Symposium EN07—Thin-Film Compound Semiconductor Photovoltaics

Kesterite or CZTSSe (Cu₂ZnSnS_xSe_{4 - x}) solar cells has been steadily emerging third-generation thin-film solar cells as cost-effective CZTSSe solar cells have shown <12.6 % efficiency. In particular, we reported that CZTSe Ge bi-layers devices have shown greatly improved performance with the enhanced grain and grain boundary properties. CZTSe: Ge bi-layers devices consist of several different functional parts such as carrier generation and carrier separation, associated with electric field in the depletion region. The current-voltage (I-V) measurements can provide the device performance of photovoltaic devices (DC characteristics), but the electrical property associated with each interface junction of CZTSe: Ge bi-layers devices cannot be readily accessed with DC characteristics. Conversely, AC characteristics (modulus spectroscopy) can provide each interface junction's properties in the frequency domain as a certain range of frequency separately reveals them.
In this study, we fabricated and analyzed CZTSe: Ge bi-layers devices to accurately evaluate the electrical structure at hetero and back contact junctions by modulus spectroscopic characteristics. Moreover, we studied the dependence of light and voltage biases with respect to impedance and modulus behaviors of hetero and back contact junctions. We used an equivalent circuit model for CZTSe: Ge bi-layers devices that describes the junction modulus and impedance characteristics. This model is utilized to extract the most appropriate device parameters under various voltage bias conditions, which agrees with Sentaurus TCAD numerical simulation results.
Fabricated devices are analyzed with an equivalent circuit model, and LEVM/LEMW software resolved two different parallel circuit branch components. Modeling with LEVM/LEMW software accurately fits our model to modulus spectroscopy combined with impedance spectra in the given testing frequency domain. Using TCAD numerical simulations, we examine a variety of device models with a set of built-in potentials for appropriate parallel circuits with two capacitance and two resistance (R²=0.99). Our study shows that heterojunction built-in potential is 801 meV, which is ~10 % smaller than a theoretical value, 894 meV. Conversely, the built-in potential of the back contact junction is 428 meV. The total built-in potential is 373 meV. All in all, modulus data combined with impedance spectra agree well with modeling and numerical simulation results.

Introduction to current density-voltage (J-V) and quantum efficiency (QE) measurements including essentials to measurement setup and data analysis. Expanded application of J-V and QE data to photovoltaic performance loss quantification: how the combination of meaningful measurements and fundamental photovoltaic concepts such as the Shockley-Queisser limit and non-ideal diode equation can be used to understand dominant loss mechanisms in solar cells. Basic outline of SCAPS 1-D modeling as a companion analysis tool to supplement interpretation of device operation.

Introduction into tandem and multijunction devices: boosting solar cell efficiencies, tandems with Perovskites, and 2T vs 4T. Semitransparent electrodes and recombination layers for tandem fabrication. Challenges in analyzing electrical J-V measurements of so-called two terminal (2T) devices and how to properly measure EQE with bias illumination and voltage.

●Theoretical background on semiconductor photoluminescence
●Basic photoluminescence measurement setup
●Spectrally resolved photoluminescence: impact of absorptance spectrum, temperature, injection level, and defect states
●Time-resolved photoluminescence: measurement technique, competing emission dynamics within samples, potential and pitfalls
●Quantitative photoluminescence: accessing the quasi-Fermi-level splitting from a rapid nondestructive measurement
Photoluminescence playbook case study: combining spectrally resolved, quantitative, and time-resolved photoluminescence to understand to origin of voltage losses in cadmium selenium telluride solar cells.

Introduction into Raman spectroscopy: regular, resonance, and polarization measurements. How to conduct proper Raman measurements, the most common mistakes. Application of Raman spectroscopy for structural and crystal quality assessments, as well as defect and secondary phase identification. Building Raman-based methodologies for defect identification and correlation with optoelectronic properties through combinatorial studies.

Application of semiconductor thin film materials in photovoltaic devices requires optimized optoelectronic properties and thus understanding of the underlying physical mechanisms and limits governing carrier transport and recombination. Although material parameters such as doping density, recombination lifetimes and mobilities belong to the most basic properties eventually defining the solar cell operation, they can be quite elusive or ambiguous in their characterization. Here, chalcogenides (e.g. chalcopyrite and kesterite) and halide perovskites share some similarities, but also a number of differences. Considerable discussion prevails about the role of interface versus bulk recombination in devices, the values of carrier mobilities, doping levels and minority carrier lifetimes. Combining different characterization techniques such as time-resolved luminescence, photoluminescence quantum yield as well as pump-probe terahertz spectroscopy we attempt to draw a relatively consistent picture of material properties and device function, exposing current limits and paths to improvement.

The fabrication of cost-effective multi-junction (MJ) device architectures is presently considered the strategy to increase the Power Conversion Efficiency (PCE) of the solar cells in order to lower the levelized cost of energy (LCOE), decreasing therefore the price of an installed photovoltaic (PV) system. The outstanding advancement of metal halide perovskite absorber (PSC) technologies in the last years motivates the PV community to investigate tandem hybrid architectures, consisting of a perovskite based top cell with a commercial bottom PV device such as crystalline silicon (c-Si) or copper indium gallium selenide (CIGS). The combination with the latter, in particular, offers unique advantages: i) compatibility with lightweight flexible plastic or metal foils as substrates, in view of a roll to roll (R2R) manufacturing line, further reducing the manufacturing and the installation costs, as well as the environmental footprint of the KWh generated; ii) large scale integration of PV devices in, amongst others, buildings (BIPV and vehicles (VIPV).
This contribution will provide an overview of the activities within the Solliance consortium on 4 terminal tandem and preliminary work on 2 terminal architectures, on rigid, as well as flexible substrates. In order to increase the overall PCE of the tandem hybrid configuration, the optimization of the sub-cells’ performance with respect to their spectral responses is required. Several combinations of suitable complementary band gaps of the two absorbers are explored, as well as the optimization of the near infrared transparency of the top perovskite cell by tuning the free carrier absorption of the TCO adopted. Furthermore, in the case of the 2-terminal architecture, the development of an efficient interconnected layer among the two sub-cells is utmost crucial to provide the monolithic series interconnection and, at the same time, to minimize the parasitic absorption.
Optimized TCO, transport layers and anti-reflective properties lead to a near-infrared transmittance of about 90% (an average taken from the wavelength range of 700-800 nm to 1200 nm) on glass or plastic based top PSC optimizing therefore the response of the filtered CIGS. As a result, coupling a semitransparent glass-based PSC and a flexible CIGS we calculate a 4T tandem PCE of 25.2% on small area devices (<0.1cm²). When using a flexible semi-transparent perovskite solar cell, a record-breaking power conversion efficiency of 23.0% is calculated. We also evaluated the PSC-CIGS combination on a larger area (> 80 cm²), currently adopting a no-optimized (in the NIR region) large area perovskite module. We calculated already a 1.9% absolute increase in the efficiency with respect to the single CIGS device, with a further improvement that can be achieved by improving the light management and fine tuning the perovskite band gap. Whereas the 4T architecture offers already a significant increment in performance with respect to the single junction devices, the development of the 2T tandem PSC-CIGS exhibits more challenges. As observed by the first results, the efficiency of the flexible 2T tandem PSC-CIGS, still lags behind the power conversion efficiency of the single CIGS. New upscalable perovskite formulations as well as the optimization of the recombination layer will also be addressed as crucial topics to boost the overall performance.

Great progress in perovskite photovoltaics has been enabled by the excellent intrinsic material properties of lead-halide perovskites. These relatively newly-available materials are already opening paths to higher performance in mass-market PV technologies, including tandem PV modules. Nonetheless, reaching competitive levelized energy costs will hinge on achieving comparable robustness to existing silicon-based technology. This is a high and rising benchmark, which has not yet been approached by halide perovskite technologies. This makes it less likely that investors will support the upscaling and deployment of perovskite-based solar technology.

Chalcogenide perovskites, in which halide anions are replaced by chalcogenides (O, S, Se) could offer a new path forward. Examples include BaZrS₃ and SrHfS₃, among several others. The solar cell-relevant properties of these materials are projected to be comparable to halide perovskites. More critically though, their chemistry provides them with excellent stability under conditions for which halide perovskites would rapidly degrade, e.g. in water, humid air or at high temperatures.

The major challenge with chalcogenide perovskites has been to synthesise them in thin film form with good quality. Therefore, despite impressive measurements from bulk materials, it has not been possible to validate their application in thin film solar cells. There are several reasons for the synthetic difficulties, connecting to intrinsic material properties as well as to the use of inappropriate synthesis methods. To date, we still have incomplete understanding of the growth processes of these materials.

In this presentation, we briefly summarise the potential of the chalcogenide perovskites based on prior literature. Then, our own progress toward thin films suitable for solar-cell integration will be presented. A number of sputter-based growth routes to BaZrS₃ have been investigated and evaluated using combinatorial approaches. Materials have been characterised using XRD, STEM-EDS, SEM/TEM imaging, XAFS, Photoluminescence mapping and Ion-beam techniques including RBS and ERDA, to construct a picture of the growth sequence and final compositional, microstructural and optoelectronic characteristics of the films. This is a means to establish the first “cookbook” for chalcogenide perovskite films suitable for solar cells, i.e. the optimal growth conditions for high quality material. Film formation will be discussed in the context of thermodynamic and kinetic aspects, applicable to any type of growth method. The aim of this talk is to promote and support research into these highly interesting materials in other labs, and stimulate collaboration on this exciting topic.

Spiro-OMeTAD, CuSCN and PTAA are attractive hole transport materials (HTM) in conventional n-i-p architectures for perovskite cells¹. In this work, we combine these three HTMs on semi-transparent (ST) solar cells with different sputtered electrodes made of In₂O₃-SnO₂ (ITO) or In₂O₃-ZnO (IZO) to study their impact on the optical and electrical properties of the cells.

First, thanks to an optical model based on the Transfer Matrix Method and careful material characterizations^2,3, a good agreement has been obtained between the transmission, reflection and absorption spectra of the complete structure of the ST cell with Spiro and ITO. This model allows to simulate the efficiency of a tandem cell with a silicon bottom cell, to identify optical losses and to optimize the thickness and properties of the each layer. It is shown that the Spiro and ITO layers have a non-negligible parasitic infrared absorption in the IR, resulting in a lower efficiency for the filtered silicon cell. The gain in efficiency enabled by the use of PTAA, CuSCN and IZO, which are almost transparent in the IR region, is assessed.

Second, triple cation perovskite solar cells with these three HTMs are synthetized and characterized in detail using scanning electron microscopy, ellipsometry spectroscopy, glow discharge emission optical spectroscopy, X-ray diffraction and transmission spectroscopy. Regarding the electrical properties, the Spiro is a layer known for its sensitivity, and the poor reproducibility obtained on ST cells is often associated with degradation due to energetic sputtering⁴. Investigations using CuSCN as HTM showed a slightly better reproducibility than Spiro but strong light soaking effects also appeared, and the resulting efficiencies are low. The replacement of the Spiro by PTAA has improved the reproducibility of cells, while maintaining good performances. A 17.4% efficiency perovskite semi-transparent cell with PTAA was obtained and which is significantly above our previous results with Spiro (16.4%)⁵.

After electrode sputtering deposition, a characteristic S-shape appears in IV characteristics. By regularly following the cells over time while keeping them under vacuum atmosphere and in the dark, IV curves recovered until the S-shape disappeared completely. For example, a cell with Spiro and ITO have 8.7% efficiency and 43% Fill Factor (FF) the first day and reach 16.8% with FF=70% after 66 days. When the Spiro is replaced by PTAA, the S-shape is less pronounced and the recovery time markedly reduced to 15 days. These effects are not visible on cells with CuSCN.

Lee et al.⁶ obtained a similar S-shaped curve on perovskite cell with a Spiro/silver electrode, with a short recovery time of about one day. A local reaction between the dopant Li-SIF and the silver could cause the deoxygenation of the Spiro at the surface, which would create a temporary electrical barrier. It is interesting to note that this dopant is common to PTAA and Spiro but not to CuSCN. Preliminary results obtained by time resolved photoluminescence (TRPL) shows a decrease of the PL signal over time. Experiments are underway to determine if the cause of this decrease is due to improved extraction at the HTM/TCO barrier as suggested by Lee et al.⁶, or if it comes from a reduction in radiative recombination related to defect healing. We will discuss the reaction mechanisms for these three HTM.

[1] Z. Shariatinia, Renew. Sustain. Energy Rev., 119, 109608, 2020
[2] E. Raoult et al., 36th Eur. Photovolt. Sol. Energy Conf. Exhib. 757, 2019
[3] E. Raoult et al., SPIE: Physics, Simulation, and Photonic Engineering of Photovoltaic Devices IX, 2020
[4] H. Kanda et al., J. Phys. Chem. C, 120, 50, 28441, 2016
[5] F. J. Ramos et al., Sci. Rep., 8, 1, 2018
[6] D. G. Lee et al.,, ACS Appl. Mater. Interfaces, 11, 51, 48497, 2019

Tandem photovoltaic (PV) devices offer great advantages over individual photovoltaic cells including the potential to reach power conversion efficiencies (PCEs) greater than that of the composing individual cells. The current record PCE for a PV tandem cell is over 29%, set by a perovskite/Si tandem. Perovskite solar cells (PSCs) are ideal for tandem devices because they have proven to be highly efficient and the bandgap can be tuned. However, PSCs are extremely sensitive to air and light. Another great candidate for tandem devices are chalcopyrite thin-film solar cells. Chalcopyrite cells also have tunable bandgaps and high efficiencies. The manufacturing process for chalcopyrite cells includes parameters that would degrade the performance of the bottom cell in a monolithic tandem device and the architecture of the cell prohibits a sufficient amount of energy below the bandgap from reaching the bottom cell. Thus, we present a strategy for fabricating a semi-monolithic chalcopyrite/perovskite tandem device by employing a transparent polymer impregnated with conductive particles (TPCP).
The TPCP was designed for electrically coupling solar cells into tandem devices. The TPCP can be employed in three ways: (i) fully-bonded, (ii) semi-bonded , and (iii) freestanding. When fully-bonded, the TPCP permanently bonds two surface allowing for passage of light and electrical current. TPCPs have been used to fabricate a CuGaSe₂/Si tandem device with an open circuit voltage of 1.04 V and PCE of 2.45%. Fully-bonded TPCPs were used to exfoliate the CuGaSe₂ cell at the MoSe₂ layer with an FTO-coated glass handle. The exfoliated CuGaSe₂ cell was then fully bonded with the TPCP to a Si bottom cell. The semi-bonded TPCP passes light and electrical current while also providing the protective features of a polymer over a surface. The semi-bonded TPCP has successfully been employed as an encapsulation technique for preserving PSCs. The TPCP encapsulation did not affect the performance of the PSC. An encapsulated PSC exposed to air and light for 14 days did not show any visible signs of degradation. The freestanding TPCP enables the coupling of PV cells without permanently bonding the cells. Ultraviolet-visible spectroscopy measured over 90% transmittance across the visible spectrum. The resistance of the TPCP measured by linear sweep voltammetry yielded a series resistance to the order of 10^-1 Ω●cm².
Based on the successful fabrication of the CuGaSe₂/Si tandem device and encapsulation of the PSC, we will investigate the fabrication of a chalcopyrite/perovskite tandem device composing of a wide bandgap chalcopyrite top cell and a narrow bandgap PSC. Additionally, the application of the freestanding TPCP to fabricate a tunable tandem device with exchangeable cells will be attempted.

There are good reasons to abandon the usual p-type absorbers in CdTe-based solar cells in favour of n-type ones. Despite recent advances in CdTe PV (e.g. eliminating CdS, the use of CdSeTe and intentional p-type doping using CdTe:As), the basic design of ‘superstrate’ CdTe-based devices remains unchanged since the 1960s: light enters through the glass/transparent conducting electrode/window layer and generation takes place in a shallow homojunction in the CdTe-based absorber layer - which is otherwise p-type. These cells have limited V_oc. The use of n-type CdTe absorbers sidesteps two fundamental physical limitations of p-CdTe and replaces them with technological challenges: For p-type CdTe it is well known that formation of Ohmic contacts requires that a barrier be overcome (this being the value of the electron affinity plus a large fraction of the bandgap). Also, wide- and medium-gap semiconductors cannot stably be doped p-type at a high density due to self-compensation limitations imposed by native defects. Such factors limit the V_oc achievable from CdTe-based devices. For n-type CdTe, doping with indium can achieve close to 100% activation and allows higher stable doping densities to be achieved than with CdTe:As for example. Ohmic contacts to it may be formed without issue. Both the n-doping and Ohmic contacts are well-known for single crystal n-CdTe.

Hence the challenge for n-type devices, lies elsewhere, notably in the choice of partner layers and their electrodes, the selection of a device architecture and the control of n-doping in the context of thin film growth and processing. There are essentially four choices of device architecture: A) ‘substrate’ devices in which the light enters through the glass/TCO combination and then first encounters either a) n-CdTe or b) a p-type partner layer or B) ‘superstrate’ devices in which the light enters through a transparent electrode and then first encounters either a) n-CdTe or b) its p-type partner layer. To sum them up in other words: there is either an n- or a p-type window layer. We ruled out using p-type TCOs immediately - since they have low conductivity - and instead we considered other options. In a preliminary simulation study (using SCAPS) we identified a range of materials choices and architectures for n-CdTe devices that gave V_oc ≥ 1V. This indicated that experimentation would be worthwhile.

For the CdTe:In doping we have been trialling post-growth introduction of indium into CSS-grown CdTe using elemental indium (with and without pre- or post-treatment using MgCl₂), both aqueous and methanolic InCl₃ and also treatments with and without nitric/phosphoric etching. We will report conductivity measurements, transfer length measurements (contacts), SIMS and XPS results. There is some indication that diffusion can be blocked by surface oxides. As an alternative we are trialling intentionally pre-doped CdTe:In source material for use in close space sublimation film growth directly. At the time of writing device work is in progress using a variety of wafer (test structure) and thin film partner layers including organics. This part of the work is at an early stage and we will report on progress at the conference.

The ability to accurately model, understand and predict the behaviour of crystalline defects would constitute a significant step towards improving photovoltaic device efficiencies and semiconductor doping control, accelerating materials discovery and design. In this work, we apply state-of-the-art ab initio techniques - hybrid Density Functional Theory (DFT) including spin-orbit coupling - to accurately model the atomistic behaviour of the cadmium vacancy (V_Cd) in cadmium telluride (CdTe). In doing so, we resolve several longstanding discrepancies in the extensive literature on this subject.

CdTe is a champion thin-film absorber for which defects, through facilitation of non-radiative recombination, significantly impact photovoltaic (PV) performance, contributing to a reduction in efficiency from an ideal (Shockley-Queisser) value of 32% to a current record of 22.1%. Despite over 70 years of experimental and theoretical research, many of the relevant defects in CdTe are still not well understood, with the definitive identification of the atomistic origins of experimentally-observed defect levels remaining elusive.^1–3

In this work, through identification of a tellurium dimer ground-state structure for the neutral Cd vacancy, we obtain a single negative-U defect level for V_Cd at 0.35 eV above the VBM, finally reconciling theoretical predictions with experimental observations. Moreover, we reproduce the polaronic, optical and magnetic behaviour of V_Cd^-1 in excellent agreement with previous Electron Paramagnetic Resonance (EPR) characterisation.⁴

The origins of previous discrepancies between theory and experiment, namely incomplete mapping of the defect potential energy surface (PES) and inherent qualitative errors in lower levels of electronic structure theory, are analysed in detail. Accordingly, this work helps to establish robust procedures for accurate and reliable modelling of defect processes in emerging materials, informing future investigations and enabling the acceleration of materials discovery and design procedures.

