Home Search Collections Journals About Contact us My IOPscience

Optimal wavelength scale diffraction gratings for light trapping in solar cells

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2012 J. Opt. 14 024012

(http://iopscience.iop.org/2040-8986/14/2/024012)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 137.99.26.43
The article was downloaded on 02/08/2013 at 15:10

Please note that terms and conditions apply.

IOP PUBLISHING JOURNAL OF OPTICS A: PURE AND APPLIED OPTICS
J. Opt. 14 (2012) 024012 (9pp) doi:10.1088/2040-8978/14/2/024012

Optimal wavelength scale diffraction

gratings for light trapping in solar cells
Teck Kong Chong1 , Jonathan Wilson2 , Sudha Mokkapati2 and
Kylie R Catchpole2
1
School of Electrical and Information Engineering, University of South Australia, Adelaide,
South Australia 5001, Australia
2
Centre for Sustainable Energy Systems, College of Engineering and Computer Science,
The Australian National University, Canberra, ACT 0200, Australia

Received 25 August 2011, accepted for publication 15 November 2011

Published 12 January 2012
Online at stacks.iop.org/JOpt/14/024012

Abstract
Dielectric gratings are a promising method of achieving light trapping for thin crystalline
silicon solar cells. In this paper, we systematically examine the potential performance of thin
silicon solar cells with either silicon (Si) or titanium dioxide (TiO2 ) gratings using numerical
simulations. The square pyramid structure with silicon nitride coating provides the best light
trapping among all the symmetric structures investigated, with 89% of the expected short
circuit current density of the Lambertian case. For structures where the grating is at the rear of
the cell, we show that the light trapping provided by the square pyramid and the checkerboard
structure is almost identical. Introducing asymmetry into the grating structures can further
improve their light trapping properties. An optimized Si skewed pyramid grating on the front
surface of the solar cell results in a maximum short circuit current density, Jsc , of
33.4 mA cm−2 , which is 91% of the Jsc expected from an ideal Lambertian scatterer. An
optimized Si skewed pyramid grating on the rear performs as well as a rear Lambertian
scatterer and an optimized TiO2 grating on the rear results in 84% of the Jsc expected from an
optimized Si grating. The results show that submicron symmetric and skewed pyramids of Si
or TiO2 are a highly effective way of achieving light trapping in thin film solar cells. TiO2
structures would have the additional advantage of not increasing recombination within the
cell.

Keywords: light trapping, diffraction gratings, solar cells, photovoltaics

(Some figures may appear in colour only in the online journal)

1. Introduction Crystalline silicon is an indirect bandgap material and

therefore it has a low absorption coefficient, α, for light
The cost of photovoltaic power remains relatively high due wavelengths in the near-infrared region, with the absorption
to the amount of silicon used in conventional wafer-based length 1/α increasing from 10 µm to about 3 mm in the
wavelength range 800–1100 nm. However, there are 36%
solar cells. Thin crystalline silicon cells are a promising
of photons with energies above the bandgap of Si in this
candidate for future renewable energy applications because
wavelength range [5]. Thus, effective light confinement is
of the greatly reduced amount of pure silicon used as the essential for thin crystalline silicon solar cells.
active material, together with the stability of the film and the There are many ways to enhance the photon absorption
ease of processing of the cell, making it extremely suitable for wavelengths in the near-IR region. The standard approach
for mass production [1, 2]. Thin films are also attractive due for wafer-based cells is to use pyramid structures on a silicon
to the potential for lower bulk recombination [3, 4] and the surface which are usually several microns or more in size
potential for increasing open circuit voltage (Voc ) if a high to trap light within the solar cell [6]. Features of this size
level of optical confinement can be achieved [4]. can be understood using geometrical optics, but such large

2040-8978/12/024012+09$33.00 1 c 2012 IOP Publishing Ltd Printed in the UK & the USA
J. Opt. 14 (2012) 024012 T K Chong et al

Figure 1. Schematic diagram of checkerboard (a) and square pillar structure (b).

