Optimization of post-deposition annealing in Cu2ZnSnS4 thin film solar cells and its impact on device performance

Solar Energy Materials and Solar Cells 170 (2017) 287–294

Contents lists available at ScienceDirect

Solar Energy Materials and Solar Cells journal homepage: www.elsevier.com/locate/solmat

Optimization of post-deposition annealing in Cu2ZnSnS4 thin film solar cells and its impact on device performance

MARK

⁎

M.G. Sousaa, , A.F. da Cunhaa, J.P. Teixeiraa, J.P. Leitãoa, G. Otero-Iruruetab, M.K. Singhb a

I3N and Department of Physics, University of Aveiro, Campus Universitário de Santiago, 3810-193 Aveiro, Portugal Center for Mechanical Technology and Automation-TEMA, Department of Mechanical Engineering, University of Aveiro, Campus Universitário de Santiago, 3810-193 Aveiro, Portugal b

A R T I C L E I N F O

A B S T R A C T

Keywords: Kesterites Cu2ZnSnS4 Thin film solar cells Post-deposition annealing Device performance

In this work we present an optimization of the post-deposition annealing, in Cu2ZnSnS4 (CZTS) thin film solar cells, applied at different stages of the solar cell preparation, namely, bare CZTS absorber, CZTS/CdS heterojunction and CZTS/CdS/i-ZnO/ITO complete solar cell. We performed current-density measurements, scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), X-ray diffraction (XRD), Raman scattering, photoluminescence (PL) and X-ray photoelectron spectroscopy (XPS) studies to enlighten the mechanisms by which solar cells performance improvement comes about. As a result, we concluded that the optimum postdeposition annealing for CZTS is at 300 °C for 15 min and at atmospheric pressure. The highest efficiency gain was obtained when the absorber layer composition is close to the ideal one and when a single annealing step is performed on complete solar cells, where, we obtained efficiency improvements from below 1% to over 6.6%. Despite the observed improvement in device performance for annealing at intermediate stages it is, however, less pronounced than for full cell annealing. In this process we demonstrate very substantial cell performance improvements. XRD results show a shift of all Bragg peaks to lower diffraction angle values, after post-deposition annealing. Also, the intensity of the peaks decreases and their full width at half maximum increases. PL measurements show that, post-deposition annealing, leads to a clear reduction of the non-radiative recombination channels and that the electronic structure is dominated by fluctuating potentials. XPS measurements reveal an interdiffusion of Cu, Zn and possibly Cd across the interface between buffer and CZTS absorber layers as the source of the significant observed cell performance enhancement.

1. Introduction Cu2ZnSnS4 (CZTS) has been object of intense research due to its nearly ideal band gap of about 1.5 eV, a high absorption coefficient and containing only earth abundant and non-toxic elements [1,2]. The efficiency of CZTS thin film solar cells is known to depend strongly on the concentration of Cu on Zn antisites, Cu vacancies and defect clusters in the kesterite-type structure. However, the relationship of order-disorder, point defects and their influence on device performance is not yet clearly understood [3]. Calculations according to the Shockley-Queisser photon balance have estimated the theoretical conversion efficiency limit of single-junction CZTS solar cells to be as high as 32.2% [4]. Significant developments have been made on CZTS based thin film photovoltaic (PV) solar cells in the past few years, reporting solar cells with 9.2% efficiency [5]. However, CZTS PV technology requires extensive research to become marketable in the near future. So far, high efficiency CZTS based solar cells were found to have slightly Cu-poor

⁎

and Zn-rich composition [6], which corresponds to a composition ratio of 0.8–0.9 for Cu/(Zn+Sn) and 1.1–1.2 for Zn/Sn, irrespective of the deposition technique or absorber preparation method. During the absorber layer fabrication process, secondary phases may form and depending on their fraction, have a big impact on the characteristics of the cell. Nagoya et al. [7] and Maeda et al. [8] have theoretically predicted ZnS to be the predominant secondary phase under the Cu-poor and Znrich growth condition with CuZn antisite being the most stable defect in the stability region of CZTS [9]. ZnS has a wider bandgap and is usually less conductive. Therefore is not considered to be responsible for reduced open-circuit voltage (VOC) or reduced shunt resistance, but can lead to high series resistance and reduced charge carrier collection efficiency depending on its location in the solar cell [10,11]. Other secondary phases such as Cu-Sn sulfides, SnSx or CuSx are considered to be more detrimental because of their lower bandgap and high conductivity, which can significantly reduce the open-circuit voltage and decrease the shunt resistance leading to much inferior photovoltaic

Corresponding author. E-mail address: [email protected] (M.G. Sousa).

http://dx.doi.org/10.1016/j.solmat.2017.05.065 Received 7 September 2016; Received in revised form 18 March 2017; Accepted 27 May 2017 0927-0248/ © 2017 Elsevier B.V. All rights reserved.

