Research · Memories

Memristors and neuromorphic hardware

Halide perovskite resistive-switching memories on rigid and flexible substrates, integrated into crossbar arrays and artificial synapses for neuromorphic computing.

A memristor is a two-terminal device whose resistance can be switched between a high-resistance and a low-resistance state and is retained without power. In halide perovskites, switching is attributed to the formation and rupture of a conducting filament by mobile ions and vacancies. These memories are low-cost, operate at low voltages and can be fabricated on flexible substrates. As their conductance can be modulated in small increments, they can also emulate biological synapses.

Background

A memristor is non-volatile, exhibits a pinched hysteresis loop in its I–V characteristics and can support multiple resistance levels in addition to binary states. These properties enable multi-bit storage, in-memory computing and devices that emulate synaptic weights. Since the conductance can be incrementally increased or decreased, memristors can exhibit spike-timing-dependent plasticity and Hebbian learning. Switching energies as low as the femtojoule range have been reported, well below those of SRAM and DRAM, and the two-terminal structure is well suited to dense 3D crossbar integration.

Applications include non-volatile memory, neuromorphic and in-memory computing, physically unclonable functions (PUFs), analog signal processing and logic-in-memory. Broader directions in the field include 3D memristor arrays (3D RRAM, VRRAM), edge-AI hardware for wearable and IoT devices, radiation-tolerant memory for space applications, and stochastic neuromorphic systems. Our work focuses on flexible, high-endurance crossbar RRAMs based on emerging materials such as halide perovskites.

Key results

  • Lead-free, flexible and air-stable memristors. Cs₃Bi₂I₉ memristors on PET show an ON/OFF ratio above 10⁵ and an endurance of 10⁴ cycles. They remain functional after more than 30 days at over 80% relative humidity without encapsulation, and under bending to a 5 mm radius. A flexible 4×4 crossbar array was fabricated with 81% device yield. The devices exhibit synaptic potentiation and depression at about 0.1 µJ per event.
  • Two switching modes in a single cell. In MAPbI₃ memristors, the compliance current selects between a field-driven (redox) mode and a Joule-heating mode of opposite polarity. In combination, the two modes provide complementary resistive switching in a single cell, which can suppress sneak currents in crossbar arrays.
  • Measurement methodology. The scan rate, compliance current, pulse parameters and device area all influence the switching voltages and the ON/OFF ratio. These dependences were characterised systematically, and a standard measurement protocol was proposed.
  • Defects involved in switching. DLTS on flexible MAPbI₃ memristors identifies methylammonium vacancies (activation energy of about 0.48 eV) that are expected to assist filament formation.
  • From device to array. M.Tech projects have addressed the design of read/write peripheral circuits for passive ReRAM crossbars at the 180 nm and 65 nm nodes, as well as sneak-path currents and crosstalk.

Ongoing work includes larger crossbar arrays based on all-inorganic perovskites and a memristor-based artificial olfactory system.

Applications of hybrid perovskite memristors in memory, synaptic and logic devices based on inorganic, organic and hybrid materials.
Applications of hybrid perovskite memristors in memory, synaptic and logic devices based on inorganic, organic and hybrid materials.
Flexible perovskite memristor on PET, with its switching characteristics and synaptic plasticity
Flexible perovskite memristor on PET, with its switching characteristics and synaptic plasticity. From M. Patel et al., Adv. Electron. Mater. 9, 2200908 (2023).
Paired-pulse facilitation, potentiation and depression in a flexible Cs₃Bi₂I₉ memristor synapse
Paired-pulse facilitation, potentiation and depression in a flexible Cs₃Bi₂I₉ memristor synapse. From R. Yumlembam et al., Flex. Print. Electron. 10, 015007 (2025).
Selection of two opposite-polarity switching modes by the compliance current in a MAPbI₃ memristor
Selection of two opposite-polarity switching modes by the compliance current in a MAPbI₃ memristor. From H. J. Gogoi and A. T. Mallajosyula, ACS Appl. Electron. Mater. 4, 1039 (2022).

Methods and tools

Spin coating and thermal / e-beam evaporation · conductive AFM · DC and pulsed I–V (Keysight B1500A, Keithley 4200) · DLTS · SPICE (Yakopcic) models · ReRAM peripheral circuits at 180 nm and 65 nm

Papers in this area

  1. Artificial synapse using flexible and air-stable Cs₃Bi₂I₉ perovskite memristors R. Yumlembam, H. J. Gogoi, R. P. Palathinkal, A. T. Mallajosyula Flexible and Printed Electronics 10, 015007 (2025) · doi:10.1088/2058-8585/adacf9
  2. Hybrid perovskite-based flexible and stable memristor by complete solution process for neuromorphic computing M. Patel, D. D. Kumbhar, J. Gosai, M. R. Sekhar, A. T. Mallajosyula, A. Solanki Advanced Electronic Materials 9(4), 2200908 (2023) Collaboration with PDEU · doi:10.1002/aelm.202200908
  3. Multifunctional bipolar and complementary resistive switching in HOIP memristors by the control of compliance current H. J. Gogoi, A. T. Mallajosyula ACS Applied Electronic Materials 4(3), 1039–1046 (2022) · doi:10.1021/acsaelm.1c01200
  4. Advances in flexible memristors with hybrid perovskites H. J. Gogoi, K. Bajpai, A. T. Mallajosyula, A. Solanki The Journal of Physical Chemistry Letters 12(36), 8798–8825 (2021) Perspective · doi:10.1021/acs.jpclett.1c02105
  5. Enhancing the switching performance of CH₃NH₃PbI₃ memristors by the control of size and characterization parameters H. J. Gogoi, A. T. Mallajosyula Advanced Electronic Materials 7(11), 2100472 (2021) · doi:10.1002/aelm.202100472
  6. A comparative study on the forming methods of chalcogenide memristors to optimize the resistive switching performance H. J. Gogoi, A. T. Mallajosyula Journal of Physics D: Applied Physics 53, 445108 (2020) · doi:10.1088/1361-6463/aba56e
  7. Ionic defect analysis in flexible hybrid perovskite memristor using deep level transient spectroscopy H. J. Gogoi, A. B. Andrabi, M. R. Sekhar, A. T. Mallajosyula IEEE International Flexible Electronics Technology Conference (IFETC) (2022) · doi:10.1109/IFETC53656.2022.9948530
  8. Effect of antisolvent method on the performance of HOIP based memristive devices H. J. Gogoi, A. Solanki, A. T. Mallajosyula 4th IEEE International Conference on Emerging Electronics (ICEE) (2018) · doi:10.1109/ICEE44586.2018.8937971
  9. Lead-free all-inorganic perovskite memristor by thermal evaporation R. Yumlembam, H. J. Gogoi, R. P. Palathinkal, A. T. Mallajosyula International Conference on Condensed Matter and Device Physics (ICCMDP), Gandhinagar (2023)
  10. Analysis of ionic defects in hybrid perovskite memristors using deep level transient spectroscopy H. J. Gogoi*, A. B. Andrabi*, A. T. Mallajosyula 1st International Conference on Low-Energy Digital Devices and Computing (ICLED), Singapore (2023)
  11. SPICE simulation of hybrid organic-inorganic perovskite memristive devices H. J. Gogoi 7th International Conference on Advanced Nanomaterials and Nanotechnology, flash talk (2021)
  12. Measurement process-dependent performance and conduction mechanism of hybrid perovskite memristors H. J. Gogoi, A. T. Mallajosyula nanoGe International Conference on Perovskite Memristors and Electronics (ICPME), oral (2021) · doi:10.29363/nanoge.icpme.2021.009