Measurement of mobility in dual-gated MoS2 transistors

TL;DR

This paper investigates the measurement techniques of mobility in dual-gated MoS2 transistors, highlighting discrepancies and potential overestimations in reported mobility values, and emphasizes the importance of accurate measurement methods for reliable device characterization.

Contribution

The study provides a detailed analysis of measurement techniques in dual-gated MoS2 transistors, identifying sources of overestimation and proposing more accurate assessment methods.

Findings

Double-gated measurements often overestimate mobility due to measurement artifacts.

Significant discrepancy exists between top-gate and back-gate mobility measurements.

Proper measurement techniques are crucial for accurate mobility evaluation in MoS2 devices.

Abstract

Atomically thin semiconducting MoS2 is of great interest for high-performance flexible electronic and optoelectronic devices. Initial measurements using back-gated field-effect transistor (FET) structures on SiO2 yielded mobility of 1-50 cm^2/Vs for few-layer MoS2[1, 2]. Initial measurements using back-gated field-effect transistor (FET) structures on SiO2 yielded mobility of 1-50 cm^2/Vs for few-layer MoS2[1, 2]. However, greatly increased mobility - as high as 900 cm^2/Vs - was recently reported for monolayer MoS2 by Radisavljevic, et al.[Ref. 3; see also Refs. 4-6], and for multilayer MoS2 by others[7, 8], in devices covered by a high-kappa dielectric layer and a metal top gate. Similar increases were reported for electrolyte top-gated MoS2[9]. A puzzling common aspect of these reports is that the double-gated devices consistently show much higher mobility when measured using the weakly-coupled back gate (typically 300 nm of SiO2) than using the strongly-coupled top gate. While this difference might be attributed to a different quality of the top and bottom surfaces, the large discrepancy motivates close examination of the measurement techniques to rule out spurious effects. Notably, it has been shown that the mobility may be greatly overestimated in this geometry[10, 11]. There is good evidence that this is the case here[4-9]; below we discuss Radisavljevic, et al.[4] as an illustrative example.