- Email: [email protected]
Correlation of the near-infrared optical absorption with Cu concentration in coevaporated Cu–In–S films
Thin Solid Films 517 (2009) 2260–2263
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Correlation of the near-infrared optical absorption with Cu concentration in coevaporated Cu–In–S films J.F. Trigo ⁎, A. Bollero, J. Herrero, M.T. Gutiérrez Dep. of Energy, CIEMAT, Avda. Complutense 22, 28040 Madrid, Spain
a r t i c l e
i n f o
Available online 7 November 2008 Keywords: CuInS2 Thin film In situ monitoring Electro-optical properties
a b s t r a c t We have deposited Cu–In–S films by an innovative modulated flux deposition (MFD) procedure performed in a static physical evaporation chamber in compositional ranges as diverse as: [In] ranging 10–70%, [Cu] = 2–43% and [S] = 20–50%. We propose optical absorption (A% = 100 − R% − T%) in the near-IR (about 0.5 eV or 2500 nm) as a non-contact measurement which could be indirectly correlated with chemical composition. The films, characterized by spectrophotometry (A%) and sheet resistance (R) by four-terminal sensing measurement, were classified in compositional groups by the use of the two-dimensional parameter (R, A%). Interestingly, the optical absorption A% was linearly correlated with Cu content in our compositional range while R presents a jump back related with the variation of S concentration. The independence of R and A% in this case is discussed in terms of different conduction mechanisms. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Elemental coevaporation of copper indium and sulfur (CuInS2) for photovoltaic application is an increasingly used technique where the stoichiometry control is the main challenge [1,2]. Several control methods have been demonstrated but much of them result in expensive and complex setups. There is a long list of real time monitoring and insitu diagnostic methods for thin film fabrication that has been applied for fabrication of CuInS2/CIGS films (EDXRD, in-situ XPS, UPS, Raman, ellipsometry, light scattering, pyrometry…) [3–5]. The last one is now a routine instrumentation in CIGS coevaporation and its introduction represented an important advance in cell efficiencies [4]. Sheet resistance (Rs in Ω/□ or simply R in Ω) is an easy to implement in-situ measurement, but it is not univocally related with CuInS2 composition. As a ternary compound, CuInS2 composition is a two-dimensional space and at least another independent measurement should be used. Additionally, no need to optical models, calculation or thickness knowledge, could make it useful for in-situ control. Drude's free electron theory has been extensively utilized to understand the relationship between the optical and electrical properties of thin films. Its use has been very successful in the field of transparent contact oxides (TCO) where this relation is the key of the application [6,7]. In the last work, the density of free carriers and their mobility were independently estimated from optical free carrier absorption and compared with that obtained by Hall measurements. The agreement was notable for the concentration of free carriers and the discrepancies in carriers mobility was satisfactory explained due to ⁎ Corresponding author. Tel.: +34 913466669; fax: +34 913466037. E-mail address: [email protected] (J.F. Trigo). 0040-6090/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2008.10.111
different spatial range that the optical excitation and DC conductivity are probing. The grain boundaries hinder the carrier transport if the carrier density is not too high. It looks like IR optical absorption (or pyrometry) and DC conductivity are strongly related but could be complementary measurements in some manner because DC conductivity would be more sensitive to the morphology. In the present work we have tried to explore this complementarity in a phenomenological procedure that correlates these measurements with averaged composition of CuInS2 films grown by coevaporation. We have not paid attention to the deposition conditions as no relation with fabrication parameters will be drawn and we needed the wider compositional space possible. Although we have used ex-situ measurements of A% and R, we realize that many technical problems would arise for the measurements to become compatible with the deposition process, especially if non transparent conductive layers are added to the substrate. We have not tried to solve this questions but a suggestion would be the use of removable glassy test samples.
2. Experimental Several CuInS2 films have been prepared onto 10 × 10 cm soda-lime glass substrates by the MFD procedure in a static physical evaporation chamber [8,9]. They have been grown in very different conditions (simultaneous or sequential fluxes, constant and ramping temperature of the sources, and substrates temperature ranging 250–500 °C) and with final thickness varying from 125 to 890 nm. The flashing of the sulfur source produced oscillations of the chamber pressure from 9 · 10− 3 to 3 · 10− 2 Pa. A 2 × 2 cm central part of the samples were ex-situ optically, electrically and chemically characterized.