1 A. Lindström, S. Mirbt, B. Sanyal and M. Klintenberg, J. Phys. D: Appl. Phys., 2015, 49, 035101.
2 J.-H. Yang, W.-J. Yin, J.-S. Park, J. Ma and S.-H. Wei, Semicond. Sci. Technol., 2016, 31, 083002.
3 A. Shepidchenko, B. Sanyal, M. Klintenberg and S. Mirbt, Scientific Reports, 2015, 5, 1–6.
4 P. Emanuelsson, P. Omling, B. K. Meyer, M. Wienecke and M. Schenk, Phys. Rev. B, 1993, 47, 15578–15580.

In this work we investigate the use of pre-doped n-type CdTe source material to fabricate PV devices via close-space sublimation (CSS). The more readily dopable n-type CdTe offers the potential advantages of easier Ohmic contacting and higher V_oc than are achieved with conventional p-type CdTe material.
Metallic indium was incorporated into bulk CdTe source material (Sigma Aldrich 6N) in quartz capsules sealed under vacuum and heated above the melting point of 1091°C for 3 days. In order to achieve doping in the range 10¹⁸ – 10¹⁶ cm^-3 successive dilution steps were required. The resulting material was qualified by chemical analysis using EDX (to determine homogeneity for the highly doped compositions) and by ICP-OES to determine the absolute indium concentrations in the bulk material more generally. XRD was used to confirm the phase purity.
Thin films of CdTe:In deposited using CSS onto NSG TEC-AB substrates (highly resistive SnO₂ on glass) were examined. Incorporation of the indium during the CSS deposition was investigated by quantitative SIMS of the films, and by scraping off the material for ICP-OES analysis. Growth on the resistive TEC-AB substrates also allowed for in-plane conductivity and contact resistance measurements using the transmission line method (TLM). C-V measurements were carried out on glass/Mo/CdTe:In/In structures and were used to give estimates of the carrier concentration. The carrier type was confirmed using the hot probe method.
In addition to the characterisation of the films, we also investigated the use of n-CdTe in test-structures and devices based on p-type semiconductor wafers. Simple evaluations of the band line-ups between n-CdTe and p-type wafers of Si, GaAs, and InP, using the Anderson model, indicated favourable junction properties. We therefore grew these structures by CSS transport of CdTe:In onto (001) oriented wafers. Initial deposition tests indicated that the films adhered well to their substrates and were crystalline in appearance. The junction properties and PV performances of these structures will be reported.

The motivation of this work is to deploy intentionally doped n-type CdTe in place of the more conventional p-type: this offers the possibility of high, stable doping in the absorber layer which has the potential to give high V_oc in photovoltaic devices. In this work we explore the use of post-growth doping of CSS-grown CdTe films, both to investigate the feasibility of doping and in preliminary devices.

Several approaches to post-growth doping with indium were trialled, including in-diffusion of evaporated indium metal, and the annealing of films that had been sprayed with InCl₃ in either aqueous or methanolic solutions. Annealing was conducted either in air or under flowing nitrogen.

The surface chemistry of the treated samples was investigated using XPS. The instrument used was a Kratos Axis Ultra-DLD photoelectron spectrometer with a monochromatic AlKα x-ray source. The elements/peaks screened for were Cd3d, Te3d, In3d, Cl2p and O1s. The In MNN Auger line was also measured since it exhibits larger chemical shifts than the In3d XPS line. A pass energy of 20 eV was used for the XPS lines and 80 eV for the Auger.

Chemical incorporation of the indium in the CdTe films was evaluated by quantitative SIMS. Implantation standards were created using single crystal THM-grown B surfaces of CdTe (Japan Energy) with xxx-In and yyy-Cl. SIMS measurements were done by Loughborough Surface Analysis using a Cameca IMS 7F-auto SIMS instrument using O²⁺ and Cs⁺ as the primary ion species at an energy of 10 keV. Profiles were obtained for Cd, Te, In, Cl, O and Sn. It was found that indium could be incorporated throughout an approx. 5 µm thick film at a density of 6 x 10¹⁸ cm^-3. The conductivity of the films and contact resistances were measured using the transmission line method (TLM) and the hot probe method was used to verify the carrier type.

A variety of ‘superstrate’ device structures were considered, for example glass/FTO/n-CdTe/ZnTe/P3HT/Au and SCAPS simulations indicated that many were capable of achieving V_ocs of > 1 V in principle. Preliminary devices were fabricated on NSG TEC15 FTO/glass. ZnTe was evaporated from a powder source while the CSS-grown CdTe films were post-growth doped using the indium metal route. Device and diode characterisation results will be presented.

Due to the low material and manufacturing costs of polycrystalline CdTe-based solar cells, it is the leading thin-film photovoltaic technology. Magnesium zinc oxide (Mg_xZn_1-xO) has proven to be a highly effective emitter layer due to its transparency and ability to tune the conduction band alignment with CdTe. State-of-the-art devices employ graded CdSe_yTe_1-y (CST) layer at the interface, and emerging absorbers employ group V doping with significantly higher carrier concentrations. The goal of this work is to engineer MZO to adapt to the evolving nature of the absorber. Combinatorial libraries of MZO were fabricated by reactive sputtering and characterized by ellipsometry, XRD, UV-visible spectroscopy, 4-point probe, and Hall effect. After screening to identify promising compositions, uniform MZO films are used to optimize device performance. For standard CdTe the optimum Mg content was found to be x~0.15, producing >16% devices. With CST based devices the optimum Mg content increased to x~0.25, producing >19% devices. The as-deposited MZO is highly insulating, so Ga was added to increase carrier concentration. GMZO libraries are fabricated with band gaps spanning 700 mV and whose electrical resistivity may be systematically varied >6 orders of magnitude by controlling the rate of Ga deposition. We plan to report on the efficacy of these promising GMZO libraries to optimize band gap and carrier concentration for both CST and CST:GrV based devices.
<gdiv></gdiv>

Polycrystalline cadmium telluride (CdTe) and cadmium selenide telluride (CdSeTe) photovoltaic devices suffer from a sizable voltage deficit when compared with competing technologies (e.g., crystalline silicon, perovskite, III-V). Record-efficiency devices have not yet broken the 900-mV open-circuit voltage limit, even though, given the wide direct bandgap of CdSeTe absorbers (1.4–1.5 eV depending on the selenium content), a voltage above 1100 mV could be theoretically expected. Yet, a precise understanding of the origin of this voltage deficit and, more importantly, a clear pathway to overcome it are currently lacking.
The historically poor material quality of polycrystalline CdTe absorbers with, until recently, minority carrier lifetimes τ in the nanoseconds at best, has long been held responsible for these poor V_oc performances. To remedy this, extensive work has focused on increasing both the lifetime and the acceptor (i.e. activated dopant) concentration N_A, as the quasi-Fermi-level separation is a direct function of the N_A×τ product. Double heterostructure CdSeTe samples, with both surface passivated with either magnesium zinc oxide (MZO) or Al₂O₃, have reached lifetimes in the hundreds of nanoseconds, indicating that excellent material quality was achievable with CdSeTe absorbers and suggesting that poor surface passivation—i.e., passivation of the absorber front and back interfaces, as opposed to passivation of defects at the grain boundaries—was an important contributor to the limited voltages measured on finished devices. Similarly, arsenic doping of the absorber has led to acceptor concentrations in excess of 10¹⁶ cm^-3.
Yet, these improvements in both lifetime and acceptor concentration have not translated into meaningful increases in V_oc. Two questions thus arise:
1) What limits the voltage of Cd(Se)Te devices?
2) How can we easily and systematically decouple the sources of voltage loss in finished devices?
In this work, we answer these two questions by comparing the thermodynamic voltage limit V_oc,ideal of finished solar cells with their quasi-Fermi-level splitting (QFLS)—also referred to as the internal or implied open-circuit voltage iV_oc (QFLS=q×iV_oc with q the elementary charge)—and their external or terminal voltage V_oc. We calculate V_oc,ideal from the external quantum efficiency (EQE) of the device and access iV_oc through measurement of the device external radiative efficiency (ERE) using a calibrated photoluminescence tool, while V_oc is measured through standard J–V curve tracing.
We show that As-doped cells with Al₂O₃-passivated contacts can exhibit EREs in excess of 1%—surpassing the best crystalline silicon devices and on par with the best perovskite devices—but, concurrently, band tails limit their thermodynamic voltage limit. Nevertheless, our best device achieves an internal voltage iV_oc>980 mV. This is close to 100 mV higher than the external open-circuit voltage (V_oc) of the record-efficiency polycrystalline CdSeTe solar cell. Thus, in these highly luminescent devices, the voltage is limited by the contact selectivity rather than by the material quality of the absorber. We then propose a novel passivated and selective back contact structure—based on an Al₂O₃ passivating layer followed by a p-type boron-doped hydrogenated amorphous silicon (a-Si:H) hole selective layer—with the goal of translating these high iV_oc values into high V_oc values and achieving V_oc>1 .

First Solar is the largest solar manufacturer in Western Hemisphere. While possessing CdTe cell and module efficiency records of 22.1% and 18.2%, respectively, we continue research to further push the boundaries of CdTe cell technology. In particular, we conduct fundamental studies of losses and instabilities within active layers of CdTe cells.

The current Cu-doped CdTe technology suffers from insufficient absorber doping caused by atomic interactions between Cu and Cl. Besides limiting the maximum doping achievable in practical fabrication process, defect reactions between Cu and Cl cause further reduction in absorber doping during field operation, which contributes to slow degradation of Cu-doped devices.

High and stable absorber doping of 1e16 cm^-2 and above could be achieved in chlorinated CdTe films by replacing Cu with arsenic acceptors. However, taking the full advantage of heavily-doped CdTe absorbers requires high quality front interface with small conduction band offset. Our atomic simulations predict strong conduction band offset at CdTe/SnO₂ interface, which prevents achieving high Voc values. Additional performance improvement of arsenic-doped cells could be achieved by further increase in doping activation, which is currently limited by formation of kinetically-stabilized deep donors that not only reduce the built-in potential, but may also limit the bulk lifetime.

Another important aspect of CdTe cell performance is reversible short-range metastabilities present both in Cu- and As-doped devices. Our kinetic device modeling suggests that observed metastable phenomena are caused by the defect reactions at the front interface.

Understanding the nature of recombination and its dependence on defects and interfaces is essential for engineering materials and contacts for higher V_oc and power conversion efficiency in photovoltaic (PV) devices. Time-resolved photoluminescence (TRPL) has conventionally been used to evaluate recombination, but not all materials are strongly emissive or otherwise suitable. Pump – probe spectroscopy presents valuable complementary information, wherein an optical pump pulse photoexcites carriers within the absorber and the transient photoconductivity is probed with a terahertz (THz) pulse. Until now, experimental constraints have limited the use of THz probes to interrogate full device stacks, and measurements were instead made on exfoliated films or films grown on unconventional substrates. However, interfaces are critical to the behavior of photoexcited carriers in solar cells, and the substrates themselves often influence the film growth and bulk properties. Therefore, it is important to probe the behavior of PV absorbers as close as possible to their normal operating conditions.
Here we report on cross-cutting metrology tools based on time-resolved THz spectroscopy (TRTS) and TRPL that enable in operando characterization of carrier dynamics and recombination mechanisms in working PV devices. Through a combination of experiments and modeling, we can obtain key parameters including Shockley-Read-Hall (SRH) lifetime, minority carrier mobility and diffusion length, and interface recombination velocity using non-contact probes. By varying pulsed photoexcitation conditions, the contributions of interface and bulk recombination can be determined as a function of applied bias.
TRTS has been applied to thin film absorbers but has not been used for full devices because the conductive contacts absorb or reflect the THz radiation. Here we create wire-grid patterns using conventional laser scribing to increase the signal-to-noise ratio by at least 5-fold compared to unscribed samples. The TRTS signal corresponds to transient photoconductivity, while TRPL arises from radiative recombination. We can simulate both of these by numerical solution of the continuity equations and Poisson’s equation. Best-fit values of the SRH lifetime and surface and interface recombination velocities are then determined by global fitting of multiple TRTS and TRPL data sets having different photoexcitation conditions.
We have developed and validated this approach using CdTe thin film PVs. For instance, we found a bulk lifetime of 1.6 ns, CdTe/CdS interface recombination velocity of less than 100 cm/s, and back surface recombination velocity of ~2x10³ cm/s for a CdTe device treated with CdCl₂. In contrast, as-deposited films without CdCl₂ treatment had bulk lifetime of only ~50 ps and higher interface and back-surface recombination velocities of ~2x10⁴ cm/s. We are now applying this approach to understand how dynamics change depending on processing conditions designed to enable high doping densities and on choice of buffer material.
By determining the locus and mechanisms of performance-limiting recombination, we can accelerate the development of thin film PVs with high higher Voc and efficiency. While the method has been demonstrated here using CdTe, it is also applicable to perovskites, CIGS, CZTSSe, and other emerging technologies.

Polycrystalline CdTe thin films for solar energy applications are deposited by fast high throughput deposition techniques on glass coated with nanocrystalline transparent conducting oxides. While this approach makes polycrystalline CdTe solar panels a cost leader, the deposition has resulted in small grain sizes on the order of microns for decades.
Grain boundaries (GBs) are active recombination centers that reduce overall carrier lifetime and hence the device performance of CdTe solar cells [1,2]. Therefore, increasing grain size and reducing GB density, especially at the critical junction region, can improve carrier lifetime and solar cell performance [3-5].
The CdTe community relies on CdCl₂ treatments to facilitate grain growth and passivate GBs, increasing the aggregate lifetime from ps to tens of ns. This passivation combined with CdSeTe bandgap engineering has improved CdTe record efficiency to 22.1%. Material challenges are that CdTe grain growth during CdCl₂ anneals is driven by Ostwald ripening, which has historically been limited to the order of the film thickness (i.e. a few microns). Furthermore, CdCl₂ treatments do not completely passivate GBs [4]. For example, correlated GB recombination velocities have been measured to be as high as 10⁵–10⁶ cm/s [6]. Finally, Cl compensation may reduce activation ratios in Group-V doped CdTe devices resulting in microscopic potential fluctuations and lower V_OC [7].
Colossal grain growth (CGG) was recently reported in micron-thick Cd(Se,Te) thin films deposited on glass [8]. The resulting CdTe structures had extremely large ~ 1-mm lateral size grains that were much larger than anything previously reported and a layer structure relevant for CdTe solar cells.
In this work, we will illustrate that the resulting, near single-crystal thin films can be used as a growth template for epitaxy of subsequent CdTe layers by techniques appropriate for making devices. Starting device structures have been made using Corning 7059/75-nm Cd₂SnO₄/60-nm MgZnO substrates upon which a CdSe_0.10Te_0.9 layer was evaporated for CGG. Because of its lower bulk resistivity, Cd₂SnO₄ films could be made thin enough for devices and smooth for CGG of the Cd(Se,Te) layer. After the Cd(Se,Te) was deposited, CdCl₂ treatments with temperatures ranging from 420 °C to 500 °C (with and without oxygen) were carried out before and after the CGG anneal, and subsequent CdTe was deposited epitaxially by using either close-spaced sublimation or vapor transport deposition. Lifetime, device performance, and cathodoluminescence results will be presented for both glass/Cd₂SnO₄/MgZnO/CGG-Cd(Se,Te)/CdTe and glass/Cd₂SnO₄/MgZnO/CGG-Cd(Se,Te) structures at the conference.
References:
1. Moseley, J. et al., IEEE J. Photovolt. 4, 1671–1679 (2014)
2. Moseley J. et al., J. Appl. Phys, 118, 025702 (2015)
3. Kanevce A. et al., J. Appl. Phys, 121, 214506 (2017)
4. Amarasinghe M. et al., Adv. Energy Mater. 2018, 8, 1702666
5. Amarasinghe M. et al., IEEE J. Photovolt. 8 (2), 600-603 (2018)
6. Moseley J. et al., J. Appl. Phys, 124, 113104 (2018)
7. Moseley J. et al., Journal of Applied Physics, 128, 103105 (2020).
8. Albin D. et al., J. Phys Energy

Inhomogeneous microscopic carrier transport is difficult to study but important in many condensed-matter applications. For example, the role of grain boundaries (GBs) in polycrystalline semiconductors has been controversial for 20 years. In cadmium telluride (CdTe) solar cells, electron-beam-induced current (EBIC) measurements consistently demonstrate enhanced current collection along GBs, which is argued as evidence for interpenetrating CdTe p-n current-collection networks critical to high efficiency. Conversely, cathodoluminescence (CL) measurements consistently indicate that GBs are deleterious low-lifetime regions. Here, we apply transport imaging (TI) in conjunction with spatially correlated EBIC, CL, and scanning Kelvin probe force microscopy measurements to understand carrier drift, diffusion, and recombination in polycrystalline CdTe. We simultaneously observe GB potential wells, reduced carrier lifetime at GBs, and seemingly contradictory enhanced GB current collection, then describe their coexistence with microscopic TI and physical arguments. The results provide visualization of inhomogeneous transport that is critical to understanding and engineering polycrystalline solar technology.

Recent developments in cadmium telluride (CdTe)-based thin film technology has demonstrated record efficiencies for both CdTe cells (22.1 %) and modules (18.6 %). Even with this progress, record-efficiency CdTe solar cells have open-circuit voltage VOC ~ 880 mV, which is considerably lower than the theoretical limit of VOC ~ 1.15V. Much of the focus on CdTe has been associated with increasing its net p-type doping concentration in order to increase the device VOC. High p-type doping in CdTe has been challenging, while various studies have shown 1016 - 1019 cm-3 n-type doping in crystalline CdTe using In grown by the Bridgman or molecular beam epitaxy (MBE) methods. Recent studies of In doped crystalline CdTe have shown lifetimes exceeding 2 µs and solar cells fabricated with this material demonstrated VOC greater than 1 V.
In this paper the effect of indium (In) doping on CdTe thin film solar cells will be discussed. CdTe thin films were deposited using the elemental vapor transport (EVT) technique under various Cd/Te gas phase ratios and In vapor concentrations. Solar cells of the superstrate configuration (glass/TCO/CdS/n-CdTe/p-ZnTe/BC) have been fabricated and characterized. There was a correlation between the concetration of In in the vapor phase and net n-type doping for CdTe devices fabricated near Cd/Te stoichiometric ratio. Increasing the amount of In vapors resulted in higher n-type doping in CdTe. From C-V measurements doping levels >1016cm-3 were achieved. Devices were also fabricated at various Cd/Te vapor ratios at a constant In vapor concentration. Films deposited under Cd-rich conditions exhibited higher lifetimes than those deposited under Te-rich conditions. On the other hand, films deposited at under Te-rich conditions exhibited higher n-type doping. To-date n-type CdTe solar cells yielded devices with VOC= 730 mV, JSC= 20 mA/cm2, FF= 62 % and efficiency near 9 %. SCAPS and AMPS solar cell simulation tools were used to better understand and aid the development of n-type CdTe solar cells. Doping, carrier lifetimes, and interface states of the CdTe-absorber and ZnTe window layer are being investigated.

Improvement in polycrystalline CdTe photovoltaics efficiency is primarily limited by the deficit in open-circuit voltage. The deficit can be overcome by reducing the carrier recombination while improving carrier concentration in the film. Traditionally, CdTe-based devices have been doped with copper. However, increasing carrier concentration using copper while reducing recombination is fundamentally limited by various complexes copper may form. To overcome this issue, doping with group V elements has been of interest to the research community for decades and more recently it has been one of the primary focuses of the CdTe research community.
Using sublimation as the deposition method, our research has shown that arsenic can be effectively used to improve carrier concentration while simultaneously reducing recombination. Doping with arsenic is certainly not devoid of challenges as various defects such as the presence of dimers, tetramers, AX-centers, and arsenic clusters in the device pose challenges. However, using an advanced fabrication approach guided by state-of-the-art predictive computational simulation models, we have been able to overcome several of these limitations. Using the method that will be discussed, we have been able to demonstrate carrier lifetimes greater than 2 µs, carrier concentration over 5x10¹⁵ cm^-3, and surface recombination velocity less than 100 cm/s. Following fabrication of films with characteristics suitable for high performing devices, several other factors in the film fabrication process and device structure had to be refined to employ the advantages of doping CdTe-based devices with arsenic as shown by various numerical simulations.
In addition to improvement in device performance, using arsenic as the doping element for CdTe-based photovoltaics may have advantages in terms of longevity of the modules in the field as well. With optimized doping in CdTe with copper, large volumes of modules have been commercially manufactured and installed. However, arsenic doping in polycrystalline films is found to have superior metastability. This is one of the critical factors to be taken into consideration as the elimination of copper may improve projected life of CdTe-based photovoltaic panels from 25 years currently to over 50 years.

