Thin film CdTe solar cells

Thin film CdTe solar cells

512 Journal of Crystal Growth 72 (1985) 512—524 North-Holland, Amsterdam THIN FILM CdTe SOLAR CELLS C. COHEN-SOLAL. M. BARBE. H. AFIFI and G. NEU *...

1001KB Sizes 3 Downloads 91 Views

512

Journal of Crystal Growth 72 (1985) 512—524 North-Holland, Amsterdam

THIN FILM CdTe SOLAR CELLS C. COHEN-SOLAL. M. BARBE. H. AFIFI and G. NEU

*

Laboraroire de Physique des So/ides, CNRS, / Place Arisnde Briand, F- 92/95 Meudon Principal Cedex, Fran

This paper reviews some of the most promising CdTe-based solar cells, with efficiencies in excess of 10%, prepared by various film deposition methods including close-spaced vapor transport, close-spaced sublimation, hot-wall vacuum evaporation, electrodeposition and screen-printing. One of the major problems with CdTe films has always been the difficulty of achieving high p-type doping in order to provide a low resistance frontal layer as well as to solve the ohmic contact problem through tunneling effects in a Schottky-barrier contact. Heavily p-type doping experiments with phosphorus and arsenic impurities are described. The results of the diffusion treatment on homoepitaxial CdTe films, grown by the close-spaced vapor transport method are investigated by measuring the concentration and profile of the impurities, and by analyzing the excitation spectra of luminescence at 1.8 K; it is shown that the doping efficiency is related to the structural location and the electronic activity of the impurity.

1. Introduction For 15 years a considerable amount of research has been directed toward terrestrial photovoltaic conversion of solar energy. A major objective has been to lower the manufacturing cost of cells and one obvious solution was the use of thin films devices with a creditable efficiency goal of at least 10%. At present four types of cells have reached this value in laboratory samples of small area about 1 cm2. Apart from one type based on amorphous silicon, the others involved Il—VI semiconductor compounds. The devices being studied most intensively are: heterojunctions based on CuInSe 2 using CdS or Cd0 5Zn() 2S windows and based on CdTe coupled with CdS or ITO. This is clearly indicated by the status and future prospects of DOE—SERI—Polycrystalline Thin Film Photovoltaic Program, as presented by Hermann et al. [1,5]. Sustained work is also being done on Schottky barrier, MIS-type structure and CdTe homojunction cell. A survey of Il—VI based cells has been given in Wagner [2] and Loferski [3] and in recent extensive review of CdTe solar cells published by Bube [4] and the group of Stanford University [3~I.

*

LPSES, CNRS, F-06560 valbonne, France.

CdTe is a good alternative candidate for solar cells because: Taking into account the solar spectrum it has been shown [6] that the theoretical conversion efficiency versus energy gap peaks at about 1.5 eV. The energy band gap is direct with an optical absorption coefficient greater than 3 X iO~crnt at the band edge [7,8], large enough to require only one micron of material for quasi total absorption of light. The compound sublimes congruently [9,10], both components having sufficiently high vapor pressures for growth through a vapor phase process. CdTe can be prepared in both n- and p-type —



conductivity, in contrast with all the wide band-gap Il—VI materials. Some difficulties, however, have to be overcorned: Achieving low resistivity p-type material despite severe self-compensation phenomena probably through the formation of neutral associates and precipitates [11,12,32]. Reducing the large concentration of uncontrolled residual acceptors [27] such as Cu, which is thought to be amphoteric impurity. Obtaining low resistance ohmic contact to p-type material in spite of the high value of the electron affinity [13—IS]. In addition, the particular problems associated —



0022-0248/85/$03.30 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

G. Cohen - So/al et a!.

/

Thin film CdTe solar cells

513

with the polycrystalline nature of the thin films such as photocarrier recombinations at grain boundaries and decreasing of the open-circuit voltage by junction shunts should be taken into account [16]. Nevertheless it has been generally assumed that a reasonable efficiency of 15% can be expected from well developed CdTe thin film cells. Indeed it was around 1982 that cells with efficiencies over 10% were announced, thus stimulating a renewed interest in CdTe based cells and initiating a sustained basic research effort, particularly on p-type doping of CdTe films. Many film deposition techniques have been used to prepare cells either in thin film/single crystal form or, following the line of the major objective, in totally thin-film structures. These are: Homojunction devices with p-CdTe deposited by Close Spaced Vapor Transport (CSVT) on large-grain polycrystalline n-CdTe with efficiencies in excess of 13% [17,18].

erties of the cells made by the described techniques will be analysed and the last part of the paper will be devoted to the major problem of high doping of p-CdTe films with a study of acceptor states in the epilayer volume using luminescence depth profile.