features are not suitable for a thin film cell which may have silicon or titanium dioxide (TiO2 ) on silicon thin films.
a total thickness of only a few microns. Diffractive structures TiO2 gratings were studied because they allow a surface
can also be used to increase the equivalent optical path texture to be applied to a finished cell, avoiding the reduction
length [7] and distributed Bragg reflectors (DBRs) can be in material quality and increase in surface recombination
used to reflect the light back into the substrates [8]. However, associated with texturing the active Si region [18]. The
only a few investigations of diffractive structures for solar cell potential short circuit current density (Jsc ) for the optimized
applications have been done to date. Feng et al investigated structures is compared against the planar cell and a cell
design optimization of a combined diffractive/DBR light with an ideal Lambertian scatterer. A Lambertian scatterer
trapping structure [7]. Morf et al have investigated groove is a perfectly randomizing surface. Absorption in Si with a
gratings as summarized in [9], while Llopis and Tobias Lambertian scatterer is calculated following the approach of
studied the effects of feature size for triangular grooves [10] Goetzberger [19]. Regardless of the material used to construct
and Abouelsaood et al have investigated the shape and the grating, the effective thickness of the cell is fixed at 3 µm,
size dependence of the anti-reflective and light trapping where the effective thickness is the height of a planar structure
performance of triangular and rectangular grooves [11]. of equivalent volume per unit area.
Meanwhile, Sai et al have investigated the light trapping effect
of submicron surface textures in crystalline solar cell [12].
2. Method
Some numerical studies have been done by Mellor et al
as summarized in [13]. Kroll et al, on the other hand,
We investigate two single-period structures, rectangular
have investigated numerically the application of dielectric
grooves and sinusoidal grooves. The rest of the structures
diffractive structures in solar cells [14]. Clearly, up until now
are biperiodic and included pillars, pyramids, grid and
only particular structures of interest have been studied and no
checkerboard structures. The grid structures are the inverse
comparison of a wide range of structures has been done.
The number and the direction of the propagation of of pillar gratings. For biperiodic structures, both periods were
diffracted orders depend on the period of the grating and the varied independently of each other. The periods, d1 and d2 ,
wavelength of the light. The grating diffraction equation is and the height, h, of the structures were varied between 0.2
and 1.2 µm for all the structures. However, the step size of the
mλ periods and height was set at either 0.05 or 0.1 µm. Figure 1
n1 sin θ1 = n2 sin θ2 = (1)
d shows the schematic diagram of some of the biperiodic
where n1 and n2 are the refractive index of air and silicon, structures. The structure on the left (a) is a checkerboard
respectively, θ1 and θ2 are the angles of propagation in air structure and the square pillar structure (b) is on the right.
and silicon, m is the diffracted order, λ is the wavelength The unit cell used to define the periods is also indicated in
and d is the grating period. For light trapping, incident the diagram (shaded gratings) by the period (d1 or d2 ) and
light needs to be coupled to diffraction orders propagating height, h.
outside the escape cone of Si. For very small periods (d  The height of a pyramid structure is constructed by
λ) higher-order diffraction modes do not exist in Si, so stacking a number of strata (staircase-like layers). As such,
small period gratings cannot provide light trapping. While different potentials Jsc that can be generated by a pyramid
there are conceptual models available for optimization of structure could be significantly affected by the number of
very large period gratings [15], and for particular structures strata due to different light diffraction properties. As a
such as rectangular grooves [16] and pillars [17], there is result, convergence tests of the number of strata for pyramid
no general conceptual model for gratings with wavelength structures were performed. It was found that the Jsc did
scale periodicity. Diffraction gratings with wavelength scale not change provided that the number of strata was 15 or
features need to be optimized using numerical studies and this greater (with reference to the maximum diffraction order, m,
optimization is the subject of this study. considered in this study). As a result this number has been
In this paper, we examine the behavior and the potential used for our study and it is valid for the ranges of height for
performance of diffractive structures constructed from either pyramid structures mentioned in this paper.

