Solar Cells, 30 (1991) 109-122

109

Growth and process optimization of CdTe and CdZnTe polycrystalline films for high efficiency solar cells
A. Rohatgi, R. Sudharsanan, S. A. Ringel and M. H. MacDougal
School of Electrical Engineering and Microelectronics Research Center, Georgia Institute of Technology, Atlanta, GA 30332 (U.S.A.)

(Received October 25, 1990)

Abstract Polycrystalline CdTe solar cells with efficiencies of approximately 10% were achieved by metal organic chemical vapor deposition growth of CdTe on glass/SnO2/CdS substrates. An in situ pre-heat treatment of the CdS substrate at 450 C in an H2 ambient was found to be essential for high performance devices because it removes oxygen-related defects on the CdS surface. This heat treatment also results in a cadmium-deficient CdS surface which may, in part, limit the CdTe cell efficiency to 10% owing to cadmium vacancy related interface defects. The CdC12 treatment used in CdTe cell processing was found to promote grain growth, reduce series resistance and interface state density, and change the dominant current transport mechanism from thermally assisted tunneling and recombination via interface states to recombination in the depletion region. These beneficial effects resulted in an increase in the CdTe/CdS cell efficiency from around 2% to approximately 9%. In addition to the CdTe cells, polycrystalline 1.7 eV CdZnTe films were grown by molecular beam epitaxy for tandem cell applications. CdZnTe/CdS cells processed using the standard CdTe cell fabrication procedure resulted in 4.4% efficiency, high series resistance, and a band gap shift to 1.55 eV. Formation of Zn-O at and near the CdZnTe surface was found to be the source of high contact resistance. A saturated dichromate etch instead of the Br2:CH3OH etch prior to contact deposition was found to solve the contact resistance problem. The CdC12 treatment was identified to be the cause of the observed band gap shift owing to the preferred formation of ZnC12. A model for the band gap shift along with a possible solution using an overpressure of ZnC12 in the annealing ambient is proposed. Development of a sintering aid which promotes grain growth and preserves the optimum 1.7 eV band gap is shown to be the key to successful wide gap CdZnTe cells.

1. Introduction T h i n film p o l y c r y s t a l l i n e cell e f f i c i e n c i e s h i g h e r t h a n 2 0 % c a n b e a c h i e v e d u s i n g a t a n d e m cell d e s i g n . F o r a t w o - c e l l t a n d e m d e s i g n , t h e o p t i m u m b a n d g a p s f o r t h e t o p a n d b o t t o m c e l l s a r e 1.7 eV a n d 1 eV r e s p e c t i v e l y [1]. C o n s i d e r a b l e p r o g r e s s h a s b e e n m a d e i n t h e a r e a o f t h e b o t t o m cell, w i t h s m a l l - a r e a C u I n S e 2 cell e f f i c i e n c i e s i n e x c e s s o f 1 4 % h a v i n g b e e n r e p o r t e d r e c e n t l y [2]. H o w e v e r , a c o m p a t i b l e w i d e b a n d g a p ( 1 . 7 eV) m a t e r i a l f o r t h e t o p cell h a s n o t y e t b e e n d e v e l o p e d . C d T e a n d C d Z n T e a l l o y s a r e c o n s i d e r e d f o r t o p cell a p p l i c a t i o n s . C d T e is a p r o m i s i n g p o l y c r y s t a l l i n e w i d e

0379-6779/91/$3.50

Elsevier Sequoia/Printed in The Netherlands

110 band gap material that has given small-area single-cell efficiencies in excess of 12% [3]. However, the CdTe bandgap is 1.45 eV, instead of 1.7 eV which is the optimum for the top cell. Thus, there are two approaches for realizing the goal of 20% efficient polycrystalline cells with a tandem structure. The first approach involves improving CdTe cell efficiencies in excess of 15% and augmenting that with a small amount of power from the bottom cell. The second approach involves developing a 1.7 eV bandgap CdZnTe top cell with an efficiency of approximately 10% to take greater advantage of the bottom cell efficiency owing to higher subgap transmission. This paper describes the investigation of growth and process optimization of CdTe and CdZnTe cells in order to understand the efficiency-limiting defects and mechanisms. Future directions are suggested to improve their efficiencies to a point where they can be a potential candidate for approximately 20% efficient tandem cells.