Electron beam evaporated indium—tin oxide (ITO) on p-type single crystal CdTe with presently 10.5% efficiency [19]. Chemical vapor deposition (CVD) of CdS on p-CdTe single crystal with 11.7% efficiency [20]. And for all thin-film cells: Polycrystalline CdS and CdTe layers deposited by close-spaced sublimation (CSS) showing a 10.9% (AM2) efficiency [21]. Electrodeposited CdS—CdTe heterojunction devices with 8.6% [22] and above 9% efficiencies [23]. Backwall CdS—CdTe cells made by screen-printing with a reported efficiency of 12.8% [61]. It is to be noted that all the results summarized above depend critically on the ability to increase hole density in p-type CdTe material. In this paper we shall review the different growth systems limiting ourselves to few of them which have contributed significantly in achieving the required quality thin films for solar cells; close-spaced vapor transport, close-spaced sublimation, hot wall vacuum evaporation (HWVE) and screen printing. For the others methods not detailed here one can refer to the review papers of Lopez-Otero [25] and Bube [4] and for the very promising electrodeposition process, reference should be made to the papers reported by Monsolar Inc. group [26—31]and by Ametek Inc. group [16,22,32]. The electrical prop-

He, N2 or Ar is used to either as “carrier gas” or to react with the source to form volatile compounds which are decomposed at the surface of the substrate. A schematic diagram of the usual CSVT apparatus, from ref. [56], is given in fig. Ia; the mean characteristics of the CSVT apparatus, developed at CNRS [36], are the control and the stability of the temperatures, the rapid stabilisation of the thermal program and the possibility of in-situ operations such as thermal etching, growth and doping processes (fig. lb)













2. Close-spaced vapor transport 2.]. Introduction The use of a close-spacing in chemical transport systems for the growth of epitaxial layers of pure or compound semiconductors has been emphasized by Sirtl [33], Robinson [34] and Nicoll [35]. The most important features of the technique are the use of a close-spacing between source and substrate (1 mm) between which is maintained a temperature gradient, and the fact that a gas H2,

2.2. CdTe film deposition and doping Although the deposition rate is strongly sensitive to the surface preparation and the crystallographic orientation of the substrate [37,38] as well as to the pressure or the flow rate of the reactive gas, the substrate and the source temperatures are of major importance in the process only in so far as the single crystallinity of the layer is not concerned. Saraie et al. [39] have reported the first use of the CSVT method for the growth of CdTe. Mimila-Arroyo et al. [40] and Lincot [41] made a detailed study of the influence of water vapor as an effective transport agent for the deposition

