Monte Carlo Simulation of Solar Cells

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Solar Cells, I0 (1983) 211 - 222 211

MONTE CARLO SIMULATION OF SOLAR CELLS


C. MANFREDOTTI and M. MELIGA
Istituto di Fisica Superiore, Corso Massimo D'Azeglio 46, 10125 Turin (Italy)
(Received December 8, 1982; accepted May 10, 1983)
Summary
The Monte Carlo met hod was applied for the first time to calculate the
performance of a solar cell. It was proved that the met hod is self-consistent
even for a relatively low number of histories and, in the simple case of a
Schottky barrier, the results are in good agreement with the existing simple
theory. The possible application of this met hod is relatively straightforward
and could give much more information than existing numerical methods.
1. Introduction
Even though solar cells are semiconductor devices that are easy to pro-
duce, a complete analysis which takes into account the morphology of the
cell and the influence of various parameters is rather complex. Generally it is
necessary, for a complete understanding of the problem, to solve Poisson's
equation and the continuity equation simultaneously for electrons and holes.
Under illumination, these are non-linear partial differential equations. A
problem of this kind may be solved essentially in two ways: by numerical
analysis and by the Monte Carlo met hod (MCM).
Numerical analysis, which has many examples in the literature [1 - 3],
presents the advantage of short computation times, but it cannot avoid
problems due to boundary conditions, discontinuities, complex models or
structures. The MCM, which differs from numerical methods by its simplic-
ity and versatility, can also overcome very easily the most complicated prob-
lems connected with structural complexity, electrical potential behaviour,
discontinuities etc. without increasing the program complexity and, by tak-
ing the appropriate precautions, without even a marked increase in computa-
tion time. In fact this time can be kept to the same order of magnitude as for
numerical analysis; moreover, it is not proportional to the complexity of the
problem. Finally, the MCM, being generally based on a microscopic treat-
ment, leads to an easy visualization of the physical processes occurring in
solar cells and for this reason it can give various results without modification
of the program itself.
0379-6787/ 83/ $3. 00 Elsevier Seauoia/Print~rl i,, ~ ~,^,L . . . . . . .
212
In our work, a Schot t ky barrier n-type Au-Si solar cell was chosen
because of bot h structural simplicity and lack of literature of previous work
in this field. However, t he program is quite general and with slight modifica-
tions it may be applied to other kinds of solar cells, such as p- n j unct i on
cells, met al -i nsul at or-semi conduct or (MIS) cells or het eroj unct i on cells, and
it can easily be used for more compl ex structures, e . g . t andem cells.
2. The met hod
The MCM is very extensively used in various fields of physics, whenever
processes or experi ment s have to be simulated. The best advantages result in
cases where a few t ypes of process are repeated several times (cooling down
of neut rons, electromagnetic or nucleonic cascades etc.). In these cases each
particle is followed unt i l it has lost all its energy and t he various processes
t hat occur are i nt erpret ed on a probability basis. Since the sum of t he prob-
abilities of t he various processes at each step is of course equal to uni t y, a
random number between zero and uni t y is consequent l y chosen in order to
decide t he kind of process which should occur. In our case, for instance, in
order to select a phot on from t he solar spect rum, t he integral of its energy
distribution is normalized to uni t y and t he random number chosen by the
program is compared wi t h t he normalized partial integrals of the energy dis-
t ri but i on. For t he subsequent process, light absorption in t he cell, t he prob-
lem can be solved analytically since t he probability of absorption between x
and x + dx is given by
d/
p ( x ) d x = - - = o~ exp(--ax) dx
I0
which is obviously normalized to uni t y, and t hus t he random number R is
given by t he following expression:
d
f exp(--ax) dx = 1 -- exp(--~d) = 1 - - R
0
Therefore t he penet rat i on dept h of t he involved phot on can be obt ai ned
from t he equat i on
In R
d- - - -
~(~,)
a(~) being t he absorption coeffi ci ent at t he selected phot on wavelength.
