Oscillatory neural networks (ONNs) represent a neuromorphic computing paradigm that leverages the phase dynamics of coupled oscillators to encode and process information [1]. A vanadium dioxide (VO₂) oscillator network is a specific type of neuromorphic architecture that utilizes the phase transition properties of VO₂ to create oscillating neurons interconnected with resistive or capacitive components. Coupling elements based on resistive random-access memory (ReRAM) enable programmable and trainable network connectivity [2]. These circuits are being explored for applications such as associative memory and pattern recognition [3], where the network's ability to synchronize in phase and frequency is utilized. Additionally, oscillation-based computing is highly effective for solving complex optimization problems (COPs), which typically require extensive computing resources, long processing times, and significant energy consumption [4].

In this work, we explore ONNs based on VO₂ oscillators and HfO₂-based analog ReRAM. The fabrication processes for both types of devices have been optimized to be back-end-of-line (BEOL) compatible, allowing for post-processing on top of an underlying CMOS circuit [5]. Regular VO₂ undergoes a reversible insulator-to-metal transition (IMT) at approximately 68°C, which can be increased by 20°C or more through doping or alloying. VO₂ devices excel in scalability and low power consumption due to their crossbar geometry, and they demonstrate long endurance of more than 10¹² cycles. The ReRAM devices used in this work consist of a conductive metal oxide (CMO) and a dielectric HfO_x, stacked between TiN electrodes. Their programmable multilevel resistance states make them ideal candidates for trainable ONNs.

We have demonstrated applications such as pattern or image recognition and COP tasks with VO₂ ONNs. Specifically, we have solved fundamental optimization problems like Graph Coloring, Max-cut, and Max-3SAT. By leveraging the natural tendency of oscillators to stabilize into defined state relationships, we achieve solution convergence within fewer than 25 oscillation cycles, a significantly faster process than traditional computers testing all possible combinations. We successfully mapped graph problems to ONNs with 9 VO₂ oscillators, attaining optimal solutions with high probability. The integration of VO₂ oscillators and HfO₂ ReRAM coupling arrays enables flexible re-programming of ONNs using the switching capability of ReRAM. The multi-level resistance tuning of our ReRAM allows fine adjustment of the coupling strength between individual oscillators in the ONN.

Biological neurons routinely execute higher-order computations that artificial neural networks usually require multiple layers to achieve. Replicating this versatility in hardware demands circuits that extend beyond simple relaxation oscillators to include (i) self-resonant, inductive elements and (ii) ultrasmooth negative-differential-resistance (NDR) devices. Here we leverage ion-mediated recombination in metal-halide perovskite diodes which introduces a phase-lag between voltage and current. This enables the diode with an intrinsic electrochemical inductance. For the first time, we quantify a quality factor of Q ~ 3 and observe both a fundamental resonance and its harmonics in the perovskite diode. Unlike conventional abrupt switching NDR, a continuous low-hysteresis NDR can be achieved by using an electrostatically gated silicon thyristor. The resulting perovskite–thyristor combination supports rich nonlinear dynamics, including saddle-node and Hopf bifurcations, thereby enabling both integrator- and resonator-type artificial neurons. Integrator neurons encode stimulus onset and termination with high temporal precision, whereas resonator neurons exhibit sharp frequency selectivity and native XOR (anti-coincidence) detection. These findings introduce a framework for designing versatile neuromorphic systems with unique hysteretic properties from memristive materials, offering a practical route to compact neurons that match the computational richness of their biological counterparts.