514

G. Cohen -Solal et a!.

/

Thin film CdTe solar cells

650W QUARTZ-HALOGEN LAMP

I

GAS INLET QUARTZ TUBE ________________________________________________________

a

SUBSTRATE BLOCK

S S COOLING TUBE ___________

MASK

____________

COUPLE

QUARTZ SPACER

Lfll80~

___________________

5.5. SUPPORT

SOURCE OURCE

GAS OUTLET

~

QUARTZ PLATE

BLOCK

~~~

7isolated vertical motion guides

~

or t graphite mobile heater block]

~f

______________________

r~

~

lex ib Ic cu re~ supply cable —~substrate quartz fixed ho’der

____

____

quartz mobile substrate holder

~ ____________

mobile

I

ri_I_I

crucible

~for

I

+etching thermal

~

_______________

powder doping~

for~

fixed

H

I

I

1______________

gas outlet

to vacuum pump

1~~~ourc~

graphite block heater

gas inlet

Fig. 1. Schematic diagram of the close-spaced vapor transport (CSVT) systems. (a) the usual one as described in ref. [56], (b) the CNRS apparatus as described in ref. [361.

process. Indeed water partial pressure in H2 atmosphere had a considerable effect on the growth rate, as shown in fig. 2. The dependence of the growth rate with substrate and/or source temperatures is illustrated in fig. 3. The experimental data were in good agreement with the curve calculated

using a simplified theoretical model proposed by Bailly et al. [42] in terms of thermodynamic constants of the transport reaction. Applying the model, the enthalpy variation and the activation energies of condensation and sublimation were determined L~Heq 22, Q~0~d 9.2 and QS~hl =

=

=

G. Cohen -So/al et a!. I

ID .

previously by Saraie et al. [39] and Fahrenbruch et al. [43]. The growth rate might agree with a diffusion limited transport model, as shown by Anthony et al. [44],the growth occurring through a sublimation process. At lower source temperature the “activation energy” is smaller (35 kcal/mol) and the growth occurs through a chemical reaction with H2 [41]. A probable explanation of the role

I-

4

z Qi I-

(1)

0 u.i



1O~ 10—’

iO-~ H2O

1O-~

PRESSURE

10-s

latml

Fig. 2. Growth rate of epitaxial CdTe layer versus water partial pressure in I atm of H2 from ref. [411.

10 8 7 6 C

.~

515

Thin film CdTe solar cells

31.2 in kcal/mol. It has been shown from the variation of growth rate versus source termperature (fig. 4) that a pseudo “activation energy” can be evaluated at about 45 kcal/mol as reported

I

TOTAL H2 PRESSURE: lotm E ~ E

/

SOURCE TEMPERATURE T ~6OO°C

of H20 would be that the polar molecule H20 acts as a catalyst and decreases the surface activation energies. Nevertheless Bailly’s theory was applied successfully to CSVT transport of CdS [45], of Si [46], as shown in fig. 5, and more recently to CSVT transport of GaAs [47] in agreement with the conclusions reached by Shaw [48]. A salient feature of the CSVT method is that the growth rate can be unusually high, typically 5 ~Lm/min for source and substrate temperatures of 650 and 600°C,respectively. Deposited on glass, the layers are generally

. EXPERIMENTAL

______________________________________

J \

5

CALCULATED

\ S





10.

E S

S..

E I—

4

I

U.’

4

I

I.

I

-~-

I

S

I I

0

~

z 0

I 2

~

2

0,1.

aU.’ 0.01 1_.._1__ 1.1

I

1.2

I

1.3

I

14

I

1.5

I

1.6

I

1.7 3/T K)

TEMPERATURE Fig. 3. VariationSUBSTRATE of the growth rate of CdTe(10 with substrate temperature and comparison with theory. CSVT technique, separation 1 mm from ref. [421.

II______________________________________ 1 1.1 1.2 1.3 14 1.5 1.6 SOURCE TEMPERATURE (103/TK I

Fig. 4. Variation of the deposition rate of CdTe with source temperature. CSVT technique, ~T’~ 50*C.

516

G. Cohen -So/al et a!.

CALCULATED

E 0.1

I

/

/

-

Thin film CdTe solar cells

[42,49]. Significant improvements have been achieved in preparing CdTe shallow homojunctions with conversion efficiencies in excess of 13% [17,18]. They were fabricated by deposition of an epitaxial p-type layer (0.1—0.3 ~tm thick) on an n-type CdTe substrate, generally constituted of several crystal grains. Before growth, the substrate surface is thermally etched, and a gradient of n-type impurities is obtained by outdiffusion of donors. After growth, the epilayer is submitted to

SUBSTRATE TEMPERATURE

z E 0.01

/

/

/

diffusion of As and a heavily p-type doped layer is obtained on the front. One example of impurity profiles from SIMS measurements is given in fig. 7. For AMI illumination the best cell showed 2 and a fill factor 0.85 V, ~ 20 mA/cm of 70%. The reverse saturation current was low, 1.5 x 10—10 A/cm2 and an ideality factor ~ 1.8 has been obtained. It has been noted that the photocurrent was a decreasing function of the applied voltage (it would be a constant under the classical assumptions [50]). The cells have shown =

• EXPERIMENTAL

I

(from ref.40)

I

I ~65

0.7 ~Th SOURCE TEMPERATURE (103/T K)

Fig. 5. Dependence on growth rate of silicon of source temperature. Iodine pressure 0.7 Torr. Separation I mm. CSVT technique. Experimental data from ref. [401.

=

=

=

-

no apparent degradation in performance after two years. 2.4. Thin film CdS—CdTe solar cell

