Photoconductivity relaxation processes in Cu1−xZnxInS2 solid solutions

Materials Science in Semiconductor Processing 39 (2015) 665–670

Contents lists available at ScienceDirect

Materials Science in Semiconductor Processing journal homepage: www.elsevier.com/locate/mssp

Photoconductivity relaxation processes in Cu1
xZnxInS2 solid solutions V.V. Bozhko a,n, A.V. Novosad a, O.V. Parasyuk a, N. Vainorius b, V. Vertelis b, A. Nekrošius b, V. Kažukauskas b,n a

Lesya Ukrainka Eastern European National University, 43025 Lutsk, Ukraine Semiconductor Physics Department and Institute of Applied Research, Vilnius University, Saulėtekio al. 9, bldg.3, LT-10222 Vilnius, Lithuania b

a r t i c l e i n f o

PACS: 72.20 72.40 72.80 73.20 Keywords: Cu1
xZnxIn2S4 solid solutions Photoconductivity Relaxation Recombination centers

abstract Photoconductivity and its relaxation in Cu1
xZnxIn2S4 alloys (x ¼ 0–16) single crystals were investigated in the temperature range 30–100 K. The long-lasting relaxation processes (τ  103 s) and induced photoconductivity phenomena were identified. The main parameters characterizing the photoconductivity kinetic were determined. The observed phenomena were analyzed by taking into account effects of the trapping levels. It is shown that relaxation of the photoconductivity and induced photoconductivity are controlled by the multicenter recombination in which both 'fast' and 'slow' recombination centers take part. & 2015 Elsevier Ltd. All rights reserved.

1. Introduction CuInS2 compound and CuInS2-based alloys have found application in optoelectronics because of the optimal combination of their optical and electronic properties. Particularly, they are promising materials for the absorbing layers in thin film heterojunction solar cells [1–3]. Numerous experimental data show that the physical properties of CuInS2 compound and CuInS2-based alloys are controlled to a large extent by point defects of the crystal lattice [4–7]. Thus, the in-depth studies of the behavior of defects in these semiconductors can reveal new ways of controlling the (photo-)electrical characteristics, and spectral sensitivity in particular. The long-lasting ( 103 s) photoconductivity relaxations in Cu1
xZnxInS2 alloys single crystals were first reported by

n

Corresponding author. E-mail addresses: [email protected] (V.V. Bozhko), [email protected] (V. Kažukauskas). http://dx.doi.org/10.1016/j.mssp.2015.06.007 1369-8001/& 2015 Elsevier Ltd. All rights reserved.

Novosad et al. [8]. Semiconductors, in which the long-lasting relaxations of photoconductivity appear, are widely used as materials for the slow-relaxing photoresistors and photomemory systems [9]. Moreover, study of the photoconductivity and its relaxations is one of the most used methods for determining parameters of the localized states in the band gap, as well as the nonequilibrium charge carrier lifetimes [10,11]. We have demonstrated earlier that photoconductivity is a sensitive tool for the analysis of charge transport in defective and inhomogeneous materials as well [12–17]. The purpose of this study is investigation of the photoconductivity spectra and its relaxation in Cu1
xZnxInS2 alloys single crystals in the temperature range from 30 up to 100 K. Attention is paid to the influence of the cationvacancy defects on the photoelectrical properties of crystals. As described in detail in [18] the horizontal version of the Bridgman–Stockbarger method was used for the solid solution crystal growth. Initial compositions were selected within the homogeneity range of the solid solutions at 870 K. We have demonstrated in [18] that solid solubility of

666

V.V. Bozhko et al. / Materials Science in Semiconductor Processing 39 (2015) 665–670

ZnIn2S4 in the low temperature modification of CuInS2 with chalcopyrite structure can be achieved in the ranges of up to 12 mol% ZnIn2S4 at the annealing temperature. Thus samples with different ratios of ZnIn2S4 were grown in order to compare their structural, physical, (photo-)electrical and photocurrent transient properties as influenced by the cation-vacancy defect states.