2. Experimental details
2.1. F i l m growth CdTe films were grown on CdS/SnO2/glass substrates by metal organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE). In the MOCVD growth dimethylcadmium was used as the cadmium source, and diallyltellurium and diisopropyltellurium were used as the tellurium sources. The CdTe films were grown at a substrate temperature in the range 3 0 0 - 4 0 0 C with a reactor pressure of 250 Torr. The Cd/Te ratio in the growth ambient was varied from 0.4 to 4.0. MBE CdTe and Cdl_~ZnxTe ( x = 0 . 3 5 , will be referred to as CdZnTe in the paper) films were grown using a Varian Gen II MBE system and elemental sources were used for all constituents having a purity of at least 99.999%. The substrates were baked out at 250 C for 3 - 4 h before film growth. The substrate temperature was kept at 275 C for 30 min to initiate film growth and was increased to 300 C for the remainder of the run [4]. 2.2. Cell fabrication P - i - n front-wall solar cells were fabricated at AMETEK Applied Materials Laboratory. After a dip in a saturated CdC12:CHaOH solution, the CdTe/CdS structures were annealed at 400 C for 30 min in air. Annealed films were etched in b r o m i n e - m e t h a n o l solution for 5 s followed by an evaporation of 1 0 0 0 / ~ of copper-doped p + - Z n T e film. Small-area (8 mm 2) nickel contacts were evaporated to form an ohmic contact [5]. P - n CdTe cells were fabricated by treating the CdTe films with CdC12 followed by an air anneal at 400 C for 30 min. Contact to the p-CdTe was made by a sequential evaporation of 100 /~ of c o p p e r and 400 /~ of gold followed by an anneal at 150 C in argon atmosphere for 90 min. Finally, b r o m i n e - m e t h a n o l etch was performed on the entire structure to etch off any residual cadmium and tellurium oxides [6].

111
2.3. M a t e r i a l a n d d e v i c e c h a r a c t e r i z a t i o n Electrochemical surface photovoltage (SPV) measurement, Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS), and transmission measurements were used for optical, physical and chemical characterization of the films. Current-voltage-temperature ( J - V - T ) , frequencydependent capacitance-voltage (C-V), and bias-dependent spectral response measurements were used to characterize device properties.

3. R e s u l t s a n d d i s c u s s i o n

3.1. CdTe s ol ar cells Figures 1 and 2 show the comparison of light current-voltage (I-V) and bias-dependent spectral response of 9.7% efficient p - i - n and 9.9% efficient p - n CdTe cells. The cell parameters are similar, and in both cases the spectral response is fairly uniform with sharp cut-off indicating that bulk recombination does not play a major role in the collection efficiency in either cell. However, the spectral responses of both cells show a strong wavelengthindependent bias dependence, suggesting that defects at the CdS-CdTe interface are limiting the performance of these cells by making the interface recombination velocity sensitive to applied bias [7]. Attempts were made to improve the interface quality by (a) controlling the CdTe film deposition conditions, (b) adjusting CdTe stoichiometry and (c) varying the i n s i t u pre-heat treatment of CdS in hydrogen atmosphere prior to CdTe deposition. Figure 3 shows the light I - V curves for CdTe cells with CdS substrates annealed at different temperatures. It clearly shows that a preheat treatment improves both the open-circuit voltage Voc and the fill factor FF but does not change the short-circuit current density Jsc. The best results, Voc=0.68 V, Js=20.5 mA cm -2, and F F = 0 . 5 6 , were obtained for substrates annealed at 450 C for 15 rain. The fact that J~c did not change and Vo~ and FF increased with heat treatment suggests that improved cell performance results from better interface quality. Capacitance vs. frequency measurements confirmed that the interface quality was improved after the preheat treatment. Cells fabricated with no pre-heat treatment of CdS not only showed higher absolute capacitance but also a stronger frequency dependence which is generally indicative of the active role of interface states [81. The fact that the 10% efficient cells subjected to this pre-heat treatment at 450 C for 15 min showed a strong bias dependence in the spectral response, suggests that pre-heat treatment is necessary but not sufficient to eliminate the detrimental effects of interface recombination. In order to understand the source of these defects, XPS and AES measurements were performed on various CdS films before and after the pre-heat treatment. XPS data (Fig. 4) showed that the pre-heat treatment was able to remove a significant amount of oxygen from the CdS surface. Since it is known that oxygen adsorbed on the surface of CdS gives rise to recombination centers