polycrystalline with a dark resistivity as high as I0~—10~ ~ cm (decreasing by one or two orders of magnitude under illumination). An activation energy of about 0.7 eV was obtained from the dependence of resistivity with temperature [86].Using single crystal substrates such as CdTe or sapphire sheet, epitaxially grown layers are obtained, as it clearly seen in the (111) orientation of the diffraction pattern given in fig. 6. From a detailed study of the electrical properties of CdTe films deposited on graphite sheets Bube et al. [86] and Anthony et al. [78] have concluded that attempts to deliberately dope the films during growth ha~venot been successful. We will see later that high doping can be achieved by a post-growth diffusion of As carried out in-situ, 2.3. Solar cell

The first application of the CSVT process to CdTe solar cell fabrication led to devices (thin film CdTe on single CdS crystal) with 4% efficiency

The most interesting work in CdS—CdTe heterojunctions has been achieved by Tyan and Perez-Albuerne [21]. The CdS and CdTe films, with a thickness of 0.1 ~tm and 4 ~smrespectively, were prepared by a close-spaced sublimation technique (CSS) using the usual CSVT apparatus, the most significant difference between the two methods being that the material transport occurred by a sublimation—condensation mechanism in a low vacuum I Torr) environment. It is reported that a small amount of oxygen introduced during the deposition of the layers enhances the p-type character of the CdTe film. This unusual effect of oxygen as efficient dopant for p-CdTe, whereas the more conventional p-type dopants failed, has to be understood. The mechanism of the oxygen effect was studied by analysis of the characteristics of the cells and more especially the spectral collection efficiency [51]. Indeed oxygen had drastic effects on the performance of the cell, the efficiency varying from almost 0% up to 10% with the amount of oxygen present during the films (—

G. Cohen - So/a! et a!.

/

517

Thin film CdTe solar ce/Is

Fig. 6. Laue X-ray diffraction pattern of an epitaxial CdTe layer grown on a sapphire substrate.

10

THEORY

E

z

8.

0 I—~

z

*

\_Arsenic 6.~

-~

Indium

,,.‘

LU

I

U

•-

z

O

~

/

2

As.

~

test sample

5. S

.

~

%5

0



.5



I

.

1

1.5

2

DEPTH (micron) Fig. 7. Experimental As and In profiles from SIMS measurements and calculated curves, cf. refs. [17,24].

518

G. Cohen - So/a! et a!.

deposition. Although it was demonstrated that oxygen ensures a shallow heterojunction rather than a less efficient buried junction in the CdTe, the doping mechanism however is not clear. TzuMm Hsu et al. [52] have recently shown by wavelength modulation reflectance spectroscopy that oxygen atoms exhibit a local level in the gap with an ionization energy of about 0.1 eV. They found from JR measurements that the oxygen content was unusually high (1019_1020 cm3) but this did not result in band gap narrowing. It is to be noted that the same phenomenon, no better understood, is found in high-efficiency CuInSe 2 cells which always requires oxygen heat treatments [53,54]. The solar cell have the following structure: glass with 1n203 (0.3 ~m) + CdS (0.1 ~m) + CdTe (4 ~tm) + Au (0.05 ~sm). Under simulated illumination 2) the cells have aAM2 typical output of (75 V mW/ cm 0~. 750 2, FF 0.62 and efficiency mV, ‘sc 17 mA/cm 10.5% at room temperature. The diode factors are estimated to be 1.8 and J 0— 5.4 X 10’° 2. A simple monolithically integrated array A/cm for low-cost large-scale terrestrial use has been —

proposed by Tyan and Perez-Albuerne [55].

3. Screen printing technique 3.1. Introduction Entirely screen-printed CdS—CdTe solar cells have been prepared by the group of Wireless Res.

LIGHT

,~,

~

Glass

______________________________ CcIS ,25 ~im ~ CdTe ,5 ~im

I

~

eI~ctrode(-)

I

Carbon Subsidiary Ag grid electrode (~)

+

Cu electrode

ref. Fig. [61]. 8. Cross section structure of CdS—CdTe solar cell, from

/

Thin film CdTe solar cells

Lab. (Japan) [20]. After eight years of continuous improvements in the preparation conditions (sintering and heating cycles) and in reducing the contact resistance between the carbon electrode and the CdTe layer, a cell was obtained with a reported efficiency of 12.8% for an active area of 0.78 cm2 [57—65]. 3.2. Ce//fabrication The structure of an all screen-printed cell is shown schematically in fig. 8. A borosilicate glass was used as substrate. A CdS paste was obtained by mixing CdS powder with 9.1% by weight of CdCl2 powder (flux) and an appropriate amount of propylene glycol (as binder). After screen printing (polyester screen-80 mesh) the paste was first dried at 120°Cfor 1 h and thenThe sintered N2prepared atomosphere 690°Cfor 90 mm. CdS in film in thisat way had a thickness of about 30 ~sm and a sheet resistance about Cd 100 and Q/D. Te powders of nearly equimolar ratio were mixed with CdCl 2 (0.5 wt%) and propylene glycol then the past was screen printed (s-steel screen-400 mesh) and dried before sintered in N2 atmosphere at 600—700°C for 1 h. During the sintering Cd and Te reacted to form CdTe, and the layer thickness was 5—10 jsm. Screen printable electrode materials were developed, carbon paste containing 10 ppm Cu and an organic binder for CdTe, Ag + In paste for CdS and a Ag past for the carbon electrode. The silver electrode is connected to the silver plus indium bus-bar electrode of next cell by Ag paste applied by screen-printing to connect two cells in series. The spectral responses of the cells were strongly influenced by the preparation condition for each film and particularly by the CdTe sintering temperature; the best photoresponse was obtained for a sintering temperature of 620°Cat which correspond a sharp peak in efficiency. It was suggested that this temperature was suitable to improve the quality of the junction interface by reducing the number of interface states. The highest efficiency 2 solar simusolar cell lator was under 12.8%AM corresponding 1.5, 100 mW/cm to Voc 0.75 V, =

