Physics of Solar Cells

Peter Würfel
© 2005 WILEY-VCH Verlag GmbH & Co.

4 Conversion of Thermal Radiation into Chemical Energy

In Chapter 2 we have seen what a Carnot engine can do, which operates with the heat gained
from a black or a monochromatic absorber of solar radiation. The absorber was characterized
by a high temperature TA and a chemical potential of its electron–hole excitations of µeh = 0.
A solar cell remains at about T0 = 300 K and if it had µeh = 0 as well, it would not be able to
convert heat into an entropy-free form of energy. But as was shown in Chapter 3, electron–hole
pairs in an illuminated semiconductor have a chemical potential of µeh 6= 0, which means that
conversion of solar heat into chemical energy has already taken place in the semiconductor.
In order to examine this conversion in more detail, we will follow the generation and ther-
malization of electrons and holes step by step. These steps are shown in Figure 4.1. In the first
step, photons absorbed from fully concentrated solar radiation establish an energy distribution
of the electron–hole pairs, which is identical to the energy distribution of the absorbed photons
reflecting the high temperature of the sun, TS . Radiative recombination of electron–hole pairs
results in the emission of photons with the same spectrum and filling (for maximal concen-
tration) the same solid angle as the absorbed photons. This will not be changed, when, in a
next step, scattering of electrons and holes between each other is allowed (the interaction with
the lattice vibrations is still switched off), since scattering preserves the number of electrons
and holes and the average energy per electron–hole pair. The temperature of the electron–hole
pairs defined by their energy distribution is still Teh = TS . In this situation the emitted photons
and the electron–hole pairs in the semiconductor are in thermal and chemical equilibrium with
the solar radiation and have a chemical potential of µγ = µeh = 0.

Figure 4.1: Electrons and holes generated by photons with energy h̄ω lose energy by thermalization,
which produces chemical energy per electron–hole pair µeh .

Physics of Solar Cells: From Principles to New Concepts. Peter Würfel

Copyright 2005
c WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 3-527-40428-7
86 4 Conversion of Thermal Radiation into Chemical Energy

When concentration is not at maximum and emitted photons cover a larger solid angle than
the absorbed photons, the emitted photon current density per solid angle is smaller than the
absorbed, but still has the same spectrum. In this situation Teh = TS because of the spectrum,
but µγ = µeh < 0.
Returning to maximum concentration, the electrons and holes are in thermal and chemical
equilibrium with the sun with no splitting of the Fermi energies. In a rough estimate, the
average energy of the electron–hole pairs is then εG + 3kTS . From Eq. (3.29) it follows that,
at Teh = TS
 
NC NV
µeh (TS ) = εFC − εFV = εG − kTS ln =0,
ne nh
from which kTS ln (NC NV /ne nh ) = εG .
In the next step, the interaction with the lattice vibrations is switched on. This leads to
a cooling of the electrons and holes, until after about 10−12 s the lattice temperature T0 is
reached, while the concentrations of electrons and holes remain constant. For Teh = T0
 
NC NV
µeh (T0 ) = εFC − εFV = εG − kT0 ln .
ne nh
Because of the constant concentrations of electrons and holes and neglecting the temperature
dependence of the effective densities of states NC and NV , this finally gives us an estimate of
the chemical energy per electron–hole pair produced by the cooling process
µeh (T0 ) = εFC − εFV = εG (1 − T0 /TS ) .
This relation looks like a Carnot efficiency, but it is different. For the efficiency we have to
compare εFC − εFV not with εG but with the energy εG + 3kTS per electron–hole pair before
the cooling. We can then see a large energy loss resulting from thermalization, which can be
attributed to the production of entropy by the lattice vibrations. The cooling of electrons and
holes at constant concentrations is therefore far from an ideal process for the production of
chemical energy, if we start from a broad energy distribution. Although in this consideration,
no electron–hole pairs have been extracted from the semiconductor absorber, an equilibrium
between the sun and the semiconductor does not exist, if the semiconductor is at a different
temperature. A non-zero chemical potential of its electron–hole pairs reduces the entropy
production compared with a zero chemical potential absorber, as it enables the absorber to emit
as many photons as it absorbs, but it does not fully compensate the temperature difference,
since the emitted spectrum is different from the absorbed spectrum, and less energy is emitted
than is absorbed.
Figure 4.2 shows the exact result of the thermalization of electron–hole pairs generated
by non-concentrated sunlight and by 300 K background radiation under open circuit condi-
tions, when there is only radiative recombination. Applying the generalized Planck law from
Eq. (3.101), the comparison of the average energy hh̄ωabs i of the absorbed photons which
is invested for the generation of each electron–hole pair with the average energy hh̄ωemit i of
the emitted photons, shows the energy loss by thermalization which, however, results in the
chemical energy per electron–hole pair µeh . This chemical energy can, e. g., initiate further
chemical reactions as in photosynthesis, where a long-term storage of energy is achieved. In
a solar cell, the chemical energy is an intermediate product, from which electrical energy can
be obtained by a further step.
 87

