Quantum-Dot-Based Thermometry Using 12-nm FinFET and Machine Learning Models

In this work, we demonstrate the use of a bulk FinFET designed in a 12-nm CMOS technology node, as a quantum dot (QD)-based thermometer at cryogenic temperatures. Although the operational principles of the proposed sensor bear a resemblance to that of a standard QD-based thermometric device such as quantum-dot thermometer (QDT), considerable attention to its operation is essential as the Fermi edge smearing of both the source and drain contributes to the drain current. To tackle the existing drawbacks of slower QDTs, we demonstrate the reliable and high-speed operation of the proposed sensor without compromising accuracy. Using FinFETs with identical fin dimensions, the QDs formed in the channel are highly uniform in size, nonspurious, and hence serve as a coherent, accurate, and yet reliable CMOS-based platform for designing QD-based devices. The Coulomb blockade (CB) regime of these FinFETs can be fine-tuned to any gate bias range using its threshold voltage and the number of fins. Use of machine learning (ML) models for QDT is also proposed by simplifying the approach of obtaining the operational temperatures from model-hardware correlations of the device. These models can be efficiently trained to the standard Fermi-function-based equation governing the device operation. Results show that replacing these Fermi-function-based equations with a linear equation significantly reduces the training time and the number of iterations, promising real-time evaluation of ON-chip heating in cointegrated quantum systems.