Doping is used in many pn junction devices, such as polycrystalline solar cells, to increase the strength of the junction field to assist charge carrier collection and thus partially mitigate nonradiative recombination losses. We demonstrate a different doping characteristic for inorganic solar cells: using dopants to reduce charge carrier trapping and electronic band tails. Alloying CdTe with Se to form CdSeTe semiconductor reduced recombination, but CdSeTe has more complex defect states which can limit further efficiency gains due to charge carrier trapping and trap-limited mobility. Doping CdSeTe with P (but not N, As, or Sb in this study) reduces band tails (Urbach energies) and lessens the impact of the near valence band trap states, with ambipolar mobilities improving to >50 cm2V-1s-1, fill factor increasing from 76% to 79%, and efficiencies increasing by 0.9% absolute. Simulations are used to show how such defect reduction improves performance in the radiative limit.

Progress of state-of-the-art and next-generation thin-film photovoltaic devices is often stymied by open-circuit voltage (Voc) that is significantly lower than theoretical and practical limits. Yet, effectively diagnosing the primary sources of voltage loss remains challenging. Herein, a sequence of device-level characterization techniques and simulations are employed to identify and rank loss mechanisms. For the research-based Cd(Se, Te) device under study, most of the loss was at the front semiconductor heterointerface due to a clifflike conduction-band offset that lowered the recombination activation energy. Additional losses due to band tails were quantified by photoluminescence analysis. The latter provided the absorption coefficient and activation energy reduction associated with band tails as inputs to device models. Simulations showed that alleviating front-interface issues would improve Voc, but it would then be limited by bulk recombination. Further improvement of the bulk would then lead to back-contact limitations. Reducing band tails is beneficial in any circumstance. This analysis provides guidance for reaching toward the radiative Voc limit.

As polycrystalline photovoltaic device efficiencies continue to rise, performance loss mechanisms must be further scrutinized. Conventional mechanisms, including recombination and poor carrier collection are important, but losses can also occur solely due to nonuniformity in optoelectronic properties, such as spatially varying band gap, doping, and lifetimes. The disorder effects act in parallel to the conventional mechanisms. Herein, a simulation approach is developed based on the finite element method that can predict the behavior of a solar cell with any spatial distribution of parameters that vary on a grain-to-grain basis. Open circuit voltage (Voc) loss calculations are found to be in accordance with analytical predictions that disorder-induced losses are directly proportional to the square of the standard deviation of the spatial distribution. This approach also enables model inputs based on spatially mapped characterization data instead of assuming a probability distribution. It can be extended to allow coupling of the electronic physics with thermal physics and time dependence.

A comprehensive simulation platform was developed for tandem, bifacial photovoltaic (PV) devices using COMSOL Multiphysics (R) software and demonstrated with a CdSe top cell and Sb2Se3 bottom cell. The model allows for monolithic schemes based on a tunnel junction, a recombination interface, or an Ohmic, idealized interconnection. The Ohmic connection is facilitated by integrating the simulated top and bottom cells in a customized electric circuit model. Rigorous optical physics based on Maxwell's wave equation enables detailed optimization required for these multi-layer, bifacial systems. We explore the impacts of the recombination interface on device performance and determine the optimal conditions for current matching between the top and bottom cells. The simulated device has an efficiency of 29% for one-sided illumination and marginally increases to 29.25% in bifacial mode. We note that optimizing the tandem device for bifacial operation requires revisiting the absorber band gaps and thicknesses. Avenues for significant improvement are discussed.