2. Experimental We have investigated CuInS2–ZnIn2S4 solid solutions containing 4, 8, 12 and 16 mol% of the ZnInSe4. In our previous publications we have proposed the single crystal growth technology of CuInS2–ZnIn2S4 and CuInS2–ZnIn2Se4 crystals, presented results of their X-ray structural analysis and investigations of the physical and electrical properties in the steady state mode at  30 K and the room temperature [18–21]. In the present paper we will focus on the photoconductivity transient properties at different temperatures in the temperature range from  30 up to 100 K as they are influenced by the cation-vacancy defects. All the crystals of CuInS2–ZnIn2S4 alloys had n-type conductivity. The crystals containing 4, 8, and 12 mol% of ZnIn2S4 were the single crystalline ones, demonstrating the same structure in all points. In contrast, samples containing as much as 16 mol% ZnIn2S4 were the two-phase ones, i.e., different structural phases could be identified and the crystallites of ZnIn2S4 were formed in the grown samples. The composition of grown crystals was controlled by EDX analysis. It had confirmed that the content of the elements in the solid solutions was coinciding well with the composition of the initial charge. For example, in the СuInS2–ZnIn2S4 alloy with 12 mol% of ZnIn2S4, the determined content of the elements Cu:Zn:In:S was 20.68:2.55:25.43:52.99 at%; while in the initial mixture, the corresponding ratio was 20.17:2.76:25.69:51.38 at%. The measurements of the (photo-) electrical properties were carried out using the samples fabricated in the form of regular parallelepipeds  6  2  1 mm3 in size. The samples were mechanically and afterwards chemically polished. The relaxation processes and photoconductivity spectra were measured in the temperature regions TE30–100 K. In order to avoid pre-excitation of the samples by highenergy photons, photoconductivity spectra were first measured starting from the low quanta energies of the exciting light. After that, the measurements were repeated in the reverse direction starting from the highest quanta energies. To assure the stationary measurement conditions, duration of the single scans was as long as 2600 s. Kinetics of the photoconductivity growth and decay were recorded after the excitation of the samples by light corresponding to the maxima of the induced photoconductivity. The energy position of these peaks was found to be 1.32 eV, 1.35 eV and 1.30 eV for the samples with 8, 12 and 16 mol% of ZnIn2S4, respectively [18]. The light excitation of the samples lasted for 3000 s till the photocurrent saturation was reached, after that light was turned off and relaxation kinetics were measured. The electrical measurements were performed using Keithley 6430 Sub-Femtoamp SourceMeter.

3. Results and discussion 3.1. Photoconductivity We have observed in Ref. [18] that structural and photoelectrical properties of the СuInS2–ZnIn2S4 crystals with as low as 4 mol% of ZnIn2S4 are very similar to that of СuInS2 crystals, indicating that in the samples with 4 mol% of ZnIn2S4 effect of the vacancy type cationic defects is negligible because of the very low amount of Zn. Thus, we also did not observe any peculiarities in the transient behavior of their photoconductive properties. Therefore we will focus on the crystals with larger amounts of ZnIn2S4 in which defect-related features were expressed the best. The induced impurity-related photoconductivity was observed in Cu1
xZnxInS2 single crystals, containing  8 and 12 mol% ZnIn2S4 at low temperature (T E30 K) in [18]. The explanation and model of the induced impurity-related photoconductivity was first proposed in [22]. To explain the experimental results, in [18] we had proposed the two-defect-center recombination model. According to the model, the impurity and induced photoconductivity phenomena are caused by the transitions to the valence band from the acceptor levels formed by VCu and VIn, respectively. It was argued that VIn play the role of the fast recombination s-centers, and the slow recombination r-centers are formed by VCu [18,21]. Meanwhile two donor levels, which are caused by VS, play the role of shallow traps (t-levels). The thermal activation energies (ED) of these levels were determined from the spectra of thermally stimulated currents in [18]. They were found to be ED1: 0.016 eV and 0.023 eV, ED2: 0.081 eV and 0.117 eV for the samples containing 8 and 12 mol% ZnIn2S4, respectively. The photoconductivity spectra at different temperatures are shown in Fig.1. Increase in the height of the induced photoconductivity peak compared to that of the impurity photoconductivity (shortwave maximum) can be explained by the fact that upon light excitation trap levels become filled, thus more generated electrons can take part in the photoconductivity without being trapped. This means that the trapping time constant of electrons exceeds the electron lifetime. Reduction of the difference in heights of the maxima of the impurity photoconductivity and induced photoconductivity at higher temperatures in Fig. 1 is caused by the onset of full or partial thermal activation of electrons from t-levels. This assumption is supported by the fact that the thermally stimulated conductivity maxima are observed at  70 K and  80 K in the samples with 8 and 12 mol% ZnIn2S4, respectively [18]. Characteristically, in Fig.1, depicting compound with 12 mol% of ZnIn2S4, two distinctive photoconductivity maxima appear by scanning spectra down. The induced maximum appears at  1.36 eV. To explain this phenomenon it can be assumed that it is caused by the increased number of acceptors VCu, resulting in higher recombination probability of nonequilibrium electrons. Thus, after the preliminary light excitation, some electrons will still remain on r-centers, and their transitions to the conductivity band will cause the second (high-energy) maximum.

