SQUID magnetometry from nanometer to centimeter length scales
- Author(s): Hatridge, Michael Jonathan
- Advisor(s): Clarke, John
- et al.
Information stored in magnetic fields plays an important role in everyday life. This information exists over a remarkably wide range of sizes, so that magnetometry at a variety of length scales can extract useful information. Examples at centimeter to millimeter length scales include measurement of spatial and temporal character of fields generated in the human brain and heart, and active manipulation of spins in the human body for non-invasive magnetic resonance imaging (MRI). At micron length scales, magnetometry can be used to measure magnetic objects such as flux qubits; at nanometer length scales it can be used to study individual magnetic domains, and even individual spins. The development of Superconducting QUantum Interference Device (SQUID) based magnetometer for two such applications, in vivo prepolarized, ultra-low field MRI of humans and dispersive readout of SQUIDs for micro- and nanoscale magnetometry, are the focus of this thesis.
Conventional MRI has developed into a powerful clinical tool for imaging the human body. This technique is based on nuclear magnetic resonance of protons with the addition application of three-dimensional magnetic field gradients to encode spatial information. Most clinical MRI systems involve magnetic fields generated by superconducting magnets, and the current trend is to higher magnetic fields than the widely used 1.5-T systems. Nonetheless, there is ongoing interest in the development of less expensive imagers operating at lower fields. The prepolarized, SQUID detected ultra-low field MRI (ULF MRI) developed by the Clarke group allows imaging in very weak fields (typically 132 μT, corresponding to a resonant frequency of 5.6 kHz). At these low field strengths, there is enhanced contrast in the longitudinal relaxation time of various tissue types, enabling imaging of objects which are not visible to conventional MRI, for instance prostate cancer. We are currently investigating the contrast between normal and cancerous prostate tissue in ex vivo prostate specimens in collaboration with the UCSF Genitourinary Oncology/Prostate SPORE Tissue Core. In characterizing pairs of nominally normal and cancerous tissue, we measure a marked difference in the longitudinal relaxation times, with an average value of cancerous tissue 0.66 times shorter than normal prostate tissue. However, in vivo imaging is required to definitively demonstrate the feasibility of ULF MR imaging of prostate cancer. To that end, we have worked to improve the performance of the system to facilitate human imaging. This is accomplished by increasing the prepolarizing field amplitude, and minimizing magnetic noise in the SQUID detector. We have achieved polarizing fields as high as 150 mT and SQUID effective field noise below 1 fT Hz-1/2, enabling us to demonstrate proof-of-principle in vivo images of the human forearm with 2 x 2 x 10 mm3 resolution in 6 minutes.
On a much smaller spatial scale, there is currently fundamental and technological interest in measuring and manipulating nanoscale magnets, particularly in the quantum coherent regime. The observation of the dynamics of such systems requires a magnetometer with not only exceptional sensitivity but also high gain, wide bandwidth and low backaction. We demonstrate a dispersive magnetometer consisting of a two-junction SQUID in parallel with an integrated, lumped-element capacitor. Input flux signals are encoded as a phase modulation of the microwave drive tone applied to the magnetometer, resulting in a single quadrature voltage signal. For strong drive power, the nonlinearity of the resonator results in quantum limited, phase sensitive parametric amplification of this signal. We have achieved a bandwidth of 20 MHz---approximately two orders of magnitude higher than dispersive devices of comparable sensitivity---with an effective flux noise of 0.29 μΦ0Hz-(1/2). This performance is in excellent agreement with our theoretical model.