3. The model
In t he present work, t he program describes an n-type Au-Si Schot t ky
barrier solar cell with a surface area of 1 cm 2. However, the various physical
213
a n t i r ef / ~
coating
Paramet er S y mb o l Value
Metal thickness SM
Solar cell thickness SL
Barrier height qb B
{lowered by Schottky effect}
Energy gap Eg(T)
(as a funct i on of temperature)
Resistivity p
Electron lifetime T e
Hole lifetime Tp
Hole diffusion length Lp
100 A
300 pm
0.78 eV
1.2 -- (3 X 10-4T) eV
1 ~c m
3.95 x 10 -4 s
2.62 x 10-Ss
150 pm
mE
0.766
!
barrier
tal
W
0.780.
0.778
0 . 7 7 4
uJ
r
base 0.770
20 40 60 8'0
T A B L E 1
S y m b o l s a n d parameters used in the M o n t e Carlo m e t h o d
$L d( A }
Fig. 1. Geometry of the solar cell with the various zones considered by the program. The
figure is not to scale.
Fig. 2. The met al -semi conduct or pot ent i al barrier used in the program as a funct i on of
depth. The Schottky effect is included and the lowest energy corresponds to a t unnel l i ng
transmission of 3%.
and geometrical paramet ers are given as external dat a and can be varied
according t o t he characteristics of t he device under investigation. In Table 1
t he paramet ers which were used in this wor k are given.
Fr om a geomet ri cal poi nt of view, t he model st ruct ure may be divided
i nt o f our di fferent parts (Fig. 1).
In t he a n t i r e f l e c t i o n c oat i ng and me t a l e l e c t r o d e z o n e onl y t he trans-
mission of i nci dent light is t aken into account [4]. In fact, for t ypi cal met al
thicknesses of a Schot t ky barrier solar cell t he phot oexci t at i on of el ect rons
from t he met al t o t he semi conduct or is negligible.
The barri er z o n e thickness is evaluated in t he program [5]. The shape
of t he pot ent i al barrier is shown in Fig. 2: it was obt ai ned by considering
214
TABLE 2
Variation in Iasc and Voc as funct i ons of the t i me interval used in the calculation of the
current
At Numbe r o f Iasc Voc
boxes (mA) (V)
(1/30)Tp 8 24.28 0.281
(1/100)Tp 14 26.52 0.283
(1/150)Tp 17 27.38 0.284
(1/ 200)rp 20 27.63 0.284
(1/250)Tp 22 27.86 0.284
a barrier height of 0.8 eV and det ermi ni ng t he lowering due to the Schot t ky
effect. In any case, it is given as an external vector and t herefore it is possible
to consider barriers of any shape and height. The barrier zone is divided into
boxes in such a way as to obt ai n a const ant electrical field in each box.
In t he base z one , carrier di ffusi on takes place and t herefore t he char-
acteristic paramet er will be t he mi nori t y {hole) diffusion length Lp. The base
zone is also divided into boxes and t he box wi dt h depends on t he time
interval At chosen to calculate t he instantaneous current. The diffusion
length {given in terms of t he number of boxes) is det ermi ned by simulating a
semi-infinite crystal with t he same physical characteristics as the base and by
considering hole injection at each t i me interval (see Table 2) and t hus the
number of boxes necessary for t he base zone is obt ai ned.
The back el ect rode is i mport ant because it gives rise to phot on reflec-
t i on and it is characterized by a reflectivity coeffi ci ent t hat depends on t he
metal of t he electrode and varies with t he phot on wavelength.
4. The program
The flow chart of t he program is shown in Fig. 3. The MCM, if rigorous-
ly applied t o our program, would t ake a l ot of central processing uni t (CPU)
time. In fact, under air mass {AM) 1 condi t i ons about 3 101~ phot ons
should impinge in 1 s on a cell of area 1 cm 2. Four of five carrier lifetimes
are enough for t he semi conduct or t o reach a st at i onary state. In fact it was
proved t hat t he cell reaches st at i onary condi t i ons in a t i me shorter t han t he
lifetime and t herefore t he calculation is limited to an interval equal to one
lifetime. A furt her reduct i on in t he number of phot ons can be obt ai ned by
considering a number of histories sufficient to give a good representation of
t he solar spectrum and of t he absorption coefficient. Usually, 10 4 histories
are enough from this poi nt of view and t herefore a mul t i pl i cat i on coefficient
is applied to t he number of carriers reaching t he barrier zone. This poi nt will
be discussed furt her below. The phot on source considered exhibits an AM 1
solar spectrum which, for simplicity, is restricted to a wavelength interval
215

Fig. 3. Flow chart of the program (for nomenclature see Appendix A).