112 Z0

'

'

'

'

I'

'

'

'

'

1.6 I 0.8
0.4
F_,ff]ciency = 9.7 %

PIN

0.0
, I , I , 1 , I , I, I , I ,

\
I ,

-0.4

(a)
25

VOLTAGE (volts)

20

15

~E
10

I-Z LU EC nO

E v

!~ ~ ! . ! ! ! 8i20.0t 7 r r ~ ~ o / o \ i
5 0

Ei i n y 9.9 / fce c= %

PN

-5

-10
I I I I ~ I I I I

Co)

V L A Eo) OT G ( s vt l

Fig. 1. Comparison of light I - V data of (a) a 9.7% efficient p - i - n cell and Co) a 9.9/o p - n MOCVD-grown CdTe cell.

[9], it is reasonable to state that oxygen-induced defects on the CdS surface could be responsible for the large number of interface states and poor performance of the untreated cells. However, depth-resolved AES measurements made on the annealed CdS film showed that 450 C heat treatment in hydrogen causes the CdS surface to become cadmium deficient ([Cd]/[S] < 1) in addition to removing oxygen [8]. It is well known that cadmium deficiency in CdS gives rise to an acceptor type defect level which is situated at 0.26 eV above the valence band [10].

113
1.2 .... I''''1''''1''''1 .... I'''
, I

J 1.0
0.8
-

PIN

0.6

0.4

o.z
0.0

,,,,ll,,lllJ,ol

....

I ....

ILl
O,

c ~

(a)

Wavelen~h

(urn)

1.2
: o

o>,
O.B

.5 V

E:
LU

E .-I
0.6

PN

E "R
LLI

(~

0.4

0.2

i
6

Wavelength (um)
Fig. 2. Comparison of bias-dependent spectral response data of (a) a 9.7% p - i - n cell and (b) a 9.9% p - n CdTe cell.

This can p r oduc e a high resistivity CdS surface layer, which is known to r ed u ce Voc and FF in CuInSe2/CdS solar cells [11]. A similar mechanism is possible in CdTe/CdS cells, which may in part be responsible for limiting the efficiency to 10%. Since cadmium deficiency in CdS is a potential source of interface defects, one way of reducing such defects is to grow CdTe in cadmium-rich conditions. This was att em pt e d by gradually increasing the [Cd]/[Te] ratio in the vapor from 0.4 to 4.0. The 10% efficient cells were grown with a [Cd]/[Te] ratio of 0.4, in a cadmium-deficient ambient. The SPV responses of these films

114 24 20
~'16

E
c

~4
--4

-B

-12

; '011 '012 '013 '0:4 'ols

Voltage (volts)

016

0.7

Fig. 3. Light I - V data of CdTe/CdS solar cells fabricated on CdS/SnO2/glass substrates annealed at different temperatures.

~s0oc oeol

:: ~ ":

~or~o<o

I 1090"0

I Binding energy(eV)

I -100

Fig. 4. XPS data on the surfaces of CdS substrates before and after annealing at 450 C in hydrogen atmosphere inside the MOCVD reactor.

were measured with light incident on the CdTe film side. Figure 5 shows electrochemical SPV spectra of CdTe films grown on CdS with various [Cd]/ [Te] ratios. In order to separate the electrolyte-CdTe junction contribution from the CdTe-CdS interface contribution, only the SPV response in the long wavelength region is shown in the figure. It is interesting to note that the cadmium-rich films show a significant increase in the SPV, suggesting that the excess cadmium in the CdTe film is eliminating the cadmium deficiency at the CdS-CdTe interface. Since the SPV response is an indicator of Vo, it is possible to expect higher Voc on films grown under cadmium-rich conditions c o m p a r e d with 9.7% efficient CdTe cells grown in a telluriumrich ambient. The films are now being subjected to the standard postdeposition 400 C for 30 rain air anneal for cell fabrication. If the higher SPV response is maintained throughout the cell processing, then we should

115

I~" , ~

(o) TelCd =20 (b) Te/Cd = 10

~ ~( ~ v

u ~\ c ~~

(d) Te/Cd :0'5 (e) Te/Zd= 0"25

B00 900 Wavelength (nm) Fig. 5. Surface photo voltage spectra of CdTe films grown by MOOVI) on CdS/SnO2/glass substrates with different [Cd]/[Te] ratios.

