Accelerated discovery of CO2 electrocatalysts using active machine learning.

The rapid increase in global energy demand and the need to replace carbon dioxide (CO2)-emitting fossil fuels with renewable sources have driven interest in chemical storage of intermittent solar and wind energy1,2. Particularly attractive is the electrochemical reduction of CO2 to chemical feedstocks, which uses both CO2 and renewable energy3-8. Copper has been the predominant electrocatalyst for this reaction when aiming for more valuable multi-carbon products9-16, and process improvements have been particularly notable when targeting ethylene. However, the energy efficiency and productivity (current density) achieved so far still fall below the values required to produce ethylene at cost-competitive prices. Here we describe Cu-Al electrocatalysts, identified using density functional theory calculations in combination with active machine learning, that efficiently reduce CO2 to ethylene with the highest Faradaic efficiency reported so far. This Faradaic efficiency of over 80 per cent (compared to about 66 per cent for pure Cu) is achieved at a current density of 400 milliamperes per square centimetre (at 1.5 volts versus a reversible hydrogen electrode) and a cathodic-side (half-cell) ethylene power conversion efficiency of 55 ± 2 per cent at 150 milliamperes per square centimetre. We perform computational studies that suggest that the Cu-Al alloys provide multiple sites and surface orientations with near-optimal CO binding for both efficient and selective CO2 reduction17. Furthermore, in situ X-ray absorption measurements reveal that Cu and Al enable a favourable Cu coordination environment that enhances C-C dimerization. These findings illustrate the value of computation and machine learning in guiding the experimental exploration of multi-metallic systems that go beyond the limitations of conventional single-metal electrocatalysts.

Glycerol binding at the narrow channel of photosystem II stabilizes the low-spin S2 state of the oxygen-evolving complex.

The oxygen-evolving complex (OEC) of photosystem II (PSII) cycles through redox intermediate states Si (i = 0-4) during the photochemical oxidation of water. The S2 state involves an equilibrium of two isomers including the low-spin S2 (LS-S2) state with its characteristic electron paramagnetic resonance (EPR) multiline signal centered at g = 2.0, and a high-spin S2 (HS-S2) state with its g = 4.1 EPR signal. The relative intensities of the two EPR signals change under experimental conditions that shift the HS-S2/LS-S2 state equilibrium. Here, we analyze the effect of glycerol on the relative stability of the LS-S2 and HS-S2 states when bound at the narrow channel of PSII, as reported in an X-ray crystal structure of cyanobacterial PSII. Our quantum mechanics/molecular mechanics (QM/MM) hybrid models of cyanobacterial PSII show that the glycerol molecule perturbs the hydrogen-bond network in the narrow channel, increasing the pKa of D1-Asp61 and stabilizing the LS-S2 state relative to the HS-S2 state. The reported results are consistent with the absence of the HS-S2 state EPR signal in native cyanobacterial PSII EPR spectra and suggest that the narrow water channel hydrogen-bond network regulates the relative stability of OEC catalytic intermediates during water oxidation.

Learning-in-Templates Enables Accelerated Discovery and Synthesis of New Stable Double Perovskites.

In the past three years, machine learning (ML) in combination with density functional theory (DFT) has enabled computational screening of compounds with the goal of accelerated materials discovery. Unfortunately, DFT+ML has, until now, either relied on knowledge of the atomic positions at DFT energy minima, which are a priori unknown, or been limited to chemical spaces of modest size. Here we report a strategy that we term learning-in-templates (LiT), wherein we first define a series of space group and stoichiometry templates corresponding to hypothesized compounds and, orthogonally, we allow any list of atoms to take on any template. The LiT approach is deployed in combination with previously established position-dependent representations and performs best with the representations that rely least on the atomic positions. Since the positions of the atoms in templates are known and do not change, LiT enables us to infer the properties of interest directly; additionally, LiT allows working with increased chemical spaces, since the same elements can take on a large number of templates. Only by using LiT were we able to span 5 × 106 double-perovskite compounds and achieve an acceleration factor of 700 compared to brute-force DFT, allowing us to predict never-before-screened compounds. Our findings motivated us to synthesize a new BaCu yTa(1- y)S3 perovskite, which we show using an electron probe microanalyzer has a 5:3 molar ratio of Cu to Ta and, using powder X-ray diffraction (XRD) analysis combined with a DFT-based XRD simulation and fitting, indicate a new phase having an I4/ m space group.