J.F. Trigo et al. / Thin Solid Films 517 (2009) 2260–2263
2261
samples, a thickness value d measured by profilometry has been operated to obtain resistivity and absorption coefficient following: π0; 94 ρðX cmÞ = RðXÞ d dðcmÞ d lnð2Þ
α ðcm−1 Þ =
ln 100−Rk0:5eV Tk0:5eV
dðcmÞ
The last equation is an approximation where the absorption of the substrate has not been separated. Finally their chemical composition was determined by energy dispersive X-ray (EDX) analysis in a scanning electron microscope Hitachi S-2500. 3. Results and discussion
Fig. 1. Sulfur composition related with Cu (In on the inset) content from EDX (at.%). Marked point corresponds to a sample after thermal annealing in sulfur atmosphere.
The optical properties were carried out with a Perkin-Elmer Lambda 9 spectrophotometer operating at room temperature. The near normal Reflectance and Transmittance were measured in the dual channel configuration and by means of an integrator sphere in the 250–2500 nm range. The static electrical sheet resistance was measured using a fourpoint probe Veeco FPP5000 device. For simplicity we have used direct R =V /I values in Ohms and no geometrical constant and thickness information has been operated. For a more precise study of some
As our initial goal was to define the general working ranges of our deposition system we have studied our compositional map for all the films in Fig. 1 where we observed a strong correlation between S and Cu contents, increasing together probably due to our simultaneous efforts introducing improvements of the process toward the stoichiometric CuInS2. To check if it was possible to unlink these two parameters we tried a thermal treatment to a sample with low (4 at.%) Cu content. 35 min at 300 °C in sulfur atmosphere at a deposition process pressure were enough to have the S content rising in the marked sample on Fig. 1. Another goal was to look for a fast mechanism to control the basic stoichiometric parameters (average content of Cu and S related to In). In this order we have tried with optical absorption in its more primitive expression: the energy conservation for the whole sample (including the substrate) A% =100 −R% −T%. There are other formulations intended to calculate the absorption coefficient of the material, but the thickness of the film and several data from the substrate are needed. If the substrate is transparent enough and the film is highly absorbent and with a low refractive index, as is the case with CuInS2, the effect of film interferences
Fig. 2. Optical absorption defined as A% = 100 − R% − T% for selected samples with increasing Cu content (at.%) as determined by EDX.
2262
J.F. Trigo et al. / Thin Solid Films 517 (2009) 2260–2263
disappears above 100 nm film thickness. A harder approximation is not to consider the thickness of the films, but with the same hypothesis than before we can argue that for thickness in the micrometer range the T% is near zero, so what we are considering is the reflectance or, as it can be seen in pyrometry, the thermal emittance of the film ε = 1 −R. This magnitude has been often related (in longer wavelengths than here) with the conductivity through the Hagen–Rubens relation ε = 2[4πε0cρ0 /λ]1/2 [10] as an approximation that equals static and optical conductivities. When we plotted A% against photon energy (Fig. 2) for several samples we observe the growing shoulders at 1 and 2 eV with the increment of Cu content. Several of our samples are in fact copper doped indium sulfide and the first A% spectrum (thermal treated 4% Cu) resembles In2S3 optical absorption with these additional IR–VIS absorptions into the In2S3 transparent window. Looking only at the IR end (0.5 eV) of our spectra a constant AIR% value has been extracted by the averaging of ten wavelength measurements due to the high noise in this end of the spectra. This noise is due to the low sensitivity of the spectrometer at its measurement edge so it could be avoided with more appropriate instrumentation but it is still smaller than samples differences. This energy is far enough from interband transitions as we are looking for free carrier absorption dominated effects. From now on we will associate this energy independent value to each sample and will name it A%. A higher A% value is observed for averaged compositions closer to CuInS2 and also for Cu rich samples, as is expected for a higher absorber and better conducting material. In Fig. 3 we have plotted A% values against Cu concentration and we can see a noisy linear correlation till CuInS2 composition (25%) is reached. We do not have much information for rich Cu CuInS2 but the IR optical absorption seems to stabilize around 60%. The correlation of Cu composition with A% was not good enough to produce accurate predictions. It looks like this A% measurement could be useful to distinguish contents of Cu very far down the stoichiometry but the information is confusing when stoichiometric or rich Cu CuInS2 are compared. In order to check whether the correlation of A% with Cu is drawn by the S–Cu correlation, we have plotted A% versus S in Fig. 4 and we have observed the behavior of the thermal treated sample (rounded with a mark in the figure) showing that at least in the low Cu content range A% is sensible to the S content as well. We studied the films resistance (R) measurements and we did not find a clear correlation with the Cu content for all the samples. Despite of it, we were still able to find some variations after grouping the samples by their sulfur content.
