Symposium ES20 : Thin-Film Chalcogenide Semiconductor Photovoltaics

Thin-film Ag-CIGS (or ACIGS) solar cells and modules produced by MiaSolé Hi-Tech yield high efficiencies rivaling polycrystalline silicon-based modules but are based on a high-throughput, PVD-based deposition process on flexible stainless steel substrates. The process is unique among thin-film solar manufacturers in that a complete device stack is produced without a vacuum break in less than one hour. Using this manufacturing system, a high level of process control and stability is possible via real-time control of the deposition process.

To continue performance improvements that close the gap between manufacturing material and champion small devices, a better understanding of the fundamental device properties is required. This requires development of a device model that accounts for electrical performance characteristics observed under light and voltage stress over time for comparison to real-world solar cell performance.

Our recent effort to generate a fundamental device model began with employing a wide variety of characterization results from our own measurements, those of our research partners, and features of many CIGS materials that have previously been excluded in prior device models. Additionally, device parameters must be based on actual device measurements and constituents, as heterojunction ACIGS solar cell devices are complex and errors compound quickly. The overall objective was to keep the model as simple as possible and look for generally good agreement among a variety of simulated measurements.

Most of the effort has been devoted to understanding the defect structure of the ACIGS absorber layer. Conventional capacitance-voltage profiling is supplanted in our measurements by high-speed C-V profiling (HSCV) that avoids many complications of transient charge effects in highly defective materials such as ACIGS. An important consequence of these measurements is the discovery that MiaSolé absorber layers (as well as other CIGS devices) can be defined by an acceptor-rich layer (ARL) adjacent to the CdS buffer layer. This ARL has been shown to have profound consequences for device electrical performance, as it can degrade fill factor and reduce efficiency.

The physical basis for the ARL is a matter of some debate. A variation in static shallow acceptor concentration is one possible source, but it predicts neither device metastability nor measured high concentrations of DLTS-visible defects. Charge accumulation at high concentrations of deep acceptors would be possible, but accounts for neither metastability nor carrier capture kinetics observed in DLTS. Instead we hypothesize that V(Se)-V(Cu) divacancy defects are responsible for charge accumulation in the ARL, and their simulated response to light- and voltage-stressing is consistent with our measurements. This hypothesis is qualitatively consistent with positron annihilation spectroscopy (PAS) measurements of our absorber. Predicted shallow acceptor and donor levels of the divacancy also provide an explanation for the long-observed high dopant compensation in CIGS, and at the concentrations predicted by our model, it is possible that the divacancies alone can be used to explain most of the charge and doping profile of the absorber.

Emphasis of the divacancy defect in our device model can guide MiaSolé Hi-Tech to new directions for device improvement that leverage our ability to directly manipulate absorber stoichiometry in our unique thin-film solar manufacturing system.

In this work, we present a suite of characterization tools developed to accurately and systematically analyze losses in non-Si photovoltaic solar cell technologies, in particular thin-film chalcogenide devices. This characterization platform will enable easy access to the implied open-circuit voltage (iV_oc) and fill factor (iFF) of devices, to their pseudo fill factor (pFF) and thus series resistance (R_s) at maximum power point, and to their complete-stack contact resistance (R_c). Together with calculated detailed-balance performance limits and traditional J-V measurements, these metrics will enable a precise accounting of recombination and resistive losses, from fundamental efficiency limits to experimental device performance. The characterization platform includes systematic determination of the absorber optical parameters (n & k), optical modeling of the device, calculation of the iV_oc from measurement of the external radiative efficiency (ERE) of the absorber, J_sc-V_oc measurements, and measurement on the contact stack resistance using through-the-absorber transfer line method (TLM).
Cd(Se)Te solar cells, our test system, provide a prime example of the potential impact of the techniques we propose to develop and implement: although they exhibit strong photoluminescence response, record poly-Cd(Se)Te cells have bandgap-voltage deficits (W_oc) of approximately 550 mV, as compared with below 400 mV for all other mature PV technologies. Similarly, these record Cd(Se)Te devices have FFs below 80%, when other mature cells are near or above 85%. Frustratingly, a systematic identification of the origin of these sub-par performances – for example recombination or resistive losses – has been lacking, thus slowing down the development of these technologies. Similarly, it is often asserted that Cd(Se)Te cells need a better back (hole) contact. Although we believe this is true, it is currently unknown how high the V_oc and FF could be for a given cell if it had a perfect back contact.
Using the proposed set of characterization techniques, users will be able to identify the sources of inefficiencies in thin-film chalcogenide solar cells, from bulk material quality and surface passivation to contact selectivity and resistivity. Rapid absorber and contact improvement will thus be possible.