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

80v Fourier transform infrared (FTIR) spectrometer, equipped with an InGaAs photodetector. The samples were inserted in a helium gas flow cryostat and the measurements were performed at 70 K. The 514.5 nm line of an Ar+ ion laser was used as the excitation source and the excitation power was measured at the front of the spectrometer entrance window. The PL spectra are presented as measured. For these PL measurements, a CdS thin film reference sample was deposited by CBD on SLG, using the same conditions as those for the deposition of the buffer layer on the solar cells. X-ray photoelectron spectroscopy (XPS) measurements were performed in an Ultra High Vacuum (UHV) system with a base pressure of 2 × 10−10 mbar. The system is equipped with a hemispherical electron energy analyser (SPECS Phoibos 150), a delayline detector and a monochromatic AlKα (hν ~ 1486.74 eV) X-ray source. High resolution spectra were recorded at normal emission takeoff angle and with a pass-energy of 20 eV, which provides an overall instrumental peak broadening of 0.5 eV. Ar+ ions (1.5 kV), with an incidence angle of 45°, were used for XPS depth profiling. The solar cell performance was characterized through current-voltage (J-V) measurements under simulated standard test conditions in which the light source consisted of a tungsten-halogen lamp combined with an infrared filter for spectrum conditioning.

performance of the cell [12]. New concepts for further increasing the performance and reducing the costs by, for example, improved solar cell architectures and processing, are needed. In this work, development, optimization and understanding the effects of post-deposition annealing in Cu2ZnSnS4 thin film solar cells and its impact on device performance are presented. The main goal is to improve the efficiency of the Cu2ZnSnS4 thin film solar cells produced by rapid thermal processing (RTP) [13], in an atmosphere containing H2S, of multi-period precursor layers deposited on Mo coated soda-lime glass (SLG) by RF magnetron sputtering. Two types of precursors were prepared with the following structures: (ZnS/SnS2/Cu) and (ZnS/SnS2/Cu)(8_periods). For each type of precursors two sets of absorbers have been prepared and subjected to post-deposition annealing at various stages of the cell preparation. The effects of this post-deposition annealing were studied.

2. Experimental In this work, the method employed for the growth of Cu2ZnSnS4 thin films consisted on the annealing of RF-magnetron sputtered precursors deposited on Mo coated SLG. Two types of precursors were prepared with the following structures: (ZnS/SnS2/Cu) and (ZnS/SnS2/ Cu)(8_periods). For each type, two sets of absorbers have been prepared, one with excess Zn and another with the ideal composition [6]. With the resulting absorbers, sets of cells have been prepared and subjected to the post-deposition annealing at various stages of the cell preparation, namely, bare CZTS absorber, CZTS/CdS heterojunction and CZTS/ CdS/i-ZnO/ITO complete solar cells. The resulting final cell structure was SLG/Mo/absorber/CdS/i-ZnO/ITO. A DC-magnetron sputtered single layer Mo film with a thickness around 0.5 µm was used as standard back contact. The precursor stacks were annealed in a rapid thermal processing (RTP) furnace with an atmosphere of 95% N2 + 5% H2S at a pressure of 1 atm and at 520 °C for 30 mins. The p-type CZTS absorber layer had a thickness ranging from 1 to 1.5 µm. Before the buffer layer deposition, potassium cyanide (KCN) etching was employed to remove possible unwanted CuS secondary phases formed on the absorber surface. A thin n-type CdS layer of 70 nm thick was deposited on the p-type CZTS film by chemical bath deposition (CBD). Subsequently, the device structure was completed by depositing a 60 nm highly resistive intrinsic ZnO (i-ZnO) layer followed by the deposition of an indium tin oxide (ITO) layer with a thickness around 0.3 µm. The sheet resistance of the window layer control sample was ~ 23 Ω. The optimum post-deposition annealing was carried out on a hot plate at atmospheric pressure, in an atmosphere of N2 and at 300 °C for 15 mins. The top surface, cross-sectional morphology and average composition of the absorber layers were analysed by SEM/EDS using a TESCAN Vega3 SBH SEM microscope, operated at an acceleration voltage of 15.0 kV for image acquisition and 25 kV for chemical analysis. The crystalline structure was analysed by X-ray diffraction (XRD) with a Philips PW 3710 system, in the Bragg-Brentano configuration (θ-2θ), using the CuKα line (λ ~ 1.54060 Å) and the generator settings were 50 mA and 40 kV. A LabRam Horiba, HR800UV spectrometer, equipped with a solid state laser oscillating at 532 nm was used for Raman scattering measurements. Raman spectra were calibrated using a single crystal Si reference sample, before the actual measurements, by setting the position of the dominant Si peak at 520 cm−1. The photoluminescence (PL) measurements were performed with a Bruker Vertex

