Final Report Summary - JUMPKEST (JUnction iMProved KESTerite based solar cells for cost efficient sustainable photovoltaic technologies)
With increasing market penetration and cut-backs of market incentives the photovoltaic (PV) technologies are facing new levels of harsh competition and are currently undergoing a painful consolidation phase due to heavily dropped module prices. European PV industry and research is struggling to hold its leading expertise in 2nd generation thin film PV technologies. Kesterite (Cu2SnZnSe4, CZTSe) thin film solar cells comprise a newly emerging technology with promising perspectives, because the absorber is made up entirely by earth-abundant elements of low toxicity. However, efficiency of kesterite solar cells has been lacking behind the more mature other chalcogenide technologies such as Cu(In,Ga)Se2 or CdTe.
Compared with these technologies, the non-radiative recombination and therefore the voltage losses of CZTSe solar cells are much more pronounced and one of the main reasons of their inferior performance. The common model of these cells states strong interface recombination, recombination centers at certain internal grain boundaries and/or potential fluctuations in the absorber as the most probable origin of these losses.
The main goal of the JumpKEST project was to characterize state-of-the-art kesterite solar cells, identify main limiters by comparison with numerical modelling and to develop strategies to improve the heterojunction of these solar cells. The aim was a jump in the power conversion efficiencies to ≥ 10% using new processes, cell designs and alternative buffer window layers.
For this purpose, state-of-the-art Cu2SnZnSe4 solar cells prepared at the IREC laboratory were in addition to standard opto-electronic, structural and compositional characterization methods analyzed by advanced methods such as multi-wavelength excitation Raman spectroscopy analysis (MW Raman), High Resolution Transmission Electron Microscopy / Electron Energy Loss Spectroscopy (HR TEM/EELS, in collaboration with the University of Uppsala) or High Kinetic Energy X-ray Photoelecton Spectroscopy (HIKE XPS, in collaboration with the Helmholtz-Zentrum Berlin).
In order to reduce parasitic absorption and interface recombination at the absorber/buffer junction, several different Cd-free buffer layers (e.g. In2S3, Zn(O,S), ZnSe, ...) as well as hybrid double layer structures with reduced Cd thickness (ZnS/CdS, In2S3/CdS, ...) with a great variation of deposition techniques (e.g. atomic layer deposition (ALD), sputtering, evaporation, chemical bath deposition (CBD)) have been tested. Best Cd-free results were obtained for ZnS buffer layers by CBD (5.2 %) and ZnS/CdS hybrid devices with reduced Cd content (8.9%). The deposition of very thin (Cd,Zn)S buffer layers from chemical bathes with mixed Cd and Zn salts resulted in promising efficiencies (>8 %) and exceeded the pure CdS reference devices, however homogeneity still remains as an issue. These new ternary buffer layers are expected to be especially interesting for wide gap absorber such as Cu2ZnSnS4, as for these absorbers the band alignment between the absorber and the standard CdS buffer is expected to exhibit an unfavourablecliff-like offset. However, the work on the wide gap absorbers was not within the scope of the JumpKEST project.
Within the framework of the JumpKEST project, Cd-containing devices still outperformed the Cd-free alternatives. Comparison of numerical modelling with HIKE-XPS characterization revealed a near optimum band alignment with only a small (0.0-0.3 eV) cliff-like conduction band offset between the CZTSe absorber and the CBD CdS buffer prepared from Cd nitrate salts. We concluded, that modifications of the CZTSe absorber surface will be a more effective mean to improve the open circuit voltage than further variations of the buffer layer. In order reduce the density of active defects present at the absorber/buffer junction, several attempts to passivate the surface were undertaken. The most effective ones resulted to be 1) Dipping of the absorber in (NH4)2S solutions: The bath in (NH4)2S solutions effectively cleans the surface of the absorber, helps to remove secondary phases (SnxSe, ZnSe) and passivates the surface. 2) Post-deposition soft annealing (PDA): The performance of CZTSe devices is substantially improved by a soft annealing. Maximum solar cell efficiencies were obtained for PDAs at 250ºC for 20 min. A detailed MW Raman analysis showed that the main effect of this PDT is a redistribution of Cu/Zn atoms between absorber bulk and absorber surface, resulting in an electronically more favorable Cu-poor Zn-rich absorber surface. 3) Addition of nanometric Ge layers to the Cu/Zn/Cu/Sn precursor stack: The incorporation of small amounts of Ge on top of the precursors was shown to substantially modify the surface region of the absorber, leading to large grains and a boost in the open circuit voltage. HR TEM measurements revealed the presence of at least two types of grain boundaries in CZTSe absorbers: “straight” grain boundaries, mainly perpendicular to the substrate separate large grains in upper part of the absorber and are clearly Cu-enriched. The “meandering” grain boundaries showed oxygen inclusions and separated mainly the upper part from the lower part of the absorber parallel to the substrate.