Several characterization techniques are usually performed on a localized region of any material because of convenience in a routine basis. However, lateral non-uniformities in chemical composition, structural defects and even the presence of voids can be found in a polycrystalline solar cell, i.e. CI(G)S, influencing the local carrier properties of the material. This implies that some of the material properties extracted from local characterization methods may not be representative of the full device performance complicating the correlation between measurements. For example, time-resolved photoluminescence (TRPL) is typically measured locally, and the effective lifetime values determined (including the influence of front/back recombination, etc.) are closely related to the open circuit voltage which is a macroscopic parameter that might be affected by any carrier lifetime non-uniformity. Therefore, information about the influence of spatial inhomogeneities on the carrier transport is important to assess whether it is enough to use local parameters to correlate different characterization results —and to predict the performance of the solar cell—, or to what extent it is required to consider non-uniform parameters that could be more representative of the whole device.

In this contribution, we use transient PL measurements spatially resolved in the micrometer-scale range to access the carrier properties of different absorbers. The TRPL setup is coupled into a microscope enabling to operate in confocal or widefield illumination mode. Widefield mode is used to map the samples in a broader range while confocal mode attempts to resolve the influence of the grains inside the structure with the highest optical resolution of the system (< 1 µm). Besides all the mechanisms involved in the determination of the effective carrier lifetime such as surface recombination, charge separation, etc., the use of confocal mode requires the consideration of lateral diffusion of carriers as well as the proper evaluation of the sample performance due to possible high injection effects. Accordingly, the interpretation of the measurements in both modes is aided by 2D simulations, which are also compared to the macroscopic I-V parameters of the devices. Specifically, 2D device simulations allow us to quantify the lateral diffusion of carriers as well as to include other key parameters involved in TRPL measurements, i.e. surface recombination or carrier mobility. All simulations use as input measured absorption coefficient as a function of the bandgap grading for improved modelling. Finally, results from other spatially resolved techniques will be shown to correlate device performance with TRPL mapping and support the main findings.

Numerical simulation of solar cell device physics for predicting the performance, design and optimization of solar cells is a well explored active area of research. There are many tools and software packages like AFORS-HET, SCAPS-1D, AMPS-1D, PC1D, Silvaco TCAD, Sentaurus TCAD, etc., which are available freely or commercially, to perform numerical simulation of solar cells. The fundamental physics equations solved are continuity equations for charge carriers, namely electrons and holes, and the Poisson equation for electrostatic potential. In the continuity equation for electrons and holes the generation and recombination term is modeled through radiative processes, the SRH process, the Auger process etc. For the SRH generation recombination process an effective lifetime is assumed in the modeling. Combining this model with the boundary conditions, light and temperature simulation conditions, one can calculate I-V curves, efficiencies, fill factors etc., for the solar cell.
As the solar cell ages performance is affected. This can be microscopically explained through the transport of different defects present in the solar cell. The transport of defects can also cause metastability in solar cells. In our previous work we explained the metastable behavior of CdTe solar cells by studying the Cu related defect transport along with the carrier transport.
In this work, we present a novel Unified Solver for studying carrier and defect transport on an equal footing. The generation recombination term in the continuity equation for defects corresponds to the formation and transformation of defects. This formation and transformation of defects along with generation and recombination process for charge carriers is represented as a defect chemical reaction. Hence, we call our model as reaction-drift-diffusion modeling of solar cell. The drift-diffusion equations for defects require the diffusion constants and activation energies of the defect to be known and the defect chemical reaction require reaction rate constants to be known. These parameters are calculated using Density Functional Theory (DFT).
Since the main goal of our research work is to study short time metastability and long-time reliability concerns of cadmium telluride (CdTe) photovoltaics, special attention has been placed in the design of the solver to be able to produce results ranging from ns to hours/days/years. The solver gives us possibilities to explicitly account for all transient effects with free carriers (simulation of time resolved photoluminescence) and defects (simulation of performance instabilities, IV hysteresis etc). Various generation recombination processes can be represented as additional defect chemical reactions. Moreover, the Unified Solver supports accurate treatment of interfaces and grain boundaries that are crucial for the explanation of the operation of CdTe and other chalcogenide PV technologies.
The Unified Solver is benchmarked against Silvaco simulations of a homojunction and heterojunction solar cell. Excellent agreement is observed between the Unified Solver and Silvaco results for the key solar-cell parameters (short-circuit current and open-circuit voltage). Next, the Unified Solver is employed in constant temperature 2D simulation of chlorine diffusion annealing in a cadmium telluride (CdTe) system under insulating boundary conditions (isolated system). Chlorine is introduced in the system as a neutral interstitial at a half corner of the (1um×1.2um) structure. The concentration of chlorine interstitial is 1e16 cm^-3 and the system is kept at a temperature of 750K. The sample is annealed for 240s using the test case of chlorine defect reactions. The time evolution of chlorine substitutional defect (Cl_A⁺) is presented. Emulation of process temperature profiles is also presented in the talk.