One of the major problems in advancing Internet of Things (IoT) technology is the need for fast complex data processing, feature extraction and classification tasks.^[1]  A further challenge is the need for powerful computing facilities as close as possible to the physical interface, implemented in edge computing systems, to decrease the power management, scalability and sustainability of cloud computing infrastructure.^[2] A notable example of edge  system is Brain-Computer Interface (BCI), that is a rapidly emerging field with applications in domains as prosthetic devices, robotics, communication technology, and security.^[3] To efficiently interface the brain with electronic devices for the recording, and possibly the in loco processing of signals, a major problem is represented by the real-time processing of raw neuronal signals, which imposes excessive requirements on bandwidth, energy, and computation capacity, often asking for a severe pre-processing task.[4] Neural networks can be employed in edge computing solutions for classification tasks.^[5] However, to leverage the inference capabilities of these learning machines, time-series data must first be flattened and then encoded into spike trains. This step is not trivial, as it introduces an additional layer of complexity in the processing chain and significantly reduces the temporal resolution originally present in the raw time-series data.^[6] Among various strategies developed to overcome these issues, cluster-assembled thin films are here proposed as novel hardware data processing solutions to efficiently perform reprogrammable computation and signal processing on the edge of the physical system under investigation. Metallic cluster-assembled materials, deposited by Supersonic Cluster Beam Deposition (SCBD), are characterized by a complex network composed by a high density of defects and grain-boundaries.^[7] These nonlinear electrical properties^[7–9] can be exploited for the development of novel paradigm of computation, as reconfigurable nonlinear Threshold Logic Gates.^[10] The engineering of the metal cluster-assembled thin films can be further developed and implemented in hybrid computing architectures, used for processing signals recorded on edge.^[11] As a case of interest, we report an in materia approach to perform on edge real-time series classification tasks,^[12] based on cluster-assembled thin films and a time-series analysis method proposed by Fulcher et al.^[13]. We used a nanocomposite resistive switching device, based on gold and zirconia (Au/ZrO_x) thin film,^[14,15] to project the input time-series into a higher dimensional-space, allowing the resulting output time-series to be further analyzed by a linear classifier. We demonstrated the potential of this method to classify with high accuracy and in real-time neuronal traces, recorded by a neural probe in the barrel cortex of a rat, in spontaneous and elicited conditions. The classification was carried out with limited datasets for training and memory storage, and characterized by higher interpretability and accuracy with respect to artificial neural networks used on the same neuronal traces^[16]. The proposed methodology is well-suited for its extension to other neuromorphic devices, in particular to all those systems with a fast response to stimuli to classify highly resolved temporal time-series.

On conventional computers, the performance of AI models is limited by the data transfer between the memory and the processor. Compute-in-Memory architectures offer a new paradigm: Vector-Matrix Multiplications may be performed by a voltage drop through a matrix of programmable resistances, the “synaptic weights”. Ferroelectric materials are excellent candidates for their realization:[1] in a two- or three-terminals geometry and in combination with a semiconducting oxide,[2], [3], [4] the conductance is programmed by controlling the configuration of the ferroelectric domains.

The unique fluorite unit cell of HfZrO₄ allows for the stabilization of ferroelectricity below 3 nm,[5] facilitating the scaling of synaptic weights. The mechanisms governing the resistive switching in WO_x / HZO-SL (5 nm) bilayers are discussed. The effect of the programming pulse duration and amplitude on the polarization switching are investigated, from milliseconds to nanoseconds timescales. Devices of different sizes and shapes are measured down to 500 nm in dimension. For an device size of 1 micrometer square, an On/Off ratio as high as 8 is obtained for 20 ns pulses, a 4-fold improvement compared to 40 um devices.

The relatively low crystallization temperature of polycrystalline hafnium oxide / zirconium oxide superlattices (HZO-SL) is compatible with the Back-End-Of-Line (BEOL) of CMOS transistors.[6], [7] These results not only demonstrate the functionalization of the BEOL with synaptic weights, but also pave the way for the integration of ferroelectric field-effect transistors with Beyond CMOS semiconductors.

[1]           T. Mikolajick, et al., “From Ferroelectric Material Optimization to Neuromorphic Devices,” Advanced Materials, 2023,  doi: 10.1002/adma.202206042.

[2]           L. Bégon-Lours et al., “Scaled, Ferroelectric Memristive Synapse for Back-End-of-Line Integration with Neuromorphic Hardware,” Advanced Electronic Materials, 2022, doi: 10.1002/aelm.202101395.

[3]           M. Halter et al., “Back-End, CMOS-Compatible Ferroelectric Field-Effect Transistor for Synaptic Weights,” ACS Appl. Mater. Interfaces, 2020, doi: 10.1021/acsami.0c00877.

