Donhee Ham

We present a compact, IC-based low-field NMR spectrometer that achieves diffusion-sensitive measurements without bulky gradient hardware, by using the inherent static magnetic gradient of a 0.5-T Halbach magnet. A custom 180-nm CMOS NMR ASIC integrates an arbitrary pulse sequencer, an RF transmitter, and a low-noise receiver within a die area of 5 mm2, enabling fully portable operation. Using the static gradient (gavg = 0.5 T/m), we demonstrate two-dimensional (2D) diffusion–relaxation (D–T2) experiments on solid peanuts during drying at 60°C. These measurements reveal oil molecular mobility and moisture loss without oil extraction. The system maintains a signal-to-noise ratio (SNR) of 29 dB and stable operation across off-center regions with strong static gradients. These results establish a chip-scale, gradient-electronics-free NMR platform that bridges integrated-circuit design and magnetic resonance sensing, enabling field-deployable, non-destructive analysis of agricultural, food, and biological materials.

– Intracellular recording offers exquisite access to the electrical interior of a neuron––detecting not only spikes but also the small synaptic signals that reflect synaptic connections and their properties––but it had been limited to only one cell at a time. In 2020, a CMOS nanoelectrode array broke this constraint and massively parallelized intracellular recording, opening a path to capturing synaptic signals across a neuronal network. This shift led us in 2021 to propose the copy framework: population-scale intracellular recording is a process of copying a biological network, because the resulting data––rich with synaptic signals from across the network––reveal its synaptic connectivity map. In this Perspective , we revisit that framework because the recent CMOS intracellular microelectrode array (iMEA), building on the 2020 nanoelectrode array, transforms copy from a demanding demonstration into a practical, high-yield technology. By routinely recording intracellular signals across thousands of neurons, the iMEA uncovers synaptic organization with a breadth and clarity previously unreachable. We outline the development of the iMEA as an improved copy platform, what it now enables, and what must still be advanced to build a richer synaptic connectivity map––one that is indispensable to understanding how the brain functions, and provides a firmer foundation for neuromorphic models grounded in the brain’s own wiring.

Patch-clamp electrodes capture intracellular signals from few neurons, while microelectrode arrays (MEAs) record many neurons extracellularly with limited sensitivity. Bridging these approaches has been a long-standing goal. We report a 4,096-channel semiconductor microhole electrode array that records intracellularly from thousands of neurons in parallel. From over 2,000 neurons, more than 70,000 plausible synaptic connections were identified and classified as electrical, inhibitory, or excitatory, with about 5% error. This platform enables large-scale mapping of neuronal network connectivity with single-cell resolution.

Obtaining a high Rabi oscillation frequency homogeneously across a spatially extended population of nitrogen-vacancy (NV) center electronic spins in diamond is useful for efficient spin-state manipulation of the NV ensemble and in using NVs to detect ensembles of other spin species. Here, we achieve a high, homogeneous Rabi frequency for a dense NV ensemble by enhancing the microwave magnetic fields in the center region of a diamond-coupled planar metallic loop via systematic engineering that increases the microwave current driving of the loop, while avoiding off-center proximity to the loop that gives strong but inhomogeneous microwave fields. With such enhanced microwave fields at 2.55 GHz, we achieve a 136.3 MHz NV Rabi frequency with 1.5% inhomogeneity over a 40 μm × 40 μm diamond area, and we use the NV ensemble to detect a ∼30-MHz magnetic signal, similar to a nuclear magnetic resonance (NMR) signal at a tesla-scale bias magnetic field, with Hz-scale spectral resolution.