3. Results and discussion The work reported in this paper consisted on a systematic study of the effect of the post-deposition annealing on the performance of CZTS based thin film solar cells, applied at different stages of cell preparation. For both types of precursors, (ZnS/SnS2/Cu) and (ZnS/SnS2/ Cu)(8_periods), the maximum temperature and the dwelling time at maximum temperature for the post-deposition annealing were varied from 200 to 400 °C and from 15 to 20 mins, respectively. The application of the post-deposition annealing to the full cells based on the first type of precursors (ZnS/SnS2/Cu) has shown that annealing at 200 °C leads to an improvement of the solar cell efficiency by a factor of 2 while a post-deposition annealing at 300 °C leads to an efficiency improvement by a factor of 11. The changes in the solar cell performance parameters are shown in Table 1. Annealing at temperatures lower than 300 °C also improves cell performance, however, in a less pronounced way, while, annealing at 400 °C is detrimental for the solar cell and leads to a clear degradation of the performance possibly due to CZTS decomposition and deterioration of the CdS layer. For the annealing time, in the interval considered, no significant changes have been observed. The results have shown that the optimum post-deposition annealing, for the CZTS case, should be performed at 300 °C for 15 mins. In an effort to improve beyond the best results and realizing that the open circuit voltage (VOC) was still too low we decided to test precursors with 8 periods of (ZnS/SnS2/Cu) in order to reduce undesired phase segregation. For this type of precursors the effect of Zn content was also studied. While the solar cell performance results, with this new type of precursors, did not show significant gains before post-deposition annealing, after the annealing major gains were obtained. The highest efficiency gain was obtained when the absorber layer composition is close to the ideal one and when a single annealing step is performed on complete solar cells (Table 2). However, in all the cases, the post-deposition annealing on full cells promotes an improvement of all device parameters, such as, open circuit voltage (VOC), short circuit current

Table 1 Comparison of performance, before and after the post-deposition annealing, at 200 °C and 300 °C for 15 mins, on complete solar cells (CZTS/CdS/i-ZnO/ITO) with ideal composition absorbers. The ratios of all device parameters corresponds to CZTS/CdS/i-ZnO/ITO 200 °C or 300 °C annealed divided by CZTS/CdS/i-ZnO/ITO not annealed. Complete cells CZTS/CdS/i-ZnO/ITO

Not annealed 200 °C annealed 300 °C annealed

VOC (mV)

VOC ratio

ISC (mA cm−2)

ISC ratio

FF (%)

FF ratio

Eff. (%)

Eff. ratio

174.8 182.6 399.2

1.04

4.6 6.4 12.8

1.4

26.3 38.6 45.8

1.5

0.21 0.45 2.35

2.1

2.3

288

2.8

1.7

11.2

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

Table 2 Comparison of performance, before and after the post-deposition annealing, of solar cells obtained from annealed bare CZTS absorber, annealed CZTS/CdS heterojunction and CZTS/CdS/ i-ZnO/ITO complete cell for a) CZTS absorber layers with excess Zn and b) CZTS absorber layers with ideal composition. The ratios of all device parameters corresponds to CZTS/CdS/iZnO/ITO 200 °C or 300 °C annealed divided by CZTS/CdS/i-ZnO/ITO not annealed. a) Absorber layer with excess Zn

Cu/Zn

Cu/Sn

Zn/Sn

1.04

1.81

1.78

Incomplete cells

Complete cells

VOC (mV)

VOC ratio

ISC (mA cm−2)

ISC ratio

FF (%)

FF ratio

Eff. (%)

Eff. ratio

(1)

Not annealed 300 °C annealed Not annealed 300 °C annealed Not annealed 300 °C annealed

176.4 490.6 342.1 527.1 150.2 462.9

2.8

3.8 11.4 8.1 16.1 1.7 10.3

2.9

23.5 56.4 28.0 48.6 21.7 50.8

2.4

0.16 3.16 0.78 4.12 0.06 2.43

19.8

Annealed bare CZTS absorber (2) Annealed CZTS/CdS heterojunction (3) CZTS/CdS/i-ZnO/ ITO

1.5 3.1

1.9 6.1

b) Absorber layer with ideal composition

1.7 2.3

5.3 40.5

Cu/Zn

Cu/Sn

Zn/Sn

1.47

1.74

1.18

Incomplete cells

Complete cells

VOC (mV)

VOC ratio

ISC (mA cm−2)

ISC ratio

FF (%)

FF ratio

Eff. (%)

Eff. ratio

(1)

Not annealed 300 °C annealed Not annealed 300 ºC annealed Not annealed 300 ºC annealed

161.9 455.4 414.0 520.8 161.1 517.7

2.8

4.5 18.1 15.4 18.1 4.1 33.1

3.5

22.7 46.9 25.4 38.9 21.5 38.6

2.1

0.17 3.44 1.62 3.67 0.14 6.63

20.2

Annealed bare CZTS absorber Annealed CZTS/CdS heterojunction (3) CZTS/CdS/i-ZnO/ ITO (2)

1.3 3.2

1.2 8.1

1.5 1.8

2.3 47.4

Note: (1) Annealed bare CZTS absorber: Mo/CZTS/post-annealing/CdS/i-ZnO/ITO/(study J-V curve)/post-annealing/(study J-V curve) (Double annealing step) (2) Annealed CZTS/CdS heterojunction: Mo/CZTS/CdS/post-annealing/i-ZnO/ITO/(study J-V curve)/post-annealing/(study J-V curve) (Double annealing step) (3) CZTS/CdS/i-ZnO/ITO: Mo/CZTS/CdS/i-ZnO/ITO/post-annealing /(study J-V curve) (Single annealing step)

