- Email: [email protected]
Electrochemical interface modification of CuInS2 thin films
Thin Solid Films 403 – 404 (2002) 57–61
Electrochemical interface modification of CuInS2 thin films b ¨ M. Aggoura, U. Storkel , C. Murrellc, S.A. Campbelld, H. Jungblutc, P. Hoffmanne, R. Mikaloe, D. Schmeißere, H.J. Lewerenzc,* a
Universite´ Ibn Tofail, Faculty of Sciences, LPMC, Kenitra, Morocco b ¨ Mikrotechnik, Mainz, Germany Institut fur c Hahn-Meitner-Institut, Department of Interfaces, Glienicker Str. 100, 14109 Berlin, Germany d University of Portsmouth, Portsmouth, UK e ¨ Cottbus, Germany Brandenburgisch-Technische Universitat,
Abstract We studied two novel electrochemical treatments of CuInS2 solar-cell absorber films, introduced to remove the deleterious segregated CuS phase. Their influence on surface topography, chemistry and electronic properties was investigated using in situ atomic force microscopy (AFM) and photoelectron spectroscopy, performed in part at the U49y2 undulator beam line at BESSY II. The results are examined in order to develop an improved understanding of the chemical–electrochemical surface transformation processes. 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Thin-film solar cells; Copper indium disulfide; Atomic force microscopy (AFM); Photoelectron spectroscopy; Synchrotron radiation
1. Introduction
2. Experimental
The application of photovoltaic solar cells with the chalcopyrite CuInS2 is limited, due to the comparably low photovoltage and toxic chemical treatment steps. In the latter, the removal of a deleterious CuS (covellite) phase by a KCN etch is considered problematic. We were able to remove CuS based on an electrochemical treatment, but found residual CuxS phases on the surface by X-ray diffraction (XRD) w1x, and the photoeffect observed was lower than after KCN etching. Therefore, an improved electrochemical treatment and an analysis of the initial electrochemical conditioning procedure are desirable. In this work, we analyzed the original electrochemical treatment by X-ray photoelectron spectroscopy, including synchrotron radiation and in situ AFM, and present a modified electrochemical treatment that results in a photoeffect as good as that obtained after the KCN etching employed at present.
Electrochemical treatments were carried out in a glass cell in the standard three-electrode potentiostatic arrangement with a Pt counter and a saturated calomel electrode (SCE) as the reference. All solutions were prepared from analytical or ultrapure chemicals, ultrafiltered water (18 MV) and were N2-purged. Cu-rich CuInS2 thin-film samples were prepared either by coevaporation or by sequential processing w2,3x. XRD measurements were performed in the Q–2Q geometry (Cu anode; 45 kV; is30 mA) with a secondary graphite monochromator. In situ AFM experiments (Digital Instruments, Nanoscope III) were carried out in a specifically designed electrochemical cell. Imaging was carried out in water after electrochemical processing. Photoelectron spectroscopy was carried out using an established combined electrochemistryyultrahigh vacuum analysis system w4x. Measurements were made using AlKa (hns1486.6 eV) and synchrotron light (BESSY II) at the U 49y2 undulator beam line.
* Corresponding author. Fax: q49-30-8062-2434. E-mail address: [email protected] (H.J. Lewerenz).
0040-6090/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 0 - 6 0 9 0 Ž 0 1 . 0 1 5 3 2 - 2
58
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
Fig. 1. Cyclic voltammogram of as-grown CuInS2 in 0.1 M K2SO4, pH 10. The pH was adjusted by adding 0.1 M KOH. The strong cathodic peak in the first scan is reduced during subsequent scans. The shape of the curve is stabilized after 10 cycles. Scan rate, 20 mV sy1.
3. Results Fig. 1 shows a cyclic voltammogram of sequentially prepared CuInS2 in slightly alkaline solution. The decay of the cathodic current after cycling shows a transient reaction. This is also observed as a current transient when a potential of y0.95 V is applied to a new sample. The slope of this transient is shown in Fig. 2, indicating a reaction increase and a fast decrease, followed by a leveling out. The overall charge flux is Q(100 mC cmy2. The corresponding surface changes monitored by AFM are shown in Fig. 3. A typical surface area of the polycrystalline samples is displayed. The images correlate with the current transient: (1) shows the as-grown sample, (2) is at the current maximum (shaded region in Fig. 2) and (3) at the leveling-out of the current (situation 4 in Fig. 2). In the lower left side, for instance, pronounced changes can be observed. From (1)´(2), platelets become visible. After the current decay (situation 3), the crystallite in the center appears more pronounced and the platelets are less visible. In other areas, distinct changes are also notable (see arrows). The photoelectron spectroscopy results are summarized in Table 1. We show the position and respective full width at half-maximum (FWHM) for each line. Whereas the Cu 2p line shifts by 0.4 eV towards higher EB, the In 3d line remains unchanged, as does the S 2p line. The In signal, however, is strongly attenuated after this electrochemical treatment. The FWHM of In is unaltered, but is reduced for the Cu and S lines by the treatment. Since the photoeffect observed using this method was inferior to that obtained after a KCN etch, the new procedure developed earlier was slightly modified. In a new, second step, the electrode was immersed in an acidic solution containing In ions (pH 3) at a potential where a peak, present at y0.45 V, disappeared after one
Fig. 2. Slope of the current transient obtained at fixed potential of y 0.95 V vs. SCE in 0.1 M K2SO4, pH 10 (see text).