The speed of a plasmonic wave in the presence of electron drift in a conductor depends on the wave's propagation direction, with the wave traveling along the drift (“forward wave”) faster than the wave traveling against the drift (“backward wave”). Phenomena related to this plasmonic non-reciprocity—which is relatively more pronounced in two-dimensional conductors than in bulk conductors and could lead to solid-state device applications—have been studied in THz and optical spectral regimes. Here, we demonstrate the plasmonic non-reciprocity at microwave frequencies (10–50 GHz). Concretely, we conduct, at 4 K, a microwave network analysis on a gated GaAs two-dimensional electron gas with electron drift (i.e., DC), directly measuring the forward and backward wave speeds via their propagation phase delays. We resolve, for example, forward and backward wave speeds of 4.26×10−3±8.97×10−6 (normalized to the speed of light). Sufficient consistency between the electron drift speed obtained from the microwave measurement and that alternatively estimated by a DC transport theory further confirms the non-reciprocity. We conclude this paper with a discussion on how to enhance the non-reciprocity for real-world applications, where degeneracy pressure would play an important role.

Obtaining a high Rabi oscillation frequency homogeneously across a spatially-extended population of nitrogen-vacancy (NV) center electronic spins in diamond is useful for efficient spin-state manipulation of the NV ensemble and in using NVs to detect ensembles of other spin species. Here, we achieve a high, homogeneous Rabi frequency for a dense NV ensemble by enhancing the microwave magnetic fields in the center region of a diamond-coupled planar metallic loop via systematic engineering that increases the microwave current driving of the loop, while avoiding off-center proximity to the loop that gives strong but inhomogeneous microwave fields. With such enhanced microwave fields at 2.55 GHz, we achieve a 136.3 MHz NV Rabi frequency with 1.5% inhomogeneity over a 40 $\times$ 40 $μm^{2}$ diamond area; and use the NV ensemble to detect a ~30-MHz magnetic signal, similar to a nuclear magnetic resonance signal at a tesla-scale bias magnetic field, with Hz-scale spectral resolution.

Always-on AI sensor applications–based on deep neural networks (DNNs)–with sparse inference require low power consumption during both computing and idle phases. In-memory computing (IMC) with non-volatile memory crossbar arrays could meet this demand. Concretely, the co-location of memory (DNN weight storage) and computing (analog matrix multiplications (MMs)) in crossbar arrays obviates the need to shuttle weight data, thus reducing the computing power consumption, while the use of the non-volatile memory minimizes the power consumption during idle states. Resistive, phase-change, and magneto-resistive random-access memory (RRAM, PRAM, and MRAM) and flash memory are well-known non-volatile memory types, with their own merits and drawbacks. Of these, MRAM boasts high endurance and low switching energy (with the drawback being 1b storage per cell) [1] and has been commercially embedded in CMOS logic technology, and thus MRAM is a good non-volatile memory candidate for IMC. However, previous MRAM IMC works [2]–[5] have been focused on individual crossbar arrays, which correspond only to a small fraction of a DNN.

The massive parallelization of neuronal intracellular recording, which enables the measurement of synaptic signals across a neuronal network, and thus the mapping and characterization of synaptic connections, is an open challenge, with the state of the art being limited to the mapping of about 300 synaptic connections. Here we report a 4,096 platinum/platinum-black microhole electrode array fabricated on a complementary metal-oxide semiconductor chip for parallel intracellular recording and thus for synaptic-connectivity mapping. The microhole–neuron interface, together with current-clamp electronics in the underlying semiconductor chip, allowed a 90% average intracellular coupling rate in rat neuronal cultures, generating network-wide intracellular-recording data with abundant synaptic signals. From these data, we extracted more than 70,000 plausible synaptic connections among more than 2,000 neurons and catalogued them into electrical synaptic connections and into inhibitory, weak/uneventful excitatory and strong/eventful excitatory chemical synaptic connections, with an estimated overall error rate of about 5%. This scale of synaptic-connectivity mapping and the ability to characterize synaptic connections is a step towards the functional connectivity mapping of large-scale neuronal networks. The parallel intracellular recording of neurons via an array of 4,096 microhole electrodes on a semiconductor chip enables the large-scale mapping of synaptic connections, as shown with rat neuronal cultures.

