitzi, C.A. Feild, Z. Schlesinger, R.B. Laibowitz, Transport, optical, and magnetic
properties of the conducting halide perovskite CH3NH3SnI3. J. Solid State Chem. 114, 159–
163 (1995)
3. S. Wang, D.B. Mitzi, C.A. Feild, A. Guloys, Synthesis and characterization of [NH2C(I) =
NH2]3MI5 (M = Sn, Pb): stereochemical activity in divalent tin and lead halides containing
single (110) perovskite sheets. J. Am. Chem. Soc. 117, 5297–5302 (1995)
4. D.B. Mitzi, Solution-processed inorganic semiconductors. J. Mater. Chem. 14, 2355 (2004)
5. A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka, Organometal halide perovskites as visible-light
sensitizers for photovoltaic cells. J. Am. Chem. Soc. 131, 6050–6051 (2009)
6. J.-H. Im, C.-R. Lee, J.-W. Lee, S.-W. Park, N.-G. Park, 6.5% Efficient perovskite quantum-
dot-sensitized solar cell. Nanoscale 3, 4088 (2011)
7. I. Chung, B. Lee, J. He, R.P.H. Chang, M.G. Kanatzidis, All-solid-state dye-sensitized solar
cells with high efficiency. Nature 485, 486–489 (2012)
8. H.-S. Kim et al., Lead iodide perovskite sensitized all-solid-state submicron thin film meso-
scopic solar cell with efficiency exceeding 9%. Sci. Rep. 2, 591 (2012)
9. M.M. Lee, J. Teuscher, T. Miyasaka, T.N. Murakami, H.J. Snaith, Efficient hybrid solar cells
based on meso-superstructured organometal halide perovskites. Science (80-.). 338, 643–647
(2012)
10. S.D. Stranks, et al., Electron-hole diffusion lengths exceeding 1 micrometer in an organometal
trihalide perovskite absorber. Science (80-.). 342, 341–344 (2013)
11. L.M. Fraas, Low-Cost Solar Electric Power. Low-Cost Solar Electric Power 9783319075,
(Springer International Publishing, 2014)
12. S.D. Stranks, H.J. Snaith, Metal-halide perovskites for photovoltaic and light-emitting devices.
Nat. Nanotechnol. 10, 391–402 (2015)
13. N.-G. Park, M. Grätzel, T. Miyasaka, K. Zhu, K. Emery, Towards stable and commercially
available perovskite solar cells. Nat. Energy 1, 16152 (2016)
14. V.M. Goldschmidt, Die Gesetze der Krystallochemie. Naturwissenschaften 14, 477–485
(1926)
15. W. Travis, E.N.K. Glover, H. Bronstein, D.O. Scanlon, R.G. Palgrave, On the application of
the tolerance factor to inorganic and hybrid halide perovskites: a revised system. Chem. Sci.
7, 4548–4556 (2016)
16. K. Wu et al., Temperature-dependent excitonic photoluminescence of hybrid organometal
halide perovskite films. Phys. Chem. Chem. Phys. 16, 22476–22481 (2014)
17. M.A. Green et al., Solar cell efficiency tables (version 51). Prog. Photovolt. Res. Appl. 26,
3–12 (2018)
18. U.K. Mishra, J. Singh, Semiconductor Device Physics and Design 28–91 (Springer Nether-
lands, 2008). https://doi.org/10.1007/978-1-4020-6481-4_2
19. A.S. Lin, et al., High mechanical strength thin HIT solar cells with graphene back contact.
IEEE Photonics J. 9, (2017)
20. K. Yoshikawa, et al., Silicon heterojunction solar cell with interdigitated back contacts for a
photoconversion efficiency over 26%. Nat. Energy 2, (2017)
21. W.-J. Yin, J.-H. Yang, J. Kang, Y. Yan, S.-H. Wei, Halide perovskite materials for solar cells:
a theoretical review. J. Mater. Chem. A 00, 1–17 (2015)
22. P. Umari, E. Mosconi, F. De Angelis, Relativistic GW calculations on CH3NH3PbI3 and
CH3NH3SnI3 perovskites for solar cell applications. Sci. Rep. 4, 4467 (2015)
23. M. Kot, K. Wojciechowski, H. Snaith, D. Schmeißer, Evidence of nitrogen contribution to the
electronic structure of the CH3NH3PbI3 perovskite. Chem. A Eur. J. 24, 3539–3544 (2018)
24. W. Tress, Perovskite solar cells on the way to their radiative efficiency limit—insights into
a success story of high open-circuit voltage and low recombination. Adv. Energy Mater. 7,
1602358 (2017)
10 Perovskite Photovoltaics: From Laboratory to Industry 251

25. P. Gao, M. Grätzel, M.K. Nazeeruddin, Organohalide lead perovskites for photovoltaic appli-
cations. Energy Environ. Sci. 7, 2448–2463 (2014)
26. S. A. Kulkarni, et al., Band-gap tuning of lead halide perovskites using a sequential deposition
process. J. Mater. Chem. A 2, 9221 (2014)
27. G.E. Eperon et al., Formamidinium lead trihalide: a broadly tunable perovskite for efficient
planar heterojunction solar cells. Energy Environ. Sci. 7, 982 (2014)
28. J.H. Noh, S.H. Im, J.H. Heo, T.N. Mandal, S.I Seok, Chemical management for colorful,
efficient, and stable inorganic-organic hybrid nanostructured solar cells. Nano Lett. 13, 1764–
1769 (2013)
29. A.R. Denton, N.W. Ashcroft, Vegard’s law. Phys. Rev. A 43, 3161–3164 (1991)
30. D.J. Slotcavage, H.I. Karunadasa, M.D. McGehee, Light-induced phase segregation in halide-
perovskite absorbers. ACS Energy Lett. 1, 1199–1205 (2016)
31. K. Yan et al., Hybrid halide perovskite solar cell precursors: colloidal chemistry and coor-
dination engineering behind device processing for high efficiency. J. Am. Chem. Soc. 137,
4460–4468 (2015)
32. T.M. Koh et al., Formamidinium-containing metal-halide: an alternative material for near-IR
absorption perovskite solar cells. J. Phys. Chem. C 118, 16458–16462 (2014)
33. N.J. Jeon et al., Compositional engineering of perovskite materials for high-performance solar
cells. Nature 517, 476–480 (2015)
34. M. Saliba, et al., Incorporation of rubidium cations into perovskite solar cells improves pho-
tovoltaic performance. Science (80-.) 354, 206–209 (2016)
35. M. Abdi-Jalebi et al., Maximizing and stabilizing luminescence from halide perovskites with
potassium passivation. Nature 555, 497–501 (2018)
36. G.E. Eperon et al., Inorganic caesium lead iodide perovskite solar cells. J. Mater. Chem. A 3,
19688–19695 (2015)
37. M. Saliba et al., Cesium-containing triple cation perovskite solar cells: improved stability,
reproducibility and high efficiency. Energy Environ. Sci. 9, 1989–1997 (2016)
38. P. Gratia et al., The many faces of mixed ion perovskites: unraveling and understanding the
crystallization process. ACS Energy Lett. 2, 2686–2693 (2017)
39. J. Liang, et al., CsPb0. 9Sn0. 1IBr2 based all-inorganic perovskite solar cells with exceptional
efficiency and stability. J. Am. Chem. Soc. 139, 14009–14012 (2017)
40. H. Chen et al., Inorganic perovskite solar cells: a rapidly growing field. Sol. RRL 2, 1700188
(2018)
41. B. Brunetti, C. Cavallo, A. Ciccioli, G. Gigli, A. Latini, on the thermal and thermodynamic
(in) stability of methylammonium lead halide perovskites. Sci. Rep. 6, (2016)
42. B. Conings, Intrinsic thermal instability of methylammonium lead trihalide perovskite. Adv.
Energy Mater. (2015)
43. C.C. Stoumpos et al., Ruddlesden-popper hybrid lead iodide perovskite 2D homologous semi-
conductors. Chem. Mater. 28, 2852–2867 (2016)
44. H. Tsai et al., High-efficiency two-dimensional ruddlesden-popper perovskite solar cells.
Nature 536, 312–316 (2016)
45. I.C. Smith, E.T. Hoke, D. Solis-Ibarra, M.D. McGehee, H.I. Karunadasa, A layered hybrid
perovskite solar-cell absorber with enhanced moisture stability. Angew. Chem. Int. Ed. 53,
11232–11235 (2014)
46. G. Grancini et al., One-year stable perovskite solar cells by 2D/3D interface engineering. Nat.
Commun. 8, 15684 (2017)
47. P.K. Nayak et al., Mechanism for rapid growth of organic–inorganic halide perovskite crystals.
Nat. Commun. 7, 13303 (2016)
48. A. Dubey et al., Strategic review on processing routes towards highly efficient perovskite
solar cells. J. Mater. Chem. A (2018). https://doi.org/10.1039/C7TA08277K
49. Z. Song et al., A technoeconomic analysis of perovskite solar module manufacturing with
low-cost materials and techniques. Energy Environ. Sci. 10, 1297–1305 (2017)
50. D.P. McMeekin, et al., Crystallization kinetics and morphology control of formamidinium–
cesium mixed-cation lead mixed-halide perovskite via tunability of the colloidal precursor
solution. Adv. Mater. 29, (2017)
252 D. Forgacs et al.

51. S.I. Seok, M. Grätzel, N.-G. Park, Methodologies toward highly efficient perovskite solar
cells. Small 1704177 (2018)
52. J.S. Manser, M.I. Saidaminov, J.A. Christians, O.M. Bakr, P.V. Kamat, Making and breaking
of lead halide perovskites. Acc. Chem. Res. 49, 330–338 (2016)
53. J.C. Hamill, J. Schwartz, Y.-L. Loo, Influence of solvent coordination on hybrid organic-
inorganic perovskite formation. ACS Energy Lett. 3, 92–97 (2018)
54. X. Cao et al., Elucidating the key role of a lewis base solvent in the formation of perovskite
films fabricated from the lewis adduct approach. ACS Appl. Mater. Interfaces. 9, 32868–32875
(2017)
55. L. Etgar et al., Mesoscopic CH3 NH3 PbI3 /TiO2 Heterojunction Solar Cells. J. Am. Chem.
Soc. 134, 17396–17399 (2012)
56. Y.-J. Jeon et al., Planar heterojunction perovskite solar cells with superior reproducibility. Sci.
Rep. 4, 6953 (2014)
57. M. Yang et al., Perovskite ink with wide processing window for scalable high-efficiency solar
cells. Nat. Energy 2, 17038 (2017)
58. D.T. Moore, et al., Crystallization kinetics of organic–inorganic trihalide perovskites and the
role of the lead anion in crystal growth. J. Am. Chem. Soc. 150209140709004 (2015). https://
doi.org/10.1021/ja512117e
59. Y. Wu et al., Retarding the crystallization of PbI 2 for highly reproducible planar-structured
perovskite solar cells via sequential deposition. Energy Environ. Sci. 7, 2934–2938 (2014)
60. C.-C. Chueh et al., The roles of alkyl halide additives in enhancing perovskite solar cell
performance. J. Mater. Chem. A 3, 9058–9062 (2015)
61. P.-W. Liang et al., Additive enhanced crystallization of solution-processed perovskite for
highly efficient planar-heterojunction solar cells. Adv. Mater. 26, 3748–3754 (2014)
62. S.D. Stranks, P.K. Nayak, W. Zhang, T. Stergiopoulos, H.J. Snaith, Formation of thin films
of organic-inorganic perovskites for high-efficiency solar cells. Angew. Chem. Int. Ed. 54,
3240–3248 (2015)
63. S. Colella et al., MAPbI3-xClx mixed halide perovskite for hybrid solar cells: the role of
chloride as dopant on the transport and structural properties. Chem. Mater. 25, 4613 (2013)
64. A.R. Pascoe et al., Directing nucleation and growth kinetics in solution-processed hybrid
perovskite thin-films. Sci. China Mater. 60, 617–628 (2017)
65. C. Bi et al., Non-wetting surface-driven high-aspect-ratio crystalline grain growth for efficient
hybrid perovskite solar cells. Nat. Commun. 6, 1–7 (2015)
66. N.J. Jeon et al., Solvent engineering for high-performance inorganic-organic hybrid perovskite
solar cells. Nat. Mater. 13, 1–7 (2014)
67. M. Xiao et al., A fast deposition-crystallization procedure for highly efficient lead iodide
perovskite thin-film solar cells. Angew. Chem. Int. Ed. 53, 9898–9903 (2014)
68. H. Zhang et al., Understanding the effect of delay time of solvent washing on the performances
of perovskite solar cells. ACS Omega 2, 7666–7671 (2017)
69. T. Jesper Jacobsson, et al., Exploration of the compositional space for mixed lead halogen
perovskites for high efficiency solar cells. Energy Environ. Sci. 9, 1706–1724 (2016)
70. J.P. Correa-Baena, et al., Unbroken perovskite: interplay of morphology, electro-optical prop-
erties, and ionic movement. Adv. Mater. 5031–5037 (2016). https://doi.org/10.1002/adma.
201600624
71. K. Liang, D.B. Mitzi, M.T. Prikas, Synthesis and characterization of organic − inorganic
perovskite thin films prepared using a versatile two-step dipping technique. Chem. Mater. 10,
403–411 (1998)
72. J. Burschka et al., Sequential deposition as a route to high-performance perovskite-sensitized
solar cells. Nature 499, 316–319 (2013)
73. H. Chen, Two-step sequential deposition of organometal halide perovskite for photovoltaic
application. Adv. Funct. Mater. 27, 1605654 (2017)
74. X. Cao et al., Control of the morphology of PbI 2 films for efficient perovskite solar cells by
strong Lewis base additives. J. Mater. Chem. C 5, 7458–7464 (2017)
10 Perovskite Photovoltaics: From Laboratory to Industry 253