For solar technology to compete with traditional energy sources, a continued decrease in photovoltaic (PV) energy generation cost is needed. Next-generation PV technology can achieve this by utilizing earth-abundant materials and low-cost processing techniques. Chalcogenide compounds are of particular interest in this area due to their prolific use in energy applications and amenability to high-throughput, low-cost processing techniques. Chalcogenides semiconductors have shown particular success in PV applications with materials such as Cu(In,Ga)(S,Se)₂ (CIGS), Cu₂ZnSn(S,Se)₄ (CZTS), and CdTe. However, developing new materials and processing techniques require extensive research efforts to achieve the required high-performance goals. Complex device processing and non-ideal optoelectronic properties associated with early stage materials often prolong material development.

In this work, our approach to developing next-generation semiconductors for PV is shown. We utilize low-cost, scalable solution-based processing techniques which are amenable to high-throughput optimization and a variety of chalcogenide precursors. Furthermore, we utilize advanced optoelectronic and structural characterization to guide the material development process; this provides rapid feedback for accelerated material development through accurate screening of early-stage materials for relevant optoelectronic properties and optimal synthesis parameters. In particular, we focus on the extraction of optoelectronic properties relevant for device performance – without the need for device fabrication – through optical techniques. Previous work using these techniques for developing CZTS, CIGS, and perovskites will be shown, with application to new materials currently in development. Ultimately, the successful development of new semiconductor materials requires a cross-disciplinary approach linking fundamental material properties and processing to the device-relevant optoelectronic properties.

Sb₂(S,Se)₃, is becoming a relevant thin film chalcogenide semiconductor with different technological applications such as: superconductivity, electronic components, electrode for sodium-ion batteries, photodetectors and as emerging photovoltaic absorber. In particular, and for this last application, the material has shown remarkable improvements in the last few years, demonstrating solar cells in superstrate configuration with power conversion efficiencies reaching 7.6%. In fact, and similarly to CdTe, most of the devices reported in the literature so far have been prepared using this configuration. This has opened interesting perspectives for their use in solar energy conversion applications, also taking into account the 1D crystalline organization of the material, with in principle benign grain boundaries and anisotropic conduction properties. Additionally Sb₂(S,Se)₃ has shown a high flexibility degree in terms of substrate type, due to the relatively low synthesis temperatures required for optimal high quality polycrystalline growth (300-400 C), allowing deposition onto polymeric, steel, ceramic and TCO/glass substrates. This versatility makes this compound very promising for ubiquitous applications such as building integrated photovoltaics (BIPV) (flexible, bifacial, and semi-transparent), wearables, or autonomous IOT applications among others.
In this work we present a systematic optimization study of the synthesis of Sb₂Se₃ thin films using substrate configuration solar cells, by a sequential process based on reactive annealing under Se atmosphere of thermally evaporated Sb layer precursors. The study is centered in the analysis of Sb precursor thickness and reactive thermal annealing conditions (annealing temperature, time, and pressure) on the compositional, structural and morphological properties of the layers. We observe the formation of continuous layers with large and homogeneous crystals, reporting for first time a weak photoluminescence signal close to 1.3eV in agreement with the band gap value obtained by IQE, and a systematic vibrational characterization under resonant and non- resonant Raman conditions that allows report 15 peaks of the 30 expected.
After a first optimization on Mo coated soda lime glass substrates, we report a promising power conversion efficiency of 5.3% in substrate configuration with a V_OC of 403 mV (the highest value reported for this configuration to the best of our knowledge), close to the 7.6% certified world record in superstrate one.
Additionally the study is complemented with a wide characterization of the fundamental properties of Sb₂Se₃ layers and devices using morphological and physic-chemical characterization (Photoluminescence, SEM, XRF, XRD), and with a complete analysis of the impact of the absorber stoichiometry under different regimes (Se-rich, Se-poor, Sb-rich and Sb-poor conditions). All this will be correlated with the optoelectronic characterization (JV, IQE, CV) of the solar cells. Finally, the main challenges to develop Sb₂Se₃ type solar cells in substrate configuration will be reviewed in the frame of the obtained results.

