Home Search Collections Journals About Contact us My IOPscience

Theoretical limits of thermophotovoltaic solar energy conversion

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

2003 Semicond. Sci. Technol. 18 S151

(http://iopscience.iop.org/0268-1242/18/5/303)

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

Download details:
IP Address: 137.99.31.134
The article was downloaded on 23/04/2013 at 14:42

Please note that terms and conditions apply.

INSTITUTE OF PHYSICS PUBLISHING SEMICONDUCTOR SCIENCE AND TECHNOLOGY
Semicond. Sci. Technol. 18 (2003) S151–S157 PII: S0268-1242(03)59384-1

Theoretical limits of thermophotovoltaic

solar energy conversion
Nils-Peter Harder1 and Peter Würfel2
1
Centre for Third Generation Photovoltaics, University of New South Wales, Sydney,
2052 NSW, Australia
2
Institut für angewandte Physik, Universität Karlsruhe, D-76128 Karlsruhe, Germany

Received 4 November 2002

Published 4 April 2003
Online at stacks.iop.org/SST/18/S151

Abstract
Theoretical efficiencies are derived in a detailed balance calculation for
thermophotovoltaic solar energy conversion, where solar radiation is
absorbed by an intermediate absorber, which emits radiation inside an
evacuated housing towards a solar cell. For ideal components with no
optical losses and only radiative recombination in the solar cell, maximal
efficiencies are found of 85% for full concentration of the incident sunlight
on a black absorber, and of 54% for no concentration and a selective
absorber absorbing only for h̄ω > 0.92 eV. This is considerably larger than
the efficiency for directly illuminated solar cells with also only radiative
recombination, the Shockley–Queisser limit, which is 41% for full
concentration and 30% for no concentration.
In order to approach efficiency limits for real TPV systems, several
non-idealities have been introduced: (a) realistic assumptions about the
geometry of the intermediate absorber, (b) optical losses of 5% for photons
with energy below the band gap of the solar cell and (c) non-radiative
recombination in the solar cell of the same amount as radiative
recombination. This reduces the efficiency for non-concentrated sunlight to
only 32.8%, but for very high concentrations of 10 000 and above suitable
absorber geometries still seem to allow efficiencies close to 60%.

1. Introduction caused by recombination, since even radiative recombination

is a loss.
In thermophotovoltaic (TPV) solar energy conversion the solar TPV conversion is very attractive, because all of these
cell is not directly exposed to solar radiation. Instead, an losses can be avoided [2]. The loss due to non-adequate photon
intermediate absorber is heated by absorbing solar radiation energies can be avoided by illuminating the solar cell only with
and its emitted radiation is converted into electrical energy photons with energy equal to the band gap of the solar cell.
by a solar cell. Figure 1 shows the principal geometry. This This can be achieved in several ways, e.g. by making the
arrangement allows reducing losses occurring in conventional emitter selective, by placing a filter between emitter and solar
solar energy conversion in solar cells. cell or, to a lesser degree, by using a mirror behind the solar cell
Losses in conventional solar energy conversion in solar which reflects all photons with energy below the band gap for
cells consist of the non-absorbed radiation from photons with which the solar cell is transparent. Photons, which are not used
energy smaller than the band gap of the solar cell and from by the solar cell, are not lost, but are reflected back onto the
thermalization of electron–hole pairs generated by photons emitter and help to keep the intermediate absorber/emitter at a
with energy larger than the band gap. This is the reason why high temperature. Even the loss due to the photons emitted by
the efficiency for the conversion of AM0 solar radiation cannot the solar cell by radiative recombination is avoided, since the
exceed 30%, even if recombination losses have their lowest emitter also reabsorbs these photons and recycles their energy.
possible value (only radiative recombination), as was shown by The solar cell can, therefore, be operated close to open-circuit
Shockley and Queisser [1]. In addition, a directly illuminated condition, where the voltage is larger than at the maximum
solar cell must be operated at its maximum power point, i.e. at power point of the solar cell without recycling of photons
voltages below the open-circuit voltage, for minimizing losses emitted from radiative recombination.