700

be able to attain a higher Voc and efficiency compared with the 10% efficient cells. In order to estimate the maximum attainable efficiency as a result of the improved quality of the CdTe-CdS interface, model calculations were performed using a collection function approach. This was done by first calculating the collection function (ratio of zero bias quantum efficiency to quantum efficiency at different bias voltages) as a function of applied bias using bias-dependent spectral response data (Fig. 2). The losses in Voc and FF due to the collection function were calculated according to [7, 11]:

AVoc= (AKT/q)

in ~

Vo~

A F F = (Vmp]r~roc)(1 -- 77 Vmp)

where A is the ideality factor, n is the collection function, and Vmp is the voltage at the maximum power. For a 9.7% efficient MOCVD CdTe cell, the calculated losses in Vo and FF were 0.05 V and 0.12 respectively. In addition, reverse bias spectral response measurements indicate (Fig. 2) an additional loss of at least 10% (1.5 mA cm -2) in J,~ due to interface recombination. If these losses can be eliminated by improving the interface quality then we project that cell efficiencies of 13.5% can be achieved from the MOCVD CdTe cells. A further improvement in efficiency can be achieved by eliminating the loss due to absorption of high energy photons in the thick (1500 /~) CdS

116 window layer. The 10/0 cells had approximately 1500 /~ of CdS on SnO2, and the estimated loss in Jsc due to absorption is 4 mA cm -2 [5]. This results in an additional loss of 0.02 V and 0.08 in Voc and FF respectively. Thus, by eliminating this loss mechanism along with the interface recombination, it is possible to achieve 18% efficient CdTe cells. Attempts are being made to deposit thin CdS films by the chemical immersion method [12 ]. We have successfully grown CdS films in the thickness range 4 0 0 - 1 4 0 0 /~ by controlling the immersion time and have found an appreciable improvement in the transmission of 400 /~ CdS film in the short wavelength range. Thus a combination of optimum pre-heat treatment, p r o p e r [Cd]/[Te] ratio, and thin CdS films should yield a significant improvement in the MOCVD CdTe cell efficiency.

3.2. C d Z n T e s o l a r cells Polycrystalline CdZnTe films with a 1.7 eV band gap were successfully deposited by MBE onto CdS/SnO2/glass substrates [4]. However, CdZnTe solar cells fabricated by the identical process sequence used successfully for high efficiency CdTe cells gave efficiencies of only around 4.4%. In addition, the CdZnTe band gap shifted from 1.7 eV to 1.55 eV and the series resistance (around 2 - 6 ~ cm 2) was 3 - 5 times higher than in the counterpart CdTe solar cells. Detailed investigations were conducted to understand and remove the source of these problems. In order to find and eliminate the source of high resistance, depthresolved AES and XPS measurements were p e r f o r m e d near the CdZnTe surface after annealing (400 C for 30 min in air without the CdC12 treatment) and subsequent chemical etching to investigate process-induced changes in the CdZnTe surface which can prevent the formation of good ohmic contacts [13]. XPS analysis and AES profiles in Fig. 6 show that without any postanneal chemical etch, a significant amount of Cd-O, T e - O and Zn-O are present at and below the CdZnTe surface. After an etch in Brg.:CHaOH, which was used in the standard cell fabrication prior to ZnTe/Ni back-contact deposition, both cadmium and tellurium oxides were rem oved from the surface but the Zn-O remained at and below the surface, responsible in part for the high series resistance. In addition, the CdZnTe surface is not as tellurium rich ( [ T e ] / [ C d + Z n ] = 1.2) as the counterpart CdTe surface where the [Te]/ [Cd] ratio is approximately 1.5-2. It has been suggested that a telluriumrich surface can facilitate the ohmic contact formation on p-type CdTe [14]. In an attempt to remove the Zn-O and make the CdZnTe surface tellurium rich, various chemical etchants were investigated. Figure 6 shows that a post-anneal saturated dichromate (K2Cr207:H2SO4) etch r e m o v e d the nearsurface region that contained Zn-O and yielded a tellurium-rich surface layer approximately 0.15 ~ m thick with little or no detectable trace of cadmium, zinc or oxygen. C - V measurements made on the dichromate-etched surface confirmed a much higher carrier concentration of around 2 1017 cm -a in the tellurium-rich surface layer which gradually dropped down to the bulk