The absorber modification by nanometric Ge layers on top of the precursors has been shown to be especially important as it effectively reduces internal interfaces through a reduction of the “meandering” CZTSe grain boundaries, leading to large crystallites and boost of 30-50 mV in the open circuit voltage.
The deposition of an additional thin Ge layer (10nm) below the precursors helped to completely inhibit the formation of “meandering” grain boundaries and let to improved short circuit current densities and device performances. A detailed combined X-ray fluorescence, X-ray diffraction and MW Raman analysis of different stages of the reactive annealing process allowed the explanation of variations in the reaction mechanisms. Ge acts as a flux agent, causes the fast formation of ternaries and alloys, prevents Sn losses during selenization, and leads to the formation of large grains with a width of several microns.
As result of the combined improvements made to the CZTSe devices in the framework of the JumpKEST project, efficiencies in the IREC laboratory have improved substantially from 8.2% at the beginning of the project (Neuschitzer et al., Prog. in Photov. 2015, DOI:10.1002/pip.2589) to a current record efficiency of 10.6% (Giraldo et al., Prog. in Photov. 2016, DOI: 10.1002/pip.2797) clearly surpassing the efficiency target of 10%.
More importantly, the research of the JumpKEST project has helped to shed light on the limitations of kesterite solar cells and improved the fundamental understanding of this emerging class of solar cells. The outcome of the project enabled the design of a strategy and an efficiency road map for the future improvement of kesterite solar cells on a combined European level, which will be tackled from 2017-2020 in the joint European H2020 project StarCell (H2020-NMBP-03-2016-720907, www.starcell.eu).
Compared with these technologies, the non-radiative recombination and therefore the voltage losses of CZTSe solar cells are much more pronounced and one of the main reasons of their inferior performance. The common model of these cells states strong interface recombination, recombination centers at certain internal grain boundaries and/or potential fluctuations in the absorber as the most probable origin of these losses.
The main goal of the JumpKEST project was to characterize state-of-the-art kesterite solar cells, identify main limiters by comparison with numerical modelling and to develop strategies to improve the heterojunction of these solar cells. The aim was a jump in the power conversion efficiencies to ≥ 10% using new processes, cell designs and alternative buffer window layers.
For this purpose, state-of-the-art Cu2SnZnSe4 solar cells prepared at the IREC laboratory were in addition to standard opto-electronic, structural and compositional characterization methods analyzed by advanced methods such as multi-wavelength excitation Raman spectroscopy analysis (MW Raman), High Resolution Transmission Electron Microscopy / Electron Energy Loss Spectroscopy (HR TEM/EELS, in collaboration with the University of Uppsala) or High Kinetic Energy X-ray Photoelecton Spectroscopy (HIKE XPS, in collaboration with the Helmholtz-Zentrum Berlin).
