Showing papers by "Jong-Ho Lee published in 2023"

Highly-scaled and fully-integrated 3-dimensional ferroelectric transistor array for hardware implementation of neural networks

Abstract: Hardware-based neural networks (NNs) can provide a significant breakthrough in artificial intelligence applications due to their ability to extract features from unstructured data and learn from them. However, realizing complex NN models remains challenging because different tasks, such as feature extraction and classification, should be performed at different memory elements and arrays. This further increases the required number of memory arrays and chip size. Here, we propose a three-dimensional ferroelectric NAND (3D FeNAND) array for the area-efficient hardware implementation of NNs. Vector-matrix multiplication is successfully demonstrated using the integrated 3D FeNAND arrays, and excellent pattern classification is achieved. By allocating each array of vertical layers in 3D FeNAND as the hidden layer of NN, each layer can be used to perform different tasks, and the classification of color-mixed patterns is achieved. This work provides a practical strategy to realize high-performance and highly efficient NN systems by stacking computation components vertically.

In‐Memory‐Computed Low‐Frequency Noise Spectroscopy for Selective Gas Detection Using a Reducible Metal Oxide

Abstract: Concerns about indoor and outdoor air quality, industrial gas leaks, and medical diagnostics are driving the demand for high‐performance gas sensors. Owing to their structural variety and large surface area, reducible metal oxides hold great promise for constructing a gas‐sensing system. While many earlier reports have successfully obtained a sufficient response to various types of target gases, the selective detection of target gases remains challenging. In this work, a novel method, low‐frequency noise (LFN) spectroscopy is presented, to achieve selective detection using a single FET‐type gas sensor. The LFN of the sensor is accurately modeled by considering the charge fluctuation in both the sensing material and the FET channel. Exposure to different target gases produces distinct corner frequencies of the power spectral density that can be used to achieve selective detection. In addition, a 3D vertical‐NAND flash array is used with the fast Fourier transform method via in‐memory‐computing, significantly improving the area and power efficiency rate. The proposed system provides a novel and efficient method capable of selectively detecting a target gas using in‐memory‐computed LFN spectroscopy and thus paving the way for the further development in gas sensing systems.

Unveiling Resistance Switching Mechanisms in Undoped HfOx Ferroelectric Tunnel Junction Using Low-Frequency Noise Spectroscopy

Abstract: We demonstrate that the resistance switching (RS) of an undoped hafnium oxide (HfOx)-based ferroelectric tunnel junction (FTJ) is affected not only by ferroelectric domain switching of HfOx but also by the redistribution of oxygen vacancies inside HfOx, known as the working principle of resistive random-access memory. It is revealed that the RS mechanism varies depending on the program bias applied to FTJ. Through low-frequency noise spectroscopy, a precise method for distinguishing two distinct RS processes intrinsic to FTJ is presented.

Recommended Implementation of Quantitative Susceptibility Mapping for Clinical Research in The Brain: A Consensus of the ISMRM Electro-Magnetic Tissue Properties Study Group

Abstract: This article provides recommendations for implementing quantitative susceptibility mapping (QSM) for clinical brain research. It is a consensus of the ISMRM Electro-Magnetic Tissue Properties Study Group. While QSM technical development continues to advance rapidly, the current QSM methods have been demonstrated to be repeatable and reproducible for generating quantitative tissue magnetic susceptibility maps in the brain. However, the many QSM approaches available give rise to the need in the neuroimaging community for guidelines on implementation. This article describes relevant considerations and provides specific implementation recommendations for all steps in QSM data acquisition, processing, analysis, and presentation in scientific publications. We recommend that data be acquired using a monopolar 3D multi-echo GRE sequence, that phase images be saved and exported in DICOM format and unwrapped using an exact unwrapping approach. Multi-echo images should be combined before background removal, and a brain mask created using a brain extraction tool with the incorporation of phase-quality-based masking. Background fields should be removed within the brain mask using a technique based on SHARP or PDF, and the optimization approach to dipole inversion should be employed with a sparsity-based regularization. Susceptibility values should be measured relative to a specified reference, including the common reference region of whole brain as a region of interest in the analysis, and QSM results should be reported with - as a minimum - the acquisition and processing specifications listed in the last section of the article. These recommendations should facilitate clinical QSM research and lead to increased harmonization in data acquisition, analysis, and reporting.