G. Cohen -So!a! et a!.

/

2, FF 61%. mA/0.78 The21.8 slightly largercmshort-circuit current density probably arose from the concentration effect on the front glass which resulted in under estimation of the total active area. With efficiencies in the range of 10% these heterojunction cells are very promising devices and significant improvements can be expected as soon as intensive research is conducted on understanding the structural and electronic properties of this granular-type material, =

4. Hot-wall vacuum evaporation The use of HWVE method for the deposition of CdS and CdTe films has been investigated by Lopez-Otero and Huber [25,66] and Bube et al. [67,86]. The hot-wall chamber, which is enclosed in a vacuum system is similar to that described by Lopez-Otero [25] except that all the quartz crucibles are independant and adjustable. A schematic diagram is shown in fig. 9. Four zones, with inde-

substrate carrier substrate\ heating wires

~

--‘

~

vsuthreaded rod orts rr

Thin film CdTe solar ce/Is

519

pendently controlled furnaces, allow the regulation of the source material flux, the dopant pressure and the control of stoichiometry by additional component flux. The substrate is placed on the top of the external quartz liner confining this way a volume were the evaporation and the mixing of the different vapors take place without waste of material. Although attempts to highly dope the CdTe films p-type with As, Sb, Na and Ag impurities during HWVE growth have not been successful [86], it was possible to obtain 1017 cm3 electron density in films doped with In [67]. All thin film solar cells CdS—CdTe on graphite sheet have been prepared with efficiencies greater than 6% [78]. Considerable improvement is expected in the near future. Evidence for this lies in the fact that epitaxially grown layers on sapphire exhibited resistivity and carrier mobility comparable to CdTe single crystals.

5. Electrochemical photovoltaic cells Electrochemical photovoltaic cells (EPC), which utilize the junction formed at the interface between a semiconductor (in single crystal or in thin film forms) and an electrolyte, appear an attractive alternative for low-cost photovoltaic conversion.

substrate-...

,JIJJ.1

zone

14~.~quortz 1JUJL tube

Till

~

source iCdTe or CdSI dopant zone (In I

fl~’ff

I H~-W~ * I I I~ ~ I

~

ii ~

bottle

~IUJ

Ii It1~L-~-* rad~a~~

cadmium

zone

thermocouple position

_~

* quartz

crucibles

Using single high crystal qualitywith n-CdSe065Te035 Bridgman grown cesium polysulfide solution Licht et a!. [68] reported a 12.7% conversion efficiency. With a CdSe08Te02 alloy thin film electrode efficiency in excess of 7% have been reported by Russak and Reichman [69]. However the major problem at present is still the chemical stability of the semiconductor electrode but this kind of device is certainly very promising. 6. Study of acceptor states in CdTe layer by luminescence It is well known that As and P are shallow acceptors (EAS 92 meV, E~= 68.2 meV) [83,84,88] when they substitute Te; if they are located in Cd or interstitial sites, likewise associated with others impurities or microprecipitates, one can expect that these elements, partly electri=

,

Fig. 9. Hot-waIl vacuum evaporation (HWVE) apparatus used in CdS and CdTe films deposition (schematic diagram).

520

G. Cohen. Solal ci al.

cally inactive, form deep energy levels reducing the doping concentration even below the limit of solubility. From secondary ion mass spectrometry (SIMS) performed on doped epitaxial CdTe layers grown by CSVT on CdTe single crystal substrate the chemical concentration and profile of As and P are obtained, and photoluminescence can be used to characterize their position in the crystal lattice. Furthermore using luminescence excited in resonance with a dye laser the analysis can be extended to the layer—substrate interface and the substrate itself, giving information about the contamination effects which take place during the growth (from the source or from the substrate) and the diffusion (from the working chamber) processes. 6.1. Substrate

/

Thin film CdTe solar cells

