Copassivation of polycrystalline CdTe absorber by CuCl thin films for CdTe solar cells

Copassivation of polycrystalline CdTe absorber by CuCl thin films for CdTe solar cells

Applied Surface Science 484 (2019) 1214–1222 Contents lists available at ScienceDirect Applied Surface Science journal homepage: www.elsevier.com/lo...

3MB Sizes 0 Downloads 7 Views

Applied Surface Science 484 (2019) 1214–1222

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Full length article

Copassivation of polycrystalline CdTe absorber by CuCl thin films for CdTe solar cells

T



Jian Huang, Dan Yang, Wei Li , Jingquan Zhang, Lili Wu, Wenwu Wang College of Materials Science and Engineering, Sichuan University, Chengdu 610064, China

A R T I C LE I N FO

A B S T R A C T

Keywords: CuCl treatment Copassivation Back contact CdTe solar cells

Copper has been introduced in CdTe solar cell manufacturing for a long time and is critical for achieving high performance devices. In this work, CuCl thin films were prepared by vacuum thermal deposition and then used as a buffer layer with a thermal annealing treatment for activation in the CdTe solar cells. The results indicate that Cu and Cl diffused into the CdTe absorber which was beneficial for doping and copassivation. The carrier density for CdTe solar cells in the vicinity of the pn junction was significantly improved from 6.1 × 1013 to and Voc was increased by > 70 mV. The cell with the CuCl thin films was more uniform and 4.3 × 1014 cm−3 , revealed higher photocurrent response than the reference cell. The results also indicate that the formation of ohmic back contacts to CdTe solar cells due to the presence of the CuxTe (Cu7Te4) thin films. An optimal device with open circuit voltage of ~820 mV, and fill factor of ~72% and efficiency reaching 16.69% was obtained by the control of the copper concentration and the thermal activation conditions of the CuCl buffer layer.

1. Introduction Cadmium telluride (CdTe) is a II-VI compound semiconductor with a direct band gap of about 1.45 eV which is nearly ideal for photovoltaic solar cells. CdTe also has a larger optical absorption coefficient (> 105 cm−1) in comparison with a-Si, which means only several micron CdTe can absorb as high as 90% photons. CdTe has attracted so much attention since the first CdS/CdTe solar cell prepared in 1982, and nowadays it becomes one of the most promising and leading thin film photovoltaic technologies [1,2]. Undergoing a stagnant period for more than a decade for CdTe solar cells with the record device efficiency of 16.5% [3],it has taken off with the world-record efficiency dramatically increased to 22.1% by First Solar [4]. Although CdTe solar cells are now competitive with Si-based photovoltaic devices, there remain some technical difficulties during the device fabrication. One is the common issue in the polycrystalline solar cells known as grain boundaries (GBs) between grains, which affect the carrier lifetime due to the recombination. Another is difficulty in realizing a good ohmic contact owing to the high work function of p-type CdTe [5]. CdCl2 treatment has been a standard process in the manufacture of CdTe based devices. It is believed that Cl tends to accumulate at GBs passivating these levels thereby enhancing the electrical properties of polycrystalline CdTe and assisting grain growth and recrystallization



[6,7]. Also, Cl improves p doping of CdTe, possibly because of the formation of VCd-ClTe complex [8,9]. Previous articles have proved that Cl prefers to segregate at the CdS/CdTe interface eliminating active states, presumably by enhancing the interdiffusion at the interface [6]. The work function of the common metals is lower than 5.7 eV, so a Schottky barrier tends to form at the back contact of CdTe solar cells, which leads to blocking hole transmission and increasing recombination. There are two ways of lowering the barrier height. The etching method using solutions such as Br2/methanol or nitric acid/phosphoric acid mixtures is commonly performed in order to form a thin Te-rich layer which can lower the valence band offset, followed by a metal contact deposition. Another widely used method is to deposit a heavily p-doped back contact which narrows the barrier width and allows tunneling from the CdTe absorber to the metal electrodes in the valence band. Among back contact materials, we can divide them into two types concerning whether there is copper containing or not. Copper containing back contacts include Cu/Au [10], CuxTe (x ≤ 2) [11–15], ZnTe:Cu [16–18], Te/Cu bilayer [19], Cu2O [20], Cu2S [21]. Another type is commonly called copper-free back contacts such as Sb2Te3 [22,23], V2O5 [24], graphene [25], MoOx [24,26,27]. Copper used in CdTe solar cells is responsible for the long-term stability due to its high diffusion. Appropriate amount of copper is beneficial for doping CdTe thus increasing carrier concentration. As we know, film deposition process (in-situ) or annealing process

Corresponding author. E-mail address: [email protected] (W. Li).

https://doi.org/10.1016/j.apsusc.2019.03.253 Received 24 December 2018; Received in revised form 7 March 2019; Accepted 23 March 2019 Available online 23 March 2019 0169-4332/ © 2019 Elsevier B.V. All rights reserved.