75. W.S. Yang, et al., High-performance photovoltaic perovskite layers fabricated through
intramolecular exchange. Science (80-.) 348, 1234–1237 (2015)
76. M. Liu, M.B. Johnston, H.J. Snaith, Efficient planar heterojunction perovskite solar cells by
vapour deposition. Nature 501, 395–398 (2013)
77. O. Malinkiewicz et al., Perovskite solar cells employing organic charge-transport layers. Nat.
Photonics 8, 128–132 (2014)
78. C. Momblona et al., Efficient vacuum deposited p-i-n and n-i-p perovskite solar cells employ-
ing doped charge transport layers. Energy Environ. Sci. 9, 3456–3463 (2016)
79. Q. Chen et al., Planar heterojunction perovskite solar cells via vapor-assisted solution process.
J. Am. Chem. Soc. 136, 622–625 (2014)
80. Y.H. Chiang, H.M. Cheng, M.H. Li, T.F. Guo, P. Chen, Low-pressure vapor-assisted solution
process for thiocyanate-based pseudohalide perovskite solar cells. Chemsuschem 9, 2620–
2627 (2016)
81. H. Snaith, Perovskites: the emergence of a new era for low-cost, high-efficiency solar cells.
J. Phys. Chem. Lett. 4, 3623–3630 (2013)
82. U. Wurfel, A. Cuevas, P. Wurfel, Charge carrier separation in solar cells. IEEE J. Photovolt.
5, 461–469 (2015)
83. Z. Guo, L. Gao, C. Zhang, Z. Xu, T. Ma, Low-temperature processed non-TiO 2 electron
selective layers for perovskite solar cells. J. Mater. Chem. A 6, 4572–4589 (2018)
84. Z. Song, S.C. Watthage, A.B. Phillips, M.J. Heben, Pathways toward high-performance per-
ovskite solar cells: review of recent advances in organo-metal halide perovskites for photo-
voltaic applications. J. Photonics Energy 6, 22001 (2016)
85. W.S. Yang, et al., Iodide management in formamidinium-lead-halide–based perovskite layers
for efficient solar cells. Science (80-.) 356, (2017)
86. J.M. Ball, M.M. Lee, A. Hey, H.J. Snaith, Low-temperature processed meso-superstructured
to thin-film perovskite solar cells. Energy Environ. Sci. 6, 1739 (2013)
87. R.W. Johnson, A. Hultqvist, S.F. Bent, A brief review of atomic layer deposition: from fun-
damentals to applications. Mater. Today 17, 236–246 (2014)
88. Y. Xiao, G. Han, H. Zhou, J. Wu, An efficient titanium foil based perovskite solar cell: using
a titanium dioxide nanowire array anode and transparent poly(3,4-ethylenedioxythiophene)
electrode. RSC Adv. 6, 2778–2784 (2016)
89. L. Calió, S. Kazim, M. Grätzel, S. Ahmad, Hole-transport materials for perovskite solar cells.
Angew. Chem. Int. Ed. 55, 14522–14545 (2016)
90. I. Lee, J.H. Yun, H.J. Son, T.S. Kim, Accelerated degradation due to weakened adhesion from
Li-TFSI additives in perovskite solar cells. ACS Appl. Mater. Interfaces. 9, 7029–7035 (2017)
91. M. Saliba et al., A molecularly engineered hole-transporting material for efficient perovskite
solar cells. Nat. Energy 1, 15017 (2016)
92. Chen, J. & Park, N.-G. Inorganic Hole Transporting Materials for Stable and High Efficiency
Perovskite Solar Cells. J. Phys. Chem. C acs.jpcc.8b01177 (2018). https://doi.org/10.1021/
acs.jpcc.8b01177
93. J.-Y. Jeng et al., CH3 NH3 PbI3 perovskite/fullerene planar-heterojunction hybrid solar cells.
Adv. Mater. 25, 3727–3732 (2013)
94. J.C. Yu et al., Highly efficient and stable inverted perovskite solar cell employing PEDOT:GO
composite layer as a hole transport layer. Sci. Rep. 8, 1070 (2018)
95. Y. Fang, C. Bi, D. Wang, J. Huang, The functions of fullerenes in hybrid perovskite solar
cells. ACS Energy Lett. 2, 782–794 (2017)
96. K.A. Bush, et al., 23.6%-efficient monolithic perovskite/silicon tandem solar cells with
improved stability. Nat. Energy 2, 17009 (2017)
97. A. Guchhait, et al., Over 20% efficient CIGS–perovskite tandem solar cells. ACS Energy Lett.
807–812 (2017). https://doi.org/10.1021/acsenergylett.7b00187
98. De A. Vos, Detailed balance limit of the efficiency of tandem solar cells. J. Phys. D. Appl.
Phys. 13, 839–846 (1980)
99. G.E. Eperon, et al. Perovskite-perovskite tandem photovoltaics with optimized band gaps.
Science (80-.). 354, 861–865 (2016)
254 D. Forgacs et al.

100. F. Hao, C.C. Stoumpos, R.P.H. Chang, M.G. Kanatzidis, Anomalous band gap behavior in
mixed Sn and Pb perovskites enables broadening of absorption spectrum in solar cells. J. Am.
Chem. Soc. 136, 8094–8099 (2014)
101. G.E. Eperon, M.T. Hörantner, H.J. Snaith, Metal halide perovskite tandem and multiple-
junction photovoltaics. Nat. Rev. Chem. 1, 0095 (2017)
102. E.T. Hoke et al., Reversible photo-induced trap formation in mixed-halide hybrid perovskites
for photovoltaics. Chem. Sci. 6, 613–617 (2015)
103. C.S. Ponseca et al., Organometal halide perovskite solar cell materials rationalized: ultrafast
charge generation, high and microsecond-long balanced mobilities, and slow recombination.
J. Am. Chem. Soc. 136, 5189–5192 (2014)
104. G.D. Scholes, G. Rumbles, Excitons in nanoscale systems. Nat. Mater. 5, 683–696 (2006)
105. V. D’Innocenzo et al., Excitons versus free charges in organo-lead tri-halide perovskites. Nat.
Commun. 5, 3586 (2014)
106. E.J. Juarez-Perez et al., Photoinduced giant dielectric constant in lead halide perovskite solar
cells. J. Phys. Chem. Lett. 5, 2390–2394 (2014)
107. Q. Lin, A. Armin, R.C.R. Nagiri, P.L. Burn, P. Meredith, Electro-optics of perovskite solar
cells. Nat. Photonics 9, 106–112 (2015)
108. G. Xing, et al., Long-range balanced electron-and hole-transport lengths in organic-inorganic
CH3NH3PbI3. Science (80-.). 342, 344–347 (2013)
109. O.D. Miller, E. Yablonovitch, S.R. Kurtz, Intense internal and external fluorescence as solar
cell approach the SQ efficiency limit. Photovolt. IEEE J. 2, 1–27 (2012)
110. T.S. Sherkar et al., Recombination in perovskite solar cells: significance of grain boundaries,
interface traps, and defect ions. ACS Energy Lett. 2, 1214–1222 (2017)
111. D. Meggiolaro et al., Iodine chemistry determines the defect tolerance of lead-halide per-
ovskites. Energy Environ. Sci. 11, 702–713 (2018)
112. J.S. Yun et al., Critical role of grain boundaries for ion migration in formamidinium and
methylammonium lead halide perovskite solar cells. Adv. Energy Mater. 6, 1–8 (2016)
113. R. Brenes et al., Metal halide perovskite polycrystalline films exhibiting properties of single
crystals. Joule 1, 155–167 (2017)
114. W.J. Yin, T. Shi, Y. Yan, Unusual defect physics in CH3NH3PbI3perovskite solar cell absorber.
Appl. Phys. Lett. 104, (2014)
115. D. Meggiolaro, et al., Iodine chemistry determines the defect tolerance of lead-halide per-
ovskites. Energy Environ. Sci. 0–30 (2018). https://doi.org/10.1039/c8ee00124c
116. C. Li et al., Iodine migration and its effect on hysteresis in perovskite solar cells. Adv. Mater.
28, 2446–2454 (2016)
117. T. Niu et al., Stable high-performance perovskite solar cells via grain boundary passivation.
Adv. Mater. 30, 1706576 (2018)
118. H.J. Snaith et al., Anomalous hysteresis in perovskite solar cells. J. Phys. Chem. Lett. 5,
1511–1515 (2014)
119. W. Tress, et al., Understanding the rate-dependent J–V hysteresis, slow time component, and
aging in CH3 NH3 PbI3 perovskite solar cells: the role of a compensated electric field. Energy
Environ. Sci. 995–1004 (2015). https://doi.org/10.1039/c4ee03664f
120. W. Tress, Metal halide perovskites as mixed electronic-ionic conductors: challenges and
opportunities—from hysteresis to memristivity. J. Phys. Chem. Lett. 8, 3106–3114 (2017)
121. E.L. Unger et al., Hysteresis and transient behavior in current–voltage measurements of
hybrid-perovskite absorber solar cells. Energy Environ. Sci. 7, 3690–3698 (2014)
122. K. Domanski, E.A. Alharbi, A. Hagfeldt, M. Grätzel, W. Tress, Systematic investigation of
the impact of operation conditions on the degradation behaviour of perovskite solar cells. Nat.
Energy 3, 61–67 (2018)
123. J.-P. Correa-Baena et al., The rapid evolution of highly efficient perovskite solar cells. Energy
Environ. Sci. 10, 710–727 (2017)
124. J.-S. Park, H. Chae, H.K. Chung, S.I. Lee, Thin film encapsulation for flexible AM-OLED: a
review. Semicond. Sci. Technol. 26, 034001 (2011)
10 Perovskite Photovoltaics: From Laboratory to Industry 255

125. A. Guerrero et al., Interfacial degradation of planar lead halide perovskite solar cells. ACS
Nano 10, 218–224 (2016)
126. E. Bi et al., Diffusion engineering of ions and charge carriers for stable efficient perovskite
solar cells. Nat. Commun. 8, 15330 (2017)
127. A. Babayigit et al., Assessing the toxicity of Pb-and Sn-based perovskite solar cells in model
organism Danio rerio. Sci. Rep. 6, 1–11 (2016)
128. A. Abate, Perovskite solar cells go lead free. Joule 1, 659–664 (2017)
129. M. Konstantakou, T. Stergiopoulos, A critical review on tin halide perovskite solar cells. J.
Mater. Chem. A 5, 11518–11549 (2017)
130. S.J. Lee et al., Fabrication of efficient formamidinium tin iodide perovskite solar cells through
SnF2 -pyrazine complex. J. Am. Chem. Soc. 138, 3974–3977 (2016)
131. S. Nagane et al., Lead-free perovskite semiconductors based on germanium-tin solid solutions:
structural and optoelectronic properties. J. Phys. Chem. C 122, 5940–5947 (2018)
132. C. Wu et al., The dawn of lead-free perovskite solar cell: highly stable double perovskite
Cs2 AgBiBr6 film. Adv. Sci. 5, 1700759 (2018)
133. X. Zhao et al., Design of lead-free inorganic halide perovskites for solar cells via cation-
transmutation. J. Am. Chem. Soc. 139, 2630–2638 (2017)
134. G. Volonakis et al., Cs2 InAgCl6 : a new lead-free halide double perovskite with direct band
gap. J. Phys. Chem. Lett. 8, 772–778 (2017)
135. M. Cai, et al., Cost-performance analysis of perovskite solar modules. Adv. Sci. 4, (2017)
136. F. Taghizadeh-Hesary, N. Yoshino, Y. Inagaki, Empirical analysis of factors influencing price
of solar modules (2018)
137. He, X. Perovskite Photovoltaics 2018–2028. (2018)
138. I. Cardinaletti et al., Organic and perovskite solar cells for space applications. Sol. Energy
Mater. Sol. Cells 182, 121–127 (2018)
139. M. Kaltenbrunner et al., Flexible high power-per-weight perovskite solar cells with chromium
oxide-metal contacts for improved stability in air. Nat. Mater. 14, 1032–1039 (2015)
140. S. De Wolf et al., Organometallic halide perovskites: sharp optical absorption edge and its
relation to photovoltaic performance. J. Phys. Chem. Lett. 5, 1035–1039 (2014)
141. S. Pang et al., NH2 CH NH2 PbI3 : an alternative organolead iodide perovskite sensitizer for
mesoscopic solar cells. Chem. Mater. 26, 1485–1491 (2014)
142. J.-H. Im, H.-S. Kim, N.-G. Park, Morphology-photovoltaic property correlation in perovskite
solar cells: one-step versus two-step deposition of CH3 NH3 PbI3 . APL Mater. 2, 081510
(2014)
143. B. Chen, M. Yang, S. Priya, K. Zhu, Origin of J-V hysteresis in perovskite solar cells. J. Phys.
Chem. Lett. 905–917 (2016)
144. K. Wojciechowski, D.Forgács, T. Rivera, Industrial Opportunities and Challenges for Per-
ovskite Photovoltaic Technology. 1900144, 1–9 (2019)
Chapter 11
Augsburger Tubular Photovoltaic
(ATPV)