Cu₂BaSn(S,Se)₄ (CBTSSe) has been gaining attention as a prospective solar absorber, since it employs low-toxicity and abundant metals, while offering low-cost manufacturing options, controllable stoichiometry, high absorption coefficient (>10⁴ cm^-1) and bandgap (E_g) tunability (1.5-2.0 eV). Besides these suitable optoelectronic properties, CBTSSe does not suffer from anti-site disorder in contrast with Cu₂ZnSn(S,Se)₄, as confirmed by Shin et al. with vacuum-deposited films [1]. The current study focuses on solution-deposited nominally stoichiometric CBTSSe films [2] with a bandgap of 1.59 eV and explores the fundamental film properties using several spectroscopic techniques. Temperature- and excitation-dependent photoluminescence studies reveal a dominant defect emission at ~1.5 eV and a second deep defect feature at 1.15 eV. Time-resolved terahertz spectroscopy measurements show few tens of picoseconds (~50 ps) surface and few nanoseconds (~3 ns) bulk lifetimes, as well as ~140 cm²/Vs mobility, the latter of which is in the range of reported values for CZTSSe. The solution-processed CBTSSe films suffer from exacerbated surface recombination at the junction of CdS and CBTSSe, due to cliff-like band alignment with 0.63 eV conduction band offset (as measured by ultraviolet photoemission spectroscopy). Employing these films in an Al-Ni/ITO/ZnO/CdS/CBTSSe/Mo photovoltaic device structure, we report open-circuit voltage (V_OC), short-circuit current density, fill factor, and efficiency of 470 mV, 14.3 mA/cm², 43.6%, and 2.93%, a record performance level among solution-processed CBTSSe devices. Device performance in this study may be mainly limited by interfacial properties and inadequately selected device structure (e.g., poorly matched buffer layer), which both enhance recombination-associated losses in V_OC. The physical measurements provided for the nominally stoichiometric solution-processed CBTSSe absorber point to critical areas for future improvement of CBTSSe and related photovoltaic cells in the quest for higher efficiency devices based on earth-abundant metals.

Reference

[1] D. Shin, B. Saparov, T. Zhu, W. P. Huhn, V. Blum, D. B. Mitzi, BaCu₂Sn(S,Se)₄: Earth-Abundant Chalcogenides for Thin-Film Photovoltaics, Chem. Mater., 28 (2016), 4771-4780.

[2] B. Teymur, Y. Zhou, E. Ngaboyamahina, J. T. Glass, D. B. Mitzi, Solution-Processed Earth-Abundant Cu₂BaSn(S,Se)₄Solar Absorber Using a Low-Toxicity Solvent, Chem. Mater., 30 (2018), 6116-6123.

Enargite (Cu₃AsS₄) is an emerging candidate for photovoltaic applications with a direct band gap of 1.41 eV, consisting of earth abundant elements with differing ionic radii and consequently lower likelihood of cation disorder, ideal predicted and experimentally verified optoelectronic properties, and demonstrated use in working thin-film solar cells. Previous experimental work on this material from our group has demonstrated that enargite films have a Shockley-Read-Hall, or defect assisted, recombination lifetime on the timescale of nanoseconds, relatively shallow defect energies, and carrier concentrations on the order of 10¹⁵ cm^-3. However, carbonaceous fine grain layers, secondary phases, pinholes in the film, and poor band alignment with n-type CdS may be part of the reason that power conversion efficiencies for enargite-based solar cells remained below 0.5%. While investigating suitable n-type junction partners is a subject of a parallel research effort, in this work, we present a novel solution-processed route to create enargite films using an amine-thiol based molecular precursor approach that eliminates carbonaceous fine grain layers, creates phase-pure films, and improves film morphology.

A mixture of amine and thiol has shown to be a very versatile solvent system for dissolving a variety of metal chalcogenides for the fabrication of thin films and nanoparticles. Arsenic sulfides and copper sulfides can undergo reactive dissolution in amine-thiol solvents to form complexes that can decompose at relatively low temperatures. These molecular precursor solutions can be combined into an ink that can be directly coated onto conductive substrates to create tennantite (Cu₁₂As₄S₁₃) films that transform and coarsen into micron-sized enargite grains with low surface roughness and no detectable secondary phases after heating in an arsenic-sulfide atmosphere at 425°C. Unfortunately, shunt resistances on devices made using this approach remained low and SEM has confirmed the presence of pin-holes in the film. To improve film morphology and increase shunt resistances, inspiration was taken from silver alloyed Cu(In,Ga)Se₂ thin films.

Silver alloying in Cu(In,Ga)Se₂ (CIGSe) solar cells has been an area of active research over the past decade. In CIGSe, Ag alloying has been shown to lower the valence band edge, increase the bandgap, increase grain size, and reduce structural defects, partially due to the lower melting temperature of these alloyed films. The amine-thiol solvent system allowed for facile incorporation of varying concentrations of silver sulfide into coating inks for the creation of silver doped enargite films. With silver incorporation, we observed significant improvements in film morphology that fused adjacent enargite grains in the film together and eliminated pinholes. Preliminary optoelectronic characterization on these films have revealed a shift to higher energy bandgaps with increasing silver content. Devices created with silver-doped enargite show increased shunt resistance and produced champion device efficiencies exceeding 0.6%. This is higher than past demonstrated efficiencies for this material and shows that silver alloying has the potential to improve film properties and device efficiencies. Further optimization of film growth conditions and silver content is expected to allow for further increases in device efficiencies.

We have used pulsed laser deposition to deposit nanoparticles (Ag, Au) to investigate
the effects of these impurities on the photovoltaic properties of the CdS/CdTe based thin films. The main
objective was to investigate how the inclusion of nanoparticles will affect light scattering in the at the interfaces
and whether the different size and shape of nanoparticles will have a positive effect on the overall electrical
performance of these thin film solar cells. In our previous studies, we have investigated the effects of the
embedded Ag nanoparticles on the photoelectric conversion efficiency on CdS/CdTe based thin film solar cells
as synthesized by Pulsed Laser Deposition (PLD). Silver was shown to enhance the photovoltaic performance
by almost doubling the photovoltaic conversion efficiency of the conventional CdS/CdTe films [1]. A careful
comparison of photovoltaic performance of Au/Si versus Ag embedded thin films of CdS/CdTe on indium tin
oxide coated glass substrates have been performed. Our results on the Ag case revealed electrical
performance of these cells have correlates with the particles density and the particle size on the CdS/CdTe
interface. This study concentrates on the Au and Ag nanoparticle deposition on the CdS/CdTe interface with
varying particle size and distributions. Structural and compositional characterization were performed using
XRD, AFM, and SEM/EDX. Photovoltaic properties were measured using a LabView assisted Keithley
Sourcemeter set-up. The comparison of Ag vs Au nanoparticles on the structure and photovoltaic
conversion efficiency will be presented. Ag and Au nanoparticles have contrasting effects on the
photovoltaic conversion efficiency in terms of their relative coverage at the interface, This will be discussed in
the light of plasmonic resonances and effective light scattering for Ag and Au particles.

[1] Olivia Rodgers, Anthony Viscovich, Yunis Yilmaz, Mehmet Sahiner, “The Effect of Embedded Ag
Nanoparticle on the Photovoltaic Conversion Efficiency in CdTe/CdS Thin Films”, American Physical Society
Bulletin, X17.13 (2018).
This work is supported by NSF Award #:DMI-0420952

In the search for solar absorber materials for thin-film photovoltaics (PV), the materials class of complex chalcogenides has maintained consistent interest. In the past two decades, the kesterite Cu₂ZnSn(S,Se)₄ (CZTSSe) has proven promising as an earth-abundant, nontoxic solar absorber. However, with an efficiency plateauing at 12.6%, a necessary search for novel earth-abundant and nontoxic chalcogenide PV candidates continues. Recently, bournonite (CuPbSbS₃) has been identified as a potential ferroelectric photovoltaic material with a bandgap (1.2-1.3 eV) appropriate for single junction photovoltaic devices (1-1.6 eV). Bournonite is an appealing candidate for a number of reasons, including its structural and electronic three dimensionality, absorption coefficients comparable to that of MAPbI₃ and GaAs, an optical dielectric constant similar to that of hybrid perovskites, high predicted defect tolerance, and the potential for Rashba/Dresselhaus splitting, a phenomenon that has been linked to slow electron hole recombination in lead-halide perovskites.¹ Additionally, as a naturally occurring mineral, bournonite is predicted to be stable, avoiding the instability challenges that face the highly popular perovskites. In the past year, progress in bournonite’s application for PV has accelerated with two studies reporting thin-film processing of bournonite.^2,3 One such study also reported the first PV device to employ bournonite as the absorber layer (PCE of 2.23%).³ In an effort to explore band gap engineering of bournonite, we report the solid state bulk synthesis of selenium-alloyed bournonite (CuPbSb(S_1-xSe_x)₃) across the full range of selenium concentrations (x = 0.0 - 1.0). We characterize the crystal structure and band gap of the samples using x-ray diffraction and diffuse reflectance spectroscopy, reporting a band gap that approaches 1.1 eV near x = 0.5. We compare these results with computational band structure predictions.

1. Wallace, S. K.; Svane, K. L.; Huhn, W. P.; Zhu, T.; Mitzi, D. B.; Blum, V.; Walsh, A., Candidate photoferroic absorber materials for thin-film solar cells from naturally occurring minerals: enargite, stephanite, and bournonite. Sustainable Energy & Fuels 2017, 1 (6), 1339-1350.
2. Koskela, K. M.; Melot, B. C.; Brutchey, R. L., Solution Deposition of a Bournonite CuPbSbS₃ Semiconductor Thin Film from the Dissolution of Bulk Materials with a Thiol-Amine Solvent Mixture. Journal of the American Chemical Society 2020, 142 (13), 6173-6179.
3. Liu, Y.; Yang, B.; Zhang, M.; Xia, B.; Chen, C.; Liu, X.; Zhong, J.; Xiao, Z.; Tang, J., Bournonite CuPbSbS₃: An electronically-3D, defect-tolerant, and solution-processable semiconductor for efficient solar cells. Nano Energy 2020, 71, 104574.

Tin sulfide (SnS) is a promising ligth-absorbing material for next-generation solar cells due to its safe and abundant constituent elements and suitable optical properties. SnS homojunction structure is expected to overcome the limitation of conversion efficiency (η) of the conventional SnS solar cells based on the heterojunction between p-type SnS and n-type CdS or Zn(O,S) with η of 4-5% at maximum[1]. However, SnS homojunction has not been reported due to difficulties in realizing n-type SnS. In recent years, we have demonstrated the n-type conductive SnS single crystals with halogen (Cl or Br) doping[2], and growth of large single crystal in centimeter-scale[3]. In this study, a p-type SnS thin film was deposited on the large size n-type crystal to form a pn homojunction, and its photovoltaic properties were evaluated.
Cl-doped large n-type SnS single crystals were prepared by the flux method using Sn as the main solvent. A p-type SnS thin film with a diameter of 1 mm and a thickness of 270 nm was deposited at 340° C on the (100) cleavage plane of the n-type single crystal by RF sputtering. The device structure was GaIn/n-type SnS single crystal/p-type SnS thin film/ZnO/ITO. Structural and electrical properties of the film and homojunction were evaluated. Photovoltaic properties were studied by I-V and C-V measurements.
The p-type SnS thin film deposited on the single crystal was <100>-oriented in the growth direction and was also gently oriented in the growth plane. In addition, there was no void at the n-type single crystal/p-type thin film interface, and the change in Cl concentration was very abrupt. The open circuit voltage (V_OC) was 360 mV, which is comparable to the highest value of the previously reported SnS homojunction cells (372 mV[1]). Since the built-in potential of the homojunction was 920 mV and much larger than the previous heterojunction (~700 mV), it should be possible to improve the V_OC further. The conversion efficiency was 1.4%, which is lower than the record η of heterojunction (4-5%), mainly due to its low closed circuit current density (J_SC) of 7.5 mA/cm². Once the device structure of the homojunction cell is optimized to efficiently collect the photogenerated carriers and achieve a comparable J_SC as the conventional heterojunction cells (~25 mAcm^-2), conversion efficiency exceeding 4-5% would be realized with homojunction with improving the V_OC.

[1] P. Sinsermsuksakul et al., Adv. Energy Mater., 4, 1 (2014).
[2] H. Yanagi et al., Appl. Phys. Express, 9, 051201 (2016); Y. Iguchi et al., Inorg. Chem., 57, 6769 (2018).
[3] S. Kawanishi et al., Cryst. Growth Des. 20, 5931 (2020).

Kesterite Cu₂ZnSnS₄ (CZTS) is a promising semiconductor material for next-generation thin film photovoltaics since it has earth-abundant and nontoxic elements, a stable structure, and strong light absorption properties. Nanoparticle, solution-based methods have been demonstrated as a potential route for the deposition of CZTS thin films, but many of these methods incorporate organic solvents or additives which can leave a carbon residue that affects device performance. In addition, these organic solvents can create problems with scalability and sustainability. The presented work demonstrates the capability of an environmentally friendly, aqueous-based nanoparticle ink to create the CZTS absorber thin film. Metal chalcogenide complexes formed with ammonium sulfide and tin sulfide are used to stabilize a mixture of component binary sulfides in water without other additives, which are deposited as a compact absorber film with spin coating. It is shown that the concentration of the ammonium sulfide content greatly affects the ink stability, the thin film morphology, and optoelectronic properties of the absorber layer. Furthermore, the incorporation of the tin sulfide complex enables a low temperature transition from binary sulfide nanoparticles to the CZTS phase.

Oxide heterostructures have drawn great attention recently, because of its environment-friendly properties and potential applications in optoelectronic devices like LEDs and Solar Cells. In this work, we have performed a simulation study of the heterojunction solar cell with SCAPS (a Solar Cell Capacitance Simulator) using Ti02 as an n-type layer and CuO as a p-type layer. The thickness and the dopant dependent simulation studies have shown that the solar cell operates at a maximum efficiency of 19.15% when the thickness of the TiO2/CuO layers are chosen 1.4µm/1.2µm. We carried out an indium doped tin oxide (ITO) vs Fluorine doped tin oxide (FTO) comparison study as well where FTO worked better as the anode in this structure. We chose gold as the cathode from the simulation study of a range of elements and observed that efficiency increases with better metal work function. Based on these simulation results we have fabricated the oxide-based heterojunction solar cell on FTO substrates using pulsed laser deposition (PLD) for the TiO2 layer and spin coating for the CuO layer. We have conducted structural-property correlations of individual layers using x-ray diffraction (XRD), Raman analysis, photoluminescence spectroscopy (PL), scanning electron microscopy (SEM), energy-dispersive x-ray spectroscopy (EDX), and different electrical measurements e.g. hall measurements, and sheet resistance measurements. Using specific PLD parameters, we have successfully grown both the rutile phase and anatase phase of TiO2 separately on several FTO substrates. The results and analysis from the characterizations confirm the successful growth of high-quality oxide thin film layers of TiO2 on FTO and CuO on TiO2. Finally, to make the cathode layer of gold, the sputtering technique is used. The solar cell characterization is performed by the I-V measurement using a standard solar simulator and UV-VIS. We have used Origin, Vesta, and MATLAB software for data analysis. The analysis of the solar cell characterization results will be presented at the conference. This facile and cost-effective process of fabrication of all oxide-based heterojunction solar cells will reduce the overall cost of solar cells and increase efficiency with a better optoelectronic performance.

Can spin property of electron lead to a novel form of solar? This question is critical because almost all the solar cells created so far are based on electronic charge. This paper reports the solar cell phenomenon based on the spin property of electrons. The spin-based solar cell effect was observed on magnetic tunnel junction based molecular spintronics devices (MTJMSD). MTJMSDs were produced by covalently bonding organometallic molecular clusters (OMCs) between the top and bottom ferromagnetic electrodes along the exposed side edges of magnetic tunnel junctions with Co/NiFe/AlOx/NiFe thin film configurations. The MTJMSD configuration showing the photovoltaic effect also exhibited OMC induced strong antiferromagnetic coupling[1] and room temperature current suppression by >five orders of magnitude[2]. Our MTJMSD was fabricated below 100-degree C temperature and employed earth-abundant transition metals like nickel and iron. This paper shows that MTJMSD’s photovoltaic effect was susceptible to the magnetic field, temperature, and light intensity. MTJMSDs exhibited three different current states termed as high (µA), medium (nA), and low (pA). In each state, light radiation produced the photovoltaic effect. OMC molecule appears to create robust exchange coupling between the two ferromagnetic electrodes of the magnetic tunnel junctions leading to significant changes in the electrical, magnetic, and optical properties of the ferromagnetic electrodes. OMC induced changes in the ferromagnetic electrodes also propagated outside the MTJ’s perimeter. Magnetic studies, KPAFM, and Raman studies suggested that OMC transformed a ferromagnetic film into photoresponsive material and produced a built-in potential in the MTJMSD. MTJMSD’s ability to absorb white light radiation and the ability to separate opposite spins in the three different current states lead to the net photovoltaic effect. MTJMSD’s photovoltaic response responded to the magnetic field. This paper mainly reports the experimental observations. Further investigation about the deeper understanding of the spin-photovoltaic effect is needed. We were also not able to provide an exact estimate of the energy conversion efficiency. It was experimentally challenging to determine the exact area responding to light radiation. Future work may focus on simultaneous KPAFM, MFM, and I-V measurements under dark and light for further understanding and new insights.

We studied the disordered-rocksalt, orthorhombic and disordered-wurtzite phases of the ternary nitride semiconductor MgSnN₂ from first principles methods using density functional theory (DFT). We find that MgSnN₂ is mechanically and dynamically stable in all three phases. However, COHP analysis suggests that the disordered rocksalt structure has anti-bonding states below the Fermi level between -5 eV to -2 eV as compared to the bonding states in other two phases, indicative thermodynamic metastability. Computed lattice constant and electronic band gap values of 4.56 Å and 2.69 eV of MgSnN₂ in disordered rocksalt structure compare well with experimentally reported values of 4.48 Å and 2.3 eV respectively. Furthermore, band gaps were computed in MgSnN_2-xO_x, with x = 0.5, 1.0, 1.5, 2.0, to elucidate the role of possible oxygen impurity. Of the three phases, disordered-rocksalt structure shows the lowest charge carrier effective masses. Moreover, this phase has promising absorption coefficient and reflectivity to be used as the absorber layer of tandem solar cells in the higher energy region of the visible portion of the solar spectrum. The other two phases can be utilized as a window layer of solar cells owing to their larger band gap values of 4.36 eV and 4.86 eV respectively.