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

Fig. 1. X-ray diffraction patterns of CuCl deposited on (a) corning glass substrates and the CdTe layer with the thicknesses of (b) 10 nm, (c) 100 nm; GIXRD patterns of CuCl deposited on CdTe with the thicknesses of (d) 10 nm, (e) 30 nm, and (f) 50 nm. Inset is a larger version of Fig. 1c at 2θ from 25 to 30°.

into CdTe and be in the form of Cu-related defects. There are four major Cu-related defects which can be described by the following three reactions in CdTe [33],

Table 1 Parameters for cells treated with varied thickness of CuCl. Thickness

Reference 5 nm 10 nm 20 nm 30 nm

Eff (%)

Rsh (Ω·cm2)

Rs (Ω·cm2)

FF (%)

Jsc (mA/cm2)

Voc (mV)

Area (cm2)

12.83 15.02 16.21 14.80 13.89

645 1387 1337 896 559

6.16 2.90 2.67 3.04 3.23

65.59 72.80 75.24 71.86 69.47

27.08 26.07 26.95 26.09 26.44

723 791 800 789 756

0.24 0.24 0.24 0.24 0.24

2− − Cui+ + VCd ⇌ CuCd

(1)

2− Cdi2 + + VCd ⇌ CdCd

(2)

− Cui+ + CdCd ⇌ CuCd + Cdi2 +.

(3)

Once excessive copper is introduced into CdTe, the situation gets complicated because copper will be in the form of not only substitutional at Cd site (CuCd) acted as acceptors but also interstitial served as donors even tightly-bounded complexes (e.g. Cui-CuCd and Cdi-CuCd) [33]. Fortunately, formation energy of CuCd is the lowest among these impurity levels [34]. As a result, controlling the Cu concentration is of vital importance for CdTe solar cells [7–17,35]. Furthermore, Zhang et al. [36] studied the copassivation effect of Cl and Cu in the CdTe solar cells using first-principles calculations. They found that the Te core creates a high defect density below the conduction band minimum, but all these levels can be removed by copassivation of Cl and Cu. Thus, CuCl binary compound material deserves more care and attention on the copassivation processing in the CdTe solar cells, rather than on material preparation or device fabrication as the precursor [37,38]. Moore, et al. [39] employed a close spaced sublimation (CSS) method to deposit a very thin CuCl layer for CdTe solar cells at the temperature of about 190 °C and followed by a post annealing treatment. However, their work focused on the effects of the Te layer on cell performance. In this work, vacuum thermal

after doping (ex-situ) can increase the carrier density in the semiconductor compound system. For example, by using RF-plasma assisted molecular beam epitaxy, nitrogen or Cl can be doped in the ZnSe wafer to form p- or n-type semiconductor compounds with high carrier densities [28]. It has been found that the charge carrier concentration of ntype nanostructured BixSb2-xTe3 thermoelectric thin films can be tuned by post-growth heat treatment [29]. However, the in-situ doping and interdiffusion for the semiconductors cannot solve the problems of the position of Fermi level and the solubility limitation [30,31]. Kennedy et al. [32] synthesized Al-doped ZnO films by ion implantation for thermoelectric applications. The resultant carrier concentration for undoped ZnO was 8.8 × 1019 cm−3, while Al doping increased the carrier concentration with a maximum value of 4.3 × 1020 cm−3 observed for 1% Al-doped ZnO thin films. As for CdTe, a commonly used way is diffusion which is involved in annealing process after doping [10,16–19]. The diffusion is the thermal process and the electronic bonding, which helps copper diffuse deeply 1215

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

Fig. 2. The box charts of devices annealed for various time.

followed by a rapid thermal annealing at 284 °C. For reference devices, ~150 nm Au was directly deposited after 0.2% (v/v) Br2/CH3OH solution etching. The structural properties of the films were investigated by X-ray diffraction (XRD; DX-2600, Dandong, China) and grazing incidence Xray diffraction (GIXRD; X'Pert Pro MPD DY129, Holland) using Cu Kα radiation (λ = 1.5405 Å). Measurements were taken at 2θ angles of 10–70° or 10–80°. X-ray photoelectron spectroscopy (XPS; AXIS UltraDLD, KRATOS, U.K.) was carried out to determine the composition of the CuCl treated CdTe back surface using Al Kα radiation (hν = 1486.6 eV) as X-ray source. All the XPS spectra were calibrated based on C 1 s peak at 284.6 eV. Secondary ion mass spectroscopy (SIMS) was performed to understand the diffusion behavior of Cu and Cl. As for devices characterization, light J-V measurement (XJCM-9, Gsolar, China) under simulated AM1.5 illumination and dark J-V measurement (4155C, Agilent, USA) were used to measure the voltagecurrent characteristics of the cells. Quantum efficiency (QE) and apparent quantum efficiency (AQE) of the devices were investigated using a QE/IPCE measurement system (QEX10, PV Measurements Inc.) over the wavelength range from 280 to1000 nm. Agilent 4155C semiconductor parameter analyzer was also employed in dark capacitance voltage (C-V) measurements. Laser beam induced current measurement (LBIC, SCS10-LBIC01, Zolix, China) was employed to characterize the uniformity for the solar cells.

