Multiple Avalanches across the Metal-Insulator Transition of Vanadium Oxide Nanoscaled Junctions

The metal-insulator transition of nanoscaled ${\mathrm{VO}}_2$ devices is drastically different from the smooth transport curves generally reported. The temperature driven transition occurs through a series of resistance jumps ranging over 2 decades in magnitude, indicating that the transition is caused by avalanches. We find a power law distribution of the jump sizes, demonstrating an inherent property of the ${\mathrm{VO}}_2$ films. We report a surprising relation between jump magnitude and device size. A percolation model captures the general transport behavior, but cannot account for the statistical behavior.

Figure 1

Main panel—8 consecutive $R-T$ cycles ($R$ in linear scale) of a $1\times 6\mu {\mathrm{m}}^2$ ${\mathrm{VO}}_2$ device zoomed in on part of the MIT, as marked in the full measurement shown in (b) (log scale of $R$). (a)—Image of 8 devices on one sample showing ${\mathrm{VO}}_2$ square of side $50\mu \mathrm{m}$, on top of which are $\mathrm{V}/\mathrm{Au}$ electrodes defining device lengths of 1,2,3, and $4\mu \mathrm{m}$ (2 devices of each) and width of $8\mu \mathrm{m}$ for all the devices. Devices with length of 1 and $4\mu \mathrm{m}$ are marked.

Figure 2

Average value of each of the three largest jumps as a function of device length. The inset plots only the average of the largest jump as a function of inverse the device length. Error bars are standard deviations.

Figure 3

Histogram of the jump sizes plotted on a log-log scale, acquired from over 100 cycles measured for a sample $1\times 6\mu {\mathrm{m}}^2$. Solid line (red online) is calculated using the maximum likelihood method, indicating a power law dependence. The exponent $\alpha =2.48\pm 0.05$. a- $\alpha$ for different channel lengths from different samples. Data for the $1\mu \mathrm{m}$ device are provided from 3 different samples deposited under the same nominal conditions (on one sample 2 devices were measured). b- $\alpha$ as a function of temperature ramp rate.

Figure 4

a- Resistance (bottom curve, left axis) and temperature (top curve, right axis) as a function of time. No avalanches are observed for constant temperatures. b- Simulation of $R-T$ (thick line, blue online) along with experimental data (thin line, black) from a $1\times 6\mu {\mathrm{m}}^2$ device, cooling branch. The simulation is of size $20\times 50$ ($\mathrm{\text{length}}\times \mathrm{\text{width}}$) on a square lattice and ${\rho }_I/{\rho }_M=1000$.