2
J. Opt. 14 (2012) 024012 T K Chong et al

Table 1. The potential short circuit current densities for all the is around 1 µm and is similar to our cell thickness. The
model structures including comparable reference structures; planar separation of the different parts of the spectrum by the grating
and Lambertian model. The effective thickness of the Si film is fixed
will also tend to increase the coherence length. Therefore it is
at 3 µm.
reasonable to include coherent interactions between the front
Name of structure Jsc (mA cm−2 ) Optimized parameters and back of the cell into account.
Reference structures In order to understand the separate effects of light
trapping and minimizing front surface reflection, the
d1 (nm) h (nm)
following cases are analyzed: (1) silicon solar cells with
Planar 13.2 N.A N.A silicon diffraction gratings without an anti-reflection (AR)
Ideal Lambertian 36.8 N.A N.A
coating, (2) silicon solar cells with silicon diffraction gratings
Uniperiodic (1D) structure with 70 nm SiN AR coating, (3) silicon solar cells with
d1 (nm) h (nm) titanium dioxide (TiO2 ) grating and (4) silicon solar cells with
titanium dioxide (TiO2 ) grating on the rear surface and a front
Rectangular groove 20.8 600 1000
Sinusoidal groove 25.8 600 800 AR coating.

Biperiodic (2D) structure

3. Results and discussion
d1 (nm) d2 (nm) h (nm)
Square pillars 26.0 600 600 600 3.1. Si gratings
Rectangular pillars 25.6 700 600 600
Square pyramid 31.6 800 800 800 The potential short circuit current densities for all the model
Rectangular pyramid 29.8 800 1200 1000
Checkerboard 26.0 800 800 1000 structures including a reference planar and Lambertian model
Square grid 23.3 600 600 1000 are illustrated in table 1.
Rectangular grid 21.7 600 400 600 From table 1, we can see that, for uniperiodic structures,
the sinusoidal groove is the most promising structure. With
a sinusoidal groove grating on the front surface of a 3 µm
The optimum combination of periods, d1 and d2 , and thick Si solar cell the Jsc increases by 95% to 25.8 mA cm−2 ,
height, h, for each of the grating structures was found by compared to the planar case.
weighted integration of each combination (d1 , d2 and h) Since we are looking at diffraction gratings on the
over the full wavelength range (400–1100 nm). The set illuminated surface of the solar cell, the increase in Jsc is
of simulations was performed twice: once with a structure partly due to the anti-reflection effect and partly due to
entirely made of silicon and once with an active planar the light trapping effect. Figure 2 shows the absorptance in
layer consisting of silicon and a grating structure of TiO2 . the solar cell as a function of wavelength for the optimal
Absorption within the TiO2 was neglected for this study. sinusoidal and rectangular grating on the front surface.
There is no back reflector used for all model structures studied Higher absorptance for the structure with the sinusoidal
in this paper, including the reference structures. grating in the long wavelength region could be either due
The maximum diffraction order m can be found from to better anti-reflection or better light trapping properties of
the condition sin θ < 1 applied to equation (1). Although m the structure. The effects of the two mechanisms cannot be
can be as large as 14, it was found that about half of the decoupled. However, in the short wavelength region, where Si
orders (m = 8–14) have very little effect on the overall results. is strongly absorbing, the higher (near-ideal) absorptance in
This is because these higher orders are only present at short the structure is entirely due to the anti-reflection property of
wavelengths where silicon is strongly absorbing and light the grating. Light trapping is not critical for this region of the
trapping is not important. However, the higher orders take spectrum as light has a very short absorption length. Since a
up a lot of memory. So, the maximum m was set to 7 for diffraction grating with a sinusoidal profile results in a gradual
computational efficiency purposes. change of refractive index compared to a rectangular groove
In order to find the power coupled into each diffracted grating, it provides better anti-reflection and hence larger Jsc
order we used rigorous coupled wave analysis (RCWA) using compared to a rectangular structure.
GD-Calc under the Matlab environment. In this work, RCWA From table 1, we can see that biperiodic structures in
is used to calculate the diffraction efficiencies for all the general perform better than single-period structures. Results
structures. The absorption can then be calculated as presented in table 1 are the average Jsc expected for TE and
" # TM polarized light. For non-symmetric biperiodic structures,
X X such as rectangular pillars, rectangular pyramids, rectangular
A=1− Rm + Tm (2) grids, etc, due to the non-symmetry of the structure periods,
m m
TE and TM polarized light result in slight performance
where Rm and Tm are the reflected and transmitted efficiencies differences. On the other hand, for a symmetric biperiodic
of the mth diffracted order. The potential short circuit current structure (d1 = d2 ) TE and TM polarized light result in
density was then calculated by integrating the absorption the same Jsc . However, from our simulations, single-period
multiplied by the photon flux per unit wavelength over the full structures are more sensitive to the polarization of incident
spectrum of 400–1100 nm. The coherence length of sunlight light and do not give equal performance for both TE and

