[Paper Review] A high-performance MoS2 synaptic device with floating gate engineering for Neuromorphic Computing

Tathagata Paul, Tanweer Ahmed+3 more|ArXiv.org|Apr 6, 2019

Advanced Memory and Neural Computing4 citations

TL;DR

This paper presents a high-performance MoS2 synaptic transistor with an extended floating gate architecture that achieves near-ideal subthreshold swing (77 mV/decade) and enables efficient charge tunneling for neuromorphic computing. The device demonstrates spike-time-dependent plasticity (STDP), long-term potentiation/depression, and ultra-low energy dissipation (~20 fJ per pulse), offering a robust, solid-state alternative to electrochemical-based synaptic devices.

ABSTRACT

As one of the most important members of the two dimensional chalcogenide family, molybdenum disulphide (MoS2) has played a fundamental role in the advancement of low dimensional electronic, optoelectronic and piezoelectric designs. Here, we demonstrate a new approach to solid state synaptic transistors using two dimensional MoS2 floating gate memories. By using an extended floating gate architecture which allows the device to be operated at near-ideal subthreshold swing of 77 mV/decade over four decades of drain current, we have realised a charge tunneling based synaptic memory with performance comparable to the state of the art in neuromorphic designs. The device successfully demonstrates various features of a biological synapse, including pulsed potentiation and relaxation of channel conductance, as well as spike time dependent plasticity (STDP). Our device returns excellent energy efficiency figures and provides a robust platform based on ultrathin two dimensional nanosheets for future neuromorphic applications.

Motivation & Objective

• To develop a solid-state synaptic transistor free from electrochemical reactions that degrade over time.
• To overcome the high voltage and energy requirements of conventional MoS2 floating gate devices for neuromorphic applications.
• To achieve near-ideal subthreshold swing and low-energy operation using an extended graphene floating gate architecture.
• To demonstrate biologically plausible synaptic plasticity, including spike-time-dependent plasticity (STDP), in a 2D semiconductor-based device.
• To enable scalable, energy-efficient neuromorphic hardware using ultrathin 2D materials like MoS2 with improved gate coupling and reduced short-channel effects.

Proposed method

• The device uses a MoS2 channel with a hexagonal boron nitride (hBN) tunnel barrier separating it from an extended graphene floating gate (FG).
• Charge tunneling between the FG and the MoS2 channel is controlled by a global back or top gate, enabling non-volatile memory behavior.
• The extended FG architecture enhances gating efficiency, reducing the required gate voltage and enabling subthreshold swing of 77 mV/decade over four decades of drain current.
• Spike-timing-dependent plasticity (STDP) is demonstrated by applying precisely timed voltage pulses to pre- and post-synaptic electrodes, with conductance changes measured as a function of time delay.
• Energy dissipation per pulse is calculated using E = I_sd × t_pulse × V_sd, with measurements taken at low drain bias (V_sd = 0.01 V).
• The device performance is characterized under varying pulse widths and amplitudes to assess energy efficiency and stability of synaptic weight modulation.

Experimental results

Research questions

• RQ1 Can an extended floating gate architecture in a MoS2 synaptic transistor achieve near-ideal subthreshold swing and low-energy operation?
• RQ2 How does the device emulate biological synaptic plasticity, particularly spike-time-dependent plasticity (STDP), in a solid-state, non-electrochemical system?
• RQ3 What is the energy dissipation per synaptic pulse in the proposed MoS2-based synaptic transistor, and how does it compare to existing neuromorphic devices?
• RQ4 Can the device maintain stable conductance modulation over repeated potentiation and depression cycles without degradation from ionic or electrochemical processes?
• RQ5 To what extent can the time constants of potentiation and depression be tuned via the device’s design and input pulse configuration?

Key findings

• The device achieves a subthreshold swing of 77 mV/decade over four decades of drain current, approaching the theoretical limit for field-effect transistors.
• Spike-time-dependent plasticity (STDP) is successfully demonstrated with characteristic time constants of τ+ = 0.34 s for potentiation and τ− = 0.6 s for depression.
• The device exhibits a 100% change in channel conductance for symmetric STDP, confirming robust and reversible synaptic weight modulation.
• Energy dissipation is measured at approximately 20 pJ per pulse for depression and scales linearly with pulse width, reaching an extrapolated value of ~20 fJ at 100 μs pulse width.
• The energy dissipation is five decades lower than two-terminal MoS2 devices and 1–2 decades lower than complementary MOS devices, highlighting superior energy efficiency.
• The device operates with low drain bias (V_sd = 0.01 V) and reduced gate voltage requirements due to enhanced gating efficiency from the extended FG architecture.