V.V. Bozhko et al. / Materials Science in Semiconductor Processing 39 (2015) 665–670

Shift of the peaks of the impurity photoconductivity towards longer wavelengths with increasing temperature is presumably due to the decrease of Eg. This shift is marked by the doted straight line in Fig.1. The thermal coefficient of the change in the position of the photoconductivity peak is 4.3∙10
4 eV K
1, being consistent with the thermal coefficient of other chalcogenide compounds [22,23].

3.2. Kinetics of photoconductivity relaxation Kinetics of the photoconductivity growth and decay of Cu1
xZnxInS2 single crystals are shown in Figs. 2–4. It can be seen that these long-lasting relaxation processes are of complex nature. In case of impurity excitation

667

photoconductivity growth in semiconductors can be described by the expression as follows [11,21]:    t Δσ ¼ Δσ st U 1
exp
: ð1Þ

τ1

Here τ1 – photoconductivity relaxation time upon excitation, Δσst – stationary nonequilibrium conductivity. Fig. 2 shows photoconductivity growth of CuInS2– ZnIn2S4 solid solutions containing 12 and 16 mol% of ZnIn2S4. In the semi-logarithmic scaling they are well described by the straight lines in a wide time interval. From the slope of these straight lines the relaxation time constants of photoconductivity τ1 were defined (Table 1). Large values of τ1 ( 102–103 s) indicate participation of the shallow traps (t-levels) or recombination barriers in the photoconductivity relaxation. To

Fig. 1. Photoconductivity spectra of the СuInS2–ZnIn2S4 alloys containing 8 and 12 mol% of ZnIn2S4 at different temperatures.

Fig. 2. Kinetics of the photoconductivity growth of the СuInS2–ZnIn2S4 alloys containing 12 and 16 mol% of ZnIn2S4 at different temperatures.

668

V.V. Bozhko et al. / Materials Science in Semiconductor Processing 39 (2015) 665–670

the second model electrons and holes become spatially separated by potential barriers caused by fluctuations of the band gap edges. Such spatial fluctuations can arise in highly compensated crystals due to local inhomogeneities of doping, or around clusters of defects, or can be caused by, e.g., ionizing. Thus, in order to take part in conductivity, charge carriers have to overcome the so-called mobility barriers. In order to recombine and/or to be trapped charge carriers of different signs have to move in space to meet each other by overcoming so-called recombination barriers. Both these phenomena prevent the effective recombination of the generated carriers, thus causing the long relaxation times. In most cases, the differentiation of both relaxation mechanisms is a challenging and complex task. As we have studied the relaxation of impurity photoconductivity induced by light quanta from the impurity region, the analysis of the experimental results based on the relaxation model due to the capture of free carriers on localized centers is more appropriate. In such case, the relaxation time constant of photoconductivity is not equal to the lifetime of the photoexcited charge carriers, and will significantly exceed it, since part of the electrons will be captured on the t-levels. Moreover, as compared to the band-to-band recombination, the relaxation time constants are much longer because the main recombination channels are transitions of the electrons to defect sites – the r or s recombination centers [11,21]. Thus, presence of

explain such long relaxation times of photoconductivity two main models can be taken into account. The first one is based on the representation of the capture of nonequilibrium charge carriers by the strongly localized states in the vicinity of point defects (traps) and/ or impurities with deep energy levels in the band gap. In

Fig. 3. Kinetics of the photoconductivity growth of the СuInS2–ZnIn2S4 alloy containing 8 mol% of ZnIn2S4 at different temperatures.