3
J. Opt. 14 (2012) 024012 T K Chong et al

Table 2. Effect of polarization on potential Jsc of different uniperiodic and biperiodic structures.
Jsc (mA cm−2 )
Name of structure TE polarizations TM polarizations Average TE and TM
Uniperiodic (1D) structure
Rectangular groove 21.2 20.4 20.8
Sinusoidal groove 25.1 26.5 25.8
Biperiodic (2D) structure
Square pillars 26.0 26.0 26.0
Rectangular pillars 25.9 25.3 25.6
Square pyramid 31.6 31.6 31.6
Rectangular pyramid 29.7 29.9 29.8
Checkerboard 26.0 26.0 26.0
Square grid 23.3 23.3 23.3
Rectangular grid 23.3 20.1 21.7

The square pyramid produced the highest Jsc of

31.6 mA cm−2 among all the structures investigated, which
corresponds to 139% enhancement with respect to the planar
structure. Both checkerboard and square pillar structures
resulted in the same Jsc of 26.0 mA cm−2 . We will see
later, however, that the difference between these structures
and the square pyramid structure is greatly decreased when
an anti-reflection coating is included. From table 1, we also
notice that biperiodic rectangular structures result in similar
or lower Jsc compared to biperiodic square structures. So for
the rest of the paper, we will only discuss biperiodic structures
with equal periods, i.e. d1 = d2 .
From figure 4, we can see that the difference in
the performance of a square pyramid grating and the
checkerboard or pillar grating is mainly in the wavelength
Figure 2. Comparison of absorption as a function of wavelength range of 400–700 nm. While the structure with a pyramid
for sinusoidal and rectangular gratings on the front surface of a solar grating has close to ideal absorption in this wavelength region,
cell. structures with checkerboard or pillar gratings result in lower
absorption.
TM polarized light, resulting in lower average performance. To investigate this further we integrated the short
Table 2 shows the effect of TE and TM polarization on circuit current over two wavelength ranges: 400–700 and
symmetric and non-symmetric structures, with the average Jsc 700–1100 nm. Table 3 summarizes the results. In the short
as shown in table 1 for comparison. wavelength region there is 2.7 mA cm−2 difference in Jsc
The periodicity of the grating also imposes constraints on between the square pyramid structure and the square pillar
the direction of the propagation of light in air and in Si. Only structure, while there is 2.3 mA cm−2 difference in the long
plane waves whose in-plane wavevectors satisfy the relation
wavelength region. There is a larger difference between the
kE = kE0 + Gm1 ,m2 (3) checkerboard structure and the pyramid structure in the short
wavelength region. In the wavelength range 400–700 nm,
where kE0 is the in-plane wavevector of incident light and anti-reflection is the dominant effect in increasing the
Gm1 ,m2 is the grating vector are supported. The wavevectors absorption in the cell, whereas in the range 700–1100 nm,
for the diffraction modes inside Si for single-period and both anti-reflection and light trapping play a role for structures
biperiodic gratings are shown in wavevector space or k-space on the front surface. The results therefore indicate that
in figure 3. The allowed modes are separated from each other better anti-reflection is responsible for most of the difference
by 2πd . In addition, propagating modes in Si are bounded by a between the three structures. Again, as in the case of
λ , where λ is the wavelength of
circle in k-space with radius n 2π single-period structures better anti-reflection properties for
incident light in air and n is the refractive index of Si. As can the pyramid grating can be attributed to a gradual change in
be seen from the figure, with a biperiodic grating incident light refractive index from the apex to the base of the pyramids.
can be coupled to more modes inside Si than a single-period To further investigate the effect of anti-reflection on the
grating, resulting in better performance. The same result was results, a layer of 70 nm thick SiN anti-reflection coating
also found by Tobias et al [20], Yu et al [21] and Mellor et al was added to the structures to analyze the effect of AR in
[22]. enhancing the photon absorption by the cells. Table 4 shows

4
J. Opt. 14 (2012) 024012 T K Chong et al

Figure 3. (a) Schematic of a single-period grating; (b) schematic of a biperiodic grating; (c) accessible diffraction modes inside Si with a
single-period grating, represented in wavevector space and (d) accessible diffraction modes inside Si with a biperiodic grating, represented
in wavevector space.