V. Petrova-Koch and J. Mayer

Abstract Nowadays, available PV modules are flat. The PV presented in this chapter
is a glass tube with a flexible PV cell inside. The tubular diversification of the PV
was initially inspired by the start-up company Solyndra in Silicon Valley CA, which
manufactured CIGS cell directly on the surface of an inner glass tube. Unfortunately,
Solyndra went bankrupt due to high manufacturing costs. In 2011, a leaner tubular
PV was invented at Gate-East in Garching, which enabled substantial simplification
compared to the Solyndra’s ideas and design. It was named Augsburger Tubular
Photovoltaics (ATPV) and patented with Osram, later Ledvance. It has numerous
advantages as follows: the cell manufacturing is roll-to-roll technology, the absorp-
tion films can be thinner at the same efficiency, because the tubular shape extends the
light path, compared to flat cells and modules. Simplified installation and operation:
low weight; semitransparent for rain, wind, and snow; better cooling compared to a
flat panel; leaner and cost-effective substructure. Agricultural application: protection
against hail but permeability for sun and rain—homogenous plant growth as gen-
eral: reduced evaporation, short-period shade and therefore, uniform growing; good
protection at high temperatures and from destructive hail—dual as well as multiple
use is possible. The concept of ATPV was successfully proven, and a small-scale
manufacturing was built at the Osram, later Ledvance site in Augsburg.

Dedicated to Prof. Fred Koch (TU Munich).

V. Petrova-Koch (B)
Gate-East, Schleissheimer Str. 17, Garching 85748, Germany
e-mail: [email protected]
J. Mayer
Ledvance GmbH, Equipment Engineering, Berliner Allee 65, Augsburg 86153, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 257

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_11
258 V. Petrova-Koch and J. Mayer

11.1 Introduction

The sun is the strongest energy source known to the humankind on one side, the
growing population uses more and more electricity on the other. The PV, if affordable,
is capable to supply with solar electricity to everybody on the planet, clean and
environmentally friendly.
The PV has the potential to become one of the biggest industries of the twenty-first
century. It is already bigger than the microelectronics, or the lighting industry. The
recent PV reached in a relatively short time, the remarkable cumulative volume of
300 GW in 2017/1/). However this is only the beginning. The PV market share is
expected to grow to approximately 400 GW in 2020 and to approximately 700 GW
in 2030/2/.
Cost reduction is one of the main goals of the recent PV industry. There are two
major ways to achieve lower cost: (i) via economy of scale in Gigawatt-sized fabs;
or (ii) via next generation lean manufacturing at a smaller scale. Vertical integration
along the value chain could further reduce the costs.
Many of the known big PV companies are upscaling the volume of their already
established PV manufacturing to Giga-scale at the moment, in order to further lower
the costs. Increasing the efficiency of the cells and of the modules at the same time,
approaching the theoretical limits, became a must in the last years. Almost all single-
junction PV cells manufactured on a big scale reached values around 20%.
Newcomers and game changer are focused on next generation PV, and on inno-
vative and lean manufacturing/14–17/. New PV concepts promise to diversify the
recent PV market. There is enough room for it. Affordable, competitive, and innova-
tive PV modules, produced in substantially smaller fabs, are on the way to establish
their market position and shares.
The Augsburger Tubular Photovoltaic (ATPV), which we are presenting in this
chapter, belongs to the second category. The ATPV targets a lean PV manufacturing
and is a shaped PV.
Today, 99% of the PV modules are flat, no matter if they are c-Si wafer- or
thin-film based. In both cases, they resemble big-size roof tiles and wall panels.
More than a decade ago, there was a start-up company Solyndra/3/in Silicon
Valley CA, which put a lot of effort to shape TF PV and to bring to the market its
patented tubular PV cells and modules/4/. Solyndra was founded in 2005, unveiled its
greenfield manufacturing side in Silicon Valley Ca in 2008, and reached in a relatively
short time, a manufacturing volume higher than 100 MWp/year. Unfortunately, it
went bankrupt in 2011, together with many other PV start-up companies. Obviously,
the cost of the tubular cells and modules were too high.
In 2011, a leaner tubular PV was invented by V. Petrova-Koch at Gate-East in
Garching, which enabled substantial simplification compared to the Solyndra’s ideas
and design. It was named Augsburger Tubular Photovoltaics (ATPV). A patent was
applied together with Osram in 2014/5/. The new concept of this PV was successfully
proven, and a small-scale manufacturing was built and explored more recently for
ATPV module-production at small-scale at the Ledvance site in Augsburg.
11 Augsburger Tubular Photovoltaic (ATPV) 259

Fig. 11.1 The Osram manufacturing site in Augsburg, known since 1961 for high scale manufac-
turing of world-class fluorescent tubes. Since 2016, it belongs to Ledvance

Osram was known since 1961 for its top-quality fluorescing tube manufacturing
on big scale. This site in Augsburg, owned since 2016 by Ledvance (Fig. 11.1) was
founded more than 100 years ago. The classical lighting business lost drastically
its market shares worldwide in the last years. Alternative ideas, technologies, and
solutions are welcome to prolong the life of excellent manufacturing sites like the
one in Augsburg.
The ATPV concept, the cells, the modules, and several innovative installations
and applications will be presented in this chapter. We compare the ATPV with the
Solyndra’s product at one side, and with the standard flat TF PV- and c-Si PV modules
on the other.

11.2 Shaped PV Versus Flat PV

11.2.1 Advantages of the Cylindrical Tubular PV Design

Putting PV cells inside a cylindrical glass tube is not popular. Almost all Thin-Film
PV (TFPV) developers and manufacturers are also focused on cells that lay flat on
a panel. The TFPV modules have similar appearance and design as the crystalline
silicon solar panels, commonly seen today.
The cylindrical shape has numerous advantages:
260 V. Petrova-Koch and J. Mayer

Fig. 11.2 Self-tracking the sun with cylindrically shaped horizontally installed PV power tubes

– It allows the PV cells to face the sun for longer periods of time in a day than a flat
solar cell (Fig. 11.2). Hence, cylindrical power tubes can produce more electricity
over the day. No trackers are needed. The self-tracking saves costs.
– The PV tubes can be installed horizontally, they don’t need any inclination. This
brings additional simplifications on system level, provides new appearance of the
PV installations, and enables new applications.
– The absorbing films can be thinner when cylindrically shaped than the absorbing
film used to manufacture flat PV cells with the same efficiency because the light
path gets prolonged when the cells are curved. This means that there is a reduction
of the material consumption, and also of the cell deposition and processing time,
which at the end results in cost reduction.
– The cylindrical shape provides a certain concentration of solar radiation on the
way to the cell surface (Fig. 11.3). The degree of concentration depends on the
distance between the outer glass tube surface and the absorption layer top surface.
– The PV power tubes cool better than a flat power module, because of the dis-
cretization of the absorber area. Also, the air moves more efficiently around the
power tubes and cools them more efficiently than a PV flat module.
– The wind, rain, and snow clean the tubes better than the flat glass panels.
11 Augsburger Tubular Photovoltaic (ATPV) 261

Fig. 11.3 Concentration

(small degree) of the solar
radiation on the way to the
surface of a cylindrically
shaped solar cell

11.2.2 Comparison Between an Array Cylindrical Glass

Tubes and a Pair of Flat Glass Plates (Without PV
Cells)

The top of almost all PV modules today is a soda-lime glass plate, solar grade.
Plastic is used to wrap only flexible and very light PV modules, which make use of
roll-to-roll TFPV cells.
Our cylindrically shaped tubular PV is making use of soda-lime glass, which is
cost-effective. It is hard to imagine that plastic tubes could compete with soda-lime
glass tubes, and be the choice to manufacture cylindrically shaped PV. Soda-lime
glass tubes are an ultimate choice for outdoor applications if the preferred shape of
the PV is tubular.
The glass tubing industry is very advanced, and it is cost-competitive to the flat
glass manufacturing industry.
The diameters of the available cylindrical soda-lime glass tubes spread in a very
broad range (from few mm in diameter to several hundreds of mm). Their wall
thickness can be varied between several hundred µm and several mm.
The diameter the tubes we chose for ATPV, and also the diameter of the Solyndra’s
tubes is approximately one inch, (25 mm). These tubes are known from the classical
lighting industry as T8 glass tubes and are very popular. This diameter is considerably
smaller than the diameter of the tubes usually used to manufacture evacuated solar
thermal tubular collectors, which is typically approximately 2 inch/6/.
The wall thickness of the Solyndra’s T8 tubes was ca. 1.8 mm. In our case, it is
around 1–1.2 mm.
The length of the tubes we and Solyndra chose is approximately one meter. We
prefer this length because the width of the standard flat c-Si PV modules is also
approximately one meter.
The weight of the ATPV tubes is about 200 grams/tube, versus ca. 300 g/tube in
the case of Solyndra.
262 V. Petrova-Koch and J. Mayer

While Solyndra was buying tubes from companies specialized on glass tubbing,
our strategy is to manufacture the glass tubes, the power tubes, and the ATPV modules
vertically integrated at the manufacturing side of Ledvance in Augsburg.
The ATPV glass tubes need a hardening of 90–120 N/m2 . They have to satisfy
also the required hail test for outdoor applications. We managed to toughen the tubes
thermally, and in-line. This reduces the cost as compared to the thermal toughening
of glass plates, which has to be performed after cutting the flat glass and c-shaping
the glass edges. The edge processing and external thermal toughening elevate the
cost by at least 30%/6.7/.
An array of thermally toughened glass tubes is cost-competitive to a sandwich of
two flat glass plates with the same area, and in many cases, even less expensive. The
comparison here concerns the glass only, without the PV cells.
The flat PV modules are usually covered on the top side with solar glass (typically
with a thickness of 2, 3.2, or 4 mm). Their backside is glass for the case of all TFPV.
Glass/glass modules also became popular for the c-Si modules more recently/8/. The
thickness of the top and the bottom glass is usually the same. The weight of one square
meter of flat glass sandwich varies in the range 10–20 kg/m2 . 2 mm glass could be
toughened ESG recently/8/, and used more widely for manufacturing glass/glass PV
modules..
One square meter area can be completely covered by 40 T8 tightly ordered T8
glass tubes (25 mm in diameter). Their weight is around 8 kg in our case, which is
substantially lower than the weight per m2 of the flat glass sandwich. This means a
significant process energy and material savings, cost reduction, and more efficient
transportation of the ATPV modules.
Many PV manufacturers believe nowadays, that using flat soda-lime glass plates
is less expensive than to wrap PV cells in soda-lime glass tubes. It could be proven
that tubular design is competitive and can even have lower costs.
The tubular lighting industry has more than half a century of successful history.
A typical glass furnace enables today the manufacturing of about 1000 km glass
tubes per day (with a diameter of approx. 1 inch). The highly automated lines used
to assemble fluorescing tubes reach values up to 7000 tubes/h!
The high degree of automation and the in-line toughening of the glass tubes allows
manufacturing a square meter of T8 tubes cost-competitive than a sandwich of two
thermally toughened soda-lime glass plates of the same area. The main advantages of
the tubes are they don’t have edges, don´t need cutting, and because of it, they could
be toughened thermally in-line! Tubes with walls as thin as 1 mm enable toughening
comparable to one of flat solar glass plates with of a 2 mm thickness or more.
A cylindrical glass body could be vacuum sealed (Fig. 11.4). Its inside stays free
of humidity this way for a very long time (for many decades). The sealing is of higher
quality if we use the tubes fresh and don’t wash them. Two glass plates in contrast
can’t be vacuum sealed so far. Numerous attempts have failed.
A vertically integrated highly automated production of Augsburger Tubular PV
(ATPV) on the side of the glass tubbing fab is our target, where we start with sand and
glass tube and end with cost-competitive ATPV Modules. High-efficient TFPV cells
11 Augsburger Tubular Photovoltaic (ATPV) 263

Fig. 11.4 Vacuum-sealed cylindrical glass tube

and stripes needed for the ATPV will face less degradation if the cell and module
manufacturing is not separated at two sides.
The advantages of a vertical integrated tubular device manufacturing have been
proven over decades by the fluorescing tube industry.
The thin-film PV manufacturing today is also vertically integrated into the 99%
of all known fabs.
Our goal is to inherit the existing glass tubing experience and enable a second
life of the glass fab and glass sealing production lines. This way, we will be able to
reduce the cost.