cycle. The resulting photoactivity matched that of a KCN-etched sample. In Fig. 4, XRD data of the asgrown, the transient current- and the In-treated samples are compared. It is evident that the first treatment removes CuS from the surface, but leaves a Cu2S species on the surface. After the so-called In treatment, no evidence for residual Cu2S or CuS is found within the limits of the method. An evaluation of the XPS data after the new two-step procedure involving In ions in solution is shown in Table 2. The positions of the Cu and In lines are shifted by 0.3 and 0.4 eV, respectively, whereas the S line remains unchanged. The FWHM of S is strongly reduced. Fig. 5 shows valence band spectra after the first and second electrochemical treatments. After the current transient decay, we observe an emission onset at the valence band maximum and do not find an In 4d line. Following the In treatment, the emission onset is shifted by 1 eV and a distinct In 4d signal at 18.7 eV is evident. 4. Discussion The XRD data (Fig. 4) show that the electrochemical treatments shown in Figs. 1 and 2 lead to the removal of covellite. From the change in slope in Fig. 2, a staged dissolution process is evident. This can be attributed either to morphological influences on the etching rate Table 1 Evaluation of binding energy (EB) and full width at half-maximum (FWHM) from XPS data obtained with AlKa As grown
Cu 2p In 3d S 2p
Electrochemical treatment, pH 10
EB (eV)
FWHM (eV)
EB (eV)
FWHM (eV)
932.1 444.4 161.4
1.62 1.4 1.6
932.5 444.4 161.3
1.5 1.4 1.2
lesc yML
5 7 8
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
59
Fig. 3. In situ AFM micrographs of CuInS2 surface changes obtained at different positions (1–3) of the current transient (insert). Z scale: 1 mm per division. Arrows indicate distinct morphological changes.
or to a chemical origin. Since CuS is a layered-type material with Cu in a trigonal planar and a tetraedral configuration and three different sites for S, we suppose that the initial, slower process (region 1) attacks the edges of layers at steps where saturated dangling bonds are likely to exist and backbonds are weakened. The faster process (region 3) could be due to the removal of more strongly bound Cu in layers that are already
partly dissolved. The leveling out is tentatively attributed to etching in the deeper parts of grain boundaries. The AFM images show the successive dissolution of material and it should be noted that at step 3, part of the surface (lower left) appears smoother than for the as-grown sample. Interestingly, the rather sharp edges of the crystallites become softened after the current decay. Inspection of Table 1 shows that the Cu line has shifted
Fig. 4. X-Ray diffractograms of an untreated (top), K2SO4-treated (center) and an additionally In-treated CuInS2 sample (bottom).
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
60
Table 2 Evaluation of EB and FWHM after two-step electrochemical treatment from XPS data obtained with AlKa As growna
Cu 2p In 3d S 2p a
intermediates postulated have not yet been found. At present, the following reaction sequence might be envisaged:
Electrochemical treatment 2
CuSqH2Oq2ey™Cu2SqHSyqOHy In a simultaneous reaction, subsulfides are formed:
EB (eV)
FWHM (eV)
EB (eV)
FWHM (eV)
932.1 444.4 161.4
1.62 1.4 1.6
932.4 444.8 161.3
1.7 1.43 1.18
See Table 1.