Table 3. The short circuit current density (Jsc ) produced by the extremely important to select an optimum thickness of AR for
cells in different wavelength ranges. maximum benefit if large facet angles are used, since the effect
Jsc (mA cm−2 ) for of AR coating is significant for triangular groove or pyramid
wavelength range structures with facet angles larger than 55◦ [23].
Name of structures 400–700 nm 700–1100 nm
Planar 9.9 3.1
3.2. TiO2 gratings
Square pyramid 18.4 13.2
Checkerboard 15.4 10.9 While silicon gratings can be very effective in improving
Square pillars 15.7 10.9 the absorptance in thin film cells, they can also lead to
increased surface [24–26], emitter recombination [24] and
bulk recombination [25, 26]. An alternative is to use a
the results. Generally an improvement of about 2.5 mA cm−2 planar silicon cell with a grating of another material on
is achievable for pillar-type structures after a thin layer of SiN the surface, avoiding this increased recombination. In the
is added. This represents at least a 10% enhancement in short second part of this study, we investigate the enhancement in
circuit current density current achievable with titanium dioxide gratings on silicon
For checkerboard and pillar gratings, there is a substantial substrates. We chose to study TiO2 gratings as they can
increase in the Jsc with the anti-reflection coating. For the be fabricated using nano-imprinting, a cheap and large area
square pyramid structure in our simulations, the reduction patterning technique, and may also be able to provide surface
in reflection is not significant. This is because multiple passivation for Si [18]. The absorption within the TiO2 was
reflections already occur due to the small facet angle, reducing neglected for this study.
the overall reflection and making the effect of the AR coating In general, Jsc produced by TiO2 gratings on a silicon
less important. As a result, the improvement in Jsc for pyramid substrate exhibited the same trends as silicon gratings on a
structure with AR coating was not large. Nevertheless, it is silicon substrate. Interestingly, although the refractive index of

Table 4. The Jsc results and optimized parameters for the best three structures.
Optimized
parameters
Name of structures Type of texturing Jsc (mA cm−2 ) d1 (nm) h (nm)
Square pyramid structure Front surface 31.6 800 800
Front surface + AR 32.7 800 1000
Checkerboard Front surface 26.0 800 1000
Front surface + AR 29.1 800 1000
Square pillars Front surface 26.0 600 600
Front surface + AR 28.6 600 200

5
J. Opt. 14 (2012) 024012 T K Chong et al

Table 5. The potential short circuit current density for the model structures constructed from TiO2 grating on silicon substrate with
effective thickness of the Si substrate fixed at 3 µm.
Name of structure Jsc (mA cm−2 ) Optimized parameters
Reference structure
d1 (nm) h (nm)
Planar 13.2 N.A N.A
Ideal Lambertian 36.8 N.A N.A
Uniperiodic (1D) structure
d1 (nm) h (nm)
Rectangular groove 21.3 400 200
Sinusoidal groove 23.2 800 600
Biperiodic (2D) structure
d1 (nm) d2 (nm) h (nm)
Square pillars 24.4 600 600 400
Rectangular pillars 24.3 700 600 400
Square pyramid 28.2 750 750 900
Rectangular pyramid 27.8 800 1200 800
Checkerboard 25.4 800 800 400
Square grid 22.3 600 600 200
Rectangular grid 21.3 600 400 200

Figure 4. Absorption of the best three Si gratings on an Si substrate Figure 5. Absorption in a 3 µm thick planar Si layer and Si layer
compared against absorption in planar Si and Si with Lambertian with a Lambertian scatterer or TiO2 gratings on the rear. The front
scatterer. surface of Si is coated with 70 nm SiN for anti-reflection.