The efficiency potential for single-junction photovoltaics (PV) is described by the detailed balance model, which requires the elimination of nonradiative recombination and perfect minority carrier collection. Improvements in GaAs, Si, and perovskite PV follow this model. It might be more complex for CdTe, a leading thin-film PV technology. While lifetime, passivation, and doping goals for 25% efficient CdTe solar cells are largely reached, voltage is approximate to 20% below the detailed balance limit. Why is that? In Se-alloyed CdSexTe1-x (Se is required for >20% efficiency) additional losses can occur due to electrostatic and bandgap fluctuations and due to electronic trap states. To understand mechanisms limiting CdSeTe solar cell performance and to suggest improvements, carrier dynamics, and transport in CdSexTe1-x with variation in Se composition and as doping is analyzed. It is shown that trapping, likely due to anion-site defects and their complexes, is correlated with low charge carrier mobility of 0.1-0.6 cm(2) (Vs)(-1). Even with 1000 ns charge carrier lifetimes, carrier diffusion length is less than the absorber thickness, reducing efficiency to approximate to 23%. Device simulations are used to analyze the performance of CdSexTe1-x solar cells; thermodynamic models are not sufficient for absorbers with electronic disorder and trapping.

AbstractIn this study, the origins of efficiency gains in Cu(In,Ga)Se2(CIGS) solar cells are investigated by introducing an Al2O3passivation layer in terms of the oxidation condition of Mo back contact, alkali‐metal diffusion, minority carrier lifetimes (τ), and charge conditions. The study reveals that introduction of an Al2O3back‐contact passivation layer into solar cells yields multiple impacts. Al2O3deposition enhances the oxidation of the Mo back contacts, increasing Na solubility in Mo and Na diffusion from Mo into the CIGS layer, thereby modifying the metastable properties of CIGS. The charge condition at the CIGS/Al2O3interface is not fixed negative charge but variable, dependent on whether electrons or holes are supplied. During solar cell operation, the interfacial charge condition is expected to be neutral or positive for Al2O3grown using plasma or thermal atomic layer deposition techniques, respectively. Moreover, the mechanical peeling off of CIGS from Mo back contact enhancedτin a similar way as with the insertion of Al2O3. Based on this study, the enhancement of alkali metal supply and the removal of direct contact of CIGS to the metal contact (Mo) can play crucial roles in improving the performance of CIGS solar cell.

Local resistance imaging can provide information on nm-scale carrier distribution in semiconductor devices. Scanning spreading resistance microscopy (SSRM), an atomic force microscopy-based nm-scale resistance mapping technique, has been developed for carrier delineation in Si microdevices. We report on the development and validation of SSRM on CdTe materials, by testing on molecular beam epitaxy (MBE) grown CdTe films. The probe/CdTe contact resistance was suppressed sufficiently below sample's spreading resistance by pressing the probe into the sample with ∼mN contact force and applying a large sample/probe forward bias voltage (Vs), which was understood by analyzing current-voltage (I-V) involving a serially connected insulating top layer with underlying spreading resistance. The carrier concentration as deduced from the resistance measurement, using a single mobility value, is consistent with Hall measurement with a standard deviation of 14% based on a set of MBE films with carrier concentrations in the range of 1015–1016/cm3. The doping polarity was readily identified by flipping Vs polarity, where the resistance with reverse Vs is orders of magnitude larger than forward Vs. While focusing on the SSRM technique validation, we also show an example on an As-doped Cd(Se,Te) polycrystalline thin film of a high-performance CdTe solar cell, which illustrates the local resistance nonuniformity with up to two orders of magnitude differences, indicating if local mobility is roughly constant, local carrier concentration can have significant nonuniformity.

When various types of disorder are present in a semiconductor, the extended states at the edge of the conduction and valence bands can spread into the band gap. Such sub-band gap states, or band tails, can be localized or delocalized depending on their range. Band tails can be detrimental to device performance by enhancing recombination rates and lowering the open-circuit voltage (Voc). They are often observed in photoluminescence (PL) spectra as peak broadening or as a sub-band gap energy peak that is red-shifted away from the band gap measured by absorption. In this work, we describe how to incorporate band tails in device simulations using mathematical models and interpretation of PL data. A rather general model is used allowing for consideration of different microscopic origins of band tails. Occupancy of the band tails is also included, which results in bias-dependent absorption. Inputs to the simulations include the band-tail-corrected absorption coefficient and the reduced band gap that governs the radiative recombination peak (PL peak). Significant Voc losses are observed in device simulations when absorption coefficients including tail effects and band gap shifts are included. It is demonstrated that band tails are increasingly detrimental as the tail extent goes beyond the thermal energy, kT. This approach allows for quantification of band-tail-related Voc and power conversion efficiency losses through a combination of PL, JV, and QE measurements and simulation. Cd(Se,Te) is used as a case study, but the approach can be applied to other technologies.