Fig. 4. Kinetics of the photoconductivity decay of СuInS2–ZnIn2S4 alloys containing 8 and 12 mol% of ZnIn2S4 at different temperatures.

Table 1 Temporal parameters of relaxation processes in SuInS2-ZnIn2S4. 8 mol% ZnIn2S4

12 mol% ZnIn2S4

16 mol% ZnIn2S4

T [K]

τ1 (s)

τ2 (s)

τ3 (s)

τ4 (s)

T [K]

τ1 (s)

τ3 (s)

τ4 (s)

T [K]

τ1 (s)

τ3 (s)

τ4 (s)

30 50 73 100

600 300 200 140

1420 1170 1040 860

80 50 30 15

1550 1150 870 680

30 40 65 80

1060 900 540 400

 10  10  10  10

1250 1000 660 500

30 63 80 100

925 600 320 210

100 60 40 25

1700 820 610 470

V.V. Bozhko et al. / Materials Science in Semiconductor Processing 39 (2015) 665–670

the trapping t-levels which capture the light generated electrons temporarily preventing them from taking part in the photoconductivity because of the slow exchange with the conduction band is the cause of the slow growth of the photoconductivity. Long-lasting photoconductivity relaxations in other semiconductor compounds were observed in Refs. [24–27]. Characteristically, in Fig. 2 deviations from the exponential dependencies of the photoconductivity vs time are seen at the initial stages in the temperature region 30– 60 K in the samples with 12 and 16 mol% of ZnIn2S4. The detailed analysis showed that after the onset of excitation photoconductivity increases linearly with time. Such linear behavior is characteristic in case of nearly empty t-levels [11,21], e.g., in highly compensated semiconductors, including Cu1
xZnxInS2 with the high content of ZnIn2S4. The reason for the linear growth is that the balance between the conduction band and the t-levels sets some time after the beginning of the photoexcitation. After the establishment of equilibrium between the conduction band and t-levels, alongside with the photoexcitation of electrons, they start recombining through the r- and scenters. In the samples with lower ZnIn2S4 content (8 mol%) two straight-line segments can be singled out in the photocon

ductivity growth process in
ln 1
Δσ =Δσ st
t scaling, see Fig. 3. The evaluated characteristic time constants of different photoconductivity growth and relaxation parts are presented in Table 1. In this Table the values of τ1 and τ2 characterize the growth of photoconductivity upon the light excitation, meanwhile τ3 and τ4 describe its decay after the light is switched off. At the initial stage the increase of photoconductivity occurs with a characteristic time τ1 (Table 1). Later on (approximately 500 s at 30 K) photoconductivity growth slows down and is described by the relaxation time τ2 4 τ1 (Table 1). This could be explained by the fact that when the trapping centers become more and more filled the recombination of the carriers via the recombination centers starts playing its role, and thus further slows the photoconductivity growth. This assumption is confirmed by the fact, that τ2 is close to the slow recombination time constant τ4 (Table 1) as will be discussed below. Both τ1 and τ2 decrease with increasing temperature. Different photoconductivity kinetics in the samples with different ZnIn2S4 content can be due to the different concentrations of VCu resulting in a redistribution of electrons at defect levels. Similar processes were observed in Cu-doped Ge [21]. In the CuInS2–ZnIn2S4 alloys with the lowest content of the second component ( 4 mol% ZnIn2S4) the long-lasting photoconductivity relaxations were not observed. This is consistent with the data of Ref. [18], according to which in СuInS2– ZnIn2S4 with 4 mol% ZnIn2S4 as well as in СuInS2 crystals the photoconductivity peak corresponds to band-to-band optical transitions. Thus, low content of Zn atoms is the reason for such variation of photoelectrical properties. As is known e.g., [18], during the formation of СuInS2– ZnIn2S4 solid solutions, two Cu atoms are substituted by one Zn atom in their crystallographic positions 4a. This causes formation of the tetrahedral voids and appearance of VCu. Concentration of VCu grows with increasing ZnIn2S4

