CdTe solar cells

CdTe solar cells

Applications of Surface Science 22/23 (1985) 1083-1090 North-Holland, Amsterdam 1083 THIN FILM CdS/CdTe S O L A R C E L L S W.J. D A N A H E R , L...

347KB Sizes 1 Downloads 36 Views

Applications of Surface Science 22/23 (1985) 1083-1090 North-Holland, Amsterdam



L.E. L Y O N S and G.C. M O R R I S

Department of Chemistry, University of Queensland, St. Lucia, Queensland 406Z Australia

Received 27 August 1984; accepted for publication 31 October 1984

The effects of processes used in making thin film ITO/CdS/CdTe/Au heterojunction solar cells have been investigated. In particular, air annealing the ITO/CdS/CdTe layer is shown, from spectral evidence, to produce a heterojunction. XPS evidence shows that this heating forms CdO and TeO2 on the CdTe surface and that these are removed by a KOH etch. A further bromine/methanol etch leaves a Te-rich surface which forms an injecting contact with the Au top electrode into which the Te mixes.

1. Introduction A n all thin film heterojunction of C d S / C d T e is one of the most promising systems for an efficient solar cell [1] and m e t h o d s of making such cells include e v a p o r a t i o n [2], screen printing [3] and electrodeposition [4]. In o u r work, C d T e is electrodeposited on CdS which was chemically deposited on indium thin oxide ( I T O ) coated glass. T h e h e t e r o j u n c t i o n is f o r m e d by heating, and an injecting contact f o r m e d at the C d T e surface by a gold layer on the chemically etched CdTe. Processes that occur during these steps are discussed in this paper, and in particular X-ray p h o t o e l e c t r o n spectroscopy is used to p r o b e some properties of the films and the changes which occur during processing.

2. Experimental Thin layers of polycrystalline CdS were chemically deposited at 90°C o n t o indium tin oxide (ITO) coated glass ( P P G Industries, U S A ) . T h e deposition solution contained 0.021 mol dm -3 CdCI 2, 0.065 mol dm -3 NH4C1, 0.665 mol dm -3 N H 3 and 0.209 tool dm -3 thiourea, p r e p a r e d from A R grade chemicals dissolved in water purified by a Millipore "Milli-Q" system. Deposition times of 15min gave films approximately 0 . 3 # m thick. Full details of the deposition process and the film properties are to be r e p o r t e d elsewhere [5]. 0378-5963/85/$03.30 O Elsevier Science Publishers B.V. ( N o r t h - H o l l a n d Physics Publishing Division)


W.J. Danaher et al. / Thin film C'dS/CdTe solar cells

The CdTe layer was cathodically deposited onto C d S / I T O in a two c o m p a r t m e n t cell from a N 2 purged, magnetically stirred solution containing 2 . 7 m o l d m - 3 C d S O 4 and 1.5× 10 4 m o l d m - ~ T e O ~ at p H - 2 . The solution was prepared using "Milli-Q'" water and was pre-electrolysed prior to use. Films were deposited at 90°C using deposition potentials between + 100 and 0 m V versus C d / C d S O 4 reference electrode, also at 90°C. Film thickness ranged from 0.3 to 1.0#m. The deposition procedure has been detailed elsewhere [6]. ITO/CdS/CdTe/Au devices were prepared by annealing the ITO/CdS/CdTe for 10 min in air at 350°C, etching in A R K O H (30% w/w, 80°C, 20 s), blow-drying, etching for 5 s in 0.1% (v/v) bromine in methanol (both A R grade), rinsing in A R methanol for 15s and then in -Milli-Q'" water for 30 s, blow-drying in N~ and then evaporating a 80 nm Au layer onto the surface. For an unheated I T O / C d S / C d T e / A u device, the only etch was A R K O H in water (30% v/v, 80°C, 20 s). Current-voltage curves for devices were taken in dark and light using standard current and voltage measuring equipment. An Oriel Solar Simulator, calibrated at 100mW/cm 2 by a Solarex Si cell traceable to NBS standards, was used as the light source. Surface analysis results were obtained using a Physical Electronics 560 Multi-technique System equipped with X-ray Photoelectron Spectroscopy (XPS) and Secondary Ion Mass Spectroscopy (SIMS), and incorporating a dual pass cylindrical mirror analyzer. D a t a acquisition and analysis for XPS was accomplished using a Physical Electronics Multiple-technique Analytical C o m p u t e r with Physical Electronics Version 6 software. The spectrometer work function was calibrated on the basis of the Au 4f7/2 peak energy of 83.8 eV and all reported binding energies are referenced to C Is at 284.6 eV. All spectra were obtained using Mg K a X-rays with energy 1253.6 eV.