0268-1242/03/050151+07$30.00 © 2003 IOP Publishing Ltd Printed in the UK S151

N-P Harder and P Würfel

absorber emitter solar cell + +

filter εe n p
pn
x
IE,S TA Iy,E IE,E εC
earth ε F,left
sun εFC
Ts T0
IE,A Iy,C IE,C
εFV
ε F,right
εV
IQ
Figure 2. Different Fermi energies in an illuminated solar cell,
Figure 1. In TPV conversion an intermediate absorber is heated by εFC for the conduction band (ε e > εC) and εFV for the valence band
absorbing solar radiation. The solar cell is illuminated by radiation (εe < εV) lead to a voltage V at the contacts with eV = ε F,right − εF,left.
from the emitter transmitted by a filter. In the analysis, the balance
of energy currents IE, photon currents Iγ from the sun (S), the The difference of the Fermi energies εFC − ε FV is the chemical
absorber (A), the emitter (E), and the solar cell (C) and the charge
current IQ from the solar cell must be considered.
energy per electron–hole pair.
In the next step chemical energy is to be transformed
into electrical energy. In order to produce an electrical
The only loss occurring in an ideal TPV process is caused charge current jQ, the electrons and the holes have to leave
by the emission of the intermediate absorber into the outside the solar cell through different ports or electrodes. As for
hemisphere. the separation of a gas mixture, semi-permeable membranes
Limiting efficiencies for solar TPV energy conversion would be helpful, one which conducts electrons and blocks
for concentrated and non-concentrated sunlight have been holes in front of one (the negative) electrode of the solar cell
determined to be 85.4% and 54%, respectively [3]. In this and one which conducts holes and blocks electrons in front of
paper, we investigate how the introduction of several non- the other (the positive) electrode. An n-type semiconductor
idealities such as (a) realistic geometry of the intermediate has the properties of a membrane permeable for the electrons
absorber/emitter, (b) optical losses and (c) non-radiative and a p-type semiconductor is a permeable membrane for the
recombination influences these limiting efficiencies. holes. Highly doped n+- and p+-regions are therefore found at
Thermodynamic limitations to TPV conversion have the electrodes in solar cells. In most Si-solar cells, the part,
recently been discussed by Baldasaro et al [4]. The power where the photons are absorbed and the electron–hole pairs are
density tradeoffs discussed by Baldasaro et al are closely generated, is moderately p-type and combines the properties of
related to effects due to realistic geometries of the intermediate the absorber and of the hole membrane. The force that drives
absorber, which are investigated in our paper. However, the electrons is the gradient of the Fermi energy εFC and the
Baldasaro et al imply that the solar cell must be operated at the holes are driven by a gradient of εFV. The voltage V between
voltage that is the maximum power point when no recycling the electrodes is given by the difference of the Fermi energies
of emitted photons is present. We find that a correct treatment at the electrodes (eV = ε F,right − ε F,left) and as can be seen in
of the emission properties of the solar cell predicts higher figure 2, is equal to the chemical energy per electron–hole pair
efficiencies by moving the point of operation closer to the ε FC − ε FV, if the gradients necessary to drive the electron and
open-circuit condition of the solar cell. hole currents are negligible, as is the case in good solar cells.
For an analysis of the potential of the TPV principle we A good solar cell is able to transform the chemical energy of
have to look in detail into the operating principle of solar electron–hole pairs into electrical energy by 100%.
cells and into the balance of absorbed and emitted photon and
energy currents. 2.1. Current–voltage relation of solar cells

2. Operating principle of solar cells In the limit of only radiative recombination the current density
of a solar cell is given by
Solar cells consist of semi-conducting material. Photons jQ = −e(jγ ,abs − jγ ,em ), (2.1)
generate electron–hole pairs, when they are absorbed. The first where the absorbed photon current density jγ ,abs determines
step of the energy conversion from solar heat into electricity the overall rate of generation of electron–hole pairs. The
is a conversion of heat into chemical energy of the generated overall recombination rate is given by the emitted photon
electron–hole pairs. Chemical energy is produced by the rapid current density jγ ,em. Assuming full absorption by the solar cell
thermalization of electrons in the conduction band and of holes for photons with energy in the transmission range of the filter
in the valence band down to kinetic energies of 3/2 kT0 per ε l < h̄ω < ε u and no absorption outside this range, the absorbed
particle on average, where T0 is the temperature of the solar and emitted photon currents are given by a generalized Planck
cell. The resulting distributions of electrons in the conduction equation [5],
band as well as of holes in the valence band are separate Fermi
εu
 (h̄ω)2 dh̄ω
distributions with the Fermi energy εFC for the conduction jγ =  F C −εF V )
 , (2.2)
band different from the Fermi energy εFV for the valence band. 4π 3h̄3 c2 εl exp h̄ω−(εkT −1