deposition method was conducted at room temperature to make the CuCl buffer layer more uniform on the CdTe back surface. And we tried to understand how CuCl plays an important role on CdTe solar cell performance. 2. Experimental details The cells with the structure of glass/SnO2:F/SnO2/ZnMgO/CdS/ CdSe/CdTe/CuCl/Au were fabricated in this work. The substrates were bi-layer SnO2 coated glass, with the first layer of SnO2: F and followed by an intrinsic SnO2 layer. ZnMgO of 60 nm was deposited onto the substrates by RF magnetron sputtering. Then a 40 nm CdS layer was deposited by chemical bath deposition and the following was a CdSe layer of about 80 nm deposited by magnetron sputtering in pure Ar ambient. For all devices a ~3 μm CdTe absorber layer was grown by CSS at 2100 Pa in an Ar/O2 ambient for 3 min with the substrate temperature of ~565 °C and the source temperature of ~646 °C, followed by the CdCl2 treatment at 388 °C for 35 min in an atmosphere of N2:O2 = 4:1 (samples for SIMS and XPS were annealed at 388 °C for 35 min without CdCl2). Prior to the CuCl treatment, the samples including those used for SIMS, XPS and XRD were etched with 0.2% (v/v) Br2/CH3OH solution for 8 s to remove the oxides and receive a Te-rich surface. Then some of the samples were put in a vacuum chamber equipped with a sample exchange room, with which it was helpful to keep the chamber in a relatively high vacuum and keep the CuCl source (99.95% purity purchased from Codow) from oxidation. When the pressure was below 5 × 10−4 Pa, a CuCl buffer layer was deposited at a deposition rate of 0.032 nm/s by evaporation at room temperature. The following was a post-annealing process to activate the back contact and ~150 nm Au was deposited for metallization. Individual devices with the area of 0.24 cm2 or 0.51 cm2 were isolated by laser scribing which was a kind of integrated advanced technology without destroying the bi-layer SnO2 compared to the etching method. Cells with Cu/Au contacts were prepared by depositing copper of 3 nm and Au of 150 nm,

3. Results and discussion 3.1. Properties of CuCl thin films XRD measurements were used to determine the structure of CuCl thin films deposited on CdTe and corning glass substrates as shown in Fig. 1. Single layer CuCl films are a zinc blende structure with a peak at 2θ angle of 28.5° corresponding to the (111) plane, while CuCl films grown on CdTe reveal a hexagonal structure with the diffraction peak of 1216

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

Fig. 3. The box charts of devices annealed at various temperatures for 15 min. The reference cell refers to the one without the CuCl treatment.

(002) at 2θ = 27.8°. An additional peak at 2θ = 26.6° belongs to SnO2 from the FTO substrates as shown in Fig. 1c. CuCl thin films grown on CdTe were also characterized using GIXRD because it could analyze only tens of nanometers near the surface with grazing-incidence angle of 1° (Fig. 1d–f). With thicker CuCl deposition, the intensity of CdTe decreases but it is still strong (not shown here). A peak at 2θ = 21.3° might be Cu7Te4 which can be observed in all samples and confirmed when another peak at 2θ = 24.7° clearly appears. A diffraction peak at about 2θ of 27.7° and a weak peak at 2θ of 50.5° for the sample with a thickness of 50 nm might be CuCl phase.

Table 2 Photovoltaic parameters of cells annealing at around 260 °C. Annealing process

Eff (%)

Rsh (Ω·cm2)

Rs (Ω·cm2)

FF (%)

Jsc (mA/cm2)

Voc (mV)

Area (cm2)

Cu/Au 255 °C 265 °C 275 °C

15.20 15.74 16.69 15.83

1082 1207 1098 887

3.57 2.91 2.38 2.34

69.60 73.20 72.59 75.53

27.13 26.85 27.96 26.73

805 801 822 784

0.24 0.51 0.51 0.51

Fig. 4. (a) J-V curve and (b) QE response for a CdTe solar cell with CuCl annealed at 265 °C for 15 min. 1217

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

becomes more sufficient naturally with improvements in Voc and FF as the duration increases. Meanwhile, copper diffusion intensifies during the period and can propagate into the main junction leading to deep level defects and recombination centers thus degrading cell performance. When the duration approaches 17 min, evident efficiency loss is observed. Fig. 3 shows the thermal activation for the CdTe solar cells at different temperatures. Annealing at 260 °C gives the apparently enhanced cell performance. The enhancements in FF and Voc might be due to the increase of the hole concentration and the formation of the back contact. The copassivation behavior of Cu and Cl could also lead to the increase in Voc. Additional experiments were carried out to optimize cell performance at annealing temperature of around 260 °C (Table 2, Fig. 4). All devices treated with CuCl exhibit much better performance due to the significant reduction of series resistance (Rs) resulting in the promotion of FF and the copassivation involved increase in Rsh and Voc. Compared with the solar cells with Cu/Au contacts, the devices treated with CuCl are more efficient in lowering Rs. Note that annealing at 265 °C for 15 min greatly enhances the cell efficiency from 12.83% to 16.69%. From Fig. 4(b) we can see that our best CdTe solar cell with CuCl thin films shows uniform collection of ~90% across its responsive range between 370–800 nm. The integral current ~27.72 mA/cm2 for the CdTe solar cell from QE is consistent with the result of J-V measurement as presented in Table 2.