Figure 4.2: Average energy of absorbed photons hh̄ωabs i from a non-concentrated AM0-spectrum and
from the 300 K background radiation, the average energy of emitted photons hh̄ωemit i and the chemical
potential of electron–hole pairs µeh of a semiconductor as a function of its band gap εG for radiative
recombination under open circuit conditions, resulting from the generalized Planck radiation law. The
difference between hh̄ωabs i and hh̄ωemit i is lost by thermalization of the electron–hole pairs.

The chemical energy per electron–hole pair is the sum of the electrochemical potentials of
the electrons and the holes, and therefore

ηe + ηh = µe + µh = εFC − εFV .

According to Eq. (3.29) this gives us

 
ne nh
µe + µh = εFC − εFV = kT ln . (4.1)
n2i

If we again separate the particle densities into dark-state and additional densities, so that
ne = ne,0 +∆ne and consider the case of a p-type semiconductor with nh ≈ n0h (weak excitation)
we then find
 
∆ne
µe + µh = kT ln 1 + 0 . (4.2)
ne
This means that for weak excitation the chemical energy per electron–hole pair is the result of
only the change in the chemical potential of the minority charge carriers (here the electrons)
or the change in their Fermi energy.
88 4 Conversion of Thermal Radiation into Chemical Energy

4.1 Maximum efficiency for the production of chemical

energy
Is it imaginable that a material at room temperature is in equilibrium with the sun, in which
it emits towards the sun as many photons, as much energy and as much entropy as it absorbs
from the sun? As we have seen in the last paragraph, the main problem is the energy loss
accompanying the entropy production during the thermalization process. Figure 4.3 shows
that an energy loss cannot occur, if the states available for electrons and holes are confined to
a narrow energy interval. Such a material could only absorb almost monochromatic radiation.
Alternatively, if in a broad band semiconductor the electrons and holes are generated by almost
monochromatic radiation, and only into those states in which they are after the thermalization,
can an energy loss be avoided. For maximum efficiency, we therefore make the following
idealized assumptions:
1. Only radiative recombination takes place.

2. There is no extraction of electrons and holes.

3. There are no thermalization losses. Absorption and emission are monochromatic, with
an energy of h̄ω = εG and an absorptivity of a(h̄ω = εG ) = 1 over an interval dh̄ω. We
can consider this condition fulfilled by a filter enclosing the semiconductor, which is
transparent only for h̄ω = εG , reflecting all other photons.

4. The rate of generation is at maximum. This condition is achieved by maximum concen-

tration, where the solid angles for absorption Ωabs and emission Ωemit , are equal.
From the first two conditions it follows that in the steady state the emitted photon current is
equal to the absorbed photon current. Referring to Eq. (2.34) and Eq. (3.101) this means that
Ωemit ε2G dh̄ω Ωabs ε2 dh̄ω
d jγ, emit =   = 3 3 2 G  = d jγ,abs . (4.3)
4π3 h3 c2 εG − µeh 4π h c εG
exp −1 exp −1
kT0 kTS

Figure 4.3: Electrons and holes generated in narrow energy ranges by monochromatic radiation have
the same energy distribution after thermalization as before.
4.1 Maximum efficiency for the production of chemical energy 89

With condition 4, Ωemit = Ωabs , under which the semiconductor would reach the temperature
of the sun TS unless we hold its temperature constant at T = T0 , we find the chemical energy
per electron–hole pair
 