Fig. 3. An averaged value of A% for 0.5 eV (wavelength = 2.5 μm) plotted against the Cu content in CuInS2 samples, and the corresponding α value for selected near stoichiometric films in the onset.
Fig. 4. An averaged value of A% for 0.5 eV (wavelength = 2.5 μm) plotted against the S content in CuInS2 samples. Marked point corresponds to a sample after thermal annealing in sulfur atmosphere.
Fig. 5. A scatter plot of all the CuInS2 samples in the A%−R (Ω) space a) (upper) the schema of the grouping classification b) showing the trend with Cu content, more precisely in the onset where a α–ρ plot of five samples with similar S content is drawn (two uppermost samples overlap).
J.F. Trigo et al. / Thin Solid Films 517 (2009) 2260–2263
2263
4. Conclusions
Table 1 Classification of CuInS2 samples in sets by their composition and properties Group
[Cu] %
AIR%
Film resistance
[S] %
A B C D E
2–9 13–15 16–21 23–24 37–43
8–18 18–25 35–45 50–55 60–70
Insulating Medium R Low R Medium R Metallic
Variable 29–39 29–39 48–49 48–49
Upper and lower instrumentation R limits has been associated with insulating or metallic behavior for clarity.
Finally a representation with both (R, A%) measurements (Fig. 5a) permitted to classify the samples and correlate with both (Cu, S) contents. We have found five groups in our samples rounded in the figure and described in Table 1. Their behavior can be summarized as follows: Samples in group A of the table have averaged compositions closed to [In]/[S] = 2 and resulted insulating and transparent till Cu doping was higher than 9%. Conductivity and IR absorption rose with higher Cu content as expected through a Drude model with increasing free carrier concentration. This system seems to saturate in an averaged composition around [Cu] = 19%, [In] = 46%, [S] = 35%. After that we observe a jump down in conductivity while IR Absorption continues growing. This corresponds to the C to D group jump with a concentration of about [Cu]= 24%, [In]= 28%, [S] = 48%. Further increase of Cu and S contents reaches the averaged CuInS2 stoichiometric composition. Further increase of Cu meanwhile the S content is limited leads to group E where a conducting behavior of the samples is probably due to Cu–S phases. In another set of samples the authors could also check that sulfur incorporation was not only correlated with Cu flux, but also depends on the history of the deposition process and the substrate temperature, which could be minimized by means of a modulated coevaporation method [11]. Finally the same information has been replotted on Fig 5b avoiding grouping. Only a separation between sulfur deficient and fully sulfurized samples by means of a horizontal dashed line has drawn. The upper (fully sulfurized) set of samples has been studied by introducing thickness information in the primitive parameters and represented in the onset. Here sulfur and thickness variations have been decoupled and an exponential like behavior of the absorption coefficient α with resistivity ρ seems to be strongly influenced by the Cu stoichiometric deviations. An open question for the in-situ monitoring use would be the influence of the high deposition temperature on the compositional correlation with the (R, A%) set of parameters.