Cadmium telluride (CdTe)-based cells have emerged as the leading commercialized thin film photovoltaic technology and has intrinsically better temperature co-efficients, energy yield, and degradation rates than Si technologies. More than 30 GW peak (GWp) of CdTe-based modules are installed worldwide, multiple com-panies are in production, modules are shipping at up to 18.6% efficiency, and lab cell efficiency is above 22%. We review developments in the science and technology that have occurred over approximately the past decade. These achievements were enabled by manufacturing innovations and scaling module production, as well as maximizing photocurrent through window layer optimization and alloyed CdSexTe1-x (CST) absorbers. Improved chlorine passivation processes, film microstructure, and serendipitous Se defect passivation significantly increased minority carrier lifetime. Efficiencies >22% have been realized for both Cu and As doped CST-based cells. The path to further efficiency gains hinges primarily on increasing open circuit voltage (Voc) and fill factor (FF) through innovations in materials, fabrication methods, and device stacks. Replacing the longstanding Cu doping with As doping is resulting in better module stability and is being translated to large-scale production. To realize 25% efficiency and >1 V Voc, research and development is needed to increase the minority carrier lifetime beyond 100 ns, reduce grain boundary and interface recombination, and tailor band diagrams at the front and back interfaces. Many of these goals have been realized separately however combining them together using scalable manufacturing approaches has been elusive to date. We review these achievements and outstanding opportunities for this remarkable photovoltaic technology.

We show that trapping impacts charge carrier dynamics in undoped CdSeTe, undoped CdSeTe/CdTe bilayers, and in As-doped CdSeTe/CdTe solar cells. Trapping has much smaller influence in CdTe-only (no Se, no As) films. Electrostatic potential fluctuation model (amplitude γ > 30 meV) applies to As-doped CdSeTe/CdTe, and defect model (defect activation energy Ea = 0.2 eV) applies to undoped CdSeTe. Unusually, undoped CdSeTe with trap states can have high radiative efficiencies and TRPL lifetimes (10-15 μs). As a result, thermodynamic CdTe solar cell analysis (radiative voltage, implied voltage, external radiative efficiency metrics) needs to be used with caution. As a metric for trap states, PL emission spectroscopy and charge carrier lifetimes at low temperatures provide distinct trap signatures. Trapping impact on devices can be evaluated in modeling and assessed in measurements outlined in this paper, perhaps providing a verifiable hypothesis when we search how to overcome voltage bottleneck in CdTe solar cells.

With the semiconductor bulk properties reaching target values for highly efficient solar cells, efforts are applied to reduce losses at solar cell interfaces and contacts. Advances in understanding back contacts in thin‐film polycrystalline CdTe solar cells, a leading thin‐film PV technology, are reported. By using X‐Ray photoelectron spectroscopy, Kelvin probe spectroscopy, time‐ and energy‐resolved photoluminescence, defects at the back contact are analyzed. Densities of recombination centers and charged defects that induce near‐back‐contact band bending, both resulting in recombination losses, were estimated. Electro‐optical and surface analysis results are integrated into a device model, simulating the performance of CdSeTe/CdTe solar cells with 902 mV open circuit voltage.

In this work, we compare the light soaking behavior and potential induced degradation in devices from two different process batches (A and C). The two batches of CIGS devices have been prepared by ZSW with different co-evaporation tools and Ga/(Ga + In) depth profiles. Two different Na barriers ( $\text{SiO}_{2}$ or AIOx-AIN) have been used for low Na devices, and two buffers namely CdS and Zn(O,S) have been used in both batches. Initial device performance before stress indicates an improvement in efficiency and open-circuit voltage (Voc) in the incorporation of Na in the absorber. Replacement of CdS buffer with Zn(O,S) results in a decrease in efficiency due to drop in V oc albeit a slight increase in short-circuit current density (Jsc). CdS buffer devices from batch A show more degradation under heat-light soaking (HLS) with short-circuit junction bias compared to batch C. Zn(O,S) devices show similar HLS behavior for both sample batches. Potential induced degradation (PID) tests with 1000 V bias show higher degradation in low N a devices with Si02 barrier (batch C) compared to $\text{AIO}_{\mathrm{x}}$ .AIN barrier (batch A).