The speed of a plasmonic wave in the presence of electron drift in a conductor depends on the wave's propagation direction, with the wave traveling along the drift (`forward wave') faster than the wave traveling against the drift (`backward wave'). Phenomena related to this plasmonic non-reciprocity -- which is relatively more pronounced in two-dimensional conductors than in bulk conductors and could lead to solid-state device applications -- have been studied in THz and optical spectral regimes. Here we demonstrate the plasmonic non-reciprocity at microwave frequencies (10 $\sim$ 50 GHz). Concretely, we conduct, at 4K, a microwave network analysis on a gated GaAs two-dimensional electron gas with electron drift (i.e., DC current), directly measuring out forward and backward wave speeds via their propagation phase delays. We resolve, for example, forward and backward wave speeds of $4.26 \times 10^{-3} \pm 8.97 \times 10^{-6}$ (normalized to the speed of light). Sufficient consistency between the electron drift speed obtained from the microwave measurement and that alternatively estimated by a DC transport theory further confirms the non-reciprocity. We conclude this paper with a discussion on how to enhance the non-reciprocity for real-world applications, where degeneracy pressure would play an important role.

Background/Aims. Microelectrode arrays (MEAs) realized on and operated by CMOS integrated circuits, typically containing thousands of densely packed electrodes, can record neural activities with high spatial (e.g., ~15 µm) and high temporal resolution (e.g., ~20 kHz bandwidth) [1-3]. The recorded voltage signals are accompanied by noise of biological and electronic origins, and by filtering out this noise, one can see electrophysiological signals, in particular, action potentials, more clearly. However, the noise may be actually exploited for the label-free and noninvasive detection of adherent cells, based on the fact that the voltage noise from the resistive adhesion cleft may vary depending on whether the cells adhere or not [4]. The cell adhesion noise (CAN) might be altered by the electrode size [5], the size of the adherent cells, the junction capacitance, the cell type, and the corresponding sampling frequency [6]. We therefore record the voltage noise of adherent cell cultures from different types of CMOS MEAs and their corresponding recording systems, each of them being optimized in a different working regime. Based on previous studies with neurons, the cleft resistance between a cell and a micron-scale recording site contributes significantly to the voltage noise, thus distinguishing this value from the one of a bare sensor site [4, 7-8]. We analyze the voltage noise in terms of power spectral density (SV) to generate electrical images from the SV-derived CAN maps with single-cell resolution to assess the reliability of the label-free cell imaging across different CMOS MEA biosensing platforms. This label-free cell detection is of broad interest for biotechnological applications, for instance, to determine the cells’ proliferation status [9] or to record sparsely distributed regions on the MEA, where densely packed neural cells require high-density arrangements [10]. In contrast to optical imaging or standard biological dead-end assays, CAN spectroscopy offers label-free and, in principle, continuous recording capability and could be used to track the neural system as it adheres to the MEA surface. Methods. The colorectal cancer (CRC) cells and E18 primary neuron cells were plated in monocultures on the CMOS MEAs coated with collagen type I and poly-D-lysine. We analyzed the recorded voltage noise power spectral density SV at 35 kHz. After detecting adherent cells, the CAN-based electrical images were compared with light microscopic images to relate the estimated cell positions to ground truth [9]. Results. We calculated the root-mean-square (RMS) and the SV of the noise levels of CMOS MEAs on the recording platforms supplied with PBS (1X) of conductivity κ=16 mS/cm. The average RMS voltage noise of hundreds of sensor sites of CMOS MEA type I showed 15 µV (in accordance with [11]) with an SV of 0.0009 µV²/Hz. The CMOS MEA type II was at 57 µVRMS (in accordance with [12]) with SV of 0.05 µV²/Hz and CMOS MEA type III at 75.8 µVRMS with SV of 0.013 µV²/Hz (s. Figure 1). The adhesion noise spectrum from different sensor sites with adherent cells consistently shows uniform profiles with elevated values compared to sensors without cells. We reproducibly accomplished adhesion noise-based cell identification across different CMOS MEA biosensing platforms independent of the noise levels thereof with high correspondence (>80 %) between electrically and microscopically estimated cell positions. Conclusion. Adhesion noise spectroscopy constitutes a potent tool for label-free and noninvasive cell detection with high accordance between electrical and brightfield microscopy imaging. Future work aims to record neural activity at a resolution of 6 µm and to detect cancer spheroids and organoids.

