No abstract

No abstract

We propose a method for learning expressive energy-based policies for continuous states and actions, which has been feasible only in tabular domains before. We apply our method to learning maximum entropy policies, resulting into a new algorithm, called soft Q-learning, that expresses the optimal policy via a Boltzmann distribution. We use the recently proposed amortized Stein variational gradient descent to learn a stochastic sampling network that approximates samples from this distribution. The benefits of the proposed algorithm include improved exploration and compositionality that allows transferring skills between tasks, which we confirm in simulated experiments with swimming and walking robots. We also draw a connection to actorcritic methods, which can be viewed performing approximate inference on the corresponding energy-based model.

Summary Conjugated polymers are emerging as promising organic photocatalysts for hydrogen evolution from water. However, it is still very challenging for conjugated polymers to realize highly efficient photocatalytic hydrogen evolution. Herein, we demonstrate an efficient strategy of hydrophilic side chain functionalization to boost the hydrogen evolution rates of conjugated polymers. By functionalizing conjugated polymers with hydrophilic oligo (ethylene glycol) monomethyl ether (OEG) side chains, a 90-fold improvement in hydrogen evolution rate has been achieved than that of alkyl-functionalized conjugated polymer. It is found that the OEG side chains interact robustly with Pt co-catalysts, resulting in more efficient charge transfer. Moreover, OEG side chains in conjugated polymers can adsorb H + from water, resulting in significantly lowered energy levels on the surfaces of conjugated polymers, which enables cascade energy levels and enhances charge separation and photocatalytic performance. Our results indicate that rational side-chain engineering could facilitate the design of improved organic photocatalysts for hydrogen evolution.

Nonfullerene acceptors (NFAs) have contributed significantly to the progress of organic solar cells (OSCs). However, most NFAs feature a large fused-ring backbone, which usually requires a tedious multiple-step synthesis, and are not applicable to commercial applications. An alternative strategy is to develop nonfused NFAs, which possess synthetic simplicity and facile tunability in optoelectronic properties and solid-state microstructures. In this work, we report two nonfused NFAs, BTCIC and BTCIC-4Cl, based on an A–D–A′–D–A architecture, which possess the same electron-deficient benzothiadiazole central core but different electron-withdrawing terminal groups. The optical properties, energy levels, and molecular crystallinities were finely tuned by changing the terminal groups. Moreover, a decent power conversion efficiency of 9.3 and 10.5% has been achieved by BTCIC and BTCIC-4Cl, respectively, by blending them with an appropriate polymer donor. These results demonstrate the potential of A–D–A′–D–A type nonfused NFAs for high-performance OSCs. Further development of nonfused NFAs will be very fruitful by employing appropriate building blocks and via side-chain optimizations.