- Email: [email protected]
Cu thin films sequentially deposited
MATERIALS SCIENCE & ENGWEERING
ELSEVIER
B
Materials Science and Engineering B45 (1997) 69-75
Physico-chemical characterization of CuAlSe, films obtained by reaction, induced by annealing, between Se vapour and Al/&/AI.. . Cu/Al/‘Cu thin films sequentially deposited S. Marsillac Equipe
Couches
Minces
et Matkiaux
*, J.C. Bernede, C. El Moctar,
J. Pouzet
Nomeaux,
44072 Nantes,
LPME,
FSTN,
2 rue de la HoussiniPre,
C6de.x
03, France
Received 24 May 1996; accepted 14 November 1996
Abstract CuAlSe, thin films have beensynthesizedby selenizationof thin Cu and Al layers sequentiallydepositedby evaporation under vacuum. The films have been characterizedby X-ray diffraction, scannin, mand transmissionelectron microscopy, microprobe analysis,photoelectronspectroscopyand Raman diffusion. The resultsare comparedto those obtained with a CuAlSe, reference powder. It is shownthat CuAlSe, films are obtained with someCu,.,Se phasespresentat the surface.Thesesurfacephasesare suppressed by chemicaletching in a KCN solution. At the end of the process,the XRD spectrumdemonstratesthat textured CuAlSe, films have beenobtained with preferential orientation of the crystallites along the (112)direction. TEM study hasshown that both randomly and (112) oriented microcrystallitesexist in the layers. The films are nearly stoichiometric,but their surface is quite rough. The Raman patterns are in good accordancewith the reference.The XPS spectrashow that the binding energies of the elementsare in good agreementwith bonds of CuAlSe,. 0 1997Elsevier ScienceS.A. Keywords:
CuAlSe,; Chemical etching; Raman patterns
2. Experimental
1. Introduction Ternary I-II-VI,
chalcopyrite semiconductors have
received considerable attention in recent years because of their applications in photovoltaic devices [l]. If the CuInSe, is the most extensively studied compound of the family, some other ternary chalcopyrites have attracted much attention in recent years because of their large potential applications. Among them, CuAlSe,, which is a large band gap material-Eg = 2.67 eV [2]-that can be used as an emitting layer of blue light photoelectroluminescent devices [3] or as a window in heterostructures photovoltaic devices [4]. If for the former application heteroepitaxial layers are needed [3], in the case of the latter application polycrystalline layers can be used. In this paper, we describe a very simple and cheap technique which allows us to obtain well crystallized CuAlSe, polycrystalline layers. * Corresponding
author.
0921-5107/97/$17.00 0 1997 Elsevier Science S.A. All rights reserved. P!ns0921-5107(96)02031-4
2.1. Film preparation CuAlSe, films were obtained by selenization (annealing under selenium atmosphere of Cu/Al.. .Al/Cu/Al thin layers sequentially deposited). The substrates were polished soda lime glass or silicon, chemically cleaned. Then the substrates were out gassed ‘in-situ’ prior to deposition by heating to 400 K for 2 h. The Cu and Al films were deposited in vacuum (pressure 5 10 -’ Pa) by evaporation from two tungsten crucibles. The purity of Al was 99.99%, that of Cu was 99.999% and that of Se was 99.999%. Layers of Cu and Al were deposited sequentially on the heated substrates (400 K < T, < 450 K). The CuAlSe, films were then synthesized by annealing in selenium atmosphere.
70
S. Marsillac
et al. /Materials
Science
During the deposition, a rotating substrate holder was successively positioned in front of the Cu evaporation source and in front of the Al evaporation source according to the sequence. The evaporation rates and film thicknesses were measured in-situ by the vibrating quartz method, the quartz head being joined to the substrate holder. The number of layers for each constituent varied from six to ten in order to deposit Al/&/AI.. .Al/Cu structures. The relative thicknesses of the layers were calculated to achieve the desired atomic ratio Cu/Al = 1.1. After deposition, the multilayer structures were introduced with a small amount of Se in a Pyrex tube. This tube was then sealed under vacuum (10 - ’ Pa). The samples were then annealed under selenium atmosphere. Selenium has a high vapour pressure, therefore during the annealing the selenium was in the vapour state. During this process the selenium pressure was about lo6 Pa. The annealing temperature varied from 823 to 853 K while the annealing time was between 24 and 45 h. The results obtained after these processes will be discussed below. The thickness of the Cu layers was about 10 nm and its evaporation rate was 0.2 nm s-l, in the case of aluminium they were 13 nm and 0.2 nm s - ’ respectively. The total thickness of the films after reaction with selenium was measured by a mechanical tiger. In order to check the results obtained with thin i?lms, a CuAlSe, powder has also been synthesized. The powder was prepared by mixing quantities of high purity copper, aluminium and selenium in the atomic ratio 1.l Cu, 1 Al, 2.5 Se. The mixture was sealed in an evacuated silica tube (10 - 3 Pa) and heated at 1373 K for 48 h. After cooling, the solid was reduced into powder. 2.2. Characterization
techniques
A large number of experimental techniques have been used for characterization of the layers: Scanning electron microscopy using the 6400 F JEOL field emission microscope (SEM). Surface topography was examined by scanning electron microscopy and checked by the mechanical finger. Another electron microscope equipped with an electron microprobe was used to check the composition of the samples. Transmission electron microscopy (TEM) was performed with a ‘JEM 120 CX’ equipped with a tungsten tiament at 120 kV to evidence the microstructure and to confirm the texture axis orientation determined by the XRD investigations. For examination by TEM, the films were chemically removed from the glass substrates and then attached to copper grids.