Fig. 5. SIMS depth profiles of Cu and Cl concentrations for the sample with the CuCl treatment.

3.2. Optimization of copassivated CdTe solar cells The deposition and post-deposition conditions of the CuCl buffer layers are of vital importance for CdTe solar cells, i.e., the thickness, annealing temperature and annealing time. CdTe solar cells with different CuCl thicknesses (i.e., 5, 10, 20, 30 nm) were prepared and annealed at 265 °C for 15 min (see Table 1). It illustrates that all samples have considerable improvement in the open circuit voltage (Voc) and fill factor (FF), which reveals a positive effect on cell performance even with only 5 nm CuCl. The sample with 10 nm CuCl has the most extraordinary enhancement among those cells. However, the thickest one behaves degradation of almost all photovoltaic parameters. Note that the shunt resistance (Rsh) declines even contrasted with the reference cell. We can infer that excessive copper diffused into the depletion region forming deep level defects in the bulk CdTe thereby reducing carrier lifetime and increasing recombination [40]. Duration of the post-annealing treatment is also significant because the unfulfilled annealing and overheating both lead to the efficiency loss [16,41]. Several experiments have been conducted as the durations range from 10 to 17 min with invariable CuCl thickness of 10 nm and annealing temperature at 265 °C with an area of 0.51 cm2. As can be seen in Fig. 2, with increasing the annealing time, there is a trend that the efficiency firstly increases and then decreases similar to the previous study on back contacts [21]. Post-annealing treatment is a heat accumulation process for activating the CdTe solar cells. The activation

3.3. Characterization of copassivated CdTe solar cells The Cu and Cl profiles taken from SIMS are shown in Fig. 5. Very high Cu concentration on the sample surface and Cu gradient after diffusion are observed, especially near the back surface of the CdTe solar cell. In the bulk CdTe, Cu shows a nearly uniform distribution at the level of 1018 cm−3. As for Cl, high concentration on the sample surface is also observed, but it is approximately one order of magnitude lower than Cu concentration. The effect of excessive Cu at the back on the CdTe solar cells will be discussed later. The flat region of Cl concentration in bulk CdTe is about 6 × 1018 cm−3. Cu diffuses into CdTe acting as a dopant. As a consequence, the carrier density will be improved. Meanwhile, note that the carrier density is just at the level of 1014 cm−3 (see Fig. 7b), which is much lower than the Cu concentration. This indicates that the compensation effect that lots of Cu involves in the formation of other defects even complexes compensating the acceptors. And it is proved that amount of Cu and Cl accumulated in the GBs will copassivate in polycrystalline CdTe [36,42]. LBIC technique is now widely used to evaluate the local variation of the photocurrent for solar cells [43–45]. Here we applied it to examine

Fig. 6. Variations of LBIC mapping for the cells (a) without CuCl; (b) with CuCl. The measured areas were 3.0 × 3.0 mm with 50-micron resolution 1218

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

Fig. 7. (a) A2/C2-V plots of typical cells with various thicknesses and annealing temperatures of CuCl, (b) related NA of typical cells; Simulated photovoltaic parameters of devices with different NA in CdTe, (c) keeping the deep donor defect density (Nt) at 5 × 1013 cm−3, and (d) Nt = NA. Table 3 The A and J0 values for the typical cells. Sample

A

J0 (mA/cm2)

Reference cell 220 °C 10 nm CuCl 240 °C 10 nm CuCl 260 °C 10 nm CuCl 280 °C 10 nm CuCl 265 °C 10 nm CuCl 265 °C 20 nm CuCl 265 °C 30 nm CuCl

1.55 1.56 1.46 1.55 1.56 1.67 1.54 1.90

1.6 × 10−7 7.0 × 10−7 8.2 × 10−8 5.3 × 10−8 5.8 × 10−6 3.9 × 10−7 2.6 × 10−7 1.4 × 10−5

the uniformity of the devices with and without the CuCl treatment at the wavelength of 633 nm under one sun illumination. In addition, the measured areas were 3.0 × 3.0 mm with 50-micron resolution. Fig. 6 shows the photocurrent mapping of two individual cells and the photocurrent normalization is carried out by setting the highest photocurrent as unity. The cell with the CuCl treatment shows excellent uniformity as most parts of the cell exhibit extremely high response (> 0.95). In contrast, the other one is less uniform and shows a relatively lower photocurrent response. We can draw a conclusion that the use of CuCl helps improve the uniformity of CdTe solar cells thus achieving a preferable FF and Voc. C-V measurement was used to calculate the carrier concentration for typical devices as plotted in Fig. 7(a)–(b). A strong correlation between effective dopant density and temperature is observed. It is noted that lower temperature annealing exhibited a reduction in the carrier

Fig. 8. The dark J-V characteristics of typical cells.

