Electroforming Processes in Metal Oxide Resistive‐Switching Cells

Experimental Demonstration of Feature Extraction and Dimensionality Reduction Using Memristor Networks

“…Here the principal components were obtained by directly calculating the eigenvectors of the covariance matrix, where the primary principal component was obtained in the direction of the largest variance, and subsequently the second orthogonal principal component from the second greatest variance, and so forth. 26 Afterward, the data become clustered by transforming the input along the obtained principal components, as shown in Figure 2b. Instead of directly solving the eigenvectors from the covariance matrix, the principal components can also be obtained through neural networks through training, using Sanger's rule (eq 1).…”

“…In general, RRAM devices require an initial electroforming process, where a high voltage is used to create the ionic distributions necessary for subsequent resistiveswitching processes. 26 In a passive memristor crossbar, the high forming voltage (typically ∼5 V, whereas the set voltage is ∼1.0 V for Ta 2 O 5 -based devices 27 ) can cause damage to the halfselected devices that are already formed and share the same row as the target device, because a voltage of ∼2.5 V will be applied to these half-selected devices in a standard protective voltage scheme. Therefore, devices that are forming-free or require only low-voltage forming are essential for the successful operation of passive crossbar systems, that is, systems without the access transistor in one-transistor one-resistor (1T1R) type implementations.…”

“…Crossbar arrays based on a forming-free, tantalum-oxide memristor structure were used in this study to experimentally implement PCA analysis. In general, RRAM devices require an initial electroforming process, where a high voltage is used to create the ionic distributions necessary for subsequent resistive-switching processes . In a passive memristor crossbar, the high forming voltage (typically ∼5 V, whereas the set voltage is ∼1.0 V for Ta 2 O 5 -based devices) can cause damage to the half-selected devices that are already formed and share the same row as the target device, because a voltage of ∼2.5 V will be applied to these half-selected devices in a standard protective voltage scheme.…”

Effect of Dielectric Thickness on Resistive Switching Polarity in TiN/Ti/HfO2/Pt Stacks

“…Nonetheless, as precedent research on redox-based RRAM presenting VCM suggests, no clear dependence of RS parameters such as LRS, HRS, LRS/HRS ratio or switching voltages with film thickness should be expected [45][46][47]. Only forming voltage has shown a dependence with thickness [48,49], with some devices even turning formingfree if the dielectric is sufficiently thin [50,51]. Forming voltage, however, is an RS parameter clearly differentiated from the rest by the fact that the electroforming process is made when the sample is in a pristine state, and thus, there are no conductive filaments, making the thickness dependence unsurprising.…”

Estudio de la conmutación resistiva a partir de la caracterización eléctrica de estructuras MIM

“…Si el dieléctrico perdiese su capacidad aislante y quedase inutilizado, el término usado es hard breakdown. Aunque la mayoría de los dispositivos MIM que presentan conmutación resistiva requieren de este proceso de forming, en algunos casos se ha observado que no es necesario [75]. En caso de necesidad de la realización de un electroforming, la corriente debe limitarse para evitar la ya mencionada hard breakdown que destruiría el dispositivo, a esta corriente se la suele conocer como corriente límite o current compliance (I cc ).…”