and Engineering
B45 (1997)
69-75
The structure of the films was examined using an analytical X-rays system type DIFFRACT AT V3. 1 SEMENS. which uses a graphics program EVA. The wave length 1 was 0.15 406 nm. The grain size, D, was estimated from the full width at half maximum (FWHM) of the diffraction peaks. Photoelectron spectroscopy (XPS) measurements were performed with a magnesium X-ray source (1253 eV) operating at 10 kV and 10 mA. Data acquisition and treatment were realized using a computer and a standard program. The XPS analysis was made at the University of Nantes, CNRS. We proceeded to in-depth study by recording successive XPS spectra obtained after ion etching for short periods. Using an ion gun, sputtering was accomplished at pressures of less than 5 10B4 Pa with a 10 mA emission current and 5 kV beam energy. The Ar+ ion beam was rastered over the entire sample surface. At the surface of the films there is a carboncarbon bond corresponding to surface contamination. In the apparatus used this C-C bond had a well defined position at 284.8 eV and the carbon peak was used as a reference to estimate the electrical charge effect as the film samples were on glass. Raman spectra were measured at room temperature using a BRUKER model RFS 100; the spectra were excited by a (Nd: YAG) laser with a wavelength of 1064 nm. The laser power and the scale number necessary to obtain good signal to noise ratio were 100 and 500 mW, respectively.
Experimental
results
3.1. Checking of the metallic deposited layers
After deposition of the Al/Cu... sequences, the averaged composition of the superposed layers was estimated by microprobe analysis in order to check the validity of the atomic ratio Cu/Al before annealing under selenium atmosphere. Since it has been shown that a small Cu excess improves strongly the crystallization of the ternary chalcopyrites compounds, only the samples with composition 53 + 2 at.% of Cu were used for the thin film realization. The XRD pattern of such samples exhibits two small peaks which can be attributed to the Al and the Cu. The intensity of these peaks is small, while their full width at half maximum is quite large, which shows that the metallic layers are poorly crystallized with very small crystallites. 3.2. Structural
charncterisation
When possible,
the experimental
results obtained
S. MarsilIac
et al. /Materials
Science
and Engineering
B45 (1997)
69-75
71
on thin fYms have been compared to those obtained with the reference powder. Just after synthesis the XRD spectrum of the CuAlSe, powder exhibits only diffraction peaks which belong to the CuAlSe2 compound (Table 1). However, as we can see in Table 1, two small peaks appear in the powder but not in the reference (JCPDS 44-1269). It must be noticed that these 2 peaks appear also in CuInSe, (JCPDS 40-1487), another ternary chalcopyrite compound, and also obey to the extinction law h + Jc + I= 2 a+ 1. No other peaks, from binary compounds for example, are detected. Since the composition value given by the EPMA (Cu=26 Al=24 and Se= 50 at.%) shows an excess of Cu and Se: we can suppose that Cu,,Se exists in the powder in an amorphous state or in microcrystallites embedded in the powder (the EMPA technique will be discussed below). Therefore, as is usually done in the case of CuInSe, [5], the binary compounds present in the powder have been etched in a 0.1 M KCN solution for some minutes, afterwards only the peak intensities have changed in the XRD spectrum (Table l), while a stoichiometric composition is measured by microprobe analysis (Cu = 25, Al = 25 and Se = 50 at.%). Therefore, the XRD spectrum of the powder has been used as a reference (Table 1). The XRD of thin tirns are reported in Fig. 1. It can be seen in Fig. la that just after annealing, CuAlSe, but also CuZMsSe and Se are present in the films. The surface of these layers appears to be quite inhomogeneous (Fig. 2a), small heaps and large grains of about 1-5 urn size appear, randomly distributed at the surface of the l?lms. It has been shown earlier that annealing under selenium atmosphere in a Pyrex tube induces some selenium condensation at the surface of the layer during the cooling of the tube [6]. This selenium excess at the surface of the layer was sublimated by annealing the samples under dynamic vacuum for 6 h at 570 K After this annealing the Se peak has disappeared but those related to Cu,-,Se are still there (Fig. lb) as the large grains residing on the surface. This binary phase is etched with a KCN solution 0.1 M). The films were etched for periods of 1 min and X-rays were recorded after each treatment until the Cu,,Se peaks disappear (Fig. lc). Then the visualization of the surface of the layer shows that the large grains have disappeared (Fig. 2b). As shown by the X-ray diffraction spectrum (Fig. lc), at the end of the process, only the (112) peak is clearly visible, which shows that the majority of the large crystallites are preferentially oriented along the (112) direction. However, as shown by SEM in Fig.