We have deposited Cu–In–S films by an innovative modulated flux deposition (MFD) procedure performed in a static physical evaporation chamber in compositional ranges as diverse as: [In] ranging 10–70%, [Cu] = 2–43% and [S] = 20–50%. The films, characterized by spectrophotometry and four-terminal sensing measurement, were classified in compositional groups by the use of the two-dimensional parameter (DC resistance, near IR optical absorption A%). Interestingly, the optical absorption A% was linearly correlated with Cu content in the Cu poor compositional range while R presents a jump back related with the variation of S concentration, being more sensitive to Cu content variation when S content is fixed. The independence of R and A% in this case makes this measurement set useful as an end-point method for compositional detection of CuInS2 films with similar thicknesses, but the effect of deposition temperature and the relation with morphology and carriers mobility need to be further investigated. Acknowledgments This work has been supported by Spanish government MEC projects: CONSOLIDER GENESIS FV (CSD2006-04) and MAT2005-06738-C02-02, by Madrid Community in the framework of The IV PRICIT through the project FOTOFLEX-CM (S-0505/ENE-0213) and by Fundación Ramón Areces Project: “Células fotovoltaicas ligeras y flexibles”. A.B. acknowledges support through a “Ramón y Cajal” contract from the Spanish Ministry of Education and Science. References [1] R. Klenk, J. Klaer, R. Scheer, M.Ch. Lux-Steiner, I. Luck, N. Meyer, U. Rühle, Thin Solid Films 480 (2005) 509. [2] A. Amara, W. Rezaiki, A. Ferd, A. Hendaoui, A. Drici, M. Guerioune, J.C. Bernède, M. Morsli, Sol. Energy Mater. Sol. Cells 91 (2007) 1916. [3] E. Rudigier, Ch. Pietzker, M. Wimbor, I. Luck, J. Klaer, R. Scheer, B. Barcones, T. Jawhari Colin, J. Alvarez-Garcia, A. Perez-Rodriguez, A. Romano-Rodriguez, Thin Solid Films 431–432 (2003) 110. [4] K. Sakurai, R. Hunger, R. Scheer, C.A. Kaufmann, A. Yamada, T. Baba, Y. Kimura, K. Matsubara, P. Fons, H. Nakanishi, S. Niki, Prog. Photovolt: Res. Appl. 12 (2004) 219. [5] K. Müller, R. Scheer, Y. Burkov, D. Schmeisser, Thin Solid Films 451–452 (2004) 120. [6] I. Hamberg, C.G. Granqvist, J. Appl. Phys. 60 (11) (1986) R123. [7] H. Fujiwara, M. Kondo, Phys. Rev. B 71 (2005) 075109. [8] C. Guillén, T. García, J. Herrero, M.T. Gutiérrez, F. Briones, Thin Solid Films 451 (2004) 112. [9] J.F. Trigo, B. Asenjo, J. Herrero, M.T. Gutiérrez, Sol. Energy Mater. Sol. Cells 92 (9) (2008) 1145. [10] K. Gelin, E. Wäckelgård, J. Phys: Condens. Matter 16 (2004) 833. [11] A. Bollero, J.F. Trigo, J. Herrero, M.T. Gutiérrez, Thin Solid Films 517 (7) (2009) 2167 (this issue), doi:10.1016/j.tsf.2008.10.081.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Correlation of the near-infrared optical absorption with Cu concentration in coevaporated Cu–In–S films J.F. Trigo ⁎, A. Bollero, J. Herrero, M.T. Gutiérrez Dep. of Energy, CIEMAT, Avda. Complutense 22, 28040 Madrid, Spain
a r t i c l e
i n f o
Available online 7 November 2008 Keywords: CuInS2 Thin film In situ monitoring Electro-optical properties
a b s t r a c t We have deposited Cu–In–S films by an innovative modulated flux deposition (MFD) procedure performed in a static physical evaporation chamber in compositional ranges as diverse as: [In] ranging 10–70%, [Cu] = 2–43% and [S] = 20–50%. We propose optical absorption (A% = 100 − R% − T%) in the near-IR (about 0.5 eV or 2500 nm) as a non-contact measurement which could be indirectly correlated with chemical composition. The films, characterized by spectrophotometry (A%) and sheet resistance (R) by four-terminal sensing measurement, were classified in compositional groups by the use of the two-dimensional parameter (R, A%). Interestingly, the optical absorption A% was linearly correlated with Cu content in our compositional range while R presents a jump back related with the variation of S concentration. The independence of R and A% in this case is discussed in terms of different conduction mechanisms. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Elemental coevaporation of copper indium and sulfur (CuInS2) for photovoltaic application is an increasingly used technique where the stoichiometry control is the main challenge [1,2]. Several control methods have been demonstrated but much of them result in expensive and complex setups. There is a long list of real time monitoring and insitu diagnostic methods for thin film fabrication that has been applied for fabrication of CuInS2/CIGS films (EDXRD, in-situ XPS, UPS, Raman, ellipsometry, light scattering, pyrometry…) [3–5]. The last one is now a routine instrumentation in CIGS coevaporation and its introduction represented an important advance in cell efficiencies [4]. Sheet resistance (Rs in Ω/□ or simply R in Ω) is an easy to implement in-situ measurement, but it is not univocally related with CuInS2 composition. As a ternary compound, CuInS2 composition is a two-dimensional space and at least another independent measurement should be used. Additionally, no need to optical models, calculation or thickness knowledge, could make it useful for in-situ control. Drude's free electron theory has been extensively utilized to understand the relationship between the optical and electrical properties of thin films. Its use has been very successful in the field of transparent contact oxides (TCO) where this relation is the key of the application [6,7]. In the last work, the density of free carriers and their mobility were independently estimated from optical free carrier absorption and compared with that obtained by Hall measurements. The agreement was notable for the concentration of free carriers and the discrepancies in carriers mobility was satisfactory explained due to ⁎ Corresponding author. Tel.: +34 913466669; fax: +34 913466037. E-mail address: [email protected] (J.F. Trigo). 