TABLE 3
Comparison of the total energy and the average energy of photons in the real and in the
s i mu l a t e d s o u r c e
Data for the real source Data for the simulated source
Total number of phot ons 2.49 1017 cm -2 s -1 104 histories
Totalpower density (mW cm -2) 72.62 72.6171
Average energy of photons (eV) 1.8235 1.8230
bet ween 3000 and 12 000 A, correspondi ng t o a t ot al i nt ensi t y of 72. 62 mW
cm -2. Table 3 shows t hat t he si mul at ed source (104 histories) gives a t ot al
i nt ensi t y and an average phot one ne r gy al most coi nci dent with t hose of t he
real source.
The dark cur r ent is si mul at ed by consi deri ng t he pot ent i al barrier illus-
t r at ed in Fig. 2, t aki ng f or el ect rons in t he met al a densi t y of states wi t h a
normal di st ri but i on and including t he ef f ect of t he Fer mi - Di r ac statistics. Of
course, onl y t hermi oni c emission of el ect rons in t he met al is t aken i nt o
account . The energy interval f or el ect rons is rest ri ct ed in or der t o neglect, on
t he high energy side (at about 1 eV above t he Fermi level), t oo low an
216
o. s j /
O
0.3
0.1
0.I O.2 0/3
AE (eV)
Fig. 4. Qua nt um mechani cal ref l ect i on coef f i ci ent Q as a f unc t i on of t he ener gy di ffer-
ence bet ween el ectrons and t he t op of t he bar r i er .
occupat i on probabi l i t y and, on t he low energy side, t oo l ow a tunnelling
probabi l i t y (approxi mat el y 3%). For each hi st ory, bot h t he tunnelling prob-
ability and, for E > ~B, t he quant um mechani cal refl ect i on coeffi ci ent [6]
are t aken into account (Fig. 4). The dark current is t hen calculated as t he
dri ft current of el ect rons injected into t he semi conduct or.
The dark electric field is evaluated as a derivative of t he electrical
pot ent i al , and t he carrier mobi l i t y is cal cul at ed as a funct i on of t he dopi ng
and of t he electric field according t o ref. 7. The simulation of t he short-
circuit current Isc is obt ai ned wi t h 104 histories; for each hi st ory, t he wave-
length of an incident phot on and its pat h length d = --(l n R)/o~(~) in t he cell
(where ~(~) is t he absorpt i on coeffi ci ent and R is a random number) are
det ermi ned. If d ~< w t he phot on will be absorbed in t he depl et i on layer (the
quant um effi ci ency is t aken as equal t o uni t y); if w < d <~ SL t he phot on
will be absorbed i n t he base zone; finally, if d > SL t he phot on will be
refl ect ed by t he back el ect rode, according t o t he relative reflectivity coef-
ficient. Aft er each event, t he charge paramet ers C(I) and G(I), which rep-
resent t he phot ogener at ed carrier numbers in box I of t he depl et i on zone
and of t he base zone respect i vel y, are updat ed.
In t he cal cul at i on of t he st at i onary short-circuit current, t he di ffusi on
process f r om each box of t he base is fol l owed at each t i me interval At, whi ch
is expressed as a fract i on of t he mi nori t y carrier lifetime. This paramet er is
qui t e i mpor t ant since it implicitly det ermi nes, as previously shown, t he num-
ber of boxes in t he base zone. Subsequent l y, aft er a t i me lag equal t o t he
lifetime rp t he drift current and di ffusi on current are calculated, their sum
being Isc.