Macroscopic analysis and design of Si HFGFET gas sensor for sensitive gas detection

Abstract: In this study, we investigate gas sensing performance of horizontal floating-gate field-effect transistor (HFGFET) gas sensors having different FET channel width and length, and the number and length of FG fingers using both technology computer-aided design (TCAD) device simulation and actual gas sensing measurement. The HFGFET gas sensors have a control-gate (CG) and a FG, which are horizontally interdigitated. An indium oxide (In2O3) film is locally deposited between the CG and FG to be used as a sensing layer. Utilizing a simplified equivalent circuit and electrical characteristics of the HFGFET gas sensors, we define a new parameter G, which represents a magnitude of gas sensitivity. The effect of channel width and length, and the number and length of FG fingers of the HFGFET gas sensors on gas sensitivity is first investigated by TCAD device simulation, and then verified by NO2 and H2S gas sensing measurement.

Self‐Curable Synaptic Ferroelectric FET Arrays for Neuromorphic Convolutional Neural Network

Abstract: With the recently increasing prevalence of deep learning, both academia and industry exhibit substantial interest in neuromorphic computing, which mimics the functional and structural features of the human brain. To realize neuromorphic computing, an energy‐efficient and reliable artificial synapse must be developed. In this study, the synaptic ferroelectric field‐effect‐transistor (FeFET) array is fabricated as a component of a neuromorphic convolutional neural network. Beyond the single transistor level, the long‐term potentiation and depression of synaptic weights are achieved at the array level, and a successful program‐inhibiting operation is demonstrated in the synaptic array, achieving a learning accuracy of 79.84% on the Canadian Institute for Advanced Research (CIFAR)‐10 dataset. Furthermore, an efficient self‐curing method is proposed to improve the endurance of the FeFET array by tenfold, utilizing the punch‐through current inherent to the device. Low‐frequency noise spectroscopy is employed to quantitatively evaluate the curing efficiency of the proposed self‐curing method. The results of this study provide a method to fabricate and operate reliable synaptic FeFET arrays, thereby paving the way for further development of ferroelectric‐based neuromorphic computing.

1/f Noise in Synaptic Ferroelectric Tunnel Junction: Impact on Convolutional Neural Network

Abstract: In recent years, neuromorphic computing has been rapidly developed to overcome the limitations of von Neumann architecture. In this regard, the demand for high‐performance synaptic devices with high switching speeds, low power consumption, and multilevel conductance is increasing. Among the various synaptic devices, ferroelectric tunnel junctions (FTJs) are promising candidates. While previous studies have focused on improving reliability of FTJs to enhance the synaptic behavior, low‐frequency noise (LFN) of FTJs has not been characterized and its impact on the learning accuracy in neuromorphic computing remains unknown. Herein, the LFN characteristics of FTJs fabricated on n‐ and p‐type Si along with the impact of 1/f noise on the learning accuracy of convolutional neural networks (CNNs) are investigated. The results indicate that the FTJ on p‐type Si exhibits a far lower 1/f noise than that on n‐type Si. The FTJ on p‐type Si exhibits a significantly higher learning accuracy (86.26%) than that on n‐type Si (78.70%) owing to its low‐noise properties. This study provides valuable insights into the LFN characteristics of FTJs and a solution to improve the performance of synaptic devices by significantly reducing the 1/f noise.

Neuromorphic Computing Using Random Synaptic Feedback Weights for Error Backpropagation in NAND Flash Memory-Based Synaptic Devices

Abstract: This work proposes utilizing separate synaptic string array for error backpropagation in NAND flash memory-based synaptic architecture with random synaptic feedback weight. To enable error backpropagation, forward and backward propagations are processed in separate synaptic devices in forward and backward synaptic arrays, respectively. In addition, synaptic weights in forward synaptic array are updated at each iteration, while those in backward synaptic array are fixed to reduce burden of peripheral circuits and power consumption. The optimal conductance response is investigated considering the linearity of the conductance response and the ratio of maximum and minimum currents. Reliability characteristics are verified by retention, endurance, and pass bias disturbance measurement results. Hardware-based neural networks with random synaptic weight achieve an inference accuracy of 95.41%, which is comparable to that of 95.58% obtained with transposed weight. Hardware-based neural network simulations demonstrate that the inference accuracy of the proposed on-chip learning scheme hardly decreases compared to that of the off-chip learning even with increasing device variation.

