Electrical characteristics of a Co/ TiOx/Co resistive memory device, fabricated by two different methods, are reported. In addition to crystalline TiO2 layers fabricated via conventional atomic layer deposition (ALD), an alternative method has been examined, where TiOx nanoparticle layers were fabricated via sol-gel. The different devices have shown different hysteresis loops with a unique crossing point for the sol-gel devices. A simple qualitative model is introduced to describe the different current-voltage behaviours by suggesting only one active metal-oxide interface for the ALD devices and two active metal-oxide interfaces for the sol-gel devices. Furthermore, we show that the resistive switching behaviour could be easily tuned by proper interface engineering and that despite having a similar active material, different fabrication methods can lead to dissimilar resistive switching properties.
References
Values of the relaxation parameter kt between 10 and 40, expressed in units of the numerical integration time step, qualitatively reproduce the I-V curve of Fig. 2(b). The number of steps in these simulated I-V curves is the same as the number of voltage steps in Fig. 2(b) during a complete cycle. Since the voltage step time used in the experiment is 60 ms, kt = 10 corresponds to a decay constant of 600 ms. In order to have a direct measurement of the retention time of the top interface, we measured HSLs with different waiting times after each writing pulse (between 20 ms and 5 s). Then, by fitting the HSL current values at Vwrite = 0 with the exponential decay equation Irem(0+)/Irem(0−) − 1 = A∞ + A1 exp (−t/τ), we extract a decay constant of τ ≈ 100 ms, which is in very good agreement with the one obtained from the numerical simulations of our toy model.