12
S. Mamillac
et al. /Materials
Science
and Engineering
B45 (1997)
69-75
2(b), there is a lot of small crystallites randomly oriented. This fact has been corroborated by TEM study. A transmission electron micrograph of CuAlSe, layer is reported Fig. 3. It can be seen that larger grains are embedded in small grain domains. The SAD of these domains is shown in Fig. 4(a), it corresponds to randomly distributed microdomains. It can be seen in Table 2 that these microcrystalites have the chalcopyrite structure of CuAlSe,. The SAD of the large grains is shown in, Fig. 4(b), rectangles with d, = 3964 A and L&= 1864 A as characteristic distances appear; these
Fig. 2. Scanning electron micrograph of a CuAlSe, thin film (a) after annealing under selenium atmosphere and (b) after chemical etching by a 0.1 mole KCN solution.
values lead to the zone axis [I121 with (1iO) and (222) as basis vectors. The extinctions which occur are the usual extinction for tetragonal compounds, i.e. for 12+ k+1=2n+l.
Fig. 1. XRD pattern of Al/Cu/Al...Cu/Al/Cu thin film (a) after annealing under Se atmosphere, (b) after 1st post-treatment (annealing under dynamic vacuum) and (c) after 2nd post-treatment (etching by a KCN solution).
Fig. 3. TEM microphotograph.
S. MarsilIac
et al. /Materials
Science
and Engineering
B45 (1997)
73
69-75
4;: bp*=_n..-. ;. $g+;io& =I;-
-
c---. p=/-:
...-I._---_-
i)LI. _
iy
“-hu.‘.
--., ~A&$,‘~
.,.;-.:. +-.*~
‘UN
I /
120 ,... ~-
i
1 _
9ll ._-..--.--...-
;
i, ~“*jyy
t
t
--..
_-._
:j,
i (L-.,j
-+
sb cE(eV) __ -.---“A
f :
Fig. 5. A12p, Cu3p, Na2s and Se3d XPS peaks of (a) CuAlSe, thin films before KCN treatment and (b) CuAlSe, thin films after KCN treatment.
Fig. 4. (a) SAD of a disordered domain and (b) SAD of a large crystallite.
3.3. Chemical characterization
It has been shown above that CuAISe, films crystallized in the chalcopyrite structure can be obtained by our procedure. The chemical state of these films has been studied by microprobe analysis, X-ray photoelectron spectroscopy and Raman diffusion. Since the Se L, and Al K, rays overlap, a long acquisition time ( > 400 s) is necessary to get reproducible results for quantitative measurements by EPMA. It can be seen that at the end of the process the films are stoichiometric 24 < Al < 26; 24 < Cu < 26; 49 < Se < 51. The samples have also been studied by XPS; however, quantitative information is difficult to obtain because the binding energies for Cu and Al are adjacent to each other (Cu3s and A12s; Cu3p and A12p; etc). Moreover, the binding energy and the sensibility factors of Cu2p,,,, A12p and Se3d peaks, which are usually used for quantitative analysis, are very different, which induce uncertainties. XPS lines are shown in Fig. 5 for CuAlSe, thin tims before and after Table 2 XPS results for CuAlSe, powder and thin film after KCN treatment Sample
Cu2psis (ev)
Se Al
cu CuAlSe, Powder CuAlSe, Thin Elm
A12psjz (ev)
Se3d (eV) 55.5
72.7 932.4 932.3 932.4
74.7 74.6
55 55.2
KCN treatment. The binding energies are reported in Table 3 as well as the position of the lines for the elements. It can be seen that there is a good agreement between the reference powder spectra and those of the layers. For the discussion below the carbon peak of contamination at the surface of the samples has been taken as reference: Cls = 284.8 eV. The Se3d peak is situated at 55 eV while the binding energy of the Cu2p,,, is 932.3 eV. The Al peaks are adjacent to Cu peaks (A12p and Cu3p; Al2s and CUES), therefore, we have proceeded to a decomposition. It can be seen in Fig. 6 that the peak situated at 74.5 eV can be attributed to A12p while the other situated at around 77 eV corresponds to Cu3p. The binding energy of the different elements Se3d, A12p and CU~P~,~ are, respectively, 55.3, 72.65 and 932.4 eV [5]. The electronegativity of Se being the highest of the three elements, its binding energy in the CuAlSe, compound should be smaller than that of the element alone, while that of the cations should be higher. This is the case of the A12p peak. In the case of Cu2p,,,, it is well known that its binding energy is poorly sensitive to its oxidation state. Moreover, the binding energies obtained for Se and Cu are in good accordance with the values measured on other compounds of the same ternary chalcopyrite family such as CuInSe, [6]. The iilms have also been studied by XPS before KCN chemical etching. The result depends on the substrate. When Si is used as a substrate the XPS spectra are the same as those discussed above. When a soda lime glass is used a new peak appears at 66 eV (Fig. 5a). In fact, simultaneous to this peak, another one situated at 1072 eV is also detected. This last peak has been already discussed in the case of CuInSe, [7]: it corresponds to Na diffusion from the soda lime substrate towards the surface of the layer. Therefore the peak situated at 66 eV can be attributed to Na 2s. The intensity of these Na peaks decreases when the etching