(ISC), fill factor (FF) and, consequently, in efficiency (Eff.). Considering the Zn-rich and ideal composition sets of absorbers, the most pronounced improvement in efficiency was obtained for the cells named bare CZTS absorber and CZTS/CdS/i-ZnO/ITO, with improvements in the order of 20 times and 40 times, respectively. In both cases, the Zn-rich and the one with ideal composition, the ratio between the parameter VOC_300 °C and VOC_not_annealed shows an increase by a factor of 3, when the annealing is carried out on a full cell. The FF ratios show an increase by a factor of 2. The biggest gain is that associated with increased ISC, changing by a factor of 3 and 6, for the absorber layer with excess Zn and, by a factor of 3 and 8, for the absorber layer with ideal composition. In the case of the absorbers with excess Zn, the efficiency gain was by a factor of 40 for the sample called CZTS/CdS/i-ZnO/ITO, although the best efficiency has been obtained for the CZTS/CdS heterojunction, with 4.12%. This is explained by the higher values of VOC (527.1 mV) and ISC (16.1 mA cm−2) parameters. When the annealing is performed on incomplete cells and without any subsequent annealing, in both sets of absorbers, the post-deposition annealing has a most significant impact, in cell efficiency, in the CZTS/ CdS heterojunction case in comparison with bare CZTS absorber. Improvements are also observed on cells named annealed bare CZTS absorber when compared with CZTS/CdS/i-ZnO/ITO. This indicates that the beneficial effects of post-deposition annealing are due to changes in the CZTS absorber itself and due to improvements of the CdS layer. Despite the observed improvement in device performance for annealing at intermediate stages of cell preparation (incomplete cells) it is, however, less pronounced than for full cell annealing. Fig. 1 shows the illuminated current density-voltage (J-V) curves for exactly the same cell before and after the optimum post-deposition annealing. Comparing the two curves, one can clearly see that before annealing the shape of the curve shows a double diode behaviour. Both curves start approximately at similar bias voltage values of −0.6 V but for the curves before post-deposition annealing the photogenerated

current increases very fast as the bias voltage is increased in the forward direction. The post-deposition annealing at 300 °C clearly removes the double diode behaviour resulting in the enhancement of all cell performance parameters. In the best case (Fig. 1(f)) the annealing step led to an improvement of the device efficiencies from below 1–6.6%. Given that the CZTS absorber is grown at temperatures in the order of 520 °C it is expected that the interface between CZTS and Mo should not be substantially affected by the post-deposition annealing. However, the interface between absorber and CdS buffer layer is produced at low temperature and therefore it is likely that it is affected by the annealing at 300 °C. Raman scattering measurements confirm that substantial changes occur in the latter interface, as shown later in Fig. 3(b)), where both CdS and CZTS main peaks change considerably in relative intensity. In order to clarify the physical origins of the beneficial effects of the post-deposition annealing on the device performance, a careful characterization was performed. SEM images representative of the samples, for which the best cell results were obtained, are shown in Fig. 2. Fig. 2(a) shows a compact top surface with a relatively small grain structure uniformly distributed, crack free and few pinholes resulting from the KCN etching. Through careful point EDS analysis it was possible to see that the bright particles on the surface correspond to ZnS. Fig. 2(b) shows the cross-section of a solar cell where it is possible to confirm that the CZTS absorber was compact and with few voids near the Mo back contact. Two representative XRD diffractograms are shown in Fig. 3(a). It shows the XRD results for the samples named CZTS/CdS heterojunction before and after the post-deposition annealing. Generally, both diffractograms point to CZTS being the dominant phase present as supported by the presence of several characteristic low intensity CZTS diffraction peaks, which also suggest good crystallinity of the samples. This, however, does not exclude the possibility of there being additional phases such as ZnS and Cu2SnS3 (CTS), since, they all have similar 289

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

Fig. 1. Illuminated current density-voltage (J-V) curves for solar cells obtained with post-deposition annealing steps at bare CZTS absorber (a, b), CZTS/CdS heterojunction (c, d) and CZTS/CdS/i-ZnO/ITO cell structure (e, f). Curves (a), (c) and (e) correspond to absorbers with excess zinc and curves (b), (d) and (f) correspond to absorbers with ideal composition.

post-deposition annealing carried out on full devices. Interesting changes in the relative intensity of the peaks occur. Besides the typical features of the CZTS thin films the Raman spectra also present a peak at 303 cm−1 assigned to the CdS buffer layer. This assignment is based on the fact that our previous studies show that bare CZTS layers do not show a peak in this region with such prominence [14]. As such, even though there could be a small contribution from residual CTS phases the dominant contribution must be from the CdS layer. Furthermore, this peak appears every time we perform Raman scattering measurements on complete CZTS cells [13] and no other cell components have Raman modes in this region. It shows that the intensity of the peak

structure and the diffraction peaks often overlap. Comparing the XRD diffractograms before and after post-deposition annealing no additional peaks are observed. A shift of all the Bragg peaks to lower diffraction angle values is observed after post-deposition annealing. This indicates an increase in the interatomic plane spacing, possibly due to structural relaxation, since the angular position of the peaks decreases to the relaxed powder diffraction values. Also, the intensity of the peaks decreases for annealed CZTS/CdS samples and their full width at half maximum (FWHM) increases suggesting a slight degradation of the CZTS crystal structure, as shown in Fig. 3(c) and Table 3(a). Fig. 3(b) shows representative Raman spectra of CZTS solar cells before and after

Fig. 2. SEM micrographs of: a) top surface of the CZTS absorber after KCN etching; b) Cross-section of the CZTS/CdS/i-ZnO/ITO solar cells after post-deposition annealing.