Infrared (IR) photodetectors play a major role in numerous applications, including communications, environmental monitoring and security [1]. Various materials have been used for the purposes of infrared photo detection. Notable among them include Ge, InGaAs and InAs [2]. Research on lead based chalcogenide (PbTe, PbS, and PbSe) for infrared applications have been extensively explored by the scientific community due of the possibility to tune the bandgap in the infrared wavelength range even up to the Terahertz (THz) region [3]. Various photodetectors have been commercialized; photoconduction type which employ the use of an external bias and photovoltaic type detectors which operate with no external bias.
Owing to high transmission in the infrared region, high electrical conductivity, excellent atomic lattice structure and the broad applications for industrial purposes, great attention from the scientific community has been focused on graphene. Graphene is also a very good material for the purposes of transparent contacts in optoelectronic devices operating in the mid - to - far - infrared range, where other transparent oxides such as ITO are no longer desirable due to their free carrier absorption [4]. The increasing interest in the need for bendable, flexible and high performance optoelectronic devices makes graphene a good candidate and has therefore led to various functional materials/graphene hybridized materials [5-6]. Great efforts on the achievement of high-performance graphene-based photodetectors have been concentrated on the development of graphene (G) hybrid structure such as G-semiconductors, G-quantum dots (QD) and G-polymer. Studies have also been conducted to form epitaxial junctions by directly depositing various functional materials on large area CVD-graphene [7].
We fabricated vertical photovoltaic type G/PbS/Ti device by taking advantage of Ti/PbS Schottky junction and discussed the photocurrent transient characteristics. Lead sulfide (PbS), which has a bandgap of 0.42 eV at room temperature, was deposited directly on post-annealed large-area CVD (Chemical vapor deposition) graphene by CBD (Chemical bath deposition). Using a metal mask and an e-beam evaporator, we deposited Ti on glass/G/PbS. Temperature dependent photocurrent spectra of our G/PbS/Ti photovoltaic devices were measured by a Fourier transformed infrared (FTIR) set-up. Post-annealing was found to be an important activity for the adhesion of PbS films on G/glass. As the bandgap energy of PbS decreases with decreasing temperature, the cut-off wavelength increases to longer wavelength, as temperature decreases.
We also fabricated a lateral type device which was comprised of two (2) Ti electrodes separated by a gap of 7 mm and a PbS layer and discussed the laser power dependent photocurrent transient. Because we could obtain fast photo-response from the lateral type device, photocurrent spectrum at room temperature using a FTIR equipment was measured. The cut off wavelength of the detector was found to be ~3.2 μm at room temperature (300 K) which is typical for PbS devices. Our photovoltaic PbS devices can detect not only short - to – mid - infrared but also terahertz radiation at room temperature, which is highly applicable in many fields.

Since tin (II) sulfide (SnS) is a semiconductor composed of abundant and safe elements and possesses a bandgap (1.1-1.3 eV) suitable for solar cells, it is expected as a light absorber for the next-generation thin film solar cells. The acceptor-type defects, such as Sn antisite (Sn_S) and Sn vacancy (V_Sn), are easy to be formed in SnS due to their low formation enthalpy, SnS typically exhibits p-type conduction and it is difficult to convert it into n-type [1]. SnS solar cells, therefore, have adopted a heterojunction structure with p-type SnS and n-type semiconductors such as CdS, however, the conversion efficiency is still 4-5% at maximum [2]. Such low efficiencies are attributed to an unfavorable band offset and interface defects at the hetero interfaces. Homojunction structure is expected to exhibit higher conversion efficiency and therefore many efforts have been made to realizing n-type SnS. In recent years, it has been reported that n-type SnS sintered compact and single crystals are obtained by halogen (Cl or Br) doping, which paved the way for SnS homojunction [4]. However, the n-type SnS thin films have not been reported so far and it remains as a barrier towards realizing homojunction devices. In this study, we succeeded in obtaining n-type Cl-doped SnS thin films by RF sputtering deposition under a use of RF atomic sulfur supplier (sulfur cracker).
Cl-doped SnS thin films were deposited by RF puttering on a SiO₂ glass substrate heated at 220-340 °C. Atomic sulfur was supplied to the deposition field of the thin film at the same time from the sulfur cracker installed next to the sputtering cathode. The electrical conductivity, Hall coefficient, chemical compositions, and optical absorption spectra of the obtained thin films were studied.
All of the obtained SnS thin films were p-type conductive without using sulfur cracker and n-type conductive with sulfur cracker, which indicates that supply of the atomic sulfur to the deposition suppresses the formation of acceptor-type defects. Whereas the n-type films exhibited sharp absorption due to the direct band gap around 1.3 eV, p-type films additionally exhibited a broad absorption in the range of 0.9 to 1.2 eV. This broad absorption indicates that there are defect levels with a large density of states within the band gap, which supports the above speculation that the acceptor-type defects were suppressed by performing the deposition with the sulfur cracker.

[1]J. Vidal et al., Appl. Phys. Lett., 2012, 100, 032104.
[2]P. Sinsermsuksakul et al., Adv. Energy Mater. 2014, 4, 1400496.
[3]H. Yanagi et al., Appl. Phys. Express, 2016, 9, 051201.

The electronic structure properties of interfaces are of central importance to photovoltaics. Computational approaches can provide information complementary to experimental characterization. In particular, first principles methods can establish an unambiguous link between atomic structures features and their electronic consequences. This strength is also a challenge, because realistic atomic structure models must be available for such calculations. Only for highly idealized situations, such as the epitaxial interface between two reasonably lattice matched materials with the same crystal structure (coherent interface), can the atomic interface structure be trivially constructed. The situation becomes much more complicated when the interface must accommodate incommensurate surface unit cells for the two materials. In this case, the interface becomes incoherent, i.e., exhibiting a reduced periodicity requiring a larger surface unit cell or even forming an essentially (2-dimensionally) amorphous non-periodic structure.
In this work, we adopted the kinetically limited minimization (KLM) approach, previously utilized for unconstrained crystal structure prediction, to sample atomic structures for SnO₂/CdTe interfaces without and with the presence of CdCl₂. The simulations treat SnO₂ as a substrate, and CdTe as a thin film, with a variable number of atomic layers. No initial assumptions are made about the CdTe atomic structure. Both the correct zinc-blende bulk structure and the known surface structures of the (110) and (001) back-surfaces are reproduced over the course of the sampling. The inclusion of CdCl₂ results in improved bonding and periodicity, and considerably lowers the total interface energy. While the full sampling of the compositional and configurational space remains challenging, this work provides a proof of concept for first principles atomic structure prediction for incoherent interfaces.

The advancement of thin film photovoltaic devices requires the successful implementation of a series of high-performing, low-cost scribes for efficient module interconnections. These scribes, which serve to electrically isolate the front electrode (P1) and rear electrode (P3), and provide the electrical interconnection (P2) between adjacent cells, have been explored using a wide variety of scribing options, primarily laser-based mechanisms. Laser scribing offers the benefits of high tunability and alignment accuracy, allowing for the selective removal of specific layers while mitigating dead area losses. Different laser wavelengths and process parameters are often employed to achieve this selectivity; however, the demand for multiple unique laser systems for complete module fabrication results in high equipment costs and increased process complexity.

Our work demonstrates the successful implementation of a single-source scribing mechanism capable of performing the P1, P2, and P3 scribes, enabling further simplification and cost reduction over other laser scribing methods. This mechanism utilizes one of the most commercially available laser wavelengths (1064nm) without the need for expensive yet commonly employed ultrashort laser pulsed systems, relying exclusively on interactions between the laser and front electrode material and thus demonstrating broad applicability across a variety of materials and architectures. This unique scribing mechanism is compatible with extremely high throughputs on the order of several meters per second, an order of magnitude increase over the speeds of other laser scribing mechanisms. An improvement at this scale further enables reductions in cost of over 90% for the laser scribing steps in a MW-scale photovoltaic system.

Quantifying the performance impacts of band tails on CdTe, CIGSe, and CZTS thin-film solar cells remains a significant challenge to further development of these technologies. Band tails are prevalent in CdTe, CIGSe, and CZTS absorption, quantum efficiency, and luminescence data and are believed to reduce the open-circuit voltage (V_OC). However, quantifying band-tail related performance impacts is not straightforward. Tailing present in macroscopic optoelectronic data may arise from inhomogeneous impurity/matrix-element distributions, from structural defects such as dislocations and grain boundaries, or from intentional compositional grading [e.g., Cu(In_xGa_1-x)Se₂]. Furthermore, inhomogeneities may span from nm to mm in scale. Thus, quantifying performance impacts includes the challenges of separating tailing due to inhomogeneities and structural defects from tailing due to compositional grading; determining the characteristic length scale of inhomogeneity; and reconciling macroscopic and microscopic data and models.

The thin-film solar cell community knows well that fluctuations in dopant densities produce electrostatic potential fluctuations. However, the community still relies heavily on early theoretical work (from Shklovskii and Efros and others) to estimate potential fluctuation magnitudes. Commonly used expressions were only meant to provide first-order estimates and are potentially inadequate for describing fluctuations occurring in thin-film solar cell materials. In this talk, I will summarize our recent efforts to develop realistic estimates of fluctuation impacts on performance, using Cd(Se)Te solar cells as an example.

Optoelectronic data on Cd(Se)Te cells doped with arsenic show an increasing tail below the bandgap that is correlated with increasing arsenic incorporation level. These cells have As-dopant activation levels of only 1-5%, and an inverse relationship exists between dopant activation and incorporation, consistent with self-compensation mechanisms. Macroscopic photoluminescence and microscopic cathodoluminescence measurements were conducted at room and low temperature. The data indicate that, to a large extent, the tailing observed in macroscopic measurements, and the inhomogeneity that causes it, is present within individual µm-size grains.

Consistent with this picture, we consider that tails are created by randomly distributed As donor and acceptor (shallow) defects. The densities of donor and acceptor defects are based on incorporation (secondary-ion mass spectrometry) and carrier-density (capacitance-voltage) data. Donor and acceptor defects are randomly distributed in a cube and each is assigned a screened coulomb potential. Fluctuations in the potential can be clearly seen and we confirm that this numerical approach reproduces results from early semiclassical band-tail theory. Furthermore, we apply semiclassical theory to estimate V_OC losses in Cd(Se)Te solar cells.

Addressing the weaknesses of semiclassical theory, we develop self-consistent numerical methods to analyze random dopant fluctuations. We distribute donor and acceptor defects randomly within a 2D domain and solve Poisson’s equation with density-gradient corrections to the electron and hole densities. The results show how compensation can enhance potential fluctuation amplitudes and induce large carrier-density fluctuations. Interestingly, we find that the effective band gap energy is not constant but has a distribution with a low energy tail. Finally, we use simple formulas to show how the effective bandgap tail can produce tails in absorption and luminescence spectra. These methods can potentially be used to form realistic models of fluctuation impacts on thin-film solar cell performance and facilitate progress.

Cadmium telluride (CdTe) photovoltaic (PV) modules are relatively easy to produce at a low cost, making them one of the most competitive commercially available PV technologies. Recently, the world-record PV cell has a power conversion efficiency of 22.1%, which is well below the theoretical limits. The improvement of efficiency without increasing the production costs will make PV generation more competitive with fossil fuels. The practical strategy for higher efficiency is improvements of open-circuit voltage (V_OC) and short-circuit current (J_SC).
Recently, it was discovered that group-V element doping (P, As, Sb, or Bi) and Cd-rich composition are key technologies for increasing V_OC, and these results were attributed to high p-type doping and long minority carrier lifetime. PV devices fabricated with P-doped CdTe single crystals as absorbing layer and with Cd-rich composition could result in hole concentration of ~10¹⁷ cm^-3, minority carrier lifetime of several hundreds of nano second, and V_OC exceeding 1 V. However, it is difficult to obtain group-V doping higher than 10¹⁷ cm^-3 because doping efficiency typically becomes lower with higher dopant incorporation (above ~10¹⁶ cm^-3). The origin of this doping limit has not been established, but theoretical studies suggest that it may be caused by the formation of self-compensating AX center or defect complex. The AX center is formed due to a large lattice relaxation of the substitutional dopant which results in conversion to a donor state. Recently, with in situ As-doping in polycrystalline thin film CdTe-based PV devices larger than 20% power conversion efficiency was be achieved, but dopant activation remained low (2-4%).
On the other hand, CdTe alloying with Se (CdTeSe) has attracted strong interest for improving J_SC because lower bandgap leads to increased absorption in the long wavelength part of the spectrum. Bandgap tuning through alloying is widely used in PV. For example, high-efficiency Cu(In_xGa_1-x)Se₂ (CIGS) absorbers employ bandgap grading by controlled variation of In/Ga composition.
We combine both approaches to develop more efficient CdTe solar cells: group V doping and alloying with Se. We have grown high-quality As-doped Cd-rich CdTeSe single crystals from metal Cd solvent near thermodynamic equilibrium using the traveling heater method (THM). In this presentation, we will report growth and characterization of group-V doped Cd-rich CdTeSe single crystals from THM. It is not clear how the formation of CdTeSe alloy affects the activation of group-V doping, defect compensation mechanism and PV performance, and we propose some possible answers to these questions.

Widespread research in the field of wearable electronics, has ignited interest in stand-alone and efficient power generators. Photovoltaic devices are promising candidates for such applications. Due to their high stability and biocompatibility, Si solar cells are an attractive choice. However, the absorption coefficient of Si is highest in the visible and NIR region of the EM spectrum. Thus, efficient Si cells are opaque, which makes them undesirable for a wide range of applications that require the on-board electronics to be transparent. Additionally, the thickness of the conventional Si solar cells is ~200-300 μm which makes them rigid and unsuitable for applications that require the system components to be thin and flexible. Here, we present the design, fabrication, and performance evaluation of semi-transparent (T ~ 30% and 60%) and ultra-thin (15 µm) Si solar cells fabricated by surface-bulk micromachining (SBM) process.
A <111> boron-doped Si wafer with an average bulk resistivity of 5-10 Ω.cm was used as the substrate and base region. 400 nm of SiO₂ was thermally grown which acts as the diffusion barrier for subsequent doping steps. Phosphorous doping was performed at 1000 °C for 60 minutes to define the emitter region (sheet resistance = 8 Ω.sq^-1). Micro holes (with a diameter of 170 µm) were then lithographically patterned in a hexagonal arrangement to achieve different levels of transparency in the fabricated Si solar cell. Next, deep reactive ion etching (DRIE) was performed on these patterns to realize trenches of 15 μm depth. This defines the final thickness of the solar cell. After this, 300 nm of LPCVD low-stress Si nitride was deposited on the sidewalls of the trenches as an etch resistant layer. This was followed by patterning and a second DRIE (DRIE 2) to dig deeper trenches than those achieved previously. The final transparency and effective surface area of the Si solar cell is determined during this step. The diameter of DRIE 2 micro holes was designed to be ~150 µm with a filling ratio of 25% and 50% that corresponds to an inter-hole distance (s) of 138 µm and 54 µm, respectively. This inter-hole spacing determines the required minimum depth of DRIE 2 trenches (h), which is crucial to release the ultra-thin device from the underlying substrate. Since (111) family planes have a negligible etch rate in KOH/TMAH, they act as an effective etch-stop layer. Therefore, to ensure that the etchant completely undercuts the device before reaching the etch-stop layer, the DRIE-2 depth (h) must be higher than s . tan (19.47°). The DRIE-2 trench depth for 25% and 50% filling ratio devices were designed to be 394 µm and 154 µm, respectively. Finally, the samples were dipped into a 10 wt% tetramethylammonium hydroxide (TMAH) solution at 100 °C for completely releasing the structures. The time taken to release devices with 25% filling ratio and 50% filling ratio was 10 hours and 4 hours, respectively. After fishing out the released devices from the TMAH solution and removing the remaining Si nitride by CF₄/O₂ RIE dry-etching, 20 nm of SiO₂ was deposited as a passivation layer on the bottom of the solar cell and the sidewalls of DRIE trenches.
15 µm thick Si solar cells with a transparency of ~ 30% and ~ 60% were fabricated using the proposed method. Photocurrent-voltage measurements were conducted under indoor white light condition with an input power density of 13 mW.cm^-2. I-V results demonstrate the open-circuit voltage of 300, 120 mV and the short circuit current density of 34, 1.4 µA.cm^-2 for the 30% and 60% transparent silicon solar cells, respectively.

The Kesterite Cu₂ZnSn(S,Se)₄ (CZTSSe) semiconductor is a compelling emerging light-harvesting material for low-cost, environment-benign, and high-efficiency thin-film photovoltaics. The highest power conversion efficiency so far for kesterites are 11% and 12.6 % for pure sulfide kesterite and Se-containing kesterites, respectively. The current state-of-the-art kesterite devices suffer from the challenge of controlling defects (e.g. cation—disordering defects, defect clusters), which generally results in severe potential fluctuation, low minority carrier lifetime, and thereby unsatisfactory device performance. Identifying effective ways of controlling these defects is of vital importance for further efficiency breakthroughs. In our recent work, we demonstrated that some of these defects can be suppressed by either controlling the local chemical environment during the chalcogenisation synthesis of kesterite, or lithium post-deposition-treatment. By controlling the local chemical environment in time when the kesteirte phase starts to form, we demonstrated an independently confirmed 12.5% efficiency for CZTSe solar cells with high V_OC of 491 mV, showing the promise of these defect control strategies.

The chalcopyrite family has emerged as a material with many possibilities and has yielded high efficiencies above 23 %[1]. One of the additions to this family is alloying with silver to form (Ag,Cu)(In,Ga)Se₂ (ACIGS). Among the features of Ag alloying as compared to Cu(In,Ga)Se₂ (CIGS) without Ag, are a lower melting temperature, that leads to large grains also for relatively low deposition temperatures, but also lower voltage losses as compared to the Shockley Queisser limit[2], and a slight increase in bandgap, due to Ag, accompanied by lowering of both the conduction band minimum and of the valence band maximum energy[3]. These advantages open up interesting applications for ACIGS as e.g. top cells in tandem devices, either in combination with low bandgap CuInSe₂ (with or withot Ag) or with Si as bottom cells. Due to its electronic band structure, ACIGS with high bandgap up to 1.5 eV yields high efficiencies with a conventional buffer layer like CdS, but also with ZnSnO and Zn(O,S)[4]. In addition to high efficiency, also long-term performance is essential for all applications and there are some questions related to phase stability for the ACIGS material, that need to be investigated further, but where we have obtained new insights with the help of first principle calculations[5]. Among the studies made in our group during the last few years are ACIGS with Ag/(Ag+Cu) ratios varying from 10 to 100 % and with varying Ga/(Ga+In) from 0 to 100 %. We have also applied ACIGS to solar cells with submicron thick absorber layers, as well as investigated the impact of post deposition treatments to this type of materials. In the talk, a short review of the state-of-the art of ACIGS will be given together with some of our insights from this line of research.
[1] Nakamure et al, IEEE Journal of Photovoltaics, 2019, 9 (6), 1863-1867
[2] Edoff et al, IEEE Journal of Photovoltaics, 2017, 7 (6), 1789-1794
[3] Boyle et al, Journal of Applied Physics, 2014, 115 (22), 223504
[4] Keller et al, Progress in Photovoltaics, 2020, 28 (4), 237-250
[5] Sopiha et al, Journal of Materials Chemistry A, 2020, 8 (17), 8740-8751

In order to obtain long-term stable and low-cost photovoltaic modules, it is important to know possible degradation mechanisms. In order to identify long term behavior, unpackaged Cu(In,Ga)Se₂ solar cells were simultaneously exposed to damp heat and illumination (settings: 85^oC/85% relative humidity and 1000 W/m² BAA illumination) [1]. In-situ monitoring of their electrical parameters demonstrated a rapid decrease of the efficiency, mainly driven by changes in the series and shunt resistances. Moreover, non-degraded and degraded solar cells were studied by in-depth XRD and SIMS to investigate the material changes leading to efficiency loss.
SIMS revealed the migration of sodium and potassium, likely leading to a severe decrease in the shunt resistance (from 500±100Ω to 20-30Ω within 200 hours) and voltage. It also displayed the ingression of hydroxide, especially in the ZnO:Al film. Extensive XRD measurements showed that molybdenum oxide was formed, which likely affected the Ohmic contact between Mo and CIGS. Moreover, an in-plane stress increase in the ZnO:Al film from -183±29 to -450±40 MPa was observed. This stress increase is most likely due to the incorporation of species like hydroxides and carbonates in the grain boundaries of the ZnO:Al film. These phenomena could lead to the observed increased series resistance in the solar cells [2].