Relative stability of the S2 isomers of the oxygen evolving complex of photosystem II.

The oxidation of water to O2 is catalyzed by the Oxygen Evolving Complex (OEC), a Mn4CaO5 complex in Photosystem II (PSII). The OEC is sequentially oxidized from state S0 to S4. The S2 state, (MnIII)(MnIV)3, coexists in two redox isomers: S2,g=2, where Mn4 is MnIV and S2,g=4.1, where Mn1 is MnIV. Mn4 has two terminal water ligands, whose proton affinity is affected by the Mn oxidation state. The relative energy of the two S2 redox isomers and the protonation state of the terminal water ligands are analyzed using classical multi-conformer continuum electrostatics (MCCE). The Monte Carlo simulations are done on QM/MM optimized S1 and S2 structures docked back into the complete PSII, keeping the protonation state of the protein at equilibrium with the OEC redox and protonation states. Wild-type PSII, chloride-depleted PSII, PSII in the presence of oxidized YZ/protonated D1-H190, and the PSII mutants D2-K317A, D1-D61A, and D1-S169A are studied at pH 6. The wild-type PSII at pH 8 is also described. In qualitative agreement with experiment, in wild-type PSII, the S2,g=2 redox isomer is the lower energy state; while chloride depletion or pH 8 stabilizes the S2,g=4.1 state and the mutants D2-K317A, D1-D61A, and D1-S169A favor the S2,g=2 state. The protonation states of D1-E329, D1-E65, D1-H337, D1-D61, and the terminal waters on Mn4 (W1 and W2) are affected by the OEC oxidation state. The terminal W2 on Mn4 is a mixture of water and hydroxyl in the S2,g=2 state, indicating the two water protonation states have similar energy, while it remains neutral in the S1 and S2,g=4.1 states. In wild-type PSII, advancement to S2 leads to negligible proton loss and so there is an accumulation of positive charge. In the analyzed mutations and Cl- depleted PSII, additional deprotonation is found upon formation of S2 state.

The O2-Evolving Complex of Photosystem II: Recent Insights from Quantum Mechanics/Molecular Mechanics (QM/MM), Extended X-ray Absorption Fine Structure (EXAFS), and Femtosecond X-ray Crystallography Data.