117

~'~

Te

3O

(.)

~
~
~< 0

Te
Zn ,0 30
(b)

Te

cd

30

(c) Te

e$

Te
ca

s'o

~ ~ < 0
Sputter

SPUTTER (rain.) TIME

Time(rain.)

Zn 3~ '

Fig. 6. Auger d e p t h profiles of air-annealed CdZnTe films after (a) no post-anneal etch, (b) Br2:CH3OH etch, and (c) saturated dichromate etch. Fig. 7. Auger d e p t h profiles of air-annealed CdZnTe films (a) without CdC12 t r e a t m e n t a n d Co) with CdC12 treatment. 2-

1.8: 1.6:

gl,!
~

1.4: 1.

n 0.8:

o.61 o.41 o.21 640 680 720

o.ea,
760
wavelength

\
840 880

800 (nm)

Fig. 8. Electrochemical surface photovoltage spectra of air-annealed CdZnTe/CdS structures with and without the CdC12 treatment.

d o p i n g c o n c e n t r a t i o n o f a r o u n d 5 x 10 z5 c m -8 o v e r a d i s t a n c e o f 0 . 1 5 /~m [13]. This s h o u l d eliminate t h e c o n t r i b u t i o n f r o m high c o n t a c t r e s i s t a n c e to the measured high series resistance. T h e n e x t s t e p w a s to i n v e s t i g a t e t h e p r o c e s s - i n d u c e d b a n d g a p shift o b s e r v e d in p r o c e s s e d CdZnTe films. AES d e p t h profiles s h o w n in Fig. 7 a n d t h e SPV s p e c t r a in Fig. 8 a f t e r t h e s t a n d a r d air anneal, with a n d w i t h o u t

118 the CdC12 treatment, clearly demonstrate that it is not the air anneal itself, but the CdC12 treatment coupled with the air anneal that is responsible for the band gap shift. The cells were fabricated with the CdCI 2 treatment. The Auger profiles show that after the CdC12 treatment, out-diffusion of zinc from the bulk toward the surface occurs which reduces the bulk zinc content and decreases the band gap, as demonstrated by the shift in the cut-off edge toward the longer wavelength in the SPV spectra of Fig. 8. A proposed model for the p h e n o m e n o n stems from the fact that, thermodynamically, the formation of ZnCI2 is preferred over CdC12 [ 15 ]. Therefore, introducing CdC12 onto the surface and into the CdZnTe bulk triggers the formation of ZnCI2 which has a much higher vapor pressure and lower melting point (318 C [15]) than CdC12, resulting in the out-diffusion of zinc. This process, where cadmium from the CdC12 substitutes for lattice zinc, resulting in ZnC12 formation can be described by the following equation: Cdi_xZnxTe+yCdC12. Cdl_~+vZnx_vTe+yZnC12