0040-6090/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2008.10.111
different spatial range that the optical excitation and DC conductivity are probing. The grain boundaries hinder the carrier transport if the carrier density is not too high. It looks like IR optical absorption (or pyrometry) and DC conductivity are strongly related but could be complementary measurements in some manner because DC conductivity would be more sensitive to the morphology. In the present work we have tried to explore this complementarity in a phenomenological procedure that correlates these measurements with averaged composition of CuInS2 films grown by coevaporation. We have not paid attention to the deposition conditions as no relation with fabrication parameters will be drawn and we needed the wider compositional space possible. Although we have used ex-situ measurements of A% and R, we realize that many technical problems would arise for the measurements to become compatible with the deposition process, especially if non transparent conductive layers are added to the substrate. We have not tried to solve this questions but a suggestion would be the use of removable glassy test samples.
2. Experimental Several CuInS2 films have been prepared onto 10 × 10 cm soda-lime glass substrates by the MFD procedure in a static physical evaporation chamber [8,9]. They have been grown in very different conditions (simultaneous or sequential fluxes, constant and ramping temperature of the sources, and substrates temperature ranging 250–500 °C) and with final thickness varying from 125 to 890 nm. The flashing of the sulfur source produced oscillations of the chamber pressure from 9 · 10− 3 to 3 · 10− 2 Pa. A 2 × 2 cm central part of the samples were ex-situ optically, electrically and chemically characterized.
J.F. Trigo et al. / Thin Solid Films 517 (2009) 2260–2263
2261
samples, a thickness value d measured by profilometry has been operated to obtain resistivity and absorption coefficient following: π0; 94 ρðX cmÞ = RðXÞ d dðcmÞ d lnð2Þ
α ðcm−1 Þ =
ln 100−Rk0:5eV Tk0:5eV
dðcmÞ
The last equation is an approximation where the absorption of the substrate has not been separated. Finally their chemical composition was determined by energy dispersive X-ray (EDX) analysis in a scanning electron microscope Hitachi S-2500. 3. Results and discussion
Fig. 1. Sulfur composition related with Cu (In on the inset) content from EDX (at.%). Marked point corresponds to a sample after thermal annealing in sulfur atmosphere.
The optical properties were carried out with a Perkin-Elmer Lambda 9 spectrophotometer operating at room temperature. The near normal Reflectance and Transmittance were measured in the dual channel configuration and by means of an integrator sphere in the 250–2500 nm range. The static electrical sheet resistance was measured using a fourpoint probe Veeco FPP5000 device. For simplicity we have used direct R =V /I values in Ohms and no geometrical constant and thickness information has been operated. For a more precise study of some
As our initial goal was to define the general working ranges of our deposition system we have studied our compositional map for all the films in Fig. 1 where we observed a strong correlation between S and Cu contents, increasing together probably due to our simultaneous efforts introducing improvements of the process toward the stoichiometric CuInS2. To check if it was possible to unlink these two parameters we tried a thermal treatment to a sample with low (4 at.%) Cu content. 35 min at 300 °C in sulfur atmosphere at a deposition process pressure were enough to have the S content rising in the marked sample on Fig. 1. Another goal was to look for a fast mechanism to control the basic stoichiometric parameters (average content of Cu and S related to In). In this order we have tried with optical absorption in its more primitive expression: the energy conservation for the whole sample (including the substrate) A% =100 −R% −T%. There are other formulations intended to calculate the absorption coefficient of the material, but the thickness of the film and several data from the substrate are needed. If the substrate is transparent enough and the film is highly absorbent and with a low refractive index, as is the case with CuInS2, the effect of film interferences
Fig. 2. Optical absorption defined as A% = 100 − R% − T% for selected samples with increasing Cu content (at.%) as determined by EDX.
