Design of portable CMOS NMR system

Hong, Sungjin (Ph. D. in electrical and computer engineering)
Journal Title
Journal ISSN
Volume Title

Nuclear magnetic resonance (NMR) is the physical phenomenon that illustrates the behavior of the atomic nucleus under resonance condition. It has been an important tool for chemical spectroscopy. Recently, NMR spectroscopy becomes more popular with its wide variety of applications. Accordingly, the demand on the NMR spectrometers increases. However, the size and cost limit the widespread of the NMR spectroscopy. The limitations are resolved by miniaturization of the NMR spectrometers with CMOS chips and permanent magnets. Along with miniaturization, there are several problems arising from CMOS circuits and the magnets. This work addresses the existing problems in the portable NMR systems and proposes the circuit solutions. First, the portable NMR systems exploit the zero-intermediate frequency (IF) architecture by having the single frequency for sample excitation and local oscillator. From the architecture, the tradeoff between the inaccurate excitation and the 1/f noise is inevitable. This can be avoided by employing the separate frequencies with the IF frequency of 50 kHz. However, the dual-clock architecture randomizes the NMR output phase. The output with random phase disables the direct signal averaging. To solve this problem, the phase detection method by capturing the phase alignment of two clocks is proposed. The phase detector circuit triggers the on-chip timer that controls the NMR operation. This way, the output phase can remain constant. Second, the slow settling of the receiver deteriorates the acquisition sensitivity. This problem is solved with the proposed dynamic high-pass filter cutoff frequency switching technique. By doing so, the DC recovery time is 40 times faster than that without the technique, thereby achieving 10 µs of dead time. Lastly, the frequency fluctuation arises from large temperature coefficient of the permanent magnets. To calibrate the frequency variation, the frequency calibration method based on the signal peak detection is proposed. The peak detector extracts the amplitude of the NMR spin echoes which can be translated into the frequency response. The method exploits the natural behavior of the NMR signal whose amplitude is maximized at the frequency of interest. These techniques are combined with the fully integrated CMOS NMR transceiver. The proposed system demonstrates the one-dimensional (1D) NMR relaxation experiments for measuring T₂ relaxation time constant. The proposed features of fast receiver settling and high IF frequency allow the system to acquire the NMR signal that has the decaying time constant of less than 100 µs. Furthermore, NMR relaxation experiments are performed with the short time spacing between the refocusing pulses of 0.2 ms. The programmability and the high integration level extend the usage of the portable NMR system. The system with many circuit blocks such as delay-locked loop (DLL), timing controller, analog-to-digital converter (ADC), and the gradient controller allow various NMR experiments. The programmable NMR pulse sequence supports the two-dimensional (2D) experiments that characterize the two different physical properties of the sample. The inversion recovery method is employed to find the T₁ relaxation behavior, thereby producing the T₁-T₂ correlation of the samples. The system also demonstrates diffusion NMR experiments with the pulsed-field gradient. This diffusion information can be acquired with the on-chip gradient pulse sequencer and the supporting components such as the gradient coil, gradient amplifier, and the 3D-printed NMR probe.