TiO2 is significantly lower than that of silicon, the Jsc values obtained for the case of a rear surface texture with front SiN
predicted are only slightly lower. A sinusoidal grating remains AR coating (70 nm).
the best uniperiodic structure with 23.2 mA cm−2 compared The potential short circuit current, Jsc for the three
structures is shown in table 6. The table also illustrates
to 13.2 mA cm−2 for a planar structure. For biperiodic
the optimum period and height for each structure. We can
structures, a square pyramid remains the best structure with
see from table 6 that, for the cases with the texture on
Jsc of 28.2 mA cm−2 followed by a checkerboard with Jsc of
the rear and a front AR coating, the predicted Jsc for the
25.4 mA cm−2 and pillars with Jsc of 24.3 mA cm−2 . The
checkerboard (26.1 mA cm−2 ) and square pyramid structure
complete results, together with the optimum height, h, and (26.5 mA cm−2 ) is very similar. Since the anti-reflection effect
period, d1 , are given in table 5. is the same in each case, this indicates, for structures on the
The best three models (square pyramid, checkerboard and rear of the cell, the light trapping provided by the square
square pillars) were selected for further analysis. In order to pyramid structure and the checkerboard structure is almost
investigate the potential of these gratings for different types identical.
of solar cells, a rear surface textured cell with and without a For diffraction gratings with wavelength scale periodicity,
front AR coating was constructed. Figure 5 shows the result it has been proposed that introducing asymmetry into the

6
J. Opt. 14 (2012) 024012 T K Chong et al

Figure 6. Side view and top view for the square and skewed pyramid structures.

Table 6. Summary of Jsc for front and rear TiO2 surface textures, including rear surface textures with AR coating on the front side of the
cell. All the cells have 3 µm thickness of Si.
Optimized parameters
Name of structures Type of texturing Jsc (mA cm−2 ) d1 (nm) d2 (nm) h (nm)
Pyramid structure Front surface 28.2 750 750 900
Rear surface 16.9 850 850 600
Rear surface + AR 26.5 850 850 600
Checkerboard Front surface 25.4 800 800 400
Rear surface 17.2 800 800 400
Rear surface + AR 26.1 800 800 400
Square pillars Front surface 24.3 600 600 400
Rear surface 16.7 600 600 400
Rear surface + AR 23.6 900 900 400

Table 7. A summary of Jsc for symmetric and skewed pyramid (Si or TiO2 ) gratings on the front and rear of a 3 µm thick Si solar cell.
Symmetric structures Asymmetric structures
Optimized
Optimized parameters parameters
Type Jsc (mA cm−2 ) d1 = d2 (nm) h (nm) Type Jsc (mA cm−2 ) d (nm) h (nm)
Si pyramids
Pyramid front 31.6 800 800 Skewed pyramid front 33.4 750 750
Pyramid rear 21.5 850 350 Skewed pyramid rear 22.1 650 950
TiO2 pyramids
Pyramid front 28.2 750 900 Skewed pyramid front 28.9 700 750
Pyramid rear 16.9 850 600 Skewed pyramid rear 18.5 700 1000

grating structure can further improve its light trapping or a symmetric Si pyramid grating or a skewed Si pyramid
capabilities [21]. Having determined that pyramid gratings grating on the front surface. Higher absorptance in the short
give the best performance of the symmetric structures studied, wavelength region for the cell with a Lambertian scatterer
we investigated the effect of introducing asymmetry into the shows that the difference between the performance of the
pyramid gratings to further improve the Jsc of the 3 µm thick cell with a Lambertian scatterer and a cell with an Si grating
solar cells. Figure 6 shows the side and top views of the is mainly due to better anti-reflection from the Lambertian
skewed pyramid structures studied. surface. For TiO2 gratings on the front surface of the solar
Table 7 summarizes the Jsc expected from a 3 µm thick cell, the skewed pyramids result in 2.4% enhancement in Jsc
Si solar cell with symmetric or skewed Si or TiO2 pyramid compared to the symmetric pyramid structures, from 28.2 to
gratings on the front or on the rear of the cell. For Si gratings 28.9 mA cm−2 . With optimized skewed TiO2 gratings, the
on the front surface of the solar cell, the skewed pyramids maximum expected Jsc is 79% of the Jsc expected from an
result in a maximum Jsc of 33.4 mA cm−2 , which represents ideal Lambertian scatterer.
a 7% increase from Jsc of 31.6 mA cm−2 expected from Figure 8 shows the absorptance in 3 µm thick Si with
the symmetric pyramid grating without AR coating. With a Lambertian scatterer, a symmetric Si pyramid grating or a
optimized Si skewed pyramid gratings on the front side of skewed Si pyramid grating on the rear surface. We can see
the solar cell, the maximum expected Jsc is 86% of the Jsc that Si gratings on the rear of the solar cell perform as well as
expected from an ideal Lambertian scatterer. Figure 7 shows a Lambertian scatterer on the rear. The maximum Jsc expected
the absorptance in 3 µm thick Si with a Lambertian scatterer from a 3 µm thick Si solar cell with an ideal Lambertian

