Research Article www.acsami.org
Sputtered Inx(O,S)y Buffer Layers for Cu(In,Ga)Se2 Thin-Film Solar Cells: Engineering of Band Alignment and Interface Properties Wei-Hao Ho, Chia-Hao Hsu, Shih-Yuan Wei, Chung-Hao Cai, Wei-Chih Huang, and Chih-Huang Lai* Department of Materials Science and Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan, Republic of China S Supporting Information *
ABSTRACT: We propose a simple approach to engineering the sputtered Inx(O,S)y/Cu(In,Ga)Se2 heterojunction, in terms of band alignment and interface properties. The band alignment was tailored by tuning the base pressure of the sputtering deposition to incorporate oxygen into deposited In2S3 layers (termed as Inx(O,S)y). The interface properties were improved by optimizing the air-annealing temperature on Inx(O,S)y/Cu(In,Ga)Se2 stacked layers. Increasing the base pressure raises the O/(S + O) ratio contained in deposited Inx(O,S)y films and thus widens the band gaps. This could effectively tailor the conduction band offset (ΔEC) at the Inx(O,S)y/ Cu(In,Ga)Se2 interface from a cliff (−0.25 eV) to a nearly flat band (0.07 eV) alignment. On the other hand, the extra air annealing at 235 °C did not significantly change the band alignment but did ameliorate the interface properties by reducing the Cu content at the Cu(In,Ga)Se2 surface and diminish the interface defect density induced by sputtering damages. The former might enhance the type of inversion and increase the hole barrier at the interface, preventing the detrimental recombination behavior. The latter could effectively strengthen the junction quality. Consequently, our approach substantially enhances the cell efficiency from 2.30% to 11.04%. KEYWORDS: CIGSe thin-film solar cells, Cd-free buffer layers, sputtered Inx(O,S)y buffer layers, band alignment, interface properties
1. INTRODUCTION Cu(In,Ga)Se2 (CIGSe)-based thin-film solar cells buffered with chemical-bath-deposited CdS layers are a promising heterostructure for high photovoltaic performance.1 However, the use of Cd has environmental concerns and current collection loss in the short-wavelength region due to its narrow band gap (Eg). Also, chemical bath deposition (CBD) is not compatible with the vacuum processes of CIGSe and window layers, lowering the production yield, which is undesirable from the production viewpoint.2 To address these issues, indium sulfide (In2S3), a nontoxic and tunable-band-gap III−VI semiconductor, has been fabricated by vacuum-based deposition as an alternative to the CdS buffer layer. Several groups have demonstrated the potential of In2S3 buffer layers, fabricated by thermal evaporation, atomic layer deposition (ALD), and sputtering, for CIGSe devices.3−5 Among these processes, sputtering is a well-developed technique for industrial application and possesses the advantages of large-area deposition, conformal coverage, high production yield, and long-term process stability.6 Hariskos et al. have demonstrated the sputtered In2S3-buffered CIGSe devices with the highest efficiency of 13.3% in lab scale and 11.2% in mini-module scale, respectively.7 While In2S3 buffer layers are applied to CIGSe devices, one of the key challenges is to engineer the buffer/absorber heterojunction, in terms of interface properties and band © XXXX American Chemical Society
alignment, which directly determines the carrier transport, recombination behavior, and thereby photovoltaic performance. For high-quality interface properties, a well-accepted way is to control the interdiffusion behavior at the buffer/CIGSe interface by optimizing the deposition and/or postdeposition annealing temperatures.8−10 The interdiffusion yields a Cudepleted CIGSe surface owing to the Cu diffusion from the surface into the buffer layer (or into CIGSe bulk), reducing the interface recombination and enhancing device performance.8−11 Thus, the post air-annealing treatment is considered as a necessary process for the In-based buffer layers.6 In contrast to the interface properties, tailoring the band alignment to improve the cell performance is relatively unexplored. The conduction band offsets (ΔEC) of the Inbased buffer/CIGSe interface were reported in a wide range from −0.45 to 0.5 eV8,11−13 and strongly depend on the growth methods of buffer layers. The O-doped In2S3 thin films have been fabricated by using various methods, including thermal evaporation, ALD, and CBD processes.14−17 Their results suggest that doping O into the In2S3 layer can enlarge the band gap and modify the conduction band minimum, which is beneficial to the band alignment with CIGSe absorbers;16,17 however, these studies only focus on the growth mechanism Received: February 8, 2017 Accepted: May 4, 2017 Published: May 4, 2017 A
DOI: 10.1021/acsami.7b01862 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 1. Schematic diagram of the fabrication of Inx(O,S)y buffer layers.