T0
µeh = µe + µh = εG 1 − . (4.4)
TS
The efficiency for the conversion of solar heat into chemical energy is
µe + µh T0
η= = 1− (4.5)
εG TS
This is the Carnot efficiency, a limiting value which is obtained for the conversion of heat into
an entropy-free form of energy when the conversion process is reversible, i. e., takes place
without the production of entropy. We can now see that an ideal semiconductor, which only
has radiative recombination, represents an ideal converter of heat into chemical energy in
monochromatic operation.
The state defined by Eq. (4.4) is a state of equilibrium between the semiconductor and the
sun, in which not only are the absorbed and emitted photon currents equal, as expressed by
Eq. (4.3), but also the energy currents which follow from the photon currents by multiplication
with the photon energy which for absorption and emission have the same value h̄ω = εG . For
the absorbed and emitted radiation, the entropy per photon state is1
 
σγ = k (1 + fγ ) ln(1 + fγ ) − fγ ln( fγ ) , (4.6)
where fγ is the Bose–Einstein distribution function
1
fγ = 
h̄ω−µγ
 (4.7)
exp kT −1

The exponent in the distribution function has the same value for the sun and the semiconduc-
tor according to Eqs. (4.3) and (4.4) with µe + µh = µγ because of the chemical equilibrium
between the electron–hole pairs and the photons by frequent absorption and emission in a
material with absorptivity a(h̄ω = εG ) = 1. In addition, absorption and emission involve the
same number of photon states because of Ωabs = Ωemit . Under these conditions, the entropy
absorbed with the photons from the sun is equal to the entropy emitted with the photons to-
wards the sun. This all shows that an equilibrium is possible between the sun and a material
at room temperature, if the absorption and emission of photons is limited to the same narrow
energy range. This equilibrium requires the right combination of temperature T and chemical
potential µ as expressed by Eq. (4.4), which is why we may call it a thermo-chemical equilib-
rium. The chemical energy produced by reversible, non-dissipative cooling of electrons and
holes is carried away by the emitted photons. The emitted photons, known as luminescent
radiation from LEDs, look the same as the solar photons. For monochromatic radiation, there
is no way to decide whether they have a high temperature T and a small chemical potential µγ ,
or a low temperature and a large chemical potential.
The process in which all the chemical energy is emitted with the photons is in fact of
no interest to us. We are instead interested in how much chemical energy can be harvested
1 C. Kittel, Elementary Statistical Physics, Wiley 1958.
90 4 Conversion of Thermal Radiation into Chemical Energy

with the electron–hole pairs, if we knew how to extract them from the absorber. From the
continuity equation for the electrons under steady-state conditions,
∂ne
= Ge − Re − div d je = 0 ,
∂t
we see that
div d je = Ge − Re (4.8)
gives just the rate with which electrons can be extracted from a volume element, since more
flow out of this element than into it when Ge > Re . We are considering a differentially small
electron current d je , since it results from absorption and emission in a differentially small
photon energy interval dh̄ω.
Integrating Eq. (4.8) over the volume of the semiconductor gives the total extracted elec-
tron current d je
Z Z
d je = Ge dx − Re dx .

If we again restrict ourselves to radiative recombination, the extracted electron current density
is given by the difference between the absorbed and emitted photon current densities, which
already includes photon recycling by re-absorption. Thus,
d je = d jγ, abs − d jγ, emit . (4.9)
The same relationship is true for the holes. The current of extracted electron–hole pairs is
d jeh = d je = d jh .
For a spatially constant chemical potential of the electron–hole pairs µeh = const, jeh
follows from the relations in Eq. (4.3) for the absorbed and emitted photon currents.
In an approximation to the generalized Planck law, when µeh < εG − 3kT allows us to
neglect the “−1” in the denominator of the Planck law, we find according to Eq. (4.1)
 