vestigated by Bube [86] and Anthony et al. [78]. They found for instance that the magnitude of the hole density did not vary either with the type or concentration of acceptor dopant in the source wafers, or with the growth rate, but did vary with the substrate temperature, with a monotonic increase from 1.5 X 1015 cm3 at 500°C to 1016 cm3 at 610°C. Nevertheless, the difficulty of making low resistance ohmic contact to p-type material used in solar cells can be solved by achieving higher p-type doping, at least in the top surface layer. It has been shown that it was possible to achieve, successfully, superficial doping to produce a hole density as high as 1.2 >< 1017 cm3 using a particular technique of solid diffusion of acceptor impurity in presence of Cd in excess. From analysis of the experimental diffusion profiles, the diffusion coefficient of As has been evaluated DA~ 2 x 10— 12 cm2/s at T= 500°C [171. Significant doping of the as-grown CdTe is achieved using a two-step diffusion process: in the first one a thin layer of dopant mixture—Cd 7As2 + As or Cd3 P2 + P in several proportions is deposited on the top of the epilayer from a source in powder form. At the end of the deposition the temperature gradient is inverted so that the doping layer is taken completely off the substrate, actually in source position. During the whole process, at substrate temperature between 500—600°C,acceptor impurities are incorporated in the CdTe epilayer by out-diffusion from the doping layer, provided the layer is not eliminated. Diffusion was investigated by photoluminescence study and the results indicated a strong dependence on the doping of the source wafers used to deposit the CdTe epilayer, although it has been shown [78] that the same layer doping was found for films deposited from differently doped sources. =

The CdTe layers are deposited on single crystal or large-grain polycrystalline CdTe wafers using the CSVT technique described previously. The substrates were prepared from a CdTe ingot In or B doped in the melt during Bridgman growth [71]. Low temperature (at 1.8 K) photoluminescence (PL) spectra exhibited strong bands of donor— acceptor pair (DAP) recombinations. The no-phonon lines peaking at 1.546 and 1.457 eV denoted a significant background of residual Group I acceptors as Li + Na [72—74]and residual Cu acceptors [75,76]. At high dopant concentrations of In or B there were no more exciton lines. Results obtained on p-type (As or P) and n-type (Al) material used as deposition sources revealed that Cu is still the major residual contaminant [76]. 6.2. Post-growth doping of CdTe films The CdTe layers were epitaxially grown on single crystal or large-grain polycrystalline CdTe substrates using the CSVT method described previously. Effective dopant transport from source to layer during CSVT growth has been reported for CdTe doped In, P on andthe Asacceptor [85]. The influence of growth with parameters density in CdTe layer grown by CSVT has been extensively in-



6.3. Arsenic doping Films grown from a p-type CdTe: As (7 x 1016 cm 3) source, and submitted to arsenic post-diffusion3)process, had a highly doped surfaceIn ~the1017 as determined by SIMS analysis. PL cm spectra are found weak ex~itonlines and a strong donor—acceptor band, the no-phonon peaking at

/

G. Cohen -So!al ci a!.

1.511 eV. Under the same conditions, films grown from n-type CdTe: Al (— 1017 cm3) source were

less doped, as indicated by radiative recombination spectra of free and neutral acceptors bound excitons (A°X) as well as excitons bound to neutral donors (D°X) with their excited states (fig. 10). Using the techniques of excitation spectroscopy (ES) and selective pair luminescence (SPL) [79,80] one can observe in fig. lithe resonant emission of donor—acceptor pairs neutralized in the 2S3/2 acceptor state. For distant DA pairs the extrapolated transition energy E(1S3/2 —9 2S3/2) 73 meV is obtained as from a two-hole spectroscopy of As bound exciton [75,87]. The peak at 1.511 eV can be attributed without ambiguity, to the As acceptor impurity on Te sites. It is to be noted here that other peaks are often observed in selectively excited PL spectra and attributed to donor—phosphorus acceptor transitions. This indicated that the deposition system was responsible for the adjunction of the observed phosphorus in spite of the high temperature bakeouts of the graphite heaters. This “memory” effect is also responsible for an As peak present in PL spectra of P diffused layers where the donor—acceptor band at 1.533 eV is related to substitutional ~Te (fig. 12). As the intensity of this band is three times smaller than the As band, it is concluded that the compensation of P in CdTe is more pronounced, probably through the formation of neutral associates and precipitates [77] as analyzed by Selim and Kroger [70] and discussed by Marfaing [81]. When the energy of the EL was reduced to

521

Thin film CdTe solar cells

B

E(ls—2s

.<

3121meV



DAPAs 72 1s...2s3~ $ 1.584

~2

~

~

101 1.53

=

~

1.587 1.590 DAP:Alcalins

1.55

PHOTON ENERGY leVi _______________________________________ Eexc teVi 1.5840 1.5867 1.5883 1.5893 1.5899 1.5906 DAPns11s3,~73.2 72.6 72.5 71.8 71.7 71.4

me Fig. 11. Selectively excited DAP Spectra of As doped layers of CdTe (excitation energy 1.5867 eV). IS1/2 —2S3/2 energy difference as a function of the excitation energy.

values smaller than the absorption energy of bound excitons (< 1.58 eV), the bulk substrate material spectra were obtained with Cu, Li and Na residual acceptors bands. Now for higher energy (1.585 eV) only the volume of the layer is involved and the copper peak vanished whereas the As peak appeared; the 1.545 eV band in fig. ii indicated that the residual group I acceptors have rapidly diffused into the layer from the substrate during the growth and diffusion processes. From these results Cd Te

Do.

~i~9AOP

2

1.6 146

1A8

1.5

1.52

154

1.56

1.58

1.6

PHOTON ENERGY 1eV) Fig. 10. Photoluminescence spectra of As-doped layer grown from n-CdTe Al source,

1.55

1.5

PHOTON ENERGY

1.45

1.4

1eV)