3. Results and discussion

3. I. The CdS layer SEM photographs showed that CdS thin films were mainly composed of small, columnar crystals about 0 . 1 - 0 . 2 # m in size, with some larger CdS particles trapped in the film. X-ray diffraction spectra revealed a hexagonal crystal structure with the 002 peak dominating. XPS results for the atomic concentrations of Cd and S for the films and for a single crystal were the same, indicating that the films were stochiometric. SIMS analysis of films revealed the presence of i m p u r i t i e s - ( C H , ) +, C d O H +, C d O +, O-, CN ,CI-, C N S - - t r a c e a b l e to contaminants from the deposition solution. Only peaks

W.J. Danaher et al. / Thin film CdS/CdTe solar cells


attributed to carbon and oxygen were detected in XPS survey spectra. Resistivities for as-deposited films were 109.(2 cm for in-plane conduction and 104-105/-2 cm for out-of-plane conduction. Annealing for 30min at 450°C in 2 0 k P a of high purity H 2 lowered the resistivity for in-plane conduction to 0.15/2 cm, and annealing under the same conditions in the presence of Cd vapour lowered the resistivity to 0.03 ,(2 cm. The same effect has been noted for CdS produced by other methods [7-9]. SIMS analysis showed that H 2 annealing also lowered the impurity level in the films but resulted in diffusion of In from the ITO into the CdS. Even though H 2 annealing produced lower resistivity CdS, solar cells incorporating annealed CdS have so far, in our work, exhibited lower conversion efficiencies than those incorporating as-deposited CdS. Cells incorporating vacuum evaporated CdS also had low efficiencies ( < 1 % ) . Both H 2 annealed CdS and vacuum evaporated CdS are known to contain a Cd excess [10,11] and are likely to be doped with In from the ITO. Subsequent annealing of the ITO/CdS/CdTe may cause these species to diffuse into the CdTe and hinder the formation of a heterojunction, but further work is needed to establish this mechanism.

3.2. The CdTe layer CdTe films were polycrystalline with columnar grains up to 1.2 # m in size orientated so that the cubic 111 peak dominated in the X-ray diffraction spectrum. As-deposited films were approximately stoichiometric (measured by atomic absorption spectroscopy), except for those deposited at potentials where free Cd also deposits. XPS analysis also gave a C d : T e ratio of approximately unity. Films exhibited an impurity level considerably less than that of a 99.999% pure single crystal. Typical film resistivities were 108 12 cm for in-plane conduction and 105-106,O cm for out-of-plane conduction [6].

3.3. Heterojunction formation Heating was used to form a n-CdS/p-CdTe heterojunction by conversion of n-CdTe to p-CdTe. Evidence for such a conversion was shown by alteration of the wavelength dependence of J~ after heating. The spectral response for an as-deposited cell and a heated cell are shown in fig. 1. Also included is a plot of the absorption coefficient for single crystal CdTe [12]. The value of J~ was measured at each wavelength and the data corrected for reflective loss. For the as-deposited cell, the peak of J~ was near 750 nm, indicating that carriers were mainly collected near the CdTe/Au interface where the weakly absorbed light penetrates. For the heated cell, the peak of J~ was near 550 nm, indicating that the more strongly absorbed light (near the CdS/CdTe barrier) gave rise to the carriers mainly collected. Mitchell


W.J. Danaher et al. / Thin film CdS/CdTe solar cells


32 (b )~"~

7 E

~ "--.

v I.--



~ "





1 0'~


o ,,~ UJ

o o z o_


F13.. riO ~9 m i






600 700 800 WAVELENGTH (NM)

Fig. 1. Spectral response for ITO/CdS/CdTe/Au (light incident through the ITO): (a) unannealed; (b) annealed in air, 10 min, 350°C; (c) absorption coefficient for single crystal CdTe [121.

[13] s h o w s a s i m i l a r d i a g r a m for a cell f o r m e d with e v a p o r a t e d C d S on single crystal p - C d T e .

