(2jq) High-Throughput Machine-Guided Hybrid Materials Exploration Via Combinatorial Resonant Infrared Matrix Assisted Pulsed Laser Evaporation

Vision. Hybrid materials with synergistic properties endowed by blending inorganic and organic components into crystalline structures represent an exciting frontier in materials science. The untapped promise of crystalline hybrid materials is enormous due to the large array of compounds that can be assembled from wide ranges of possible organic and inorganic substructures, but these combinatorial possibilities make materials exploration challenges equally vast. The pace of conventional experimental materials development is unacceptably slow (decade timescale), and many performance-critical material properties lie beyond easy reach of computation. Data-driven machine learning is a promising materials exploration paradigm, but it requires large amounts of high-quality experimental data to arrive at robust conclusions. High-throughput materials synthesis techniques producing many samples at once can greatly accelerate machine-guided materials exploration, yet popular techniques for materials library fabrication are high-energy processes ill-suited for softer hybrid materials. Efficient exploration and exploitation of hybrid materials’ composition space will require high-throughput deposition tools specially designed to manage their unique challenges. Resonant infrared matrix-assisted pulsed laser evaporation (RIR-MAPLE) is an especially promising technique to fill this gap that has only recently begun to be used as a tool for depositing crystalline hybrid thin films. The large volume of data available from combinatorial RIR-MAPLE will facilitate the use of state-of-the-art machine learning techniques to optimize material properties and analyze their links to composition and processing parameters. To explore and optimize hybrid materials, my group will follow a “make-measure-model” wherein samples are prepared using combinatorial RIR-MAPLE and the resulting materials characterization data are used to train machine learning models that inform further cycles of experimentation. Particularly promising materials will be selected for detailed scientific study and used to construct devices for applications such as optoelectronics, spintronics, radiative cooling, and beyond.

Prior Research. My previous work includes developing methods to process hybrid materials and machine learning models to forecast their properties. As a graduate student in the Mitzi group at Duke University, I pursued several projects focused on developing novel processing schemes to improve the functionality of hybrid perovskites. One of the most captivating projects concerned the use of RIR-MAPLE to deposit thin films of crystalline hybrid materials, in collaboration with the Stiff-Roberts group at Duke. We were the first researchers to deposit crystalline hybrid materials of any kind by RIR-MAPLE, obtaining device-quality films of the archetypal halide perovskite, methylammonium lead iodide.¹ Subsequently, we used RIR-MAPLE to deposit layered perovskite films containing large conjugated organic molecules, forming self-assembled tunable quantum wells.² These complex structures possess unique photophysical properties and offer exquisite control of excited-state dynamics, but they are exceedingly difficult to fabricate as thin films due to the chemical disparity between the organic and inorganic structural components. RIR-MAPLE makes them tractable by uniting the gentleness of solution processing with the fine control of vapor deposition. While at Duke, I also pursued the development of stable hybrid perovskite solar cell architectures, and a better understanding of the degradation mechanisms that can occur in these devices. From this work, I discovered several reactions between methylammonium lead iodide and semiconductors commonly used in thin film solar cell fabrication: CdS,³ NiO,⁴ and SnO₂.⁴ During my postdoctoral work in the Hillhouse group at the University of Washington, I continued my work on hybrid perovskite stability from a different angle: using data science to predict and understand degradation processes affecting hybrid perovskites. Using a unique instrument that collects photoconductivity, photoluminescence, and optical transmittance data in-situ from thin film samples exposed to controlled environmental conditions, my colleagues and I assembled a large body of perovskite degradation data. From initial observations of the evolution of perovskite films’ optoelectronic properties, we developed the world’s first predictive model of perovskite degradation, which forecasts the decay kinetics of carrier diffusion length, a key material parameter constraining photovoltaic performance.⁵ Performing many degradation experiments across a wide range of conditions also enabled the development of the first quantitative kinetic model of the chemical decomposition rate of the prototypical hybrid perovskite, CH₃NH₃PbI₃, as a function of ambient environmental conditions.⁶ The decomposition rate predicted by this model subsequently served as a crucial predictive feature in first-of-their-kind machine learning models capable of predicting service lifetimes of CH₃NH₃PbI₃ solar cells with average error of ~40%.⁷ In my current role as a postdoctoral associate in the Waldeck group at the University of Pittsburgh, I am currently investigating the mechanisms by which chiral organic molecules can imprint optospintronic properties such as circular dichroism onto colloidal halide perovskite nanoparticles. This work is anticipated to reveal design rules by which such molecules may be selected and paired with inorganic semiconductors, and this knowledge should translate to the closely related solid-state organic-inorganic hybrid materials that have been the subject of my prior work.