[4]           M. Halter, et al., “A multi-timescale synaptic weight based on ferroelectric hafnium zirconium oxide,” COMMUNICATIONS MATERIALS, 2023, doi: 10.1038/s43246-023-00342-x.

[5]           L. Bégon-Lours et al., “Effect of cycling on ultra-thin HfZrO4, ferroelectric synaptic weights,” Neuromorph. Comput. Eng., 2022, doi: 10.1088/2634-4386/ac5b2d.

[6]           L. Bégon-Lours et al., “Back-End-of-Line Integration of Synaptic Weights using HfO2/ZrO2 Nanolaminates,” Advanced Electronic Materials, 2024, doi: 10.1002/aelm.202300649.

[7]           R. Hamming-Green, et al., “Multi-Level, Low-Voltage Programming of Ferroelectric HfO 2 /ZrO 2 Nanolaminates Integrated in the Back-End-Of-Line,” in 2024 8th IEEE EDTM, Bangalore, India, 2024, doi: 10.1109/EDTM58488.2024.10511719.

In this work, we introduce CMOS-compatible self-rectifying resistive switching devices based on amorphous GaO_x grown by plasma-enhanced atomic layer depositon (PE-ALD) at a low temperature of 250℃.

First, we will present how oxygen vacancies can be introduced in Ga₂O₃ thin films by shortening the oxygen plasma exposure time at each ALD cycle [1], which results in semiconducting thin films [2].

Second, we will discuss the bipolar resistive switching properties of Ti/GaO_x/W devices [3]. These forming-free self-rectifying devices exhibit an interfacial/bulk type resistive switching. The switching process originates from a field-driven oxygen exchange between the interfacial TiO_x and the GaO_x layers as well as from the charging/discharging of interfacial trap states. Highly reproducible multi-level resistance states are obtained under identical pulses with a close-to-linear behaviour in potentiation and depression. Since the operating voltage is quite large (8 V), further engineering of the stack is needed for compatibility with the XFAB 180 nm CMOS technology.

Third, we will discuss the properties of devices with controlled bilayer stacks involving a dielectric layer – either HfO₂ or Al₂O₃ – deposited by ALD at 250℃.  For W/GaO_x/HfO₂/Ti devices, V_Set is reduced down to 3.2 V (4 nm HfO₂) and for W/GaO_x/Al₂O₃/Ti devices, V_Set is as low as 2.0 V (3 nm Al₂O₃). The devices operate at low power, with low writing (~ 260 nA/µm²) and reading currents (~100- 600 pA/µm²).

Finally, we will illustrate the potential of W/GaO_x/Al₂O₃/Ti devices for neuronal activity detection with the detection of the high frequency activity of neuronal culture cells.

Our results highlight the potential of GaO_x-based bilayer memristive devices for BEOL integration on CMOS chips, to build a hybrid memristor / CMOS Micro Electrode Array (MEA) platform that can achieve on-chip neuronal signal processing in real time.

Logic and memory technologies face increasing complexity as continued scaling necessitates consideration of multiple physical processes to increase device, circuit and system reliability. This complexity drives the need for Design-Technology-Co-Optimization (DTCO) and System-Technology-Co-Optimization (STCO) approaches [1]. In these approaches, systems, circuits and devices are co-designed to improve performance and face critical development challenges. Emerging applications in machine learning acceleration and neuromorphic computing require sophisticated simulations that exploit extended regimes of device behavior, such as low-power operation at low voltages. Memory devices in particular lack adequate compact models, which are required for circuit simulation and design. Yet, the importance of memory devices and their compact models grows with Compute-In-Memory approaches, as SRAM and embedded DRAM scaling reach fundamental limits [2]. However, when used for computation, non-volatile memory (NVM) devices are facing serious reliability challenges due to endurance limitations that need to be modelled accurately [3].