2262
J.F. Trigo et al. / Thin Solid Films 517 (2009) 2260–2263
disappears above 100 nm film thickness. A harder approximation is not to consider the thickness of the films, but with the same hypothesis than before we can argue that for thickness in the micrometer range the T% is near zero, so what we are considering is the reflectance or, as it can be seen in pyrometry, the thermal emittance of the film ε = 1 −R. This magnitude has been often related (in longer wavelengths than here) with the conductivity through the Hagen–Rubens relation ε = 2[4πε0cρ0 /λ]1/2 [10] as an approximation that equals static and optical conductivities. When we plotted A% against photon energy (Fig. 2) for several samples we observe the growing shoulders at 1 and 2 eV with the increment of Cu content. Several of our samples are in fact copper doped indium sulfide and the first A% spectrum (thermal treated 4% Cu) resembles In2S3 optical absorption with these additional IR–VIS absorptions into the In2S3 transparent window. Looking only at the IR end (0.5 eV) of our spectra a constant AIR% value has been extracted by the averaging of ten wavelength measurements due to the high noise in this end of the spectra. This noise is due to the low sensitivity of the spectrometer at its measurement edge so it could be avoided with more appropriate instrumentation but it is still smaller than samples differences. This energy is far enough from interband transitions as we are looking for free carrier absorption dominated effects. From now on we will associate this energy independent value to each sample and will name it A%. A higher A% value is observed for averaged compositions closer to CuInS2 and also for Cu rich samples, as is expected for a higher absorber and better conducting material. In Fig. 3 we have plotted A% values against Cu concentration and we can see a noisy linear correlation till CuInS2 composition (25%) is reached. We do not have much information for rich Cu CuInS2 but the IR optical absorption seems to stabilize around 60%. The correlation of Cu composition with A% was not good enough to produce accurate predictions. It looks like this A% measurement could be useful to distinguish contents of Cu very far down the stoichiometry but the information is confusing when stoichiometric or rich Cu CuInS2 are compared. In order to check whether the correlation of A% with Cu is drawn by the S–Cu correlation, we have plotted A% versus S in Fig. 4 and we have observed the behavior of the thermal treated sample (rounded with a mark in the figure) showing that at least in the low Cu content range A% is sensible to the S content as well. We studied the films resistance (R) measurements and we did not find a clear correlation with the Cu content for all the samples. Despite of it, we were still able to find some variations after grouping the samples by their sulfur content.
Fig. 3. An averaged value of A% for 0.5 eV (wavelength = 2.5 μm) plotted against the Cu content in CuInS2 samples, and the corresponding α value for selected near stoichiometric films in the onset.
Fig. 4. An averaged value of A% for 0.5 eV (wavelength = 2.5 μm) plotted against the S content in CuInS2 samples. Marked point corresponds to a sample after thermal annealing in sulfur atmosphere.
Fig. 5. A scatter plot of all the CuInS2 samples in the A%−R (Ω) space a) (upper) the schema of the grouping classification b) showing the trend with Cu content, more precisely in the onset where a α–ρ plot of five samples with similar S content is drawn (two uppermost samples overlap).
J.F. Trigo et al. / Thin Solid Films 517 (2009) 2260–2263
2263
4. Conclusions
Table 1 Classification of CuInS2 samples in sets by their composition and properties Group
[Cu] %
AIR%
Film resistance
[S] %
A B C D E
2–9 13–15 16–21 23–24 37–43
8–18 18–25 35–45 50–55 60–70
Insulating Medium R Low R Medium R Metallic
Variable 29–39 29–39 48–49 48–49
Upper and lower instrumentation R limits has been associated with insulating or metallic behavior for clarity.