We introduce a bioelectronic interface between biological electrogenic cells and a mixed-signal CMOS integrated circuit with an array of surface electrodes, where not only is the CMOS electrode array capable of electrophysiological recording and stimulation of the cells with 1,024 recording and stimulation channels, but it can also provide low-latency artificial signal pathways from cells it records to cells it stimulates. This on-chip closed-loop modulation has an intrinsic latency less than 5 µs. To demonstrate the utility of the on-chip closed loop modulation as an artificial feedback pathway between biological cells, we develop a silicon-cardiomyocyte self-sustained oscillator with a tunable frequency to which both the relevant part of the CMOS chip and cells are locked, and also a silicon-neuron interface with a silicon inhibitory connection between neuronal cells. This line of cyto-silicon hybrid system, where the boundary between biological and semiconductor systems is blurred, may find applications in prosthesis, brain-machine interface, and fundamental biology research.

Author(s): Ramesh, Ramamoorthy; Salahuddin, Sayeef; Datta, Suman; Diaz, Carlos H; Nikonov, Dmitri E; Young, Ian A; Ham, Donhee; Chang, Meng-Fan; Khwa, Win-San; Lele, Ashwin Sanjay; Binek, Christian; Huang, Yen-Lin; Sun, Yuan-Chen; Chu, Ying-Hao; Prasad, Bhagwati; Hoffmann, Michael; Hu, Jia-Mian; Yao, Zhi; Bellaiche, Laurent; Wu, Peng; Cai, Jun; Appenzeller, Joerg; Datta, Supriyo; Camsari, Kerem Y; Kwon, Jaesuk; Incorvia, Jean Anne C; Asselberghs, Inge; Ciubotaru, Florin; Couet, Sebastien; Adelmann, Christoph; Zheng, Yi; Lindenberg, Aaron M; Evans, Paul G; Ercius, Peter; Radu, Iuliana P

The bioinspired camera, comprising a single lens and a curved image sensor-a photodiode array on a curved surface-, was born of flexible electronics. Its economical build lends itself well to space-constrained machine vision applications. The curved sensor, much akin to the retina, helps image focusing, but the curvature also creates a problem of image distortion, which can undermine machine vision tasks such as object recognition. Here we report an anti-distortion single-lens camera, where 4096 silicon photodiodes arrayed on a curved surface in a nonuniform pattern assimilated to the distorting optics are the key to anti-distortion engineering. That is, the photo-pixel distribution pattern itself is warped in the same manner as images are warped, which correctively reverses distortion. Acquired images feature no appreciable distortion across a 120° horizontal view, as confirmed by their neural-network recognition accuracies. This distortion correction via photo-pixel array reconfiguration is a form of in-sensor computing.

We present a small wideband CMOS nuclear magnetic resonance (NMR) spectrometer capable of multinuclear molecular fingerprinting (molecular structure determination). The spectrometer combines a small permanent magnet and a wideband digitally assisted CMOS radio frequency (RF) transceiver integrated circuit (IC) that operates in the RF frequency range from 1 to 100 MHz. This small, thus broadly deployable, NMR system can find applications in online, on-site, or on-demand multinuclear molecular fingerprinting. Concretely, the wideband spectrometer performs NMR spectroscopy with nuclear spins from 1H, 19F, and 2H with a spectral resolution down to 0.07 ppm to determine structures of molecules containing these nuclear spins; it can also perform NMR with the nuclear overhauser effect (NOE), which is an important modality that further enhances the capability for molecular fingerprinting (we demonstrate the NOE by reading the 19F NMR signal that is enhanced via heteronuclear magnetization transfer from 1H spins). The wideband operation of the CMOS RF transceiver IC is made possible by integrating a delay-locked loop (DLL) with a broad locking range. The RF receiver (RX) part of the IC achieves an input-referred noise less than 1.2 nV/<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\surd $ </tex-math></inline-formula> Hz across the entire bandwidth of 1–100 MHz, while the RF transmitter (TX) part employs a digital power amplifier.