14
S. Mamillac
et al. /Materials
Science
and Engineerirzg
B45 (1997)
69-75
Table 3 TEM results Circles
d(lO-’
nm)
CuAlSe, Reference
d(lO-’ W)
nm)
3.21
1.96
1.66
1.37
1.13
0.97
3.21
1.96 (204)
1.66
(116)
1.37 (008)
1.13 (228)
0.97 (1 1 11)
(112)
time increases which shows that Na is accumulated mainly at the surface of the layer. Fortunately Na is soluble in KCN and the corresponding peaks disappear after chemical etching (Fig. 5b). The pollution of the thin film has been controlled by the evolution of the carbon and oxygen peaks after etching. It has been seen that these two peaks disappear after 1 or 2 min of etching. Therefore, we can conclude that there is no strong pollution in our thin films. The chemical state of the films has been also checked by Raman diffusion in order to confirm the CuAlSe, occurrence. The results obtained are presented in Fig. 7. It can be seen that the spectrum of the reference powder and that of the layers after KCN treatment are similar. The main peaks obtain are at 26, 85, 184, 211 and 234 cm-’ which is in good accordance with the results of Kamata and Co11 [8], except for the line at 26 cm-l that they did not find. Anyway, a reference spectrum of Cu,$e shows that no line appears at 26 cm- ‘; this peak has also been attributed to CuAlSe, compound.
microcrystallite disordered matrix. If the large crystallites are oriented along the (112) direction, the microcrystallites are randomly oriented. These results are corroborated by those obtained by transmission electron microscopy as well as by Raman spectroscopy. The optimum temperature for the annealing under selenium atmosphere appears to be around 820 K. The annealing duration is optimum at about 24 h. The XPS spectra show that the binding energy of the metal anions is increased while that of Se has decreased by comparison to the reference element. These modifications correspond to electron exchange from the anion to the cation. When the films are obtained on soda lime glass, there is Na at the surface of the layer. The good crystallization and texturation of the layers can be related to this Na diffusion, since it has been shown in the case of CuInSe, [9] that the crystalline quality is strongly enhanced by Na diffusion through the layer. Therefore, these layers appear to have the necessary quality to be used as a buffer layer in solar cells.
4. Discussionand conclusion
Oriented CuAlSe, thin lilms have been obtained by thermal annealing under selenium atmosphere of Cu/ Al/Cu.. .Cu/Al thin sequentially deposited metallic layers. At the end of the process (annealing under selenium atmosphere, annealing under dynamic vacuum, chemical etching) the CuAlSe, films show properties close to those of CuAlSe, reference powder. The films exhibit large grains randomly distributed in a
Wavenumben
Wn\enumbcrs Ek-.-.
Fig. 6. Decomposition of the A12p peak and the Cu3p peak.
~-------
(cm.‘)
(cm-l)
-----p--..I
Fig. 7. Raman diffusion spectra after KCN treatment of: (a) the reference powder and (b) a CuAlSe, film.
S. Marsillac
et al. /Materials
Science
Acknowledgements
The authors wish to thank Mr P. LeRay improvments in the depositing apparatus.
for the
References [I] K.J. Bachmann, H.L. Hwang and C. Schwab, Non-stoichiometry in Semiconductors, EMRS 1991, Spring Meeting, North Holland, 1992. [2] S. Shirakata, S. Chichibu, S. Matsumoto and S. Isomura, Jpn. J. Appl. Phys., 32 (1993) L167.
and Engineering
B45 (1997)
69-75
75
S. Shirakata, S. Isomura, Y. Harada, M. Yuchida, S. [31 S. Chichibu, Matsumoto, H. Higuchi, J. Appl. Phys. 77 (1995) 1225. V.Y. Rud’ and Y.V. Rud’, Tech. Phys. Lett., 20 [41 I.V. Bodnar, (1994) 317. [51 C.D. Wagner, WM. Riggs, M.E. Davis, S.F. Moulder and G.E. Muilenberg, Hand-Book of X-ray Photoelectron Spectroscopy CD, Perkin Elmer, Eden Prairie, MN, 1979. Fl R. Scheer and H.J. Lewerenz, J. Vat. Sci. Technol., AlZ(1) (1994) 56. in H.S. Stephens [71 C. Heske, R. Fink, D. Jacob and E. Umbach, (Ed.), Proc. 13th European Photovoltaic Solar Energy Conference, Nice, France, Bedford UK, 1995, p. 2003. Fl A. Kamata and S. Chichibu, J. Appl. Phys., 77 (1995) 10. Bless, F. Pfisterer, M. Schubert and T. Walter, Prog. PI H.W. Photovolt. Res. Applications, 3 (1995) 3.