290

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

Fig. 3. a) Representative XRD diffractograms of CZTS/CdS samples to study the effects of post-deposition annealing on CZTS/CdS heterojunction; b) Representative Raman spectra of CZTS solar cells before and after post-deposition annealing.

assigned to CdS at 303 cm−1 undergoes a big increase in intensity after post-deposition annealing at 300 °C while its FWHM decreases, as shown in Fig. 3d) and Table 3b). This is interpreted as the result of an improvement in the crystallinity of this layer. At the same time, also shows a large decrease in relative intensity of the most intense peak for CZTS at 337.2 cm−1. This may be the result of a degradation of Cu/Zn ordering. This decrease in relative intensities of the main CZTS Raman peaks is also observed for samples with decreasing relative Cu content.

Zn and Sn enriched surfaces show the same Raman behaviour [3]. CZTS is known to have a strong Raman peak at 338 cm−1, which is clearly present, and a less intense one to the right of the main peak at 374 cm−1. Both samples showed the structure described above, containing peaks at 337.2 cm−1 and 373 cm−1, assigned to CZTS. A broad shoulder at 254 cm−1 is also assigned to CZTS. The peaks at 601 cm−1 and 668 cm−1 are believed to be second order peaks of 303 cm−1 and 337.2 cm−1 main ones.

Table 3 a) XRD peak positions and full width at half maximum (FWHM) of CZTS/CdS heterojunctions; b) Peak positions for the main Raman CZTS and CdS peaks and full width at half maximum (FWHM) of CZTS solar cells. a)

Before annealing

Diffraction angle (deg)

FWHM (deg)

28.54 33.14 40.46 47.42 56.22

0.13 0.16 0.35 0.09 0.77

After annealing

Diffraction angle (deg)

FWHM (deg)

28.46 33 40.41 47.34 56.14

0.15 0.19 0.36 0.24 0.77

b)

Before annealing After annealing

Raman shift (CdS) (cm−1)

FWHM (cm−1)

Raman shift (CZTS) (cm−1)

FWHM (cm−1)

303 303

20.4 11.7

337.2 337.2

5.34 5.22

291

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

Fig. 4. a) PL spectra measured at 70 K, under an excitation power of 140 mW, of a CZTS based solar cell before and after a post-deposition annealing. In the inset, a comparison between the PL spectra of the sample before the post-deposition annealing and of a thin CdS layer deposited on SLG, is also shown. A multiplicative factor of 0.55 for the intensity was considered for comparison purposes with the PL spectrum measured before the treatment. The small peak at 1.164 meV is related to the SLG. b) Dependence on the excitation power of the peak energy of the luminescence from the solar cell measured after the post-deposition annealing.

In order to acquire a finer understanding of the effect of post-deposition annealing on the CZTS solar cells PL measurements have also been performed. Fig. 4(a) shows the PL spectra measured at 70 K, under an excitation power of 140 mW, of the solar cell before and after postdeposition annealing. Before the post-deposition annealing, a broad emission with a low signal-to-noise ratio is observed. The energy range and shape of this emission is very similar to the one measured from the CdS thin layer, deposited directly on SLG. Similar experimental parameters were used and the intensity was multiplied by a factor of 0.55 to allow an easier comparison of the two PL spectra. It shows that, before the post-deposition annealing, only radiative transitions involving deep defects in CdS were observed. We must note that the excitation wavelength is close to the bandgap of the CdS buffer layer [15] but allows the optical excitation of the absorber layer as shown previously [16–18]. Actually, a thickness of several tens of nm of the absorber layer should be probed in the experiment [19]. After the post-deposition annealing, the luminescence from the solar cell clearly changed as shown in Fig. 4a). A broad and asymmetric band in the range of 0.8–1.34 eV, with maximum of intensity at ~ 1.16 eV, was observed. Additionally, a higher signal-to-noise ratio was obtained, despite the caution needed in the comparison of the luminescence from different measurements due to the unavoidable differences in the optical alignments. The shape of the band after the post-deposition annealing is identical to the ones assigned in literature to CZTS [16,17,20–22]. The peak of this band shows a deviation of 0.34 eV from the accepted CZTS bandgap energy (~ 1.5 eV [23,24]), which is higher than the common values available in the literature. This higher redshift of the luminescence may be due to the higher temperature (70 K) used for our current measurements [17,20]. The dependence of the peak energy of the emission on the excitation power allows the clarification of the nature of the radiative transitions [16]. As such, a study of the dependence on the excitation power of the peak position was performed, in the 2–140 mW range, for the solar cell after the post-deposition annealing (Fig. 4(b)). All measured PL spectra were fitted with Gaussian components according to a model that requires the lowest number (4 in this study) of Gaussian components. The increase of the excitation power up to 100 mW, results in a blueshift of ~ 18 meV/decade, whereas for higher values the slope approaches zero. It has been shown in the literature [17,20,25], that an asymmetric band is characteristic of radiative transitions occurring in semiconductors with high densities of ionized defects. The interaction of these randomly distributed defects results in electrostatics fluctuating potentials along the film, resulting in the appearance of tails states in the bandgap. In the scope of this model, high blueshift values of several meV/decade are expected. The value (~ 18 meV/decade) estimated in this work, confirms the existence of fluctuating potentials in the absorber layer after the post-deposition