Hardware-Based Ternary Neural Network Using AND-Type Poly-Si TFT Array and Its Optimization Guideline

Abstract: Thin-film transistor (TFT)-type synaptic devices with poly-Si channels have the benefits of compatibility with the CMOS process, high reliability, and low power consumption. However, it is challenging to determine the optimal operating conditions of TFT arrays in hardware-based neural networks (HNNs) due to the limited device characteristics. In this work, we design hardware-based ternary neural networks (TNNs) using TFT-type synaptic devices. The electrical characteristics of the TFT array are investigated, and the effects of leakage currents are analyzed on the inference accuracy of TNNs. Based on the analysis, systematic optimization of the operating conditions is conducted to mitigate the impact of the device variation on the current sum and to maximize the accuracy. This result offers important guidelines for designing and optimizing hardware-based TNNs with not only TFT-type synaptic devices but also transistor-type synaptic devices.

Fuse Devices for Pruning in Memristive Neural Network

Abstract: In this study, we developed a fuse device for pruning implementation in a hardware neural network. A line-shaped fuse device was fabricated with aluminum metal and characterized in a one fuse and one resistive random-access memory (1F1R) structure. A blow time of 0.4 $\mu \text{s}$ and read endurance of $> 10^{{7}}$ were achieved, and the cut-off operation of the fuse was successfully verified in 1F1R. In addition, we developed a fuse design method for blow voltage and current via adjustment of the length, width, and thickness of the fuse. The adjustments were adopted to utilize the fuse in various synaptic devices. Finally, using simulations, we demonstrated a performance improvement due to the network pruning wherein the defective devices are disconnected by the fuse operations.

Structural characteristics of thermally advanced polymer with intrinsic microporosity for application as flexible OLED substrate

Abstract: The chemical and structural characteristics of solution-processed polymer with intrinsic microporosity (PIM) based on 9, 9-bis (4-hydroxyphenyl) fluorene (BHPF) film is investigated. A fully flexible and transparent PIM film exhibiting uniform and high transmittance of more than 90% within the visible light range is demonstrated. The stability of the PIM film is further demonstrated with relatively stable coefficient of thermal expansion (CTE) characteristics. Additionally, the film has high heat resistance that can withstand high-temperature processes at 250 °C. The PIM film exhibits a remarkably high glass transition temperature (Tg) of 294.7 °C, indicating the thermal stability of the film at elevated temperatures. The PIM film also exhibits relatively high surface energy, low surface roughness (Rq), and peak-to-valley values of 0.45 and 4.4 nm, respectively. This surface morphology confirms the PIM film’s superior characteristic in preventing short circuits or leakage in current paths in an organic electronic device. Finally, the PIM film is successfully tested as a substrate for a bottom-emitting organic light-emitting diode (OLED). The investigation has proven PIM film to be a good candidate to be adopted as a substrate in fabricating advanced organic electronic devices and next-generation displays.

Temperature Measurement Method Using FET Type Gas Sensor

Abstract: In this work, we propose the method about temperature measurement using transient behaviors of FET-type gas sensors based on MOSFET. The proposed sensor was manufactured using the CMOS process, and has a floating gate, p-type buried channel, control-gate, and WO3. Through appropriate pulse time and bias control, it was possible to confirm the response more than 5 times according to the temperature change from $25^{\circ}\mathrm{C}$ to $45^{\circ}\mathrm{C}$.

Floating Fin Shaped Stacked Nanosheet MOSFET for Low Power Logic Application

Abstract: In this paper, floating fin structured vertically stacked nanosheet gate-all-around (GAA) metal oxide semiconductor field-effect transistor (FNS) is proposed for low power logic device applications. To verify the electrical performance of the proposed device, three-dimensional (3-D) technology computer-aided design (TCAD) device/circuit simulations are performed with calibrated device model parameters. As a result, it is found that gate propagation delay $({\tau }_{\mathrm{ delay}})$ and dynamic power $(P_{\mathrm{ dyn}})$ are improved by 8% and 19%. respectively as compared to conventional vertically stacked lateral nanosheet (LNS). Through the rigorous analysis on the resistance and capacitance components of FNS and LNS, it is clearly revealed that the ${\tau }_{\mathrm{ delay}}$ and $P_{\mathrm{ dyn}}$ are improved at the same $P_{\mathrm{ dyn}}$ (50 $\mu \text{W}$ ) and ${\tau }_{\mathrm{ delay}}$ (187 GHz) by the reduced effective capacitance which results from the diminished gate-to-sorece/drain overlap area. Based on the TCAD simulation studies, it is expected that the FNS is suitable for next generation logic digital applications.