7
J. Opt. 14 (2012) 024012 T K Chong et al

gratings on the rear of the solar cell the Jsc enhancement is

only due to light trapping.
Introducing asymmetry into the square pyramid structure
further improves the performance of the biperiodic gratings.
For skewed Si gratings on the front surface of the solar cell, a
maximum Jsc of 33.4 mA cm−2 is expected, which is 91%
of the current density expected with an ideal Lambertian
scatterer. With an optimized TiO2 skewed pyramid grating on
the front surface, the maximum Jsc obtainable is 79% of that
expected from an ideal Lambertian scatterer. For gratings on
the rear surface of the solar cell, Si skewed pyramid gratings
perform as well as a Lambertian surface and Jsc from the
skewed TiO2 pyramid grating is 84% of the maximum Jsc
obtained using a skewed Si grating.

Figure 7. Absorptance as a function of wavelength in a 3 µm thick Acknowledgments

Si solar cell with a Lambertian scatterer, a symmetric Si pyramid
grating or a skewed Si pyramid grating on the front surface. One of the authors (TKC) would like to thank CSES of the
Australian National University for the opportunity given to
explore this topic. We would like to thank the Australian
Research Council and the Australian Solar Institute for
financial support.

References
[1] Hamakawa Y 2004 Thin-Film Solar Cells: Next Generation
Photovoltaics and Its Applications (Berlin: Springer)
chapter 2, pp 15–39
[2] Nelson J 2003 The Physics of Solar Cells (London: Imperial
College Press) chapter 8, pp 211–51
[3] Zeng L, Yi Y, Hong C, Liu J, Feng N, Duan X,
Kimerling L C and Alamariu B A 2006 Efficiency
enhancement in Si solar cells by textured photonic crystal
back reflector Appl. Phys. Lett. 89 111111
[4] Tiedje T, Yablonovitch E, Cody G D and Brooks B G 1984
Limiting efficiency of silicon solar cells IEEE Trans.
Electron Devices 31 711–6
Figure 8. Absorptance as a function of wavelength in a 3 µm thick [5] Bermel P, Luo C, Zeng L, Kimerling L C and
Si solar cell with a Lambertian scatterer, a symmetric Si pyramid Joannopoulos J D 2007 Improving thin-film crystalline
grating or a skewed Si pyramid grating on the rear surface. silicon solar cell efficiencies with photonic crystals Opt.
Express 15 16986–7000
[6] Campbell P and Green M A 1987 Light trapping properties of
scatterer on the rear is 21.1 mA cm−2 . For gratings on the rear pyramidally textured surfaces J. Appl. Phys. 62 243
[7] Ning-Ning F, Michel J, Lirong Z, Jifeng L, Ching-Yin H,
of the solar cell, the Si and TiO2 skewed pyramids result in 3%
Kimerling L C and Xiaoman D 2007 Design of highly
and 9.5% enhancement, respectively, in Jsc compared to the efficient light-trapping structures for thin-film crystalline
symmetric pyramid structures. For optimized skewed pyramid silicon solar cells IEEE Trans. Electron Devices
gratings on the rear surface of the solar cell, the TiO2 gratings 54 1926–33
result in 84% of the current expected from the Si gratings. [8] Johnson D C et al 2005 Advances in Bragg stack quantum
well solar cells Sol. Energy Mater. Sol. Cells 87 169–79
[9] Morf R H, Kiess H and Heine C 1997 Diffractive optics for
4. Conclusions solar cells Diffractive Optics for Industrial and Commercial
Applications ed J Turunen and F Wyrowski (Berlin:
We have presented a numerical study of Si and TiO2 Akademie) pp 361–89
[10] Llopis F and Tobı́as I 2005 Influence of texture feature size on
diffraction gratings with wavelength scale periodicity for light
the optical performance of silicon solar cells Prog.
trapping in 3 µm thick Si solar cells. Si gratings result in Photovolt., Res. Appl. 13 27–36
better performance compared to TiO2 gratings because of [11] Abouelsaood A A, El-Naggar S A and Ghannam M Y 2002
higher refractive index contrast. For both materials, biperiodic Shape and size dependence of the anti-reflective and
gratings result in higher Jsc from the solar cell compared light-trapping action of periodic grooves Prog. Photovolt.,
to single-period gratings. Of all the structures investigated, Res. Appl. 10 513–26
[12] Sai H, Kanamori Y, Arafune K, Ohshita Y and
pyramid structures resulted in the highest Jsc . For gratings on Yamaguchi M 2007 Light trapping effect of submicron
the illuminated surface of the solar cell, both anti-reflection surface textures in crystalline Si solar cells Prog. Photovolt.
and light trapping effects lead to increased Jsc , while for 15 415–23