The non-cubic antimony chalcogenides, i.e., Sb₂Se₃, formed by quasi-one-dimensional ribbons can enhance light absorption and carrier transport by tuning the ribbon direction using the close-space sublimation (CSS) deposition. The power conversion efficiency (PCE) ~7% is found to be associated with the ribbon direction, which was investigated with theoretical calculation and experimental measurement in the Sb₂Se₃ films and devices. The substrate temperature and film thickness are critical for the fine-tuning of ribbon orientations during the CSS deposition. Our results show that [211]-preferred orientation leads to the minimum series resistance and highest light absorbance in the device. The device reliability measurement and in-depth elemental profiling analysis suggest that the interdiffusion between window layer and absorber layer dominate the degradation mechanism. This observation demonstrates that Sb₂Se₃-like quasi-one-dimensional materials with van der Waals boundaries can achieve scalable production at low cost and hold great potential for next-generation solar cell.

Tandem cells are considered as the next generation solar cell to overpass the intrinsic efficiency limitations of single junction solar cells, fixed by the Schockley Queisser limit below 33% and more probably below 30%. Silicon solar cells are already approaching the theoretical limit of about 29%, and represent 95 % of the present PV Market. Tandem solar cells on silicon would make possible to search for 43% theoretical limit, making practical module efficiencies of more than 30%. Combinations with hybrid perovskite and III-V top cells are presently focusing a lot of attention, but an alternative is to develop top cells based on chalcogenide materials, which technologies are well proved on the market (CdTe and CIGS). In the presentation we will report on researches carried out on silicon/CIGS solar cells. Based on the experience on single junction CIGS low band solar cells (1.15 eV) at about 20% efficiency, wide gap CIGS solar cells (1,6-1,8 aV) are grown on single cristal silicon substrates by coevaporation, with lattice parameters matching with silicon, thanks to compositional adjustments (Ga to In, S to Se). In order to give more flexibility in interface electrical properties engineering, the silicon substrate is functionnalized by a thin III-V buffer layer, based on GaInAlAsP lattice matched alloys¹. Epitaxial growth of CIGS has been successfully demonstrated and first devices have been elaborated to address the formation of high efficiency top cells alone, using Si/III-V as a selective back contact. Further structural, chemical and luminescence characterization will be reported.
Ref :
¹ D. Lincot et al., Proceedings of EUPVSEC 2018

A large gap still remains between the achieved conversion efficiencies and the Shockley-Queisser limit for Cu(In,Ga)Se₂ (CIGS) based solar cells. Some of the current limitations such as parasitic absorption losses are known and solutions are being developed. For some other limitations the origin is not well understood yet, but is required for knowledge based improvements. Thorough materials and device characterization of highly efficient solar cells can help to understand the origins of the remaining losses.
We used advanced materials and device characterization including high resolution transmission electron microscopy, time-of-flight secondary ion mass spectrometry, and time resolved photoluminescence (TRPL) combined with multidimensional device modelling to confine the origins of the remaining losses. We will summarize our recent findings on optical, compositional, structural and electronic properties in multi-stage co-evaporated CIGS layers and at interfaces to adjacent layers. In particular, we will describe how surface recombination velocities, minority carrier lifetime and mobility can be extracted from TRPL measurements and what we can learn from these findings. Further, we will present the presence of undesired compositional non-uniformities & voids and discuss their formation mechanism and role on device performance. Combined with recent findings obtained on narrow band gap CIS absorbers our results indicate possible inherent limitations originating from current fabrication methods and device architecture, which leads to new strategies how the actual CIGS solar cell efficiency can be pushed closer to the Shockley-Queisser value.

Recently, 22.8% efficient Cu(In,Ga)Se₂ (CIGS) thin film solar cell has been reported, demonstrating its great potential as a competitor to silicon solar cells. Adopting solution method to prepare highly efficient CIGS light absorbing materials is crucial for reducing the cost of CIGS solar cell fabrication and achieving large-scale production. We have previously reported 10.3% copper indium selenide (CIS) solar cell with the absorber fabricated from single DMF molecular precursor solution, demonstrating the great potential of DMF as relatively environmental benign solvent for solution processed thin film solar cells. The metallic element ratio (Cu/In or Cu/(In+Ga)) is the key factor to affect the quality of absorber materials. Here, we have systematically investigated the effect of the Cu/In ratio (0.85 to 1.2, from Cu-poor to Cu-rich) in the precursor solution on the CIS device performance. We found that solar cell efficiency increases with the Cu/In ratio from 0.85 to 1.05 and then decreases from 1.05 to 1.2. The best device was achieved from a ratio of 1.05 with a peak power conversion efficiency (PCE) of 12.20%, a short circuit current density (J_sc) of 36.12 mA/cm², an open circuit voltage (V_oc) of 0.499 V, and a fill factor (FF) of 67.57%. Further, by gallium alloying, CIGS solar cell with a PCE of 13.6% has been fabricated with a V_oc of 0.600 V, J_sc of 34.78 mA/cm² and FF of 65.17% under similar Cu-rich (Cu/(In+Ga)=1.05, Ga/(In+Ga)=0.2) condition. Our results for the first time demonstrate highly efficient CIS/CIGS solar cells can be achieved from absorbers grown under Cu-rich conditions. Characterization of the absorber materials and solar cell devices are undergoing to understand the mechanism behind the phenomenon observed.