and/or fundamental properties of deposited films without reporting device performance. Although the presence of O in the In2S3 buffer layers has been observed in the CIGSe devices without intentional oxygen addition,18−21 the experimental connection between the device performance and the band alignment controlled by O doping has not yet been established. Furthermore, the engineering of the In2S3/CIGSe heterointerface may become complicated when the In2S3 layers are deposited by sputtering. The sputtering process can introduce a large number of defects at the interface and/or near the absorber surface due to the bombardment of high energy particles during deposition,22−25 known as sputtering damage. Sputtering damages have commonly been reported in the sputtered ZnMgO and plasma-enhanced ALD-deposited In2(S,O)3 buffer layers as well as the i-ZnO window layers;18,22−25 however, it is rarely discussed in the sputtered In2S3 buffer layer. In this study, we proposed a simple approach to engineering the sputtered In2S3/CIGSe heterojunction, in terms of band alignment and interface properties. The band alignment was tailored by tuning the base pressure of the sputtering deposition to incorporate oxygen into deposited In2S3 layers (hereafter termed the “Inx(O,S)y”). The interface properties were improved by optimizing the air-annealing temperature on Inx(O,S)y/CIGSe stacked layers. To our best knowledge, this is the first study about the Inx(O,S)y buffer layer fabricated by sputtering deposition with oxygen doping. Our approach substantially enhances the energy conversion efficiency of sputtered Inx(O,S)y/CIGSe devices from 2.30% to 11.04% (best). The influences of band alignment, interface properties, and sputtering damage on device performance were revealed. The corresponding results can also be applied to other sputtered Cd-free buffer layers that suffer from the poor interface properties.
CIGSe absorbers by pulsed-DC sputtering of a single indium sulfide target (In = 45.4 at. %, S = 55.6 at. %) with the power density of 1 W/ cm2. The sputtering deposition was implemented in an Ar atmosphere without extra reactive gas at room temperature. To incorporate O into deposited films, the base pressure was controlled at low base pressure (LBP) of 5.33 × 10−4 Pa and high base pressure (HBP) of 1.33 × 10−2 Pa, respectively. The working pressure was kept at 2.67 Pa. The deposition rate at HBP (LBP) condition was 23 nm/min (19.5 nm/ min). The deposition time was fixed at 2.6 min, resulting in film thickness in the range of 50 nm (LBP) ∼ 60 nm (HBP). During the sputtering deposition, optical emission spectroscopy (OES) was used to characterize the emission spectrum of each element, especially the oxygen emission spectrum. After deposition, the air-annealing treatment, carried out by a hot plate, is applied on the Inx(O,S)y/ CIGSe/Mo/SLG stacked layers rather than complete devices to avoid the interdiffusion at the window/buffer interface. The air annealing was done for 10 min with varied temperature. Subsequently, CIGSe solar cells were completed by the deposition of the i-ZnO (50 nm)/ ZnO:Al (300 nm) double window layer by using radio frequency (RF) sputtering and the evaporation of Al grid contacts. The cell area is 0.3 cm2. No annealing of the complete device is applied. 2.2. Characterization. The current density−voltage (J−V) characteristics were measured by a Keithley 4200-SCS system at 25 °C under AM 1.5 illumination with the light density of 100 mW/cm2, which was calibrated by a standard Si solar cell certified by Fraunhofer ISE, Germany. The compositional profiles of the Inx(O,S)y/CIGSe interface were characterized by X-ray photoelectron spectroscopy (XPS) using ESCA PHI 1600. The compositional profiling was done by recording the successive XPS spectra acquired at the time interval of 0.3 min by Ar sputtering etching with the energy of 2 keV. The total sputtering etching time of 8 min is required for the depth across the Inx(O,S)y/CIGSe interface. The valence band structure was measured by ultraviolet photoelectron spectroscopy (UPS) with the He I excitation source (21.2 eV) using the same instrument (ESCA PHI 1600). The morphologies were performed by a scanning electron microscope using Hitachi SU8010. For characterization of optical properties, Inx(O,S)y films were deposited on a borosilicate glass substrate rather than SLG to avoid Na diffusion. Transmittance and reflectance spectra were measured by a Hitachi U4100 spectrophotometer. The external quantum efficiency (EQE) of the CIGSe device was measured using a lock-in amplifier, chopped light, and monochromator system without background illumination (ENLI, QE-R). The light intensity was calibrated by Si and Ge photodiodes. Time-resolved photoluminescence (TR-PL) was implemented at room temperature using a picosecond pulsed diode laser (638 nm, 70 ps, 1 mW) with a repetition frequency of 10 MHz.
