The Next Generation of Magnetoresistive Devices

Magnetometers have existed in some form for almost 200 years and play a critical role in our technological environment. New applications are continually developed as the resolution capabilities of these devices increases. State-of-the-art magnetometers such as superconducting quantum interference devices (SQUIDs) can have an excellent magnetic field resolution but are encumbered by the need for cryogenic liquid He cooling, heavy magnetic shielding, and extensive ancillary equipment and software which makes their operation expensive and outside of the reach of all but the most capital intensive institutions. It is therefore of scientific and technological interest to develop inexpensive and simple magnetometers which can also provide the fine field resolution necessary for many of the most challenging applications.

One such magnetometry technique which to date remains relatively unexplored but may hold significant promise derives from a phenomenon known as extraordinary magnetoresistance (EMR). This effect occurs in hybrid systems of high mobility semiconductors interfaced with a second, higher conductivity material. The flow of current through the devices is heavily altered by a magnetic field, as a Lorentz force at the interface causes the charge carriers to deflect and travel through the higher resistance semiconductor, leading to large changes in the device resistance. Recent developments in both materials science and computational numerical simulations present an opportunity to improve the performance of EMR devices to the point where they are competitive with established high-performance magnetometers.

In this thesis I detail the development of fabrication processes for producing EMR devices from InSb thin films, InAs quantum wells, and graphene encapsulated with hBN. While these processes have existed broadly for decades, they are novel in our research group and required extensive development before yielding usable devices. InSb and InAs samples were tested in magnetic fields up to 2 T and characterized, yielding insights about the intrinsic transport properties which were buttressed by analytical calculations and numerical simulations. In particular I show how the intrinsic magnetoresistance in InSb can be explained by the presence of multiple carrier bands and how the properties of the bands change upon thermal degradation of the material. EMR devices fabricated from these materials were also tested. I show how annealing can improve the contact resistivity by two orders of magnitude and results in an increase of the magnetoresistance from 870% to over 65,000%.

Secondly, I also demonstrate how a synergistic relationship can be formed between numerical simulations and experimental data. Finite element analysis (FEA) simulations were used to predict the performance of EMR devices and showed good agreement to experimental data. The simulations could also be used to explain deviations between the experimental data and the expected behavior. In particular, I show how the behavior of one of our devices can be explained by inhomogenous material properties as a result of the diffusion of gold into our structure. Electron diffraction spectroscopy measurements later supported this interpretation. The simulations were also used to study the effect of the sensor boundary on performance, a parameter which to-date had remained unexplored. The simulations predicted that increases in the maximum sensitivity and the presence of a zero-field sensitivity could be obtained if material was removed from key areas of the device. Experimental devices were realized with these geometries with behavior that matched the predictions from simulations in most cases. I found that the model deviates from experiments for the cases where the intrinsic magnetoresistance of the material is large and the EMR effect is weak and propose a method for including intrinsic magnetoresistance into the FEA simulations. The predictive power of the simulations was tested with topology optimized designs which in theory should show magnetoresistances on the order of 10¹¹%. However, while experimental devices made with these designs did not match the simulations they did show promising results with regards to the minimum field resolution suggesting that further tuning of the algorithm could yield significant results. Finally, I extended the FEA simulations into the realm of semiconductor nanowires where we show that more accurate estimations of the transport properties can be made with the help of numerical simulations.