S152
 Theoretical limits of thermophotovoltaic solar energy conversion

which differs from Planck’s original formula by introducing of photons in the integrand by h̄ω is
the difference ε FC − ε FV of the quasi-Fermi energies, which 
characterize the energy distribution of the electrons in a solar jE = σ T 4 , (2.3)
π
cell, giving rise to the emission of luminescent radiation. The
where σ = 5.67 × 10−8 W/(m2 K4) is Stefan–Boltzmann’s
energy current density jE associated with this photon flux jγ is
constant.
obtained by multiplying the integrand of equation (2.2) with
For later reference, we use equations (2.1) and (2.2) for
the energy per photon h̄ω. Equation (2.2) for jγ (and jE) gives
conventional solar energy conversion, where the solar cell is
the energy and photon currents (Iγ and IE) indicated in figure 1 directly illuminated by the sun. For non-concentrated sunlight
when multiplying the current densities with the appropriate a maximal efficiency of 30% is found for an optimal band gap
areas (AA for the absorber and AE for the emitter) and of the solar cell of ε G = 1.3 eV, while for full concentration
using the corresponding temperatures, splitting of the quasi- the maximal efficiency is 41% for a solar cell with εG =
Fermi energies (ε FC − ε FV = 0 for the emitter and absorber), 1.1 eV [1].
threshold energies ε l and a finite upper limit ε u in the case
of using a filter or selective emitter with finite bandwidth
for illumination of the solar cell. As the filter has to be
3. Theoretical maximum of TPV efficiency
reflective outside this bandwidth for efficient TPV conversion,
3.1. Black absorber
it effectively narrows down the spectrum of the net emitted
radiation for both the emitter and the solar cell. Unless If the solar radiation is incident from a solid angle S
some means for spacial control of the emission has been the incident energy current density is, according to Stefan–
implemented (as mentioned briefly at the end of section 3.1) Boltzmann’s law,
all solid angles  correspond to the entire hemisphere for S
jE,S = σ TS4 . (3.1)
the photon and energy currents Iγ and IE leaving components π
of the TPV system. The solid angle of the radiation If the intermediate absorber is black and at temperature TA , it
received by the absorber from the sun depends on the emits an energy current density into A = π which is lost to
concentration. The intermediate absorber receives radiation the outside
from the sun, a black body subtending the solid angle S =
jE,A = σ TA4 . (3.2)
6.8 × 10−5 for non-concentrated light and S = π for
maximal concentration.  = π results from Lambert’s This does not contain the radiation emitted towards the solar
law after integration over the whole hemisphere. The cell.
temperature of solar radiation is TS = 6000 K and the We can define an efficiency for absorption which is
difference of the Fermi energies ε FC − ε FV = 0. It may be jE,S − jE,A π TA4
advantageous for the intermediate absorber not to be black, but ηabs = =1− . (3.3)
jE,S  TS4
to absorb only photons with energies above a lower threshold
energy ε l, while the upper threshold energy of the absorber If all recombination in the solar cell is radiative, the open-
circuit situation of the solar cell is simple to describe, since
ε u = ∞. For the emission of the absorber, the solid angle
the solar cell must emit as many photons as it absorbs. If, in
is A = π, its temperature is TA and ε FC − ε FV = 0 in the
addition, the filter transmits near-monochromatic radiation, the
absorber.
energy currents exchanged between emitter and solar cell are
The emitter of the intermediate absorber/emitter
equal, too. In this quasi-equilibrium situation, where the solid
combination emits into E = π towards the solar cells, has angles and photon energy range for radiation from the emitter
the same temperature TA as the absorber and the difference through the filter and absorbed by the solar cell and radiation
of the Fermi energies ε FC − ε FV = 0. The emission towards emitted by the solar cell through the filter and absorbed by
the solar cells may be restricted to an energy interval εu–ε l the emitter are identical, equal absorbed and emitted photon
by its selective properties or by an appropriate filter, which currents in (2.2) require
reflects all photons with energy outside this interval back onto
h̄ω h̄ω − (εF C − εF V )
the emitter. The emitter may even be a light-emitting diode = . (3.4)
and have ε FC − ε FV > 0, a situation analysed separately in this kTA kT0
issue by Harder and Green [6]. The efficiency of the production of electrical energy per
The solar cell absorbs all photons transmitted by the filter. electron–hole pair as the chemical energy divided by the
The lower threshold is given by its band gap, εl = ε G. Due photon energy h̄ω is for monochromatic radiation from (3.4)
to radiative recombination, the solar cell also emits photons T0
ηcell = 1 − . (3.5)
towards the emitter into the solid angle C = π according TA
to (2.2). Due to a large difference of the Fermi energies Since this is the Carnot efficiency for reversible processes,
ε FC − ε FV in the solar cell, the emitted photon current density no other device could perform better in converting heat
may be large, although the temperature of the solar cell is than a solar cell with only radiative recombination operated
only T0 = 300 K. The selectivity of the emitter, or the filter in with monochromatic light. It also shows that under these
between the solar cell and the emitter, would again limit the conditions, no entropy is generated in the solar cell.
net emission spectrum to the photon energy range εu–ε l. Derived for open-circuit conditions, this efficiency also
For black-body radiation (ε l = 0 and ε u = ∞) the energy holds for small deviations from open circuit, i.e. for electrical
current density obtained from (2.2) by multiplying the number currents, which are much smaller than the short-circuit current.