Fig. 12. Photoluminescence spectra of P-doped layer showing an As peak due to a “memory” effect of the deposition chamber.

522

G. Cohen -Solal e a!.

it is concluded that Cu is not transported from source to layer in the CSVT method. 6.4. Residual impurities The properties of residual donors in the layer were difficult to analyze. Very close to the surface the D°X peak was quenched as the penetration depth increased in favour of a D°h band (i.e. holes recombination on neutral donors) peaking at 1.592 eV) as observed in bulk CdTe crystal. The resonant excitation of D°h generates two-electron transitions, displaced from excitation light (EL) by the difference in energy between the ground and excited 2S and 3S states, 11.2 meV and 13.4 meV respectively. This led to an estimation of 14.8 meV for the binding energy of the residual donor (fig. 13). Slightly deeper than In, this donor is not yet identified. When the energy of the EL is reduced, the latter two-electron lines disappear and are replaced by a broad replica (2.5 meV in width), with a peak shifted by 12.3 meV by the excitation energy. Attributing this peak to another residual donor never observed either in source or in substrate material it would be expected to obtain a “chemical shift” of more than 3 meV. In fact the origin of this replica is likely to be related to internal electrical effect, near the junction [82]. Indeed the binding energy of the fundamental and excited —



/ Z1000

D(ls—2s)

z zU.’

u

+1 D)ls—3 8)A / -‘

~

10 1:58

Thin film CdTe solar cells

states of donors in an internal field are reduced (Pool—Frenkel effect) and even at low bias donors could no more link electrons in the excited states. Since the excitation of neutral donors, moved the bound electron directly toward the continuum state, a lower transition energy is indeed needed.

References [1] A. Hermann, K. Zweibel and R. Mitchell, in: Conf. Rec. 17th IEEE Photovoltaic Specialists Conf., 1984 (IEEE, New York, 1984) p. 910. [2] S. Wagner, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 685. [3] J.J. Loferski, Conf., in: Proc. 1st Intern. Science and Engineering Kobe, 1984, p.Photovoltaic 373. [41RH. Bube, in: Proc. Symp. on Materials and New Process Technology for Photovoltaics, Eds. J.A. Amick, V.K. Kapur and J. Dietl. Vol. 83—11 (Flectrochem. Soc., Pennington, NJ, 1983) p. 359. [5] A. Hermann, L. Fabick, K. Zweibel and R. Hardy, in: Conf. Rec. 17th IEEE Photovoltaic Specialists Conf., 1984 (IEEE, New York, 1984) p. 840. [6] J.J. Loferski, J. Appi. Phys. 27 (1956) 777. [7] K.W. Mitchell, AL. Fahrenbruch and RH. Bube, J. AppI. Phys. 48 (1977) 829. [8] K. Semiconductors Semimetals, 13, Eds.Zanio, R.K. in: Willardson and AC.andBeer (AcademicVol. Press, New York, 1978). [9] P. Goldfinger and M. Jeunehomme, Trans. Faraday Soc. 59 (1963) 2851.

[101A.J.

Strauss and R.F. Brebrick, J. Physique 29 (1968)

C4-21.

[11] T.J. Magee, J. Peng and J. Bean, Phys. Status Solidi (a) 27 (1975) 557. [12] J. Saraie, M. Kitagawa and T. Tanaka, J. Electrochem. Soc. 126 (1979) 2226. [13] T.C. Anthony, AL. Fahrenbruch and RH. Bube, J. Fleetron. Mater. 11 (1982) 89. [14] C. An, H. Tews and G. Cohen-Solal, J. Crystal Growth 59 (1982) 289. [15] H. Tews and C. An, J. App!. Phys. 53 (1982) 5339. and ES. Yang, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 917. [161 G.F. Fulop, P.V. Meyers, C.H. Lju, S.M. So, W. Huang [171G. Cohen-Solal, D. Lincot and M. Barbê, in: Proc. 4th European Photovoltaic Solar Energy Conf., Stresa, 1982 (Reidel, Dordrecht, 1982).

,— .

100 100

/

/

1.59

PHOTON ENERGY leVi Fig. 13. Two-electron transition for: (a) 1.5940 eV (near D°X) excitation energy, (b) 1.5933 eV of excitation energy.

[18] M. Barbé, F. Bailly, D. Lincot and G. Cohen-Solal, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 1133. [19] J.G. Werthen, AL. Fahrenbruch, RH. Bube and J.C. Zesch, J. Appi. Phys. 54 (1983) 2750. [20] K. Yamaguchi, N. Nakayama, H. Matsumoto and S. Ikegansi, Japan. J. Appi. Phys. 16 (1977) 1203.

G. Cohen - Solal ci at

/

[21] Y.S. Tyan and E.A. Perez-Albuerne, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 794. [22] G.F. Fulop, M. Doty, P. Meyers, J. Betz and C.H. Liu, AppI. Phys. Letters 40 (1982) 327. [23] B.M. Basol, J. Appl. Phys. 55 (1984) 601. [24] F. Bailly, M. Barbé, G. Cohen-Solal and D. Lincot, J. Physique 44 (1983) 489. [25] A. Lopez-Otero, Thin Solid Films 49 (1978) 3. [26] B.M. Basol, O.M. Stafsudd, R.L. Rod and E.S. Tseng, in: Proc. 3rd European Photovoltaic Solar Energy Conf., Cannes, 1980 (Reidel, Dordrecht, 1980) p. 878. [27] B.M. Basol and O.M. Stafsudd, Solid State Electron. 24 (1981) 121. [28] B.M. Basol and O.M. Stafsudd, Thin Solid Films 78 217 (1981). [29] B.M. Basol, ES. Tseng, R.L. Rod, S. Ou and O.M. Stafsudd, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 805. [301 B.M. Basol, R.L. Rod and ES. Tseng, in: Proc. 4th European Photovoltaic Solar Energy Conf., Stresa, 1982 (Reidel, Dordrecht, 1982) p. 719. [31] S.S. Ou, O.M. Stafsudd and B.M. Basol, Solid State Eleciron. 