In situ monitoring of bacterial growth can greatly benefit human healthcare, biomedical research, and hygiene management. Magnetic resonance imaging (MRI) offers two key advantages in tracking bacterial growth: non-invasive monitoring through opaque sample containers and no need for sample pretreatment such as labeling. However, the large size and high cost of conventional MRI systems are the roadblocks for in situ monitoring. Here, we proposed a small, portable MRI system by combining a small permanent magnet and an integrated radio-frequency (RF) electronic chip that excites and reads out nuclear spin motions in a sample, and utilize this small MRI platform for in situ imaging of bacterial growth and biofilm formation. We demonstrate that MRI images taken by the miniature--and thus broadly deployable for in situ work--MRI system provide information on the spatial distribution of bacterial density, and a sequential set of MRI images taken at different times inform the temporal change of the spatial map of bacterial density, showing bacterial growth.

High-content imaging for compound and genetic profiling is popular for drug discovery but limited to endpoint images of fixed cells. Conversely, electronic-based devices offer label-free, live cell functional information but suffer from limited spatial resolution or throughput. Here, we introduce a semiconductor 96-microplate platform for high-resolution, real-time impedance imaging. Each well features 4096 electrodes at 25 µm spatial resolution and a miniaturized data interface allows 8× parallel plate operation (768 total wells) for increased throughput. Electric field impedance measurements capture >20 parameter images including cell barrier, attachment, flatness, and motility every 15 min during experiments. We apply this technology to characterize 16 cell types, from primary epithelial to suspension cells, and quantify heterogeneity in mixed co-cultures. Screening 904 compounds across 13 semiconductor microplates reveals 25 distinct responses, demonstrating the platform's potential for mechanism of action profiling. The scalability and translatability of this semiconductor platform expands high-throughput mechanism of action profiling and phenotypic drug discovery applications.

We report a mixed-signal CMOS chip with an array of surface electrodes, which is capable of not only electrophysiological recording and stimulation of biological cells but also low-latency closed-loop modulation between the recorded and stimulated cells. To demonstrate the utility of the on-chip closed-loop modulation as an artificial feedback pathway between biological cells, we have developed a silicon-cardiomyocyte self-sustained oscillator with a tunable locked frequency and a silicon-neuron interface that offers an artificial (silicon) inhibitory connection between neurons. These chip-cell interfaces smear the boundary between biological and semiconductor systems.

We demonstrate quantum logic enhanced sensitivity for a macroscopic ensemble of solid-state, hybrid two-qubit sensors. We achieve over a factor of 30 improvement in the single-shot signal-to-noise ratio, translating to an ac magnetic field sensitivity enhancement exceeding an order of magnitude for time-averaged measurements. Using the electronic spins of nitrogen vacancy (NV) centers in diamond as sensors, we leverage the on-site nitrogen nuclear spins of the NV centers as memory qubits, in combination with homogeneous and stable bias and control fields, ensuring that all of the ∼10^{9} two-qubit sensors are sufficiently identical to permit global control of the NV ensemble spin states. We find quantum logic sensitivity enhancement for multiple measurement protocols with varying optimal sensing intervals, including XY8 and DROID-60 dynamical decoupling, as well as correlation spectroscopy, using an applied ac magnetic field signal. The results are independent of the nature of the target signal and broadly applicable to measurements using NV centers and other solid-state spin ensembles. This work provides a benchmark for macroscopic ensembles of quantum sensors that employ quantum logic or quantum error correction algorithms for enhanced sensitivity.

Aqueous Ionic Circuits An ionic circuit developed by Woo-Bin Jung, Donhee Ham, and co-workers in article number 2205096 computes in water. It can execute a core of neural-net computing in an analog manner fully based on electrochemical principles in an aqueous solution of quinone. This ionic circuit thus demonstrates a step toward sophisticated aqueous ionics.

We report a wideband (1-100MHz) CMOS spectrometer capable of multinuclear nuclear magnetic resonance (NMR) spectroscopy and demonstrate it by performing 119, and 2H NMR spectroscopy with high spectral resolution down to 0.07 ppm. This is made possible by developing a wideband, digitally assisted CMOS RF transceiver, where a delay-locked loop (DLL) with a broad locking range is the most critical functional block for enabling operation across the full experimental bandwidth. This portable multinuclear NMR system can enhance the capability of online molecular fingerprinting.