In this talk we present Heracles [4], an efficient and dynamic compact model for ferroelectric HfO2 that enables analog and digital circuit simulations of both FeCap and FeFET devices with CMOS circuits. It further allows accurate modelling and simulation of variability and mismatch in both devices and circuits. This enables modelling of reliability phenomena and the exploration of their effects on the performance of memory circuits [5]. The physics-based model parameters are extracted using experimentally acquired characterization data and Technology Computer-Aided Design (TCAD) simulations. This allows direct correlation between reliability and scaling challenges with underlying material properties, such as parasitics or defects. In conclusion, our modeling framework provides essential tools for advancing ferroelectric memory technologies in next-generation computing architectures.

Memristors are one of the four fundamental electrical components, linking key quantities such as current, voltage, charge (the integral of current), and magnetic flux (the integral of voltage). First theorized in 1971 and experimentally realized in 2009, memristors have attracted considerable research attention due to their distinctive properties, particularly in neuromorphic computing and memory storage.

While memristive behavior is often studied in dedicated devices, it can also arise as a parasitic effect in other systems [1,2]. Despite its significance, research on parasitic memristive phenomena remains sparse, especially in comparison to resistors and capacitors.

In this work, we present a modeling approach for integrating parasitic memristive effects into real device simulations. Focusing on solar cells, we extend the standard single-diode model to incorporate second-order parasitic memristive effects. We show how interface-related charge accumulation can induce memristive behavior [3], along with a zero-bias voltage shift. Additionally, we include a memristor component to account for material-level changes in the semiconductor, particularly noticeable in the diode’s reverse-bias region.

As a case study, we analyze a Mo/MoSe₂/Sb₂Se₃(10 nm)/CdS(2 nm)/ITO solar cell structure, revealing unexpected memristive properties. These are investigated through triangular waveform and pulsed voltage experiments, with transient current responses recorded. Our enhanced single-diode model, modified to include parasitic memristive effects, shows strong agreement with experimental data, validating the presence of memristive behavior in the system.

Facing the energy and computational demands of large artificial intelligence (AI) models, significant efforts have focused on overcoming the memory bandwidth bottleneck by integrating memory and processor units [1]. Analog in-memory computing (AIMC), particularly with resistive array-based architectures, is a promising approach, enabling massively parallel, energy-efficient vector matrix multiplication (VMM) operations directly where data resides [2]. Resistive crossbar arrays efficiently map deep neural network (DNN) architectures onto real hardware, realizing the synaptic interconnects where the cross-point resistive devices store synaptic weights as conductance values [3].

In this talk, we present a CMOS-integrated analog resistive memory (ReRAM) technology based on fab-friendly conductive metal oxide and HfOx materials [4], enabling fully parallel in-memory compute (inference and training) operations in crossbar circuits. The results highlight the potential of our technology for scalable, energy-efficient analog AI hardware for both inference and training - in one platform. 

As neuromorphic computing moves toward energy-efficient, event-driven architectures,
the use of oscillator-based neurons has gained renewed interest. These systems aim to
emulate the core dynamic features of biological spiking neurons while leveraging
physical models that allow compact and tunable implementations. In this context, we
investigate a neuromorphic system governed by a set of two coupled differential
equations, structurally analogous to the Morris–Lecar model [1,2]. This reduced
framework captures key excitability and oscillatory features characteristic of spiking
neurons, while remaining analytically tractable.
By drawing a formal analogy with an elementary electrical circuit composed of a resistor,
a capacitor, and an inductor, we derive an analytical expression for the system’s
impedance function [3]. This complex function describes the linear frequency response
of the system to small periodic perturbations, and provides a natural bridge between the
time-domain dynamics of the model and its frequency-domain characteristics. The
impedance reveals how the system processes inputs across a range of frequencies,
exhibiting features such as resonance and phase lag.
Through numerical simulations, we explore how the system responds as a control
parameter is varied. We observe a Hopf bifurcation that marks the transition from a stable
fixed point to a regime of self-sustained oscillations [4]. Importantly, we find that this
bifurcation is accompanied by a qualitative transformation in the impedance spectrum:
the emergence of resonance peaks, frequency selectivity, and distinct shifts in phase
response signal the onset of oscillatory behavior. These changes reflect a reorganization
of the system’s internal time scales and nonlinear feedback structure.
Our findings underscore the utility of impedance spectroscopy as a diagnostic and
classification tool for neuromorphic oscillators. The method provides insight into critical
transitions—such as the onset of spiking—and allows identification of dynamical regimes
using experimentally accessible quantities. This approach is rooted in well-established
techniques from electrochemistry and neuroscience, where impedance measurements
have long been used to probe the behavior of both chemical and biological oscillators