density which might be due to the interstitial Cui, a shallow donor, compensating the p-dopants and thus lowering the hole density. It has been reported that increasing the hole density will efficiently improve Voc for CdTe solar cells [46], which is consistent with our simulation results as shown in Fig. 7(c)–(d). The Voc efficiently improves with the increase of NA, whereas FF and Eff will be affected negatively to a great 1219

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

Fig. 9. The AQE experimental results for (a) an efficient cell treated with CuCl under various voltage bias; (b) cells annealed at various temperatures under Vbias = Voc + 100 mV; (c) cells with various thicknesses of CuCl under Vbias = Voc + 100 mV.

inverse phenomenon could be observed for the Voc of those samples. Combined with the reduction in Rsh of those devices, it suggested that a stronger recombination took place both in the space charge region and the bulk of CdTe layer (see Table 3). From the studies mentioned above, we can see that the role of copassivation of Cu and Cl on CdTe solar cells. As we know, many metals form large Schottky barriers with CdTe resulting in a high contact resistance. Thus, we wonder whether and how the ohmic back contact to CdTe formed in the CdTe solar cells. Current density-voltage (J-V) curves for typical cells are illustrated in Fig. 8 and the corresponding diode ideality factor (A) and dark saturation current (J0) are listed in Table 3. The reference device showed an obvious roll-over owing to the Schottky barrier which resulted from the high work function CdTe in direct contact with metal. The roll-over decreased after the CuCl treatment and even nearly invisible at voltage bias as high as 1.2 V, which demonstrated that the CuCl treatment is an effective way in lowering the back barrier height and thus decreasing Rs as indicated in Tables 1–2. For a well-behaved device, it is essential for the ideality factor in the range of 1.3–2 which indicates that the dominant recombination through trap states is in the depletion region of the CdTe layer [47]. As can be seen in Table 3, all devices were well-behaved according to the A between 1.4 and 1.9. However, the dark saturation current densities of devices annealed at high temperature (280 °C) and with the thickest CuCl layers (30 nm) were 5.8 × 10−6 mA/cm2 and 1.4 × 10−5 mA/cm2, respectively. It was likely due to the excessive copper diffusion into the depletion region which was involved in the formation of deep level defects, as a result, climbing recombination took place and carrier lifetime decreased. This was in agreement with a reduction in Voc which had a strong dependence on minority carrier lifetime [46], as shown in Fig. 3 and Table 1.

Fig. 10. The (a) Te 3d and (b) Cu 2p photoelectron spectra for samples before and after annealing, Te 3d5/2 photoelectron spectra for (c) as-deposited sample and (d) annealed sample. The dash lines stand for the fitting curves.

extent once the deep defect density is large. The most efficient doping at 265 °C obviously increased Voc by > 70 mV compared to the reference cell as given in Table 1. As for the devices with thicker CuCl, the carrier concentration as high as 1.2 × 1015 cm−3 was obtained, however, an 1220

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

solar cell in the vicinity of the pn junction increased remarkably from 6.1 × 1013 to 4.3 × 1014 cm−3. The copassivation effect of Cu and Cl contributes to the enhancement of carrier collection thus increasing the Voc and the uniformity of cells. Moreover, the formation of Cu7Te4 as a back contact could responsible for the decrease of Rs < 2.34 Ω·cm2 and the increase of FF > 75%. By performing the CuCl treatment, an optimal device with the efficiency of 16.69%, Voc > 820 mV and FF > 72% was successfully obtained with the area of 0.51 cm2.

QE measurement is a typical tool to measure the light-generated current under standard conditions, in which one sun illumination and zero bias are required. AQE, instead, is under bias light or voltage bias. An analytical model of AQE under voltage bias following the equations is described by Sites and Gloeckler et al. [48,49], −1

AQE (V ) = AQE (0) ⎡1 + ⎢ ⎣

JF (RL + RS ) R + L⎤ AkT / q rsh ⎥ ⎦

JF = J00 exp [(V − Vbi − JRS )/ AkT ]

(1) (2)