0 ne nh 0 µe + µh
d je = d jγ,abs − d jγ 2 = d jγ,abs − d jγ exp , (4.10)
ni kT
where jγ0 is the photon current density absorbed and emitted in equilibrium with the 300 K
background radiation.
The condition µe + µh = constant is fulfilled, for example, when electrons and holes are
produced uniformly throughout the volume as a result of weak absorption or when they are
uniformly distributed throughout the volume as a result of large diffusion constants and life-
times.
For a given current of absorbed photons d jγ,abs , the current d jeh of extracted electrons
and holes is shown in Figure 4.4 as a function of the chemical energy extracted along with
the electron–hole current. We see that the extracted electron–hole current is nearly equal to
the absorbed photon current for small values of µeh = µe + µh , where almost no photons are
emitted. The emitted photon current d jγ,emit rises exponentially with µe + µh , until the open
circuit situation is reached at µeh,oc , where absorbed and emitted photon currents are equal and
no electron–hole pairs are extracted. The current of chemical energy extracted along with the
electron–hole pairs is
d jµ = d jeh (µe + µh ) .
4.1 Maximum efficiency for the production of chemical energy 91

Figure 4.4: Current d jγ,emit of emitted photons as a function of the chemical energy µeh = µe + µh
of the electron–hole pairs. The dark rectangle is the current of chemical energy extracted along with
the current d jeh = d jγ,abs − d jγ,emit of extracted electron–hole pairs. It is produced from the current of
absorbed energy represented by the grey rectangle.

Its maximum value is the largest rectangle (dark grey in Figure 4.4), which can be fitted be-
tween the absorbed and emitted photon currents. The absorbed energy current in the interval
dh̄ω is d jγabs h̄ω and is given by the lightly shaded rectangle with the broken border in Fig-
ure 4.4. The efficiency with which chemical energy is obtained from the absorbed radiation is
finally given by the ratio of the dark and lightly shaded rectangles.
Figure 4.5 shows the efficiency η with which chemical energy is extracted as a function of
the chemical potential of the electron–hole pairs. It starts with zero at µeh = 0, when electrons
and holes are so effectively extracted that none can accumulate, then it rises linearly and
reaches a maximum at the point of maximum power (mp), after which it drops sharply down
to zero at the open circuit situation, where all electron–hole pairs recombine to emit photons.

Figure 4.5: Efficiency η with which chemical energy is extracted as a function of the chemical potential
µeh of the electron–hole pairs of monochromatic absorbers for different photon energies.
92 4 Conversion of Thermal Radiation into Chemical Energy

Figure 4.6: Monochromatic efficiency ηmono (εG ) for obtaining chemical energy µe + µh as a function of
the photon energy h̄ω of fully concentrated (full line) and non-concentrated (broken line) monochromatic
solar radiation.

In order to find the maximum efficiency with which chemical energy can be obtained from
fully concentrated solar radiation, we proceed as follows. From the generalized Planck law in
Eq. (3.101) the absorbed and emitted photon currents are found as a function of the monochro-
matic photon energy. For each photon energy the chemical potential µeh in the emitted photon
current is varied until the value µeh,mp is found for which the current of extracted chemical
energy is maximal. Dividing this chemical energy current by the absorbed monochromatic
energy current gives the efficiency ηmono (εG ) as a function of the band gap.
The result of this calculation is shown in Figure 4.6 for non-concentrated radiation and for
fully concentrated radiation. The efficiencies are rather large and rise with the band gap εG ,
demonstrating the high value of solar energy. In order to make proper use of the total solar
spectrum, very (infinitely) many monochromatic absorbers have to be employed. The overall
efficiency then follows from an integration of the efficiency ηmono (εG ) over the spectrum after
weighting each interval dh̄ω with its share of the absorbed energy current.
Z ∞
ηmono (εG ) h̄ω d jγ,abs
0
η= Z ∞
h̄ω d jγ,abs
0
The total efficiency is 86% for fully concentrated radiation from a 5800 K sun and 67% for
non-concentrated radiation from a solid angle of 6.8 × 10−5 .2
Although we were able to calculate with what efficiency chemical energy can be obtained
from an illuminated semiconductor by an extraction of some of the generated electron–hole
pairs, we still have no idea, how a device should look, which allows this extraction of electron–
hole pairs and their energy. This is the subject of the next chapter.

2 A. deVos, Proceedings 5. E.C. Photovoltaic Solar Energy Conf., Athens 1983, p. 186.