[1] M. Theelen et al., JoVE 140 (2018)
[2] M. Theelen et al., Sol. Mat. Sol. Cells 157 (2016) 943–952

Thin-film photovoltaic (PV) technologies, and Cu(In,Ga)Se₂ (CIGS) thin-film solar cells in particular, have been the subject of increasingly rigorous study of late. There are several motivating factors for the development of thin-film photovoltaics, such as the reduction in raw material usage, the decrease in solar cell weight, and the possibility of deposition on flexible substrates. Furthermore, CIGS-based thin-film solar cells hold several advantages compared to rival solar cell technologies. Specifically, they have the highest conversion efficiencies among chalcogenide thin-film PV technologies (23.35% for cells and 17.5% for modules), high radiation resistance, and outstanding stability.

The typical CIGS-based solar cell consists of a soda-lime glass (SLG) substrate, a Mo back contact, CIGS as the p-type absorber layer, CdS as the n-type buffer layer, and i-ZnO/ZnO:Al as the decoupling and transporting window layers, respectively. However, in the state-of-the-art CIGS-based solar cells, each of these layers is deposited with a different method; co-evaporation for the CIGS layer, chemical bath deposition for the CdS layer, and reactive sputtering for the i-ZnO/ZnO:Al bilayer.

This work reports on the utilization of pulsed laser deposition (PLD) as a single technique for the preparation of the aforementioned layers of a complete CIGS-based solar cell stack. Employing a single deposition technique greatly reduces manufacturing complexity. Furthermore, it potentially decreases processing time and associated fabrication costs through streamlined production lines.

The presented results discuss the challenges faced in the completion of this task. The properties of the PLD-grown thin films with respect to structure, composition and morphology are parametrically investigated through X-ray diffraction (XRD), energy-dispersive X-ray spectroscopy (EDS), scanning electron microscopy (SEM) and atomic force microscopy (AFM). Hence, the influence of PLD process parameters on film growth is evaluated. TCAD simulations realized in Synopsys Sentaurus Device are utilized for the optimization of the function of the cell. Electrical and optical measurements carried out in a solar simulator are used to assess the photovoltaic behavior of the complete structure and, critically, the CdS/CIGS junction heterointerface. The solutions implemented, involving modifications to the typical PLD process, are discussed. Finally, the resulting improvement in conversion efficiency is demonstrated.

CuIn_0.7Ga_0.3S₂ semi-conductor has a bandgap energy of 1.7 eV, which is well adapted to applications as top cell in tandem structures with c-silicon bottom cell. Recently, we have shown that solar cells with standard structure, namely Glass/Mo/CuIn_0.7Ga_0.3S₂/CdS/ZnO/AZO, can reach efficiencies above 14 %; the CuIn_0.7Ga_0.3S₂ layer being grown by co-evaporation following a 3-stage process. Determination of compositional depth profile of these absorbers revealed that the distribution of group III elements (i.e. In and Ga) is not homogeneous throughout the layers, as expected for films grown by means of sequential process. However, in contrast to V-shaped GGI evolution commonly observed in selenide absorbers, these sulfide films appear composed of two distinct layers, a first with GGI ~ 0.7/0.8 near the back contact and a second with GGI ~ 0.2/0.3 close to the surface; note that the abrupt interface between these layers is located at about half of total film thickness. A large part is the present contribution will focus on the origins for such bilayered compositional structure, which will be discussed accounting relative phase stability while copper content is increased during the second stage. In addition, although optimal bulk material quality could be achieved, device performance still appears limited by dominant interface recombination. To further explore that issue, the first results on epitaxial CuIn_0.7Ga_0.3S₂ layers will be presented.

Thin film solar cells have the possibility to be made flexible, semi-transparent and/or may be applied for tandem structures or building integrated. Having multiple usages, it is of interest to make these thin film solar cells as fast and cheap as possible. From the available thin films, Cu(In,Ga)Se2 (CIGS) has one of the best solar cell performance [1]. There are concerns about the usage of indium though when CIGS will be widely applied. Therefore, making the CIGS layer thinner and thereby reducing the amount of In, is an interesting option to explore. When the absorber material becomes thin, interfaces are generally limiting the performance and the path length to absorb all the incoming light may be too short to absorb the longer wavelengths. There are various approaches to tackle these problems. Passivation of the back contact by applying a dielectric is the main route investigated. Not only reduces it the interface recombination at the Mo/CIGS back contact but it also increases the reflection. We will present here various approaches which are aimed to be industrial viable and easy to make to reduce the losses in ultrathin CIGS solar cells. This also holds for the absorber layers which are made by a single stage coevaporation process. In this case, the Ga gradient at the back is replaced by a passivation layer and no copper rich stage is applied. The effect of the simplified growth is mostly visible in the grain boundaries [2]. As passivation layer, AlOx is often used. As this layer blocks the current flow, it has to be either sufficiently thin to allow tunneling or requires holes allowing for the current to flow to the Mo back contact. We will show that the latter approach can be applied for layers up to 6 nm without lithography steps by simply adding NaF before CIGS growth [3]. When applying a thin Ag layer under the AlOx the cells also show optical improvement. This is likely due to formation of scattering particles, making the CIGS rougher and thereby increasing the optical path length. To improve the absorber layer itself alkali treatments are applied [4]. At the front, the CIGS/buffer interface may also need to be adapted. Application of a passivation layer at the front, either with holes or thin enough for tunneling, is investigated. Before applying a passivation layer on the CIGS surface, cleaning treatments need to be applied [5]. To analyze the effect of the various treatments a combinatorial approach based on bias dependent admittance spectroscopy and (time resolved (TR)) photoluminescence (PL) is developed. Generally, with (TR)-PL a quick analysis whether the treatments are improving the absorber/interface quality before finishing into a device is possible. Features like interference and blue shifts of the spectra may be observed as well and can be related to the emission profile and scale of the potential fluctuations in the absorber layer [6]. However, improvement due the various treatments observed in the (TR)-PL does not necessarily translates into better solar cell performance. This can indicate that the treatments are sensitive to the processing conditions of the window layers for instance and/or that barriers are formed. With bias-dependent admittance spectroscopy losses at interfaces and bulk can be found by introducing a so-called CVf loss map [7]. These maps visualize the losses in the solar cell, which makes it possible to distinguish defects in the bulk from the interface and barriers at front or back interfaces. The bottlenecks observed when developing ultrathin CIGS solar cells will be discussed and possible options to mitigate them are investigated.

[1] M.A. Green, DOI: 10.1002/pip.3303
[2] T. Kohl, DOI: 10.1021/acsaem.0c00610
[3] G. Birant, DOI: 10.1016/j.solener.2020.07.038
[4] J. de Wild, DOI: 10.1021/acsaem.9b01370
[5] D. G. Buldu, DOI:10.1002/pssa.202000307
[6] J. de Wild, DOI: 10.1063/5.0024840
[7] G. Brammertz, DOI: 10.1109/JPHOTOV.2020.2992350

Oxide-based materials could be candidates for buffer, passivation, or high resistive (HR) layers in Cu(In,Ga)Se₂ (CIGS) thin-film solar cells. In recent years, oxide materials like (Zn,Mg)O, Al₂O₃, Zn_xTi_yO [1], (In,Ga)₂O₃ [2] and Sn_1-xGa_xO_y [3] were under investigation. Most of them were deposited by atomic layer deposition. However, from an industrial point of view a very fast method, with deposition times <1 min for a complete layer, would be preferable for buffer, passivation, or HR layer growth.
In this study, we present our results with rf-magnetron sputtered Ga₂O₃ buffer layers as a substitute for our solution-grown CdS buffer in combination with i-ZnO as HR layer and ZnO:Al front contact. So far, best cells achieved efficiencies up to 14% without post-annealing compared to the CdS-buffered reference cells, which exhibit efficiencies up to 17%. As a result of the very high bandgap energy around 4.7 eV of Ga₂O₃, as determined on quartz glass by optical transmittance measurements, and reduced parasitic absorption even higher Jsc values could be achieved compared to the CdS-buffered reference cells. Nevertheless, the difference in efficiency is mainly due to reduced Voc and FF values of the CIGS cells with Ga₂O₃.
The solar cell performance of cells with Ga₂O₃ buffers strongly depends on the deposition conditions as well as on the thickness of the sputtered Ga₂O₃ layers. We observed poor cell efficiencies for cells with Ga₂O₃ sputtered at room temperature and good values for films grown at elevated substrate temperatures (120 – 200 °C). This result is similar to the behavior we observed for CIGS cells with sputtered Zn(O,S) and In_xS_y buffers [4]. The Ga₂O₃ layers deposited in the relevant temperature range for solar cell applications are X-ray amorphous.
When using CIGS absorber layers with a RbF post-deposition treatment, a wet chemical treatment of the CIGS surface before the Ga₂O₃ deposition has a significant positive impact to achieve decent efficiencies. Non-rinsed samples could exhibit poor efficiencies, mainly due to bad FF values.
In addition, we could achieve a comparable good efficiency of 19.4% (with anti-reflective coating) with Ga₂O₃ applied as an HR layer instead of i-ZnO (19.7%) in combination with CdS as a buffer and ZnO:Al as front contact.

[1] J. Löckinger et al., ACS Appl. Mater. Interfaces 10 (2018) 43603
[2] T. Koida et al., IEEE J. Photovolt. 5 (2015) 956
[3] F. Larsson et al., J. Vac. Sci. Technol. A 37 (2019) 030906
[4] D. Hariskos et al., Appl. Sci. 10 (2020) 1052

The introduction of post-deposition treatments (PDTs) has increased the Cu(In,Ga)Se₂-based (CIGSe) thin-film solar cell efficiencies to >23 % on a laboratory scale. Such high-efficiency devices are often processed with a solution-deposited CdS buffer layer. However, the CdS band gap leads to absorption losses in the buffer layer, reducing the solar-cell efficiency. Hence, there is a strong interest to replace CdS with an alternative wide(r) band gap material. In addition to merely increasing the buffer layer band gap, using such an alternative material would also lead to various changes at the buffer/absorber interface, including the electronic and chemical structure. A detailed knowledge of these changes is crucial for further device optimization.

In this contribution, we report on industrially relevant CIGSe absorbers with and without an NH₃-based rinsing step applied after the RbF PDT. Their interface with a wide band-gap GaO_x buffer layer is studied using x-ray photoelectron spectroscopy (XPS), hard XPS (HAXPES) at the synchrotron, and a combination of UV photoelectron spectroscopy (UPS) and inverse photoemission spectroscopy (IPES). The results paint a comprehensive picture of the GaO_x/CIGSe interface and will be discussed in view of the performance of the corresponding CIGSe solar cells.

We investigate atomic layer deposited zinc magnesium oxide films with varying Mg content as transparent conducting films and as electron transport layers for chalcopyrite solar cells. Optimizing a ternary process by mixing two binary ALD process has its challenges [1]. We achieve high degree of control on composition by optimizing the growth conditions with varying deposition temperatures and supercycle parameters such as pulse ratios and bilayer period. We examine the effect of these films on the performance of high-bandgap solar cells based on Cu(In,Ga)S₂ absorbers.

Previous studies on the impact of Zn_1-xMg_xO:Al as transparent electrodes and Mg doped ZnO thin films for the window layer of CIGS Cu(In,Ga)(S,Se)₂ solar cells employed co-sputtering , electrodeposition and ALD respectively [2][3][4][5]. With the advantage of low temperature and highly conformal thin film growth, we study ALD grown Zn_1-xMg_xO with x varying from 0.1 to 0.4 and elucidate the effect of doping on the band alignment, electrical and optical properties. With variation in Mg content we were able to achieve 11% efficient Cu(In,Ga)S₂ solar cell with an open-circuit voltage of 941 mV.



[1]. Mackus, Adriaan JM, et al. "Synthesis of doped, ternary, and quaternary materials by atomic layer deposition: a review." Chemistry of Materials 31.4 (2018): 1142-1183.
[2]. Kuwahata, Yoshihiro, and Takashi Minemoto. "Impact of Zn_1-xMg_xO: Al transparent electrode for buffer-less Cu (In, Ga) Se2 solar cells." Renewable energy 65 (2014): 113-116.
[3]. Wang, Mang, et al. "Electrodeposition of Mg doped ZnO thin film for the window layer of CIGS solar cell." Applied Surface Science 382 (2016): 217-224.
[4]. Inoue, Yukari and Hala, Matej et al. "Optimization of buffer layer/i-layer band alignment" in 42nd IEEE Photovoltaic Specialist Conference (IEEE, New Orleans, 2015), pp. 1
[5]. Hiroi H, Iwata Y, Adachi S, Sugimoto H, Yamada A. New World-Record Efficiency for Pure-Sulfide Cu(In,Ga)S2; Thin-Film Solar Cell With Cd-Free Buffer Layer via KCN-Free Process. IEEE J Photovolt. 2016;6(3):760-763.

Solution processing of inorganic thin-film solar cells is a key challenge in the growing area of system integrated photovoltaics. The ability of processing high quality materials at temperatures below 200 °C enables the fabrication of devices onto flexible composite materials. Semiconductor compounds with high degree of defect tolerance are an exciting class of materials particularly well-suited to these applications. Indeed, compounds based on Bi³⁺ can lead to attractive opto-electronic properties similar to those observed in Pb²⁺ hybrid perovskites, such as large spin-orbit coupling, dielectric constant and band dispersion.¹
In this contribution, we will discuss the structure and opto-electronic properties of phase-pure BiI₃ and BiSI obtained by solution based method. BiI₃ is prepared by spontaneous gas-phase iodination of Bi₂S₃ films at 200 °C,² while BiSI is generated by thermolysis of a precursor solution composed of Bi(NO₃)₂, thiourea and NH₄I.³ In addition to accurate structure refinement from XRD data of the thin-films, we find excellent correlation between experimental Raman spectra and Raman modes calculated by density functional perturbation theory. Interestingly, BiI₃ exhibits p-type conductivity (acceptor density of the order of 10¹⁵ cm^-3) and a band gap of 1.7 eV, while BiSI is n-type with a donor density in the range of 10¹⁹ cm^-3. Quasi-particle G₀W₀ calculations of both materials show that the conduction band is more dispersed than the valance band due to spin-orbit coupling promoted by Bi³⁺. Band edge energy values are estimated by electrochemical impedance spectroscopy, enabling the design of PV devices with appropriate band alignment. Devices with the structure Glass/FTO/TiO₂/BiI₃/F8/Au and Glass/FTO/SnO₂/BiI₃/F8/Au , where F8 is Poly(9,9-di-n-octylfluorenyl-2,7-diyl), display open-circuit voltage above as high as 600 mV and record power conversion efficiency of 1.32% under AM 1.5G illumination. We will expand the discussion to other complex Bi chalcogenides,⁴ addressing the key limiting factor in the PV performance of these materials, i.e. their short carrier lifetimes.
References:
1- A.M. Ganose et al. Chem. Commun. 2017, 53, 20
2- D. Tiwari and D.J. Fermin, ACS Energy Lett. 2018, 3, 1882
3- D. Tiwari et al. ACS Appl. Energy Mater. 2019, 2, 3878
4- D. Tiwari et al. Chem. Mater. 2020, 32, 1235

Replacing the established Molybdenum back contact in Cu(In,Ga)Se₂ solar cells by a transparent conducting layer would enable a variety of new applications: Bifacial solar modules, semitransparent building integration without back side mirroring, and ultrathin solar cells with scattering back reflector. In this work, we review the status of alternative back contacts for CIGSe solar cells and present own work using ITO. It is shown that for low temperature CIGSe deposition, the ITO back contact is on par with the Mo standard regarding PCE, and may even surpass Mo for ultrathin devices. Admittance measurements on ITO back contacts indicate the absence of a back contact barrier and thus the avoidance of the wellknown reach-through effect. Using bifacial illumination, the back contact recombination velocity is determined to around 10⁵ cm/s for the ITO/CIGSe interface formed at around 480°C by co-evaporation of CIGSe on sputtered ITO. Cells of 300 nm CIGSe thickness demonstrate quantum efficiency of 55% if illuminated from the back side. If the ITO layer is grown on top of a reflective back contact of 100 nm Aluminum, ultrathin devices can be constructed. Here, the Aluminum acts as the optical reflector and the ITO serves as a diffusion barrier and electronic back contact. This back contact can be further functionalized by light scattering elements. In experiment, a 300 nm structure element height below a 500 nm CIGS film gives the highest current density, which amounts to 88% of that of a 2.8 µm reference solar cell on Molybdenum.

Lead-halide perovskites have emerged as a leading thin film solar absorber over the past decade. One of the key enabling properties is their tolerance to point defects, which enables them to achieve long charge-carrier lifetimes despite high defect densities when grown by low-temperature, simple fabrication methods. However, the toxicity of the water-soluble lead content may limit their widescale adoption. This talk examines bismuth-based semiconductors as a low-toxicity and stable alternative to lead-halide perovskites, with particular focus on bismuth oxyiodide (BiOI). We show that BiOI replicates the tolerance of lead-halide perovskites to point defects using experimental and computational methods [1,2]. An all-inorganic device structure is developed, and we achieve photovoltaic devices with external quantum efficiencies of up to 80% at 450 nm wavelength. Both the BiOI absorber and devices are stable in air. Although the 1.9 eV band gap is too wide for single-junction devices under 1-sun illumination, this band gap is ideal for indoor light harvesting. We demonstrate BiOI devices with comparable performance to hydrogenated amorphous silicon (the industry standard for indoor photovoltaics) under fluorescent and white LED lighting. We show that these devices, with lab-scale millimetre-squared areas, are sufficient to power carbon nanotube-based inverters [3]. Finally, we discuss the key limiting factors that need to be overcome to achieve further improvements in performance [1,3,4], as well as the potential of the broader family of bismuth-based perovskite-inspired materials for applications in solar energy and indoor light harvesting [3].

[1] R. L. Z. Hoye, et al. Adv. Mater. 2017, 29,1702176
[2] T. N. Huq, L. C. Lee, …, R. L. Z. Hoye, Adv. Funct. Mater. 2020, 30, 1909983
[3] Y. Peng, T. N. Huq, J. Mei, …, R. L. Z. Hoye, V. Pecunia, Adv. Energy Mater. 2020, Under Revision
[4] R. A. Jagt, …, R. L. Z. Hoye, J. Mater. Chem. C 2020, 8, 10791

Zinc phosphide (Zn₃P₂) is an earth-abundant semiconductor material capable of addressing the rising demand for low-cost and efficient optoelectronic devices. Its 1.5 eV direct bandgap, long minority carrier diffusion length (5-10 μm), and high absorption coefficient make it very promising for applications as an absorber in thin films solar cells. Furthermore, it has the added benefit of being a binary phase over more complex multinary materials.
Two crucial challenges need to be overcome to produce high-efficiency devices. First, poor crystal quality when grown on commercially available substrates due to high interface defect densities and high coefficient of thermal expansion. This challenge has been recently solved in our laboratory, as we have demonstrated different fabrication methods of high-quality reproducible thin films, with the possibility of facile transfer onto commercially available substrates.[1] [2] The second challenge is the control of bulk defects and doping, which is crucial for optoelectronic performance. Usually, uncontrolled growth results in p-doped material due to the formation of intrinsic acceptor defects.
To solve the challenge of defect control, we have investigated structural defects in the room-temperature-stable phase of zinc phosphide (α-Zn₃P₂, space group P4₂/nmc) using Raman spectroscopy. For that purpose, we have first fully characterized the lattice dynamics of zinc phosphide by combining density functional theory (DFT) calculation with polarized micro-Raman measurements on single-crystalline zinc phosphide. This is the first complete analysis of phonons in Zn₃P₂, essentially providing a reference for further characterization of this material. We identified 33 of the 39 expected Raman lines and have calculated the atomic displacement in all vibrational modes[3].
Then, using factorial and combinatorial studies of Zn₃P₂, along with the micro-Raman spectroscopy and previously mentioned results, has allowed us to probe different structural defects within Zn₃P₂. Systematic changes in the peak position, intensity, and widths in correlation with the presence of Zn and P interstitials and vacancies are presented and discussed. This work demonstrates the possibility of defect engineering in Zn₃P₂, which is of the foremost importance for the improvement of solar cell performance, as well as the potential of Raman scattering for point defect assessment in this system.

