The growing demand for electric vehicles and portable devices has brought significant attention to the scalability, recyclability, and economics of both traditional and emerging battery technologies. Lithium-ion batteries (LIB) are approaching their theoretical energy density yet remain the most widespread and promising technology for the next 5-10 years, as alternative (e.g., Li-metal) chemistries and solid-state architectures are still in relatively early stages of commercial scale-up. An alternative route to improve LIB performance and therefore increase energy density, is to redesign the cell geometry to incorporate thick electrodes. Batteries with thick planar (2D) electrodes suffer from power limitations and capacity loss due to increased Li-ion diffusion distances and tortuosity. 3D electrode designs compensate for this weakness by providing micro-scale channels within the electrode to enable rapid charge transport.Prototype LIBs using 3D electrodes have been fabricated using a wide variety of laboratory techniques (e.g., lithography, imprinting, direct-ink writing), which have afforded study of diverse geometries (e.g., pillar arrays, lattices, interpenetrating structures) and active materials (e.g., Si, Sn, carbon-based, lithium transition metal oxides). However, these methods do not have requisite scalability or compatibility with industrial LIB processing techniques, which demand: (1) integration with cost-effective metal substrates as current collectors; (2) compatibility with commodity anode/cathode materials and electrolytes; and (3) high-rate continuous production and cell assembly. Moreover, the materials used in 3D electrodes must be mechanically robust, electrically conductive, patterned at the microscale level, and have sufficient porosity to accommodate expansion and contraction upon charging and discharging.We are developing thick 3D “honeycomb” battery electrodes using patterned, vertically aligned carbon nanotubes (VA-CNTs) as current collectors. CNTs are widely known to have high thermal and electrical conductivities, and to be mechanically durable. VA-CNTs (“forests”) form by self-organization of CNTs during chemical vapor deposition (CVD) on substrates using common hydrocarbon sources (e.g., C2H2, C2H4). However, well-established CNT growth techniques utilize rigid non-conductive substrates, such as silicon wafers, necessitating an additional transfer step to a suitable substrate. Recently, we translated these insights from CNT growth on silicon wafer substrates to thin metal foils (Cu) suitable for battery electrode fabrication, using optimized in-furnace moisture conditions, and a thin-film interface that forms high-density nanoparticles upon annealing, and prevents diffusion of Cu that is known to poison CNT growth. The resulting CNT forests can reach up to 500 μm thickness and can be patterned to have regular hole arrays with feature sizes as small as ~5 μm.Thick composite electrodes are then created by coating CNT forests with Si thin films by CVD. The inherent nanoporosity of the CNT forests (>95%) allows for precursor diffusion into the entirety of the forest cross-section. The CVD conditions are tuned to create dense electrodes with tailored Si loading ranging from ~10 to ~90 at.%. Half-cells using monolithic and patterned Si-CNT electrodes (~200 μm tall), Li-metal foil and a liquid electrolyte/separator combination are cycled over a range of currents, demonstrating electronic connection between the deposited Si and Cu foil via the aligned CNTs. At low currents (C/5) these electrodes demonstrate large gravimetric capacities, indicating the lithiation of CNTs beneath the deposited Si. We will also investigate cycling performance at high currents (>5C) and perform post-mortem characterization of the Si-CNT interface to study how the CNTs and their patterning govern the stability of the electrode. Future work will look to optimize electrode performance by manipulating the geometric parameters of the CNT forests and assess the economics of CNT-based electrode manufacturing at commercial scales.
Read full abstract