669

content; this leads to an increase in the degree of compensation of the electrons. In Fig. 4 photoconductivity decay after the light excitation is switched off is presented. It is seen that the photoconductivity relaxation in crystals with 8 and 12 mol% ZnIn2S4 is characterized by the presence of at least two recombination channels with different characteristic recombination times. This supports our conclusion about the presence of two different recombination centers: that of the fast recombination – s-centers and of the slow ones – r- centers. In the two-phase Cu1
xZnxInS2 alloys with 16 mol% of ZnIn2S4 the results are similar. The photoconductivity relaxation in this case is described well by the sum of two exponents as follows:     t t Δσ ¼ A U exp
þ B U exp
ð2Þ

τ3

τ4

here A  B  Δσ st . From the slopes of the straight lines (Fig. 4.) the relaxation time constants τ3 and τ4 were estimated. The fast and the slow components with characteristic relaxation times: τ3 10 s and τ4  103 s, respectively, could be singled out. Both these values decrease with increasing temperature, as can be seen from Table 1. The fast recombination channel, which is observed at the initial stages could be due to electron capture by the s-fast recombination centers. The slow relaxation component is given by r-recombination centers. In the presence of electron trapping centers the photoconductivity relaxation time decrease with increasing temperature can be described as in Ref. [11]:    N E ð3Þ τ ¼ τn U 1 þ t Uexp t Nc kT here τn – nonequilibrium electron lifetimes, Nt – concentration of trapping centers, Nc – effective density of electron states in the conduction band, Et – energy distance from the electron trap level to the conduction band. As follows from Eq. (3), a temperature increase of the photoconductivity relaxation time decreases with increasing temperature, approaching τn. Thus, the lower the temperature and the higher the concentration of traps, the more the photoconductivity relaxation time constants differ from the lifetime of nonequilibrium charge carriers. The high concentration of traps in the investigated compounds is evidenced by the high induced photoconductivity. Thus, assuming that τ⪢τn in the used temperature interval, Eq. (3) can be rewritten as   N E ð3aÞ τ  τn U t U exp t Nc kT Therefore Et could be determined from the slope of the temperature dependencies of the relaxation time constants in ln τ
1=T scaling. The resulting τ dependencies on temperature are well described by straight lines in Fig. 5. From the slope of these lines Et values of 0.11 eV, 0.15 eV and 0.18 eV were evaluated for CuInS2–ZnIn2S4 alloys with 8, 12 and 16 mol% of ZnIn2S4, respectively. These values characterizing t-trapping centers are close to those, evaluated in Ref. [18] from the thermally stimulated current spectra.

670

V.V. Bozhko et al. / Materials Science in Semiconductor Processing 39 (2015) 665–670

Fig. 5. Dependecies of the relaxation time constants of the slow relaxation component on the temperature of CuInS2–ZnIn2S4 solid solutions with different ZnIn2S4 content of, mol%; 1–8; 2–12; 3–16.

4. Conclusions It was shown that the photoconductivity relaxation kinetics of Cu1
xZnxInS2 compounds is characterized at low temperatures by two types of relaxation processes: the fast and the slow ones with characteristic relaxation times of  101 s and 103 s, respectively. These two relaxation processes are controlled by the s-centers of fast recombination and r-centers of slow recombination. The thermal activation energies of the trapping t-levels formed by Vs were evaluated from the temperature dependencies of the relaxation time constants of photoconductivity.

Acknowledgments The partial financial support from the Research Council of Lithuania (Project no. TAP LU 02-2014) and the State Agency on Science, Innovations and Informatization of Ukraine (Grant no. M/106-2014) is acknowledged. References [1] R. Scheer, T. Walter, H.W. Schock, M.L. Fearheiley, H.J. Lewerenz, Appl. Phys. Lett. 63 (1993) 3294–3296.