Recent Advances and Future Prospects for Memristive Materials, Devices, and Systems.

Abstract: Memristive technology has been rapidly emerging as a potential alternative to traditional CMOS technology, which is facing fundamental limitations in its development. Since oxide-based resistive switches were demonstrated as memristors in 2008, memristive devices have garnered significant attention due to their biomimetic memory properties, which promise to significantly improve power consumption in computing applications. Here, we provide a comprehensive overview of recent advances in memristive technology, including memristive devices, theory, algorithms, architectures, and systems. In addition, we discuss research directions for various applications of memristive technology including hardware accelerators for artificial intelligence, in-sensor computing, and probabilistic computing. Finally, we provide a forward-looking perspective on the future of memristive technology, outlining the challenges and opportunities for further research and innovation in this field. By providing an up-to-date overview of the state-of-the-art in memristive technology, this review aims to inform and inspire further research in this field.

Systematic Investigation of Subtractive Contact Patterning for High-Resolution Organic Electronic Devices

Abstract: We systematically investigated the mechanism of subtractive contact patterning (SCP) and demonstrated high-resolution organic light-emitting diodes (OLEDs) using this SCP process. Owing to the application of the SCP process to various types of organic semiconducting thin films, crystalline thin films (e.g., pentacene) generated patterns via the delamination mechanism and amorphous small molecule thin films (e.g., tris(8-hydroxyquinolinato)aluminum) generated patterns via the diffusion mechanism. In the pattern generation via diffusion, we reported that the higher the processing temperature and the longer the processing time were, the deeper was the pattern depth. In particular, the patterning speed was proportional to the temperature in an exponential function. SCP can be applied to doped small molecule-based thin films and limitedly applied to polymer thin films. Finally, a high-resolution OLED pattern of less than 10 μm wide was fabricated by applying the SCP process without any detrimental effects on the device performance. Furthermore, this SCP process was applied to flexible and curved thin films, which verified its potential use for roll-to-roll processes. This study provides a scientific and technological basis for the fabrication of high-resolution patterns of organic electronic devices such as OLEDs.

Transducer-Aware Hydroxy-Rich-Surface Indium Oxide Gas Sensor for Low-Power and High-Sensitivity NO2 Gas Sensing.

Abstract: Low-power metal oxide (MOX)-based gas sensors are widely applied in edge devices. To reduce power consumption, nanostructured MOX-based sensors that detect gas at low temperatures have been reported. However, the fabrication process of these sensors is difficult for mass production, and these sensors are lack uniformity and reliability. On the other hand, MOX film-based gas sensors have been commercialized but operate at high temperatures and exhibit low sensitivity. Herein, commercially advantageous highly sensitive, film-based indium oxide sensors operating at low temperatures are reported. Ar and O2 gases are simultaneously injected during the sputtering process to form a hydroxy-rich-surface In2O3 film. Conventional indium oxide (In2O3) films (A0) and hydroxy-rich indium oxide films (A1) are compared using several analytical techniques. A1 exhibits a work function of 4.92 eV, larger than that of A0 (4.42 eV). A1 exhibits a Debye length 3.7 times longer than that of A0. A1 is advantageous for gas sensing when using field effect transistors (FETs) and resistors as transducers. Because of the hydroxy groups present on the surface of A1, A1 can react with NO2 gas at a lower temperature (∼100 °C) than A0 (180 °C). Operando diffuse reflectance infrared Fourier transform spectrometry (DRIFTS) shows that NO2 gas is adsorbed to A1 as nitrite (NO2-) at 100 °C and nitrite and nitrate (NO3-) at 200 °C. After NO2 is adsorbed as nitrate, the sensitivity of the A1 sensor decreases and its low-temperature operability is compromised. On the other hand, when NO2 is adsorbed only as nitrite, the performance of the sensor is maintained. The reliable hydroxy-rich FET-type gas sensor shows the best performance compared to that of the existing film-based NO2 gas sensors, with a 2460% response to 500 ppb NO2 gas at a power consumption of 1.03 mW.

