Search

Classification is an important task at which both biological and artificial neural networks excel 1,2 . In machine learning, nonlinear projection into a high-dimensional feature space can make data linearly separable 3,4 , simplifying the classification of complex features. Such nonlinear projections are computationally expensive in conventional computers. A promising approach is to exploit physical materials systems that perform this nonlinear projection intrinsically, because of their high computational density 5 , inherent parallelism and energy efficiency 6,7 . However, existing approaches either rely on the systems' time dynamics, which requires sequential data processing and therefore hinders parallel computation 5,6,8 , or employ large materials systems that are difficult to scale up 7 . Here we use a parallel, nanoscale approach inspired by filters in the brain 1 and artificial neural networks 2 to perform nonlinear classification and feature extraction. We exploit the nonlinearity of hopping conduction 9-11 through an electrically tunable network of boron dopant atoms in silicon, reconfiguring the network through artificial evolution to realize different computational functions. We first solve the canonical two-input binary classification problem, realizing all Boolean logic gates 12 up to room temperature, demonstrating nonlinear classification with the nanomaterial system. We then evolve our dopant network to realize feature filters 2 that can perform four-input binary classification on the Modified National Institute of Standards and Technology handwritten digit database. Implementation of our material-based filters substantially improves the classification accuracy over that of a linear classifier directly applied to the original data 13 . Our results establish a paradigm of silicon-based electronics for small-footprint and energy-efficient computation 14 .

The outstanding optoelectronics and photovoltaic properties of metal halide perovskites, including high carrier motilities, low carrier recombination rates, and the tunable spectral absorption range are attributed to the unique electronic properties of these materials. While DFT provides reliable structures and stabilities of perovskites, it performs poorly in electronic structure prediction. The relativistic GW approximation has been demonstrated to be able to capture electronic structure accurately, but at an extremely high computational cost. Here we report efficient and accurate band gap calculations of halide metal perovskites by using the approximate quasiparticle DFT-1/2 method. Using AMX 3 (A = CH 3 NH 3 , CH 2 NHCH 2 , Cs; M = Pb, Sn, X = I, Br, Cl) as demonstration, the influence of the crystal structure (cubic, tetragonal or orthorhombic), variation of ions (different A, M and X) and relativistic effects on the electronic structure are systematically studied and compared with experimental results. Our results show that the DFT-1/2 method yields accurate band gaps with the precision of the GW method with no more computational cost than standard DFT. This opens the possibility of accurate electronic structure prediction of sophisticated halide perovskite structures and new materials design for lead-free materials.

We report charge transport measurements in nanoscale vertical pillar structures incorporating ultrathin layers of the organic semiconductor poly(3-hexylthiophene) (P3HT). P3HT layers with thickness down to 5 nm are gently top-contacted using wedging transfer, yielding highly reproducible, robust nanoscale junctions carrying high current densities (up to 10 6  A/m 2 ). Current-voltage data modeling demonstrates excellent hole injection. This work opens up the pathway towards nanoscale, ultrashort-channel organic transistors for high-frequency and high-current-density operation.

Coulomb-blockade transport--whereby the Coulomb interaction between electrons can prohibit their transport around a circuit--occurs in systems in which both the tunnel resistance, Rb between neighbouring sites is large (>>h/e2) and the charging energy, E(C) (E(C) = e2/2C, where C is the capacitance of the site), of an excess electron on a site is large compared to kT. (Here e is the charge of an electron, k is Boltzmann's constant, and h is Planck's constant.) The nature of the individual sites--metallic, superconducting, semiconducting or quantum dot--is to first order irrelevant for this phenomenon to be observed. Coulomb blockade has also been observed in two-dimensional arrays of normal-metal tunnel junctions, but the relatively large capacitances of these micrometre-sized metal islands results in a small charging energy, and so the effect can be seen only at extremely low temperatures. Here we demonstrate that organic thin-film transistors based on highly ordered molecular materials can, to first order, also be considered as an array of sites separated by tunnel resistances. And as a result of the sub-nanometre sizes of the sites (the individual molecules), and hence their small capacitances, the charging energy dominates at room temperature. Conductivity measurements as a function of both gate bias and temperature reveal the presence of thermally activated transport, consistent with the conventional model of Coulomb blockade.