Investigating the open-circuit voltage deficit in Cu2ZnSn(S,Se)4 solar cells
Tai, Kong Fai
Date of Issue2016
School of Physical and Mathematical Sciences
Energy Research Institute
Thin film solar cells (TFSC) such as Cu(In,Ga)Se2 (CIGS) and CdTe technology have been successfully commercialized with cost-per-watt on par with the conventional Si photovoltaics. Additional benefits of TFSC include shorter energy payback time, better light absorbance, and high flexibility (if deposited on flexible substrate) for a variety of applications. Among the industrialized TFSC, CIGS is less toxic and possesses the highest power conversion efficiency (PCE), but scarcity of indium and gallium could limit its growth potential. Therefore, research on earth-abundant and non-toxic Cu2ZnSn(S,Se)4 (CZTSSe) solar cell has generated tremendous interest in the photovoltaic community. However, large open-circuit voltage deficit (VOC,def) in CZTSSe solar cell remains unresolved even after a decade of research and development. Although open-circuit voltage (VOC) can be enhanced by increasing its bandgap (Eg) through the tuning of S/Se ratio, VOC,def (Eg/q - VOC) generally becomes more severe accompanied by significant loss in short-circuit current density (JSC) and fill factor (FF). These losses limit the power conversion efficiency (PCE) of CZTSSe solar cell. Therefore, in-depth investigations on the fundamental limitations are required to tackle the VOC,def issue. Other issues such as the JSC and FF losses, specifically in the high bandgap CZTSSe solar cells, need to be investigated as well to further improve the device performance. This thesis focuses on investigating the fundamental limitations of CZTSSe solar cells, in particular the large VOC,def, by electrical, optical and physical characterizations. Full bandgap range CZTSSe solar cells from 1.0 to 1.5 eV were studied in this study to provide a comprehensive understanding on the limitations of the CZTSSe technologies. The loss factors were investigated from two perspectives: (i) the intrinsic electrical and defect properties of the CZTSSe absorber layer, and (ii) the extrinsic loss factors arising from the interfacial secondary phases and interfacial resistances such as high shunt conductance (which lowers VOC) and high series resistance (which lowers JSC). The intrinsic limitations were investigated by self-developed electrical AC Hall measurement system and optical photoluminescence (PL) technique. From the AC Hall measurement, it was found that CZTSSe thin films (regardless of their bandgaps) typically have low carrier mobility of less than 1 cm2/V.s. This pinpoints one major factor to account for the high VOC deficit. In addition, PL studies indicate that CZTSSe material has a high defect density in the order of 1020 cm-3, which results in a deep fluctuating potential (γ) at the conduction and valence band given its low dielectric constant. The upper bound of γ for CZTSSe ranges from 115 meV (Se-rich CZTSSe) to 144 meV (S-rich CZTSSe). The fluctuating conduction and valence band (2γ) of 230 meV and 288 meV partly accounts for the VOC,def of 588 meV and 933 meV in Se-rich and S-rich CZTSSe, respectively. Adding a general thermodynamic loss of ~350 meV to 2γ, the VOC,def in Se-rich CZTSSe can be easily explained, but the VOC,def in S-rich CZTSSe remains largely unaccountable. The deep fluctuating potential will essentially reduce the energy of the photo-generated carriers and results in much lower PL photon energy compared to its bandgap. This is consistent with the low carrier mobility which partly account for the large VOC deficit. Temperature-dependent PL further suggests that non-radiative recombination centers could be the dominant defects in CZTSSe absorbers, and the defect levels are much deeper compared with the benchmark CIGS absorber. From the Arrhenius relation, the activation energy (Ea) which includes the effect of defect energy level and non-radiative recombination rate can be extracted. The Ea for CIGSSe, Se-rich CZTSSe and S-rich CZTSSe were found to be 20 meV, 48 meV and 67 meV, respectively. The particularly high Ea in S-rich CZTSSe serves as an additional factor to explain the largely unaccountable VOC,def (from the sum of thermodynamic loss and 2γ mentioned above). These undesired defect properties will exacerbate carrier recombination which already suffers from low carrier mobility. This will essentially lower the VOC of the CZTSSe solar cells. Extrinsic loss factors were investigated by electrical colour I-V (current-voltage sweep)¬, Suns-VOC and various physical characterizations including Glancing-Incidence X-Ray Diffraction (GIXRD), Field-Emission Scanning Electron Microscopy (FE-SEM), Raman mapping, X-ray Photoelectron Spectroscopy (XPS), and XPS depth profile. As shunt conductance was severe in high bandgap S-rich CZTSSe solar cell, it further lowers the VOC by external device leakage path. Physical characterizations suggested the formation of shunting secondary phases at the CZTSSe/CdS interface which provides such leakage path. Besides the VOC deficit issue, JSC loss resulted from the high series resistance (RS) was also observed in the high bandgap CZTSSe solar cell. Suns-Voc measurement indicates the formation of non-ohmic back contact, which could be due to Schottky contact formation arising from the low carrier density in high bandgap CZTSSe (as probed by AC Hall measurement), and hole blocking secondary phases forming at the CZTSSe/Mo interface which inhibits the carrier injection into Mo electrode. The latter factor was investigated extensively by various physical characterizations, and significant presence of high bandgap secondary phases was indeed found at the back contact interface. High bandgap CZTSSe, which are vulnerable to the formation of secondary phases, could suffer more in VOC deficit, JSC loss, and FF loss. With better understanding on the intrinsic limiting factor of VOC deficit, defect engineering was attempted in order to improve the crystal quality of CZTSSe. It was demonstrated that the defect properties could be improved, specifically by reducing and passivating the non-radiative defects through extrinsic Sb doping. With the reduction in total defect density by approximately half, the depth of the potential fluctuation was reduced from ~115 meV to ~86 meV. The reduction of γ corresponded well with the increase of VOC. Other approaches such as partial or full elemental substitution could also be pursued to engineer the defect properties, as elaborated in Chapter 7. Specifically, the high concentration of the deep CuZn antisite should be inhibited and shallower defects such as Cu-vacancy (VCu) should be promoted for improved device performance. Defect engineering should lift the kesterite solar cell research into a new phase which could potentially overcome the VOC deficit bottleneck.