Low-frequency noise in gas sensors: A review

Abstract: Semiconductor-based gas sensors have been applied to a variety of applications, including environmental, safety, and health monitoring. Extensive efforts have been made to improve sensing performance outcomes, with the majority of these efforts focusing on the sensor response, sensitivity, and selectivity issues, mainly by optimizing the sensing materials and sensor structures. However, low-frequency noise (LFN), which has a considerable impact on the stability and reliability of sensors, has received far less attention in gas sensor research. In gas sensing applications, the noise in the sensing signal is determined by the LFN due to the slow reaction process. Thus, it is necessary to characterize the LFN in semiconductor-based gas sensors. This review article presents an overview of the LFN in semiconductor-based gas sensors. First, the history of LFN in gas sensor studies is explored briefly. Then, we discuss noise generation mechanisms in resistor-type, thin-film transistor-type, and horizontal floating-gate field-effect-transistor-type gas sensors. On the basis of this information, the signal-to-noise ratio, which determines the limit of detection, is examined, and the method to optimize the SNR in each sensor platform is discussed. Finally, LFN spectroscopy for selective gas detection is introduced, and its working principle is analyzed. This review article provides a foundation for understanding the LFN in semiconductor-based gas sensors and methods to control it based on application requirements.

Optimizing Post-Metal Annealing Temperature Considering Different Resistive Switching Mechanisms in Ferroelectric Tunnel Junction

Abstract: We investigate the effect of post-metal annealing temperature ( ${T}_{\text {PMA}}$ ) on ferroelectric (FE) resistive switching (RS) and non-FE RS in HfOx ferroelectric tunnel junctions. Through conductance analysis and low-frequency noise spectroscopy, the effects of ${T}_{\text {PMA}}$ on RS mechanisms are demonstrated. It is revealed that the non-FE RS, redistribution of oxygen vacancies, is suppressed with an increase in ${T}_{\text {PMA}}$ . The effects of different RS mechanisms on the tunneling electroresistance and cycling endurance characteristics are systematically investigated.

Sinus augmentation with poly(ε)caprolactone-β tricalcium phosphate scaffolds, mesenchymal stem cells and platelet rich plasma for one-stage dental implantation in minipigs

Abstract: Purpose This study evaluated the efficacy of a tube-shaped poly(ε) caprolactone - β tricalcium phosphate (PCL-TCP) scaffold with the incorporation of human umbilical cord-derived mesenchymal stem cells (hUCMSCs) and platelet-rich plasma (PRP) for bone regeneration in the procedure of single-stage sinus augmentation and dental implantation in minipigs. Methods Implants were placed in the bilateral sides of the maxillary sinuses of 5 minipigs and allocated to a PCL-TCP+hUCMSCs+PRP group (n=5), a PCL-TCP+PRP group (n=5), and a PCL-TCP-only group (n=6). After 12 weeks, bone regeneration was evaluated with soft X-rays, micro-computed tomography, fluorescence microscopy, and histomorphometric analysis. Results Four implants failed (2 each in the PCL-TCP+hUCMSCs+PRP and PCL-TCP+hUCMSC groups). An analysis of the grayscale levels and bone-implant contact ratio showed significantly higher mean values in the PCL-TCP+hUCMSCs+PRP than in the PCL-TCP group (P=0.045 and P=0.016, respectively). In fluoromicroscopic images, new bone formation around the outer surfaces of the scaffolds was observed in the PCL-TCP+hUCMSCs+PRP group, suggesting a tenting effect of the specially designed scaffolds. Bone regeneration at the scaffold-implant interfaces was observed in all 3 groups. Conclusions Using a tube-shaped, honeycombed PCL-TCP scaffold with hUCMSCs and PRP may serve to enhance bone formation and dental implants’ osseointegration in the procedure of simultaneous sinus lifting and dental implantation.