Neuromorphic electronics aims to replicate the functionality of biological neurons, offering a promising route to energy-efficient data processing. One of the key application areas is bioelectronic interfacing, where neuromorphic systems can enable efficient, real-time analysis of biological signals at the edge. Organic electrochemical transistors (OECTs) are biocompatible, operate at low voltages, and support flexible, low-cost fabrication, making them ideal for bioelectronic interfaces and edge-processing in biosensing applications. Additionally, OECTs are particularly well-suited for such applications due to their unique ability to couple ionic and electronic transport for integration of electronic circuits with ion-driven biological systems.

Despite these advantages, implementing artificial neuron circuits that emulate the functional characteristics of their biological counterparts using organic materials remains a major challenge. In this work, we present an organic artificial neuron circuit based on a multivibrator – a well-known oscillatory circuit. The proposed neuron exhibits spiking activity with the ability of tuning intrinsic neuron excitability, which is considered to be involved in the learning process alongside synaptic plasticity. Apart from providing input-dependent spiking dynamics, the neuron demonstrates short-term memory based on its previous spiking activity. This study advances the development of accessible, neuromorphic hardware by introducing a new class of organic artificial neurons.

The force-flux relationship is a cornerstone of non-equilibrium thermodynamics and serves as the foundation for understanding how mass, heat, and charge are transported under external gradients. According to the Onsager reciprocal relations, these transport processes are governed by linear relationships between generalized fluxes (such as particle current or heat flow) and their conjugate forces (such as chemical potential, temperature, or electrical potential gradients). Importantly, these relations also account for cross-coupling effects—for example, how a voltage gradient can drive not only charge flow but also mass transport via ion migration. In this presentation, I will examine ion transport specifically under an applied voltage gradient and show how its behavior critically depends on the diffusion medium. I will focus in particular on layered materials with van der Waals (vdW) gaps, where ions can move through channels with minimal steric hindrance and low diffusion barriers. These unique structural characteristics enable directional, selective, and tunable ion transport. Such materials are especially promising for next-generation technologies: in semiconductors, they can be used for ion-based memory or neuromorphic devices; in metallic systems, for reconfigurable interconnects or electrochemical switching; and in insulators, for ultra-thin membranes that achieve ion sieving with high selectivity and low energy cost. Through this exploration, I aim to highlight the deep interplay between transport theory and material design in developing functional ion-based devices.

With the rise of artificial intelligence (AI) and its deep integration into modern life, efficient big data processing and bio-inspired technology for neuromorphic computing become crucial for overcoming the inherent bottleneck of the von Neumann architecture. Among next-generation memories, halide perovskites (HP)-based memristors have emerged as strong candidates for multi-functional and neuromorphic computing electronic devices due to their ionic-electronic migration properties, low power consumption, facile fabrication, non-volatile switching behavior, etc. However, a clear understanding of their operating mechanisms and physical behavior analysis remains limited, hindering their optimized design for multifunctional bio-inspired applications.

This presentation outlines the fundamental electrical characteristics, material design strategies, switching mechanisms, and physical dynamics of HP-based memristors, with a focus on their applicability in multifunctional and bio-inspired systems. Compositional engineering of HP materials has been shown to significantly influence switching behavior by modulating switching mechanisms in terms of ionic and defect migration pathways. The classification of switching modes and types is discussed in relation to the formation and modulation of conductive pathways involving mobile ions within the HP layer. These mechanisms are examined in the context of their relevance to reliable memory and neuromorphic functionality. To further elucidate device dynamics, various physical analysis techniques are introduced, including impedance spectroscopy (IS) and time-domain response modeling, which provide insight into the ionic drift-diffusion dynamics and relaxation behavior of the system. These techniques enable evaluation of essential properties such as nonlinearity, memory retention, and synaptic-like plasticity. Equivalent circuit models derived from frequency-domain analysis are also considered, offering a foundation for understanding complex behaviors such as hysteresis, rectification, and adaptive conductance tuning in HP-based memristors.