S153
N-P Harder and P Würfel

100 10

dj E /d ω / W/(cm eV)
2
80
1
ηTPV /%

60
0.1

40

0.01
20

0 1E-3
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0 1000 2000 3000 4000 5000 6000
ω /eV
TA/ K
Figure 4. Energy current density spectrum as a function of photon
Figure 3. Maximum TPV conversion efficiency ηTPV as a function energy h̄ω for non-concentrated solar radiation (TS = 6000 K, solid
of the absorber temperature TA for a black absorber illuminated by line) and for radiation emitted by a black body of TA = 865 K into a
fully concentrated sunlight and using a monochromatic selective hemisphere (broken line).
emitter/filter.
60
In this situation, most of the electron–hole pairs generated by
50
the photons from the emitter recombine. As recombination is
assumed to be entirely radiative, and as all photons from the
40
solar cell are absorbed again by the emitter, no energy is lost.
ηTPV /%

This is quite different from normal photovoltaic conversion, 30

where all the energy of recombining electron–hole pairs is lost.
It is also remarkable that the efficiency of a solar cell does not 20
depend on its band gap, if it is operated in a TPV geometry,
where photons emitted by the solar cell are not lost. 10
From (3.3) and (3.5) the efficiency of TPV conversion is
   0
π TA4 T0 200 400 600 800 1000 1200
ηTPV = 1 − 1− . (3.6)
S TS4 TA TA/ K
Figure 3 shows this efficiency as a function of TA . For full Figure 5. TPV conversion efficiency ηTPV as a function of the
concentration of the incident radiation of temperature TS = absorber temperature TA for non-concentrated sunlight on a black
6000 K (S = π), it reaches remarkably large values and has absorber (solid line) and on a selective absorber (broken line)
a maximum value of 85% if the intermediate absorber/emitter absorbing only photons with energy h̄ω > 0.92 eV.
is operated at TA = 2544 K.
The efficiency of TPV conversion is more than twice as 3.2. Selective absorber for non-concentrated radiation
large as the efficiency for direct illumination of solar cells, Figure 4 shows why a black absorber performs so poorly with
although both are based on entirely radiative recombination. non-concentrated radiation. It shows the energy spectrum
The improvement results from the operation of the solar cell (energy current density per photon energy interval), i.e. the
with monochromatic light (with h̄ω close to the band gap integrand in (2.2) multiplied by h̄ω for the absorbed non-
which avoids thermalization losses) and at almost open-circuit concentrated solar radiation (solid line) and for the radiation
conditions. Both are possible in a TPV geometry, where emitted by the absorber for a temperature of TA = 865 K
unused photons are perfectly recycled between the emitter (broken line) as an example. Up to a photon energy of
and the solar cell. h̄ω = 0.92 eV more energy is emitted than is absorbed by
For no concentration (S = 6.8 × 10−5) the temperature the absorber. The energy balance of the absorber would be
TA of a black absorber must be reduced to below 400 K to positive, if photons with h̄ω smaller than a lower threshold
keep the emitted below the absorbed energy current. Since energy of ε l = 0.92 eV would neither be absorbed nor
this results in small values of the Carnot efficiency of the be emitted by making the absorber reflecting for these
solar cell, the maximum TPV efficiency for non-concentrated photon energies. Figure 5 shows the TPV efficiency for
radiation on a black absorber is only 6.7% for TA = 356 K (see non-concentrated sunlight as a function of the absorber
solid line in figure 5). temperature, if for each temperature an optimum value of εl is
As for full concentration a maximum TPV conversion applied. A maximal TPV efficiency of 54% is obtained for a
efficiency of 85% can also be obtained for non-concentrated selective absorber at TA = 865 K and an absorption threshold
solar radiation, if the surface of a black absorber has a structure of ε l = 0.92 eV.
that permits the emission of radiation only into the solid angle As for the case of full concentration of the incident
which is subtended by the sun (S = 6.8 × 10−5). In contrast radiation the TPV efficiency is remarkably larger than the
to maximum concentration this is a case of minimal emission. efficiency for direct illumination of the solar cell, although

S154
 Theoretical limits of thermophotovoltaic solar energy conversion

both are based on the assumption that the recombination in (a)

the solar cell is entirely radiative. From this observation absorber area mirror
one could hope that even for real solar cells and real optical
components, TPV conversion may be superior to conversion solar
by direct illumination of solar cells. cell
It is interesting to note that the same efficiency curves
as for TPV have been obtained for solar energy conversion
