Datasets
Standard Dataset
Resistive Drift of Titanium Dioxide Memristor Devices
- Citation Author(s):
- Submitted by:
- Waleed El-Geresy
- Last updated:
- Thu, 05/30/2024 - 11:55
- DOI:
- 10.21227/4qbk-zh17
- Data Format:
- License:
52 Views- Categories:
- Keywords:
Abstract
Dataset Description
A dataset containing data collected from measuring the drift characteristics titanium dioxide memristors, for a variety of initial states, under zero-bias conditions.
The devices used were TiO2 devices, with Al2O3 interstitials - the same device type as used in [1].
Data Collection
Devices were electroformed using a series of steadily increasing voltage pulses, ranging in magnitude from 3V to 10V with a step of 0.5V. The electroforming pulse widths ranged from 10µs to 100µs.
Following electroforming, devices were randomly programmed to an initial resistance value less than 100kΩ.
30ms, 0.3V read pulses were used to measure the device state every 10 seconds using Kelvin sensing following an initial programming step for each resistive drift series.
Following data collection, a small number of series were excluded where devices were deemed to have been incorrectly electroformed, or to have failed, due to displaying highly atypical drift characteristics. As a further post-processing step, anomalous read values within series, identified as being those exceeding 100kΩ, were adjusted in order to denoise the data.
References
[1] Strukov, D., Snider, G., Stewart, D. et al. The missing memristor found. Nature 453, 80–83 (2008).
The zip file contains directories named as "dX_sa20_Y", where X is the crossbar identifier (either 67 or 69) and Y is the experiment number. Thus, all CSV files within a directory will contain data for retention measurements of the same experiment duration (although the duration varies between experiments).
Each of these directories contains a number of CSV files labelled "Z_A_proc.csv", where Z and A reference the coordinates of the memristor to which the data in that CSV file corresponds (in the respective array).
Each CSV has headers "t" and "r", with the data corresponding to continuous series of pairs of values of the time in seconds of each measurement and the measured instantaneous resistance value in Ohms.