As shown in Fig. 5 (where At = (1/ 30)rp) I,c increases rapidly with
t i me, reaching f or t = Tp a value wi t hi n 2% of t he st at i onary value. Moreover,
it t urns out t hat t he average current Ia,~ calculated in t he interval 0 - Tp is
also within 3% of t he st at i onary value. The rapid increase in Is can be under-
217
15:
E
l{:t
I~J/30 2 0 / 3 0 3 0 / 3 0
t(Vp}
F i g . 5 . S h o r t - c i r c u i t c u r r e n t t r a n s i e n t c a l c u l a t e d a t t i m e i n t e r v a l s A t = ( I / 3 0 ) ~ p .
s t o o d if it is considered that t h e current transient is affected n o t o n l y b y t h e
m i n o r i t y carrier lifetime b u t a l s o b y t h e drift t i m e in t h e depletion layer,
w h i c h is m u c h s h o r t e r t h a n t h e lifetime. E v e n t h e diffusion t i m e in t h e base
z o n e is practically " w e i g h t e d " a c c o r d i n g to t h e e x p o n e n t i a l distribution o f
t h e p h o t o g e n e r a t e d carriers. A s far as A t is c o n c e r n e d , it is s h o w n in T a b l e 2
that it is n o t necessary to c h o o s e t h e shortest A t since, for e x a m p l e b y
c h a n g i n g f r o m A t = (1/100)rp t o A t = (1/150)rp, o n l y a 3 % increase in t h e
average Isc value is obtained. Neverthless, in o u r calculation A t = ( 1 / 2 5 0 ) r p
w a s used.
T h e present version o f t h e p r o g r a m calculates t h e short-circuit current,
t h e d a r k current, t h e o p e n ~ i r c u i t voltage, t h e current-voltage ( I - V ) charac-
teristics, t h e fill f a c t o r F F a n d t h e c o n v e r s i o n efficiency; t h e C P U t i m e for
1 0 4 histories is a b o u t 5 rain. M o r e o v e r , t h e p r o g r a m also provides t h e pos-
sibility o f evaluating all these d a t a at various temperatures.
5. Results a n d discussion
Since in t he model t he number of histories is necessarily l ow, some
checks are needed in order to confi rm t hat this number is sufficient t o
account for t he various physical processes involved wi t h good precision.
Among t he possible tests, t he solar spect rum and t he absorpt i on coeffi ci ent
at di fferent wavelengths have been consi dered. The mi nori t y carrier di ffusi on
(since it is simulated in a semi-infinite crystal) and t he carrier t ransport in
t he depl et i on layer are clearly less dependent on t he t ot al number of events.
218
!
0.084
!
i
0.034
'I ii
i I - -
~' iii
I Iii
I i - i I
J
I - - I f - - ' I - - II i
-] . . . . . . . . I.~
I ~ I I I I
- T I - ~ I E -
l ] I- J'i-
I ~, -i ~
- z -r I
I I (
r
-, I I
I -i
I
3'
I
T [
0.3 0.7 I~
I
II
I {
1
F i g . 6 . C o m p a r i s o n b e t w e e n t h e r e a l a n d s i m u l a t e d s o l a r s p e c t r u m i n t h e w a v e l e n g t h
r a n g e 0 . 3 - 1 . 2 p r o . O n t h e y a x i s t h e r a t i o o f t h e t o t a l n u m b e r o f p h o t o n s i n t h e r e a l
s o u r c e t o t h e t o t a l n u m b e r o f p h o t o n s i n t h e s i m u l a t e d s o u r c e is r e p o r t e d . T h e s p e c t r a
h a v e b e e n s h i f t e d f o r c o m p a r i s o n .
TABLE 4
Comparison between the simulated and the real absorption coefficient at three wave-
lengths and for t w o t h i c k n e s s e s
Thickness X N u m b e r o f ~ s i n m l a ~ d ~ m ~
( ~ m ) (A) absorbed photons
1 4000 9876 4.39 x 104
6000 3429 4.199 x 103
8000 952 1.0004 x 103
10 4000 10000 --
6000 9850 4.1997 103
8000 6320 0.99967 103
4.4 x 104
4.2 x 103
1.0 103
4.2 x 103
1 . 0 x 1 0 3
Figure 6 shows t hat t he solar spect rum in t he interval 0.3 - 1.2 pm is
very well r epr oduced by si mul at i on. It must be r emember ed (see Table 3)
t hat t he t ot al solar energy si mul at ed is, within 0. 004%, coi nci dent with t hat
delivered t o t he cell by t aki ng i nt o account t he real number of phot ons.