Cadmium telluride (CdTe) thin film solar cells are renowned photovoltaic materials for its high absorption coefficient, suitable band gap and low manufacturing cost. The fabrication of highly efficient devices includes reduced intrinsic defects and improved interfaces at front and back contacts. The back-contact processing includes the formation or deposition of tellurium (Te) layer to make an ohmic contact with a metal. Here, we report wet-chemical treatments of CdTe using various iodine compounds and sodium tetrafluoroborate (NaBF₄). The iodine compounds tested are elemental iodine (I₂), ammonium iodide (NH₄I), mixture of iodine and ammonium iodide (I^-/I₃^-) and formamidine iodide (FAI). The treated surfaces were studied using Raman spectroscopy, X-ray diffraction (XRD), scanning electron microscopy (SEM), and energy dispersive X-ray spectroscopy (EDS). The treatment with iodine compounds produced Te rich layer on the back surface based on Raman, XRD patterns and changed surface morphology. The fabrication of the devices after these treatments, except I₂-propanol, improved the open-circuit voltage (V_OC) and fill factor (FF) of the devices. The I₂-propanol etching produced tellurium iodide (TeI) in β-phase on the surface which reduced device performance. The photoconversion efficiency of CdTe devices after treatment is up to 14.0% (V_OC 841 mV and FF 78%) while untreated devices have an efficiency of about 12.7% (V_OC 814 mV and FF 73%). In case of BF₄^- treated samples, it did not produce Te rich layer, but it seems like BF₄^- ions are passivating CdTe surface with enhanced V_OC and FF compared to standard devices.

CdTe solar technology is producing electricity at costs less than conventional fuels in many regions. Yet despite its commercial maturity, there is still headroom to increase performance significantly by addressing fundamental material challenges including compensation, hole density, carrier lifetime, and interfaces. This presentation will describe ongoing work to understand and overcome these challenges.

Last advances on kesterite (Cu₂ZnSn(S,Se)₄) corroborate the importance of doping and alloying strategies, not only for tuning the properties of this family of materials, but also for achieving higher efficiencies and broadening their possible range of application. Alkaline (Li, Na, K) and Ge doping; as well as Ag, Cd and Ge alloying are demonstrating a high potential by enhancing different aspects of these materials. In particular, the partial or total substitution of Sn by Ge in kesterites is an emerging approach aiming to improve their efficiency and broaden their band gap up to 2.2 eV with direct application in advanced solar cell tandem concepts, or semi-transparent photovoltaics for building integrated photovoltaics (BIPV) among others. Many studies have proved that the use of Ge as doping or alloying element influences drastically the overall parameters of Sn-based kesterite devices improving carrier lifetime, boosting the V_OC, and increasing the conversion efficiency in solar cell devices. The highest efficiency reported until now for pure Cu₂ZnGeSe₄ (CZGSe) compound is a promising 7.6% achieved with an annealing under H₂Se environment and employing very similar process parameters to those typically used for the pure Sn alloy. This work aims to study and optimize the main parameters affecting the synthesis of CZGSe comparing their main characteristics with the pure Sn compound.
For this purpose, Cu₂ZnGeSe₄ thin films were synthesized onto Glass/Mo substrates by a sequential process, based on the sputtering of Cu, Zn and Ge metallic layers, followed by a reactive annealing under a Se+Ge atmosphere inside a graphite box. Parameters like metallic stack order; annealing temperature (450-550 C), pressure (1-1000 mbar), routine (one or two-step annealing) and time; as well as the composition and of the use of chemical etchings were investigated. Additionally, a combinatorial sample with compositional gradients was prepared covering the full Cu/Zn, Cu/Ge and Zn/Ge range of interest, in order to correlate composition with the formation of secondary phases and defects. All the samples were fully characterized using a comp