[2] A. Mere, O. Kijatkina, H. Rebane, J. Krustok, M. Krunks, J. Phys. Chem. Solids 64 (2003) 2025–2029. [3] D. Lee, K. Yong, J. Phys. Chem. C 118 (2014) 7788–7800. [4] A.W. Verheijen, L.J. Giling, J. Bloem, Mater. Res. Bull. 14 (1979) 237–240. [5] H.Y. Ueng, H.L. Hwang, J. Appl. Phys. 22 (1987) 434–438. [6] N. Lablou, G. Massé, J. Appl, Phys. J. Appl. Phys. 52 (1981) 978–982. [7] R. Garuthara, R. Wijesundara, W. Siripala, Sol. Energy Mater. Sol. Cells 79 (2003) 331–338. [8] A.V. Novosad, V.V. Bozhko, G. E. Davidyuk, V.R. Kozer, O.V. Parasyuk, V. Kažukauskas. Kinetics of Photoconductivity of Solid solutions Cu1
xZnxInS2 (x ¼0–0,16). in: Proceedings of the VI International Conference On Relaxational, Nonlinear and Acoustooptical Processes and Materials. Lutsk, Ukraine, 2–5 June 2012, p. 230 (In Russian). [9] V.S. Vavilov, P.C. Euthymiou, G.E. Zardas, Phys.-Uspekhi 42 (1999) 199–201. [10] Yu. V. Vorobjev, V.N. Dobrovolskij, V.I. Stricha, Methods of Investigation of Semiconductors, Kiev, Vysshja Shkola, 1988 In Russian. [11] V.V. Serdyuk, G.G. Cheremesyuk, M. Terek., Phototelectrical Processes in Semiconductors, Kiev-Odessa, Vysshja Shkola, 1982 In Russian. [12] V. Kažukauskas, J. Vaitkus, Z. Phys. B 94 (4) (1994) 401–407. [13] J. Vaitkus, V. Kažukauskas, R. Kiliulis, J. Storasta., Inst. Phys. Conf. Se. 136 (136) (1994) 755–760. [14] R. Kiliulis, V. Kažukauskas, J. Storasta, J.V. Vaitkus, J. Phys. I 6 (9) (1996) 1165–1187. [15] V. Kažukauskas, G. Kuhnel, W. Siegel, Appl. Phys. Lett. 70 (13) (1997) 1751–1753. [16] V. Kažukauskas, J. Appl. Phys. 84 (4) (1998) 2053–2061. [17] A. Affolder, A. Aleev, P.P Allport, et al., Nucl. Instrum. Methods A 658 (1) (2011) 11–16. [18] V.V. Bozhko, A.V. Novosad, G.E. Davidyuk, V.R. Kozer, O.V. Parasyuk, N. Vainorius, V. Janonis, A. Sakavičius, V. Kažukauskas, J. Alloy. Compd. 553 (2013) 48–52. [19] V.R. Kozer, V.V. Bozhko, O.V. Parasyuk, O.V. Novosad, A.O. Fedorchuk, J. Mater. Sci. Mater. Electron. 25 (1) (2014) 163–167. [20] V.V. Bozhko, A.V. Novosad, O.V. Parasyuk, O.Y. Khyzhun, N. Vainorius, A. Nekrošius, V. Vertelis, V. Kažukauskas, J. Phys. Chem. Solids 82 (2015) 42–49. [21] V.A. Ivanov, V.F. Gremenyuk, E.P. Zareckaja, O.N. Sergejeva, I.A. Viktorov, V.B. Zalesskij. Internatinal Conference on Topical Problems of Solid State Physics, Minsk, 26–28 October 2005, p. 426–429 (In Russian). [22] S.M. Ryvkin, Photoelectrical Phenomena in Semiconductors, Fizmatgiz, Moscow, 1963 In Russian. [23] L.L. Kazmerski, C.C. Shien., Thin Solid Films 41 (1977) 35. [24] M.S. Yunusov, M. Karimov, B.L. Oksengendler, Phys. Tekh. Poluprovodn. 32 (1998) 264–267. In Russian. [25] T.L. Mayorova, V.G. Kliujev, T.V. Samophalova, Phys. Tekh. Poluprovodn. 45 (2011) 577–581. In Russian. [26] S.E. Malchanov, Phys. Tekh. Poluprovodnikov. 33 (1999) 567–568. In Russian. [27] E.P. Skipetrov, E.A. Zvereva, A.V. Golubev, V.E. Slynko, Appl. Phys. 2 (2006) 72–77. In Russian.