Degradation in RbF post-deposition-treated (PDT) CIGS solar cells under heat/light/bias stress is evaluated for two different buffers, CdS and Zn(O,S). Initial efficiency of baseline CdS buffer devices is about 17.7% with and without RbF PDT. However, Zn(O,S) buffer devices show considerable improvement in net carrier concentration (NCV), VOC, and efficiency from 16.3 to 17.9% with RbF PDT. Under accelerated stress tests, CdS buffer devices show an increase in VOC under open circuit (OC) light soaking irrespective of the alkali treatments, and the devices are relatively stable under short circuit (SC) stress. Zn(O,S) buffer devices, however, show a decrease in VOC under all stress conditions irrespective of the alkali treatments. All performance metrics for Zn(O,S) buffer devices typically decreased with stress, resulting in significant efficiency loss. Device models show that an increase in recombination at the CIGS/Zn(O,S) interface and near interface doping can explain the degradation in these devices.

In this paper, we report on the stability of CdTe devices with a structure of TCO/MZO/CdSeTe/CdTe/back-contact. The device showed reversible transitions between the light-soak state (LSS) with the best device efficiency and the dark-soak state (DSS) with an inferior efficiency. However, it showed an irreversible degradation state (DgS) driven by long-hour light soaking at an elevated temperature. We have investigated transitions between these three states from the perspective of the electric field by nm-resolution potential imaging across the devices using Kelvin probe force microscopy (KPFM). The results exhibit different anomalous electric field profiles. At the LSS, the electric field exhibits a main peak inside the CdSeTe layer instead of the MZO/CdSeTe heterointerface, illustrating a buried homojunction (BHJ) of the device. At the DSS, a large electric field peak at the MZO/CdSeTe interface was measured, which probably resulted in the inferior fill factor at the DSS. At the DgS, the electric field peak at the MZO/CdSeTe interface increased further and a third electric field was measured at the back contact of the device. Device modeling using COMSOL software, in alignment with both the electric field and device current-voltage curves, elucidates that a low n-doped CdSeTe in the region near the MZO/CdSeTe interface caused the BHJ in the LSS and a loss of MZO doping and/or increase of the conduction band offset spike due to long-term stress caused the increased electric field near the MZO/CdSeTe interface at the DgS.

Advancing optoelectronic and emerging technologies increasingly requires control and design of interfaces between dissimilar materials. However, incommensurate interfaces are notoriously defective and rarely benefit from first-principles predictions, because no explicit atomic-structure models exist. Here, we adopt a bulk crystal structure prediction method to the interface geometry and apply it to SnO2/CdTe heterojunctions without and with the addition of CdCl2, a ubiquitous and beneficial, but abstruse processing step in CdTe photovoltaics. Whereas the direct SnO2/CdTe interface is highly defective, we discover a unique two-dimensional CdCl2 interphase, unrelated to the respective bulk structure. It facilitates a seamless transition from the rutile to zincblende lattices and removes defect-states from the interface bandgap. Implementing the predicted interface electronic structure in device simulations, we demonstrate the theoretical feasibility of bufferless oxide-CdTe heterojunction solar cells approaching the Shockley–Queisser limit. Our results highlight the broader potential of designing atomically thin interlayers to enable defect-free incommensurate interfaces.