Abstract 2404: Patient-derived organoid platform for the prediction of radiation response and modeling radiation resistance in oral cancer

Abstract: Oral cancer is rare cancer that accounts for roughly 1.5% of all cancers in Korea. The standard treatment for the patient with oral cancer is surgery followed by radiotherapy. However, 30-50% of patients had local recurrence and metastasis to lymph nodes within 2 years. Numerous radiation-resistant cell lines have been created and utilized for research due to the significance of biomarkers for predicting response to radiation therapy; however, the homogenous 2D cell lines limit the practical use of the identified molecules. Patient-derived organoid (PDO) systems, as opposed to 2D cell lines, have significant advantages for the preclinical model in that they are similar to the genetic heterogeneity seen in patient tumors and reflect the clinical characteristics of patients. In this study, we generated organoid models that mimic radiation therapy from the oral PDO library in order to identify molecular profiles associated with radiation resistance. From January 2021 to August 2022, we prospectively collected 164 normal tissues and 179 tumor tissues from enrolled patients with oral cancer. 60 tumor organoids and 66 normal organoids were maintained over 4 passages and cryopreserved, which is the largest PDO repository with normal and malignant ever published for oral cancer to date. Each organoid was identified through long-term clinical information follow-up in their respective patients, notably recurrence after radiation therapy. In the oral PDO library, organoids were categorized as derived from tumor tissues of non-recurred patients (nPDOs), derived from primary tumor tissues of patients who recurred after radiotherapy (pPDOs), and derived from recurrent tumor tissues (rPDOs). There was also one pair of pPDO and rPDO for the same patient. A total of 60 Gy of radiation was irradiated to the organoids for the construction of a radiation-resistant PDO model, and finally, four cases of radiotherapy mimic organoid models were successfully established. Survival analysis for radiation dose-response and post-irradiation calcein-AM staining were used to validate these models. The pPDO and rPDO models of radiation resistance are well-established, whereas nPDO was not. Given the rarity of oral cancer, this platform is the first preclinical model to closely resemble the clinical radiation pipeline for patients with oral cancer. It would help make a more accurate prediction of radiation response in patients with oral cancer, as well as the development of treatment guidelines. (This research was supported by National Cancer Center, Korea (No. 2210980) and National Research Foundation of Korea (NRF) grant, funded by the Korean government (MSIT) (No. 2020M3A9A5036362)) Citation Format: Sumin Kang, Mi Rim Lee, Sun-Young Kong, Jong-Ho Lee, Dohyun Kwon, Ikjae Kwon, Soung-Min Kim, Youngwook Kim, Wonyoung Choi, Hye Won Shon, Yu-Sun Lee, Joo Yong Park, Sung Weon Choi, Yun-Hee Kim. Patient-derived organoid platform for the prediction of radiation response and modeling radiation resistance in oral cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2404.

Universal Stretchable Conductive Cellulose/PEDOT:PSS Hybrid Films for Low Hysteresis Multifunctional Stretchable Electronics.

Abstract: Skin-attachable conductive materials have attracted significant attention for use in wearable devices and physiological monitoring applications. Soft, skin-like conductive films must have excellent mechanical and electrical characteristics with on-skin conformability, stretchability, and robustness to detect body motion and biological signals. In this study, a conductive, stretchable, hydro-biodegradable, and highly robust cellulose/poly(3,4-ethylene dioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) hybrid film is fabricated. Through the synergetic interplay of a conductivity enhancer, nonionic fluorosurfactant, and surface modifier, the mechanical and electrical properties of the stretchable hybrid film are greatly improved. The stretchable cellulose/PEDOT:PSS hybrid film achieves a limited resistance change of only 1.21-fold after 100 stretch-release cycles (30% strain) with exceptionally low hysteresis, demonstrating its great potential as a stretchable electrode for stretchable electronics. In addition, the film shows excellent biodegradability, promising environmental friendliness, and safety benefits. High-performance stretchable cellulose/PEDOT:PSS hybrid films, which have high biocompatibility and sensitivity, are applied to human skin to serve as on-skin multifunctional sensors. The conformally mounted on-skin sensors are capable of continuously monitoring human physiological signals, such as body motions, drinking, respiration rates, vocalization, humidity, and temperature, with high sensitivity, fast responses, and low power consumption (21 μW). The highly conductive hybrid films developed in this study can be integrated as both stretchable electrodes and multifunctional healthcare monitoring sensors. We believe that the highly robust stretchable, conductive, biodegradable, skin-attachable cellulose/PEDOT:PSS hybrid films are worthy candidates as promising soft conductive materials for stretchable electronics.

