Search

This paper reports an industry standard monolithic 555-timer circuit designed and fabricated in the in-house silicon carbide (SiC) low-voltage bipolar technology. This paper demonstrates the 555-timer integrated circuits (ICs) characterization in both astable and monostable modes of operation, with a supply voltage of 15 V over the wide temperature range of 25 °C–500 °C. Nonmonotonic temperature dependence was observed for the 555-timer IC frequency, rise time, fall-time, and power dissipation. [ABSTRACT FROM AUTHOR]

This paper provides an industry perspective on the present state of compact models for MOS transistors. It highlights the complexity of layout-dependent effects in modern transistors, reviews some common misunderstandings of MOS transistor capacitances, and introduces a new, structurally symmetric, “gate-as-input” equivalent network for these capacitances. Areas of focus for the future model developments are also proposed. [ABSTRACT FROM AUTHOR]

A novel vertical architecture for all-printed organic electrochemical transistors, based on poly(3, 4-ethylenedioxythiophene):poly(styrene sulfonate), realized on flexible substrates, is reported. The transistors are manufactured along both faces of plastic or paper substrates and via connections are realized using laser ablation or simple punch through using a pin. