The effect of sodium on antimony selenide thin film solar cells

Abstract

Sodium (Na) has been identified as a benign contaminant in some thin film solar cells. Here, we have studied the impact of sodium on the properties of CdS/Sb₂Se₃ superstrate thin film solar cells. NaF deposited at the CdS/Sb₂Se₃ heterojunction severely degraded the device performance, yet NaF introduced into the Sb₂Se₃ film after the completion of the heterojunction showed a negligible influence on the film conductivity and device performance. We concluded that sodium could easily diffuse into the Sb₂Se₃ film and probably locates in the gap between (Sb₄Se₆)_n ribbons, being inert in Sb₂Se₃ solar cells.

---

1. Introduction

The antimony selenium thin-film solar cell has shown potential for low-cost and low-toxic photovoltaic applications.¹ A certified efficiency as high as 5.6% was reported using a fluorine-doped tin oxide (FTO) substrate with an Sb₂Se₃ absorber layer produced by rapid thermal evaporation (RTE).² The Sb₂Se₃ thin films were deposited at 550 °C (source temperature) with the substrate temperature above 300 °C, and such high temperature processing can facilitate the inter-diffusion between different layers, including diffusion of impurities from the substrate to the absorber. Since soda-lime glass is a mixture of SiO₂, Na₂O and CaO, many impurities could diffuse into the active layers of solar cells from soda-lime glass substrate during the high temperature processing. Particularly, alkaline metals are highly mobile elements with small activation energies far below 1 eV for diffusion in CdTe,³ thus it's easier for sodium to diffuse into the active layers and influence the performance of the devices. In addition, in contrast to most semiconductors like Si or copper indium gallium selenide (CIGS) that are tightly bonded through covalent bonds in three dimensions, Sb₂Se₃ possesses peculiar one dimensional crystal structure. It is made up of (Sb₄Se₆)_n ribbons in the [001] direction through strong covalent Sb–Se bonds yet loosely stacked at the [010] and [100] directions by weak van der Waals force. As a result, sodium can diffuse easily from the glass substrate into the space between the (Sb₄Se₆)_n ribbons. For this reason, it is necessary to identify the impact of sodium on the properties of Sb₂Se₃.

Lots of researches about the impact of sodium on the performance of thin film solar cells, such as CIGS,^4,5 CdTe,^6,7 SnS,⁸ Cu₂O⁹ and so on, have been reported. For CIGS solar cells, incorporation of sodium resulted in the great enhancement of device performance through the improvement of all relevant device parameters: open-circuit voltage (V_oc), short-circuit current density (J_sc), and fill factor (FF). Although the exact mechanism is still controversial, it could be categorized as enhancing the p-type doping of CIGS absorber layer, promotion of grain growth¹⁰ or passivating grain boundaries.^11–14 The CdTe solar cell fabricated in the presence of sodium showed grain size enlargement yet deteriorated photovoltaic performance.¹⁵ Differently, Hall test and free carrier absorption showed that intentional sodium doping improved carrier concentrations in SnS thin film but demonstrated minor influence on device performance.⁸ Also for Cu₂O, sodium doping can adjust the resistivity and holes concentration in the range of 10³ to 10⁻² Ω cm, and 10¹³ to 10¹⁶ cm⁻³, respectively.⁹ Clearly, the effect of sodium doping varies significantly with the materials to which it applied.

In this work, we studied the impact of sodium on Sb₂Se₃-based solar cells, with the particular attention on grain growth and devices performance. Only plastic or quartz containers and holders were used during device fabrication process to minimize any unintentional sodium contamination. Devices were fabricated on the substrates with a 25 nm SiO₂ diffusion barrier layer between the soda-lime glass and indium-doped oxide layer to avoid uncontrollable introduction of sodium from the substrate. Superstrate ITO/CdS/Sb₂Se₃/Au devices configuration and RTE produced Sb₂Se₃ film were applied for the study because this generates the best performing device. A NaF film produced by vacuum evaporation was either introduced onto the CdS buffer layer before the Sb₂Se₃ deposition or introduced onto the as-fabricated Sb₂Se₃ film followed by annealing. X-ray photoelectron spectroscopy (XPS) and second ion mass spectrometry (SIMS) were applied to prove the existence of sodium in Sb₂Se₃ film. Device performance and temperature-dependent conductance measurement were carried out to analyze the impact of sodium on the electrical properties of Sb₂Se₃. Our results revealed that sodium showed minor influence on grain size, film conductivity and device performance, being largely inert in Sb₂Se₃ film.

2. Experimental

2.1 Device fabrication

Firstly, chemical bath deposition (CBD: 0.015 M CdSO₄, thiourea, ammonium hydroxide, DI water; reacted at 65 °C for 16 min) was used to deposit CdS buffer layer of about 60 nm thickness on soda-lime glass coated with indium-doped tin oxide (ITO, purchased from Zhuhai Kaivo Optoelectronic Technology Co., Ltd in China, Na⁺ diffusion is ≤150 μg per day per m² in the boiling water experiments), which has a SiO₂ diffusion barrier layer of ∼25 nm thickness to block the diffusion of sodium from the glass. Plastic containers were used in this step to avoid unintentional diffusion of sodium from the container's wall into the solution. A post CdCl₂-treatment was applied to the CdS layer following the same process as reported.¹⁶ To study the effectiveness of sodium on Sb₂Se₃ grain growth, a NaF layer of about 2 nm was deposited by thermal evaporation on the surface of CdS layer with the vacuum of 5 × 10⁻⁴ Pa. Subsequently, an approximately 390 nm thick Sb₂Se₃ absorber layer was deposited on top of the NaF modified CdS buffer layer using our standard RTE process, as described in our previous publications.^2,17 Alternatively, 2 nm (monitored by a calibrated quartz crystal microbalance from Fli-tech) thick NaF layer was deposited onto the back surface of Sb₂Se₃ layer for studying the impact of sodium on the performance of Sb₂Se₃ solar cells. The devices were subsequently annealed on a hotplate at 200 °C for half an hour in the glove box, facilitating the diffusion of sodium from the back surface into the Sb₂Se₃ bulk. Residual NaF on the back surface of Sb₂Se₃ film was removed by immersing the device into deionized water for half an hour at room temperature. For both modified devices (NaF on CdS or NaF on Sb₂Se₃), Au was subsequently deposited by thermal evaporation onto Sb₂Se₃ absorber layer as contact electrode (∼100 nm thickness, active area 0.092 cm²) to finish the device.

For the temperature-dependent conductance measurement, Sb₂Se₃ films were deposited on bare quartz by RTE and subsequently doped with sodium using the method mentioned above. Au bars were deposited with 15 mm × 0.2 mm dimensions on the Sb₂Se₃ thin film, and the spacing between Au bars is 0.2 mm.

2.2 Characterization and measurements

The morphology and composition of Sb₂Se₃ films were studied by scanning electron microscopy (SEM, FEI nova NanSEM450, Hillsboro, OR, USA) and energy dispersive spectroscopy (EDS, FEI Quanta 600 scanning electron microscope, 20 kV), X-ray photoelectron spectroscopy (XPS, ESCALab250) respectively. Compositional depth profile was determined by secondary ion mass spectrometry (SIMS) using Cs⁺ primary beam in positive detection mode with an analyzed area of 100 × 100 μm inside an etching area of 300 × 300 μm. The structure of Sb₂Se₃ film was investigated by X-ray diffraction (XRD) with Cu Rα radiation (Philips, X′ pert pro materials research diffractometer, Amsterdam, The Netherlands).

Solar cells were characterized by current density versus voltage (J–V) measurements using a Keithley 2420 Source Meter under standard AM 1.5G (NewPort Sol3A Class AAA Solar Simulator) illumination at room temperature. Temperature-dependent conductance measurement of the Sb₂Se₃-based film was studied by an Agilent B1500A in the dark at different temperature (range from 150 K to 370 K), while the test sample was put in liquid nitrogen trap system (Janis Research Co., Inc.) in conjunction with an temperature controller (LakeShore, 325 Temperature Controller), providing precise control over sample temperature.

3. Result and discussion

Firstly, a NaF layer of about 2 nm thickness was deposited by thermal evaporation on the surface of CdS layer, and Sb₂Se₃ thin film was subsequently deposited on the surface of NaF modified CdS layer with RTE process to explore the effect of sodium on Sb₂Se₃ grain growth. The control is Sb₂Se₃ film deposited on CdS buffer layer without NaF modification. Strictly identical RTE process was applied for both Sb₂Se₃ film depositions. Plan view SEM images of Sb₂Se₃ control and Sb₂Se₃ on sodium modified CdS buffer layer are shown in Fig. 1a and b, and the corresponding statistics of grain sizes are showed in Fig. 1c and d, respectively. For both cases, compact Sb₂Se₃ films free of pinholes and cracks were obtained. The average grain size of control and sodium-doped Sb₂Se₃ film was 255 nm and 240 nm respectively, demonstrating that sodium slightly inhibited grain growth. The grain size distribution in the sodium doped sample was also wider than control, as evidenced by the broader fitting curve. XRD patterns of these two samples were shown in Fig. 1e. Clearly, phase pure Sb₂Se₃ (JCPDS-150861) without any diffraction peaks from other crystalline impurities were obtained in both cases. Both XRD patterns showed strong (211) and (221) diffraction peaks yet very weak (020) and (120) diffraction peaks, confirming sodium doping had negligible influence on the orientation of Sb₂Se₃ films. A careful comparison of enlarge XRD patterns in Fig. 1f indicated the peak positions for both cases were identical, consistent with unchanged lattice parameter after sodium doping. The full width at half maximum (FWHM) of XRD diffraction peaks of the sodium-doped sample was also larger than the control, echoing previous SEM results that sodium slightly suppressed Sb₂Se₃ grain growth. Based on these observations, we hypothesized that sodium probably diffused into the gap between (Sb₄Se₆)_n ribbons, thus maintaining the preferred [211] and [221] orientation but inhibiting (Sb₄Se₆)_n ribbons coalescence into large grains.

Fig. 1 Impact of sodium on Sb₂Se₃ grain growth. (a) Top-view SEM images of pure Sb₂Se₃ film, (b) sodium-doped Sb₂Se₃ film. (c and d) The statistic grain size of pure and sodium-doped Sb₂Se₃ film, respectively. (e) XRD patterns and the enlarged part (f) of the sodium-doped and pure Sb₂Se₃ film deposited by RTE.

Surprisingly, a 2 nm NaF deposited on the CdS buffer severely deteriorated device performance. As shown in Table 1, the average performance for the control device was 3.84%, and the sodium doped device was only 1.86%, a significant drop in device efficiency. This might be due to inhibited grain growth upon sodium doping, but one more likely explanation is that the presence of a thin insulating NaF layer at the CdS/Sb₂Se₃ interface damages the heterojunction and ruins efficient charge separation and collection. This is quite different in contrast with traditional solar cells such as CIGS and CZTS. In these solar cells, devices of substrate configuration were employed to study the impact of sodium, and the NaF layer was always deposited onto the substrate before the deposition of absorber and buffer layer, thus circumventing influencing the quality of the heterojunction as encountered in our case.

Table 1 The comparison of device performance from the control and sodium at the of Sb₂Se₃/CdS interface device

	Voc (V)	Jsc (mA cm^−2)	FF	Efficiency
Control	0.38	22.0	46.4%	3.84%
NaF on CdS	0.29	16.0	38.1%	1.86%

---

To avoid damaging the heterojunction, we alternatively deposited NaF onto the Sb₂Se₃ layer after completion of the CdS/Sb₂Se₃ heterojunction (Fig. 2a). 20 nm thick NaF layer was deposited onto the back surface of Sb₂Se₃ by thermal evaporation, followed by a 30 min annealing carried out in the N₂ gas filled glove box to promote Na diffusion into the bulk. Before NaF evaporation, Sb₂Se₃ film was compact with a few protrusions (Fig. 2b); after NaF deposition, some “white” dots uniformly dispersed on the surface, as shown in Fig. 2c. Energy dispersive X-ray spectrum (EDS) analysis of the “white” dots showed the presence of F, Na, Se and Sb peaks at 0.69 keV, 1.1 keV, 1.5 keV and 3.5 keV, respectively, confirming its composition as NaF. To minimize the interference of the NaF leftovers, such Sb₂Se₃ film was immersed into the deionized water for half an hour to remove the sodium residual, as shown in Fig. 2d. XRD patterns of sodium-doped Sb₂Se₃ and Sb₂Se₃ control were also included in Fig. 2f. Again, all diffraction peaks agreed well with the standards JCPDS 15-0861 and (211) and (221) were dominant diffraction peaks, confirmed that both phase purity and film orientation were unchanged upon NaF post-treatment.

Fig. 2 Sodium doping introduced by post-treatment. (a) Schematic of device configuration for sodium doping study. NaF was introduced onto the back surface of Sb₂Se₃ film by thermal evaporation. Plan-view SEM images of Sb₂Se₃ film with different treatments: (b) before NaF deposition; (c) with 20 nm NaF layer; (d) with 20 nm NaF layer and follow-up DI water soaking. (e) EDS spectrum of Sb₂Se₃ film after NaF layer deposition and post-annealing treatment but without DI water soaking. (f) XRD patterns of Sb₂Se₃ films with and without NaF treatment.

To confirm the successful diffusion of sodium into Sb₂Se₃ thin films, X-ray photoelectron spectroscopy (XPS) was carried out. Samples were etched for a minute with Ar ions before XPS measurement to exclude the possible influence of the residual NaF particles. The spectra of antimony, selenium and sodium in the control and sodium-doped Sb₂Se₃ were showed in Fig. 3a–c, respectively. Compared with the control which has an obvious random noise within XPS detection limit, as shown in Fig. 3c, the sodium-doped Sb₂Se₃ showed a perfect Gaussian–Lorentzian peak at 1071.6 eV, which is in agreement with the Na 1s from the ref. 18. The presence of the signal from Na 1s orbit reveals that the added sodium diffused into the bulk of Sb₂Se₃, showcasing the success of sodium doping by our strategy. Despite there is 0.2% atomic percentage of sodium in the Sb₂Se₃ film based on the full element spectrum analysis, the change in the binding energy of Sb and Se¹⁹ was negligible, suggesting minor influence of sodium doping on the chemistry environment of Sb₂Se₃.

Fig. 3 Confirmation of sodium presence in target film: (a–c) XPS spectra of antimony, selenium and sodium, respectively. (d) Secondary ion mass spectroscopy (SIMS) depth profile of the Na-doped Sb₂Se₃ thin film.

Secondary ion mass spectrometry (SIMS) measurement was applied on a representative Na-doped Sb₂Se₃ thin-film to study the distribution of the sodium in Sb₂Se₃ thin film. The result was shown in Fig. 3d. The SIMS result indicated that sodium diffused from the back surface deep into the Sb₂Se₃ bulk. As sodium ionizes more readily than Sb or Se, its signal intensity isn't a direct measurement of element concentration, but can be only recognized as the depth profile of relative concentration. Clearly, post-annealing facilitated sodium diffusion into the bulk Sb₂Se₃ film from the white NaF nanoparticles, displaying a concentration gradient toward the CdS/Sb₂Se₃ heterojunction as dictated by the diffusion law. Note that the post-annealing was quite gentle, only at 200 °C for 30 min. However, this resulted in apparent sodium presence at the middle of the Sb₂Se₃ film (∼200 nm deep), a consequence of fast sodium diffusion in our Sb₂Se₃ possessing one dimensional crystal structure.

To investigate the effect of sodium on the device performance, the control and sodium-doped Sb₂Se₃ solar cells were fabricated employing superstrate device configuration using CdS layer as the buffer layer. Device fabrication and testing followed identical procedures. The statistical results of devices performance were showed in Fig. 4. The average short-circuited current density (J_sc) of sodium-doped devices was 0.2 mA cm⁻² larger than that of the control, and the FF is also 0.1% larger. The difference of the J_sc and FF combined leaded to a 0.2% improvement in device efficiency upon that of sodium doping, indicating that post sodium doping at least has no detrimental effect on device performance.

Fig. 4 Device performance statistics from 30 Sb₂Se₃ solar cells: (a) open-circuited voltage (V_oc), (b) short-circuited current density (J_sc), (c) Fill Factor (FF) and (d) photoelectric conversion efficiency (Eff).

We performed electrical measurements on both sodium doped and control Sb₂Se₃ films. As shown in Fig. 5a and b, both devices showed completely overlapped J–V curves during the forward and reversed scanning. The J–V curves showed no hysteresis, demonstrating that no significant ion movement driven by the external bias was observed in sodium doped devices.

Fig. 5 Electrical properties of sodium doped Sb₂Se₃ film: the forward and reverse scanned J–V curves of control (a) and sodium doped (b) solar cells. The I–V curve measured at room temperature (c) and temperature-dependent conductance (d) of the control and Na doped Sb₂Se₃ film. The inset in panel (c) is the device structure (Au–Sb₂Se₃–Au) for these conductance measurements.

Devices based on the structure of Au/Sb₂Se₃/Au on the quartz substrates were also fabricated for temperature-dependent conductance measurement. Sodium was introduced from the back surface of Sb₂Se₃ thin film by NaF evaporation and then post-annealing, the same procedure used for previous sodium doping in device. Room temperature I–V curves comparison between the control and sodium-doped samples showed that the two curves almost overlapped, indicating that sodium doping had negligible influence on film resistance (Fig. 5c). The geometry of our resistor was 200 μm long, 1.5 cm wide and 390 nm thick. Following the Ohm's law, the conductivity of the control and the sodium-doped Sb₂Se₃ was 6.7 × 10⁻⁵ S m⁻¹ and 6.9 × 10⁻⁵ S m⁻¹, respectively.

Temperature dependent conductance measurements were also carried out on this resistor, with the temperature ranged from 150 K to 370 K. The results are shown in Fig. 5d, demonstrating minuscule difference of conductivity between the control and the Na-doped one. Based on the well fitted data, the activation energy E_a was also calculated using the Arrhenius relation (eqn (1)).²⁰

where σ_dc is the DC conductivity at temperature T, σ₀ is the pre-exponential factor, k_B is Boltzmann's constant and E_a is the activation energy. It is interesting to note that ln σ_dc ∼ 1000/T can be fitted into only one linear relation during the whole temperature regimes with a very subtle change in slope for both curves, and the corresponded values of E_a were 0.539 eV and 0.552 eV, respectively. Both values were half of the band gap of Sb₂Se₃ film, suggesting that intrinsic thermal excitation of carriers from valence band to conduction band dominated the conductivity.^21–24 This implies that the doped sodium is most likely inert, introducing neither a donor nor an acceptor level in Sb₂Se₃. Since the radius of Na⁺ (0.102 nm) is smaller than the gap of the ribbons (∼0.35 nm), we hypothesized that the sodium diffused into the Sb₂Se₃ and preferred to stay at the space between the (Sb₄Se₆)_n ribbons. The Sb–Se ribbons framework determines the properties of Sb₂Se₃, such as band structure, mobility, and carrier concentration. Thus, while the doped sodium stays between (Sb₄Se₆)_n ribbons instead of occupying sites in the covalently bonded (Sb₄Se₆)_n ribbons, it doesn't modify the band structure and defect physics of Sb₂Se₃ film. We thus conclude that sodium doping is inert in Sb₂Se₃, at least for our case studied here.

4. Conclusions

In this work, we studied the effect of sodium on the grain growth and device properties of superstrate Sb₂Se₃ solar cells. By depositing the Sb₂Se₃ on the NaF modified CdS buffer layer, we found out that NaF could slightly inhibit the grain size based on the XRD and SEM results. Alternatively, sodium doping was introduced from the back surface of Sb₂Se₃ film to study its impact on device performance without ruining the CdS/Sb₂Se₃ heterojunction. XPS and SIMS study confirmed the successful incorporation of sodium into the bulk Sb₂Se₃ film. Both device performance and temperature-dependent conductance measurement showed negligible difference between the control and the sodium doped device. We thus hypothesized that sodium locates in the gap between (Sb₄Se₆)_n ribbons, being largely inert for film conductivity and device performance.

Acknowledgements

This work was supported by the seed project of Wuhan National Laboratory for Optoelectronics (WNLO), the Major State Basic Research Development Program of China (2016YFA0204000), the National Natural Science Foundation of China (91433105, 61322401), the One Thousand Young Talent Program, and the Self-determined and Innovative Research Funds of HUST (2016JCTD111). We thank the Analytical and Testing Center of HUST and the Center for Nanoscale Characterization and Devices of WNLO for the characterization support.

References

1. Y. Zhou, M. Leng, Z. Xia, J. Zhong, H. Song, X. Liu, B. Yang, J. Zhang, J. Chen, K. Zhou, J. Han, Y. Cheng and J. Tang, Adv. Energy Mater., 2014, 4, 8.
2. Y. Zhou, L. Wang, S. Chen, S. Qin, X. Liu, J. Chen, D.-J. Xue, M. Luo, Y. Cao, Y. Cheng, E. H. Sargent and J. Tang, Nat. Photonics, 2015, 9, 409–415.
3. L. Kranz, J. Perrenoud, F. Pianezzi, C. Gretener, P. Rossbach, S. Buecheler and A. N. Tiwari, Sol. Energy Mater. Sol. Cells, 2012, 105, 213–219.
4. L. Kronik, D. Cahen and H. W. Schock, Adv. Mater., 1998, 10, 31–36.
5. J. H. Yun, K. H. Kim, M. S. Kim, B. T. Ahn, S. J. Ahn, J. C. Lee and K. H. Yoon, Thin Solid Films, 2007, 515, 5876–5879.
6. J. Sites and J. Pan, Thin Solid Films, 2007, 515, 6099–6102.
7. R. Yang, D. Wang, L. Wan and D. Wang, RSC Adv., 2014, 4, 22162–22171.
8. V. Steinmann, R. E. Brandt, R. Chakraborty, R. Jaramillo, M. Young, B. K. Ofori-Okai, C. Yang, A. Polizzotti, K. A. Nelson, R. G. Gordon and T. Buonassisi, APL Mater., 2016, 4, 026103.
9. T. Minami, Y. Nishi and T. Miyata, Appl. Phys. Lett., 2014, 105, 212104.
10. J. Eid, H. Liang, I. Gereige, S. Lee and J. V. Duren, Progress in Photovoltaics: Research and Applications, 2015, 23, 269–280.
11. A. Chirila, P. Reinhard, F. Pianezzi, P. Bloesch, A. R. Uhl, C. Fella, L. Kranz, D. Keller, C. Gretener, H. Hagendorfer, D. Jaeger, R. Erni, S. Nishiwaki, S. Buecheler and A. N. Tiwari, Nat. Mater., 2013, 12, 1107–1111.
12. A. Laemmle, R. Wuerz, T. Schwarz, O. Cojocaru-Mirédin, P.-P. Choi and M. Powalla, J. Appl. Phys., 2014, 115, 154501.
13. W. Li, S. R. Cohen and D. Cahen, Sol. Energy Mater. Sol. Cells, 2014, 120, 500–505.
14. E. S. Mungan, X. Wang and M. A. Alam, IEEE Journal of Photovoltaics, 2013, 3, 451–456.
15. R. Dhere, K. Ramanathan, J. Keane, J. Zhou, H. Moutinho, S. Asher and R. Noufi, in Conference Record of the Thirty-First Ieee Photovoltaic Specialists Conference, 2005.
16. L. Wang, M. Luo, S. Qin, X. Liu, J. Chen, B. Yang, M. Leng, D.-J. Xue, Y. Zhou, L. Gao, H. Song and J. Tang, Appl. Phys. Lett., 2015, 107, 143902.
17. H. Song, X. Zhan, D. Li, Y. Zhou, B. Yang, K. Zeng, J. Zhong, X. Miao and J. Tang, Sol. Energy Mater. Sol. Cells, 2016, 146, 1–7.
18. See http://srdata.nist.gov/xps/EngElmSrchQuery.aspx?EType=PE%26CSOpt=Retri_ex_dat%26Elm=NaThe for Binding Energy of Sodium element.
19. B. Yang, L. Wang, J. Han, Y. Zhou, H. Song, S. Chen, J. Zhong, L. Lv, D. Niu and J. Tang, Chem. Mater., 2014, 26, 3135–3143.
20. Y. Pu, Z. Dong, P. Zhang, Y. Wu, J. Zhao and Y. Luo, J. Alloys Compd., 2016, 672, 64–71.
21. C. Chen, W. Li, Y. Zhou, C. Chen, M. Luo, X. Liu, K. Zeng, B. Yang, C. Zhang, J. Han and J. Tang, Appl. Phys. Lett., 2015, 107, 043905.
22. S. S. Hegedus and E. A. Fagen, J. Appl. Phys., 1992, 71, 5941.
23. R. Meaudre, Philos. Mag. B, 2006, 51, 57–61.
24. H. Overhof and W. Beyer, Philos. Mag. B, 2012, 47, 377–392.

---

Footnote

† These authors equally contributed to this work.

---