Search for: All records

In the ‘Beyond Moore’s Law’ era, with increasing edge intelligence, domain-specific computing embracing unconventional approaches will become increasingly prevalent. At the same time, adopting a variety of nanotechnologies will offer benefits in energy cost, computational speed, reduced footprint, cyber resilience, and processing power. The time is ripe for a roadmap for unconventional computing with nanotechnologies to guide future research, and this collection aims to fill that need. The authors provide a comprehensive roadmap for neuromorphic computing using electron spins, memristive devices, two-dimensional nanomaterials, nanomagnets, and various dynamical systems. They also address other paradigms such as Ising machines, Bayesian inference engines, probabilistic computing with p-bits, processing in memory, quantum memories and algorithms, computing with skyrmions and spin waves, and brain-inspired computing for incremental learning and problem-solving in severely resource-constrained environments. These approaches have advantages over traditional Boolean computing based on von Neumann architecture. As the computational requirements for artificial intelligence grow 50 times faster than Moore’s Law for electronics, more unconventional approaches to computing and signal processing will appear on the horizon, and this roadmap will help identify future needs and challenges. In a very fertile field, experts in the field aim to present some of the dominant and most promising technologies for unconventional computing that will be around for some time to come. Within a holistic approach, the goal is to provide pathways for solidifying the field and guiding future impactful discoveries.

 

The increasing complexity of deep learning systems has pushed conventional computing technologies to their limits. While the memristor is one of the prevailing technologies for deep learning acceleration, it is only suited for classical learning layers where only two operands, namely weights and inputs, are processed simultaneously. Meanwhile, to improve the computational efficiency of deep learning for emerging applications, a variety of non-traditional layers requiring concurrent processing of many operands are becoming popular. For example, hypernetworks improve their predictive robustness by simultaneously processing weights and inputs against the application context. Two-electrode memristor grids cannot directly map emerging layers’ higher-order multiplicative neural interactions. Addressing this unmet need, we present crossbar processing using dual-gated memtransistors based on two-dimensional semiconductor MoS 2 . Unlike the memristor, the resistance states of memtransistors can be persistently programmed and can be actively controlled by multiple gate electrodes. Thus, the discussed memtransistor crossbar enables several advanced inference architectures beyond a conventional passive crossbar. For example, we show that sneak paths can be effectively suppressed in memtransistor crossbars, whereas they limit size scalability in a passive memristor crossbar. Similarly, exploiting gate terminals to suppress crossbar weights dynamically reduces biasing power by ∼20% in memtransistor crossbars for a fully connected layer of AlexNet. On emerging layers such as hypernetworks, collocating multiple operations within the same crossbar cells reduces operating power by ∼ 15 × on the considered network cases. 

Mixed-dimensional van der Waals heterojunctions involve interfacing materials with different dimensionalities, such as a 2D transition metal dichalcogenide and a 0D organic semiconductor. These heterojunctions have shown unique interfacial properties not found in either individual component. Here, we use femtosecond transient absorption to reveal photoinduced charge transfer and interlayer exciton formation in a mixed-dimensional type-II heterojunction between monolayer MoS2 and vanadyl phthalocyanine (VOPc). Selective excitation of the MoS2 exciton leads to hole transfer from the MoS2 valence band to VOPc highest occupied molecular orbit in ∼710 fs. On the contrary, selective photoexcitation of the VOPc layer leads to instantaneous electron transfer from its excited state to the conduction band of MoS2 in less than 100 fs. This light-initiated ultrafast separation of electrons and holes across the heterojunction interface leads to the formation of an interlayer exciton. These interlayer excitons formed across the interface lead to longer-lived charge-separated states of up to 2.5 ns, longer than in each individual layer of this heterojunction. Thus, the longer charge-separated state along with ultrafast charge transfer times provide promising results for photovoltaic and optoelectronic device applications.

 

α-RuCl3 is a layered transition metal halide that possesses a range of exotic magnetic, optical, and electronic properties including fractional excitations indicative of a proximate Kitaev quantum spin liquid (QSL). While previous reports have explored these properties on idealized single crystals or mechanically exfoliated samples, the scalable production of α-RuCl3 nanosheets has not yet been demonstrated. Here, we perform liquid-phase exfoliation (LPE) of α-RuCl3 through an electrochemically assisted approach, which yields ultrathin, electron-doped α-RuCl3 nanosheets that are then assembled into electrically conductive large-area thin films. The crystalline integrity of the α-RuCl3 nanosheets following LPE is confirmed through a wide range of structural and chemical analyses. Moreover, the physical properties of the LPE α-RuCl3 nanosheets are investigated through electrical, optical, and magnetic characterization methods, which reveal a structural phase transition at 230 K that is consistent with the onset of Kitaev paramagnetism in addition to an antiferromagnetic transition at 2.6 K. Intercalated ions from the electrochemical LPE protocol favorably alter the optical response of the α-RuCl3 nanosheets, enabling large-area Mott insulator photodetectors that operate at telecommunications-relevant infrared wavelengths near 1.55 μm. These photodetectors show a linear photocurrent response as a function of incident power, which suggests negligible trap-mediated recombination or photothermal effects, ultimately resulting in a photoresponsivity of ≈2 mA/W. 

Abstract There is accelerating interest in developing memory devices using antiferromagnetic (AFM) materials, motivated by the possibility for electrically controlling AFM order via spin-orbit torques, and its read-out via magnetoresistive effects. Recent studies have shown, however, that high current densities create non-magnetic contributions to resistive switching signals in AFM/heavy metal (AFM/HM) bilayers, complicating their interpretation. Here we introduce an experimental protocol to unambiguously distinguish current-induced magnetic and nonmagnetic switching signals in AFM/HM structures, and demonstrate it in IrMn 3 /Pt devices. A six-terminal double-cross device is constructed, with an IrMn 3 pillar placed on one cross. The differential voltage is measured between the two crosses with and without IrMn 3 after each switching attempt. For a wide range of current densities, reversible switching is observed only when write currents pass through the cross with the IrMn 3 pillar, eliminating any possibility of non-magnetic switching artifacts. Micromagnetic simulations support our findings, indicating a complex domain-mediated switching process.