Future Plans. I will synthesize my experience in advanced thin film deposition techniques for hybrid materials with my experience in automated data collection and machine learning to develop a program for characterizing, screening, and optimizing next-generation hybrid materials. A crucial first step will be developing combinatorial RIR-MAPLE-based approaches for deposition of hybrid materials libraries. Projects I am interested in pursuing once this capability is validated include discovery and optimization of chiral hybrid perovskite compositions for efficient spin valves and spin-LEDs; manipulating photon upconversion properties of lanthanide-containing hybrid materials towards improving efficiency of photocatalysis and photovoltaic energy conversion; and designing hybrid materials for radiative cooling systems to address the increasingly urgent need for alternatives to conventional air conditioning technology. Successful execution of these projects will establish a new blueprint for efficient, high-volume research in a new frontier of materials science. While I intend to cultivate scholars who are well-equipped for these multidisciplinary challenges within my group, I also look forward to the prospect of collaborating with researchers with complementary expertise, such as organic chemists, spectroscopists, and theorists.

Teaching Interests. Gaining a broad education in engineering and the physical sciences has been one of the most rewarding parts of my life, and I am invested in guiding future generations on equally edifying journeys. With my broad background in engineering, chemistry, and physics, I am most interested in teaching courses that exist at the nexus of these disciplines, such as thermodynamics and statistical mechanics, introductory courses in materials science and solid-state chemistry, or quantum mechanics. I would also be interested in developing upper-level courses on photovoltaics, advanced materials characterization, and the design and construction of custom experimental apparatus. In my view, there are three principal assets of a classroom STEM education: marketable skills, personal empowerment, and joy. Students are better able to retain knowledge and engage with difficult material when they have strong intrinsic motivation, and my courses will be designed to stoke their desire for active learning. By making it clear that diversity and differences are welcomed, I will promote an inclusive environment in which students feel that their unique identities, and those of their classmates, are an asset to their educational goals. I will encourage a growth mindset to combat implicit bias and internalized prejudice and give students the confidence they need to persevere when they are struggling. Students will take informal surveys at the beginning of each course to tailor material to their interests where possible (e.g., worked examples), and periodically afterward to provide feedback on what is working and what is not. Hands-on experience from laboratory work and design projects will be an integral part of courses that I teach, as well as weekly design problems. These experiences are good representations of the sorts of problems students will encounter in their future careers and provide the opportunity for students to exercise critical thinking and judgment, as well as to document their reasoning.

Selected Publications.

(1) Dunlap-Shohl, W. A.,* Barraza, E. T.,* ..., Stiff-Roberts, A. D., Mitzi, D. B. ACS Energy Lett. 2018, 3, 270–275.

(2) Dunlap-Shohl, W. A., ..., Mitzi, D. B. Mater. Horiz. 2019, 6, 1707–1716.

(3) Dunlap-Shohl, W. A., ..., Mitzi, D. B. J. Phys. Chem. C 2016, 120, 16437–16445.

(4) Dunlap-Shohl, W. A., ..., Mitzi, D. B. ACS Appl. Energy Mater. 2019, 2, 5083–5093.

(5) Stoddard, R. J.,* Dunlap-Shohl, W. A.,* ..., Hillhouse, H. W. ACS Energy Lett. 2020, 5, 946–954.

(6) Siegler, T. D.,* Dunlap-Shohl, W. A.,* ..., Hillhouse, H. W. J. Am. Chem. Soc. 2022, 144, 5552-5561.

(7) Dunlap-Shohl, W. A., ..., Hillhouse, H. W. ChemRxiv preprint, DOI: 10.26434/chemrxiv-2022-01p42/

* Equal credit.