5 Dielectrophoretic Architectures

Electric programmability has been the basis for decades of advances in integrated computer component performance. However, once fabricated or assembled, such components—whether blade servers in a data center or microprocessors on a circuit board—are typically stuck in place and require human intervention for reconfiguration, removal, or replacement. For continued advances at the architectural level, mechanical programmability of components may also be needed. One generally promising approach for electromechanical manipulation at the nanoscale and microscale is dielectrophoresis, or the net force experienced by a neutral dielectric object in a nonuniform electric field. In this chapter, we review recent advances in dielectrophoretic architectures for computation, focusing particularly on the experimental demonstration of fully reconfigurable nanowire interconnects. Programmability in electronic systems originates from the ability to form and reform nonvolatile connections. Devices in modern programmable architectures typically derive this ability from controlled internal changes in material composition or charge distribution [1]. However, for ‘‘bottom-up’’ nanoelectronic systems it may be advantageous to derive programmability not only from internal state, but also from the mechanical manipulation of mobile components. Proposed applications that require component mobility include neuromorphic networks of nanostructure-based artificial synapses [2], breadboards for rapid prototyping of nanodevice circuits [3, 4], and fault-tolerant logic in which broken subsystems are replaced automatically from a reservoir [5]. In this chapter, we review recent advances in dielectrophoretic architectures for enabling such computational component mobility, focusing particularly on the experimental demonstration of fully reconfigurable nanowire interconnects—the simplest nanoelectronic components.