Efficient photoelectrochemical water oxidation may open a way to produce energy from renewable solar power. In biology, generation of fuel due to water oxidation happens efficiently on an immense scale during the light reactions of photosynthesis. To oxidize water, photosynthetic organisms have evolved a highly conserved protein complex, Photosystem II. Within that complex, water oxidation happens at the CaMn4O5 inorganic catalytic cluster, the so-called oxygen-evolving complex (OEC), which cycles through storage "S" states as it accumulates oxidizing equivalents and produces molecular oxygen. In recent years, there has been significant progress in understanding the OEC as it evolves through the catalytic cycle. Studies have combined conventional and femtosecond X-ray crystallography with extended X-ray absorption fine structure (EXAFS) and quantum mechanics/molecular mechanics (QM/MM) methods and have addressed changes in protonation states of µ-oxo bridges and the coordination of substrate water through the analysis of ammonia binding as a chemical analog of water. These advances are thought to be critical to understanding the catalytic cycle since protonation states regulate the relative stability of different redox states and the geometry of the OEC. Therefore, establishing the mechanism for substrate water binding and the nature of protonation/redox state transitions in the OEC is essential for understanding the catalytic cycle of O2 evolution. The structure of the dark-stable S1 state has been a target for X-ray crystallography for the past 15 years. However, traditional X-ray crystallography has been hampered by radiation-induced reduction of the OEC. Very recently, a revolutionary X-ray free electron laser (XFEL) technique was applied to PSII to reveal atomic positions at 1.95 Å without radiation damage, which brought us closer than ever to establishing the ultimate structure of the OEC in the S1 state. However, the atom positions in this crystal structure are still not consistent with high-resolution EXAFS spectroscopy, partially due to the poorly resolved oxygen positions next to Mn centers and partial reduction due to extended dark adaptation of the sample. These inconsistencies led to the new models of the OEC with an alternative low oxidation state and raised questions on the protonation state of the cluster, especially the O5 µ-oxo bridge. This Account summarizes the most recent models of the OEC that emerged from QM/MM, EXAFS and femtosecond X-ray crystallography methods. When PSII in the S1 state is exposed to light, the S1 state is advanced to the higher oxidation states and eventually binds substrate water molecules. Identifying the substrate waters is of paramount importance for establishing the water-oxidation mechanism but is complicated by a large number of spectroscopically similar waters. Water analogues can, therefore, be helpful because they serve as spectroscopic markers that help to track the motion of the substrate waters. Due to a close structural and electronic similarity to water, ammonia has been of particular interest. We review three competing hypotheses on substrate water/ammonia binding and compile theoretical and experimental evidence to support them. Binding of ammonia as a sixth ligand to Mn4 during the S1 → S2 transition seems to satisfy most of the criteria, especially the most compelling recent EPR data on D1-D61A mutated PSII. Such a binding mode suggests delivery of water from the "narrow" channel through a "carousel" rearrangement of waters around Mn4 upon the S2 → S3 transition. An alternative hypothesis suggests water delivery through the "large" channel on the Ca side. However, both water delivery paths lead to a similar S3 structure, seemingly reaching consensus on the nature of the last detectable S-state intermediate in the Kok cycle before O2 evolution.

Erratum: Role of Tensorial Electronic Friction in Energy Transfer at Metal Surfaces [Phys. Rev. Lett. 116, 217601 (2016)].

This corrects the article DOI: 10.1103/PhysRevLett.116.217601.

Ammonia Binding in the Second Coordination Sphere of the Oxygen-Evolving Complex of Photosystem II.

Ammonia binds to two sites in the oxygen-evolving complex (OEC) of Photosystem II (PSII). The first is as a terminal ligand to Mn in the S2 state, and the second is at a site outside the OEC that is competitive with chloride. Binding of ammonia in this latter secondary site results in the S2 state S = (5)/2 spin isomer being favored over the S = (1)/2 spin isomer. Using electron paramagnetic resonance spectroscopy, we find that ammonia binds to the secondary site in wild-type Synechocystis sp. PCC 6803 PSII, but not in D2-K317A mutated PSII that does not bind chloride. By combining these results with quantum mechanics/molecular mechanics calculations, we propose that ammonia binds in the secondary site in competition with D1-D61 as a hydrogen bond acceptor to the OEC terminal water ligand, W1. Implications for the mechanism of ammonia binding via its primary site directly to Mn4 in the OEC are discussed.

S3 State of the O2-Evolving Complex of Photosystem II: Insights from QM/MM, EXAFS, and Femtosecond X-ray Diffraction.

The oxygen-evolving complex (OEC) of photosystem II has been studied in the S3 state by electron paramagnetic resonance, extended X-ray absorption fine structure (EXAFS), and femtosecond X-ray diffraction (XRD). However, the actual structure of the OEC in the S3 state has yet to be established. Here, we apply hybrid quantum mechanics/molecular mechanics methods and propose a structural model that is consistent with EXAFS and XRD. The model supports binding of water ligands to the cluster in the S2 → S3 transition through a carousel rearrangement around Mn4, inspired by studies of ammonia binding.

Fundamental Role of Oxygen Stoichiometry in Controlling the Band Gap and Reactivity of Cupric Oxide Nanosheets.