Power and Area-Efficient XNOR-AND Hybrid Binary Neural Networks Using TFT-Type Synaptic Device

Abstract: A novel XNOR- AND hybrid binary neural network (BNN) using a TFT-type synaptic device is proposed to reduce the area and power consumption of the synaptic array. Replacing some parts of the network from the XNOR operation to the AND operation results in expressing weight and input with a single cell and single word line. In the case of replacing the operation of the fully-connected (FC) layer with the AND operation, the accuracy of VGG9 BNN for CIFAR-10 datasets drops by about 1%, while the number of cells in the synaptic array decreases by 33.7%. Using the previously proposed TFT-type synaptic devices, the proposed method reduces power consumption by ~25%.

Super-steep synapses based on positive feedback devices for reliable binary neural networks

Abstract: This work proposes positive feedback (PF) device-based synaptic devices for reliable binary neural networks (BNNs). Due to PF operation, the fabricated PF device shows a high on/off current ratio (2.69 [Formula: see text] 107). The PF device has a charge-trap layer by which the turn-on voltage ( Von) of the device can be adjusted by program/erase operations and a long-term memory function is implemented. Also, due to the steep switching characteristics of the PF device, the conductance becomes tolerant to the retention time and the variation in turn-on voltage. Simulations show that high accuracy (88.44% for CIFAR-10 image classification) can be achieved in hardware-based BNNs using PF devices with these properties as synapses.

Design Strategy to Improve Memory Window in Ferroelectric Transistors With Oxide Semiconductor Channel

Abstract: Oxide semiconductors are promising channel materials for hafnia-based ferroelectric transistor memories because they can constrain the formation of an unwanted interfacial layer that can deteriorate the stability of the device. A major obstacle is the limited memory window, originating from insufficient polarization switching because ${n}$ -type oxide semiconductors cannot provide sufficient hole carriers to realize ferroelectric polarization switching. To solve this issue, a novel design strategy is proposed to achieve increased polarization switching while maintaining the stability of oxide semiconductor-based ferroelectric thin-film transistors (FeTFTs). By inserting an additional ${p}$ -type CuOx layer between the ${n}$ -type oxide semiconductor InZnOx and ferroelectric HfZrOx, increased polarization switching is achieved owing to the high electron and hole densities in the InZnOx and CuOx layers, respectively. Thus, a memory window of 4 V is achieved, which cannot be obtained using a single oxide-semiconductor channel. We also demonstrate that the proposed method is viable for three-dimensional ferroelectric NAND (3D FeNAND) devices. In 3D FeNAND, replacing the dielectric filler with ${p}$ -type CuOx maximizes polarization switching and enlarges the memory window. The results demonstrate a novel structure and fabrication method for high-performance FeTFTs for advanced 3D non-volatile memory applications.

Inorganic Perovskite Quantum Dot-Mediated Photonic Multimodal Synapse.

Abstract: Artificial synapse is the basic unit of a neuromorphic computing system. However, there is a need to explore suitable synaptic devices for the emulation of synaptic dynamics. This study demonstrates a photonic multimodal synaptic device by implementing a perovskite quantum dot charge-trapping layer in the organic poly(3-hexylthiophene-2,5-diyl) (P3HT) channel transistor. The proposed device presents favorable band alignment that facilitates spatial separation of photogenerated charge carriers. The band alignment serves as the basis of optically induced charge trapping, which enables nonvolatile memory characteristics in the device. Furthermore, high photoresponse and excellent synaptic characteristics, such as short-term plasticity, long-term plasticity, excitatory postsynaptic current, and paired-pulse facilitation, are obtained through gate voltage regulation. Photosynaptic characteristics obtained from the device showed a multiwavelength response and a large dynamic range (∼103) that is suitable for realizing a highly accurate artificial neural network. Moreover, the device showed nearly linear synaptic weight update characteristics with incremental depression electric gate pulse. The simulation based on the experimental data showed excellent pattern recognition accuracy (∼85%) after 120 epochs. The results of this study demonstrate the feasibility of the device as an optical synapse in the next-generation neuromorphic system.