Profiling compounds and genetic perturbations via high-content imaging has become increasingly popular for drug discovery, but the technique is limited to endpoint images of fixed cells. In contrast, electronic-based devices offer label-free, functional information of live cells, yet current approaches suffer from low-spatial resolution or single-well throughput. Here, we report a semiconductor 96-microplate platform designed for high-resolution real-time impedance "imaging" at scale. Each well features 4,096 electrodes at 25 µm spatial resolution while a miniaturized data interface allows 8× parallel plate operation (768 total wells) within each incubator for enhanced throughputs. New electric field-based, multi-frequency measurement techniques capture >20 parameter images including tissue barrier, cell-surface attachment, cell flatness, and motility every 15 min throughout experiments. Using these real-time readouts, we characterized 16 cell types, ranging from primary epithelial to suspension, and quantified heterogeneity in mixed epithelial and mesenchymal co-cultures. A proof-of-concept screen of 904 diverse compounds using 13 semiconductor microplates demonstrates the platform's capability for mechanism of action (MOA) profiling with 25 distinct responses identified. The scalability of the semiconductor platform combined with the translatability of the high dimensional live-cell functional parameters expands high-throughput MOA profiling and phenotypic drug discovery applications.

Nuclear magnetic resonance (NMR) is a paramount analytical tool for chemistry, biology, medicine, and geology, and has a fundamental importance in physics. The recent years have seen a wealth of efforts to miniaturize NMR systems by combining permanent magnets and CMOS radio frequency (RF)-integrated circuits (ICs) to make the benefit of NMR more broadly available beyond dedicated facilities, which have resulted in systems capable of NMR relaxometry first, and later NMR spectroscopy, the two key NMR modalities. Here we report a small NMR system comprising a digitally assisted CMOS RF transceiver IC and a 0.51-T permanent magnet, which not only enhances spectroscopy and relaxometry performance but also includes magnetic resonance imaging (MRI), a powerful variant of relaxometry. The system achieves a spectral resolution of < 0.05 ppm (1.1 Hz), the highest reported in a portable CMOS-based NMR system, and an imaging resolution of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$67\times 67 \times 83\,\, \boldsymbol {\mu }\text{m}^{3}$ </tex-math></inline-formula> .

Flash memory is distinct from other types of data storage––such as hard disk drives (HDDs) that are still in active use and floppy disks, CDs, and DVDs that became either extinct or unpopular––in that it is a semiconductor data storage. Smaller, lighter, and lower-power than HDD, flash memory made its first major commercial success as a storage for smartphones and has since made mobile devices truly mobile. It is now a major alternative to HDD even in personal computers, and is penetrating other markets HDD once dominated. Here, we review how flash memory enabled the data-driven world through a sequence of innovations that helped sustain the continued increase of storage capacity, with the highlight being the vertical integration that opened a new era for the memory industry. We will then discuss the emerging cold data storage based on DNA that can lie below flash memory in the memory hierarchy to possibly handle the exploding data. It mimics living organisms that store genetic codes in DNA with unparalleled density and stability. DNA data storage requires the means for high-throughput writing (synthesis). We will review efforts for arrayed DNA synthesis using a highly scalable semiconductor chip. Graphical abstract

Over the past decade, significant progress in quantum technologies has been made and, hence, engineering of these systems has become an important research area. Many researchers have become interested in studying ways in which classical integrated circuits can be used to complement quantum mechanical systems, enabling more compact, performant, and/or extensible systems than would be otherwise feasible. In this article—written by a consortium of early contributors to the field—we provide a review of some of the early integrated circuits for the quantum information sciences. CMOS and BiCMOS integrated circuits for nuclear magnetic resonance, nitrogen-vacancy-based magnetometry, trapped-ion-based quantum computing, superconductor-based quantum computing, and quantum-dot based quantum computing are described. In each case, the basic technological requirements are presented before describing proof-of-concept integrated circuits. We conclude by summarizing some of the many open research areas in the quantum information sciences for CMOS designers.