11.2.3 Cell Designs and Semiconductor Materials, Suitable

to Serve Tubular PV

The wafer-based c-Si PV cells are cost-effective and widely used today, but because
of their rigidity, they can’t serve a tubular PV in a simple manner.
The tubular PV is restricted to thin-film PV cells, which can be shaped much
easier. A transparent counter electrode deposited on the top side of the absorbing
layer (Fig. 11.5) is the preferred design for the cylindrically shaped cell.
The semiconductor CdTe, which is known for its low -ost TFPV manufacturing,
is not in our focus. The widely spread architecture of the CdTe PV cells is inverted,
their top electrode is not transparent; Also the issues around manufacturing and using
PV cells containing Cd and their environmentally friendly recycling are open.
Several other semiconductor materials enable shaped architecture and would allow
manufacturing of high-efficient shapeable cells and PV power tubes. The most mature
and efficient is CIGS/9–12/, followed by perovskite TFPV/13–15/. Perovskite PV
264 V. Petrova-Koch and J. Mayer

Fig. 11.5 Preferred cell

design to serve a
cylindrically shaped PV

became a hype in the last several years because of its rapidly improving efficiency
at very low cost of manufacturing at the same time. Unfortunately, the perovskites
with the highest efficiency obtained so far contain lead (Pb).
Numerous companies are manufacturing CIGS TFPV on a large scale//nowadays.
Solar Frontier produces in Japan more than 1 GW/y flat CIGS PV Modules; Avancis
and CNBM manufactures in China ca. 300 MW/y this year and plan on upscaling
to more than one GW/y in the next 2 years; Manz, Shanghai Electric Group, and
Shenhua Group in China will produce 250 MW this year, Hanergy, which owns
Solibro’s production side in Germany for rigid CIGS on glass, MiaSole’s in CA,
and Global Solar’s in AZ, for flexible CIGS PV on stainless steel, manufactures a
volume of more than 250 MW/y total; Solo Power in Oregon reached 250 MW/y for
flexible CIGS PV on stainless steel, SIVA Power in CA ramping up a unique 200 MW
CIGS novel highly productive CIGS manufacturing on glass, Ascent Solar in CO,
US, manufactures flexible CIGS Modules on Steel at low manufacturing volume;
Flisom in CH manufacturing 15 MW/y Flexible monolithically integrated CIGS PV
on Polyimid foils and is going to build a much bigger plant in a very near future. All
these companies are aiming to prove that CIGS TFPV is a mature PV technology,
which is competitive to c-Si PV. There is a belief even that CIGS is “the next Big
thing in PVs” because this TFPV reached efficiency in the lab higher than 20%, low
manufacturing cost, and better environmental food print compared to the c-Si PV.
The c-Si wafer-based PV still dominates more than 80% of the PV market at the
moment. A serious competition race is taking place in the last years, and TFPV is
gaining more and more shares.
A remarkable recent achievement in the field of TFPV and particularly for the
CIGS TFPV is the laser processing and monomythical interconnection of the cells,
including the flexible once/10/. Laser Processing improves the reproducibility, and
the productivity of PV manufacturing, which saves costs. Stringing, and wiring,
which is unavoidable in the case of c-Si wafer PV technology is not needed in the
case of TFPV in general.
CIGS, which is a direct gap semiconductor, needs a thickness of 1.5–2 µm in
order to totally absorb the solar radiation, which is roughly 100 times less than the
thickness of a c-Si wafer-based Pv cell, which is in the range of 150–200 µm. c-Si
11 Augsburger Tubular Photovoltaic (ATPV) 265

is a semiconductor with non-direct gap. Because of this, Si absorbs visible light one
to two orders of magnitude that is less efficient, a thick absorber is needed.
A CIGS could either be deposited on curved surfaces (for example on so-called
inner tubes, as Solyndra did) or via roll-to-roll on thin foils (our choice in the case
of ATPV).

11.2.3.1 Solyndra’S CIGS PV Cells Deposition and Laser Processing

on an Inner Glass Tube

Solyndra made use of CIGS deposition on permanently rotating soda-lime glass inner
tubes (Fig. 11.6). The typical diameter of the inner tubes was 15–16 mm (T5 tubes).
The company has developed a unique way to deposit and structure efficient Copper
Indium Gallium Selenium (CIGS) cells on these tubes/10, 11/. The manufacturing
equipment was specially designed and constructed for Solyndra. This probably was
one of the serious obstacles to reduce cell manufacturing costs.

11.2.3.2 Roll-to-Roll Flexible CIGS Cells and Stripes for the ATPV

The Augsburger Tubular PV makes use of flexible CIGS PV cells (Fig. 11.7), man-
ufactured by Roll-to-Roll (R2R) Technology.
The R2R CIGS technology has the following advantages:
1. The R2R deposition of semiconductor films is an established technology, which
promises shortening of the deposition time and a substantial cost reduction. R2R
is known from the textbooks as an example for lean and fast manufacturing.
2. High-efficient CIGS PV cells can be deposited on flexible foils. Several research
labs like NRLEL in CO, US, ZSW in Stuttgart, and EMPA already demonstrated
this. At ETH Zurich is holding the world record of 20.2% on Polyimid//.
3. A number PV start-up companies like MiaSole, Global Solar SoloPower, Ascent
Solar in the US together with Hanergy Holding Group; Flisom in Switzerland,
and Solar Frontier in Japan are manufacturing flexible CIGS PV cells and mod-
ules on industrial scale. All these companies except Flisom are depositing the
cells on stainless steel foils. Monolithically interconnection of the cells is not

Fig. 11.6 Solyndra laser

scribed monolithically
interconnected CIGS cells,
deposited on an inner glass
tube
266 V. Petrova-Koch and J. Mayer

Fig. 11.7 CIGS R2R cells

and stripes for ATPV on
Polyimid or Steel foil

demonstrated so far on stainless steel foils. On Polyimid, Flisom has demon-

strated recently good laser processing, and monolithically interconnection of the
CIGS cells. For ATPV we focus at the moment mainly on CIGS cells, R2R
deposited on Polyimid foils, because of the possibility to vertically interconnect
the cells.

11.2.3.3 Other Cells Capable to Serve the ATPV

We expect that other cells will mature and serve the ATPV in the near future. These
are
High-efficient CIGS cells on steel foils, monolithically interconnected. Such cells
need a high-temperature insulating layer between the steel and the CIGS absorbing
layer.
Flexible Perovskite PV Cells. Perovskite PV became a hype in the field of PV
for two main reasons: the cells reached extraordinary high efficiency (around 20%
in the last few years), while their manufacturing remains very easy and the costs are
low as compared to the once of c-Si and CIGS PV cells. Recently, efficient flexible
perovskite PV modules (ink-jet printed on PET) have been demonstrated (see Chap.
10 for more details).
There are two disadvantages of perovskite; they degrade easily when facing
humidity and UV light, and the wanted high efficiency of about 20% is observed
only with perovskite cells containing Pb.
If the OPV can reach efficiencies higher than 12–13% and the manufacturing
technology remains low cost, then these cells will be a very promising candidate for
ATPV too. At the moment, the cell efficiencies remain much lower than for example
in the case of Perovskite cells. However, OPV has two major advantages: easy and
11 Augsburger Tubular Photovoltaic (ATPV) 267

low-cost manufacturing is possible in the case of printed OPV, the composition of

the OPV layers is not hazardous and poisoning; and the OPV is easy to recycle.

11.3 Power Tubes

11.3.1 Solyndra Power Tubes, the Forerunner of the Tubular

PV

Solyndra caught a lot of attention in the solar world because it managed not only to
deposit and process Copper Indium Gallium Selenium (CIGS) cells with an efficiency
of about 13% on a glass tube but also to encase it in a second glass tube. Each Solyndra
power tube is made up of two tubes. Silicone oil is added as an “optical coupling
agent”. The sunlight shines through the outer tube and concentrates on the way to
the inner tube (Fig. 11.8a). After inserting the inner tube into the outer tube, each
cylinder is hermetically sealed with glass and metal to keep out the moisture, which
otherwise can seriously erode the CIGS’ performance (Fig. 11.8b). Solyndra’s power
tubes need white back reflection foil to make use of the downside of the power tube
(Fig. 11.8c).
With this technology, Solyndra reached efficiencies of the cells up to 12.5% in
average, and 14% in the best case, enabling power tubes with a power of ca. 5 W each,
which was used to assemble the 40 tube-modules, described later in this chapter.

11.3.2 Koch Power Tubes

A flexible CIGS stripe containing monolithically interconnected cells is placed and

glued on the upper half of the Koch power tube, which is facing the sun (Fig. 11.9).
If bifacial power tubes of this type are needed, they could be realized optionally. No
white back refection foil is needed. The Koch Power Tube doesn’t need an inner
tube.
The silicon oil used as liquid optical coupling agent by “Solyndra” is replaced
in the case of the Koch Power Tube with a solid encapsulation foil. It provides the
needed optical matching of the cells to the glass tube, and also mechanical stability
in the case that a power tube breaks. We cover the tube-inside completely with
encapsulation foil and glue it, which helps to hold the glass shards in an emergency
case (Fig. 11.10).
The Koch power tubes are vacuum sealed. We are making use of technology
explored for a long time and proven by the lighting industry. The vacuum-sealed
power tubes are filled at the end with a defined atmosphere (N2 or dry ear, slight
overpressured), enabling a better cooling of the PV cells. The vacuum sealing of the
glass body helps to protect the cells from humidity for decades. The long-lasting
268 V. Petrova-Koch and J. Mayer

Fig. 11.8 a The way of light to the inner tube. b Solyndra power tube. c Solyndra power tube
system

extremely dry conditions inside of the tubes are proven in the case incandescent
lamps and in case of fluorescing lighting.
11 Augsburger Tubular Photovoltaic (ATPV) 269

Fig. 11.9 General view of a typical Koch power tube

Fig. 11.10 The solid encapsulation helps to optically match the CIGS cells to the glass tubes, and
also to hold the shards if some break

11.4 Tubular Modules

11.4.1 ATPV Versus Solyndra Modules

11.4.1.1 Similarities

The tubular power modules, no matter if Solyndra or ATPV, give a new meaning of
rooftop and field PV installations and bring several unique characteristics:
– Each power cylinder is placed apart (at least one inch) and is connected at each
end with wiring. The harnesses lie parallel to the cylinders. This design is intended
to keep broken cylinders from affecting the performance of those that still work.
– Unlike many solar-power installations with flat PV modules, which have to be
tilted to better capture sunlight, and tightly anchored, the tubular modules don’t
270 V. Petrova-Koch and J. Mayer

need to be tilted and also don’t need to be anchored on the roof with ballasts and
adhesives.
– The cylindrical design and spacing between the power tubes make the tubular
modules semitransparent: for light (Fig. 11.11a); for wind (Fig. 11.11b); for rain
(Fig. 11.11c); and for snow (Fig. 11.11d).
The major advantages for the solar conversion to electricity resulting from the
semitransparency of the modules for solar radiation, wind, rain, and snow of the
modules could be summarized as follows:
– The photons get scrambled and become better absorbed in thin films due to its
oblique incidence.
– The wind comes through the modules and cools the tubes and the cells better than
this in the case of the flat modules. The harnessing of the sun radiation improves.
We measured a better cooling compared to the standard flat modules by 5 to 8 C.
In addition to it, the CIGS TFPV has a substantially lower temperature coefficient
of 0.3%/C, compared to the ca. 0.5%/C for the c-Si flat modules//, and generates
more current than c-Si-based module at equal temperature.
– The rain washes the T8 tubes better, and the wind penetration helps too so that
much better self-cleaning effects take place.
– The snow sticks much less to the tubes in winter. More electricity can be generated
over the year by the tubular PV modules.