Acknowledgements

where JF, rsh, A and q represent the forward diode current, the shunt resistance, the diode quality factor and elementary charge, respectively. While Rs is the series resistance of the solar cell and RL is the input impedance of the measurement circuit. Vbi in Eq. (2) is the build-in potential of the main junction in CdTe solar cells. At higher forward bias, the forward current may become limited resulted from the backbarrier diode which is in reverse bias at this time thus drifting holes back into the CdTe. Bätzner et al. [50] reported that the depletion layer width of the Schottky diode gradually increased at high forward bias, and consequently, in the depletion region, a current of opposite direction to the junction current (indicated by a 180° phase shift in the AQE signal) was yielded by collected carriers generated by the deeply penetrating long wavelength irradiation. Gloeckler et al. [49] found that large negative peaks in the Eg region follow from a modulation of the back contact saturation current which may be affected by deep donor defect density and increased compensation in the CdTe layer. Fig. 9 shows the results of AQE for the devices annealed at different temperatures and with various thicknesses of CuCl. Compared to the dark J-V measurement, the AQE measurement is more sensitive to the back barrier because it shows much difference in Eg region. Fig. 9(a) exhibits an efficient cell treated with CuCl under various voltage bias with no negative peaks even at voltage bias as high as 1 V. Fig. 9(b)–(c) shows the AQE results under Vbias = Voc + 100 mV. Annealing at lower and higher temperatures both show large negative Eg peaks even greater than unity in magnitude which implies the existence of a back barrier and/or a high deep donor defect density. Annealing at 265 °C presents a curve following the Eqs. (1) and (2), which reveals a quasiohmic contact and indicates that the CuCl treatment is a highly effective process for lowering the back barrier height. Although devices with various thicknesses of CuCl exhibit positive response on the whole wavelength range, a few distinctions have been found that the thickest one shows a sudden decrease in the Eg region while not the case for the thinner ones. Maybe a small reverse diode still exists at the back or a high deep defect density in CdTe. XPS is a powerful tool to characterize the surface electronic state and chemical composition so it was employed to analyze the superficial variation of chemical components of CuCl treated CdTe. High-resolution XPS spectra for Te 3d and Cu 2p are shown in Fig. 10. Nonlinear Shirley-type background subtraction was carried out, together with XPS-peak-differentiating analysis. The shifts of Te 3d5/2 of ~0.4 eV towards lower binding energy and Cu 2p3/2 of ~0.2 eV towards higher binding energy indicate a chemical reaction of Te and Cu. Besides, the peaks near 572.0 and 582.5 eV become strong after annealing due to the diffusion of Te and Cu. Fig. 10 shows the fitting curves of Te 3d5/2 before and after annealing. It can be concluded that excessive Cu on the surface mentioned in SIMS analysis reacts chemically with Te from the changes in the peak position and intensity for the strong peak at 572.0 eV. Cu7Te4 formed at the back surface after the thermal activation, in combination with the result of XRD analysis.

This work was supported by the Science and Technology Program of Sichuan Province, China (Grant Nos. 2019YFG0262, 2017GZ0414) and the Open Projects of National Energy Novel Materials Center of China (Grant No. NENMC-II-1702). The authors would like to thank Ms. Suilin Liu of Analytical and Testing Center of Sichuan University for her assistance and discussions in XPS measurements. Conflict of interest The authors declare that there is no conflict of interests regarding the publication of this paper. References [1] J. Britt, C. Ferekides, Thin-film CdS/CdTe solar cell with 15.8% efficiency, Appl. Phys. Lett. 62 (1993) 2851–2852. [2] Y.S. Tyan, E. Perez-Albuerne, Efficient thin-film CdS/CdTe solar cells, 16th Photovoltaic Specialists Conference, (1982), pp. 794–800. [3] X. Wu, J. Keane, R. Dhere, C. DeHart, D. Albin, A. Duda, T. Gessert, S. Asher, D. Levi, P. Sheldon, 16.5%-efficient CdS/CdTe polycrystalline thin-film solar cell, Proceedings of the 17th European Photovoltaic Solar Energy Conference, 2001. [4] Martin A. Green, Yoshihiro Hishikawa, Ewan D. Dunlop, Dean H. Levi, Jochen Hohl-Ebinger, Anita W.Y. Ho-Baillie, Solar cell efficiency tables (version 52), Prog. Photovolt. Res. Appl. 26 (2018) 427–436. [5] M.H. Du, First-principles study of back-contact effects on CdTe thin-film solar cells, Phys. Rev. B 80 (2009) 205322. [6] I. Dharmadasa, Review of the CdCl2 treatment used in CdS/CdTe thin film solar cell development and new evidence towards improved understanding, Coatings 4 (2014) 282–307. [7] J.D. Poplawsky, N.R. Paudel, C. Li, C.M. Parish, D. Leonard, Y. Yan, S.J. Pennycook, Direct imaging of Cl- and Cu-induced short-circuit efficiency changes in CdTe solar cells, Adv. Energy Mater. 4 (2014) 1400454. [8] F.H. Seymour, V. Kaydanov, T.R. Ohno, D. Albin, Cu and CdCl2 influence on defects detected in CdTe solar cells with admittance spectroscopy, Appl. Phys. Lett. 87 (2005) 153507. [9] B.A. Korevaar, G. Zorn, K.C. Raghavan, J.R. Cournoyer, K. Dovidenko, Cross-sectional mapping of hole concentrations as a function of copper treatment in CdTe photo-voltaic devices, Prog. Photovolt. Res. Appl. 23 (2015) 1466–1474. [10] H. Chou, A. Rohatgi, E. Thomas, S. Kamra, A. Bhat, Effects of Cu on CdTe/CdS heterojunction solar cells with Au/Cu contacts, J. Electrochem. Soc. 142 (1995) 254–259. [11] G. Luo, B. Lv, W. Li, L. Feng, J. Zhang, L. Wu, G. Zeng, Characterization of Cu1.4Te thin films for CdTe solar cells, International Journal of Photoenergy 2014 (2014) 1–5. [12] B. Späth, K. Lakus-Wollny, J. Fritsche, C.S. Ferekides, A. Klein, W. Jaegermann, Surface science studies of Cu containing back contacts for CdTe solar cells, Thin Solid Films 515 (2007) 6172–6174. [13] T. Wang, S. Du, W. Li, C. Liu, J. Zhang, L. Wu, B. Li, G. Zeng, Control of Cu doping and CdTe/Te interface modification for CdTe solar cells, Mater. Sci. Semicond. Process. 72 (2017) 46–51. [14] X. Wu, J. Zhou, A. Duda, Y. Yan, G. Teeter, S. Asher, W.K. Metzger, S. Demtsu, S.H. Wei, R. Noufi, Phase control of CuxTe film and its effects on CdS/CdTe solar cell, Thin Solid Films 515 (2007) 5798–5803. [15] Y. Yang, T. Wang, C. Liu, W. Li, J. Zhang, L. Wu, G. Zeng, W. Wang, M. Yu, Singlephase control of CuTe thin films for CdTe solar cells, Vacuum 142 (2017) 181–185. [16] J. Li, J.D. Beach, C.A. Wolden, Rapid Thermal Processing of ZnTe:Cu Contacted CdTe Solar Cells, 40th IEEE Photovoltaic Specialist Conference, (2014). [17] J.V. Li, J.N. Duenow, D. Kuciauskas, A. Kanevce, R.G. Dhere, M.R. Young, D.H. Levi, Electrical characterization of Cu composition effects in CdS/CdTe thin-film solar cells with a ZnTe: Cu back contact, IEEE Journal of Photovoltaics 3 (2013) 1095–1099. [18] S. Uličná, P.J.M. Isherwood, P.M. Kaminski, J.M. Walls, J. Li, C.A. Wolden, Development of ZnTe as a back contact material for thin film cadmium telluride solar cells, Vacuum 139 (2017) 159–163. [19] W. Xia, H. Lin, H.N. Wu, C.W. Tang, I. Irfan, C. Wang, Y. Gao, Te/Cu bi-layer: a lowresistance back contact buffer for thin film CdS/CdTe solar cells, Sol. Energy Mater. Sol. Cells 128 (2014) 411–420.