Finally a representation with both (R, A%) measurements (Fig. 5a) permitted to classify the samples and correlate with both (Cu, S) contents. We have found five groups in our samples rounded in the figure and described in Table 1. Their behavior can be summarized as follows: Samples in group A of the table have averaged compositions closed to [In]/[S] = 2 and resulted insulating and transparent till Cu doping was higher than 9%. Conductivity and IR absorption rose with higher Cu content as expected through a Drude model with increasing free carrier concentration. This system seems to saturate in an averaged composition around [Cu] = 19%, [In] = 46%, [S] = 35%. After that we observe a jump down in conductivity while IR Absorption continues growing. This corresponds to the C to D group jump with a concentration of about [Cu]= 24%, [In]= 28%, [S] = 48%. Further increase of Cu and S contents reaches the averaged CuInS2 stoichiometric composition. Further increase of Cu meanwhile the S content is limited leads to group E where a conducting behavior of the samples is probably due to Cu–S phases. In another set of samples the authors could also check that sulfur incorporation was not only correlated with Cu flux, but also depends on the history of the deposition process and the substrate temperature, which could be minimized by means of a modulated coevaporation method [11]. Finally the same information has been replotted on Fig 5b avoiding grouping. Only a separation between sulfur deficient and fully sulfurized samples by means of a horizontal dashed line has drawn. The upper (fully sulfurized) set of samples has been studied by introducing thickness information in the primitive parameters and represented in the onset. Here sulfur and thickness variations have been decoupled and an exponential like behavior of the absorption coefficient α with resistivity ρ seems to be strongly influenced by the Cu stoichiometric deviations. An open question for the in-situ monitoring use would be the influence of the high deposition temperature on the compositional correlation with the (R, A%) set of parameters.
We have deposited Cu–In–S films by an innovative modulated flux deposition (MFD) procedure performed in a static physical evaporation chamber in compositional ranges as diverse as: [In] ranging 10–70%, [Cu] = 2–43% and [S] = 20–50%. The films, characterized by spectrophotometry and four-terminal sensing measurement, were classified in compositional groups by the use of the two-dimensional parameter (DC resistance, near IR optical absorption A%). Interestingly, the optical absorption A% was linearly correlated with Cu content in the Cu poor compositional range while R presents a jump back related with the variation of S concentration, being more sensitive to Cu content variation when S content is fixed. The independence of R and A% in this case makes this measurement set useful as an end-point method for compositional detection of CuInS2 films with similar thicknesses, but the effect of deposition temperature and the relation with morphology and carriers mobility need to be further investigated. Acknowledgments This work has been supported by Spanish government MEC projects: CONSOLIDER GENESIS FV (CSD2006-04) and MAT2005-06738-C02-02, by Madrid Community in the framework of The IV PRICIT through the project FOTOFLEX-CM (S-0505/ENE-0213) and by Fundación Ramón Areces Project: “Células fotovoltaicas ligeras y flexibles”. A.B. acknowledges support through a “Ramón y Cajal” contract from the Spanish Ministry of Education and Science. References [1] R. Klenk, J. Klaer, R. Scheer, M.Ch. Lux-Steiner, I. Luck, N. Meyer, U. Rühle, Thin Solid Films 480 (2005) 509. [2] A. Amara, W. Rezaiki, A. Ferd, A. Hendaoui, A. Drici, M. Guerioune, J.C. Bernède, M. Morsli, Sol. Energy Mater. Sol. Cells 91 (2007) 1916. [3] E. Rudigier, Ch. Pietzker, M. Wimbor, I. Luck, J. Klaer, R. Scheer, B. Barcones, T. Jawhari Colin, J. Alvarez-Garcia, A. Perez-Rodriguez, A. Romano-Rodriguez, Thin Solid Films 431–432 (2003) 110. [4] K. Sakurai, R. Hunger, R. Scheer, C.A. Kaufmann, A. Yamada, T. Baba, Y. Kimura, K. Matsubara, P. Fons, H. Nakanishi, S. Niki, Prog. Photovolt: Res. Appl. 12 (2004) 219. [5] K. Müller, R. Scheer, Y. Burkov, D. Schmeisser, Thin Solid Films 451–452 (2004) 120. [6] I. Hamberg, C.G. Granqvist, J. Appl. Phys. 60 (11) (1986) R123. [7] H. Fujiwara, M. Kondo, Phys. Rev. B 71 (2005) 075109. [8] C. Guillén, T. García, J. Herrero, M.T. Gutiérrez, F. Briones, Thin Solid Films 451 (2004) 112. [9] J.F. Trigo, B. Asenjo, J. Herrero, M.T. Gutiérrez, Sol. Energy Mater. Sol. Cells 92 (9) (2008) 1145. [10] K. Gelin, E. Wäckelgård, J. Phys: Condens. Matter 16 (2004) 833. [11] A. Bollero, J.F. Trigo, J. Herrero, M.T. Gutiérrez, Thin Solid Films 517 (7) (2009) 2167 (this issue), doi:10.1016/j.tsf.2008.10.081.