11.4.1.2 Major Advantages of ATPV Compared to Solyndra Modules

The major advantages of the ATPV compared to the Solyndra modules are the fol-
lowing:
– cell material saving: Only the tube side directed to the sun is covered with PV
cells, which saves roughly 50% on cell material (Fig. 11.9).
– no back reflection foils needed: In contrast, Solyndra’s bifacial power tubes, which
are fit into aluminum-framed modules, have to be placed on a white mounting
device a foot off the roof, and the clearing allows the CIGS tubes to capture
sunlight reflected from the rooftop.
– the weight of the ATPV modules is roughly one-third of the weight of the Solyndra
module because no inner tube and also no silicone oil optical agent is used, plus
the outer tube is significantly thinner in the ATPV case.
– dual use of the encapsulation foil: The optical coupling agent used in the ATPV
case is solid, enables but an optical coupling as valuable as the silicone oil in the
case of Solyndra, plus it provides safety if a glass tube brakes. The shards are not
spilled, they stick to the encapsulation foil.
– extremely dry conditions in the tubes were enabled, thanks to the vacuum sealing:
The thin-film cells and their transparent conductive electrodes (usually made out
of ZnO, which has low cost but is hygroscopic) remain protected against humidity
during the entire life of the modules.
11 Augsburger Tubular Photovoltaic (ATPV) 271

Fig. 11.11 a, b, c, d ATPV, semitransparent for solar radiation wind, rain, and snow
272 V. Petrova-Koch and J. Mayer

11.5 ATPV Rooftop Installation Versus Rooftop

Installation with Flat PV Modules

Because of the spacing between the tubes, there is no lift when the wind blows
through. “One can set the modules on the roof like a clothes horse”. The mounting
systems are very simple (Fig. 11.11) compared to the one used to install standard flat
modules.
Because of the substantially lower wind resistance and the self-tracking effect,
unlike many conventional solar-power installations, the tubular modules system
doesn’t need to be tilted and doesn’t need to be anchored on the roof with bal-
lasts and adhesives. The lightweight modules, particularly in the case of ATPV, are
very suitable for flat roofs including light industrial and private house roofs.
Experienced system integrators say that tubular modules take one-third of the
time to install.
Unlike many standard solar-power installations, which have to be tilted to better
capture sunlight, the tubular system doesn’t need to be anchored on the roof with
ballasts and adhesives (Fig. 11.12).
The tubular modules perform optimally when mounted horizontally and packed
closely together, covering significantly more of the typically available roof area and
producing more electricity per rooftop on an annual basis than a conventional panel
installation.
The transparency for wind is also a very big advantage, for example, when the
tubular modules are used for rooftop-installations. PV Modules could be installed
also on the rooftops of very tall buildings. In regions visited frequently by hurricanes,
the wind semitransparency of the modules offers advantages too. The tubular modules
can withstand winds up to 130 m/h, 200 km/h, on roofs and in solar farms as it was
demonstrated first by Solyndra in Florida/US several years ago.
Green roofs could be combined with the ATPV. Particularly in big cities, where
a lot more electricity and PV is needed, and the green area has to grow in order to
clean the air, tubular modules can be combined with green roofing. A green roof
combined with PV double-harvests the rain, for cleaning the tubular PV at one side,

Fig. 11.12 Tubular modules as rooftop installation with service path; a combination with green
roofing possible
11 Augsburger Tubular Photovoltaic (ATPV) 273

and for irrigation of the plants on the other side. The plants need less water under
the shadows of the tubular modules, and the modules will be better cooled, which is
an obvious advantage and a win/win situation.

11.6 Agro-PV Under an ATPV Power Pergola

The semitransparency of the modules for light, wind, rain, and snow brings substantial
advantages. A continuous power pergola (Fig. 11.13) can be with any size: small,
medium, and big. The mounting systems could be much leaner than the one used
today to build flat module free-field installations (Fig. 11.13).
The elevated ATPV modules allow plants, animals, and fish to grow, people to
work, agro- machines having space to move. Agro-PV, sometimes also named in
the literature as Sun-sharing, Double harvesting, etc. gets another appearance and
meaning with tubular modules. The standard free-field PV installations need dis-
ruption in rows and strong anchoring of the module-blocks to withstand the wind.
Normally, the blocks of flat modules are mounted very low above the ground, so that
Agro-PV becomes impossible, their optical look is not very attractive and appealing,
plus it does not allow simple maintenance of the solar farms The PV is frequently
criticized in the last years for competing with the food chain, and in many countries,
PV Farming gets banned because of it.
The ATPV power pergola enables more homogeneous growth of plants and harvest
compared to an Agro-PV installation based on standard modules (Fig. 11.10) because
the solar radiation gets scrambled by the glass tube surfaces before reaching the plants
(only shadows with short periods interfere during a sunny day with the crop). Also,
the rain can reach the plants under the pergola homogeneously, and the snow too.
An advantageous microclimate appears under the power pergola, which stimulates
the growth of the protected crop. The shades reduce the need for irrigation, which
is very attractive particularly for regions facing strong global climate change and
desertification. In Bavaria, we observed that saving of water for irrigation was more
than 20%. This issue is more attractive for regions with more sun like Germany, South
Europe, CA, Africa, Middle East, etc. Compared with the systems build with standard
flat modules, the advantages in the case of ATPV are obvious. ATPV modules could
also be installed also in the standard way practiced at the moment. The pergola type
of installation allows dual and even triple use of the land.

11.7 Other ATPV Applications

ATPV could improve the appearance of facades in combination with autonomous

LED lighting.
ATPV could serve the street lighting sector.
274 V. Petrova-Koch and J. Mayer

Fig. 11.13 ATPV power Pergola build in 2014 in Weihenstephan, Freising

11 Augsburger Tubular Photovoltaic (ATPV) 275

ATPV could power water desalination installations particularly on the sea coast,
where the winds are strong.
Individual power tubes could serve many other niche-markets, which will not be
discussed here.

11.8 Conclusions

The Augsburger Tubular PV(ATPV) presented here is a relatively photovoltaic tech-

nology. Its unique characteristics are as follows:
– the PV cells needed can be manufactured cost-effectively relying on R2R tech-
nology, laser processing, and monolithically interconnection of the thin-film cells.
The efficiency of the flexible cells is competitive with the efficiency of many rigid
cells and modules known from the flat PV business.
– Soda-lime glass tubes, known for their competitiveness from classical lighting
industry, can find a new application and serve the PV industry on a large scale.
– the ATPV power tubes are very lean, lightweight, and easier to manufacture com-
pared to the Solyndra power tubes, known already to the PV community.
– the tubular modules are lightweight and transparent for solar radiation, rain, wind,
and snow. They allow to install easily in power pergola-manner for both: rooftops
and Agro-PV free fields. The better combination of the tubular modules (power
pergola) with plants, animals, fish, etc. is obvious, compared to installations based
on flat PV modules.
– A lot more needs could be done for the manufacturing of high-efficient low-cost
single and tandem flexible cells, needed for the manufacturing of Augsburger
tubular PV modules.
– A lot more needs to be done to establish the ATPV and to contribute for market
diversification. Innovative affordable applications in the field of PV are possible.
We work to establish ATPV as a new product and as a part of this growing and
booming PV market.

Acknowledgements We are thankful to all colleagues and partners supporting the ATPV efforts
so far: Osram and more recently Ledvance in Augsburg, particularly the Equipment Engineering
department and the GlassFab and also to our partners: Flisom in Switzerland; ZSW Stuttgart; the
HM in Munich; and the HSWT in Weihenstephan/Freising, Germany.

Appendix

Worldwide, more than 16.330.000.000.000 kWh electricity is consumed by the seven

billion inhabitants of our planet in 2016, with a rapidly increasing tendency for the
feature.
276 V. Petrova-Koch and J. Mayer

The farming area around the world is estimated of the order.

To produce food, 4 060 000 km2 of the US territory and 166500 km2 of the German
territory is used as farming land.
There are approximately 30 billion square feet (2.8 billion square meters) of flat
roofs in the US., an area large enough to collect the sunlight needed to power 18
million American homes or replace 38 conventional coal-fired power plants.
The flat roofs in Germany have an area of one million square meters, this area is
enough to supply electricity roughly to 5 Million German homes.
There is a hope that ATPV can be manufactured competitive and on the right scale
in Bavaria in a very near future. There is no doubt that affordable PV will serve than
at least a small share of this huge electricity and food demand on our planet.

References

1. E. Bellini, PV-Magazine (2017)

2. W. Hoffmann, The Economic Competitiveness of Renewable Energy, Scrivener Publishing, J.
Wiley, New York (2014)
3. DOE invested 535 Million US-Dollar, Obama was proud oft the Solyndra unique PV. About
$600 million in equity from investors since its inception in 2005 Investors: Virgin Green
Fund, Madrone Capital Partners, RockPort Capital Partners, Argonaut Private Equity, Masdar,
Redpoint Ventures, U.S. Venture Partners, Artis Capital Management and CMEA Ventures
4. Some Solyndra Patents 1. B Buller, ME Beck - US Patent 7235736, 2007 (Monolitical integra-
tion of cylindrical solar cells); 2. E. Milshtein, B. Buller - US Patent 7879685, 2011 (System and
method for creating isolation between layers comprising solar cells); 3 B Buller, C Gronet, K
Truman, T Brezoczky - US Patent 8,443,558, 2013 (Support system for solar energy generator
panels)
5. Patent application US 7235736B1. Petrova-Koch, R. Mackh, J. Mayer (2014)
6. S. Arora, S. Chitkara, R. Udayakumar, M. Ali, J. Pet. Gas Eng. 2(4), 74–82 (2011)
7. Lisec, Austria, Private Communications
8. Solar World-innovations, Private Communications
9. A. Tiwari et al., Review of progress toward 20% efficiency flexible CIGS Solar cells and
manufacturing issues of solar modules, in 2012 IEEE 38th Photovoltaic Specialists Conference
(PVSC), vol. 2 (2012)
10. F. Kessler, D. Hariskos, S. Spiering, E. Lotter, H.P. Huber, R. Wuerz, Chapter 9 of this in book
11. B. Buller, M. Back, Solyndra US Patent 7235736B1 (2007)
12. E. Milshtein, B. Buller, Solyndra US Patent 787968B2 (2011)
13. D. Forgacs, K. Wojciechowski, O.Malinkiewicz, Chapter 10 in this book
14. Tesla Solar Roof Tile Manufacturing: Business News (2018)
15. Hevel Solar: HJT c-Si bifacial cells and glass/glass Module manufacturing started in Russia
(2017)
16. Longi c-Si PERC cells and halfcell module manufacturing started in 2017
17. Energetica in Austria announced Next Gn one Gigawatt Fab in Austria in August (2018)
Chapter 12
Fluorescent Solar Energy Concentrators:
Principle and Present State
of Development

Adolf Goetzberger

Abstract A review of the history and recent developments of fluorescent concen-

trators is given. The principle of this innovative device is based on light guiding
in a transparent matrix doped with fluorescent centers, mainly organic dyes. The
dyes strongly absorb a certain band of the solar spectrum and emit at a red-shifted
frequency for which the concentrator is transparent. Since the light is emitted isotrop-
ically a large part is guided to the edges of the concentrator and arrives in concentrated
form at the smaller side-edge. This concentrated light could be easily guided to an
adapted solar cell. Work was done in the late seventies and early eighties, but then
interest vanished because of several difficulties like unstable dyes and nonavailability
of dyes in the infrared, which would be best adapted to silicon solar cells. Theoret-
ically very high efficiencies can be reached if several plates with different colors
are stacked and equipped with adapted solar cells. In recent years new interest has
emerged in fluorescent concentrators. Two concepts are being most actively inves-
tigated now. The first is a band pass mirror on the front surface of the concentrator
that reflects selectively the emitted light but transmits the light to be absorbed. In
this manner also rays not within internal reflection are retained. It has been shown
that only in this way the ultimate efficiency can be reached for a single plate concen-
trator. This efficiency is not much lower than for a directly illuminated ideal solar
cell. A second new approach consists of replacing organic dyes with quantum dots.
They offer the hope of greater stability, they can be tuned to different absorption
wavelength by their size and by the spread of sizes their redshift can be tailored. A
third approach is trying to put the dye centers in a surface coating, which opens the
possibility to apply as a wafe guiding plates also more temperature resistive materials
like glass.

A. Goetzberger (B)
Fraunhofer Institute for Solar Energy Systems, Heidenhofstr. 2, 79110 Freiburg, Germany
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 277

V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4_12
278 A. Goetzberger

12.1 Principle

The use of transparent sheets doped with fluorescent dyes for the concentration of
sunlight was suggested first in the 1970s [1, 2]. The principle itself is much older; it
was first used in scintillation counters for atomic physics [3, 4]. Although significant
advances were made in early work, after some years the further progress of work
was limited by the materials available at that time—in particular the dyes—and
interest was dormant for decades. Only recently new progress in materials, as well
as theoretical advances, rekindled interest. In this article the historic work and the
present state of the art will be reviewed.
Figure 12.1 shows the principle of the fluorescent concentrator, sometimes also
called the luminescent collector: I represent an incident beam of light interacting
with a dye molecule dissolved in a matrix of index of refraction n. The incident light
will be absorbed and emitted at a different wavelength. (In the normal case of Stokes
fluorescence the emitted wavelength will be shifted to a longer wavelength.) If the
probability of emission is equal in all directions and part of the light will leave the
transparent medium (F 1 ) while another part (F 2 ) will be reflected back because it
intersects the surface at an angle leading to total internal reflection. It is important
to note that this reflection is in principle lossless. Thus, the captured light is guided
within the transparent sheet which will be also called the collector. Concentrated light
can thus be obtained at the edge of the concentrator. The edges of the concentrator
not contacted by solar cells have to be covered with mirrors. The fraction of light
contained in the concentrator can be quite high. The fluorescent concentrator is
the only concentrator known that can achieve high values of concentration without
tracking. In contrast, concentrators based on geometric optics have to be tracked in
order to achieve concentration of more than 2, and they can only use direct sunlight.
This is a fundamental limitation following from nonimaging optics.