In order to reduce parasitic absorption and interface recombination at the absorber/buffer junction, several different Cd-free buffer layers (e.g. In2S3, Zn(O,S), ZnSe, ...) as well as hybrid double layer structures with reduced Cd thickness (ZnS/CdS, In2S3/CdS, ...) with a great variation of deposition techniques (e.g. atomic layer deposition (ALD), sputtering, evaporation, chemical bath deposition (CBD)) have been tested. Best Cd-free results were obtained for ZnS buffer layers by CBD (5.2 %) and ZnS/CdS hybrid devices with reduced Cd content (8.9%). The deposition of very thin (Cd,Zn)S buffer layers from chemical bathes with mixed Cd and Zn salts resulted in promising efficiencies (>8 %) and exceeded the pure CdS reference devices, however homogeneity still remains as an issue. These new ternary buffer layers are expected to be especially interesting for wide gap absorber such as Cu2ZnSnS4, as for these absorbers the band alignment between the absorber and the standard CdS buffer is expected to exhibit an unfavourablecliff-like offset. However, the work on the wide gap absorbers was not within the scope of the JumpKEST project.
Within the framework of the JumpKEST project, Cd-containing devices still outperformed the Cd-free alternatives. Comparison of numerical modelling with HIKE-XPS characterization revealed a near optimum band alignment with only a small (0.0-0.3 eV) cliff-like conduction band offset between the CZTSe absorber and the CBD CdS buffer prepared from Cd nitrate salts. We concluded, that modifications of the CZTSe absorber surface will be a more effective mean to improve the open circuit voltage than further variations of the buffer layer. In order reduce the density of active defects present at the absorber/buffer junction, several attempts to passivate the surface were undertaken. The most effective ones resulted to be 1) Dipping of the absorber in (NH4)2S solutions: The bath in (NH4)2S solutions effectively cleans the surface of the absorber, helps to remove secondary phases (SnxSe, ZnSe) and passivates the surface. 2) Post-deposition soft annealing (PDA): The performance of CZTSe devices is substantially improved by a soft annealing. Maximum solar cell efficiencies were obtained for PDAs at 250ºC for 20 min. A detailed MW Raman analysis showed that the main effect of this PDT is a redistribution of Cu/Zn atoms between absorber bulk and absorber surface, resulting in an electronically more favorable Cu-poor Zn-rich absorber surface. 3) Addition of nanometric Ge layers to the Cu/Zn/Cu/Sn precursor stack: The incorporation of small amounts of Ge on top of the precursors was shown to substantially modify the surface region of the absorber, leading to large grains and a boost in the open circuit voltage. HR TEM measurements revealed the presence of at least two types of grain boundaries in CZTSe absorbers: “straight” grain boundaries, mainly perpendicular to the substrate separate large grains in upper part of the absorber and are clearly Cu-enriched. The “meandering” grain boundaries showed oxygen inclusions and separated mainly the upper part from the lower part of the absorber parallel to the substrate.
The absorber modification by nanometric Ge layers on top of the precursors has been shown to be especially important as it effectively reduces internal interfaces through a reduction of the “meandering” CZTSe grain boundaries, leading to large crystallites and boost of 30-50 mV in the open circuit voltage.
The deposition of an additional thin Ge layer (10nm) below the precursors helped to completely inhibit the formation of “meandering” grain boundaries and let to improved short circuit current densities and device performances. A detailed combined X-ray fluorescence, X-ray diffraction and MW Raman analysis of different stages of the reactive annealing process allowed the explanation of variations in the reaction mechanisms. Ge acts as a flux agent, causes the fast formation of ternaries and alloys, prevents Sn losses during selenization, and leads to the formation of large grains with a width of several microns.
As result of the combined improvements made to the CZTSe devices in the framework of the JumpKEST project, efficiencies in the IREC laboratory have improved substantially from 8.2% at the beginning of the project (Neuschitzer et al., Prog. in Photov. 2015, DOI:10.1002/pip.2589) to a current record efficiency of 10.6% (Giraldo et al., Prog. in Photov. 2016, DOI: 10.1002/pip.2797) clearly surpassing the efficiency target of 10%.
More importantly, the research of the JumpKEST project has helped to shed light on the limitations of kesterite solar cells and improved the fundamental understanding of this emerging class of solar cells. The outcome of the project enabled the design of a strategy and an efficiency road map for the future improvement of kesterite solar cells on a combined European level, which will be tackled from 2017-2020 in the joint European H2020 project StarCell (H2020-NMBP-03-2016-720907, www.starcell.eu).