CuO is a nonhazardous, earth-abundant material that has exciting potential for use in solar cells, photocatalysis, and other optoelectronic applications. While progress has been made on the characterization of properties and reactivity of CuO, there remains significant controversy on how to control the precise band gap by tuning conditions of synthetic methods. Here, we combine experimental and theoretical methods to address the origin of the wide distribution of reported band gaps for CuO nanosheets. We establish reaction conditions to control the band gap and reactivity via a high-temperature treatment in an oxygen-rich environment. SEM, TEM, XRD, and BET physisorption reveals little to no change in nanostructure, crystal structure, or surface area. In contrast, UV-vis spectroscopy shows a modulation in the material band gap over a range of 330 meV. A similar trend is found in H2 temperature-programmed reduction where peak H2 consumption temperature decreases with treatment. Calculations of the density of states show that increasing the oxygen to copper coverage ratio of the surface accounts for most of the observed changes in the band gap. An oxygen exchange mechanism, supported by (18)O2 temperature-programmed oxidation, is proposed to be responsible for changes in the CuO nanosheet oxygen to copper stoichiometry. The changes induced by oxygen depletion/deposition serve to explain discrepancies in the band gap of CuO, as reported in the literature, as well as dramatic differences in catalytic performance.

Role of Tensorial Electronic Friction in Energy Transfer at Metal Surfaces.

An accurate description of nonadiabatic energy relaxation is crucial for modeling atomistic dynamics at metal surfaces. Interfacial energy transfer due to electron-hole pair excitations coupled to motion of molecular adsorbates is often simulated by Langevin molecular dynamics with electronic friction. Here, we present calculations of the full electronic friction tensor by using first order time-dependent perturbation theory at the density functional theory level. We show that the friction tensor is generally anisotropic and nondiagonal, as found for hydrogen atom on Pd(100) and CO on Cu(100) surfaces. This implies that electron-hole pair induced nonadiabatic coupling at metal surfaces leads to friction-induced mode coupling, therefore, opening an additional channel for energy redistribution. We demonstrate the robustness and accuracy of our results by direct comparison to established methods and experimental data.

NH3 Binding to the S2 State of the O2-Evolving Complex of Photosystem II: Analogue to H2O Binding during the S2 → S3 Transition.

Ammonia binds directly to the oxygen-evolving complex of photosystem II (PSII) upon formation of the S2 intermediate, as evidenced by electron paramagnetic resonance spectroscopy. We explore the binding mode by using quantum mechanics/molecular mechanics methods and simulations of extended X-ray absorption fine structure spectra. We find that NH3 binds as an additional terminal ligand to the dangling Mn4, instead of exchanging with terminal water. Because water and ammonia are electronic and structural analogues, these findings suggest that water binds analogously during the S2 → S3 transition, leading to rearrangement of ligands in a carrousel around Mn4.

Analysis of the radiation-damage-free X-ray structure of photosystem II in light of EXAFS and QM/MM data.

A recent femtosecond X-ray diffraction study produced the first high-resolution structural model of the oxygen-evolving complex of photosystem II that is free of radiation-induced manganese reduction (Protein Data Bank entries 4UB6 and 4UB8 ). We find, however, that the model does not match extended X-ray absorption fine structure and QM/MM data for the S1 state. This is attributed to uncertainty about the positions of oxygen atoms that remain partially unresolved, even at 1.95 Å resolution, next to the heavy manganese centers. In addition, the photosystem II crystals may contain significant amounts of the S0 state, because of extensive dark adaptation prior to data collection.

Structural changes in the oxygen-evolving complex of photosystem II induced by the S1 to S2 transition: A combined XRD and QM/MM study.

The S1 → S2 transition of the oxygen-evolving complex (OEC) of photosystem II does not involve the transfer of a proton to the lumen and occurs at cryogenic temperatures. Therefore, it is commonly thought to involve only Mn oxidation without any significant change in the structure of the OEC. Here, we analyze structural changes upon the S1 → S2 transition, as revealed by quantum mechanics/molecular mechanics methods and the isomorphous difference Fourier method applied to serial femtosecond X-ray diffraction data. We find that the main structural change in the OEC is in the position of the dangling Mn and its coordination environment.