to the value for Cu in Cu2S, in accordance with the observation of a corresponding peak in the XRD data at 48.58. In addition, a pronounced decrease in the FWHM for S and a simultaneous decrease in the FWHM for Cu are found. We attribute this to the reduction of the different bonding sites after dissolution of CuS. Interestingly, the In 4d peak in Fig. 5 is suppressed after the electrochemical treatment at pH 10. Since the In 3d signal for AlKa excitation is also considerably weakened (not shown), the surface must be rather homogeneously covered by the Cu2S phase. This is also supported by the changed morphological features in the AFM image. In an earlier work, we tentatively proposed a reaction scheme for the transformation of the surface phase from CuS to one that is almost Cu2S-free w5x. The reaction
(1qx)CuSqxH2Oq2xey™Cu1qxSqxHSy qxOHy
(1)
(2)
These phases are assumed to dissolve, due to their transient formation and the fact that monoclinic chalcocite Cu2S remains in small amounts on the surface. Hydrogen evolution would likely occur as a result of catalysis of the reaction by the cuprous sulfide covering the surface. The removal of the residual Cu2S phase after the new In treatment has the electronic effect that the Fermi level no longer coincides with the valence band maximum, as observed in Fig. 5. The shift of 0.9 eV is attributed to a chemical change, i.e. the formation of a phase for which EF is located at 0.9 eV above EV. This argument is supported by the unchanged binding energy of S. The In and the Cu lines correspond to the position of In and Cu in CuInS2 with respect to EV (444.0 and 931.6 eV). It thus appears that a Cu–In–S phase resides on the surface after this treatment and the XRD results show no phases other than CuInS2.
Fig. 5. Valence band spectrum of CuInS2 obtained with photoelectron spectroscopy (synchrotron radiation) after the new electrochemical treatment (dashed line) in comparison to the valence band spectrum obtained after pure K2SO4 treatment (solid line); hns200 eV.
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
References w1x M. Aggour, H.J. Lewerenz, J. Klaer, U. Storkel, ¨ Electrochem. Solid-State Lett. 3 (2000) 399. w2x R. Scheer, T. Walter, H.W. Schock, M.L. Fearheiley, H.J. Lewerenz, Appl. Phys. Lett. 63 (1993) 3294.
61
w3x J. Klaer, J. Bruns, R. Henninger, K. Siemer, R. Klenk, K. Ellmer, ¨ D. Braunig, Semicond. Sci. Technol. 13 (1998) 1456. w4x H.J. Lewerenz, H. Jungblut, in: S.A. Campbell, H.J. Lewerenz (Eds.), Semiconductor Micromachining, 1, Wiley, Chicester, New York, 1998, pp. 217–275. w5x U. Storkel, ¨ M. Aggour, C.P. Murrell, H.J. Lewerenz, Thin Solid Films 384 (2001) 182.
Electrochemical interface modification of CuInS2 thin films b ¨ M. Aggoura, U. Storkel , C. Murrellc, S.A. Campbelld, H. Jungblutc, P. Hoffmanne, R. Mikaloe, D. Schmeißere, H.J. Lewerenzc,* a
Universite´ Ibn Tofail, Faculty of Sciences, LPMC, Kenitra, Morocco b ¨ Mikrotechnik, Mainz, Germany Institut fur c Hahn-Meitner-Institut, Department of Interfaces, Glienicker Str. 100, 14109 Berlin, Germany d University of Portsmouth, Portsmouth, UK e ¨ Cottbus, Germany Brandenburgisch-Technische Universitat,
Abstract We studied two novel electrochemical treatments of CuInS2 solar-cell absorber films, introduced to remove the deleterious segregated CuS phase. Their influence on surface topography, chemistry and electronic properties was investigated using in situ atomic force microscopy (AFM) and photoelectron spectroscopy, performed in part at the U49y2 undulator beam line at BESSY II. The results are examined in order to develop an improved understanding of the chemical–electrochemical surface transformation processes. 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Thin-film solar cells; Copper indium disulfide; Atomic force microscopy (AFM); Photoelectron spectroscopy; Synchrotron radiation
1. Introduction
2. Experimental
The application of photovoltaic solar cells with the chalcopyrite CuInS2 is limited, due to the comparably low photovoltage and toxic chemical treatment steps. In the latter, the removal of a deleterious CuS (covellite) phase by a KCN etch is considered problematic. We were able to remove CuS based on an electrochemical treatment, but found residual CuxS phases on the surface by X-ray diffraction (XRD) w1x, and the photoeffect observed was lower than after KCN etching. Therefore, an improved electrochemical treatment and an analysis of the initial electrochemical conditioning procedure are desirable. In this work, we analyzed the original electrochemical treatment by X-ray photoelectron spectroscopy, including synchrotron radiation and in situ AFM, and present a modified electrochemical treatment that results in a photoeffect as good as that obtained after the KCN etching employed at present.