ELSEVIER
B
Materials Science and Engineering B45 (1997) 69-75
Physico-chemical characterization of CuAlSe, films obtained by reaction, induced by annealing, between Se vapour and Al/&/AI.. . Cu/Al/‘Cu thin films sequentially deposited S. Marsillac Equipe
Couches
Minces
et Matkiaux
*, J.C. Bernede, C. El Moctar,
J. Pouzet
Nomeaux,
44072 Nantes,
LPME,
FSTN,
2 rue de la HoussiniPre,
C6de.x
03, France
Received 24 May 1996; accepted 14 November 1996
Abstract CuAlSe, thin films have beensynthesizedby selenizationof thin Cu and Al layers sequentiallydepositedby evaporation under vacuum. The films have been characterizedby X-ray diffraction, scannin, mand transmissionelectron microscopy, microprobe analysis,photoelectronspectroscopyand Raman diffusion. The resultsare comparedto those obtained with a CuAlSe, reference powder. It is shownthat CuAlSe, films are obtained with someCu,.,Se phasespresentat the surface.Thesesurfacephasesare suppressed by chemicaletching in a KCN solution. At the end of the process,the XRD spectrumdemonstratesthat textured CuAlSe, films have beenobtained with preferential orientation of the crystallites along the (112)direction. TEM study hasshown that both randomly and (112) oriented microcrystallitesexist in the layers. The films are nearly stoichiometric,but their surface is quite rough. The Raman patterns are in good accordancewith the reference.The XPS spectrashow that the binding energies of the elementsare in good agreementwith bonds of CuAlSe,. 0 1997Elsevier ScienceS.A. Keywords:
CuAlSe,; Chemical etching; Raman patterns
2. Experimental
1. Introduction Ternary I-II-VI,
chalcopyrite semiconductors have
received considerable attention in recent years because of their applications in photovoltaic devices [l]. If the CuInSe, is the most extensively studied compound of the family, some other ternary chalcopyrites have attracted much attention in recent years because of their large potential applications. Among them, CuAlSe,, which is a large band gap material-Eg = 2.67 eV [2]-that can be used as an emitting layer of blue light photoelectroluminescent devices [3] or as a window in heterostructures photovoltaic devices [4]. If for the former application heteroepitaxial layers are needed [3], in the case of the latter application polycrystalline layers can be used. In this paper, we describe a very simple and cheap technique which allows us to obtain well crystallized CuAlSe, polycrystalline layers. * Corresponding
author.
0921-5107/97/$17.00 0 1997 Elsevier Science S.A. All rights reserved. P!ns0921-5107(96)02031-4
2.1. Film preparation CuAlSe, films were obtained by selenization (annealing under selenium atmosphere of Cu/Al.. .Al/Cu/Al thin layers sequentially deposited). The substrates were polished soda lime glass or silicon, chemically cleaned. Then the substrates were out gassed ‘in-situ’ prior to deposition by heating to 400 K for 2 h. The Cu and Al films were deposited in vacuum (pressure 5 10 -’ Pa) by evaporation from two tungsten crucibles. The purity of Al was 99.99%, that of Cu was 99.999% and that of Se was 99.999%. Layers of Cu and Al were deposited sequentially on the heated substrates (400 K < T, < 450 K). The CuAlSe, films were then synthesized by annealing in selenium atmosphere.
70
S. Marsillac
et al. /Materials
Science
During the deposition, a rotating substrate holder was successively positioned in front of the Cu evaporation source and in front of the Al evaporation source according to the sequence. The evaporation rates and film thicknesses were measured in-situ by the vibrating quartz method, the quartz head being joined to the substrate holder. The number of layers for each constituent varied from six to ten in order to deposit Al/&/AI.. .Al/Cu structures. The relative thicknesses of the layers were calculated to achieve the desired atomic ratio Cu/Al = 1.1. After deposition, the multilayer structures were introduced with a small amount of Se in a Pyrex tube. This tube was then sealed under vacuum (10 - ’ Pa). The samples were then annealed under selenium atmosphere. Selenium has a high vapour pressure, therefore during the annealing the selenium was in the vapour state. During this process the selenium pressure was about lo6 Pa. The annealing temperature varied from 823 to 853 K while the annealing time was between 24 and 45 h. The results obtained after these processes will be discussed below. The thickness of the Cu layers was about 10 nm and its evaporation rate was 0.2 nm s-l, in the case of aluminium they were 13 nm and 0.2 nm s - ’ respectively. The total thickness of the films after reaction with selenium was measured by a mechanical tiger. In order to check the results obtained with thin i?lms, a CuAlSe, powder has also been synthesized. The powder was prepared by mixing quantities of high purity copper, aluminium and selenium in the atomic ratio 1.l Cu, 1 Al, 2.5 Se. The mixture was sealed in an evacuated silica tube (10 - 3 Pa) and heated at 1373 K for 48 h. After cooling, the solid was reduced into powder. 2.2. Characterization
techniques
A large number of experimental techniques have been used for characterization of the layers: Scanning electron microscopy using the 6400 F JEOL field emission microscope (SEM). Surface topography was examined by scanning electron microscopy and checked by the mechanical finger. Another electron microscope equipped with an electron microprobe was used to check the composition of the samples. Transmission electron microscopy (TEM) was performed with a ‘JEM 120 CX’ equipped with a tungsten tiament at 120 kV to evidence the microstructure and to confirm the texture axis orientation determined by the XRD investigations. For examination by TEM, the films were chemically removed from the glass substrates and then attached to copper grids.