If y > x, all of the zinc will be consumed and CdZnTe will reduce to CdTe. SPV data (Fig. 8) taken on the CdZnTe/CdS structures show that if the CdC12 treatment is bypassed to preserve the band gap, a very weak photoresponse is observed, resulting in only 1%-2% efficient cells. However, the SPV data clearly show that incorporation of CdC12 during the processing of CdZnTe cells resulted in much higher photovoltages but the cut-off edge (or band gap) shifted from 1.7 V to 1.55 eV. The CdC12 treatment was also found to reduce the series resistance of the CdZnTe/CdS cells by a factor of about 10. In order to investigate both the need for and the beneficial effects of the CdC12 treatment on CdTe-based polycrystalline solar cells, experiments were first conducted on p - i - n CdTe solar cells fabricated with and without the CdC12 treatment to eliminate the complication due to the presence of zinc. We have shown elsewhere [16] that the CdC12 treatment is essential for high efficiency MBE-grown CdTe cells because it prom ot es CdTe grain growth from approximately 0.1 tLm to approximately 1 t~m, resulting in an increase in Vo from 385 to 720 mV, Jsc from 10.5 to 23.1 mA cm -2, FF from 0.32 to 0.51 and cell efficiency from 1.3% to 8.6%. Other investigators have observed similar improvements due to the CdCle treatment [6, 17]. In addition to quantifying the improvements in cell parameters, attempts were made to understand the effects of the CdC12 treatment on bulk properties, interface quality and carrier transport mechanisms. The CdTe/CdS SPV response in Fig. 9 showed a significant increase over the entire long wavelength range of 7 0 0 - 8 8 0 nm for cells annealed with CdC12 c o m p a r e d with cells annealed without CdC12. Since in our SPV measurements, light is incident on the back side (not the sun side) of the CdTe cell structure, the long wavelength photons near the CdTe cut-off (around 840 nm) are absorbed in the CdTe region near the CdTe-CdS interface. Hence, the significant increase in SPV response clearly supports an improved interface quality due to CdC12-induced grain growth. This is not surprising since grain growth

119
5

3
> D_ o9

700

720

740

760 780 800 820 wavelength (nm)

840

860

880

900

Fig. 9. Electrochemical surface photovoltage spectra of air-annealed CdTe/CdS structures with and without the CdC12 treatment. 0.7
0.6

f
"*nCdCI2 ~ "
~ A A A A h ~" ~

v=05vV=OV

0.5

~-

0.4 0.3

.~ 0.2
0.1~

ovo*oooo**oo.oooooooO**O,o

~~

OV

Oo

:hl
wavelength (nm)

-- T

Fig. 10. Bias-dependent spectral r e s p o n s e of completed Ni/ZnTe/CdTe/CdS/SnO2/glass solar cells with CdC12 (solid lines) and without CdC12 (symbols). The applied reverse bias values are noted.

reduces the n u m b e r of grain boundaries intersecting the interface which should result in r educed interface state density. Spectral response measurements shown in Fig. 10 also revealed a significant increase in bulk and interface quality due to the CdC12 treatment. The spectral response of a p - i - n device is generally dominated by interface recombination since the bulk is essentially depleted. Figure 10 shows that even after the CdC12 treatment, the spectral response showed an appreciable bias d e p e n d e n c e with mu ch higher response at 0.5 V applied reverse bias. This is indicative of the fact that even though the CdC12 treatment significantly improves the interface quality, it does not completely eliminate those interface states which can increase the interface recombination velocity. T e m p e r a t u r e de pe nde nt dark J - V m e a s u r e m e n t s were p e r f o r m e d to identify and quantify changes in the fundamental current transport properties