employing hot electrons [7]. The reason for this identity earth
is that in both cases, a temperature TA is established (by emitter area
entropy generating processes) and the heat energy at TA is
then converted by a Carnot engine. In the TPV case, the absorber area
Carnot engine is a monochromatically operated solar cell with (b)
only radiative recombination. In the hot electron case, the solar celll
Carnot engine is realized by making the electron and hole (convoluted) emitter area
membranes in figure 2 allow only mono-energetic electron (convoluted)
and hole transfer to the electrodes.
mirror
3.3. The problem of the emitter area
Although not explicitly stated, the maximum TPV efficiency, earth
as derived above, requires an infinitely large area of the emitter.
The intermediate absorber/emitter must emit as much as it Figure 6. TPV system designs using different geometries of the
absorbs. A part is emitted by the absorber into the outside intermediate absorber/emitter where the area size of the emitter is
hemisphere and is lost, and the rest must be absorbed by larger than that of the absorber.
the solar cell. Even if the emitter is black and could emit
unimpeded by a filter or a partly reflecting or reemitting solar absorber/emitter temperature increases and most of the energy
cell, the balance of absorbed and emitted energy currents absorbed from the sun is emitted back to the outside. As a
requires for all absorber/emitter temperatures TA , not too close result, the TPV efficiency is only 0.04% for full concentration
to the temperature TS of the sun, a large emitter area and 23.7% for no concentration. If the selectivity of the emitter
  or the filter is strongly reduced and all photons with h̄ω > εG
S 4 are transmitted to and absorbed by the solar cell, the problem of
IE,S − IE,A = AA σ TS − TA4 = AE σ TA4 = IE,E ,
π insufficient energy transfer is relaxed and TPV efficiencies of
(3.7) 23.2% for full concentration and of 44.8% for no concentration
are found.
from which the ratio of emitter area AE to absorber area AA is These results were found for an absorber emitting into the
AE S TS4 outside hemisphere. In the minimal emission case, where the
= − 1. (3.8) emission is directed into the small solid angle of S = 6.8 ×
AA π TA4
10−5 subtended by the sun, a maximal efficiency of 84.9% is
For full concentration and the optimal absorber temperature of
found for equal emitter and absorber areas.
2544 K AE /AA would have a value of 30. However, due to the
filter, which ensures monochromatic illumination of the solar
cell, most of the emission is reflected back onto the emitter. 4. TPV conversion with non-ideal components
The resulting small net emission must be compensated by an
even larger emitter area which, depending on the bandwidth The advantage of TPV conversion of avoiding thermalization
of the monochromatic illumination and the band gap of the losses in the solar cell relies on the ability to recycle
solar cell, may have to be even orders of magnitude larger unfavourable photons by a filter or avoid their emission
than prescribed by (3.8) [8]. In principle, large ratios of the by making the emitter selective. For near monochromatic
emitter area and the absorber area AE /AA could be realized operation of the solar cell, most of the photons must be
as sketched in figure 6. Figure 6(b) shows that even very large recycled and even small absorption losses may have a great
area ratios are theoretically conceivable in compact designs. impact. In an approach towards reality we have considered the
However, due to issues such as efficient heat transfer from the effect of non-ideal behaviour of the components of a TPV
absorber to the emitter area and—in designs similar to that converter.
shown in figure 6(b)—efficient cooling of the solar cell, it The first point concerns the radiation transfer between
appears that practicable area ratios will always be of the order emitter and solar cell. For the photons with hω < εG a loss of
of 1. In section 4, absorber/emitter systems are considered 5% is assumed. This could be caused by an absorption of 5%
having area ratios AE /AA between 0.1 and 100. of sub-band gap photons in the filter or in the solar cell and its
As an example, we have calculated ideal TPV efficiencies backside reflector. A loss of 5% was the lower limit that was
for an emitter area equal to the absorber area and equal to the achieved for Si solar cells with a back reflector by Swanson [9],
area of the solar cell. For quasi-monochromatic radiation with when reflection from the solar cell was used for the recycling of
a bandwidth of 1 meV incident on the solar cell the energy sub-band gap photons in TPV conversion. Swanson observed
transfer from the emitter to the solar cell is so small that the an efficiency of 29% with Si solar cells, however, in a closed