8
J. Opt. 14 (2012) 024012 T K Chong et al

[13] Mellor A, Tobı́as I, Martı́ A and Luque A 2011 A numerical [20] Tobias I, Luque A and Marti A 2008 Light intensity
study of Bi-periodic binary diffraction gratings for solar cell enhancement by diffracting structures in solar cells J. Appl.
applications Sol. Energy Mater. Sol. Cells 95 3527–35 Phys. 104 134502
[14] Kroll M, Fahr S, Helgert C, Rockstuhl L, Lederer F and [21] Yu Z, Raman A and Fan S 2010 Fundamental limit of light
Pertsch T 2008 Employing dielectric diffractive structures trapping in grating structures Opt. Express 18 A366–80
in solar cells—a numerical study Phys. Status Solidi a [22] Mellor A, Tobias I, Marti A, Mendes M and Luque A 2011
205 2777–95 Upper limits to absorption enhancement in thick solar cells
[15] Mokkapati S, Beck F J and Catchpole K R 2011 Analytical using diffraction grating Prog. Photovolt., Res. Appl.
approach for design of blazed dielectric gratings for light 19 676–8
trapping in solar cells J. Phys. D: Appl. Phys. 44 055103 [23] Gangopadhyay U, Dutta S K and Saha H 2008 Recent
[16] Catchpole K R 2007 A conceptual model of the diffuse research trends in texturization and light trapping in silicon
solar cells Solar Cell Research Progress ed J A Carson
transmittance of lamellar diffraction gratings on solar cells
(New York: Nova Science Publisher Inc) pp 53–129
J. Appl. Phys. 102 013102 [24] Basore P A 1990 Numerical modeling of textured silicon solar
[17] Catchpole K R and Green M A 2007 A conceptual model of cells using PC-1D IEEE Trans. Electron Devices 37 337–43
light coupling by pillar diffraction gratings J. Appl. Phys. [25] McCann M J, Catchpole K R, Weber K J and Blakers A W
101 063105 2001 A review of thin-film crystalline silicon for solar cell
[18] Barbé J, Thomson A F, Wang E-C, McIntosh K and applications. Part 1: native substrates Sol. Energy Mater.
Catchpole K 2012 Nanoimprinted TiO2 sol–gel passivating Sol. Cells 68 135–72
diffraction gratings for solar cell applications Prog. [26] Rockstuhl C, Fahr S, Lederer F, Haug F-J, Soderstrom T,
Photovolt., Res. Appl. at press Nicolay S, Despeisse M and Ballif C 2011 Light absorption
[19] Goetzberger A 1981 Optical confinement in thin Si solar cells in textured thin film silicon solar cells: a simple scalar
by diffuse back reflectors 15th Photovoltaic Specialists scattering approach versus rigorous simulation Appl. Phys.
Conf. (Kissimmee, FL) pp 867–70 Lett. 98 051102