Altogether, these perspectives support the advancement of halide perovskite memristors as promising building blocks for energy-efficient, scalable, multi-functional bio-inspired computing systems.

 Halide perovskites, widely recognized for their advantages in optoelectronics, are now being actively explored as resistive switching materials for nanoelectronics due to their mixed ionic–electronic conductivity, low power consumption, and facile ion migration [1]. While three-dimensional (3D) halide perovskites exhibit memristive behavior, their polycrystalline nature and poor environmental stability have hindered practical implementation [2]. Recent advances in crystallographic orientation control [3], phase engineering [4], and mobile ion species engineering [5] have addressed these critical limitations, enabling significant progress in memory and neuromorphic technologies. Furthermore, the long-standing challenges related to phase and environmental instability in 3D halide perovskites have been effectively overcome through the development of two-dimensional (2D) halide perovskite systems. Array-level demonstrations confirm stable device operation, underscoring the feasibility of large-area integration. These findings position halide perovskites as versatile and scalable materials for next-generation neuromorphic hardware, with strong potential for energy-efficient, high-performance artificial intelligence systems. They also open pathways toward the integration of sensing and computing, enabling future advances in sensor fusion and neuromorphic intelligence.

Subtle material variations in halide-perovskite devices can lead to profound changes in their electrical behavior, transitioning from rectifying, diode-like responses to rich memristive characteristics. In our recent work, we observe that specific modifications - without altering the device architecture - induce strong inverted hysteresis, conductivity potentiation under voltage pulse trains, and inductive features in impedance spectroscopy. These findings suggest that intrinsic ionic-electronic interactions govern the switching behavior, offering insights into the origin of memristive effects in this material system. To interpret these results, we build on our earlier modeling of relaxation dynamics, where a voltage-dependent characteristic time was shown to mediate the crossover from volatile to non-volatile response. This framework helps us understand how local dynamic processes relate to global device hysteresis and impedance features. The combination of experimental observation and modeling advances our understanding of memristive switching and may contribute to identifying design principles for robust, filament-free memory devices.

Understanding and harnessing the internal physical mechanisms that govern memristive behavior is key to advancing their integration into neuromorphic computing systems. In this work, we present a modeling framework rooted in the concept of the chemical inductor [1], a reactive element originally discovered in ionic solar cells, adapted to describe memristive devices through a minimal model based on a single internal memory variable, x, governing the conductance state of the system. When x transitions between 0 and 1, it induces a low-to-high conductance change in the device, establishing a robust mechanism for generating hysteresis. This switching behavior is linked to a relaxation time governed by the chemical inductor, enabling control over the dynamic response of the system.

We illustrate this framework through an experimentally validated model of a memristive system based on a conical nanopore membrane embedded in an electrochemical cell. Static current-voltage measurements allowed the extraction of the stationary activation state of x, while impedance spectroscopy fitting to the minimal chemical inductor model yielded experimental values for the circuit elements. From these, we derived voltage-dependent relaxation times for the internal state—marking, the first experimental quantification of this property in a memristor. Building on these experimental insights, we developed a physically grounded dynamic model capable of reproducing both quasi-static I-V hysteresis and prototypical neuromorphic functions such as potentiation-depression measurements and paired-pulse facilitation, achieving excellent agreement with experimental data [2].

Building on this foundation, we extend the modeling framework to halide perovskite-based memristors. Experimental measurements have revealed the presence of additional capacitive features that cannot be captured by the single-variable memory model. To account for this, we incorporate an extra capacitive element that successfully reproduces the emergence of novel hysteretic responses at intermediate timescales, which are distinct from those governed by the primary activation variable x [3]. These observations point to more intricate internal dynamics that demand a refined modeling approach, capable of capturing the interplay between fast and slow processes in the system. Simultaneously, we are exploring the conditions under which halide perovskite memristors exhibit self-sustained oscillations, focusing on the emergence of negative differential resistance (NDR) as a key enabling feature [4]. Our models and early device-level simulations confirm that under specific voltage conditions, these systems can naturally transition into oscillatory regimes, offering a platform for integrated memristive oscillators.

