High-temperature superconducting magnetometers for on-scalp MEG

Doktorsavhandling, 2020

In the growing field of on-scalp magnetoencephalography (MEG), brain activity is studied by non-invasively mapping the magnetic fields generated by neuronal currents with sensors that are flexibly placed in close proximity to the subject's head. This thesis focuses on high-temperature superconducting magnetometers made from YBa2Cu3Ox-7 (YBCO), which enables a reduction in the sensor-to-room temperature standoff distance from roughly 2 cm (for conventional MEG systems) down to 1 mm. Because of the higher neuromagnetic signal magnitudes available to on-scalp sensors, simulations predict that even a relatively low-sensitivity (higher noise) full-head on-scalp MEG system can extract more information about brain activity than conventional systems.

In the first part of this thesis, the development of high critical temperature (high-Tc) superconducting quantum interference device (SQUID) magnetometers for a 7-channel on-scalp MEG system is described. The sensors are single layer magnetometers with a directly coupled pickup loop made on 10 mm × 10 mm substrates using bicrystal grain boundary Josephson junctions. We found that the kinetic inductance strongly varies with film quality and temperature. Determination of all SQUID parameters by combining measurements and inductance simulations led to excellent agreement between experimental results and theoretical predictions. This allowed us to perform an in-depth magnetometer optimization. The best magnetometers achieve a magnetic field noise level of 44 fT/√Hz at 78 K. Fabricated test SQUIDs provide evidence that noise levels below 30 fT/√Hz are possible for high quality junctions with fairly low critical currents and in combination with the optimized pickup loop design. Different feedback methods for operation in a densely-packed on-scalp MEG system were also investigated. Direct injection of current into the SQUID loop was identified as the best on-chip feedback method with feedback flux crosstalk below 0.5%. By reducing the operation temperature, the noise level can be further reduced, however, the effective area also decreases because of the decreasing kinetic inductance contribution. We present a method that allows for one-time sensor calibration independent of temperature.

In the second part, the design, operation, and performance of the constructed 7-channel on-scalp MEG system based on the fabricated magnetometers is presented. With a dense (2 mm edge-to-edge) hexagonal head-aligned array, the system achieves a small sensor-to-head standoff distance of 1-3 mm and dense spatial sampling. The magnetic field noise levels are 50-130 fT/√Hz and the sensor-to-sensor feedback flux crosstalk is below 0.6%. MEG measurements with the system demonstrate the feasibility of the approach and indicate that our on-scalp MEG system allows retrieval of information unavailable to conventional MEG.

In the third part, two alternative magnetometer types are studied for the next generation system. The first alternative is magnetometers based on Dayem bridge junctions instead of bicrystal grain boundary junctions. With a magnetometer based on the novel grooved Dayem bridge junctions, a magnetic field noise level of 63 fT/√Hz could be achieved, which shows that Dayem bridge junctions are starting to become a viable option for single layer magnetometers. The second alternative are high-Tc SQUID magnetometers with an inductively coupled flux transformer. The best device with bicrystal grain boundary junctions reaches a magnetic field noise level below 11 fT/√Hz and outperforms the best single layer device for frequencies above 20 Hz.

In the last part, the potential of kinetic inductance magnetometers (KIMs) is investigated. We demonstrate the first high-Tc KIMs, which can be operated in fields of 9-28 µT and achieve a noise level of 4 pT/√Hz at 10 kHz.