and Engineering
B45 (1997)
69-75
The structure of the films was examined using an analytical X-rays system type DIFFRACT AT V3. 1 SEMENS. which uses a graphics program EVA. The wave length 1 was 0.15 406 nm. The grain size, D, was estimated from the full width at half maximum (FWHM) of the diffraction peaks. Photoelectron spectroscopy (XPS) measurements were performed with a magnesium X-ray source (1253 eV) operating at 10 kV and 10 mA. Data acquisition and treatment were realized using a computer and a standard program. The XPS analysis was made at the University of Nantes, CNRS. We proceeded to in-depth study by recording successive XPS spectra obtained after ion etching for short periods. Using an ion gun, sputtering was accomplished at pressures of less than 5 10B4 Pa with a 10 mA emission current and 5 kV beam energy. The Ar+ ion beam was rastered over the entire sample surface. At the surface of the films there is a carboncarbon bond corresponding to surface contamination. In the apparatus used this C-C bond had a well defined position at 284.8 eV and the carbon peak was used as a reference to estimate the electrical charge effect as the film samples were on glass. Raman spectra were measured at room temperature using a BRUKER model RFS 100; the spectra were excited by a (Nd: YAG) laser with a wavelength of 1064 nm. The laser power and the scale number necessary to obtain good signal to noise ratio were 100 and 500 mW, respectively.
Experimental
results
3.1. Checking of the metallic deposited layers
After deposition of the Al/Cu... sequences, the averaged composition of the superposed layers was estimated by microprobe analysis in order to check the validity of the atomic ratio Cu/Al before annealing under selenium atmosphere. Since it has been shown that a small Cu excess improves strongly the crystallization of the ternary chalcopyrites compounds, only the samples with composition 53 + 2 at.% of Cu were used for the thin film realization. The XRD pattern of such samples exhibits two small peaks which can be attributed to the Al and the Cu. The intensity of these peaks is small, while their full width at half maximum is quite large, which shows that the metallic layers are poorly crystallized with very small crystallites. 3.2. Structural
charncterisation
When possible,
the experimental
results obtained
S. MarsilIac
et al. /Materials
Science
and Engineering
B45 (1997)
69-75
71
on thin fYms have been compared to those obtained with the reference powder. Just after synthesis the XRD spectrum of the CuAlSe, powder exhibits only diffraction peaks which belong to the CuAlSe2 compound (Table 1). However, as we can see in Table 1, two small peaks appear in the powder but not in the reference (JCPDS 44-1269). It must be noticed that these 2 peaks appear also in CuInSe, (JCPDS 40-1487), another ternary chalcopyrite compound, and also obey to the extinction law h + Jc + I= 2 a+ 1. No other peaks, from binary compounds for example, are detected. Since the composition value given by the EPMA (Cu=26 Al=24 and Se= 50 at.%) shows an excess of Cu and Se: we can suppose that Cu,,Se exists in the powder in an amorphous state or in microcrystallites embedded in the powder (the EMPA technique will be discussed below). Therefore, as is usually done in the case of CuInSe, [5], the binary compounds present in the powder have been etched in a 0.1 M KCN solution for some minutes, afterwards only the peak intensities have changed in the XRD spectrum (Table l), while a stoichiometric composition is measured by microprobe analysis (Cu = 25, Al = 25 and Se = 50 at.%). Therefore, the XRD spectrum of the powder has been used as a reference (Table 1). The XRD of thin tirns are reported in Fig. 1. It can be seen in Fig. la that just after annealing, CuAlSe, but also CuZMsSe and Se are present in the films. The surface of these layers appears to be quite inhomogeneous (Fig. 2a), small heaps and large grains of about 1-5 urn size appear, randomly distributed at the surface of the l?lms. It has been shown earlier that annealing under selenium atmosphere in a Pyrex tube induces some selenium condensation at the surface of the layer during the cooling of the tube [6]. This selenium excess at the surface of the layer was sublimated by annealing the samples under dynamic vacuum for 6 h at 570 K After this annealing the Se peak has disappeared but those related to Cu,-,Se are still there (Fig. lb) as the large grains residing on the surface. This binary phase is etched with a KCN solution 0.1 M). The films were etched for periods of 1 min and X-rays were recorded after each treatment until the Cu,,Se peaks disappear (Fig. lc). Then the visualization of the surface of the layer shows that the large grains have disappeared (Fig. 2b). As shown by the X-ray diffraction spectrum (Fig. lc), at the end of the process, only the (112) peak is clearly visible, which shows that the majority of the large crystallites are preferentially oriented along the (112) direction. However, as shown by SEM in Fig.