2. EXPERIMENTAL SECTION 2.1. Fabrication of Inx(O,S)y/CIGSe Solar Cells. 1.6 μm-thick CIGSe absorbers were fabricated on Mo-coated soda-lime glass (SLG) substrates by selenization of Cu−In−Ga metallic precursors, which were sputtered from Cu−Ga and In targets, respectively. The selenization process was implemented at 550 °C with elemental Se vapor. The Cu/(In + Ga) and Ga/(In + Ga) ratios of selenized CIGSe films measured by X-ray fluorescence (XRF) were 0.77 ± 0.02 and 0.36 ± 0.02, respectively. Figure 1 shows the schematic diagram of fabrication procedures for Inx(O,S)y buffer layers. The Inx(O,S)y layers were deposited onto
3. RESULTS AND DISCUSSION Inx(O,S)y buffer layers were fabricated by the sputtering process with varied base pressure and air-annealing temperature. These B
DOI: 10.1021/acsami.7b01862 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces parameters and the corresponding abbreviations used in this article are listed in Table 1. Table 1. Sample Names and Their Corresponding Process Parameters sample names LBP HBP HBP, air-ann 235 °C HBP, air-ann 280 °C
base pressure of In2S3 deposition (Pa)
air-annealing temperature (°C)
5.33 × 10−4 1.33 × 10−2 1.33 × 10−2
w/o w/o 235 °C
1.33 × 10−2
280 °C
3.1. Device Performances of Sputtered Inx(O,S)y/ CIGSe Solar Cells. Figure 2a−d shows the statistical comparisons of photovoltaic parameters for CIGSe devices with sputtered Inx(O,S)y buffer layers fabricated by various conditions. Clearly, the device performance shows the distinct dependence on both base pressure and air-annealing temperature. CIGSe devices with Inx(O,S)y buffer layers deposited at LBP condition show the lowest efficiency, which cannot be notably increased by the extra air-annealing treatment. As base pressure is raised from LBP to HBP, open-circuit voltage (VOC) and fill factor (FF) are significantly enhanced, improving the average efficiency from 2.30% to 6.50%. In addition, when the extra air annealing at 235 °C is applied to the HBP samples, all photovoltaic parameters including VOC, FF, and short-circuit current density (JSC) are further enhanced, increasing the average efficiency up to 10.93%. However, all parameters are deteriorated, as the air-annealing temperature reaches 280 °C. The corresponding J−V curves are also shown in Figure 3a. These results suggest that tuning the base pressure and air-
Figure 3. Typical (a) J−V characteristics and (b) EQE spectra of the sputtered Inx(O,S)y/CIGSe solar cells with varied base pressure and air-annealing temperature.
annealing temperature for the sputtered Inx(O,S)y/CIGSe device is crucial, which substantially improves the average efficiency from 2.30% to 10.93%. The EQE spectra in Figure 3b further reveal that optimizing the fabrication of Inx(O,S)y buffer layers ameliorates the carrier transport across the buffer/absorber interface, as evident from
Figure 2. Statistical comparison of photovoltaic parameters, including the (a) VOC, (b) JSC, (c) FF, and (d) efficiency, for sputtered Inx(O,S)y/CIGSe devices with various base pressures and air-annealing temperatures. C
DOI: 10.1021/acsami.7b01862 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces the enhanced photocurrent collection in both short (400 ∼ 570 nm) and long (570 ∼1200 nm) wavelength regions. The carrier transport in the CIGSe device is related to their energy band alignment and interface properties.6,11,26 Thus, in the following discussion, the morphological and optical properties of the Inx(O,S)y films were first studied. Then, the properties of the Inx(O,S)y/CIGSe heterojunction, including the elemental diffusion, the TR-PL lifetime, and energy band alignment, were investigated. Combined with these results, the influences of base pressure and air-annealing temperature on the device performance can be revealed. 3.2. Morphological and Optical Characterizations of Sputtered Inx(O,S)y Films. Figure 4 shows typical scanning
Figure 5. (a) Optical transmittance spectra and (b) band gaps of the sputtered Inx(O,S)y films processed with varied base pressure and airannealing temperature.
The optical band gap was extracted by linear extrapolation of the (αhν)2 versus photon energy (hν) plot to energy axis. The obtained band gap values are summarized in Table 2. As the base pressure is raised from LBP to HBP, the transmittance is increased in all wavelength regions, and the absorption edge shifted toward short wavelength, suggesting a widened band gap from 2.18 to 2.35 eV. The band gap widening of 0.17 eV is physically meaningful for band alignment, which will be discussed later. In contrast, as the extra air annealing at either 235 or 280 °C is applied, the transmittance is nearly unchanged, yielding a very small band gap variation of ±0.02 eV with respect to that of HBP films (without air annealing). These observations indicate that the band gap of Inx(O,S)y films is mainly affected by base pressure rather than airannealing temperature. 3.3. Interface Properties of Sputtered Stacked Inx(O,S)y/CIGSe. The buffer/absorber interface properties could be investigated from the viewpoints of elemental distributions and defect density at the interface, respectively. The elemental distributions were carried out by the XPS measurement. The XPS compositional depth profiles of Inx(O,S)y/CIGSe stacked layers were acquired in Figure S2. For easier comparison, the profiles of O/(S + O) ratios and Cu contents are shown in Figure 6a and b, respectively. The Se profile is also included to identify the locations of the Inx(O,S)y layer and CIGSe surface. As shown in Figure 6a, within the Inx(O,S)y layer, the oxygen content (O/(S + O)) is substantially increased from