Theoretical prediction of S-H bond rupture in methanethiol upon interaction with gold.

Organic thiols are known to react with gold surface to form self-assembled monolayers (SAMs), which can be used to produce materials with highly attractive properties. Although the structure of various SAMs is widely investigated, some aspects of their formation still represent a matter of debate. One of these aspects is the mechanism of S-H bond dissociation in thiols upon interaction with gold. This work presents a new suggestion for this mechanism on the basis of DFT study of methanethiol interaction with a single gold atom and a Au(20) cluster. The reaction path of dissociation is found to be qualitatively independent of the model employed. However, the highest activation barrier of S-H bond dissociation on the single gold atom (12.9 kcal/mol) is considerably lower than that on the Au(20) cluster (28.9 kcal/mol), which can be attributed to the higher extent of gold unsaturation. The energy barrier of S-H cleavage decreases by 4.6 kcal/mol in the presence of the second methanethiol molecule at the same adsorption site on the model gold atom. In the case of the Au(20) cluster we have observed the phenomenon of hydrogen transfer from one methanethiol molecule to another, which allows reducing the energy barrier of dissociation by 9.1 kcal/mol. This indicates the possibility of the "relay" hydrogen transfer to be the key step of the thiol adsorption observed for the SAMs systems.

Machine Learning Accelerates Discovery of Optimal Colloidal Quantum Dot Synthesis.

Colloidal quantum dots (CQDs) allow broad tuning of the bandgap across the visible and near-infrared spectral regions. Recent advances in applying CQDs in light sensing, photovoltaics, and light emission have heightened interest in achieving further synthetic improvements. In particular, improving monodispersity remains a key priority in order to improve solar cells' open-circuit voltage, decrease lasing thresholds, and improve photodetectors' noise-equivalent power. Here we utilize machine-learning-in-the-loop to learn from available experimental data, propose experimental parameters to try, and, ultimately, point to regions of synthetic parameter space that will enable record-monodispersity PbS quantum dots. The resultant studies reveal that adding a growth-slowing precursor (oleylamine) allows nucleation to prevail over growth, a strategy that enables record-large-bandgap (611 nm exciton) PbS nanoparticles with a well-defined excitonic absorption peak (half-width at half-maximum (hwhm) of 145 meV). At longer wavelengths, we also achieve improved monodispersity, with an hwhm of 55 meV at 950 nm and 24 meV at 1500 nm, compared to the best published to date values of 75 and 26 meV, respectively.

Pulsed axial epitaxy of colloidal quantum dots in nanowires enables facet-selective passivation.

Epitaxially stacking colloidal quantum dots in nanowires offers a route to selective passivation of defective facets while simultaneously enabling charge transfer to molecular adsorbates - features that must be combined to achieve high-efficiency photocatalysts. This requires dynamical switching of precursors to grow, alternatingly, the quantum dots and nanowires - something not readily implemented in conventional flask-based solution chemistry. Here we report pulsed axial epitaxy, a growth mode that enables the stacking of multiple CdS quantum dots in ZnS nanowires. The approach relies on the energy difference of incorporating these semiconductor atoms into the host catalyst, which determines the nucleation sequence at the catalyst-nanowire interface. This flexible synthetic strategy allows precise modulation of quantum dot size, number, spacing, and crystal phase. The facet-selective passivation of quantum dots in nanowires opens a pathway to photocatalyst engineering: we report photocatalysts that exhibit an order-of-magnitude higher photocatalytic hydrogen evolution rates than do plain CdS quantum dots.

Color-stable highly luminescent sky-blue perovskite light-emitting diodes.

