Silicon Chips Bridging Electronics and NMR

English Text | Lei Ka Meng
Photo | Provided by the author
Chinese Translation | Davis Ip

Complementary metal oxide semiconductor (CMOS) technology miniaturises electronic circuits onto a monolithic integrated circuit (IC) with a size of a matter of square millimetres. Since the invention of the CMOS technology in the 1960s, its performance has improved, radically bringing down the cost, footprint, and power consumption of electronic devices. Today, we are surrounded by CMOS ICs in our daily lives — whether or not we notice it. Common devices that contain CMOS ICs include cell phones, vehicles, Bluetooth headsets, and remote control devices.

CMOS as a Transformative Technology

Aside from the applications for computing and communications, CMOS technology is also useful in multidisciplinary research, which leverages the advantages of CMOS ICs to further exploration of science. The compact and versatile ICs can be customised to befit a wealth of applications, rendering it a powerful solution for scientific research.

Nuclear Magnetic Resonance (NMR) is an essential method to observe the molecular information and dynamics of the samples in the laboratory. It plays a crucial role in chemistry, physics, and biology. FiveNobel Prizes (in Chemistry, Physics, and Physiology or Medicine) have been awarded to NMR‑related development, manifesting the significance of NMR in scientific research. The equipment for the NMR
experiment mainly includes a magnet that generates astatic magnetic field to magnetise the nuclei for interest(1H in our platform), a radio‑frequency coil to couple between the magnetic field for the nuclei and electrical signal for the electronics, and transceiver (transmitter +receiver) electronics to excite the nuclei and record their responses for subsequent analysis.

Picture 1: The hardware for the portable NMR platform with parallelism. The core of the platform is the CMOS NMR transceiver, which was packaged onto a chip carrier for testing2.

Conventional NMR equipment is bulky and heavy because it employs a superconducting magnet to generate an intense magnetic field (> 1 Tesla) for magnetisation. Also, the discrete electronics occupy a significant area and consume a large amount of power.Recently, there is a thrust to miniaturise the hardware of NMR equipment by using a permanent magnet (< 0.5 Tesla) and a customised CMOS IC to replace the superconducting magnet and the discrete electronics.While the conventional NMR equipment provides unparalleled resolution for delicate analysis such as fingerprinting the macromolecules, the slim footprints of the miniaturised NMR systems open up a lot of potential NMR applications such as well‑logging and on‑site chemical screening.

Our CMOS NMR Platform

Our research team at the State Key Laboratory of Analog
and Mixed‑Signal VLSI at UM has been working on the miniaturisation of the NMR system over the past eight years and has developed different generations of portable NMR systems. Each generation has its respective innovation. For example, the first generation of the miniaturised NMR system focuses on automated sample management by integrating a digital microfluidic platform with the NMR system (Pictures 1 and 2)1, while the latest generation, on which we collaborated with the research group of Prof Donhee Ham at Harvard University, focuses on portable NMR with parallelism to expedite the NMR experiment for faster screening(Picture 3). Herein, we briefly discuss the key features of this latest research2.

Picture 2: The photo of theCMOS IC for our first portable NMR system1

At the heart of the portable NMR system is the customised CMOS IC. It mainly includes a pulse sequence synthesiser, a transmitter, and a receiver. The pulse sequence synthesiser generates different kinds of pulse sequences for the NMR experiments. The pattern of the pulse sequence manipulates the spinning of the nuclei and determines the responses from them. Hence, different information from the samples can be obtained by varying the pulse sequences. The transmitter amplifies the excitation signals generated by the pulse sequence synthesiser and drives the signals into the radio‑frequency coil, which excites the nuclei. The receiver, which is connected to the same radio‑frequency coil, picks up the diminutive signals from the coil and amplifies the signals for subsequent processing, such as filtering and frequency‑conversion. The size of the CMOS IC die is only 4 mm2.

A critical drawback of the NMR, whether for the conventional type or the miniaturised type, is its inherently low throughput. Since the NMR signals from the samples are very weak – usually at the order of microvolts, a lot of experiments are required and accumulated to improve the signal‑to‑noise ratio (SNR)of the acquisition. Yet, after the excitation from the transmitter, the nuclei takes several seconds to return to their thermal equilibrium. This long recovery time limits the repetition rate of the acquisition and throttles down the throughput of the experiment. Depending on the type of NMR experiment, it may take a matter of hours or even days to acquire the NMR signals with the desired resolution.