[1] R. Paul et al., “Van der Waals Epitaxy of Earth-Abundant Zn₃P₂ on Graphene for Photovoltaics,” Cryst. Growth Des., vol. 20, no. 6, pp. 3816–3825, 2020.
[2] S. Escobar Steinvall et al., “Multiple morphologies and functionality of nanowires made from earth-abundant zinc phosphide,” Nanoscale Horizons, vol. 5, pp. 274–282, 2020.
[3] E. Z. Stutz et al., “Raman spectroscopy and lattice dynamics calculations of tetragonally-structured single crystal zinc phosphide (Zn₃P₂) nanowires,” Nanotechnology (accepted)

Adamantine-type compounds, including kesterites, are currently the most promising material for a fully inorganic thin film photovoltaic technology that is free of critical raw materials and thus provides sustainable solutions.
Adamantines are compounds crystallizing in a structure, in which every atom is tetrahedrally bonded to four nearest neighbours [1]. To find new compounds for absorber layers as well as window materials for thin film solar cells, the Α^IB^IIIX^VI₂ chalcopyrite compound family, belonging to the Adamantine compounds, can be extended by chemical substitution. Coming from the chalcopyrite-type structure (space group I-42d), one A⁺¹ (occupying the structural site 4a) and one B⁺³ cation (occupying the structural site 4b) are replaced by one four-valent cation (A⁺¹ + B⁺³ ↔ C⁺⁴). This substitution will consequently lead to the formation of cation vacancies. In this way a quaternary “defect adamantine” [1] such as I-[]-III-IV-VI₄ is formed.
.
Here we report on the growth of Cu[]BCX₄ single crystals, with B = Ga, In, C = Ge, Sn and X = S, Se such as CuGaGeS₄, CuInGeS₄, and CuGaSnS₄, and the analysis of their structural as well as optoelectronic properties.
The single crystals were grown by chemical vapor transport using iodine as transport agent. The evolved material and the grown crystals were characterized by X-ray fluorescence spectroscopy (XRF) for chemical composition as well as by X-ray diffraction (XRD) at 298K. The obtained powder patterns were analyzed by LeBail refinement to determine the lattice parameter. The chalcopyrite-like crystal structure was used as a structural model in the refinement.
The band gap energy of the material was revealed by solid-state UV/Vis reflectance spectroscopy. The obtained data were analyzed using the Kubelka−Munk pseudo-absorption function and the Tauc-plot method. The structure-property relations of CuGaGeS₄ and CuGaSnS₄ will be discussed in detail in the presentation.
We present first results on the growth and characterization of the novel compound semiconductor CuGaGeS₄ as well as CuGaSnS₄. These indicating a potential for future applications.

[References]
[1] B.R. Pamplin, Prog. Crystal Growth Charact. vol.3, pp. 179-192 (1981)

Zinc phosphide (Zn₃P₂) is an earth-abundant semiconductor with optoelectronic properties suitable for photovoltaic applications, such as a direct bandgap (1.50 eV), long minority carrier diffusion lengths (up to 10 μm), and high absorption in the visible range.^1,2 However, applications of the material has been limited due to challenges introduced by the lack of substrate with a matching lattice constant (for epitaxial growth) and thermal expansion coefficient, and the doping due to off-stoichiometry composition resulting in the formation of self-interstitials. Molecular beam epitaxy overcomes this by providing a low-temperature growth route with precise control over the composition, facilitating the fabrication of high-quality zinc phosphide.
A technique which can be used to further improve the material quality is selective area epitaxy (SAE).^3,4 Using a nano-patterned oxide mask on indium phosphide substrates we demonstrate the selective growth in the holes, limiting the interface area to improve its quality.⁵ The SAE grown zinc phosphide first grows as nanopyramids. By controlling the growth time and pitch they can be made to overgrow the oxide and coalesce, forming a thin film. The structural properties of the material and the epitaxial relationship were investigated through transmission electron microscopy, which also showed the influence of hole size on the composition of the zinc phosphide and the formation of rotational core-shell structures. Furthermore, using conductive atomic force microscopy (CAFM) and photoluminescence spectroscopy (PL) we evaluated the functional properties of the material. The CAFM measurements showed a diode-like behavior when the zinc phosphide is grown on an n-type substrate, whilst the PL showed clear bandgap emission at 1.53 eV, ideal for photovoltaic applications. This approach has shown great potential in producing high-quality zinc phosphide, and upcoming studies will use it to transfer the growth to an earth-abundant substrate (silicon), and making prototype photovoltaic devices based on SAE grown zinc phosphide.
References
1. G. M. Kimball et al. Appl. Phys. Lett., 95, 112103 (2009).
2. M. Y. Swinkels et al. Phys. Rev. Applied, 14, 024045 (2020).
3. M. Friedl et al. Nano Lett., 18, 2666–2671 (2018).
4. C. –Y. Chi et al. Nano Lett., 13, 2506–2515 (2013).
5. F. Glas Phys. Rev. B, 74, 121302 (2006).

The effects of Na and RbF alkali treatment on the metastability behavior of Cu(In,Ga)Se₂ solar cells with two different buffers have been investigated with stress factors of heat, junction bias and illumination.

For CdS buffer, four device types with and without Na or RbF treatments have been subjected to heat- and light-soaking under open- and short-circuit (OC, SC) junction bias. Low-Na devices show a higher bandgap due to increased minimum Ga content, higher recombination current and lower open-circuit voltage (V_OC). Devices with RbF post-deposition treatment (PDT) show an improvement in net doping density ~ 10¹⁶ cm^-3, V_OC and efficiency. Heat- and light-soaking under OC junction bias provokes an increase in net carrier concentration and V_OC irrespective of the alkali treatments. After SC stress, a decrease in V_OC and net carrier concentration are observed which can be stabilized by RbF-PDT. ToF-SIMS measurements reveal an increase in Na and O concentration in CIGS for baseline and reduced-Na devices, respectively, after OC stress. Oxygen concentration in CdS decreases after heat- and light-soaking for devices without RbF-PDT, whereas it remains unchanged for devices with RbF-PDT. The atomic concentration profiles in CIGS significantly stabilize as a function of stress with the addition of RbF-PDT.

Devices under open-circuit (OC) light soaking show an increase in V_OC with CdS buffer, however, V_OC decreases for ZnOS buffer devices. All devices show an increase in net carrier concentration after OC light soak irrespective of RbF-PDT or the buffer layers, even though ZnOS buffer devices show degradation. Under short-circuit (SC) light soak CdS buffer devices with RbF-PDT show stabilization of V_OC attributed to the increase in ionized acceptors in ordered vacancy compound (OVC) or buffer donor density. ZnOS buffer devices, on the other hand, show a decrease in V_OC under both OC and SC light soak irrespective of RbF treatment. A comprehensive discussion of effect of interface defects, conduction band offset and near interface doping density will be presented for ZnOS based CIGS solar cells.

The solution processing of thin-film solar cells is of great interest due to the potential of this method to decrease costs and dramatically increase manufacturing throughput compared to the vacuum analogue. By utilizing hydrazine as a solvent, researchers have fabricated solution-processed Cu(In,Ga)(S,Se)₂ devices with efficiencies greater than 17%. However, hydrazine is highly toxic and explosive, potentially limiting its use is large-scale manufacturing. Several alternative solution processing methods are under investigation, though none have matched hydrazine in efficiency. Because of the difficulty in producing a soluble selenium species, these methods will often deposit a nanocrystalline sulfide precursor film. The precursor film is then converted to a large-grain selenide film by heating in a selenium atmosphere. During this conversion, grain growth is achieved as the nanocrystalline sulfide material dissolves in a liquid selenium front and then recrystallizes into micron-scale selenide grains. This process often results in a distinct “fine-grain” layer, particularly as film thickness increases, limiting researchers to absorber layer of 1-1.5μm in thickness.
The amine-thiol solvent system has emerged as one of the most promising alternatives to hydrazine, producing Cu(In,Ga)(S,Se)₂ devices with efficiencies above 15%. With its ability to dissolve metals and metal chalcogenides, the amine-thiol solvent system is able to avoid potential anionic impurities that arise when using other metal salts. Additionally, because selenium and many metal selenides can be dissolved in amine-thiol, this system has the potential to tune the sulfur to selenium ratio of the precursor film, as is done in the hydrazine system. One challenge, however, is that the metal-thiolates produced in this dissolution can easily decompose into metal sulfides upon heating, often leading to significant sulfur content in the precursor film, along with some residual carbon.
In this research, we will discuss investigations into amine-thiol dissolutions of metal chalcogenide species through the use of ¹H-NMR, Raman, and ESI-MS. By developing a deeper understanding of the chemistry in this reactive dissolution, we can then make carefully controlled modification to the resulting metal complexes. We will present the various methods we have developed to alter the metal-chalcogenide bonding in the soluble precursor species, thereby changing the decomposition products and allowing for additional control over precursor film fabrication. We will also discuss the implications of the precursor film chalcogen composition on the further processing of the absorber layer, particularly as it relates to grain growth and will show how it can help reduce the carbon content in the film. To this end, we have utilized this new chemistry to produce the first ever 2μm-thick, solution-processed absorber films with no fine-grain layer. Finally, we will discuss how these strategies can be implemented to further the ultimate goal of high efficiency, solution-processed solar cells.
In conclusion, this research focuses on developing an understanding of the fundamental chemistry of the metal complexes created with the amine-thiol solvent system and connecting this knowledge with the practical application of solution-processed thin-film PV.

The conversion efficiency of Cu(In,Ga)Se₂-based (CIGSe) thin-film solar cells has reached values above 23% on a laboratory scale. To further increase the solar cell performance, higher open-circuit voltages can be achieved by increasing the Ga/(Ga+In) (GGI) ratio to create a larger absorber band gap and/or by applying a post-deposition treatment (PDT) to the absorber surface. Consequently, it is crucial to understand how a high GGI and a PDT affects the chemical structure and band alignment at the buffer/absorber interface.

We have investigated in-line deposited and industrially relevant CIGSe absorbers with high bulk and surface GGI ratios. Two sample series, one with and one without RbF PDT, were prepared, and their interfaces with a chemical bath deposited CdS buffer layer were characterized with a combination of electron and soft x-ray spectroscopies. For this purpose, laboratory-based x-ray and UV photoelectron spectroscopy (XPS and UPS, respectively), inverse photoemission (IPES), and x-ray-excited Auger electron spectroscopy (XAES) were combined with synchrotron-based soft x-ray emission spectroscopy (XES). A detailed picture of the electronic and chemical structure at and near the interface can be painted, which will be discussed in relation to the performance of the corresponding CIGSe solar cells.

The reliability of photovoltaics is commonly studied at the module level. Many reliability problems do originate from module attributes, such as metal interconnections to cells, junction boxes, etc. However, significant work in reliability can also be done before an entire module is designed. In this talk I summarize how we investigated three reliability concerns in Cu(In,Ga)Se₂ (CIGS) photovoltaics at the cell level. The first is metastability, which in CIGS presents as an increase in V_OC over time during prolonged light exposure. Although increasing V_OC is not detrimental to module performance, it does complicate accurately measuring a module’s performance and subsequent rating. The second is shading-induced hot spots where cells are irreversibly damaged by reverse voltage. This can happen in less than 1 second of partial shading, and the damage is often visible to the eye. Finally, we looked at potential-induced degradation (PID) or a decrease in performance caused by voltage differentials which drive ion diffusion. PID is a growing concern as the number of modules wired in series rises, hence increasing voltage differentials. Examining these perceived problems required developing robust measurement protocols including the fabrication of novel testing structures. These efforts ensured that we were incorporating the right metrics for investigating the phenomena and provided insights for improvement strategies.
NREL is in a unique position to test reliability at the cell level as we are making improvements to PV technology. If we look at reliability earlier in the research cycle, we have the potential to avoid commonly known module reliability problems before cell changes are implemented on a large scale. Cell-level reliability studies could thus lower the rates of module failures in the field and provide confidence to investors that new technologies will perform as advertised.

In recent years, passivation of Cu(In,Ga)Se₂ (CIGSe) surface and/or interfaces with insulator materials has attained lots of attention to further increase efficiency. Such passivation is usually achieved with a thin layer of Al₂O₃ (<50 nm) deposited either before or after fabrication of CIGSe, most predominantly by atomic layer deposition method. The application of Al₂O₃ layer has been shown to reduce the number of electrically active defects at the semiconductor surface, which in turn lowers surface recombination velocities. This led to a decrease in interface recombination and resulted in improved efficiency. Passivation with Al₂O₃ has also been applied successfully to other thin film absorbers such as Cu₂ZnSn(S,Se)₄ and CdTe.
In this report, we present a novel approach to improve solution-processed CuIn(S,Se)₂ (CISSe) solar cell efficiency by in-situ incorporation of Al₂O₃ . Specifically, AlNO₃ was added to inks containing CuCl, InCl₃, and thiourea dissolved in methanol. After spin coating of these solutions performed in air, samples were subjected to a selenization process. Our study showed that the Al formed amorphous nanosized-Al₂O₃ covering parts of the film top and bottom surfaces as well as within bulk and grain boundaries. Power conversion efficiency (PCE) as high as 11.6% was measured on cells using CISSe incorporated with Al₂O₃, a value higher than that measured on CISSe fabricated from Al-free ink (maximum PCE: 8.3%). This efficiency boost stemmed primarily from an increase in both in open-circuit voltages and fill factors. Defect passivation via this in-situ formed Al₂O₃ is thought to play a major role in the improvement of the solar cell performance through defect passivation, while permitting Cd diffusion to take place at the absorber sub-surface. Details of the fabrication process, materials, and photovoltaic characterization will be discussed.

An area of interest in the thin film photovoltaics community that has grown in recent years is the alloying of Cu(In,Ga)(S,Se)₂ (CIGS) with silver to create (Ag,Cu)(In,Ga)(S,Se)₂ (ACIGS). Alloying of silver in CIGS has resulted in favorable device characteristics such as increasing the bandgap to optimize devices for top cells in tandem architectures, reduced structural disorder (observed by lower Urbach energies) and lower open circuit voltage losses. However, we anticipate that the alloying of silver may only complicate the fabrication of CIGS devices by introducing a sixth element that must be controlled in the material. Indeed, a recent report¹ showed that high bandgap ACIGS devices have a low tolerance to off-stoichiometry, and another recent study² has suggested that the ACIGS system may be prone to phase segregation at certain temperatures and compositions. These observations foreshadow possible complications in the controlled scale-up of ACIGS devices.
In this work, we propose that some advantages of silver incorporation in chalcopyrite materials could be harnessed differently, by moving away from the complex stoichiometry of ACIGS to the far simpler AgInSe₂ (AIS). AIS has been studied only briefly in the literature, but initial findings suggest that the material is worth investigating further as a photovoltaic absorber material. By eliminating copper and gallium, the possibility for phase segregation is greatly reduced and the challenges presented by ACIGS are avoided. In addition, AIS has a direct bandgap in the ideal range for single-junction cells at 1.24eV, has been shown to have a carrier mobility much higher than CIGS, and has been predicted to have higher defect formation energy than CIGS.
To explore the potential of this material, we have fabricated thin films of AgInS₂ through the amine-thiol solution processing route by dissolving silver sulfide and metallic indium in a monoamine-dithiol solvent mixture, creating a solution of bis(1,2-ethanedithiolate) indium (III) and oligomeric Ag(I) thiolates. These solutions were cast on to molybdenum-coated glass substrates and annealed to form orthorhombic AgInS₂ films. These were then selenized in an elemental selenium atmosphere to create chalcopyrite thin films.
Our preliminary electronic characterization of AgIn(S,Se)₂ thin films suggest that the carrier concentration is roughly 1 x 10¹³ cm^-3. In previous literature³ it has been shown that the carrier concentration can be controlled by changing the concentration of selenium in the annealing atmosphere. We find that carrier mobilities and photoluminescence characteristics are superior to similarly fabricated CIS thin films. From these initial results, we expect AIS to have superior electronic properties and defect characteristics when compared to CIS/CIGS films.
A side effect of the use of silver instead of copper is a change of the majority carriers from holes (p-type) to electrons (n-type). Taking full advantage of AIS will require a redesign in the optimal device architecture, as incorporation in to a heterojunction with CdS or Zn(O,S,OH) (as is done with CIGS) is no longer a viable approach. Our ongoing work is exploring possible junction partners for AIS to propose viable device architectures for further investigation.
In this work, we have developed a solution processing route for the synthesis of AgIn(S,Se)₂ thin films and demonstrated exciting electrical properties of this under-studied material. Along with our results, we point out the promise that this material holds for the thin film PV community.

(1) Keller, J.; Stolt, L.; Sopiha, K. V.; Larsen, J. K.; Riekehr, L.; Edoff, M. Sol. RRL 2020, 2000508 (1), 2000508
(2) Sopiha, K. V.; Larsen, J. K.; Donzel-Gargand, O.; Khavari, F.; Keller, J.; Edoff, M.; Platzer-Björkman, C.; Persson, C.; Scragg, J. J. S. J. Mater. Chem. A 2020, 8 (17), 8740
(3) Tell, B.; Shay, J. L.; Kasper, H. M. J. Appl. Phys. 1972, 43 (5), 2469

The development of flexible solar cells is necessary for achieving market competitiveness through the implementation of low cost solar cells and for applying customized business models, such as building integrated photovoltaics (BIPV) and mobile applications. In addition, low-cost flexible substrates can be used to lower manufacturing costs, which can contribute to the expansion of the renewable energy market by shortening the energy payback time. Solar cells with the CZTS-based (Cu₂ZnSnS₄, Cu₂ZnSnSe₄, Cu₂ZnSn(S,Se)₄) absorbers are advantageous for lowing the cost, but the development of high efficiency solar cells based on flexible substrates has been relatively unexplored. In this work, flexible CZTSSe solar cells applying the Mo and SUS foil substrate were developed. When we used the metal precursor, double CZTSSe layer, voids and ZnSSe secondary phase were observed. The ZnSSe layer formed during soft annealing plays important role as a layer that blocks the volatilization of Zn. Large void can cause the formation of unwanted secondary phase and non-uniform composition. Applying SnS precursor is a good strategy as a way to suppress void formation. To optimize the processes of the flexible CZTSSe thin film solar cells, Na doping, back contact modifying, blister formation suppression were studied. Applying NaF decreased the defect level and Voc deficit(Eg – qVOC). This improvement can be explained by the associated decrease in defects, which are considered the recombination centers in the CZTSSe absorber layer. The weak adhesion between CZTSSe and Mo layers and the high stress of CZTSSe layer can form blisters. The increase of adhesion between CZTSSe and Mo layers can prevent the formation of blister. The best power conversion efficiency of the prepared CZTSSe solar cell on flexible Mo foil is 11.59%, which is world top efficiency record on flexible substrate.

Acknowledge: This work was supported by the DGIST R&D Program of the Ministry of Science and ICT (19-BD-05) and the Korea Institute of Energy Technology Evaluation and Planning (KETEP) and the Ministry of Trade, Industry & Energy(MOTIE) of the Republic of Korea (No. 20173010012980).

Chalcogenide CdTe and Cu(In,Ga)(S,Se)₂ (CIGS) and halide-perovskite-related (e.g. CH₃NH₃PbI₃) semiconductors represent the fastest growing commercial and the most vigorously studied thin-film photovoltaic (PV) technologies, respectively. Alternatives to these semiconductors are also being sought to mitigate toxicity and elemental abundance concerns, for improved compatibility with pervasive scaling of the technologies. Inherent to these pursuits, is the need to understand materials and device limitations, especially as they relate to defects. In this talk, we will consider additive engineering and doping in CH₃NH₃PbI₃-related semiconductors [1,2], as well as proposed defect-resistant Cu₂Ba(Sn,Ge)(S,Se)₄ materials [3,4], and employ a range of characterization techniques, including carrier-resolved photoHall, conventional Hall effect, terahertz spectroscopy and other time-resolved spectroscopies, to understand associated properties and performance-limiting or -enhancing mechanisms.