annealing. A possible residual contribution from the CdS buffer layer to the measured luminescence after the post-deposition annealing of the solar cell cannot be excluded. However, that possibility does not affect the influence of fluctuating potentials in the observed luminescence from the CZTS layer. The drastic change of the PL with the post-deposition annealing suggests a huge reduction in the density of defects on the absorber layer. Before the post-deposition annealing, the density of non-radiative channels should be very high because no luminescence related to the CZTS layer was observed. Thus, upon the optical excitation, the generated charge carriers are captured by different defects that promote a non-radiative de-excitation. With the post-deposition annealing, those channels are suppressed allowing the establishment of radiative channels in the absorber layer. To further investigate the mechanisms involved in promoting the efficiency enhancement, in depth XPS studies were performed, before and after the post-deposition annealing, on samples named CZTS/CdS heterojunction. Typically, the mean free path of photoelectrons is a few atomic layers. Thus, XPS is a very surface sensitive technique giving information that concerns a few atomic layers. XPS depth profile curves corresponding to Cu, Zn, Sn, Cd and S elements are shown in Fig. 5. Fig. 5(1a) and (1c) show the results obtained before post-deposition annealing, while, Fig. 5(1b) and (1d) correspond to CZTS/CdS absorber after post-deposition annealing. Despite very similar XPS depth profiles (Fig. 5(1a) and (1b)), a non-negligible difference is detected in the measurements. In the case of the sample with the post-deposition annealing small quantities of Cu and Zn about 0.4–0.5 at% are detected in the CdS layer (Fig. 5(1d). These quantities, even though being very small, are clearly detected by XPS in the samples with post-deposition annealing, while, they are not detected in the samples before post-deposition annealing (Fig. 5(1c)). Furthermore, the depth profile curves indicate that these small quantities of Cu and Zn are almost constant along the depth of the CdS thin film, indicating a homogeneous migration of those elements from the CZTS film to the CdS layer due to the post-deposition annealing. Interestingly, in the case of Sn no migration was detected (Fig. 5(1d)). Fig. 5(2) shows the XPS depth profile for Cd, before and after post-deposition annealing. Before post-deposition annealing the atomic concentration of Cd in the CdS region is higher than the atomic concentration of S, which is consistent with the n-type doping of CdS through sulphur vacancies (see Fig. 5(1a)). After postdeposition annealing the atomic concentration of Cd in the CdS region decreased while the S increased, as a result of Cd out diffusion towards the CZTS layer and S in diffusion from the CZTS surface (see Fig. 5(1b)). This is consistent with improvement in crystallinity of the CdS layer, as suggested by the Raman results, discussed above. Fig. 5(2) also highlights the Cd reduction in the CdS region and its increase in the CZTS surface region, as shown in the encircled area of the figure. The 292

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

Fig. 5. XPS depth profile analysis of a sample consisting of the SLG/Mo/CZTS/CdS structure before (5 1a)) and after (5(1b)) post-deposition annealing. Fig. 5(1c) and (1d) show the respective zoom in of the depth profile curves. Fig. 5(2) shows the XPS depth profile analysis for the Cd element before and after the post-deposition annealing.

4. Conclusions

diffusion of Cd through CdS/CZTS interface may lead to the formation of a buried homojunction due to the conductivity type inversion in the CZTS surface via the formation of donor states corresponding to the filling of Cu vacancy sites with Cd. This may be the mechanism responsible for the high solar cell performance enhancement observed in this work. This possibility is also consistent with the observations from the PL measurements, which show a big reduction in non-radiative recombination. The formation of a buried homojunction may also explain the removal of the double diode behaviour observed in the J-V curves before post-deposition annealing, since this is often due to heterojunction interface defects. However, further studies are required to confirm this hypothesis.