Electrochemical treatments were carried out in a glass cell in the standard three-electrode potentiostatic arrangement with a Pt counter and a saturated calomel electrode (SCE) as the reference. All solutions were prepared from analytical or ultrapure chemicals, ultrafiltered water (18 MV) and were N2-purged. Cu-rich CuInS2 thin-film samples were prepared either by coevaporation or by sequential processing w2,3x. XRD measurements were performed in the Q–2Q geometry (Cu anode; 45 kV; is30 mA) with a secondary graphite monochromator. In situ AFM experiments (Digital Instruments, Nanoscope III) were carried out in a specifically designed electrochemical cell. Imaging was carried out in water after electrochemical processing. Photoelectron spectroscopy was carried out using an established combined electrochemistryyultrahigh vacuum analysis system w4x. Measurements were made using AlKa (hns1486.6 eV) and synchrotron light (BESSY II) at the U 49y2 undulator beam line.
* Corresponding author. Fax: q49-30-8062-2434. E-mail address: [email protected] (H.J. Lewerenz).
0040-6090/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 0 - 6 0 9 0 Ž 0 1 . 0 1 5 3 2 - 2
58
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
Fig. 1. Cyclic voltammogram of as-grown CuInS2 in 0.1 M K2SO4, pH 10. The pH was adjusted by adding 0.1 M KOH. The strong cathodic peak in the first scan is reduced during subsequent scans. The shape of the curve is stabilized after 10 cycles. Scan rate, 20 mV sy1.
3. Results Fig. 1 shows a cyclic voltammogram of sequentially prepared CuInS2 in slightly alkaline solution. The decay of the cathodic current after cycling shows a transient reaction. This is also observed as a current transient when a potential of y0.95 V is applied to a new sample. The slope of this transient is shown in Fig. 2, indicating a reaction increase and a fast decrease, followed by a leveling out. The overall charge flux is Q(100 mC cmy2. The corresponding surface changes monitored by AFM are shown in Fig. 3. A typical surface area of the polycrystalline samples is displayed. The images correlate with the current transient: (1) shows the as-grown sample, (2) is at the current maximum (shaded region in Fig. 2) and (3) at the leveling-out of the current (situation 4 in Fig. 2). In the lower left side, for instance, pronounced changes can be observed. From (1)´(2), platelets become visible. After the current decay (situation 3), the crystallite in the center appears more pronounced and the platelets are less visible. In other areas, distinct changes are also notable (see arrows). The photoelectron spectroscopy results are summarized in Table 1. We show the position and respective full width at half-maximum (FWHM) for each line. Whereas the Cu 2p line shifts by 0.4 eV towards higher EB, the In 3d line remains unchanged, as does the S 2p line. The In signal, however, is strongly attenuated after this electrochemical treatment. The FWHM of In is unaltered, but is reduced for the Cu and S lines by the treatment. Since the photoeffect observed using this method was inferior to that obtained after a KCN etch, the new procedure developed earlier was slightly modified. In a new, second step, the electrode was immersed in an acidic solution containing In ions (pH 3) at a potential where a peak, present at y0.45 V, disappeared after one
Fig. 2. Slope of the current transient obtained at fixed potential of y 0.95 V vs. SCE in 0.1 M K2SO4, pH 10 (see text).
cycle. The resulting photoactivity matched that of a KCN-etched sample. In Fig. 4, XRD data of the asgrown, the transient current- and the In-treated samples are compared. It is evident that the first treatment removes CuS from the surface, but leaves a Cu2S species on the surface. After the so-called In treatment, no evidence for residual Cu2S or CuS is found within the limits of the method. An evaluation of the XPS data after the new two-step procedure involving In ions in solution is shown in Table 2. The positions of the Cu and In lines are shifted by 0.3 and 0.4 eV, respectively, whereas the S line remains unchanged. The FWHM of S is strongly reduced. Fig. 5 shows valence band spectra after the first and second electrochemical treatments. After the current transient decay, we observe an emission onset at the valence band maximum and do not find an In 4d line. Following the In treatment, the emission onset is shifted by 1 eV and a distinct In 4d signal at 18.7 eV is evident. 4. Discussion The XRD data (Fig. 4) show that the electrochemical treatments shown in Figs. 1 and 2 lead to the removal of covellite. From the change in slope in Fig. 2, a staged dissolution process is evident. This can be attributed either to morphological influences on the etching rate Table 1 Evaluation of binding energy (EB) and full width at half-maximum (FWHM) from XPS data obtained with AlKa As grown
Cu 2p In 3d S 2p
Electrochemical treatment, pH 10
EB (eV)
FWHM (eV)
EB (eV)
FWHM (eV)
932.1 444.4 161.4
1.62 1.4 1.6
932.5 444.4 161.3
1.5 1.4 1.2
lesc yML
5 7 8
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
59
Fig. 3. In situ AFM micrographs of CuInS2 surface changes obtained at different positions (1–3) of the current transient (insert). Z scale: 1 mm per division. Arrows indicate distinct morphological changes.
