资源环境科技发展态势分析平台

Strain engineering is a powerful tool with which to enhance semiconductor device performance(1,2). Halide perovskites have shown great promise in device applications owing to their remarkable electronic and optoelectronic properties(3-5). Although applying strain to halide perovskites has been frequently attempted, including using hydrostatic pressurization(6-8), electrostriction(9), annealing(10-12), van der Waals force(13), thermal expansion mismatch(14), and heat-induced substrate phase transition(15), the controllable and device-compatible strain engineering of halide perovskites by chemical epitaxy remains a challenge, owing to the absence of suitable lattice-mismatched epitaxial substrates. Here we report the strained epitaxial growth of halide perovskite single-crystal thin films on lattice-mismatched halide perovskite substrates. We investigated strain engineering of a-formamidinium lead iodide (alpha-FAPbI(3)) using both experimental techniques and theoretical calculations. By tailoring the substrate composition-and therefore its lattice parameter-a compressive strain as high as 2.4 per cent is applied to the epitaxial alpha-FAPbI(3) thin film. We demonstrate that this strain effectively changes the crystal structure, reduces the bandgap and increases the hole mobility of alpha-FAPbI(3). Strained epitaxy is also shown to have a substantial stabilization effect on the alpha-FAPbI(3) phase owing to the synergistic effects of epitaxial stabilization and strain neutralization. As an example, strain engineering is applied to enhance the performance of an alpha-FAPbI(3)-based photodetector.

Epitaxial heterostructures based on oxide perovskites and III-V, II-VI and transition metal dichalcogenide semiconductors form the foundation of modern electronics and optoelectronics(1-7). Halide perovskites-an emerging family of tunable semiconductors with desirable properties-are attractive for applications such as solution-processed solar cells, light-emitting diodes, detectors and lasers(8-15). Their inherently soft crystal lattice allows greater tolerance to lattice mismatch, making them promising for heterostructure formation and semiconductor integration(16,17). Atomically sharp epitaxial interfaces are necessary to improve performance and for device miniaturization. However, epitaxial growth of atomically sharp heterostructures of halide perovskites has not yet been achieved, owing to their high intrinsic ion mobility, which leads to interdiffusion and large junction widths(18-21), and owing to their poor chemical stability, which leads to decomposition of prior layers during the fabrication of subsequent layers. Therefore, understanding the origins of this instability and identifying effective approaches to suppress ion diffusion are of great importance(22-26). Here we report an effective strategy to substantially inhibit in-plane ion diffusion in two-dimensional halide perovskites by incorporating rigid pi-conjugated organic ligands. We demonstrate highly stable and tunable lateral epitaxial heterostructures, multiheterostructures and superlattices. Near-atomically sharp interfaces and epitaxial growth are revealed by low-dose aberration-corrected high-resolution transmission electron microscopy. Molecular dynamics simulations confirm the reduced heterostructure disorder and larger vacancy formation energies of the two-dimensional perovskites in the presence of conjugated ligands. These findings provide insights into the immobilization and stabilization of halide perovskite semiconductors and demonstrate a materials platform for complex and molecularly thin superlattices, devices and integrated circuits.

An epitaxial growth strategy that improves the stability of two-dimensional halide perovskites by inhibiting ion diffusion in their heterostructures using rigid pi-conjugated ligands is demonstrated, and shows near-atomically sharp interfaces.

Layered nanocomposites fabricated using a continuous and scalable process achieve properties exceeding those of natural nacre, the result of stiffened matrix polymer chains confined between highly aligned nanosheets.

Memristor-enabled neuromorphic computing systems provide a fast and energy-efficient approach to training neural networks(1-4). However, convolutional neural networks (CNNs)-one of the most important models for image recognition(5)-have not yet been fully hardware-implemented using memristor crossbars, which are cross-point arrays with a memristor device at each intersection. Moreover, achieving software-comparable results is highly challenging owing to the poor yield, large variation and other non-ideal characteristics of devices(6-9). Here we report the fabrication of high-yield, high-performance and uniform memristor crossbar arrays for the implementation of CNNs, which integrate eight 2,048-cell memristor arrays to improve parallel-computing efficiency. In addition, we propose an effective hybrid-training method to adapt to device imperfections and improve the overall system performance. We built a five-layer memristor-based CNN to perform MNIST10 image recognition, and achieved a high accuracy of more than 96 per cent. In addition to parallel convolutions using different kernels with shared inputs, replication of multiple identical kernels in memristor arrays was demonstrated for processing different inputs in parallel. The memristor-based CNN neuromorphic system has an energy efficiency more than two orders of magnitude greater than that of state-of-the-art graphics-processing units, and is shown to be scalable to larger networks, such as residual neural networks. Our results are expected to enable a viable memristor-based non-von Neumann hardware solution for deep neural networks and edge computing.