27 (1984) 21. [32] H. Hwang, H.C. Card and ES. Yang, AppI. Phys. Letters 36 (1980) 315. [33] E. Sirt!, J. Phys. Chem. Solids 24, 1285 (1963). [34] PH. Robinson, RCA Rev. 24 574 (1963). [35] F.H. Nicoll, J. Electrochem. Soc. 110 (1963) 1165. [36] J.M. Laroche and G. Cohen-Solal, Rev. Physique App!. 18 (1983) 515. [37] D.W. Shaw, in: Proc. 2nd Intern. Symp. on GaAs and Related Compounds, Dallas, TX, 1968, Inst. Phys. Conf. Ser. 7 (Inst. Phys., London, 1969). [38] A.E. Blakeslee, Trans. Met. Soc. AIME 00 (1969) 245. [39] J. Saraie, M. Akiyama and T. Tanaka, Japan. J. Appl. Phys. 11(1972)1758. [40] J. Mimila-Arroyo, A. Bouazzi and G. Cohen-Sola!, 2nd Intern. Symp. on CdTe, Strasbourg, 1976. [41] D. Lincot, Thesis, P. et M. Curie University Paris VI (1981). [42] F. Bailly, G. Cohen-Solal and J. Mimila-Arroyo, J. Electrochem. Soc. 126 (1979) 1604. [43] AL. Fahrenbruch, V. Vasilchenko, F. Buch, K. Mitchell and RH. Bube App!. Phys. Letters 25 (1974) 605. [44] T.C. Anthony, AL. Fahrenbruch and RH. Bube, Progress Report No. 7, Basic Energy Sciences, DOE, DE-AT 03-76 ER 700—47, Dec. 1982, cited in ref. [4]. [45] B.J. Curtis and H. Brunner, J. Crystal Growth 6 (1970) 269. [46] J.E. May, J. Electrochem. Soc. 112 (1965) 710. [47] F. Chavez, J. Mimila-Arroyo, F. Bailly and J.C. Bourgoin. J. App!. Phys. 54 (1983) 6646. [48] D.W. Shaw, J. Electrochem. Soc. 117 (1970) 683. [49] K. Mitchell, AL. Fahrenbruch and RH. Bube, J. Vacuum Sci. Technol. 12 (1975) 16.

Thin film CdTe solar cells

523

[50] L.M. Ki!gren, in: Proc. 2nd European Photovoltaic Solar Energy Conf., Berlin, 1979 (Reidel, Dordrecht, 1979) p. 344. [51] Y.S. Tyan, F. Vazan and T.S. Barge, in: Conf. Rec. 17th IEEE Photovoltaic Specialists Conf., 1984 (IEEE, New York, 1984) p. 840. [52] Tzun-Min Hsu, Rush-Jiun Jih, Shu-Fang Fan, Pei-Chi Lin, Herng-Sin Ueng and Huey-Liang Hwang, in: Conf. Rec. 17th IEEE Photovoltaic Specialists Conf., 1984 (IEEE, New York, 1984) p. 846. [53] R.A. Mickelsen, W.S. Chen and L.F. Buldhaupt, Final Report Boeing Aerospace Co., Seattle, WA, 1982. [54] YR. Hsiau, W.S. Chen, R.A. Mickelsen, J.M. Stewart V. Lowe and H. Rotwarf, Final Report Boeing Aerospace Co., Seattle, WA, 1982. [55] Y.S. Tyan and E.A. Perez-Albuerna, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 928. [56] J.G. Werther, T.C. Anthony, AL. Fahrenbruch and R.H. Bube, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 1138. [57] H. Uda, H. Taniguchi, M. Yoshida and T. Yamashita, Japan. J. App!. Phys. 17 (1978) 585. [581 S. Ikegami, H. Matsumoto, H. Uda, Y. Komatsu, A. Nakano and K. Kuribayashi, in: Proc. 5th European Photovoltaic Solar Energy Conf., 1983 (Reidel, Dordrecht, 1983) p. 740. [59] H. Matsumoto, K. Kuribayashi, H. Uda, Y. Komatsu, A. Nakano and S. Ikegami, Solar Cells 11(1984) 367. [60] H. Matsumoto, A. Nakano, Y. Komatsu, H. Uda, K. Kuribayashi and S. Ikagami, Japan. J. AppI. Phys. 22 (1983) 269. [61] H. Uda, H. Matsumoto, Y. Komatsu, A. Nakano and S. Ikegami, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 801. [62] K. Kuribayashi, H. Matsumoto, H. Uda, Y. Komatsu, A. Nakona and S. Ikegami, Japan. J. App!. Phys. 22 (1983) 1828. [63] H. Matsumoto, K. Kuribayashi, Y. Komatsu, A. Nakano, H. Uda and S. Ikegami, Japan. J. Appl. Phys. 22 (1983) 891. [64] H. Uda, A. Nakano, K. Kuribayashi, Y. Komatsu, H. Matsumoto and S. Ikegami, Japan. J. App!. Phys. 22 (1983) 1822. [65] H. Matsumoto, A. Nakano, Y. Komatsu, H. Uda and S. Ikegami, in: Proc. 1st Intern. Photovoltaic Science and Engineering Conf., Kobe, 1984, p. 393. [66] A. Lopez-Otero and W. Huber, Surface Sci. 86 (1979) 167. [67] W. Huber, AL. Fahrenbruch, C. Fortman and RH. Bube, J. App!. Phys. 54 (1983) 4038. [68] S. Licht, R. Tenne, G. Dagan, G. Elodes, J. Manassen, D. Cahen, R. Triboulet, J. Rioux and C. Levy-Clement, AppI. Phys. Letters 46 (1985) 608. [69] MA. Russak and J. Reichman, in: Conf. Rec. 16th IEEE Photovoltaic Specialists Conf., 1982 (IEEE, New York, 1982) p. 1057.

524

G. Cohen - Sola! ci at

[70] F.A. Selim and F.A. KrOger, J. Electrochem. Soc. 124 (1977) 401. [711 The CdTe sing!e crystal ingots were grown like usual!y by R. Triboulet. [72] F.T.J. Smith, Met. Trans. 1 (1970) 617. [73] G. Neu, Y. Marfaing, R. Legros, R. Triboulet and L. Svob, J. Luminescence 21(1980) 293. [74] J.L. Pautrat, N. Magnea and J.P. Faurie, J. App!. Phys. 53 (1982) 8668. [75] E. Molva, J.P. Chamona! and J.L. Pautrat, Phys. Status Solidi (b) 109 (1982) 635. [76] D. de Nobel, Philips Res. Rept. 14 (1959) 361, 430. [77] J.C. Pfister, Rev. Physique App!. 15 (1980) 707. [78] T.C. Anthony, A.L. Fahrenbruch, MG. Peters and R. Bube, J. App!. Phys. 57 (2) (1985) 400. [79] R.A. Street and W. Senske, Phys. Rev. Letters 37 (1976) 1292. [801 H. Tews, H. Venghaus and P.J. Dean, Phys. Rev. B 19 (1979) 5178.

/

Thin film CdTe solar cells [81] Y. Marfaing, Progr. Crystal Growth Characterization 4 (1981) 317. [82] R.E. Hollingsworth and JR. Sites, in: Conf. Rec. 17th IEEE Photovoltaic Specialists Conf., 1984 (IEEE, New York, 1984) p. 1404. [83] H. Venghaus and P.J. Dean, Phys. Rev. B21 (1980) 1596. [84] T. Hattori, S. Nakashima and A. Mitsuihi, in: Proc. 13th Intern. Conf. on the Physics of Semiconductors, Ed. F.G. Fumi (North-Holland, Amsterdam, 1976) p. 635. [85] J. Mimila-Arroyo, Y. Marfaing, G. Cohen-So!a! and R. Triboulet, Solar Energy Mater. 1(1979)171. [86] RH. Bube, AL. Fahrenbruch, R. Sinclair, T.C. Anthony, C. Fortman, W. Huber, Chun-Teh Lee, T. Thorpe and T. Yamashita, IEEE Trans. Electron Devices ED-31 (1984) 528. [87] P.J. Dean, H. Venghaus, J.C. Pfister, B. Schaub and J. Marine, J. Luminescence 16 (1978) 363. [88] J.P. Chamonal, E. Molva and J.L. Pautrat, Solid State Commun. 43 (1982) 801.