A flexible characterization methodology of RRAM: Application to the modeling of the conductivity changes as synaptic weight updates

Abstract

In this work, an automatic and flexible measurement setup, which allows a massive electrical characterization of single RRAM devices with pulsed voltages, is presented. The evaluation of the G-V maps under single-pulse test-schemes is introduced as an example of application of the proposed methodology, in particular for neuromorphic engineering, where the fine analog control of the synaptic device conductivity state is required, by inducing small changes in each learning iteration. To describe the obtained data, a timeindependent compact model for memristive devices is used. The fitting parameters statistical distributions are further studied.

Memristive State Equation for Bipolar Resistive Switching Devices Based on a Dynamic Balance Model and its Equivalent Circuit Representation

“…4.a and 4.b. In the continuous limit, (7) naturally generates the sigmoidal hysteron structure -V (positive and negative ridge functions) introduced in [11,29,30]. In the present approach, this mathematical structure is no longer required.…”

Self-Organizing Neural Networks Based on OxRAM Devices under a Fully Unsupervised Training Scheme

“…The main results of a previous work [30] show that small changes in the conductivity at the low resistance state (LRS) can be induced by means of controlling the maximum current driving the devices during the SET process, proving their plasticity property, and thus indicating that the tested devices are suitable to play the synaptic role in a neuromorphic crossbar-array. In [29], a pulse-programming setup was proposed, with the aim of analyzing in which ways fine changes in the conductivity of the device can be induced by the application of single pulses. The proposed setup allowed obtaining the experimental G–V characteristics of the tested devices, by means of the application of increasing and decreasing amplitude single pulses with a fixed pulse-width over time.…”

“…The proposed setup allowed obtaining the experimental G–V characteristics of the tested devices, by means of the application of increasing and decreasing amplitude single pulses with a fixed pulse-width over time. Results from [29] are shown in Figure 5, where the pulse amplitude and the conductivity measured after every single applied pulse (in G o units, being G o = 77.5 µS the quantum of conductance unit) are plotted against the number of applied pulses. The conductivity state G was measured after the application of every pulse (Figure 5a, red pulses), by means of applying 50mV (Figure 5b, gray pulses) and reading the current flowing through the device.…”

“…In order to increase (decrease) the conductivity state of the device, a positive (negative) voltage has to be applied so that λ shifts towards g max (g min ), describing the Γ + (Γ − ) trajectories. The pair of logistic ridge functions Γ + and Γ − can be modeled with two cumulative distribution functions (cdf) [29], related to the pulse amplitudes applied to the non-linear memristive device, being V + , σ + and V − , σ − the average and standard deviation values of the cdf related to Γ + (for dV/dt > 0) and Γ − (dV/dt < 0) curves, respectively. Both of them define the boundaries of the possible conductivities of the device within a range limited by the minimum and maximum conductivity states, g min and g max , respectively.…”

“…For the sake of simplicity, a very simple input dataset is used as an example, for which a color identification task is provided. First of all, a model from a previous work [29] is used to analyze the plasticity property of the tested devices, which is further verified experimentally. A methodology for tuning the STDP function, which is a key element to control the learning process, is proposed.…”