4. Conclusions A promising CuCl treatment in CdTe solar cell fabrication via vacuum evaporation to modify the CdTe/Au interface was introduced and it was beneficial for enhancing the carrier density and passivating the GBs. After the CuCl treatment, the effective carrier density for CdTe 1221

Applied Surface Science 484 (2019) 1214–1222

J. Huang, et al.

[34] K.K. Chin, T.A. Gessert, S.H. Wei, The roles of Cu impurity states in CdTe thin film solar cells, 35th IEEE Photovoltaic Specialists Conference, 2010, pp. 1915–1918. [35] K.A. Aris, K.S. Rahman, F.M.T. Enam, M.I. Bin Kamaruzzaman, I. Bin Yahya, N. Amin, An investigation on copper doping to CdTe absorber layers in CdTe thin film solar cells, International Conference on Advances in Electrical, Electronic and Systems Engineering, 2016, pp. 405–408. [36] L. Zhang, J.L. Da Silva, J. Li, Y. Yan, T.A. Gessert, S.H. Wei, Effect of copassivation of Cl and Cu on CdTe grain boundaries, Phys. Rev. Lett. 101 (2008) 155501. [37] J. Tang, Z. Huo, S. Brittman, H. Gao, P. Yang, Solution-processed core-shell nanowires for efficient photovoltaic cells, Nat. Nanotechnol. 6 (2011) 568–572. [38] A. Ashour, The physical characteristics of Cu2S/CdS thin-film solar cell, J. Optoelectron. Adv. Mater. 8 (2006) 1447–1451. [39] A. Moore, T. Song, J. Sites, Improved CdTe solar-cell performance with an evaporated Te layer before the back contact, MRS Advances 2 (2017) 3195–3201. [40] S.H. Demtsu, D.S. Albin, J.R. Sites, W.K. Metzger, A. Duda, Cu-related recombination in CdS/CdTe solar cells, Thin Solid Films 516 (2008) 2251–2254. [41] J. Li, D.R. Diercks, T.R. Ohno, C.W. Warren, M.C. Lonergan, J.D. Beach, C.A. Wolden, Controlled activation of ZnTe:Cu contacted CdTe solar cells using rapid thermal processing, Sol. Energy Mater. Sol. Cells 133 (2015) 208–215. [42] J.H. Yang, W.J. Yin, J.S. Park, W. Metzger, S.H. Wei, First-principles study of roles of Cu and Cl in polycrystalline CdTe, J. Appl. Phys. 119 (2016) 045104. [43] J. Carstensen, G. Popkirov, J. Bahr, H. Foll, CELLO: an advanced LBIC measurement technique for solar cell local characterization, Sol. Energy Mater. Sol. Cells 76 (2003) 599–611. [44] B. Moralejo, M.A. González, J. Jiménez, V. Parra, O. Martínez, J. Gutiérrez, O. Charro, LBIC and reflectance mapping of multicrystalline Si solar cells, J. Electron. Mater. 39 (2010) 663–670. [45] R.M. Geisthardt, J.R. Sites, Nonuniformity characterization of CdTe solar cells using LBIC, IEEE Journal of Photovoltaics 4 (2014) 1114–1118. [46] J.N. Duenow, J.M. Burst, D.S. Albin, M.O. Reese, S.A. Jensen, S.W. Johnston, D. Kuciauskas, S.K. Swain, T. Ablekim, K.G. Lynn, A.L. Fahrenbruch, W.K. Metzger, Relationship of open-circuit voltage to CdTe hole concentration and lifetime, IEEE Journal of Photovoltaics 6 (2016) 1641–1644. [47] S.S. Hegedus, W.N. Shafarman, Thin-film solar cells: device measurements and analysis, Prog. Photovolt. Res. Appl. 12 (2004) 155–176. [48] J.R. Sites, H. Tavakolian, R.A. Sasala, Analysis of apparent quantum efficiency, Solar Cells 29 (1990) 39–48. [49] M. Gloeckler, J.R. Sites, Apparent quantum efficiency effects in CdTe solar cells, J. Appl. Phys. 95 (2004) 4438–4445. [50] D. Bätzner, G. Agostinelli, A. Romeo, H. Zogg, A. Tiwari, Voltage dependent carrier collection in CdTe solar cells, MRS Proc. 668 (2001), https://doi.org/10.1557/proc668-h5.17.