Fig. 12.1 Principle of fluorescent collection. Dye molecule D interacts with incoming light beam
I. Secondary beams are partly lost (F 1 ) and partly guided in a transparent material (F 2 –F 4 )
12 Fluorescent Solar Energy Concentrators: Principle and Present … 279

In [5, 6] it has been shown that loss L due to light leaving the concentrator through
the two boundary planes is given by:

L = 1 − (n 2 − 1)1/2 /n (12.1)

where n is the index of refraction of the concentrator. For n = 1.5 for instance, L =
0.2546, for n = 2, L is 0.134 (it will be shown below that by the optimized design of
concentrators these losses can be reduced significantly). The solid angle determined
by the onset of total internal reflection is called the loss cone.
A basic requirement for efficient collection is that the incident wavelength has
to have a short absorption length, the emitted wavelength a very long absorption
length. This is accomplished by selecting a dye, whose absorption and emission
wavelengths are well separated. Also, the fluorescence efficiency has to be high. The
latter requirement is more easily met than the former. Quite a number of dyes are
known to have a fluorescence efficiency of close to 100%. New dyes with better
spectral properties and stability are being developed by the chemical industry. The
emission and absorption characteristics are not only dependent on the nature of the
molecule, but also on that of the solvent. The material of the collector has to be
highly transparent and a good solvent for dyes. Plastics, glass, or organic solvents
contained between plastic or glass sheets are possible candidates for this purpose.
It is also possible to deposit the dye dissolved in a thin film on the surface of a
completely transparent sheet.
Although prices for solar cells, including c-Si and other cells on wafers have come
down considerably in recent years the cells are still the most expensive part of a flat
plate solar generator. Conversion of concentrated solar light leads to a reduction of
the cost of the PV systems if the concentrator applied has a significantly lower cost
than the PV cells.
Fluorescent concentrators have the following advantages:
• Concentration of Sunlight without tracking
• Concentration of direct and diffuse light
• The concentrators are particularly well adapted to overcast conditions that occur
frequently in temperate climates
• Possibility of spectrum splitting by use of several sheets doped with different dyes
shown below
A single semiconductor can never convert sunlight with the highest possible effi-
ciency because quanta with higher energy than the band gap lose their energy by
thermalization and those with lower energy are not absorbed. Therefore, multijunc-
tion cells employing different semiconductors achieve the highest efficiency today.
They consist of complex layers of III–V compounds but are very expensive and so
they can only be used in concentrating systems.
280 A. Goetzberger

12.2 Concentrator Stacks

Fluorescent concentrators offer the possibility to separate different portions of the

solar spectrum just like multijunction cells and concentrate them at the same time
[2]. Figure 12.2a shows a stack of three concentrators c1 –c3 contacted by three dif-
ferent solar cells S 1 –S 3 . The ideal absorption and emission spectra of the fluorescent
molecules in the collectors are sketched in Fig. 12.2b. Because every concentrator is
fully transparent to the unabsorbed part of the spectrum a rather complete separation
of the solar spectrum is possible in this manner. The solar cells S l –S 3 have band gaps
adjusted to the emission bands of the concentrators. A more detailed analysis carried
out below demonstrates that the theoretical energy conversion efficiency is greater
than 30% versus 24% for single-junction silicon cells (for AM 1.5).
For the calculation of quantum efficiency we put aside the energy loss due to
Stokes shift and are only interested in quantum losses. It is assumed that the quan-
tum efficiency of fluorescence is 100%. Reflection losses can be minimized with
antireflection layers but even without antireflection coatings, the stack offers possi-
bilities for minimizing reflection losses.
Much progress has occurred with antireflective coatings in the last 30 years. They
have become more efficient and also more economical. Although such coatings have
a lower index of refraction than the collector sheet they do not interfere with total
internal reflection because only the difference between medium in which the light is
emitted and air is of importance.

Fig. 12.2 a Fluorescent concentrator stack. Collectors cl , c2 , c3 are combined with solar cells S 1 ,
S 2 , S 3 of different bandgaps. b Absorption and emission (shaded) spectra of dyes in c1 –c3 , λl –λ3
are peak emission wavelengths of c1 –c3
12 Fluorescent Solar Energy Concentrators: Principle and Present … 281

For stacks, the following facts apply: The spectral range for these antireflective
coatings becomes narrower and, therefore, easier to realize for the lower plates of
the stack. It can be seen that c1 has to be transparent for the entire solar spectrum,
c3 only has to operate in the long-wavelength range. An additional possibility is the
use of photonic structures to decrease the loss cone angle as described below.
Now the quantum efficiency for a stack will is calculated. Again normal incidence
is assumed. The procedure applied here is the following: The incident spectrum is
divided into m parts such that each part contains an equal number of photons. This
requirement is convenient for computation but not necessary for practical applica-
tions. An advantage of the stack is the fact that one-half of the escape cone losses of
the preceding plate is recovered by the following one. This is seen from Fig. 12.2b.
Collector c1 emits at wavelength λ1 . If the absorption band of c2 is arranged as shown
in Fig. 12.2b, c2 absorbs not only the solar radiation at the wavelength band λ1 com-
ing from above but also the radiation emitted by the dye in c1 in the direction of c2 .
The same is true for the lower parts of the stack. A recursion formula will now be
derived for the quantum efficiency of a stack versus the number of collector plates
in the stack. The symbols used in this calculation are given in Fig. 12.3.
The incident radiation S suffers reflection losses R at every interface traversed.
Loss of fluorescent radiation not retained in the collectors is indicated by L. As was
pointed out already, the loss cone losses directed towards c2 are recoverable. A minor
detail also indicated in Fig. 12.3 is the fact that reflection of this radiation in leaving
c1 does not have to be taken into account because exactly the same loss occurs at the
top surface of the collector thus canceling the former loss (dashed lines in Fig. 12.3)
Let S 1 be the amount of radiation entering collector c1 at the absorption band of
this collector and C l be radiation collected there S 2 …S m and C 2 …C m are defined
accordingly. Then S 1 (λ1 ) = (l − R)/m

C1 = S1 (1 − L)

Fig. 12.3 Definition of

quantities used in calculation
of quantum efficiency. For
details see text
282 A. Goetzberger

Table 12.1 Quantum efficiency ηQ of collector stack

m 1 2 3 4
ηQ 0.7126 0.7299 0.7204 0.7030

The amount of radiation entering the second collector in the appropriate band is
one mth of the incident spectrum attenuated by reflection (multiple reflections are
neglected) and augmented by half of the loss from the preceding collector attenuated
by reflection upon entering the second collector and so on:

S2 (λ2 ) = (1 − R)3 /m + (1 − R)S1 L/2; C2 = S2 (1 − L);

S3 (λ3 ) = (1 − R) /m + (l − R) S2 L/2;
5
C3 = S3 (1 − L);
S4 (λ4 ) = (1 − R) /m + (1 − R)S3 L/2;
7
C4 = S4 (1 − L);
..
.
Sm (λm ) = (1 − R)2m−1 /m + (l − R)Sm−1 L/2; Cm = Sm (1 − L).

The quantum efficiency is the sum:


m
ηQ = Ck (12.2)
k=1

Equation (12.5) will now be evaluated for a realistic case: A common high trans-
parency plastic-like Plexiglas with an index of refraction of 1.49 without antireflec-
tion coating and L = 0.2587 (Table 12.1).
Obviously, the stacking of collectors does not degrade quantum efficiency.
It should be pointed out that the order of dyes in the different sheets as shown in
Fig. 12.2 can also be reversed—the longest wavelength on top and the highest at the
bottom. This may be advantageous. In this case, a band pass mirror as described in
the next section can be applied.

12.3 Light Guiding by Photonic Band Pass Mirrors

The loss of fluorescent light through the surfaces of the collector can be entirely
avoided by covering the front surface with a wavelength selective mirror (hot mirror)
[7, 8]. This mirror should have the following properties:
• A sharp cutoff edge at the onset wavelength of the dye’s emission. All light below
this wavelength should be transmitted, all above should be reflected.
• Near 100% reflection for light comes from all directions.
12 Fluorescent Solar Energy Concentrators: Principle and Present … 283

Fig. 12.4 Optimal design of fluorescent collector with band pass mirrors [9]. The coverage factor
f = s2 /l 2

A band pass mirror with nonideal properties can be realized with commercial
hot mirrors [8, 9]. Better results can be expected from photonic structures. First
experiments with Rugate structures have shown promising results [10]. It features a
continuously varying refractive index profile in contrast to the discrete structure of
normal Bragg reflectors. This results in the suppression of side loops, which would
cause unwanted reflection and loss of usable radiation. In [9, 11] it was shown that
the ultimate efficiency limits of fluorescent concentrators can only be reached with
selective mirrors. On the other hand, the light within the loss cone has a long path
length before reaching the edges of the concentrator. In this case an arrangement as
depicted in Fig. 12.4 is more advantageous.

12.4 Factors Determining Energy Efficiency of Fluorescent

Concentrators

For the energy conversion efficiency, all energy losses in the collection process have
to be taken into account. These are in detail:
• R: Surface reflection coefficient. This is either the Fresnel reflection coefficient or
a lower value if an antireflection coating is applied
• ηabs : Absorption efficiency of the dye due to its absorption spectrum with respect
to the solar spectrum
• ηqua : Quantum efficiency of dye
• ηstok : “Stokes efficiency”; (1 − ηstok ) is the energy loss due to Stokes shift
• ηtrap : Efficiency of light trapping by total internal reflection (Loss cone)
• ηdye : Efficiency of light conduction limited by self-absorption of dye
• ηmat : “Matrix efficiency”; (1 − ηmat ) is the loss caused by light scattering or
absorption in the matrix. ηdye and ηmat determine the mean free path of the emitted
light in the collector
• ηtref : Efficiency of the light guiding by total internal reflection. This depends on
the surface quality of the matrix
284 A. Goetzberger

The overall optical efficiency can then be written as:

ηopt = (1 − R)ηabs ηqua ηstok ηtrap ηdye ηmat ηtref (12.3)

These loss factors will now be discussed.

The surface reflection loss R is usually the Fresnel reflection loss. It is given by
[(n – 1)/(n + 1)]2 where n is the refractive index. For most transparent materials it
amounts to about 4% per surface. It can be minimized by employing an antireflection
coating. If it has a smooth surface it will not interfere with total internal reflection.
The absorption loss ηabs is determined by the fraction of the solar spectrum
absorbed by the dye. Practically all dyes absorb only part of the solar spectrum.
It is possible to incorporate more than one dye into a collector, leading to a cascade
of absorption and reemission. A disadvantage is that at every emission part of the
radiation escapes through the loss cone. If the dyes are located very close to one
another radiationless energy transfer can occur via the Förster mechanism. There-
fore, doping of a thin surface region of the collector with a high concentration of
dyes may have advantages.
The quantum efficiencies ηqua of the dyes can vary considerably. Only dyes with
efficiency close to unity are usable. Fortunately, dyes with such efficiency and good
stability are available but this refers only to the visible range.
The stokes loss ηstok is caused by the frequency shift between absorbed and emitted
light. It is inherent in the fluorescence process. On one hand this shift should be small
to minimize the energy loss, on the other it should be large enough to avoid overlap
of absorption and emission loss (related to ηdye ).
The light trapping efficiency ηtrap is given by L (from (12.1)). It increases with an
increasing refractive index of the collector material. In practice there are only little
differences of available materials.
ηdye designates the efficiency limitation caused by self absorption of the dye. All
known dyes have a certain overlap of absorption and emission spectrum as shown in
Fig. 12.5. Also shown in this figure is how the spectrum is modified by multiple reab-
sorption and emission. The spectrum at the output is then redshifted. It has also been
found that dyes cause a very small absorption within the emission region and beyond
[12]. This absorption is difficult to determine but can have a great influence on the
overall efficiency. A further effect causing unwanted absorption are photodegrada-
tion products caused by degradation of dyes [13]. These products can be annealed
by heating in the dark.
ηmat is the efficiency due to scattering or absorption in the matrix. It depends
strongly on the purity and preparation of the matrix.
Total internal reflection is theoretically lossless but in practice, it depends on the
surface finish of the matrix. This loss ηtref can in principle be completely avoided
by employing a reflecting band pass filter at the surface of the collector as pointed
out above. Ideally this filter should also serve as an antireflection coating for the
wavelength region absorbed by the dye to reduce R.
12 Fluorescent Solar Energy Concentrators: Principle and Present … 285

Fig. 12.5 Absorption and emission spectrum of typical dye. Output spectrum is modified by reab-
sorption

12.5 Theoretical Limits of Concentration and Efficiency

12.5.1 Limit of Concentration

Let us first consider a conventional concentrator based on geometric optics. In this

case, the concentration is limited by the conservation of étendue or Liouville’s theo-
rem which relates the beam divergence at the input and output aperture. The most effi-
cient concentrator is the CPC (Compound Parabolic Concentrator) that can approach
the theoretical limits.
The concentration ratio C is given (for three dimensions) by:

n 2 sin2 2
C≤ (12.4)
sin2 1

where C is the concentration factor, θ 1 = input angle of light, θ 2 = output angle at

receiver. If diffuse light is to be concentrated, θ 1 = 90° and, therefore, the maximum
concentration is limited to n2 which is about a factor of two for most available
transparent materials.
On the other hand, geometrical concentration can be very effective at very small
input angles. For solar radiation the limiting input angle is given by the viewing angle
of the sun. Therefore, tracking concentrators for direct radiation permit very high
concentration.
The maximum concentration of the fluorescent concentrator has been determined
by Yablonovitch [14] and Smestad et al. [7]. They point out that because of the
energy loss due to the Stokes effect the system operates as an optical heat pump. The
radiance at a given energy is increased by changing some of the incoming energy to
heat.
For fluorescent concentrators, the concentration factor is determined by:
286 A. Goetzberger

(v2 )2 h(v1 − v2 )
C≤ exp( ) (12.5)
(v1 ) 2 KT

where ν 1 , = frequency of absorbed light, ν 2 = frequency of emitted light. The con-

centration factor depends only on the magnitude of the Stokes shift. So concentration
occurs at the sacrifice of energy efficiency.

