Noise Measurements and Analyses of Nanodevices

Abstract

Usually, in transport experiments on 2D materials, mean values of quantities like voltage and current are measured. However, even in steady states, these quantities show fluctuations around their mean values, which is called as noise. Noise presents multiple new statistical quantities that we can measure in order to probe the microscopic processes occurring in the material. Here, we measure the Power Spectral Density of noise, which naturally separates out the processes occurring on different time scales since it is frequency resolved. Noise power spectral density is usually calculated by measuring the transport quantity at high sampling rates and performing algorithms like Welch's periodogram on the data collected. In such cases, the lowest magnitudes of the noise power spectral density that can be measured is often limited by the power spectral densities of the instruments used for the measurement, usually the preamplifier used for signal conditioning. The cross-correlation technique enables us to measure noise magnitudes below the noise floor of the measuring apparatus by making two simultaneous measurements at once. In this thesis, we implemented a cross correlation setup that can measure voltage noise power spectral density magnitudes as low as 1.6$\times 10^{-19}$ V$^2$/Hz, which is below the noise floor of the voltage amplifiers we used. Next, we probed devices made from 2D materials using noise spectroscopy. We performed temperature-dependent noise spectroscopy measurements on CrSBr in order to determine presence and type of n-type donors. We detected signatures of generation-recombination processes at low temperatures in the form of random telegraph noise whose attempt frequencies had an Arrhenius-type of variation with temperature. For tellurene, our aim was to detect signatures of presence of phase transitions. So, we performed temperature- and gate voltage-dependent noise spectroscopy. Through temperature dependent measurements, we detected increase of 1/f noise exponent at 220K, which could be a signature of critical fluctuations. However, time-domain analysis revealed that fluctuations are due to a very slow random telegraph process whose attempt frequencies were outside our measurement window. This serves to highlight the complexities involved in interpreting noise data. We also present here the data of how the 1/f noise magnitude changes with gate voltage at a fixed temperature.