[20] J. Türck, H.J. Nonnenmacher, P.M.L. Connor, S. Siol, B. Siepchen, J.P. Heimfarth, A. Klein, W. Jaegermann, Copper (I) Oxide (Cu2O) based back contact for p-i-n CdTe solar cells, Prog. Photovolt. Res. Appl. 24 (2016) 1229–1236. [21] J. Türck, S. Siol, T. Mayer, A. Klein, W. Jaegermann, Cu2S as ohmic back contact for CdTe solar cells, Thin Solid Films 582 (2015) 336–339. [22] S. Hu, Z. Zhu, W. Li, L. Feng, L. Wu, J. Zhang, J. Gao, Band diagrams and performance of CdTe solar cells with a Sb2Te3 back contact buffer layer, AIP Adv. 1 (2011) 042152. [23] B. Siepchen, B. Späth, C. Drost, V. Krishnakumar, C. Kraft, M. Winkler, J. König, K. Bartholomé, S. Peng, A simple Sb2Te3 back-contact process for CdTe solar cells, J. Electron. Mater. 44 (2015) 3354–3359. [24] N.R. Paudel, C. Xiao, Y. Yan, CdS/CdTe thin-film solar cells with Cu-free transition metal oxide/Au back contacts, Prog. Photovolt: Res. Appl. 23 (2015) 437–442. [25] Xue Bi, Linyu Zhu, Wei Li, Jingquan Zhang, Lili Wu, Bing Li, Guanggen Zeng, Wenwu Wang, Bi-containing reduced graphene oxide for CdTe solar cells, Sol. Energy 170 (2017) 820–827. [26] C. Gretener, J. Perrenoud, L. Kranz, C. Baechler, S. Yoon, Y.E. Romanyuk, S. Buecheler, A.N. Tiwari, Development of MoOx thin films as back contact buffer for CdTe solar cells in substrate configuration, Thin Solid Films 535 (2013) 193–197. [27] H. Lin, W. Xia, H.N. Wu, C.W. Tang, CdS/CdTe solar cells with MoOx as back contact buffers, Appl. Phys. Lett. 97 (2010) 123504. [28] K. Ohkawa, T. Karasawa, T. Mitsuyu, Molecular-beam epitaxial growth of p- and ntype ZnSe homoepitaxial layers, J. Cryst. Growth 117 (1992) 375–384. [29] Peter P. Murmu, John Kennedy, Sidharth Suman, Shen V. Chong, Jérôme Leveneur, James Storey, Sergey Rubanov, Ganpati Ramanath, Multifold improvement of thermoelectric power factor by tuning bismuth and antimony in nanostructured ntype bismuth antimony telluride thin films, Mater. Des. 163 (2019) 107549. [30] E. Antolín, C. Chen, I. Ramiro, J. Foley, Lopez Enrique, I. Artacho, Jae Sang Hwang, Alan Teran, E. Hernandez, C. Tablero, A. Martí, J.D. Phillips, A. Luque, Intermediate band to conduction band optical absorption in ZnTeO, IEEE Journal of Photovoltaics 4 (2014) 1091–1094. [31] N. Tang, Q. Hu, A. Ren, W. Li, C. Liu, J. Zhang, L. Wu, B. Li, G. Zeng, S. Hu, An approach to ZnTe:O intermediate-band photovoltaic materials, Sol. Energy 157 (2017) 707–712. [32] John Kennedy, Peter P. Murmu, Jérôme Leveneur, Grant V.M. Williams, Ryan L. Moody, Tanmay Maity, Shen V. Chong, Enhanced power factor and increased conductivity of aluminum doped zinc oxide thin films for thermoelectric applications, J. Nanosci. Nanotechnol. 18 (2018) 1384–1387. [33] D. Guo, T. Fang, A. Moore, D. Brinkman, R. Akis, D. Krasikov, I. Sankin, C. Ringhofer, D. Vasileska, Numerical simulation of copper migration in single crystal CdTe, IEEE Journal of Photovoltaics 6 (2016) 1–6.

1222