12
S. Mamillac
et al. /Materials
Science
and Engineering
B45 (1997)
69-75
2(b), there is a lot of small crystallites randomly oriented. This fact has been corroborated by TEM study. A transmission electron micrograph of CuAlSe, layer is reported Fig. 3. It can be seen that larger grains are embedded in small grain domains. The SAD of these domains is shown in Fig. 4(a), it corresponds to randomly distributed microdomains. It can be seen in Table 2 that these microcrystalites have the chalcopyrite structure of CuAlSe,. The SAD of the large grains is shown in, Fig. 4(b), rectangles with d, = 3964 A and L&= 1864 A as characteristic distances appear; these
Fig. 2. Scanning electron micrograph of a CuAlSe, thin film (a) after annealing under selenium atmosphere and (b) after chemical etching by a 0.1 mole KCN solution.
values lead to the zone axis [I121 with (1iO) and (222) as basis vectors. The extinctions which occur are the usual extinction for tetragonal compounds, i.e. for 12+ k+1=2n+l.
Fig. 1. XRD pattern of Al/Cu/Al...Cu/Al/Cu thin film (a) after annealing under Se atmosphere, (b) after 1st post-treatment (annealing under dynamic vacuum) and (c) after 2nd post-treatment (etching by a KCN solution).
Fig. 3. TEM microphotograph.
S. MarsilIac
et al. /Materials
Science
and Engineering
B45 (1997)
73
69-75
4;: bp*=_n..-. ;. $g+;io& =I;-
-
c---. p=/-:
...-I._---_-
i)LI. _
iy
“-hu.‘.
--., ~A&$,‘~
.,.;-.:. +-.*~
‘UN
I /
120 ,... ~-
i
1 _
9ll ._-..--.--...-
;
i, ~“*jyy
t
t
--..
_-._
:j,
i (L-.,j
-+
sb cE(eV) __ -.---“A
f :
Fig. 5. A12p, Cu3p, Na2s and Se3d XPS peaks of (a) CuAlSe, thin films before KCN treatment and (b) CuAlSe, thin films after KCN treatment.
Fig. 4. (a) SAD of a disordered domain and (b) SAD of a large crystallite.
3.3. Chemical characterization
It has been shown above that CuAISe, films crystallized in the chalcopyrite structure can be obtained by our procedure. The chemical state of these films has been studied by microprobe analysis, X-ray photoelectron spectroscopy and Raman diffusion. Since the Se L, and Al K, rays overlap, a long acquisition time ( > 400 s) is necessary to get reproducible results for quantitative measurements by EPMA. It can be seen that at the end of the process the films are stoichiometric 24 < Al < 26; 24 < Cu < 26; 49 < Se < 51. The samples have also been studied by XPS; however, quantitative information is difficult to obtain because the binding energies for Cu and Al are adjacent to each other (Cu3s and A12s; Cu3p and A12p; etc). Moreover, the binding energy and the sensibility factors of Cu2p,,,, A12p and Se3d peaks, which are usually used for quantitative analysis, are very different, which induce uncertainties. XPS lines are shown in Fig. 5 for CuAlSe, thin tims before and after Table 2 XPS results for CuAlSe, powder and thin film after KCN treatment Sample
Cu2psis (ev)
Se Al
cu CuAlSe, Powder CuAlSe, Thin Elm
A12psjz (ev)
Se3d (eV) 55.5
72.7 932.4 932.3 932.4
74.7 74.6
55 55.2
KCN treatment. The binding energies are reported in Table 3 as well as the position of the lines for the elements. It can be seen that there is a good agreement between the reference powder spectra and those of the layers. For the discussion below the carbon peak of contamination at the surface of the samples has been taken as reference: Cls = 284.8 eV. The Se3d peak is situated at 55 eV while the binding energy of the Cu2p,,, is 932.3 eV. The Al peaks are adjacent to Cu peaks (A12p and Cu3p; Al2s and CUES), therefore, we have proceeded to a decomposition. It can be seen in Fig. 6 that the peak situated at 74.5 eV can be attributed to A12p while the other situated at around 77 eV corresponds to Cu3p. The binding energy of the different elements Se3d, A12p and CU~P~,~ are, respectively, 55.3, 72.65 and 932.4 eV [5]. The electronegativity of Se being the highest of the three elements, its binding energy in the CuAlSe, compound should be smaller than that of the element alone, while that of the cations should be higher. This is the case of the A12p peak. In the case of Cu2p,,,, it is well known that its binding energy is poorly sensitive to its oxidation state. Moreover, the binding energies obtained for Se and Cu are in good accordance with the values measured on other compounds of the same ternary chalcopyrite family such as CuInSe, [6]. The iilms have also been studied by XPS before KCN chemical etching. The result depends on the substrate. When Si is used as a substrate the XPS spectra are the same as those discussed above. When a soda lime glass is used a new peak appears at 66 eV (Fig. 5a). In fact, simultaneous to this peak, another one situated at 1072 eV is also detected. This last peak has been already discussed in the case of CuInSe, [7]: it corresponds to Na diffusion from the soda lime substrate towards the surface of the layer. Therefore the peak situated at 66 eV can be attributed to Na 2s. The intensity of these Na peaks decreases when the etching