The absor pt i on coeffi ci ent ~(k) is also simulated at various wavelengths;
i t is derived f r om t he decrease in t he number of phot ons due t o t he absorp-
t i on in t he various boxes and is compar ed wi t h t he real absorpt i on coeffi-
cient. As shown in Tabl e 4, when t he absorpt i on in a cryst al ei t her 1 or 10
/~m t hi ck is consi dered t he maxi mum possible error in ~(k) is 0.2%.
Two exampl es of si mul at ed phot on at t enuat i on at 4000 f or a thick-
ness of 1 ~m and at 8000 A f or a t hi ckness of 10 #m are shown in Fig. 7
219
..~ 4 ,
Z
_.= ~i
i +
Ni xi e
l e
w
N
m
" o. +
l e e
50+
| -
[--
J r-*
l-++
I~+..
[+.
I - ] . . .
[ - - .
J - - ] - -
10 30 50 10 30 50
~x *b ox
Fig. 7. At t enuat i on of 4000 A phot ons in a thickness of 1 pm (logarithmic scale).
Fig. 8. At t enuat i on of 8000 A phot ons in a thickness of 10 pm (linear scale).
TABLE 5
Comparison between the solar cell performances obt ai ned by simulation and by theory
MCM results
With Schot t ky effect b Without Schot t ky effect
Average values Stationary values
Theoretical results a
Isc (A) 27.86 x 10 -3 27.86 x 10 -3 28.67 x 10 -3 28.84 x 10 -3
I s (A) 5.0 x 10 - 7 2.2 x 10 - 7 2.2 x 10 - 7 2.253 10 - 7
Voc (V) 0.284 0.304 0.304 0.304
FF 0.71 0.72 0.72 0.72
7) (%) 7.695 8.39 8.66 8.69
aThe t heory does not include the Schot t ky effect. The values given are stationary values.
b Average values.
and Fig. 8 respectively: a truly exponential phot on attenuation is clearly
evident.
The results obtained in the present work are summarized in Table 5,
where I~, Is, Voc, FF and W are reported and compared with theoretical
results obtained by using equations quoted in the literature [8]. In order
to make t he comparison more understandable, the results of the simulation
are also reported for a barrier without the lowering due to the Schottky
effect, since theoretical results can be obtained for this case only. For com-
pleteness, results are also presented for both the average and the stationary
values of the parameters. In the theoretical calculation, an average value of
the quantum mechanical reflection coefficient has been used [6]. The
agreement between t he Monte Carlo simulation and theoretical calculation
is extremely good, t he larger differences being 2% for I, (stationary values)
and 3% for Isc and ~ (average values).
220
10
5
V (v}
0.1 0.2 0.3
J
/ 1
2 0
3 0 l
Fig. 9 . I - V characteristics in A M 1 conditions as o b t a i n e d b y t h e M C M ( o ) (average
values) a n d b y t h e t h e o r y (4). T h e t h e o r y d o e s n o t include t h e S c h o t t k y effect.
0.6
~ 0.4
0.2
0 O. 5 0.7 O. 9 1.1
k('/~m)
Fi g. 1 0 . Re s po na i v i t y R ( mA mW - 1) as a f u n c t i o n o f wa v e l e ng t h as o bt a i ne d b y t he MCM
( e ) and b y t h e t h e o r y ( 4 ) ( r e f l e c t i o n f r o m t he bac k e l e c t r o d e wa s n o t i nc l ude d) . The
t h e o r y d o e s n o t t a ke i nt o a c c o u n t t h e S c h o t t k y e f f e c t .
The c o mp a r i s o n wa s e x t e n d e d t o t h e I - V charact eri s t i cs ( Fi g. 9 ) and t o
t h e r e s po ns i v i t y ( Fi gs . 1 0 a nd 1 1 ) . I n Fi g. 9 t h e di f f e r e nc e s can be at t r i but e d
t o t h e S c h o t t k y e f f e c t whi c h l owe r s Voe. The di f f e r e nc e s whi c h appear in
Fi g. 1 0 c an be e x pl a i ne d b y l o o k i n g at Fi g. 1 1 , wh i c h s h o ws t h e dri f t c ur r e nt
i n t h e d e p l e t i o n l ayer and t h e d i f f u s i o n current in t he bas e z o n e . In t h e
f o r me r c as e t h e di s agr e e me nt is a ma x i mu m at 4 0 0 0 A, c o r r e s po ndi ng t o a
pe ne t r a t i o n d e p t h c o mpa r a bl e wi t h t h e d e p l e t i o n l ayer t hi c kne s s , whi l e i n
t h e l at t er cas e t h e di s a g r e e me nt c o r r e s po nds t o a de e p pe ne t r a t i o n o f l i ght i n
t h e bas e z o n e ( a b o u t 3 0 p m) . I t i s t he r e f o r e pr o ba bl e t ha t t h e f o r me r di f f er-
e nc e i s d u e t o t h e l o we r s ur f ac e el ect ri c f i el d i n t h e MCM and t h e l at t er re-
s ul t s be c aus e t h e d i f f u s i o n current is c al c ul at e d by t h e t h e o r y unde r t h e
a s s u mp t i o n o f c o mp l e t e d e p l e t i o n at t h e e n d o f t h e bas e z o n e .