S155
N-P Harder and P Würfel

Shockley–Queisser limit still seem to be achievable for large

ratios of emitter to absorber areas. The optimal band gaps of
the solar cell were found to be between 0.3 eV and 0.5 eV.
Small band gaps are favourable because of smaller sub-band
gap optical losses. An investigation of the impact of non-
idealities will be given in [13].

5. Conclusions

TPV solar energy conversion has attracted attention because

of its very high theoretical efficiency of 85% for full
concentration, which is very close to the theoretical limit
for a tandem configuration of infinitely many solar cells with
different band gaps of 86.8% [14]. However, for achieving
Figure 7. Maximum solar TPV conversion efficiencies η as a this maximum TPV solar energy conversion efficiency in
function of the absorber temperature in the presence of optical addition to the absence of optical losses and of non-radiative
losses and non-radiative recombination, as defined in section 4.
Curves are shown for all combinations of the area rations
recombination in the solar cell a monochromatic selective
AE /AA = 1, 10 and 100 with the three concentrations C = 1, 100 emitter (or filter) and an infinitely large surface area of the
and 10 000. Also included is the maximum efficiency (circles) for emitter are required. Under these conditions TPV solar energy
AE /AA = 0.1 and C = 1. conversion allows operating the solar cell close to open-circuit
condition, which avoids entropy generation. The only entropy
system with an electrically heated emitter, which does not generation occurs in the absorber, since its temperature is
suffer from emission to the outside. smaller than the temperature of the radiation from the sun.
No losses of photons are considered for energies of hω > Limiting the ratio of emitter to absorber areas AE /AA
ε G. In the solar cell, non-radiative recombination is included to more practical values of 100 or below and considering
in addition to the radiative recombination. For all voltages optical losses of 5% for sub-band gap photons and non-
the non-radiative recombination is assumed to equal the radiative recombination in the solar cell of the same amount
radiative recombination, i.e. an external (electro-luminescent) as radiative recombination, reduces the maximal efficiencies
quantum efficiency of 50%. Values of this magnitude to only 32.8% for non-concentrated light which would not
have been observed in specially prepared three-dimensional justify the complications of a TPV system. Efficiencies up
samples of some III–V compounds with direct transitions to 60% are found for higher concentrations. The losses,
between valence and conduction band [10] and slightly which we have considered, are still small compared to losses
lower electro-luminescent external quantum efficiencies for in real systems, but they do not seem to be unachievable.
two-dimensional devices [11]. The presently best Si Large efficiencies at large concentrations, however, require
solar cells have recently been shown to have an external large area ratios AE /AA with values of about 100, which
electro-luminescence quantum efficiency of 1%, and future are conceivable from a manufacturing point of view, yet may