We report on investigations of reversible metastability and irreversible degradation of MZO/CdSeTe/CdTe devices from the perspective of electric field across the device using Kelvin probe force microscopy (KPFM). The device showed reversible transitions between the light-soak state (LSS) with the best device efficiency and the dark-soak state (DSS) with an inferior efficiency. However, it showed an irreversible degradation state (DgS) driven by long-hour light soaking at an elevated temperature. The nm-scale KPFM electric field imaging on cross-sections of the devices revealed different anomalous electric field profiles. The electric field at the LSS exhibits a main peak inside the CdSeTe layer but not at the MZO/CdSeTe heterointerface, demonstrating that working junction of the device is a buried homojunction (BHJ). At the DSS, a second electric field peak was observed at the MZO/CdSeTe interface with a similar strength to the main BHJ, which probably caused the decrease in fill factor at the DSS. At the DgS, the electric field peak at the MZO/CdSeTe interface increased significantly and a third electric field was measured at the back contact of the device. Device modeling using COMSOL and in alignment with both the electric field and device current-voltage curves suggest that a slightly low n-doped CdSeTe in the region near the MZO/CdSeTe interface caused the BHJ, and that either a loss of MZO doping or increase of the conduction band offset spike due to long-term stress caused the fill factor-dominated degradation and the increased electric field near the MZO/CdSeTe interface at the DgS. The former, MZO doping decrease, is more plausible by referring the related literature.

Solar cells are essentially minority carrier devices, and it is therefore of central importance to understand the pertinent carrier transport processes. Here, we advanced a transport imaging technique to directly visualize the charge motion and collection in the direction of relevant carrier transport and to understand the cell operation and degradation in state-of-the-art cadmium telluride solar cells. We revealed complex carrier transport profiles in the inhomogeneous polycrystalline thin-film solar cell, with the influence of electric junction, interface, recombination, and material composition. The pristine cell showed a unique dual peak in the carrier transport light intensity decay profile, and the dual peak feature disappeared on a degraded cell after light and heat stressing in the lab. The experiments, together with device modeling, suggested that selenium diffusion plays an important role in carrier transport. The work opens a new forum by which to understand the carrier transport and bridge the gap between atomic/nanometer-scale chemical/structural and submicrometer optoelectronic knowledge.

In this work, we present a methodology to separate effects of perovskite device metastability from irreversible degradation, using stress/rest cycling under constant current bias while collecting a series of electroluminescence images and continuously monitoring voltage. We develop a simulation model and procedures for image processing to better understand the effects of ion parameters on the transient nature of voltage and evolving electroluminescence images.

In this work, the effect of alkali treatments namely Na and RbF has been investigated on the potential induced degradation (PID) for CdS/CIGS solar cells. Four device types with variations in Na content and RbF post-deposition treatment (PDT) have been studied. Baseline Na devices (Type 1) show an average efficiency of similar to 18% which does not show an improvement with RbF-PDT (Type 1R) in this sample series. Low Na devices with and without RbF-PDT both show average efficiency of similar to 16% for this sample series. PID tests were conducted in dark at temperatures between RT-85 degrees C under 1000 V applied between back of the soda-lime glass and the Mo contact. All device types show degradation due to increase in shunt conductance, which could be partially recovered by reversing the bias voltage polarity. Recovery of PID degradation is a temperature dependent phenomenon, wherein, higher recovery is observed at higher temperature. Preliminary results show low-Na devices recover to 80%, whereas, baseline Na devices recover to 50% of initial efficiency at 85 degrees C. RbF-PDT does not change the recovery of baseline Na devices, however, low-Na devices with RbF-PDT show recovery to only 25% of initial device efficiency.

All-back-contact (ABC) architectures for perovskite photovoltaics represent untapped potential for higher efficiency and enhanced durability compared to conventional planar architectures. Interface engineering can be more complex in ABC designs, because both the electron and hole transport layers (ETLs/HTLs) are simultaneously exposed during processing. Herein, we fabricate ABC perovskite solar cells with a non-stabilized current-voltage scan power conversion efficiency >10% by developing complementary interface processing. UV-ozone exposure followed by annealing increases the work function and reduces the defect density of the NiOx HTL and removed contamination from the TiO2 ETL, which increases voltage and current collection. We measure the chemical composition of each transport layer interface using photoelectron spectroscopy and then use the resulting trends to inform a two-dimensional drift-diffusion model. The model suggests that further reduction of charged interface defect density, increase in the hole selective contact work function, and passivation of the front surface will enable >20% of ABC devices.