In the field of machine learning (ML) for materials optimization, active learning algorithms, such as Bayesian Optimization (BO), have been leveraged for guiding autonomous and high-throughput ...experimentation systems. However, very few studies have evaluated the efficiency of BO as a general optimization algorithm across a broad range of experimental materials science domains. In this work, we evaluate the performance of BO algorithms with a collection of surrogate model and acquisition function pairs across five diverse experimental materials systems, namely carbon nanotube polymer blends, silver nanoparticles, lead-halide perovskites, as well as additively manufactured polymer structures and shapes. By defining acceleration and enhancement metrics for general materials optimization objectives, we find that for surrogate model selection, Gaussian Process (GP) with anisotropic kernels (automatic relevance detection, ARD) and Random Forests (RF) have comparable performance and both outperform the commonly used GP without ARD. We discuss the implicit distributional assumptions of RF and GP, and the benefits of using GP with anisotropic kernels in detail. We provide practical insights for experimentalists on surrogate model selection of BO during materials optimization campaigns.
Metallic impurities can have a severe negative impact on the electrical properties of multicrystalline silicon. Understanding how metal impurities evolve over the course of solar cell processing is ...essential to determining their impact on final device performance, and further engineering these defects into their least detrimental configuration. Herein we present a summary of the results of several recent investigations into the evolution of metallic impurities during solar cell processing, including the types and nature of impurities present after ingot crystallization; the effectiveness of phosphorous diffusion at removing and passivating different types and levels of metallic impurities; and the relationship of the hydrogen passivation step to metal impurities, as observed in synchrotron investigations of solar cells material.
Measurements of charge-carrier lifetime in many early-stage thin-film photovoltaic materials can be arduous due to the prevalence of defects and limited information about material properties. In this ...perspective, we give a brief overview of typical techniques for measuring lifetimes and discuss the intuition involved in estimating lifetimes from such techniques, focusing on time-resolved photoluminescence as an example. We then delve into the underlying assumptions and uncertainties involved in analyzing lifetime measurements. Finally, we outline opportunities for improving accuracy of lifetime measurements by utilizing two emerging techniques to decouple different recombination mechanisms: two-photon spectroscopy, which we demonstrate on BiI3 thin films, and temperature- and injection-dependent current-voltage measurements.
Realizing general inverse design could greatly accelerate the discovery of new materials with user-defined properties. However, state-of-the-art generative models tend to be limited to a specific ...composition or crystal structure. Herein, we present a framework capable of general inverse design (not limited to a given set of elements or crystal structures), featuring a generalized invertible representation that encodes crystals in both real and reciprocal space, and a property-structured latent space from a variational autoencoder (VAE). In three design cases, the framework generates 142 new crystals with user-defined formation energies, bandgap, thermoelectric (TE) power factor, and combinations thereof. These generated crystals, absent in the training database, are validated by first-principles calculations. The success rates (number of first-principles-validated target-satisfying crystals/number of designed crystals) ranges between 7.1% and 38.9%. These results represent a significant step toward property-driven general inverse design using generative models, although practical challenges remain when coupled with experimental synthesis.
Commercial passivated emitter and rear cell (PERC) devices are critically hindered by a thus-far unidentified degradation mechanism called light- and elevated temperature-induced degradation (LeTID). ...In contrast to PERC devices, aluminum back-surface field (Al-BSF) devices are markedly more resistant to the same degradation mechanism. In this work, we employ device simulations to elucidate the differences between Al-BSF and PERC degradation behavior and thus accelerate the search for the root cause of LeTID. We find that a difference in defect activation under degradation conditions is a more likely explanation than enhanced sensitivity to bulk lifetime for PERC compared to Al-BSF devices. By employing device simulation for the two architectures under high illumination intensity, we identify a narrow parameter space for the LeTID defect precursor. When combined with experimental observations, this may yield important new information about LeTID defect formation.
The past decade's record of growth in the photovoltaics manufacturing industry indicates that global investment in manufacturing capacity for photovoltaic modules tends to increase in proportion to ...the size of the industry. The slope of this proportionality determines how fast the industry will grow in the future. Two key parameters determine this slope. One is the annual global investment in manufacturing capacity normalized to the manufacturing capacity for the previous year (capacity-normalized capital investment rate, CapIR, units /W). The other is how much capital investment is required for each watt of annual manufacturing capacity, normalized to the service life of the assets (capacity-normalized capital demand rate, CapDR, units /W). If these two parameters remain unchanged from the values they have held for the past few years, global manufacturing capacity will peak in the next few years and then decline. However, it only takes a small improvement in CapIR to ensure future growth in photovoltaics. Any accompanying improvement in CapDR will accelerate that growth.
By harvesting sub-bandgap photons, we have a path to overcome the Shockley-Queisser limit in photovoltaics (PVs). We investigate semiconductor nanocrystal (NC) sensitized upconversion via ...triplet-triplet annihilation (TTA) in organic semiconductors (OSCs). Since this process relies on optically inactive triplet states in the OSCs, we utilize PbS NCs to directly sensitize the triplet state via energy transfer. This is possible due to the strong spin-orbit coupling in PbS NCs, resulting in rapid spin-dephasing of the exciton. Current technology allows for upconversion of light with a photon energy above \sim 1.1 eV. However, while internal efficiencies are rapidly improving, the low external device efficiencies render them impractical for applications, as devices are based on a single monolayer of NCs. Our results show simply increasing the PbS NC film thickness does not show improvement in the efficiency due to poor exciton transport between PbS NCs. Here, we present a new strategy to increase the external upconversion efficiency by utilizing thin tinbased halide perovskites as the absorbing layer. Resonant energy transfer from the perovskite to the PbS NCs allows for subsequent sensitization of the triplet state in rubrene.
Combining high-throughput experiments with machine learning allows quick optimization of parameter spaces towards achieving target properties. In this study, we demonstrate that machine learning, ...combined with multi-labeled datasets, can additionally be used for scientific understanding and hypothesis testing. We introduce an automated flow system with high-throughput drop-casting for thin film preparation, followed by fast characterization of optical and electrical properties, with the capability to complete one cycle of learning of fully labeled ~160 samples in a single day. We combine regio-regular poly-3-hexylthiophene with various carbon nanotubes to achieve electrical conductivities as high as 1200 S/cm. Interestingly, a non-intuitive local optimum emerges when 10% of double-walled carbon nanotubes are added with long single wall carbon nanotubes, where the conductivity is seen to be as high as 700 S/cm, which we subsequently explain with high fidelity optical characterization. Employing dataset resampling strategies and graph-based regressions allows us to account for experimental cost and uncertainty estimation of correlated multi-outputs, and supports the proving of the hypothesis linking charge delocalization to electrical conductivity. We therefore present a robust machine-learning driven high-throughput experimental scheme that can be applied to optimize and understand properties of composites, or hybrid organic-inorganic materials.
As the photovoltaic community accelerates the development of new absorber
candidate materials towards high-performing PV devices, it is essential to
follow best practices and leverage deeper ...characterization tools. We have
identified temperature- and illumination-dependent current density-voltage
$J$($V$,$T$,$i$) and electron-beam induced current (EBIC) measurements as two
powerful PV device characterization techniques to evaluate the potential of
novel absorber candidate materials. Herein, we focus on the experimental
methods and best practices for applying $J$($V$,$T$,$i$) and EBIC, addressing
particular challenges in sample preparation and mounting. We demonstrate these
on the example of tin monosulfide, a promising PV absorber candidate material
that shares characteristics of many novel thin-film PV absorbers: mechanically
soft, polycrystalline, and used in heterojunction thin-film PV devices.