To tackle this limitation, we proposed different parallelisation schemes to accelerate the throughput of the measurement. The first one is the acceleration of the experiment via time‑interleaving. We placed multiple (N) samples (can be homogenous or heterogeneous) with their respective radio‑frequency coils within the permanent magnet. The coils are separated from each other to prevent cross interference. These N coils are connected to a CMOS IC by an array of switches. After the first coil finished the NMR scan on its sample and the nuclei are resting to return to equilibrium, the second coil will initiate another NMRs can on its sample. This process goes on sequentially until the Nth coil finishes its scan on the sample. Then the experiment starts over from the first coil, where the nuclei within the sample have already recovered to thermal equilibrium and are ready for another NMR scan. By exploiting the long recovery time of each NMR scan with the proposed time‑interleaving scheme, we can shorten the total NMR experiment time by a factor of N in the ideal case. In fact, from one of our experiments, we have used this miniaturized NMR platform with the time‑interleaving scheme (N = 2) to perform 2D 1H NMR experiment and elucidate the structure of ethyl formate and ethyl acetate (Picture4), where the total experiment time is 48 minutes(reduced by ~50%).

Picture 3: The system hardware of our first portable NMR system.The CMOS IC is mounted on the printed circuit boards and put inside the magnet1.

Inspired by the magnetic resonance imaging technique, which is based on NMR with additional magnetic field gradients to obtain spatial information, we proposed an NMR setup with gradient‑coils to spatially encode the samples. The gradient‑coils create magnetic field gradients in different directions. The spinning frequency of the nuclei is commensurate with the magnetic field they sensed. At the presence of the magnetic field gradient, nuclei at different regions of the sensing area experience dissimilar magnetic fields and thus spin at their individual rates. Hence, by performing Fourier Transform on the acquired signals to analyse their frequency contents, the amplitude of the NMR signals from a specific sample can be gathered. In this imaging‑based scheme, we put N samples inside the magnet. They share one radio‑frequency coil, which is connected to the CMOS IC. These N samples (N = 18in our demonstration) are then scanned simultaneously for probing the information of interest. These signals, which contain all information from the N samples, are then recorded and processed in the computer to extract the information for analysis. By using this imaging‑based scheme, the measurement speed is accelerated by 4.5times, as three extra scans are necessary to enhance
the SNR of the image, and it could be a full 18‑time acceleration if the SNR were not limited.

Picture 4: Measured 2D 1H NMR spectra (correlation spectroscopy) with time‑interleaving from ethyl acetate and ethyl formate2


Compared with the development of CMOS ICs for computing and communication purposes, the IC designed for multidisciplinary research is still in the early stage and research efforts are required to bridge the gap between the scientific research and the advances in micro‑ and nano‑electronics. Designing anIC for multidisciplinary research involves knowledge not only in circuit design but also in other areas such as physics and biology, along with delicate system‑level planning. Hence, it will be a challenging research area. Our team will continue to advance in this area, developing CMOS ICs for multidisciplinary research.

The views expressed in the Academic Research column are solely those of the authors, and do not necessarily reflect the views of the UMagazine or UM.

ISSUE 22 | 2020

Also in this issue

Lei Ka Meng is an assistant professor in the Institute of Microelectronics at UM. He obtained his PhD degree from UM in 2016. He was a postdoctoral fellow (visiting) in Prof Donhee Ham’s laboratory at Harvard University from 2017 to 2019. He has published nine peer‑reviewed journal articles and nine conference papers. He has given technical talks on his research in the United States, Portugal, and Italy. He has also co‑authored one book and one book chapter published by Springer. He holds one US patent.

1. K.‑M. Lei, P.‑I. Mak, M.‑K. Law, and R. P. Martins, “A μNMR CMOS transceiver using a Butterfly‑coil input for integration with a digital microfluidic device inside a portable magnet,” IEEE J. Solid‑State Circuits, vol. 51, no. 10, pp. 2274‑2286, Nov. 2016.

2. K.‑M. Lei, D. Ha, Y.‑Q. Song, R. M. Westervelt, R. P.Martins, P.‑I. Mak, and D. Ham, “Portable NMR with parallelism,” Analytical Chemistry, vol. 92, no. 2, pp.2112‑2120, Jan. 2020.