12.5.2 Limit of Efficiency

The limits of efficiency were examined by Rau et al. [9], Glaeser and Rau [10],
and Markvart et al. [15, 16]. They showed that the detailed balance principle intro-
duced into solar cell physics by Shockley and Queisser [17] can also be used for
the fluorescent concentrator solar cell system. They investigated a single-stage con-
centrator and found that very high efficiency is possible provided the concentrator
surface is covered by a perfect band pass mirror. In [9, 10] Monte-Carlo simulations
were employed to obtain relations between efficiency, band gap energy, and coverage
fraction. In Fig. 12.6 the results of these calculations are presented.
The simulated efficiencies for a fluorescent concentrator (f = 1) with PBS (open
squares) follow the radiative Shockley–Queisser efficiency limit ηrad (solid line)
when calculated with reference to the photonic threshold energy E th . The efficiencies
without PBS (open circles) are at about 0.88 × ηrad (dashed line) calculated with
reference to the band gap energy E g = E th − 0.2 eV of the underlying solar cell.
(b) Efficiencies for a band gap energy E g = 1.12 eV and threshold energy E th =

Fig. 12.6 a Monte-Carlo simulations [9, 10]. For explanations see text
12 Fluorescent Solar Energy Concentrators: Principle and Present … 287

Fig. 12.7 Conversion

efficiency of an ideal Si cell
with and without a
fluorescent concentrator

1.32 eV for ideal (rad.) and nonideal (nonrad.) solar cells. Without photonic band
stop (PBS) the efficiencies degrade monotonically with decreasing coverage fraction
f (full and open circles, for the radiative and the non-radiative case). With PBS the
systems efficiency drops much slower for an ideal solar cell (open squares) and even
achieves an optimum at f ≈ 10−2 in the non-radiative case (full squares). Comparable
results were obtained by Markvart [15, 16] as shown in Fig. 12.7. It is seen that ideal
efficiency is 26.8% versus 29.6% without concentrator.

12.6 Improvements in Basic Design

12.6.1 Optical Concentrators at the Collector Output

This is an old proposal [18]: At the edge of a collector plate a concentrating element
(taper or more effective a CPC) is attached before the solar cell (Fig. 12.8). The
additional concentration possible in this manner which is given by (12.4). Rests on
two facts: The divergence angle of incoming rays is limited by the onset angle of total
internal reflection and can then further be converted to 90°. The collector material

Fig. 12.8 Fluorescent

collector with two stage
concentration by attached
taper with refractive
index n2 > n1
288 A. Goetzberger

can be made of a material of a higher index of refraction than the collector material.
A concentration factor between 1.5 and 2 is thus possible.

12.6.2 Combination of Fluorescent Collector with Large

Area Si-Solar Cell

Because at present no dyes emitting in the IR with acceptable properties are available
a good compromise is the following concept [10] (Fig. 12.9).
A fluorescent concentrator doped with a dye emitting in the red or orange is
equipped with a GaInP cell that has its maximum response in this range. The collector
is transparent to all radiation not absorbed by the dye. The transmitted radiation is
converted by a large area silicon cell at the bottom. Figure 12.10 shows the efficiencies
of the silicon cell alone and the combined system. The silicon cell without the

Fig. 12.9 Combination of a short wavelength fluorescent concentrator with attached GaInP cell
and Si bottom cell

Fig. 12.10 Efficiency of

silicon cell alone and of total
system
12 Fluorescent Solar Energy Concentrators: Principle and Present … 289

concentrator had an efficiency of 16.7%. Under the fluorescent concentrator,

the efficiency dropped to 14.0%. The total system efficiency was 17.7%, which
is significantly higher than the silicon solar cell alone.

12.6.3 Combination of Fluorescent Concentrator

with Up-conversion

Recently great progress has been achieved with up-converters—compounds that

combine two or more quanta to a higher energy quantum [19].
This effect can also be useful for fluorescent concentrators as indicated in
Fig. 12.11.
Below the fluorescent collector, an up-converter and a mirror is arranged. The
spectral properties of the fluorescent collector and the up-converter are shown below.
The dye in the collector shifts the absorbed wavelength range to longer wavelength
above the band energy of the solar cell. The photons emitted downwards are con-
verted by the up-converter to shorter wavelength to be reabsorbed by the collector.
Furthermore, the collector is transparent to the solar spectrum within the emission
range of the dye in the collector. These photons will also be converted to higher
wavelengths.

Fig. 12.11 Fluorescent collector combined with up-converter. For an explanation of function see
text
290 A. Goetzberger

12.6.4 Combination of Collector Stack with Band Pass

Mirror

The collector stack can also be combined with a band pass mirror as described in
Sect. 12.3. For this purpose the following conditions apply:
• Longest wavelength dye on top
• Solar cells placed below each other
• Transparent electrical connections
In Fig. 12.12a the stacking of the collectors is sketched, in Fig. 12.12b we see
the wavelength diagram. Consider the top plate: The dye absorbs in the shorter
wavelength and emits at longer once. The upper mirror reflects the emitted radiation
but is transparent for all other wavelengths, downward emitted radiation is reflected
by the lower mirror which extends into this wavelength region as indicated by the
short arrows. The lower plates operate in the same manner.
The collector stack with band pass mirrors offers intriguing prospects: It is the
only device that can concentrate diffuse radiation and could also approach highest
conversion efficiency! While the original fluorescent concentrator shown in Fig. 12.2
has a theoretical efficiency of about 30%, the new design can go much higher. From
Fig. 12.7 we derive that a one plate concentrator with a silicon cell has a theoretical
efficiency of 90% of the thermodynamic efficiency. A collector stack is a spectrum
splitting device and can now achieve an efficiency close to that of a multijunction
cell. If we assume the theoretical efficiency to be 60%, the multi-stage fluorescent
concentrator has a theoretical efficiency of 54%.
Obviously, many difficulties have to be overcome to reach this goal, such as: Near
ideal band pass mirrors are needed, the dyes should have a well-defined absorption
bands without absorption in the shorter wavelengths and appropriate wide gap solar
cells have to be available. In addition the cost should be competitive.

Fig. 12.12 Fluorescent collector stack with band pass reflectors. a The geometrical arrangement,
b wavelength representation
12 Fluorescent Solar Energy Concentrators: Principle and Present … 291

12.7 Experimental Results

12.7.1 Results of the Initial Period

Experimental work can be divided into two periods: The initial period from 1977 to
about 1985 and the more recent period which started about 2000. The older results
are still meaningful because they represent benchmarks to be reached and exceeded
in present work. The most interesting parameters are the overall electrical efficiency
and the long term stability of the concentrators These are in turn influenced by the
properties of matrix and dye. In comparing efficiencies the dimensions of the samples
are very important. Results of the early period are summarized in review papers by
Wittwer et al. [20] and Zastrow [21] (Table 12.2).
The efficiency of 4% is still the highest value that has been realized for such a
large area. The second most important issue is the stability of the collector dye-system
under illumination. It could be observed that by continued development the stability
of the dyes improved during the course of the work. A representative measurement
is shown in Fig. 12.13. The cumulative illumination corresponds to more than two
years of outdoor exposure. Also clearly seen is the recovery of fluorescence during
periods of darkness.
A problem that continues to limit efficiency is the lack of useful long-wavelength
dyes. They still have low quantum efficiency and are not very stable. Figure 12.14

Table 12.2 Electrical

Dimension (cm3 ) Absorption range (nm) Efficiency
efficiencies for 40 × 40 cm2
(%)
collectors with Si and GaAs
cells. The highest efficiency Si GaAs
of 4.0% was obtained with a 40 × 40 × 0.3 360–550 2.1 2.5
double stack and GaAs cells
40 × 40 × 0.3 490–610 1.4 2.5
Stack of both 360–610 3.0 4.0

Fig. 12.13 Fluorescence

during long term light
exposure in kWh/m2 .
Fluorescence is normalized
to the maximum value
292 A. Goetzberger

Fig. 12.14 Quantum efficiencies of different dyes versus peak emission wavelength

gives a compilation of quantum efficiencies found until 1984. It is evident that there is
a general tendency is towards lower efficiency when emission wavelength increases.

12.7.2 Recent Experimental and Theoretical Work

Danos et al. [22] characterized fluorescent concentrators based on solid, liquid, and
Langmuir Blodget films. Chatten et al. [23] have developed a self-consistent 3D
thermodynamic model for planar concentrators, modules, and stacks. The results for
test concentrators containing both quantum dots and organic dyes as the luminescent
species show excellent agreement with experiment.
Organic dyes are still the best choice for fluorescent concentrators as shown in [24].
In this paper by Richards and McIntosh collectors doped with multiple dyes were
investigated by ray tracing using experimental data from newly available dyes. In such
a collector multiple are absorbed and reemitted by increasingly longer wavelength
dyes. In this manner a large part of the solar spectrum is absorbed and emission
occurs at a wavelength which is more suitable for silicon cells. They found that a
mixture of five dyes including a near IR dye of 85% quantum efficiency gave the best
result of 44% electrical efficiency. Only radiative transfer and high-efficiency solar
cells were studied. They also state that all dyes show photostability under sunlight
for at least 10 years. Lower stability was found by Reda with more conventional dyes
[25].
A new possibility to obtain better performing luminescing centers consists of
quantum dots as has been pointed out by Barnham et al. [26]. These are nanometer-
sized semiconductor crystallites. Because they are inorganic there is a hope that they
12 Fluorescent Solar Energy Concentrators: Principle and Present … 293

Fig. 12.15 Three PMMA plates doped with different color dyes. These plates are about 25 years
old but not degraded, they were however mostly stored in the dark. Although the light output looks
very spectacular only part of the light arriving at the edges is emitted because a big part is reflected
back by total internal reflection
294 A. Goetzberger

exhibit greater light stability than organic dyes. Another advantage is that the absorp-
tion threshold can be tuned by choice of the dot diameter. Furthermore, Williams
et al. showed by a thermodynamic model that the red shift of emission is related to
the spread of quantum dot sizes.
In view of these advantages, there is considerable interest in developing fluorescent
concentrators doped with quantum dots. Numerous investigations were carried out
in recent years. Mainly quantum dots based on CdS were used. Schüler [27] applied
the dots by the solgel technique to glass surfaces but also core-shell quantum dots
consisting of CdSe cores with CdS or ZnS shell were studied [28, 29]. So far the
quantum dots show continued improvement but quantum efficiency is not yet as high
as for the best organic dyes. Also some instability has been observed.
Further, it should be remarked that fluorescent concentrators have applications
beyond photovoltaics. They have been in use for advertising for many years. They
can also be used in green houses to convert the green light to red light which can be
better absorbed by plants [20]. Application for daylighting has been studied in the
early days [30] and recent work with newer dyes appears very promising [12].
Finally, I present a photograph of three collector plates as they appear in daylight
(Fig. 12.15).