or to a chemical origin. Since CuS is a layered-type material with Cu in a trigonal planar and a tetraedral configuration and three different sites for S, we suppose that the initial, slower process (region 1) attacks the edges of layers at steps where saturated dangling bonds are likely to exist and backbonds are weakened. The faster process (region 3) could be due to the removal of more strongly bound Cu in layers that are already
partly dissolved. The leveling out is tentatively attributed to etching in the deeper parts of grain boundaries. The AFM images show the successive dissolution of material and it should be noted that at step 3, part of the surface (lower left) appears smoother than for the as-grown sample. Interestingly, the rather sharp edges of the crystallites become softened after the current decay. Inspection of Table 1 shows that the Cu line has shifted
Fig. 4. X-Ray diffractograms of an untreated (top), K2SO4-treated (center) and an additionally In-treated CuInS2 sample (bottom).
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
60
Table 2 Evaluation of EB and FWHM after two-step electrochemical treatment from XPS data obtained with AlKa As growna
Cu 2p In 3d S 2p a
intermediates postulated have not yet been found. At present, the following reaction sequence might be envisaged:
Electrochemical treatment 2
CuSqH2Oq2ey™Cu2SqHSyqOHy In a simultaneous reaction, subsulfides are formed:
EB (eV)
FWHM (eV)
EB (eV)
FWHM (eV)
932.1 444.4 161.4
1.62 1.4 1.6
932.4 444.8 161.3
1.7 1.43 1.18
See Table 1.
to the value for Cu in Cu2S, in accordance with the observation of a corresponding peak in the XRD data at 48.58. In addition, a pronounced decrease in the FWHM for S and a simultaneous decrease in the FWHM for Cu are found. We attribute this to the reduction of the different bonding sites after dissolution of CuS. Interestingly, the In 4d peak in Fig. 5 is suppressed after the electrochemical treatment at pH 10. Since the In 3d signal for AlKa excitation is also considerably weakened (not shown), the surface must be rather homogeneously covered by the Cu2S phase. This is also supported by the changed morphological features in the AFM image. In an earlier work, we tentatively proposed a reaction scheme for the transformation of the surface phase from CuS to one that is almost Cu2S-free w5x. The reaction
(1qx)CuSqxH2Oq2xey™Cu1qxSqxHSy qxOHy
(1)
(2)
These phases are assumed to dissolve, due to their transient formation and the fact that monoclinic chalcocite Cu2S remains in small amounts on the surface. Hydrogen evolution would likely occur as a result of catalysis of the reaction by the cuprous sulfide covering the surface. The removal of the residual Cu2S phase after the new In treatment has the electronic effect that the Fermi level no longer coincides with the valence band maximum, as observed in Fig. 5. The shift of 0.9 eV is attributed to a chemical change, i.e. the formation of a phase for which EF is located at 0.9 eV above EV. This argument is supported by the unchanged binding energy of S. The In and the Cu lines correspond to the position of In and Cu in CuInS2 with respect to EV (444.0 and 931.6 eV). It thus appears that a Cu–In–S phase resides on the surface after this treatment and the XRD results show no phases other than CuInS2.
Fig. 5. Valence band spectrum of CuInS2 obtained with photoelectron spectroscopy (synchrotron radiation) after the new electrochemical treatment (dashed line) in comparison to the valence band spectrum obtained after pure K2SO4 treatment (solid line); hns200 eV.
M. Aggour et al. / Thin Solid Films 403 – 404 (2002) 57–61
References w1x M. Aggour, H.J. Lewerenz, J. Klaer, U. Storkel, ¨ Electrochem. Solid-State Lett. 3 (2000) 399. w2x R. Scheer, T. Walter, H.W. Schock, M.L. Fearheiley, H.J. Lewerenz, Appl. Phys. Lett. 63 (1993) 3294.
61
w3x J. Klaer, J. Bruns, R. Henninger, K. Siemer, R. Klenk, K. Ellmer, ¨ D. Braunig, Semicond. Sci. Technol. 13 (1998) 1456. w4x H.J. Lewerenz, H. Jungblut, in: S.A. Campbell, H.J. Lewerenz (Eds.), Semiconductor Micromachining, 1, Wiley, Chicester, New York, 1998, pp. 217–275. w5x U. Storkel, ¨ M. Aggour, C.P. Murrell, H.J. Lewerenz, Thin Solid Films 384 (2001) 182.