(a) Sketch of the samples geometry. 15 (b) Top view on the structured pattern for GISAXS measurements.
I-V curves demonstrating the forming behavior of the different sample systems. The first cycle is drawn in red. (a) Samples with Pt top electrodes show an irreversible forming step during the first cycle. Afterwards, stable resistive switching in eightwise polarity is observed. (b) The forming behavior of samples with Ti top electrodes depends strongly on the initial resistance which is depending on the electrode thickness. 15 This example shows a forming free electrode with a Ti thickness of 5 nm. (c) At low voltages, samples with Al top electrodes show resistive switching in eightwise polarity. The first cycle differs only slightly from the following ones. (d) If one increases the voltage, the flowing current increases suddenly. This is connected to an irreversible forming step. After this change of the sample, resistive switching in counter-eightwise polarity takes place.
Definition of angles and scattering vectors. Because of the large footprint of the X-ray beam, the sample has to be aligned perfectly parallel to it. This has been ensured by Cu structures parallel to the electrode arrays.
(a) AFM image of a bare substrate. A terrace structure can clearly be seen. (b) Modeling of the terrace structure for GISAXS simulations.
(a) The modeling of different parts of the samples which contribute to the scattering. (b) qy -cut of intensity originating from a correlated roughness. (c) and (d) qy -cut of intensity originating from cylindrical structures. (c) Influence of the mean diameter of cylinders. (d) Influence of the mean distance of cylinders.
(a) Sketch of the samples used for simulations. (b)–(d) Simulated qy -cuts of intensity for the sample system: (b) 20 nm Pt/20 nm Fe:STO/Nb:STO (c) 5 nm Ti/20 nm Fe:STO/Nb:STO, and (d) 10 nm Al/20 nm Fe:STO/Nb:STO. The intensity cuts have been taken along qz -values for which a maximum in the filament intensity was expected. Continuous lines: filament signature, (orange) x = 1, (green) x = 0.1. Dashed lines: background, (red) terraces, L = 250 nm, (blue) terraces, L = 25 nm, (black) grown islands on platinum top electrodes.
GISAXS scattering image for a reference sample which consists of 20 nm Fe:STO/Nb:STO only.
GISAXS results for sample I (20 nm Pt). (a) Scattering image and (b) qz -cut of scattering intensity along . The oscillations originate from the Pt layer. From their periodicity, the height of the top electrodes can be measured to be (19.8 ± 0.3) nm. In lateral cuts, no systematic change resulting from the electric treatment has been observed. This is due to the strongly scattering Pt top electrodes that cover any signal that might originate from the Fe:STO layer.
GISAXS results for sample II (5 nm Ti). (a) Scattering image. A laterally formed structure can be seen. (b) Vertical cut along qy = 0.2 nm−1. Arrows indicate oscillations. (c) Lateral cut along qz = 0.95 nm−1. Red line: Simulated GISAXS intensity with an arrow indicating a maximum.
GISAXS results for sample IV (10 nm Al). (a) Scattering image (b) qy -cut along . Black line: as-deposited sample. Red line: sample after applying a positive voltage of a few volts. After electroforming of the sample, a significant rise of the correlation maximum in the scattering image is observed.
Summary of results obtained during this work. Our findings clearly point at a filamentary type of the switching scenario.
Overview of top electrodes for samples prepared for GISAXS experiments. The samples in general consisted of top electrodes/20 nm Fe:STO/Nb:STO.
RMS roughness of sample layers, obtained by AFM measurements.
Article metrics loading...
Full text loading...