References

1. W.H. Weber, J. Lambe, Appl. Opt. 15, 2299 (1976)

2. A. Goetzberger, W. Greubel, Solar energy conversion with fluorescent collectors. Appl. Phys.
12, 123 (1977)
3. W.A. Shurcliff, R.C. Jones, J. Opt. Soc. Am. 39, 912 (1949)
4. J.B. Birks, The Theory and Practice of Scintillation Counting (Pergarnon Press, London, 1964)
5. G. Keil, J. Appl. Phys. 40, 3544 (1969)
6. G. Keil, Nucl. Instrum. Methods 87, 111–123 (1970)
7. G. Smestad, H. Riess, R. Winston, E. Yablonovitch, Solar Energy Mater. 21, 99 (1990)
8. B.S. Richards, A. Shilav, R. Corkish, in 19. EU PV Solar Energy Conference (2004), p. 113
9. U. Rau, F. Einsele, G.C. Glaeser, Appl. Phys. Lett. 87, 171101 (2005)
10. J.C. Goldschmidt, S.W. Glunz, A. Gombert, G. Willeke, in 21. EU PV Solar Energy Conference
(2006), p. 107
11. G.C. Glaeser, U. Rau, Proc. SPIE 6197, 143 (2006)
12. A.A. Earp, G.B. Smith, P.D. Swift, J. Franklin, Sol. Energy 76, 655 (2004)
13. A. Zastrow, H.R. Wilson, K. Heidler, V. Wittwer, A. Goetzberger, in 6th EU PV Conference
(1983), p. 202
14. E. Yablonovitch, J. Opt. Soc. Am. 70, 1362 (1980)
15. T. Markvart, J. Appl. Phys. 99, 026101 (2006)
16. T. Markvart, L. Danos, P. Kittidachachan, R. Greef, in 20. EU PV Solar Energy Conference
(2005), p. 171
17. W. Shockley, H.J. Queisser, J. Appl. Phys. 32, 510 (1961)
18. A. Goetzberger, O. Schirmer, Appl. Phys. 19, 53 (1979)
19. S. Balushev, T. Miteva, V. Yakutin, G. Nelles, A. Yasuda, G. Wegner, Appl. Phys. Lett. 14,
143903 (2006)
20. V. Wittwer, W. Stahl, A. Goetzberger, Solar Energy Mater. 11, 187 (1984)
21. A. Zastrow, SPIE 2255, 534 (1993)
12 Fluorescent Solar Energy Concentrators: Principle and Present … 295

22. L. Danos, P. Kittidachachan, P.J.J. Meyer, R. Greef, T. Markvart, in 21. EU PV Solar Energy
Conference (2006), p. 443
23. A.J. Chatten, D.J. Farrell, B.F. Buxton, A. Büchtemann, K.W.J. Barnham, in 21. EU PV Solar
Energy Conference (2006), p. 315
24. B.S. Richards, K.R. McIntosh, in 21. EU PV Solar Energy Conference (2006), p. 185
25. S.M. Reda, Sol. Energy 81, 755 (2007)
26. K. Barnham, J.L. Marques, J. Hassard, P. O’Brien, Appl. Phys. Lett. 76, 1197 (2000)
27. A. Schüler, Solar Energy (2007)
28. S.J. Gallagher, B. Norton, P.C. Eames, Sol. Energy 81, 813 (2007)
29. S.J. Gallagher, B.C. Rowan, J. Doran, B. Norton, Sol. Energy 81, 540 (2007)
30. A. Zastrow, V. Wittwer, Proc. SPIE 653, 93 (1986)
Index

A C
Absorber, 176–181, 183–190, 192, 195– Carrier mobility, 122, 191
199, 202, 203, 205, 224, 227, 233, Cell producers, 52, 115, 147
235, 238, 239, 242–244, 260, 265 Chemical vapor deposition, 46, 50, 58, 69,
Absorption, 47, 60, 66, 71, 76, 114, 121, 122, 74, 84, 120, 122, 137, 138, 192
134, 137, 138, 140, 143, 147, 177, Collector stack, 282, 290
191, 193–195, 197, 200, 203, 221– Concentrate the solar radiation, 261, 264,
226, 233, 238, 239, 245, 260, 277, 273
279–281, 283, 284, 290, 294 Concentrated sunlight, 137, 164
Aesthetic, 77, 115 Concentrator III–V PV, 5
AlAs–GaAs, 134, 139, 140 Concentrator lens-cell system with radically
AlGaAs, 135–140, 142, 143, 145, 146, 148, reduced size, 4
152
Concentrator Photovoltaic Systems (CPV),
AlGaAs/GaAs solar cells, 136–140, 142,
4–6, 19, 30, 124
152
Concentrator PV installation, 147
All-glass module design, 153
Cost competitive, 95, 261, 262
Amorphous silicon, 114, 116, 119, 124
Amorphous solar cell, 45, 47, 77, 114, 116, c-Si, 16, 18, 19, 95, 114–126, 235, 258, 261–
119, 121, 124, 191 264, 266, 270, 279
Antireflection and passivation layer, 56 Crystalline wafer silicon technology, 16
Antireflective coatings, 125, 280, 281 Customer needs, 19, 29
CZ single-crystal growth, 134

B
Back surface field, 16, 66, 71, 85, 86, 115,
D
120, 135, 137
Back surface reflectors, 16, 57, 60, 66, 71, Daylighting, 294
85, 86, 115, 120, 135, 137 Decentralized power, 25
Band gap, 133, 135, 138, 142, 146, 182, 186, Diffuse light, 68, 76, 279, 285
192, 195, 224–227, 233, 235, 236, Diode, 116, 133, 143, 146, 150
240, 244, 279, 286 Donor/acceptor, 187
Band pass mirror, 277, 282, 283, 286, 290 Double heterostructure, 134, 135
Bifacial, 50, 53–61, 65, 66, 68–71, 76–80, Dye, 19, 221, 233, 277, 279, 281–284, 288–
82, 114–116, 118, 121, 127–130, 267 292
Building-integrated PV, 115 Dye molecule, 278
Buried contact, 4, 181 Dye-sensitized electrochemical solar cell, 4
© Springer Nature Switzerland AG 2020 297
V. Petrova-Koch et al. (eds.), High-Efficient Low-Cost Photovoltaics,
Springer Series in Optical Sciences 140,
https://doi.org/10.1007/978-3-030-22864-4
298 Index

E High efficiency, 4, 5, 19, 40, 43, 44, 47, 50,

Efficiencies, 18, 36, 40, 41, 43, 48–50, 54– 52, 59, 61, 74, 83, 86, 96, 109, 113–
56, 61, 65–67, 70, 72, 73, 75, 77–80, 115, 117–121, 124, 136, 137, 142,
83, 85, 87, 95, 96, 101, 102, 110, 113, 146, 147, 151, 183, 184, 266
114, 123, 130, 134, 137, 140, 142, High efficiency application, 134
146–148, 150, 151, 159, 164, 165, HIT, 6, 16, 40, 77, 116
180, 181, 183–185, 187–190, 193, HIT c-Si/a-Si:H PV cells, 4
195, 207, 219, 233, 240, 244, 266,
267, 277, 284, 286–288, 291, 292
Electron–hole pairs, 138 I
Energy losses, 10, 38, 39, 45, 47, 283 Illuminated I–V curves, 150
Energy supply for satellites, 136 InGaAsP, 135
Epitaxial growth, 120, 133, 137, 144 InP-based cells, 137
Integrated product flow, 36
EQE, 239
Internal Bragg reflector, 137, 139, 141
EQE and FF, 239, 241
ITO, 114, 116, 121–123, 126, 129, 191
Evaporation process, 185, 232
Experience factor, 15, 22
L
Laser grooved c-Si solar cell, 4
F Lens optical efficiency, 160
Feed-in law, 4 Li-doped Si solar cells, 2
Feed-in tariff, 13 Lifetime, 11, 23, 26, 27, 40, 46, 85, 115, 118,
Fill factor, 99, 101, 121, 143, 150, 151, 160, 189, 242–244
161, 187, 188, 240 Light harvesting, 220, 224, 273
First solar cell, 2 Liquid-p hase epitaxial growth, 135
Fresnel lens, 151–158, 160, 164 Liquid-Phase-Epitaxy (LPE), 136–139, 142,
143
Local back surface field, 57, 59
Low-temperature<>, 34, 44, 47, 50, 53, 54,
G 60, 61, 116, 126, 138, 142
GaAs, 4, 5, 20, 29, 114, 134–146, 148, 152,
158–161, 164, 220, 224, 225, 237,
246, 291 M
GaAs-based solar cells, 139, 141 Market support programs, 13
GaAs/Ge, 137, 143, 158 Matrix, 99, 123, 186, 189, 277, 278, 283,
GaAsP, 135 284, 291
GaInAs/InP, 142 Mechanically stacked tandems, 137
GaInP/GaAs/Ge multijunction solar cells, Metal Organic Chemical Vapor Deposition
137, 143, 158, 288 (MOCVD), 137, 138, 191, 195
GaInP/GaInAs, 138, 144, 148, 151, 288 Metallurgical grade silicon, 32, 33
GaInP/GaInAs/Ge, 144, 148, 151 Mobility, 10, 24, 48, 122, 133, 134, 191
Gallium arsenide, 134 Monocrystalline, 16, 73, 117, 118
GaSb cells, 142 Monolithic dual-junction, 143
Ge–GaAs, 134 Monolithic tandem cells, 137, 142, 143, 236
German PV market, 4 Monosilane, 33–35, 38, 39, 41
Germanium substrates, 137 Morphology, 35, 119, 229, 230, 235
Graded-band heterostructure, 135 Multicrystalline, 33, 43, 48, 95, 110, 178,
Green houses, 1, 294 183
Multijunction solar cells, 114, 137, 143, 145,
158
H
Heterostructure solar cell, 133, 134, 164 N
Heterostructures, 133–139, 141–146, 176 Near infrared, 225
Index 299

Nonimaging optics, 278 Segregation, 39, 188, 225, 236

Sequestration and storage of carbon dioxide,
28
O Short circuit current, 184
OECO technology, 55 Siemens method, 3
One-junction solar cell, 50, 77, 137, 182 Silane, 33, 34, 36, 38–40, 47, 50–53, 120
Open circuit voltage, 88, 141, 147, 150, 151, Silicon (Si), 2, 31, 45–50, 54, 67, 69, 74, 75,
182, 184, 187, 188, 195 77, 83, 85–88, 95, 96, 101–103, 105–
Organic dyes, 19, 277, 292, 294 107, 110, 113, 116, 117, 120, 122,
124, 125, 130, 137, 145, 146, 149,
176, 220, 224, 237, 287, 288, 291
P Silicon solar cell, 3, 13, 18, 43–46, 49–51,
Passivation, 3, 16, 43, 44, 46–61, 66–77, 83– 53–61, 65–68, 73, 74, 77, 113, 114,
85, 99, 101, 114–116, 118–121, 125, 130, 135, 289
129, 130, 137, 178, 240 Single-junction, 138, 140, 141, 148
PEDOT, 234, 235 Si-nitride, 3, 16, 44–51, 53, 54, 58, 60, 67,
Photoconductivity, 2 69, 74, 114
Photocurrent, 146, 147, 150, 153, 158–163,
Small-size concentrator, 153
231
Small solar Home Systems (SHS), 13
Photoluminescence, 121, 185, 189, 225, 239
Solar array, 136, 138, 145, 151, 164
Photosensitivity, 2, 135
Solar battery, 16
Photovoltaic effect, 2
<>Plasma silicon nitride, 44, 49, 54 Solar cell, 2, 4–6, 13, 16–18, 20, 29, 36,
PMMA, 293 40, 43–54, 56–61, 65, 67–79, 81, 83–
pn-junction, 48, 55, 77, 180 87, 95, 96, 100–104, 106, 110, 113,
Point-contact, 67, 85 114, 116–125, 127, 128, 130, 141,
Poly-Si, 3, 5, 17, 239 145, 147, 148, 151, 164, 176–178,
Polysilicon, 31–41, 117, 124 180, 181, 183–185, 187–190, 192–
Price experience curve, 15, 18, 19, 22 194, 196–199, 203–205, 207, 260,
Purification, 32, 33, 36, 38, 39 261, 276, 277, 286, 287, 289
Solar silicon feedstock, 32
Space, 40, 53, 58, 124, 134–138, 146, 147,
Q 164, 177, 180, 181, 184, 241, 246,
Quantum Dot (QD) heterostructures, 294 248, 273
Quantum dots, 146, 277, 292, 294 Space cells, 137
Quantum efficiency, 188, 193–195, 239, Spectrum splitting, 137, 279, 290
280–283, 291, 292, 294 Stacking, 188, 282, 290
Stokes, 278, 280, 283–286
Subcell, 141–145, 158, 160, 161
R
Recombination, 43, 44, 46–48, 50, 54, 56–
58, 60, 65–67, 69, 70, 72, 74, 80,
83–88, 114–118, 134, 135, 138, 141, T
180–182, 184, 186, 188, 189, 207, Tandem, 6, 7, 18, 137, 141–144, 179, 180,
226, 232, 235, 236, 238–241 183, 207, 225, 232, 235–237, 245,
Reflective solar concentrators, 3 275
Ribbon silicon, 48 Temperature dependence, 108
Roll-to-roll manufacturing, 179 Terrestrial applications, 114, 134, 151, 164
Rooftop applications, 128 Thallium sulphide photocells, 2
Thin-film technologies, 176, 207
TiO2 , ZnO, n-type, 193
S Toluene, 231
Satellites, 13, 20, 133, 135, 152 Total internal reflection, 102, 279, 280, 283,
Screen printing, 67–69, 77, 78, 80, 85, 127 284, 287, 293
300 Index

Transport, 9, 22, 33, 52, 67, 85, 117, 121, V

150, 179, 180, 185, 221, 222, 227, Vapor-phase deposition, 40, 41
232–235, 238, 239, 241, 243, 244 Violet cell, 135
Trichlorosilane, 33, 37–39
Triple junction solar cells, 143
Tunnel junctions, 142, 143, 146, 147, 150 W
Wafer, 2, 16–19, 40, 41, 43, 46, 48–53, 56,
Two stage concentration, 287
59, 65–67, 69–71, 73–77, 83–88, 97,
99, 100, 109, 113, 115–122, 130, 145,
148, 153, 199, 207, 258, 263, 264

U X
Up-converter, 289 X-rays, 189