The work reported in this paper consisted on a systematic study of the effect of the post-deposition annealing on the performance of CZTS based thin film solar cells. The results showed that for CZTS thin film solar cells based on the RTP sulphurization of precursors, such as, (ZnS/ SnS2/Cu) or (ZnS/SnS2/Cu)(8_periods), in an H2S atmosphere, the optimum post-deposition annealing conditions are 300 °C for 15 mins at atmospheric pressure. The effect of post-deposition annealing is dependent on both the precursor structure and final composition of the absorber. In this regard, we conclude that the positive impact of the annealing on the performance of the solar cells is maximized when a single annealing step is applied to a cell formed with an absorber with 293

Solar Energy Materials and Solar Cells 170 (2017) 287–294

M.G. Sousa et al.

ideal composition and obtained from an 8 period precursor. Significant efficiency enhancements have been obtained. The XRD and Raman scattering structural studies showed that the structure of the cells undergoes some adjustment upon post-deposition annealing. Those studies reveal an apparent improvement of CdS buffer layer crystallinity, a slight CZTS crystal relaxation and also a slight degradation as inferred from the increase in the FWHM of the main XRD and Raman scattering peaks. The PL studies showed that a substantial reduction of non-radiative recombination channels occurs and that fluctuating potentials dominate the luminescence from CZTS after post-deposition annealing. XPS studies revealed that upon annealing, Cu and Zn migrate from the CZTS absorber to CdS buffer layer while some Cd migrates in the opposite direction. This latter observation led us to speculate that a buried homojunction will form in the CZTS top layer reducing the activity of interface defects, thus, explaining the disappearance of the double diode behaviour and consequently the important gains in cell performance. As a result of this work we were able to acquire a better understanding of the processes occurring in the cell structure upon postdeposition annealing and boost the cell efficiency from under 1% to over 6.6%.

[8]

[9]

[10]

[11]

[12] [13]

[14]

[15]

[16]

Acknowledgements The authors acknowledge the financial support from the Portuguese Science and Technology Foundation (FCT), through the Grants SFRH/ BD/102807/2014 and SFRH/BPD/90562/2012. This work is also funded by FEDER funds through the COMPETE 2020 Programme and National Funds through FCT under the project UID/CTM/50025/2013.

[17]

[18]

References [19] [1] M. Kumar, A. Dubey, N. Adhikari, S. Venkatesan, Q. Qiao, Strategic review of secondary phases, defects and defect-complexes in kesterite CZTS-Se solar cells, Energy Environ. Sci. 8 (11) (2015) 3134–3159, http://dx.doi.org/10.1039/ C5EE02153G. [2] X. Liu, Y. Feng, H. Cui, F. Liu, X. Hao, G. Conibeer, D. Mitzi, M. Green, The current status and future prospects of kesterite solar cells: a brief review, Prog. Photovolt.: Res. Appl. 24 (2016) 879–898, http://dx.doi.org/10.1002/pip.2741. [3] M. Neuschitzer, Y. Sanchez, T. Olar, T. Thersleff, S. Lopez-Marino, F. Oliva, M. Espindola-Rodriguez, H. Xie, M. Placidi, V. Izquierdo-Roca, I. Lauermann, K. Leifer, A. Pérez-Rodriguez, E. Saucedo, The complex surface chemistry of kesterites: cu/zn re-ordering after low temperature post deposition annealing and its role in high performance devices, Chem. Mater. 27 (15) (2015) 5279–5287, http:// dx.doi.org/10.1021/acs.chemmater.5b01473. [4] W. Shockley, H.J. Queisser, Detailed balance limit of efficiency of p-n junction solar cells, J. Appl. Phys. 32 (3) (1961) 510–519, http://dx.doi.org/10.1063/1.1736034. [5] H. Sugimoto, C. Liao, H. Hiroi, N. Sakai, T. Kato, Lifetime improvement for high efficiency Cu2ZnSnS4 submodules, in: Proceedings of the 39th IEEE Photovoltaic Specialists Conference at Tampa, Florida, USA, 2013, pp. 3208–3211. [6] H. Katagiri, K. Jimbo, M. Tahara, H. Ariki, K. Oishi, The influence of the composition ratio on CZTS-based thin film solar cells, in: Proceedings of Materials Research Society Symposium, vol. 1165, 2009, pp. M01–M04. 〈http://dx.doi.org/ 10.1557/PROC-1165-M04-01〉. [7] A. Nagoya, R. Asahi, R. Wahl, G. Kresse, Defect formation and phase stability of

[20]

[21]

[22]

[23]

[24] [25]

