Computational design of novel materials and nanocatalysts towards energy related applications

Abstract

In this dissertation, the computational design of novel materials and nanocatalysts toward energy-related applications, including two-dimensional (2D) SiP/SiAs/SiSb for water splitting reaction, 2D <em>ph</em>-BO, and AlO monolayers as high-temperature materials, machine-learning- screening pure silica zeolites for water purification, and graphene-supported single-atom catalysts for oxygen reduction reaction (ORR), oxygen evolution reaction (OER), and hydrogen evolution reaction (HER). After the discovery of graphene in 2004¹ , much attention has arisen on 2D materials. As the semiconductor industry is currently based on silicon (Si), the Si counterpart of graphene, named silicene,^2,3 has attracted increasing interest in last year. Silicene shares most of the outstanding electronic properties of graphene (e.g., it is also semi-metallic with a Dirac point,^4,5 ) and also possesses non-zero bandgap. Another interesting 2D material is black phosphorene monolayer (BP),^6-9 which adopts a puckered structure, possessing significant anisotropic properties.^8,10-13 Inspired by the successfully synthesize of the pnictogen-silicon analogs of benzene, namely [(PhC(NtBu)₂)₃Si₃P₃] and [(PhC(NtBu)₂)₃Si₃As₃] molecules¹⁴, we employed the quasi-planar geometry of Si₃P₃/Si₃As₃ unit as building blocks to design the 2D materials combining silicon and phosphorus/arsenic/antimony atoms. We investigated the stability and electronic and optical properties of silaphosphorene, silaarsenene, and silaantimonene (SiP, SiAs, and SiSb). They have direct bandgaps, rather high carrier mobilities, high-efficiency absorption in the visible light region. In addition, band edges of SiP and SiAs straddle the water redox potentials, showing that they are promising water-splitting photocatalysts. Therefore, the work provides a systematical study of planar SiP/SiAs/SiSb 2D materials, and exhibits their promising future in nanoelectronics, solar cells, photocatalysis for water-splitting.<br /> <br /> As the neighboring atom to carbon, boron, has been investigated as single-element 2D materials, namely borophene, and recently have been grown on Ag(111) substrates.^15,16 Inspired by the computationally designs of some low-pressure B₂O₃ polymorphs and its unit cells: BO₃ triangle and B₃O₆ boroxol rings¹⁷; and planar boron monoxide clusters (B_nO_n) ¹⁸, we designed porous hexagonal boron oxide (<em>ph</em>-BO) 2D material and found that <em>ph</em>-BO is the global minimum structure of 2D boron monoxide. <em>ph</em>-BO is also a wide bandgap semiconductor with high stabilities and mechanical strength, which is very promising for the applications in deep-UV light range.<br /> <br /> The other group III monolayers, aluminum, is one of the most abundant elements in the earth's crust, and can easily form oxides in nature (three aluminum oxides exist in the world). We used Particle Swarm Optimization (PSO) searches and systematic Density Functional Theory (DFT) computations to study the geometric structures and electronic properties of stable 2D aluminum monoxide materials. We theoretically predicted the five low-lying-energy 2D aluminum monoxide (AlO) nanosheets. Among them, the buckled <em>PmA</em>-AlO is of the lowest energy, which is closely followed by the buckled structure, <em>P</em>mm-AlO, while the other three structures, namely <em>P62</em>-, <em>PmB</em>- and <em>P6m</em>, are planar and with pores of different sizes (4.34-8.85 Å diameters). All these monolayers are thermodynamically, dynamically, and thermally stable, which strongly indicate the feasibility of their experimental realizations. In particular, the two lowest energy AlO monolayers are both wide bandgap semiconductors, which endow them enormous potentials in devices such as high-power ultra-thin electronics, green and blue LEDs, blue-violet laser diodes and photodetectors. We hope that our predicted 2D AlO and <em>ph</em>-BO nanosheets will encourage more efforts on 2D main group oxide materials with unusual structural and electronic properties.<br /> <br /> After the satisfying 2D material designs based on molecular chemistry, we were thinking over a reverse direction: figuring out what the problem is and screening materials with specific properties from big potential databases. Though significant development of computational resources has been achieved these years, trail-tests of structures are still unfeasible for DFT computations. Therefore, we solve two properties-structures problems under the assistant of Machine Learning (ML) algorithms. The first problem was to screen pure-silica zeolites for water purifications. Siloxanes are emerging organic contaminants in the water,^19-25 which refer to a class of silicones derivatives containing Si-O bonding, ²⁶ including linear and cyclic compounds. Siloxanes are widely used in medicine, personal care products, and industrial applications^27-29 . Due to their high vapor pressure,³⁰ siloxanes are persistent and prone to bioaccumulation; ^31-36 Meanwhile, the release of siloxanes has potential toxic effects^37,38 and mask effect for the presence of other contaminants. Thus, it remains a grand challenge to remove them from water. Developing suitable sorbents is an energy-cost-effective solution^39,40 to the notorious siloxane removal problem.⁴¹ We designed a two-step computational framework combining Grand Canonical Monte Carlo (GCMC) simulations and ML method to investigate the adsorption performances of pure-silica zeolites for four representative linear siloxanes and derivatives. ^42-44 We obtained essential features and screened out 230 preeminent zeolites from 50959 hypothetic PSZs (picking ratio ≈ 0.0045) with excellent adsorption performance toward all these four PCs. This work highlights the promise of combining data-driven modeling with traditional computations to predict the performance of complex systems.<br /> <br /> The second problem we solved is to predict limiting potentials from the physical properties of graphene-supported single-atom electrocatalysts for three reactions. Oxygen reduction reaction (ORR), oxygen evolution reaction (OER), and hydrogen evolution reaction (HER) are among the core electrochemical processes in clean energy conversion and storage devices, such as metal-air batteries^45,46, water electrolyzers⁴⁷, and fuel cells^48-50. Single-atom catalysts (SACs), in which the well-dispersed isolated metal atoms are anchored on appropriate substrates, have emerged as a new frontier of heterogeneous catalysts due to their highly increased coverage of active sites, much enhanced catalytic performance, and maximal (100%) metal utilization.^51-57 We depicted the underlying pattern of the physical properties of 104 graphene-supported SACs and their limiting potentials toward ORR/OER/HER and predicted the catalytic performance of 260 other graphene- supported metal-nitrogen/carbon systems (M@N_xC_y). The reliability of the ML models was confirmed by the DFT computed limiting potential (UL) values of the top ML-recommended electrocatalysts (0.61, 1.51, 0.003 V for ORR, OER, and HER, respectively). This work provides a new paradigm for directly predicting the catalytic performance from the physical properties of catalyst candidates, and vividly demonstrates the strong capability of ML in screening and design of catalysts.