3.4. Solar cell performance T h e light a n d d a r k c h a r a c t e r i s t i c s of a cell f o r m e d as d e s c r i b e d a b o v e are shown in fig. 2. F o r the cell, C d T e was e l e c t r o d e p o s i t e d at 0 m V versus C d / C d S O 4. T h e cell c h a r a c t e r i s t i c s w e r e : o p e n circuit v o l t a g e = 0.59 V; short circuit c u r r e n t d e n s i t y = 1 5 . 3 m A / c m 2 ; fill f a c t o r = 0.5; series r e s i s t a n c e in xll



E 20 o

E N 10 I~













U (Volts)

-10 j


-20 Fig. 2. Current-voltage characteristic in right and dark for I T O / C d S / C d T e / A u . Dark curve is shown in sections with corresponding expansion factors.

W.J. Danaher et al. / Thin film CdS/CdTe solar cells


the light (measured from the slope of the J - U curve at +1.0 V ) = 4 0 cm2; series resistance in the dark (measured using the deviation of the In J versus U curve from linearity at 20 mA/cm 2) = 18/2 cm:; the diode factor = 1.8; and the reverse diode saturation current density = 7.0× 10-gA/cm 2. The theoretical efficiency of such a cell is 17% [14]. As part of our programme of improving the film cell efficiencies, we have used electrical and surface analytic methods to investigate the changes occurring during processing. The surface analysis results are reported now.

3.5. Chemical changes during heterojunction formation Air annealing, used to form the heterojunction, also formed an oxide layer at the CdTe surface. The XPS spectrum in the Te 3d region for air annealed ITO/CdS/CdTe shows the doublet 3d5/2 (572.0eV) and 3d3/2 (582.4eV) for lattice Te in CdTe and corresponding peaks at 575.6 and 586.0 eV from Te in T e O 2. The energies agree to within 0.5 eV with values published elsewhere [15] and all XPS results are summarized in table 1. Oxide formation leads to only a small shift in the Cd 3d peak energies [15] and so it was necessary to use the Auger peaks to monitor the presence of CdO. The MsNasN45 Auger peak of Cd was used to look for CdO because of the chance coincidence of the Te 3pl peak and the Cd M4N45N45 peak. The











Fig'. 3. XPS spectrum of CdTe surface, Te 3d region: (a) after air anneal, 10 min, 350°C; (b) after 20s etch 30% (w/w) KOH, 80°C. Experimental conditions: M g K a radiation; X-ray voltage = 12 kV; source power = 300 W; pass energy = 25 eV. Data are shown after satellite subtraction, deconvolution (5 iterations) and smoothing (11 point).


Au evaporation

After Brzimethanol


572.7 Te/free



Lattice Te Less oxidized

571.7 575.5(weak)

After KOH


Lattice Tc Oxidized Te

572.0 S75.6

After air anneal


Lattice Te Oxidized Te

Lattice Te Free Te Oxidized Te

572.0 573.0 576.1

in the preparation Interpretation


Te 3d5,z

at various

572.2 575.6

air anneal




of XPS results obtained






solar cells








Oxidized Cd/ lattice Cd

Oxidized Cd/ lattice Cd

Oxidized Cd/ lattice Cd


878.6 877.4




Cd MsN4N4i

of ITO/CdS/CdTe/Au

Te only

Single crystal



2 6

G2 a !--

,b b

f i

W.J. Danaher et al. / Thin film CdS/CdTe solar cells


Cd MsN45N45 peak was a broad band (4 eV, FWHM, peak 877.8 eV) with shoulders at 879.0 and 876.8 eV. The reported values [15] for Cd in CdO and for lattice Cd are 878.6 and 876.6 eV respectively, and the data confirm their presence after air annealing. Air annealing left the CdTe surface Cd rich, with a Cd : Te ratio for the region probed by XPS of 1.7. Out diffusion of Cd during annealing has been reported [16] and is likely to be the mechanism by which n-CdTe is converted to p-CdTe. 3.6. C h e m i c a l changes during etching

Etching with K O H after air annealing removed much of the oxide layer. This is illustrated by fig. 3b where the intensity of the T e O 2 peaks relative to the lattice Te peaks is considerably reduced compared with that for fig. 3a. This result is not unexpected since T e O 2 is soluble in K O H [17]. The Cd MsN45N45 Auger peak shifted to a lower energy (877.1 eV) compared with that for the annealed surface (877.8 eV), suggesting that CdO was removed. Etching also resulted in a change of the energy of the O ls peak (530.2 eV unetched surface; 531.1 eV etched surface). Etching in K O H left the surface region with a C d : T e ratio of 2.3, a result consistent with a mainly T e O 2 surface formed in the heating which is then removed by the K O H etch. Etching in 0.1% (v/v) bromine in methanol of the heated and K O H treated surface resulted in the preferential removal of Cd from the surface, leaving a Te-rich surface region with a C d : T e ratio of 0.5. Similar effects have been noted for etching single crystal CdTe in 2% (v/v) bromine in methanol [18]. Further, the Cd MsN45N45 Auger peak energy was 876.4 e V in close agreement with the value for lattice Cd [15]. The Te 3d peak energy was 572.7 eV lying between the values of 573.0 and 572.0 eV for free Te and lattice Te, respectively [15]. This can be explained by the presence of a new layer of free Te on the bulk CdTe [19]. Some lattice Te is expected because of the presence of lattice Cd. 3. 7. Gold contact