Perovskite light-emitting diodes (PeLEDs) have shown excellent performance in the green and near-infrared spectral regions, with high color purity, efficiency, and brightness. In order to shift the emission wavelength to the blue, compositional engineering (anion mixing) and quantum-confinement engineering (reduced-dimensionality) have been employed. Unfortunately, LED emission profiles shift with increasing driving voltages due to either phase separation or the coexistence of multiple crystal domains. Here we report color-stable sky-blue PeLEDs achieved by enhancing the phase monodispersity of quasi-2D perovskite thin films. We selected cation combinations that modulate the crystallization and layer thickness distribution of the domains. The perovskite films show a record photoluminescence quantum yield of 88% at 477 nm. The corresponding PeLEDs exhibit stable sky-blue emission under high operation voltages. A maximum luminance of 2480 cd m-2 at 490 nm is achieved, fully one order of magnitude higher than the previous record for quasi-2D blue PeLEDs.

Insights into Photosystem II from Isomorphous Difference Fourier Maps of Femtosecond X-ray Diffraction Data and Quantum Mechanics/Molecular Mechanics Structural Models.

Understanding structure-function relations in photosystem II (PSII) is important for the development of biomimetic photocatalytic systems. X-ray crystallography, computational modeling, and spectroscopy have played central roles in elucidating the structure and function of PSII. Recent breakthroughs in femtosecond X-ray crystallography offer the possibility of collecting diffraction data from the X-ray free electron laser (XFEL) before radiation damage of the sample, thereby overcoming the main challenge of conventional X-ray diffraction methods. However, the interpretation of XFEL data from PSII intermediates is challenging because of the issues regarding data-processing, uncertainty on the precise positions of light oxygen atoms next to heavy metal centers, and different kinetics of the S-state transition in microcrystals compared to solution. Here, we summarize recent advances and outstanding challenges in PSII structure-function determination with emphasis on the implementation of quantum mechanics/molecular mechanics techniques combined with isomorphous difference Fourier maps, direct methods, and high-resolution spectroscopy.

Crystallographic Data Support the Carousel Mechanism of Water Supply to the Oxygen-Evolving Complex of Photosystem II.

Photosystem II (PSII) oxidizes water to produce oxygen through a four-step photocatalytic cycle. Understanding PSII structure-function relations is important for the development of biomimetic photocatalytic systems. The quantum mechanics/molecular mechanics (QM/MM) analysis of substrate water binding to the oxygen-evolving complex (OEC) has suggested a rearrangement of water ligands in a carousel mechanism around a key Mn center. Here, we find that the most recently reported X-ray free-electron laser (XFEL) crystallographic data obtained for the dark-stable S1 state and the doubly flashed S3 state at 2.25 Å resolution support the carousel mechanism. The features in the XFEL data and QM/MM model-simulated difference Fourier maps suggest that water displacement may occur from the so-called "narrow" channel, resulting in binding of a new water molecule to the OEC, and thus provide new insights into the nature of rearrangements of water ligands along the catalytic cycle before O=O bond formation.

X-ray Free Electron Laser Radiation Damage through the S-State Cycle of the Oxygen-Evolving Complex of Photosystem II.

The oxygen-evolving complex (OEC) catalyzes water-splitting through a reaction mechanism that cycles the OEC through the "S-state" intermediates. Understanding structure/function relationsships of the S-states is crucial for elucidating the water-oxidation mechanism. Serial femtosecond X-ray crystallography has been used to obtain radiation damage-free structures. However, it remains to be established whether "diffraction-before-destruction" is actually accomplished or if significant changes are produced by the high-intensity X-ray pulses during the femtosecond scattering measurement. Here, we use ab initio molecular dynamics simulations to estimate the extent of structural changes induced on the femtosecond time scale. We found that the radiation damage is dependent on the bonding and charge of each atom in the OEC, in a manner that may provide lessons for XFEL studies of other metalloproteins. The maximum displacement of Mn and oxygen centers is 0.25 and 0.39 Å, respectively, during the 50 fs pulse, which is significantly smaller than the uncertainty given the 1.9 Å resolution of the current PSII crystal structures. However, these structural changes might be detectable when comparing isomorphous Fourier differences of electron density maps of the different S-states. One conclusion is that pulses shorter than 15 fs should be used to avoid significant radiation damage.