120

due to the improved interface quality. CdTe/CdS cells, with and without CdC12 treatment, were analyzed by fitting the J - V - T data to equations governing various transport mechanisms such as direct and multi-step tunneling [18], thermally assisted tunneling [19, 20], depletion region recombination [21], interface recombination [22], etc. It was determined that without the CdC12 treatment, current transport is dominated by thermally assisted tunneling of holes from the CdTe into interface states followed by interface recombination, characterized by an activation energy of 0.56 eV and J0 = 4.7 10 -7 A cm -2. For cells treated with CdC12, the dominant mode of current transport was found to shift from a tunneling-interface recombination mechanism to recombination via traps within the CdTe depletion region, indicating substantial reduction in interface state density. This change is manifested in an increase in the activation energy for current transport from 0.56 eV to 0.85 eV which results a significantly lower Jo value of 2 . 6 10 -9 A cm -2, accounting for the observed increase in Voc from 385 mV to 720 mV. Having established quantitatively the beneficial effects of the CdC12 treatment on CdTe material and device properties, in addition to the detrimental (band gap shift) effects on CdZnTe, it becam e evident that a sintering aid other than straight CdC12 is essential for obtaining efficient CdZnTe solar cells. However, some combination involving CdC12 may be a good choice since the CdCI2 treatment results in partial grain growth (determined from scanning electron microscopy measurements) and significantly lower series resistance yielding a much higher phot oresponse (Fig. 8). Since the band gap shift appears to result from the substitution of cadmium for zinc, attempts were made to control the band gap shift and induce grain growth by dipping the CdZnTe films in various solutions of CdC12 + ZnC12 in methanol prior to air annealing at different temperatures. Only limited success was obtained using this m et hod since the ZnC12 readily evaporates into the anneal ambient at temperatures greater than 320 C, leaving CdC12 alone which results in a partial band gap shift. For lower anneal t em perat ures (less than 320 C), negligible band gap shift occurs but this is accompanied by little or no grain growth and poor photoresponse. Attempts are underway to anneal CdC12treated CdZnTe surfaces in a ZnC12 overpressure ambient using a closed system to prevent the outdiffusion of zinc and induce grain growth. In this system, ZnCle is heated in a separate boat next to the CdZnTe sample. Thus, development of a sintering aid that can prom ot e CdZnTe grain growth without the bandgap shift remains the major obstacle for the success of wide band gap CdZnTe cells for tandem cell applications.

4. C o n c l u s i o n s MOCVD grown polycrystalline CdTe solar cells with efficienc,ies of aproximately 10% were demonstrated using both p - i - n and p - n structures. I n s itu pre-heat treatment of glass/SnO2/CdS substrates at 450 C for 15 rain prior to CdTe deposition was found to be essential for high perform ance

121 devices b e c a u s e it r e m o v e s o x y g e n a n d r e l a t e d d e f e c t states. T h e h e a t t r e a t m e n t also s i m u l t a n e o u s l y m a k e s t h e CdS s u r f a c e c a d m i u m deficient, r e s u l t i n g in c a d m i u m v a c a n c y - r e l a t e d i n t e r f a c e s t a t e s b e c a u s e o f w h i c h e v e n t h e 10% efficient cells s h o w a s t r o n g b i a s - d e p e n d e n t s p e c t r a l r e s p o n s e a n d l o w e r Voc a n d F F c o m p a r e d w i t h t h e b e s t r e p o r t e d [5] cells. P r e l i m i n a r y m o d e l calc u l a t i o n s s u g g e s t t h a t t h e r e m o v a l o f t h e s e s t a t e s c a n i n c r e a s e the cell efficiency f r o m 10% to 13.5%. P h o t o n a b s o r p t i o n in t h e CdS film also limits t h e cell p e r f o r m a n c e a n d e l i m i n a t i o n o f this loss m e c h a n i s m c a n result in efficiencies in e x c e s s o f 18%. T h e C d T e p r o c e s s s e q u e n c e c a n n o t b e u s e d to f a b r i c a t e efficient CdZnTe cells b e c a u s e it r e s u l t s in h i g h r e s i s t a n c e a n d a b a n d g a p shift f r o m the d e s i r e d v a l u e of 1.7 eV to 1.55 eV. The c o n t r i b u t i o n to the high series r e s i s t a n c e f r o m the n o n - o h m i c b a c k c o n t a c t w a s s o l v e d b y u s i n g a s a t u r a t e d d i c h r o m a t e i n s t e a d of the Br:CH3OH p o s t - a n n e a l etch, w h i c h r e s u l t e d in a p - t y p e s u r f a c e d o p i n g c o n c e n t r a t i o n o f 2 1017 c m -3. It is s h o w n t h a t CdC12 is e s s e n t i a l for high efficiency C d T e s o l a r cells b e c a u s e it p r o m o t e s g r a i n g r o w t h w h i c h i m p r o v e s b u l k collection, r e d u c e s i n t e r f a c e s t a t e density, a n d e l i m i n a t e s t h e t u n n e l i n g of c a r r i e r s t h r o u g h the i n t e r f a c e to i m p r o v e Voc, Js a n d efficiency. H o w e v e r , the u s e of a CdCl2 dip p r i o r to air a n n e a l w a s e s t a b l i s h e d to b e t h e m a i n s o u r c e of the o b s e r v e d b a n d g a p shift in CdZnTe d u e to ZnCI2 f o r m a t i o n b y s u b s t i t u t i o n of c a d m i u m for lattice zinc. This also r e s u l t e d in i n c o m p l e t e g r a i n g r o w t h in CdZnTe since m u c h o f the CdCI2 w a s c o n s u m e d b y this p r o c e s s . D e v e l o p m e n t o f a n a l t e r n a t i v e sintering aid, w h i c h i n d u c e s g r a i n g r o w t h b u t p r e v e n t s t h e b a n d g a p shift, is the k e y to t h e s u c c e s s of w i d e g a p p o l y c r y s t a l l i n e CdZnTe s o l a r cells for t a n d e m cell applications.