After etching, a 80 nm thick Au layer was evaporated onto the surface. XPS analysis of the surface region revealed the presence of large amounts (up to 20 at%) of Te in the Au. In most cases no Cd XPS signal was detected in the same surface region. The Te 3d5/2 peak energy was 572.3 eV, lower than the value for free Te and suggested that the Te was bonded to the Au but conclusive evidence for an A u - T e compound was not obtained.

4. Conclusion

Evidence has been presented to show that heating in air produces a


W.J. Danaher et al. / Thin film CdS/CdTe solar cells

heterojunction between the CdS/CdTe films in an I T O / C d S / C d T e / A u solar cell. This heating also causes oxidation of the CdTe to form C d O and TeO 2. The oxidized products are removed by a K O H etch and further etching with bromine in methanol leaves free Te as well as lattice Te on the surface. The Cd " Te ratio measured by XPS for the surface region changes from 1 • 1 (as deposited), 1.7 (air annealed), 2.3 ( K O H etch), 0.5 (BrJmethanol etch).

Acknowledgements Financial support was from the Australian Research Grants Scheme and the National Research Postdoctorate Fellowship Scheme (to W. Danaher). Technical support was from C. Owen and P. Tanner. We thank each of these.

References [1] K. Zweibel, A. Hermann and R. Mitchell, Solar Cells 12 (1984) 257. [2] D. Bonnet and H. Rabinhorst, in: Proc. Intern. Conf. on the Physics and Chemistry of Heterojunctions, Vol. 1 (Akademiai Kiado, Budapest, 1971)p. 119. [3] H. Matsumoto, K. Kuribayashi, H. Uda, Y. Komatsu, A. Nakano and S. lkcgami, Solar Cells 11 (1984) 367. [4] F.A. Kroger, R.L. Rod and M.P. Panicker, US Patent 4,4/X),244, August 23. 1983. [5] W.J. Danaher, L.E. Lyons and G.C. Morris, Solar Energy Mater., in press. [6] L.E. Lyons, G.C. Morris, D.H. Horton and J.G. Keyes, J. Electroanal. Chem. 168 (1984) ll)l. [7] A.L. Fahrenbruch, V. Vasilchenko, F. Buch, K. Mitchell and R.H. Bube, Appl. Phys. Letters 25 (1974) 605. [8] Y.Y. Ma, A.L. Fahrenbruch and R.H. Bube, Appl. Phys. Letters 30 (1977) 423. [9] K. Yamaguchi, N. Nakayama, H. Matsumoto and S. lkegami, Japan. J. Appl. Phys. 16 (1977) 1203. [10] K. Mitchell, A.L. Fahrenbruch and R.H. Bube, J. Vacuum Sci. Tcchnol. 12 (1975) 91)9. [11] A.G. Stanley, Appl. Solid State Sci. 5 (1975) 251. [12] P. Rappaport and J.J. Wysocki, Acta Electron. 5 (1961) 364. [13] K. Mitchell, PhD Thesis, Stanford University (1977) p. 85. [14] A.L. Fahrenbruch, J. Crystal Growth 39 (1977) 39. [15] U. Solzbach and H.J. Richter, Surface Sci. 97 (1980) 191. [16] F.J. Bryant, A.K. Harris, S. Salkalachen and C.G. Scott, Thin Solid Films 105 (1983) 343. [17] R.C. Weast (Ed.), CRC Handbook of Chemistry and Physics. 63rd ed. (CRC Press, Boca Raton, FL, 1982) p. B-155. [18] J.G. Werthen, J.-P. Haring, A.L. Fahrenbruch and R.H. Bube, J. Appl. Phys. 54 (1983) 5982. [19] H.S. White, A.J. Ricco and M.S. Wrighton, J. Phys. Chem. 87 11983) 514/).