221
~ i o n
0.3 0.5 0.7 0.9 1.1
Fig. 11. The drift current and the diffusion current as funct i ons of wavelength as c a l c u -
l a t e d by the MCM (o) and by t he theory (A). Monochromatic l i g h t w i t h a p o w e r of 10
mW was incident on t h e c e l l .
6. Conclusions
Our work may be consi dered as a first exampl e of t he possibility of
using t he MCM in t he simulation of solar cells. It was proved t hat t he MCM
can wor k wi t h good precision, even taking into account a reasonable number
of phot ons and considering t he di ffusi on process in mi croscopi c detail. Since
reliable or usable experi ment al results are not available, our results were
compar ed wi t h t hose of t he usual t heor y.
The good agreement whi ch was obt ai ned can be consi dered as pr oof of
t he validity of t he MCM in a very simple physical case. However, as has been
not ed in t he i nt roduct i on, t he MCM can easily be ext ended t o mor e compl ex
st ruct ures such as t andem cells, het eroj unct i on cells, graded band gap cells
and MIS cells and t o transient effect s in solar cells. This ext ensi on will be
carried out in fut ure work, particularly if mor e experi ment al dat a are avail-
able. Fr om our poi nt of view, t he MCM is pr obabl y t he simplest met hod by
which t o st udy t he influence of t he various paramet ers (lifetime, t emper-
ature, di ffusi on profiles, thickness of t he various zones, absorpt i on coef-
ficient etc. ) on t he per f or mance of t he cell, particularly in cases where these
paramet ers are also dependent variables. Effort s will be made in fut ure wor k
t o investigate all t hese possibilities of t he MCM.
We are also conf i dent t hat a t wo- or three<limensional ext ensi on of our
program, in order t o deal wi t h compl ex probl ems such as series resistance or
vertical cells, may be at t ai ned wi t hout severe compl i cat i ons.
References
1 P. M. Dumbar and R. J. Hauser, Solid-State Electron., 19 (1976) 95.
2 G. J . Fossum, Solid-State Electron., 19 (1976) 269.
222
3 J. Michel and A. Mircea, Act a El ect ron. , 18 (4) (1975) 331.
4 M. V. Schneider, Bel l Syst . Tech. J. , 45 (1966) 1611.
5 S. M. Sze, Physi cs o f Semi conduct or Devices, Wiley-Interscience, New York, 1960,
Chap. 8, p. 363.
6 C. R. Crowell and S. M. Sze, J. Appl . Phys. , 37 (7) (1966) 2683.
7 S. Sharfetter and A. Gummel, I EEE Trans. El ect ron Devices, 16 (1969) 64.
8 H. J. Hovel, in R. K. Willardson and A. C. Beer (eds.), Semi conduct or s and Semi -
metals, Vol. 11, Sol ar Cells, Academic Press, New York, 1975, pp. 112 - 126.
Appendix A: nomenclature for Fig. 3
E(I) electric field
FF fill factor
Iasc average current
Id diffusion current
/dr drift current
Iisc instantaneous current
Is dark current
Isc short-circuit current
I - V I - V characteristic
Nbb number of boxes in the base
Nbw number of boxes in the barrier
Ndt number of time intervals
Np number of time intervals
NB number of histories
V(I) potential
Voc open,circuit voltage
w barrier width
At time interval
efficiency
Pe electron mobility
#h hole mobility

You might also like