References:
[1] O. Gunawan, S. R. Pae, D. M. Bishop, Y. Virgus, J. H. Noh, N. J. Jeon, Y. S. Lee, X. Shao,
T. Todorov, D. B. Mitzi, B. Shin, Nature 575, 151 (2019).
[2] J. Euvrard, O. Gunawan, D. B. Mitzi, Adv. Energy Mater. 9, 1902706 (2019).
[3] D. Shin, B. Saparov, D. B. Mitzi, Adv. Energy Mater. 7, 1602366 (2017).
[4] G. C. Wessler, T. Zhu, J.-P. Sun, A. Harrell, W. P. Huhn, V. Blum, D. B. Mitzi, Chem. Mater. 30, 6566 (2018).

The solar cells based on the emerging organic-inorganic hybrid halide perovskite is progressing rapidly in the last decade, but the commercialization of the perovskite solar cells still faces significant challenges. The major issues are the poor long-term stability and toxicity. In light of this, materials design and screening of novel stable perovskites is becoming an important research direction. In contrast to conventional trail-and-error processes, materials genome techniques and high-throughput calculations have played important roles in this area and accelerated materials discovery. In this talk, I will present a review that summarizes our recent progress in this field, mainly focusing on four classes of perovskites including AM²⁺X₃ halide single perovskites, AM⁴⁺Y₃ chalcogenide single perovskites, A₂M⁺M³⁺X₆ halide double perovskites, and A₂M³⁺M⁵⁺Y₆ chalcogenide double perovskites. The stability, electronic, optical, and defect properties of these materials in terms of their applications as solar cell absorbers are discussed.

Majority and minority carrier properties such as type, density and mobility represent fundamental yet difficult to access parameters governing semiconductor device performance, most notably solar cells. Obtaining this information simultaneously under light illumination would unlock many critical parameters such as recombination lifetime, recombination coefficient, and diffusion length, however this goal has remained elusive. We demonstrate here a new carrier-resolved photo-Hall technique that rests on a new equation relating hole-electron mobility difference (△μ), Hall coefficient (H), and conductivity (σ), and a rotating parallel dipole line ac-field Hall system with Fourier/lock-in detection for clean Hall signal measurement (Nature 575, 151, 2019). We successfully apply this technique to a world-record-quality perovskite film and map the results against varying light intensities, demonstrating unprecedented simultaneous access to the above-mentioned parameters.

The interest in perovskite photovoltaics has significantly increased over the last few years. High power conversion efficiencies (PCE) and low-cost manufacturing make perovskite PVs a very promising candidate for future applications. Despite the very advantageous features of perovskite materials, several issues still need to be solved before the commercialization of perovskite solar cells (PSCs). The main challenges of bringing perovskite technologies to the market are (i) scaling-up of the cells and modules dimension, (i) usage of lead in the perovskite solar panels and (iii) stability of the PSC modules. These three challenging topics will be discussed in the presentation, and possible solutions to overcome these issues will be proposed. The implementation of the proposed measures will help to demonstrate the feasibility of high-volume production. It is an important milestone towards the industrial manufacturing of perovskite photovoltaics and their future commercialization.

PV-powered vehicle application are very attractive for reducing CO₂ emission and creation of new market. Development of high-efficiency (> 30%) and low-cost solar cell modules is necessary [1]. The II-VI compound, chalcopyrite and kesterite based tandem solar cells are thought to be one of the most promising PV devices because of high-efficiency and low-cost potential. However, efficiencies of II-VI compound, chalcopyrite and kesterite based tandem solar cells (27.2% with III-V/CIGS 3-junction, 24.2% with perovskite/CIGS 2-junction, 16.8% with CdZnTe/Si 2-junction, 15.3% with CdTe/CISe 2-junction, 8.5% with CGSe/CIGSe 2-junction, 3.5% with CZTS/Si 2-junction tandem solar cells) are lower compared to 37.9% with III-V 3-junction and 35.9% with III-V/Si 3-junction tandem solar cells. Therefore, it is necessary to clarify and reduce several losses of tandem solar cells. This paper presents high efficiency potential of II-VI compound, chalcopyrite and kesterite based tandem solar cells analyzed by using our analytical procedure [2] and discusses about non-radiative recombination losses in those tandem solar cells.
One of problems to attain the higher efficiency II-VI compound, chalcopyrite and kesterite based tandem solar cells is to reduce non-radiative recombination loss. The open-circuit voltage V_oc drop compared to bandgap energy (E_g/q - V_oc) is dependent upon non-radiative voltage loss (V_{oc, nrad}) that is expressed by external radiative efficiency (ERE). Open-circuit voltage is expressed by

where the second term shows non-radiative voltage loss, is radiative open-circuit voltage and 0.23V was used as △ (= E_g/q - ) in this study.
Correlation between V_oc values for II-VI compound, chalcopyrite and kesterite based tandem solar cells reported in the references and ERE values estimated by eq. (1) suggests that II-VI compound, chalcopyrite and kesterite based tandem solar cells have larger non-radiative loss and further improvements in efficiency are thought to be possible by improving external radiative efficiency (ERE). The ERE values estimated in this study are 0.00116% for 27.2% III-V/CIGS 3-junction, 0.024% for 24.2% perovskite/CIGS 2-junction, 0.00162% for 16.8% CdZnTe/Si 2-junction, 1.17x10^-5% for 15.3% CdTe/CISe 2-junction, 5.87x10^-8% for 8.5% CGSe/CIGSe 2-junction, and 8.05x10^-10% for 3.5% CZTS/Si 2-junction tandem solar cells compared to 2.2% for 37.9% III-V 3-junction and 1.62% for 35.9% III-V/Si 3-junction tandem solar cells. According to our analysis [3], the 2-junction and 3-junction tandem solar cells have potential efficiencies of more than 36% and 42%, respectively.
Several non-radiative recombination losses such as bulk recombination, interface recombination and so forth in tandem solar cells are discussed in this paper. In addition, resistance loss of tandem solar cells including inter-connection is discussed in this study.

References
[1] M. Yamaguchi et al., Prog. Photovolt. (2020) https://doi.org/10.1002/pip.3343.
[2] M. Yamaguchi et al., J. Mater. Res. 32, 3445 (2017).
[3] M. Yamaguchi et al., J. Phys. D. Appl. Phys. 51, 133002 (2018).

Organic-inorganic hybrid halide perovskite solar cells (HPSCs) have attracted immense attention because of excellent optoelectronic properties and a record power conversion efficiency (PCE) that has surpassed 25.5% within a few years. HPSCs are competitive with conventional thin-film and c-Si photovoltaic technologies because of their high efficiency, tunable bandgap, high absorption coefficient, and easy and low-cost fabrication. However, despite the very high efficiency already attained by (FA,MA,Cs)Pb(I,Br)₃ single-junction solar cells, the toxicity of Pb and stability of these materials are challenges to commercialization of this technology. Many Pb-free technologies have been developed such as GeI₂ doped Sn based perovskites with promising PCE > 13%. However, a major challenge for Pb-free perovskites is poor power conversion efficiencies as divalent Sn²⁺ and Ge²⁺ tend to oxidize into 4+ oxidation state, generating excessive defects (Sn and Ge vacancies), and result in short carrier diffusion lengths.

Cs₂PtI₆ is a high absorption coefficient and optimal bandgap (E_g = 1.4 eV) material, which has shown promising power conversion efficiency of > 12%[1]. The material has shown high absorption coefficient, carrier lifetime > 2 μs and stability to heat and light soaking stress. ITO/SnO₂/Cs₂PtI₆/PCBM/Carbon devices have also shown to be stable in air without encapsulation for 60 days[2]. Hamdan and Chandiran have demonstrated Cs₂PtI₆ photoanode to be stable under AM1.5G illumination and extreme pH for over 12 hours, making the material promising for photoelectrochemical devices[3]. Although Pt offers an ideal model system, the search for a low-cost, stable, and high performance Pb-free halide perovskite continues. Replacement of Pt with Ni for A₂BX₆ is expected to reduce the bandgap with the reduction in cation size[4]. Here, we explore the effect of Pt precursor, solvents, and Pt-Ni solid solution on optical bandgap, absorption coefficient, structure, and morphology of Cs₂PtI₆ films.

Cs₂PtI₆ films processed with PtI₄ precursor in a solvent mixture of DMF:DMSO[5] results in crystalline films with an optical bandgap of ~ 1.4 eV. PtI₂ precursor processed with a solvent mixture of DMF:GBL also results in structure and optical bandgap albeit with higher film porosity and preferred crystallographic orientation. Both these processes result in low oxygen content in the films as confirmed by EDS and XPS. On the other hand, films processed with PtI₂ precursor in DMF:DMSO solvent mixture results in oxygen containing films with a bandgap measured as 2.2 eV using UV-Vis-NIR and photoluminescence spectroscopy. Intermixing of B cation with a 50 at% of Pt and 50 at% of Ni results in an optical bandgap of 1.48 eV with the crystal structure of Cs₂PtI₆. Detailed composition analysis shows oxygen is incorporated at anion site resulting in a mixed halide-chalcogenide phase with a tunable bandgap with oxygen concentration. Cs₂(Pt_xNi_1-x)(I_yO_z)₆ offers an interesting material with tunable bandgap for optoelectronic and sensor applications.


[1] Schwartz, D.; et al. Phys. Status Solidi Rapid Res. Lett. 2020, 14, 2000182.
[2] Yang, S.; et al. ACS Appl. Mater. Interfaces 2020, 12, 40, 44700–44709.
[3] Hamdan, M.; Chandiran, A. K. Angew. Chem., Int. Ed. 2020, 59 (37), 16033– 16038.
[4] Cai, Y.; et al. Chem. Mater. 2017, 29, 7740– 7749.
[5] Abbreviations: DMF = Dimethylformamide, DMSO = Dimethyl sulfoxide, GBL = γ-Butyrolactone, EDS = Energy-dispersive X-ray spectroscopy, XPS = X-ray photoelectron spectroscopy, UV-Vis-NIR = Ultraviolet-Visible-near-Infrared

Owing to the polycrystalline nature of Cu(In,Ga)Se₂ (CIGS) thin films in high-efficiency solar cells, the impact of grain boundaries in these thin films on the photovoltaic performance has always been of concern. The present contribution will give an overview on the current understanding of this matter and show that even in high-efficiency devices, the enhanced nonradiative recombination at CIGS grain boundaries leads to a V_oc loss of 20-30 mV. In contrast, nonradiative recombination at dislocations does not seem to be enhanced with respect to the CIGS bulk. Differences between grain boundaries and dislocations will be discussed as well as the issue of potential passivation of grain boundaries by, e.g., alkali metals.

Alkali post-deposition treatments in polycrystalline Cu(In,Ga)Se₂ absorbers have demonstrated to yield an increase in open-circuit voltage. The origin of such improvement has been attributed to reduced non-radiative recombination in the bulk, where the underlying mechanism has been proposed to be due to passivation of charged defects at grain boundaries as a result of the accumulation of heavy alkali elements conveying a reduction in band bending and tail states¹. However, the density of tail states is also impacted by disorder in the crystal itself, and in the case of Cu(In,Ga)Se₂, prompted by electrostatic potential and bandgap fluctuations. The Urbach energy describes the extension of the states into the band gap, and has recently been proposed as an indictor of the absorber’s quality since a correlation between Urbach energy and open-circuit voltage deficit has been observed².
In this contribution, we study the influence of different alkali post-deposition treatments on the crystal itself. We measure the Urbach energy and the quasi-Fermi level splitting (qFls) of NaF and KF-treated Cu(In,Ga)Se₂ single crystals grown by metalorganic vapor phase epitaxy. Our results show that the post-deposition treatments with NaF, and to a lesser extent KF, can effectively decrease the Urbach energy. In a similar way, an improvement in qFls is obtained after the post-deposition treatments. The fact that alkali incorporation into the grains have an impact on the Urbach energy, suggests that the decrease in tail states observed in polycrystalline absorbers with heavy alkali post-deposition treatments is not exclusively grain boundary-related. Both effects: the reduction of tail states and the increased qFls are explained by increased net-doping due to the alkalis. The effects on grain boundaries have to be complemented by the effects on the grain interiors, to understand the observed improvements and to devise ways for further efficiency improvements.
References
[1] S. Siebentritt et. al., Advanced Energy Materials 2020, 10, 1903752.
[2] S. De Wolf, J. Holovsky, S.-J. Moon, P. Löper, B. Niesen, M. Ledinsky, F.-J. Haug, J.-H. Yum, C. Ballif, The Journal of Physical Chemistry Letters 2014, 5, 1035.

The photoactive band gap energy E_g of a solar cell device is determined by the band gap of the absorber layer material which has to be optimized to reach high conversion efficiencies.
One of the reasons for the success of Cu(In,Ga)Se₂ (CIGS) based thin film solar cells is the remarkable flexibility of its chalcopyrite crystal structure, accepting large stoichiometry deviations. The structural flexibility is a key for the quaternary kesterite-type compounds too. Cu₂ZnSn(S,Se)₄ (CZTSSe) based thin film solar cells reach the highest efficiency with a Cu-poor and Zn-rich composition of the absorber layer. Such deviations from the stoichiometric composition lead to the formation of intrinsic point defects (vacancies, anti-sites, and interstitials) in the compound, which significantly influence the electrical and optical properties of the material.
The existence of off-stoichiometric CZTS, CZTSe and CZTSSe has been shown in extended and systematic studies of powder series [1,2]. The off-stoichiometric composition of the kesterite-type phase corresponds to certain point defects. This correlation is the basis for defining off-stoichiometry types referred to as A–L [2,3]. For example, A-type off-stoichiometric kesterites, which are Cu-poor and Zn-rich, contain copper vacancies (V_Cu) and Zn_Cu anti-sites. The off-stoichiometry type model has been confirmed experimentally on the basis of neutron diffraction studies [2, 4].
We have determined the band gap energy of these series of off-stoichiometric kesterite-type compounds (powder) by diffuse reflectant spectroscopy. The optical band gap was extracted from Tauc plots obtained by plotting (F(R)*hν)² versus the photon energy [5]. We can show, that in off-stoichiometric CZTS and CZTSe the band gap energy varies strongly in dependence on the off-stoichiometry (which can be expressed by the cation ratios) of the kesterite-type phase. For instance in A-B type CZTSe the band gap energy E_g increases from 0.88 eV to 1.0 eV in dependence on decreasing Cu/(Zn+Sn) and increasing Zn/Sn ratio (Cu/(Zn+Sn) decreases from 0.96 to 0.81 and Zn/Sn increases from 1.04 to 1.15).
Comparing such band gap energy variations in Cu-poor CZTS and CZTSe with those in off-stoichiometric Cu-poor chalcopyrites (like CuInS₂, CuInSe₂ and CuGaS₂) then surprisingly one finds a different behavior. When looking at the same off-stoichiometry range the band gap energy variations in Cu-poor chalcopyrites are very small.
This can have serious consequences for kesterite-based solar cell device. The absorber layer shows generally an inhomogeneous chemical composition thus different off-stoichiometric kesterite phases, this means different off-stoichiometry types, occur coexisting. This may cause remarkable variations in band gap energy within the absorber layer.

[1] L.E. Valle Rios, K. Neldner, G. Gurieva and S. Schorr, J. Alloys Compd. 657 (2016) 408.
[2] G. Gurieva, L.E. Valle Rios, A. Franz, P. Whitfield and S. Schorr, J. Appl. Phys. 213 (2018) 161519.
[3] A. Lafond, L. Coubrac, C. Guillot-Deuden, P. Deniard, S. Jobic, Z. Anorg. Allg. Chem. 638 (2012) 2571.
[4] S. Schorr, G. Gurieva, M. Guc, M. Dimitievska, A. Perez-Rodriguez, V. Izquierdo-Roca, C. Schnohr, J. Kim, W. Jo, J. M. Merino, J.Phys.: Energy 2 (2020) 012002.
[5] Tauc, J., R. Grigorovici and A. Vancu, Physica Status Solidi 15 (1966) 627.

Bandgap tunability of copper indium gallium disulfide Cu(In,Ga)S₂ from 1.55eV (CuInS₂) to 2.5eV (CuGaS₂) makes it an excellent top cell option for use in a tandem devices.[1] An open-circuit voltage (V_OC) of 973 mV and a power conversion efficiency of 15.5% Cu(In,Ga)S₂ demonstrates its promises for tandem solar cells in combination with silicon or Cu(In,Ga)Se₂. [2,3] However, even the best devices still suffer from a significant V_OC deficit (~600 mV) compared to its bandgap owing to non-radiative recombination in the device. Further advancements require a better comprehension of the recombination channels limiting the V_OC of the device, whether they lie in the bulk or at the interface.
We present investigation on a composition series of Cu(In,Ga)S₂ absorbers grown under different Cu concentration i.e. different [Cu]/[In+Ga] (CGI) ratio with bandgap (E_g) around ~1.6 eV, suitable for combination with record efficiency bottom cells. We have shown in the past that a variation in CGI of CuInS₂ drastically influences the recombination in the bulk. [4] Here we study higher band gap films with a band gap gradient. The opto-electrical properties are probed with photoluminescence measurements, and the electrical properties by current-voltage and capacitance measurements on solar cells fabricated with either CdS or Zn(O,S) buffer. The results of low temperature photoluminescence demonstrate suppression of deep defects as the CGI is decreased from 1.02 to 0.93, consequently leading to a maximum quasi-Fermi level splitting (qFLs) of 972 meV. Current-voltage measurements of devices prepared with CdS and Zn(O,S) buffer layer exhibit efficiencies close to 13 % with a V_oc deficit of 650 mV in Cu-poor (CGI<1) Cu(In,Ga)S₂ devices. Temperature dependent V_OC measurements show the presence of strong front interface recombinations in all devices, independent of buffer layer used, except one. The device prepared from absorber with CGI < 1 using Zn(O,S) buffer layer, temperature dependent V_OC measurements show V_OC extrapolation to the bandgap of the absorber at 0 Kelvin, demonstrating front interface passivation in the device. Recently front interface passivated Cu(In,Ga)S₂ device with high V_OC have been demonstrated, although with a lower bandgap.[5] Our study with higher bandgap allows differentiating between the influence of a defective surface layer and unfavorable band alignment. The findings demonstrate that high qFls and V_OC is achieved only with an unetched, i.e. undamaged surface and suitable band alignment. In this case, V_OC is limited by bulk recombination. We investigate the deep defects responsible. Finally, capacitance-voltage measurements conclude rather low doping values in Cu poor Cu(In,Ga)S₂, an increase in which could lead to further improvement in qFLs and hence the V_OC of the device and take it to beyond 1 V.

[1] B. Tell, J. Shay, and H. J. P. r. B. Kasper, Electrical Properties, Optical Properties, and Band Structure of CuGaS2 and CuInS2, Phys. Rev. B., vol. 4, no. 8, p. 2463, 1971.
[2] H. Hiroi, Y. Iwata, H. Sugimoto, and A. Yamada, Progress toward 1000 mV open circuit voltage on chalcopyrite solar cells, IEEE J. Photovolt., vol. 6, no. 6, pp. 1630-1634, 2016.
[3] H. Hiroi, Y. Iwata, S. Adachi, H. Sugimoto, and A. Yamada, New World Record Efficiency for Pure-Sulfide Cu(In,Ga)S₂ Thin-Film Solar Cell With Cd-Free Buffer Layer via KCN–Free Process, IEEE J. Photovolt., vol. 6, no. 3, pp. 760–763, 2016, doi: 10.1109/JPHOTOV.2016.2537540.
[4] A. Lomuscio et al., Quasi-Fermi-Level Splitting of Cu-Poor and Cu-Rich CuInS₂ Absorber Layers, Phys. Rev. Appl., vol. 11, no. 5, p. 054052, 2019.
[5] S. Kim, T. Nagai, H. Tampo, S. Ishizuka, and H. Shibata, Large open–circuit voltage boosting of pure sulfide chalcopyrite Cu (In, Ga) S2 prepared using Cu–deficient metal precursors, Prog. Photovolt., 2020.