294

Cu2ZnSnS4 photovoltaic material, Phys. Rev. B 81 (2010) 113202–113204, http:// dx.doi.org/10.1103/PhysRevB.81.113202. T. Maeda, S. Nakamura, T. Wada, First principles calculations of defect formation in in-free photovoltaic semiconductors Cu2ZnSnS4 and Cu2ZnSnSe4, Jpn. J. Appl. Phys. 50 (4S) (2011) 04DP07, http://dx.doi.org/10.1143/JJAP.50.04DPO7. S. Chen, J.H. Yang, X.G. Gong, A. Walsh, S.H. Wei, Intrinsic point defects and complexes in the quaternary kesterite semiconductor Cu2ZnSnS4, Phys. Rev. B 81 (24) (2010) 245204, http://dx.doi.org/10.1103/PhysRevB.81.245204. D.B. Mitzi, O. Gunawan, T.K. Todorov, K. Wang, S. Guha, The path towards a highperformance solution-processed kesterite solar cell, Sol. Energy Mater. Sol. Cells 95 (6) (2011) 1421–1436, http://dx.doi.org/10.1016/j.solmat.2010.11.028. A. Redinger, D.M. Berg, P.J. Dale, R. Djemour, L. Gutay, T. Eisenbarth, N. Valle, S. Siebentritt, Route toward high-efficiency single-phase CuZnSn(S,Se) thin-film solar cells: model experiments and literature review, IEEE J. Photovolt. 1 (2) (2011) 200–206, http://dx.doi.org/10.1109/JPHOTOV.2011.2168811. S. Siebentritt, S. Schorr, Kesterites- a challenging material for solar cells, Prog. Photovolt.: Res. Appl. 20 (2012) 512–519, http://dx.doi.org/10.102/pip.2156. M.G. Sousa, A.F. Cunha, P.A. Fernandes, J.P. Teixeira, R.A. Sousa, J.P. Leitão, Effect of rapid thermal processing conditions on the properties of Cu2ZnSnS4 thin films and solar cell performance, Sol. Energy Mater. Sol. Cells 126 (2014) 101–106, http://dx.doi.org/10.1016/j.solmat.2014.03.043. P.A. Fernandes, P.M.P. Salomé, A.F. da Cunha, Study of polycrystalline Cu2ZnSnS4 films by Raman scattering, J. Alloy. Compd. 509 (2011) 7600–7605, http://dx.doi. org/10.1016/j.jallcom.2011.04.097. J.M. Doña, J. Herrero, Dependence of electro-optical properties on the deposition conditions of chemical bath deposited CdS thin films, J. Electrochem. Soc. 144 (11) (1997) 4091–4098, http://dx.doi.org/10.1149/1.1838141. J.P. Teixeira, R.A. Sousa, M.G. Sousa, A.F. Cunha, P.A. Fernandes, P.M.P. Salomé, J.C. González, J.P. Leitão, Erratum: comparison of fluctuating potentials an donoracceptor pair transitions in a Cu poor Cu2ZnSnS4 based solar cell, Appl. Phys. Lett. 107 (4) (2015), http://dx.doi.org/10.1063/1.4927663. J.P. Teixeira, R.A. Sousa, M.G. Sousa, A.F. Cunha, P.A. Fernandes, P.M.P. Salomé, J.C. González, J.P. Leitão, Radiative transitions in highly doped and compensated chalcopyrites and kesterites: the case of Cu2ZnSnS4, Phys. Rev. B – Condens. Matter Mater. Phys. 90 (23) (2014) 235202, http://dx.doi.org/10.1103/PhysRevB.90. 235202. J.P. Teixeira, P.M.P. Salomé, M.G. Sousa, P.A. Fernandes, S. Sadewasser, A.F. da Cunha, J.P. Leitão, Optical and structural investigation of Cu2ZnSnS4 based solar cells, Phys. Status Solidi B 254 (11) (2016) 2129–2135, http://dx.doi.org/10.1002/ pssb.201670569. P.M.P. Salomé, P.A. Fernandes, J.P. Leitão, M.G. Sousa, J.P. Teixeira, A.F. Cunha, Secondary crystalline identification in Cu2ZnSnSe4 thin films: contributions from Raman scattering and photoluminescence, J. Mater. Sci. 49 (21) (2014) 7425–7436, http://dx.doi.org/10.1007/s10853-014-8446-2. J.P. Leitão, N.M. Santos, P.A. Fernandes, P.M.P. Salomé, A.F. Cunha, J.C. González, G.M. Ribeiro, F.M. Matinaga, Photoluminescence and electrical study of fluctuating potentials in Cu2ZnSnS4-based thin films, Phys. Rev. B – Condens. Matter Mater. Phys. 84 (2) (2011) 024120, http://dx.doi.org/10.1103/PhysRevB.84.024120. S. Levcenko, V.E. Tezlevan, E. Arushanov, S. Schorr, T. Unold, Free-to-bound recombination in near stoichiometric Cu2ZnSnS4 single crystals, Phys. Rev. B 86 (4) (2012) 045206, http://dx.doi.org/10.1103/PhysReVB.86.045206. A. Le Donne, S. Marchionna, P. Garattini, R.A. Mereu, M. Acciarri, S. Binetti, Effects of CdS buffer layers on photoluminescence properties of Cu2ZnSnS4 Solar Cells, Int. J. Photoenergy (2015) 1–8, http://dx.doi.org/10.1155/2015/583058. J.-S. Seol, S.-Y. Lee, J.-C. Lee, H.-D. Nam, K.-H. Kim, Electrical and optical properties of Cu2ZnSnS4 thin films prepared by rf magnetron sputtering process, Sol. Energy Mater. Sol. Cells 75 (1–2) (2003) 155–162, http://dx.doi.org/10.1016/ S0927-0248(02)00127-7. C. Persson, Electronic and optical properties of Cu2ZnSnS4 and Cu2ZnSnSe4, J. Appl. Phys. 107 (2010) 053710, http://dx.doi.org/10.1063/1.3318468. A.P. Levanyuk, V.V. Osipov, Edge luminescence of direct-gap semiconductors, Sov. Phys. Uspekhi 24 (1981) 187–215, http://dx.doi.org/10.1070/ PU1981v024n03ABEH004770.