Lithium‐Battery Anode Gains Additional Functionality for Neuromorphic Computing through Metal–Insulator Phase Separation

Abstract

Specialized hardware for neural networks requires materials with tunable symmetry, retention, and speed at low power consumption. The study proposes lithium titanates, originally developed as Li‐ion battery anode materials, as promising candidates for memristive‐based neuromorphic computing hardware. By using ex‐ and in operando spectroscopy to monitor the lithium filling and emptying of structural positions during electrochemical measurements, the study also investigates the controlled formation of a metallic phase (Li7Ti5O12) percolating through an insulating medium (Li4Ti5O12) with no volume changes under voltage bias, thereby controlling the spatially averaged conductivity of the film device. A theoretical model to explain the observed hysteretic switching behavior based on electrochemical nonequilibrium thermodynamics is presented, in which the metal‐insulator transition results from electrically driven phase separation of Li4Ti5O12 and Li7Ti5O12. Ability of highly lithiated phase of Li7Ti5O12 for Deep Neural Network applications is reported, given the large retentions and symmetry, and opportunity for the low lithiated phase of Li4Ti5O12 toward Spiking Neural Network applications, due to the shorter retention and large resistance changes. The findings pave the way for lithium oxides to enable thin‐film memristive devices with adjustable symmetry and retention.

Toward Controlling Filament Size and Location for Resistive Switches via Nanoparticle Exsolution at Oxide Interfaces

“…[3] Their applications range from binary memory storage [4] to neuromorphic computing hardware for implementing learning algorithms based on neural networks. [5][6][7] Though promising, the major hurdle for their industrial fabrication is device-to-device and cycleto-cycle variability. In metal-oxide-based memristive devices, a switching oxide is sandwiched between two metal electrodes and changes its resistance with an application of a voltage in a bipolar manner.…”

Recent Developments of Nanomaterials and Nanostructures for High‐Rate Lithium Ion Batteries

“…Phase boundaries are considered as another type of migration short circuit for lithium ions in solid‐state materials, as coexistence of multiple phases is common during the lithiation and delithiation of the electrodes, including the systems of Li 7 Ti 5 O 12 /Li 4 Ti 5 O 12 Li x V 2 O 5 /V 2 O 5 and LiFePO 4 /FePO 4 . For instance, phase transformation from LiFePO 4 into FePO 4 has been reported during delithiation, in which the movement of interphase boundary was observed .…”

Impact of Oxygen Non‐Stoichiometry on Near‐Ambient Temperature Ionic Mobility in Polaronic Mixed‐Ionic‐Electronic Conducting Thin Films

“…[21,22] While this mechanism was regarded as an aging effect, determining the lifetime of capacitors or gate dielectrics, progress in nanometric processing methods has enabled the design of novel devices that take advantage of the mobile ionic species, with versatile functionalities that cannot be achieved solely by electronic effects. New memory devices based on ionic defect motion have emerged, for example, in the fields of electronics (e.g., memristors), [12,[23][24][25][26][27][28][29] magnetics (e.g., magnetoionics), [30][31][32] optics (e.g., electrochromic devices), [33][34][35][36] and ferroelectrics, [37][38][39] often matching the performance of their electronic counterparts, at lower energy consumption and smaller device footprint. Moreover, these new devices are considered as promising candidates for progressing beyond traditional computational architectures.…”

“…[40][41][42] This opens up exciting opportunities, for example, in artificial intelligence hardware and neuromorphic computing, allowing for hardware component designs that can mimic basic synaptic functionalities. [29] Nevertheless, due to the much lower mobilities associated with ionic rather than electronic carriers, such MIEC materials tend to be predominately electronically conductive in their behavior at near ambient conditions, making it difficult to isolate and characterize the ionic conductivity component and specifically, ionic mobility, which plays a critical role in controlling nanoionic device response times. [43] The oxygen ion migration properties of metal oxides have primarily been studied at elevated temperatures.…”

“…They do this by enabling analog behavior with time‐dependent plasticity phenomena that depend on the diffusive/drift nature of the ionic carriers that can only be emulated by purely electronic counterparts at high energy expenditure. [ 40–42 ] This opens up exciting opportunities, for example, in artificial intelligence hardware and neuromorphic computing, allowing for hardware component designs that can mimic basic synaptic functionalities [29] . Nevertheless, due to the much lower mobilities associated with ionic rather than electronic carriers, such MIEC materials tend to be predominately electronically conductive in their behavior at near ambient conditions, making it difficult to isolate and characterize the ionic conductivity component and specifically, ionic mobility, which plays a critical role in controlling nanoionic device response times [43] …”