Sulfide chalcopyrites have suitable band gaps to be used as top cells in tandem photovoltaic devices. Traditionally the sulfide solar cells were made from Cu-rich absorbers, i.e. absorbers grown under Cu-excess. The latest record device with an efficiency of 15.5% however, was made from Cu-poor material - which also results in the better solar cells in selenide chalcopyrite. Yet, in contrast to the selenides, the quasi-Fermi level splitting, i.e. the internal voltage of the absorber is higher in Cu-rich CuInS2 than in Cu-poor CuInS2. We will discuss the similarities and differences between sulfide and selenide material, in particular in terms of Cu dependent properties, as well as the defect spectra.

In this contribution, I will present an overview of the most important scanning probe microscopy results acquired on Cu(In,Ga)Se₂ (CIGSe) absorber surfaces over the last years. Then I will summarize some of our latest results acquired on polycrystalline and epitaxial CIGSe with different Cu-content. I will present a scanning tunneling microscopy (STM) study, where we analyzed the electronic effect of Cd²⁺ pre-electrolyte treatment [1]. I will show how Cd-ions effectively passivate defects and increase surface band bending. Then I will critically discuss the current status of Cu-rich CISe and present a detailed STM and Kelvin Probe Force Microscopy study, which we carried out on the rear-surface of the absorbers [2]. We do find strong evidence for a Cu_xSe precipitation at the back of Cu-rich absorbers. Finally, I compare measurements aquired on polycrystalline and epitaxial absorbers before and after K deposition and show how a short air exposure alters the work function distribution and the interpretation of grain boundary band bending.

[1] Boumenou, C. K. et al. Passivation of the CuInSe₂ surface via cadmium pre-electrolyte treatment. Phys. Rev. Mater. 4, 045405 (2020).
[2] Boumenou, C. K. et al. Electronic and compositional properties of the rear-side of stoichiometric CuInSe₂ absorbers, Prog Photovolt. Res. Appl. 2020, in press.

With the increasing use of nanocrystalline silicon layers and tunnel junctions in high-efficiency tandem and single-junction solar cells, the understanding of the transport mechanisms in these layers becomes mandatory. Moreover, some solar cells can be composed of an NPN junction: (n)nc-Si:H/(p)nc-Si:H/n-type transparent conductive oxide (TCO) with complex intern mechanisms not explained in the state-of-the-art. In this work, two approaches of tunnel junction from CSEM [1] and CEA [2] are compared and deeply characterized in a cross experiment.
We performed dark I-V measurements under variable temperature on several heterojunction test-structures. We used different n-type TCOs and nanocrystalline silicon layers to decorrelate the mechanisms at the interface with the TCO and at the interface between n-type and p-type nanocrystalline silicon layers. Structures without front-side TCO and with Aluminum metallization are also used to help this decorrelation. Moreover, we calculated contact resistivity (ρ_c) of each possible stack and corresponding contact activation energies.

Each test-structure is composed of a (n)c-Si wafer, a known and Ohmic backside and a variable frontside. By measuring symmetrical structures (with equivalent backside and frontside) and using the expanded Cox and Strack Method presented by Wang Et al. [3], we extracted the backside resistances and corresponding contact resistivity. With the poor lateral conductivity of amorphous silicon layers compared to the c-Si(n) bulk, we assume that the current spreading only occurs in the bulk, so we calculate the backside resistance, R_back with the following equations: R_back=(R_tot - R_bulk)/2 with R_bulk=(ρ/dπ).arctan(4t/d) where d is the front-side circular electrode diameter, t the wafer thickness and R_tot the total resistance extracted from I-V curves. By plotting the contact resistivity on an Arrhenius plot, we obtained contact activation energies.

Same analysis were performed for the whole test-structures of the study by using the equation R_front=R_tot-R_bulk-R_back to obtain frontside resistances. Stuctures without TCO on the frontside and Aluminum metallization on (n)nc-Si:H presents a good Ohmic behaviour, with contact resistivity under 250 mΩ.cm² at ambient temperature, which confirms a good contact between n-type nanocrystalline silicon layer and Aluminum. Contrariwise, for structures with the stack (n)nc-Si:H/(p)nc-Si:H/Al, Ohmic contacts are only obtained at high temperatures with the two thinner tunnel junctions and contact resistivities are larger than 400 mΩ.cm².
For the thicker tunnel junction, Ohmic behaviors were observed at all temperatures in all test-structures, which results in very low contact resistivity (under 40 mΩ.cm²). Moreover, in this case, we observed a very low front-resistance variation from 200K to 290K and an increase of this resistance from 275K to 350K when the bulk resistance becomes dominant. The differences in thermal behavior could be explained by a huge contribution of transport via tunneling, which is a mechanism less temperature dependent than thermoionic emission. In fact, with the vertical nanocrystalline growth, thicker is a layer; the better is the vertical conduction that permits the current to flow through the junction. A complete analysis of transport mechanisms and the results with the different front-side TCOs will be presented in the full study.

[1] B. Paviet-Salomon et al., Sol. Energy, vol. 175, pp. 60–67, Nov. 2018, doi: 10.1016/j.solener.2018.01.066.
[2] A. Puaud, A.-S. Ozanne, L.-L. Senaud, D. Munoz, and C. Roux, IEEE J. Photovolt., pp. 1–7, 2020, doi: 10.1109/JPHOTOV.2020.3038600.
[3] W. Wang et al., IEEE J. Photovolt., vol. 9, no. 4, pp. 1113–1120, Jul. 2019, doi: 10.1109/JPHOTOV.2019.2917386.

Various thin-film I₂–II–IV–VI₄ photovoltaic absorbers derived from kesterite have been synthesized, characterized, and theoretically investigated in the past few years. The availability of this homogeneous experimental/computational dataset is an opportunity to examine trends in the defect properties of kesterite-derived materials and identify criteria to (hopefully) find new defect-tolerant semiconductors in this vast chemical space.

First, we find that substitutions on the Zn site lead to a smooth decrease in band tailing as the ionic radius of the substituting cation increases. Unfortunately, this substitution strategy does not ensure the suppression of deeper defects and non-radiative recombination [1].

Then, we argue that Gaussian and Urbach band tails in kesterite-inspired semiconductors are two separate phenomena caused by two different antisite defect types. Deep Urbach tails are correlated with the calculated band gap narrowing caused by the (2I_II +IV_II) defect cluster. Shallow Gaussian tails are correlated with the energy difference between the kesterite and stannite polymorphs, which points to the role of (II_I +I_II) defect clusters involving Group IB and Group IIB atoms swapping across different cation planes. This finding can explain why in-plane cation disorder and band tailing are uncorrelated in kesterites.

Finally, we will discuss some PV-relevant properties of two relative newcomers to the I₂–II–IV–VI₄ family (Cu₂SrSnS₄ and Cu₂BaSnS₄) in a more detailed manner. The properties are: bulk defect landscape, band tailing, absolute band positions, and surface band bending [2].


[1] Crovetto et al., Assessing the defect tolerance of kesterite-inspired solar absorbers. Energy Environ. Sci., 2020, 13, 3489–3503.

[2] Crovetto et al., Experimental and first-principles spectroscopy of Cu₂SrSnS₄ and Cu₂BaSnS₄ photoabsorbers, ACS Appl. Mater. Interfaces, https://doi.org/10.1021/acsami.0c14578

Growing demand for energy sources that are cleaner and economical lead to intensive research on alternative materials for solar cells which does not require energy intensive fabrication procedures. Cuprous oxide (Cu₂O) is the among earliest semiconductor to be evaluated for photo-voltaics which is still being investigated for its earth abundant composition, cost effectiveness and bandgap. Cu₂O, a p-type semiconductor with a direct band gap of 2-2.2eV, is an attractive semiconductor material for photovoltaic conversion due to its high optical absorption coefficient in the visible range and can also be used as anode material in thin film lithium batteries. To maximize the photocurrent efficiency, it is coupled to charge transfer and transport pathways of Inorganic materials such as Vanadium pentoxide (V₂O₅),Molybdenum trioxide (MoO₃) and Nickel oxide (NiO). However, such realization requires cost intensive high vacuum plasma enhanced deposition processes and graphene is an ideal alternative to such hole transport layers.

Graphene is a transparent conductor which can be roll to roll processed in a cost effective fashion. These hole extraction layer requires a high work function to allow for the built-in electric field across the active layer and for holes to transport toward the anode. Graphene and it’s derivatives, due to their ambipolar behaviour allowing work function tunability, are suitable candidates for hole transport layer (HTL). The most important character of such an implementation is the contact resistance between P-Cu2O and the graphene film. Here we study the interface conductivity of graphene and Cu₂O films grown using electrochemical methods.

A compact, non-porous, stoichiometric Cu₂O was electrodeposited on graphene coated glass plates. The contact resistance of the Cu₂O thin film with commonly used metals such as Au, Pt, Ti was measured using a standard transfer length measurement technique. The contact resistance of graphene with the same set of metals were also measured. Finally, the crucial contact resistance of graphene-Cu₂O interfaces were measured for the first time. We draw conclusions on the electrodeposited doping density, Schottky barrier between the Cu₂O thin films and the metals and finally the efficiency of graphene as an efficient hole injector.

Activation and tuning of the useful properties of semiconductors intended for photovoltaic, electronic and/or electrochemical applications hinges on their successful release of charge carriers, a process typically achieved in appreciable quantity via substitutional incorporation of foreign ions into the host crystal structure, i.e. extrinsic doping. As it stands, the rate-limiting step in designing efficient semiconductors is the identification of the most viable dopant ions to achieve n- and/or p-type conductivity – a process that is severely clouded by our incomplete understanding of chemical bonding behavior in solids. Today, doping studies rely exclusively on resource-intensive first-principles calculations, with little chemical guidance available to computationalists regarding where to focus their resources. Below, we argue that an atomistic understanding of the factors underlying ion substitutions in solids would provide such guidance, and would allow proactive calculations to be made for host/dopant combinations prior to their synthesis in lieu of expanding costly resources in low-return compositional spaces.

Three major factors underlie doping difficulties in semiconductors: [1] dopant insolubility, where defect formation energies are prohibitively high, [2] un-ionizability of dopant, where the charge carrier is not released to the system at operating temperature, and [3] compensated dopants, where formation of an opposite-charge native defect negates the effect of the intentional dopant. Factor [3] is terminal whereby formation of compensated defects is caused by the charge carrier itself, and not the identity of the dopant. However, difficulties associated with factors [1] and [2] may be circumvented via appropriate dopant selection, whose ease of substitution in the host crystal structure (in the form of defect formation energies) dictates dopant level depth in the band structure. Factors known to influence dopant level depth include similarities in atomic orbital energies, local atomic geometry (crystal-chemical behavior), and size of the substituting ions – all of which amenable to guidance via an adequate, systematized understanding of crystal-chemical behavior. Substantial discrepancy between the dopant and substituted ions for any of these parameters results in deep dopant levels, which are generally undesirable as they provide uncontrolled recombination channels.

Systematic investigation of crystal-chemical behavior begins with wide-scale analysis of chemical bonding in solids, in the form of bond-length dispersion analyses. Results from the largest bond-length dispersion analysis for inorganic solids were recently reported for cations bonded to O^2- and N^3-,^[1] covering 177,446 reliable bond lengths hand-picked from 9210 crystal-structure refinements for oxides, and 6,770 bond lengths from 720 crystal-structure refinements for nitrides; equivalent analysis for cations bonded to S^2-, Se^2- and Te^2- is underway. We use these data to systematically identify structural, electronic and/or bond-topological effects underlying anomalous bonding behavior across the periodic table. Considering that the chemical insight derived from these bonding data holds at the local scale, our analyses enable an atomistic understanding of dopant behavior at point defects which we use to investigate and model the fundamental principles underlying the coexistence of ions in crystallographic sites, taking us one step closer to a quantitative assessment of dopant viability in compound semiconductors – and an atomistic resolution to the doping bottleneck.

References:

[1] Gagné, O. C. & Hawthorne, F. C. (2016). Acta Cryst. B72, 602–625; Gagné, O. C. (2018). Acta Cryst. B74, 49–62; Gagné, O. C. & Hawthorne, F. C. (2018a). Acta Cryst. B74, 63–78; Gagné, O. C. & Hawthorne, F. C. (2018b). Acta Cryst. B74, 79–96; Gagné, O. C. & Hawthorne, F. C. (2020). IUCrJ 7, 581-629; Gagné, O. C. (2020). ChemRxiv. 11626974

The development of traditional commercial thin film solar cells based on copper indium gallium selenide (CIGS) and cadmium telluride (CdTe) in large-scale production is limited due to the scarcity of In, Ga and Te and the toxicity of Cd.
Earth-abundant and low-toxic antimony chalcogenides Sb₂X₃ (X = S, Se, S/Se) have received an increased attention from the photovoltaics community thanks to their excellent optoelectronic properties. After only 7 years of research, the record power conversion efficiency (PCE) of Sb₂Se₃ based solar cells has reached 10.5%¹. Consequently, they are more and more considered as a promising next generation absorber materials. Despite the huge progress, the efficiency is still far lower than the values reported for CIGS and CdTe, largely owing to the absorber's low carrier density and high defects concentration². Doping is an effective method to improve the properties of semiconductors, however, only few works have been reported in this field.
In this work, we investigated the effects of several dopants (Na, In, Cu, I, and Te) on the electric and electronic properties of Sb₂Se₃ prepared using melt-quenching method. Different techniques including X-ray diffractions and observation under high-resolution electronic microscopes have been used to observe that the elements of Na and In are not doped into Sb₂Se₃ lattice, leading to phase separation. Surprisingly, two novel heterojunctions Sb₂Se₃/NaSbSe₂ and Sb₂Se₃/β-In₂Se₃ were successfully in-situ synthesized and they exhibited interesting photocatalytic activity for methyl orange degradation, taken as an example, due to the effective separation and transport of photo-generated carriers.
Most importantly, the effective p-type doping (Cu) and n-type doping (I and Te) are obtained, remarkably enhancing the photoelectric properties of Sb₂Se₃. Quasi-homojunction thin film solar cells using doped Sb₂Se₃ with Cu and I, have been fabricated via radio frequency (RF) magnetron sputtering technology. Encouraging PCE of 2.41%, fill factor (FF) of 41%, J_sc of 20 mA/cm², and V_oc of 294 mV have been achieved. The preparation, characterization, and loss mechanisms of these solar cells will be presented and discussed. This is the first work on homojunction solar cells based on Sb₂Se₃.
References
¹ X. Wang, et al., Manipulating the Electrical Properties of Sb₂(S,Se)₃ Film for High-Efficiency Solar Cell, Adv. Energy Mater. 2020, 2002341.
² C. Chen, et al., Characterization of Basic Physical Properties of Sb₂Se₃ and its Relevance for Photovoltaics. Front. Optoelectron. 2017, 10, 18−30.

As an emerging class of nontoxic and earth-abundant solar absorbers, kesterite Cu₂ZnSn(S,Se)₄ has been the focus of much attention from the photovoltaic community. However, despite years of optimization, the kesterite solar cells struggle to attain a power efficiency higher than 13%, far behind the other thin-film technologies such as CIGS and perovskites. The open-circuit voltage (V_OC) is in particular low given the material's band gap. The low efficiency and the low V_OC of kesterite solar cells are often attributed to the Cu-Zn cation disorder, although such an assignment is still a subject of debate. Here, the controversial role of cation disorder in the extraordinarily low V_OC of Cu₂ZnSnS₄ (CZTS) kesterite absorber is examined through a statistical treatment of disorder within the cluster-expansion method. It is demonstrated that the extensive Cu-Zn disorder alone cannot be responsible for the large Urbach tails observed in many CZTS solar cells. While the band gap is reduced as a result of the Gaussian tails formed near the valence-band edge due to Cu clustering, band-gap fluctuations contribute only marginally to the V_OC deficit, thereby excluding Cu-Zn disorder as the primary source of the low efficiency of CZTS devices. On the other hand, the extensive disorder stabilizes the formation of Sn_Zn antisite and complexes, which as nonradiative recombination and minority carrier trapping centers dominate the V_OC loss in CZTS. Our analysis indicates that current CZTS devices might have already approached the maximum conversion efficiency given the limited growth conditions and the remnant cation disorder even after post-annealing. In view of the improved efficiency achieved with CZTS-derived kesterite absorbers, the methodology presented in this work offers an avenue to understanding and optimizing these emerging kesterite solar devices towards higher efficiency.

In the last years quaternary chalcogenides have gained a lot of attention; especially the kesterite type semiconductor compounds Cu₂ZnSn(S,Se)₄ which consist mostly of earth abundant and non-toxic elements. These compounds are a promising low cost alternative absorber material for thin film solar cells due to its suitable criteria for photovoltaic applications: p-type semiconductor behavior, direct band-gap between 1.0-1.5 eV and absorption coefficient >10⁴ cm^-1 [1]. The record conversion efficiency of 12.6 % reported for a Cu₂ZnSn(S,Se)₄ based thin film solar cell was reached when the polycrystalline CZTSSe absorber layer exhibits an off-stoichiometric composition [2]. Deviations from stoichiometry cause intrinsic point defects (vacancies, anti-sites, interstitials), which determine the electronic properties of the semiconductor significantly [3]. It is agreed in literature that large band tailing observed in Cu-based kesterite-type semiconductors causes voltage losses limiting the efficiency of kesterite-based devices. The Cu/Zn disorder (Cu_Zn and Zn_Cu anti-sites in Cu-Zn planes at z=¼ and ¾), which is always present in these compounds [4, 5], is discussed as a possible reason for band tailing. Conventional structural characterization is done with X-ray diffraction, but in case of isoelectronic cations, like Cu⁺ and Zn²⁺, they are difficult to distinguish. Nevertheless, the neutron scattering length of Cu and Zn is different, by neutron diffraction we can distinguish between Cu⁺ and Zn²⁺ site occupation in the crystal structure [4, 5]. Kesterite-type based thin films solar cell technologies are mainly based on polycrystalline absorber layers, which makes it quite hard to correlate the crystallographic structure (determined via neutron diffraction) to the performance of these materials. A promising low cost alternative technology uses Cu₂ZnSn(S,Se)₄ (CZTSSe) monograins (single crystals of 50-100 μm size) which are fixed in a polymer matrix to form a flexible solar cell [6].
The experimental determination of the order parameter Q which is a quantitative measure of the degree of Cu/Zn disorder [7] requires a differentiation between the isoelectronic cations Cu⁺ and Zn²⁺. An in depth analysis of neutron diffraction data provides information on the cation distribution in the crystal structure allowing the determination of type and concentration of intrinsic point defects including a distinction between Cu and Zn [5]. On the other hand neutron diffraction requires large sample volumes, thus kesterite monograins offer the unique possibility to correlate structural disorder in kesterite-type absorbers with device performance parameters.
We will present a detailed structural investigation of CZTSSe monograins based on neutron powder diffraction experiments, the influence of the purity of the starting material as well as small changes in the chemical composition on the Cu/Zn disorder resulting in different power conversion efficiencies of the respective devices. A further investigation of the effect of the disordering procedure and the long low temperature – “ordering” annealing on the Cu/Zn disorder and optical properties of the CZTSSe monograins will be presented as well.

Reference:
[1] S. Delbos, EPJ Photovoltaics 2012; 3: 35004.
[2] W.Wang et al., Adv. Energy Materials, 2013.
[3] M. Guc et. al., in preparation
[4] S. Schorr, Solar Energy Materials & Solar Cells 2011; 95: 1482-1488
[5] G. Gurieva et. al., J Appl Phys 2018; 123: 161519/1-12
[6] www.crystalsol.com
[7] Toebbens et al. Phys. Stat. Sol. B 253 (2016) 1890