arXiv subjects

Matthew B. Pinson

Publications and source records attributed to Matthew B. Pinson.

4 recordsLinked to original sources

arXiv2019

Learned multi-stability in mechanical networks

We contrast the distinct frameworks of materials design and physical learning in creating elastic networks with desired stable states. In design, the desired states are specified in advance and material parameters can be optimized on a computer with this knowledge. In learning, the material physically experiences the desired stable states in sequence, changing the material so as to stabilize each additional state. We show that while designed states are stable in networks of linear Hookean springs, sequential learning requires specific non-linear elasticity. We find that such non-linearity stabilizes states in which strain is zero in some springs and large in others, thus playing the role of Bayesian priors used in sparse statistical regression. Our model shows how specific material properties allow continuous learning of new functions through deployment of the material itself.

cond-mat.soft↗

arXiv2018

Inferring Pore Connectivity from Sorption Hysteresis in Multiscale Porous Media

Vapor adsorption experiments are widely used to assess pore size distributions, but the large hysteresis sometimes observed between sorption and desorption isotherms remains difficult to interpret. Such hysteresis is influenced pore network connectivity, which has previously been modeled by percolation on infinite lattices. Our hypothesis is that percolation occurs instead through finite networks of micropores connecting accessible macropores, always exposed to the outside environment. We derive a general formula for sorption/desorption isotherms that introduces a simple measure of hierarchical pore connectivity -- the fraction $f$ of always exposed pores. The model thus accounts for "small world" connections in finite-size percolation, while also incorporating other hysteresis mechanisms, in single-pore filling, liquid insertion into the solid matrix, and cavitation. Our formula is able to fit and interpret both primary and scanning sorption/desorption isotherms for a variety of adsorbates (noble gases, water, and organics) and porous materials (cement pastes, dental enamels, porous glasses, carbon black and nanotubes), including cases with broad pore-size distributions and large hysteresis. It allows quantification of the prevalence of percolating macropores in the material, even though these pores are never filled during the sorption experiments. A distinct bump in sorption isotherms is explained as a lowering of the barrier to nucleation of the vapor phase with a universal temperature scaling.

cond-mat.mtrl-sci↗

arXiv2016

Signal Transmissibility in Marginal Granular Materials

We examine the "transmissibility" of a simulated two-dimensional pack of frictionless disks formed by confining dilute disks in a shrinking, periodic box to the point of mechanical stability. Two opposite boundaries are then removed, thus allowing a set of free motions. Small free displacements on one boundary then induce proportional displacements on the opposite boundary. Transmissibility is the ability to distinguish different perturbations by their distant responses. We assess transmissibility by successively identifying free orthonormal modes of motion that have the {\em smallest} distant responses. The last modes to be identified in this "pessimistic" basis are the most transmissive. The transmitted amplitudes of these most transmissive modes fall off exponentially with mode number. Similar exponential falloff is seen in a simple elastic medium, though the responsible modes differ greatly in structure in the two systems. Thus the marginal pack's transmissibility is qualitatively similar to that of a simple elastic medium. We compare our results with recent findings based on the projection of the space of free motion onto interior sites.

cond-mat.soft↗

arXiv2012

Theory of SEI Formation in Rechargeable Batteries: Capacity Fade, Accelerated Aging and Lifetime Prediction

Cycle life is critically important in applications of rechargeable batteries, but lifetime prediction is mostly based on empirical trends, rather than mathematical models. In practical lithium-ion batteries, capacity fade occurs over thousands of cycles, limited by slow electrochemical processes, such as the formation of a solid-electrolyte interphase (SEI) in the negative electrode, which compete with reversible lithium intercalation. Focusing on SEI growth as the canonical degradation mechanism, we show that a simple single-particle model can accurately explain experimentally observed capacity fade in commercial cells with graphite anodes, and predict future fade based on limited accelerated aging data for short times and elevated temperatures. The theory is extended to porous electrodes, predicting that SEI growth is essentially homogeneous throughout the electrode, even at high rates. The lifetime distribution for a sample of batteries is found to be consistent with Gaussian statistics, as predicted by the single-particle model. We also extend the theory to rapidly degrading anodes, such as nanostructured silicon, which exhibit large expansion on ion intercalation. In such cases, large area changes during cycling promote SEI loss and faster SEI growth. Our simple models are able to accurately fit a variety of published experimental data for graphite and silicon anodes.

physics.chem-ph↗