Abstract CT130: An open label, dose escalation, phase 1 study of AT101, a novel CD19-directed CAR-T cell therapy targeting a membrane-proximal epitope of CD19, in patients with relapsed or refractory B cell non-Hodgkin lymphoma

Abstract: Background and Preliminary Data: All the FDA-approved CD19 CAR-T cell therapies are based on an antigen-binding domain (scFv) based on the FMC63 antibody which binds to the membrane-distal region of CD19 to an epitope encoded by exons 3 and 4 (Klesmith JR, Biochemistry, 2019; Zhang Z, JITC, 2020). While these CART19 products are very effective in the clinic, the majority of patients still do not respond or eventually relapse due to several mechanisms of resistance, including T cell dysfunction and epitope CD19-negative escape. Novel strategies to enhance the activity of CART cells and reduce escape are critically needed. We recently demonstrated that modifications of the binding region of the CAR (scFv) (Singh N., Nat Med, 2021) can drastically change the interaction between the CAR T cell and the cancer cells, potentially improving the anti-tumor effect. To this goal, we developed a novel anti-CD19 antibody clone (1218) that binds to a membrane-proximal epitope of CD19 (exon 2 region K59-K63) thereby not competing with FMC63. We developed a novel CART19, called AT101, using a humanized 1218 scFv along with 4-1BB costimulatory and CD3zeta domain in a lentiviral backbone. In preclinical models, AT101 showed more potent in vitro cytotoxicity against CD19-positive B lymphoma cells in a long-term killing assay and in a B-ALL (NALM6) in vivo model as compared to the control of FMC63 based CAR-T cells. In addition, differently than FMC63-based CART, AT101 could target tumor cells expressing point mutations of CD19 that are associated with relapse post-CART19 (FMC63) (Zhang Z, JITC, 2020) and leukemic blasts aberrantly expressing FMC63 CAR19 on their surface (Ruella M, Nat Med, 2018). Based on the preclinical efficacy and safety, a phase 1 clinical trial testing autologous AT101 was started for patients with relapsed and refractory B-cell non-Hodgkin lymphoma. Trial Design and Methods: This open-label, multi-center, first-in-human Phase 1 study will assess the safety and feasibility of AT101 in patients with relapsed or refractory B cell non-Hodgkin lymphoma. Key eligibility criteria include patients aged ≥19 years of age with histologically confirmed relapsed or refractory aggressive B-cell non-Hodgkin lymphoma. In this phase 1 trial, patients (n=3 per dose level; up to n=18 in total) are treated with AT101 in 3 dose-escalation cohorts based on a standard 3 + 3 design. CART doses are 2.0 × 105, 1.0 × 106, or 5.0 × 106 CAR+T cells/kg. The primary objective is to determine the safety, the maximum tolerated dose (MTD), and the recommended phase 2 dose (RP2D) of AT101 in participants following lymphodepletion with cyclophosphamide and fludarabine (250 mg/m2 and 25 mg/m2). The secondary objective is to evaluate the preliminary efficacy assessments (overall response rate (ORR), duration of response (DOR), progression-free survival (PFS), overall survival (OS), event-free survival (EFS), and pharmacokinetics of AT101. Exploratory objectives include assessment of CD19 expression and cytokines in the blood. Patients will be followed for safety for at least 60 months post AT101 infusion. Clinical trial registry number: NCT05338931. As of January 11, 2023, AT101 has been infused to six patients in cohort 1 and three patients in cohort 2. Detailed results will be presented at the meeting. Citation Format: Yunlin Zhang, Ki Hyun Kim, Dok Hyun Yoon, Ruchi P. Patel, Jae-Cheol Jo, Hyungwoo Cho, Jong-Ho Lee, Hyun-Jong Lee, Lei-Guang Cui, In-Sik Hwang, Young Ha Lee, Jong-Hoon Kim, Yong Gu Lee, Puneeth Guruprasad, Jong-Seo Lee, Junho Chung, Marco Ruella. An open label, dose escalation, phase 1 study of AT101, a novel CD19-directed CAR-T cell therapy targeting a membrane-proximal epitope of CD19, in patients with relapsed or refractory B cell non-Hodgkin lymphoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr CT130.