14
S. Mamillac
et al. /Materials
Science
and Engineerirzg
B45 (1997)
69-75
Table 3 TEM results Circles
d(lO-’
nm)
CuAlSe, Reference
d(lO-’ W)
nm)
3.21
1.96
1.66
1.37
1.13
0.97
3.21
1.96 (204)
1.66
(116)
1.37 (008)
1.13 (228)
0.97 (1 1 11)
(112)
time increases which shows that Na is accumulated mainly at the surface of the layer. Fortunately Na is soluble in KCN and the corresponding peaks disappear after chemical etching (Fig. 5b). The pollution of the thin film has been controlled by the evolution of the carbon and oxygen peaks after etching. It has been seen that these two peaks disappear after 1 or 2 min of etching. Therefore, we can conclude that there is no strong pollution in our thin films. The chemical state of the films has been also checked by Raman diffusion in order to confirm the CuAlSe, occurrence. The results obtained are presented in Fig. 7. It can be seen that the spectrum of the reference powder and that of the layers after KCN treatment are similar. The main peaks obtain are at 26, 85, 184, 211 and 234 cm-’ which is in good accordance with the results of Kamata and Co11 [8], except for the line at 26 cm-l that they did not find. Anyway, a reference spectrum of Cu,$e shows that no line appears at 26 cm- ‘; this peak has also been attributed to CuAlSe, compound.
microcrystallite disordered matrix. If the large crystallites are oriented along the (112) direction, the microcrystallites are randomly oriented. These results are corroborated by those obtained by transmission electron microscopy as well as by Raman spectroscopy. The optimum temperature for the annealing under selenium atmosphere appears to be around 820 K. The annealing duration is optimum at about 24 h. The XPS spectra show that the binding energy of the metal anions is increased while that of Se has decreased by comparison to the reference element. These modifications correspond to electron exchange from the anion to the cation. When the films are obtained on soda lime glass, there is Na at the surface of the layer. The good crystallization and texturation of the layers can be related to this Na diffusion, since it has been shown in the case of CuInSe, [9] that the crystalline quality is strongly enhanced by Na diffusion through the layer. Therefore, these layers appear to have the necessary quality to be used as a buffer layer in solar cells.
4. Discussionand conclusion
Oriented CuAlSe, thin lilms have been obtained by thermal annealing under selenium atmosphere of Cu/ Al/Cu.. .Cu/Al thin sequentially deposited metallic layers. At the end of the process (annealing under selenium atmosphere, annealing under dynamic vacuum, chemical etching) the CuAlSe, films show properties close to those of CuAlSe, reference powder. The films exhibit large grains randomly distributed in a
Wavenumben
Wn\enumbcrs Ek-.-.
Fig. 6. Decomposition of the A12p peak and the Cu3p peak.
~-------
(cm.‘)
(cm-l)
-----p--..I
Fig. 7. Raman diffusion spectra after KCN treatment of: (a) the reference powder and (b) a CuAlSe, film.
S. Marsillac
et al. /Materials
Science
Acknowledgements
The authors wish to thank Mr P. LeRay improvments in the depositing apparatus.
for the
References [I] K.J. Bachmann, H.L. Hwang and C. Schwab, Non-stoichiometry in Semiconductors, EMRS 1991, Spring Meeting, North Holland, 1992. [2] S. Shirakata, S. Chichibu, S. Matsumoto and S. Isomura, Jpn. J. Appl. Phys., 32 (1993) L167.
and Engineering
B45 (1997)
69-75
75
S. Shirakata, S. Isomura, Y. Harada, M. Yuchida, S. [31 S. Chichibu, Matsumoto, H. Higuchi, J. Appl. Phys. 77 (1995) 1225. V.Y. Rud’ and Y.V. Rud’, Tech. Phys. Lett., 20 [41 I.V. Bodnar, (1994) 317. [51 C.D. Wagner, WM. Riggs, M.E. Davis, S.F. Moulder and G.E. Muilenberg, Hand-Book of X-ray Photoelectron Spectroscopy CD, Perkin Elmer, Eden Prairie, MN, 1979. Fl R. Scheer and H.J. Lewerenz, J. Vat. Sci. Technol., AlZ(1) (1994) 56. in H.S. Stephens [71 C. Heske, R. Fink, D. Jacob and E. Umbach, (Ed.), Proc. 13th European Photovoltaic Solar Energy Conference, Nice, France, Bedford UK, 1995, p. 2003. Fl A. Kamata and S. Chichibu, J. Appl. Phys., 77 (1995) 10. Bless, F. Pfisterer, M. Schubert and T. Walter, Prog. PI H.W. Photovolt. Res. Applications, 3 (1995) 3.