In parallel, we are investigating the intrinsic noise characteristics of halide perovskite memristors. Our measurements indicate that fluctuations in resistance are closely linked to internal transitions in the system, suggesting a stochastic component in the evolution of internal states. Through noise spectral analysis and stochastic modeling, we are identifying correlations between noise signatures and switching dynamics. By embedding these memristors into simple oscillatory circuits, we demonstrate that intrinsic noise can act as a tuning parameter for driving transitions between oscillatory and non-oscillatory states. This controlled use of noise enables probabilistic switching behavior, laying the groundwork for noise-powered computational elements such as p-bits (probabilistic bits) [5]. These results underscore the potential of perovskite memristors not only as memory elements but also as fundamental building blocks for stochastic neuromorphic hardware.

Altogether, this body of work demonstrates how starting from a minimal modeling framework, based on internal memory variables and chemical inductive behavior, enables not only reproduction of experimental memristive responses but also the interpretation of complex functionalities. By progressively incorporating additional internal dynamics, noise sources, and non-linear phenomena, we bridge the gap between material-level understanding and emergent behaviors relevant for next-generation computing. This approach paves the way for leveraging memristive systems not only in conventional neuromorphic architectures such as oscillatory or spiking neural networks, but also in probabilistic and quantum-inspired platforms.

Ion migration in halide perovskites and its relation with the external contacts has very important implications in solar cells, photodetectors, X-ray detectors and memristors.¹ Ion migration poses a negative effect in some optoelectronic applications controlling the hysteresis and the long term stability. On the other hand, the ionic conductivity of halide perovskite is responsible for a memory effect that can be used in resistive memories expanding the applications for this type of materials. Several configurations are  evaluated in which structural layers are modified systematically: formulation of the perovskite including 2D perovskites,² the nature of the buffer layer³  and the nature of the metal contact⁴. We show that in order to efficiently promote migration of metal contact the use of pre-oxidized metals greatly enhance the performance of the memristor and reduces the energy requirements. Importantly, these halide perovskite devices show potential in both volatile and non-volatile memristive devices that find applications in neuromorphic computing.⁵ Overall, the interplay between migrating ions and chemical interactions with the contacts can be extrapolated to the different optoelectronic devices fabricated with halide perovskites.

 

References

1. Sakhatskyi, K.;  John, R. A.;  Guerrero, A.;  Tsarev, S.;  Sabisch, S.;  Das, T.;  Matt, G. J.;  Yakunin, S.;  Cherniukh, I.; Kotyrba, M. J. A. E. L., Assessing the Drawbacks and Benefits of Ion Migration in Lead Halide Perovskites. ACS Energy Lett. 2022, 7 (10), 3401-3414.

2.    Solanki, A.;  Guerrero, A.;  Zhang, Q.;  Bisquert, J.; Sum, T. C., Interfacial Mechanism for Efficient Resistive Switching in Ruddlesden–Popper Perovskites for Non-volatile Memories. J. Phys. Chem. Lett. 2020, 11 (2), 463-470.

3.    Gonzales, C.; Guerrero, A., Mechanistic and Kinetic Analysis of Perovskite Memristors with Buffer Layers: The Case of a Two-Step Set Process. J. Phys. Chem. Lett. 2023, 14 (6), 1395-1402.

4.    Pérez-Martínez, J. C.;  Berruet, M.;  Gonzales, C.;  Salehpour, S.;  Bahari, A.;  Arredondo, B.; Guerrero, A., Role of Metal Contacts on Halide Perovskite Memristors. Adv. Funct. Mater. 2023, 2305211.

5.    Pendyala, N.-K.;  Gonzales, C.; Guerrero, A., Decoupling Volatile and Nonvolatile Response in Reliable Halide Perovskite Memristors. Small Structures 2024, n/a (n/a), 2400380.