improvements can be expected [12]. introduce additional losses associated with heat conduction in
For the ratio of emitter to absorber area AE /AA , values of the intermediate absorber/emitter system.
1, 10 and 100 are considered.
Finally the concentration C of the incident sunlight is References
taken as C = 1 (no concentration), 100 and 10 000, less than
full concentration, which is 46 200. [1] Shockley W and Queisser H J 1961 J. Appl. Phys. 32 510
[2] White D C, Wedlock B D and Blair J 1961 Proc. 15th Power
The TPV efficiency is then calculated as a function of the Sources Conf. p 125
absorber/emitter temperature TA, while the lower threshold ε l Swanson R M 1979 Proc. IEEE 67 446
of the absorber, the band gap ε G of the solar cell, the spectral A compilation of references on TPV conversion can be found
width of the radiation incident on the solar cell are optimized in Lars Broman 1995 Prog. Photovoltaics 3 65
in order to find maximal values of the TPV efficiency. In [3] deVos A 1992 Endoreversible Thermodynamics of Solar
Energy Conversion (Oxford: Oxford University Press)
addition, near open-circuit operation of the solar cell is not [4] Baldasaro P F, Raynolds J E, Carache G W, DePoy D M,
favourable in the presence of non-radiative recombination Ballinger C T, Donovan T and Borrego J M 2001 J. Appl.
and a ‘maximum power point’ for TPV conversion must be Phys. 89 3319
found. [5] Lasher G and Stern Frank 1964 Phys. Rev. 133A 553
Figure 7 shows the result. As expected, the TPV Würfel P 1982 J. Phys. C: Solid State Phys. 15 3967
Würfel P, Finkbeiner S and Daub E 1995 Appl. Phys. A
efficiencies are lower than without the losses. For non- 60 67
concentrated radiation, the maximal efficiency of 32.8% is [6] Harder N-P and Green M A this issue
obtained for an area ratio AE /AA = 0.1 of emitter to [7] Würfel P 1997 Sol. Energy Mater, Solar Cells 46
absorber area. This is only slightly higher than a modified 43–52
Shockley–Queisser limit (30.4%), obtained from including [8] Harder N-P, Neuhaus D H, Würfel P, Aberle A G and
Green M A 2001 Proc. 17th European Photovoltaic Solar
a non-radiative recombination rate corresponding to 50% Energy Conf. (Munich)
external electro-luminescence quantum efficiency. At higher [9] Swanson R M 1980 Proc. Int. Electron Devices Meeting
concentrations, however, efficiencies considerably above the (Washington) p 186

S156
 Theoretical limits of thermophotovoltaic solar energy conversion

[10] Schnitzer I, Yablonovitch E, Caneau C and Gmitter T J 1992 [11] Windisch R, Heremans P, Knobloch A, Kiesel P,
Appl. Phys. Lett. 62 131 Döhler G H, Dutta B and Borghs G 1999 Appl. Phys. Lett.
Krames M R, Ochiai-Holcomb M, Höfler G E, 74 2257
Carter-Coman C, Chen E I, Tan I-H, Grillot P, [12] Green M A, Zhao J, Wang A, Reece P and Gal M 2001 Nature
Gardner N F, Chui H C, Huang J-W, Stockman S A, 412 805–8
Kish F A, Craford M G, Tan T S, Kocot C P, Hueschen M, [13] Harder N-P, Würfel P and Green M A Appl. Phys. submitted
Posselt J, Loh B, Sasser G and Collins D 1999 Appl. Phys. [14] de Vos A and Vyncke Dirk 1983 Proc. 5th E.C. Photovoltaic
Lett. 75 2365 Solar Energy Conf. (Athens) p 186

S157