Acknowledgments
T h e a u t h o r s w o u l d like to t h a n k P. V. M e y e r s a n d C. H. Liu o f A m e t e k f o r s u p p l y i n g s u b s t r a t e s , R. W. B i r k m i r e a n d B. M c C a n d l e s s o f the Institute of E n e r g y C o n v e r s i o n for cell p r o c e s s i n g , a n d C. J. S u m m e r s f o r useful d i s c u s s i o n s , this w o r k w a s s u p p o r t e d u n d e r SERI c o n t r a c t X L - 7 - 0 6 0 3 1 - 1 .

References
1 J. C. C. Fan, Proc. Soc. Photo-Opt. Instm~m. Eng., 543 (1985) 30. 2 K. W. Mitchell, C. Eberspacher, J. Ermer and D. Pier, Proc. 20th IEEE Photovoltaic Specialists" Conf., Las Vegas, NV, 1988, IEEE, New York, 1989, p. 1384. 3 S. P. Albright, B. Ackerman, R. R. Chamberlin and J. F. Jordan, Proc. PolycrystaUine Thin Film Program Meet., August 16-17, Lakewood, Colorado, 1989, p. 17. 4 S. A. Ringel, R. Sudharsanan, A. Rohatgi and W. B. Carter, J. Electron. Mater., 19 (1990) 259. 5 P. V. Meyers and C. H. Liu, Proc. 4th Int. Photovoltaic Specialists Core., February 14-17, Sydney, NSW, Australia, 1989, p. 881.

122 6 R. W. Birkmire, B. E. McCandless and J. E. Phillips, Proc. PolycrystaUine Thin Film Program Meet., August 15--17, Lakewood, Colorado, 1989, p. 77. 7 R. Sudharsanan and A. Rohatgi, Sol. Cells, in the press. 8 R. Sudharsanan and A. Rohatgi, Proc. 21st IEEE Photovoltaic Specialists' Conf., Orlando, FL, 1990, IEEE, New York, 1990. 9 C. H. W u and R. H. Bube, J. Appl. Phys., 45 (1974) 648. 10 I. J. Ferrer and P. Salvador, J. Appl. Phys., 66 (1989) 2568. 11 M. Eron and A. Rothwarf, Appl. Phys. Lett., 44 (1984) 131. 12 W. J. Danaher, L. E. Lyons and G. C. Morris, Sol. Energy Mater., 12 (1985) 137. 13 S. A. Ringel, R. Sudharsanan, A. Rohatgi, M. S. Owens and H. P. Gillis, J. Vac. Sci. Technol A, 8 (1990) 2012. 14 B. M. Basol, J. Appl. Phys., 58 (1985) 3809. 15 I. Barin and O. Knacke, Thermochemical Properties of Inorganic Substances, Springer, Berlin, 1977. 16 S. A. Ringel and A. Rohatgi, J. Appl. Phys. (1991) to be published. 17 J. S. Roh and H. B. Im, J. Mater. Sci., 23 (1988) 2267. 18 A. R. Riben and D. L. Feucht, Int. J. Electron., 20 (1966) 583. 19 F. A. Padovani and R. Stratton, Solid State Electron., 9 (1966) 695. 20 W. A. Miller and L. C. Olsen, IEEE Trans. Electron. Devices, 31 (1984) 654. 21 D. L. Smith, Phys. Status Solidi, 44 (1977) 381. 22 A. Rothwarf, IEEE Trans. Electron. Devices, 29 (1982) 1513.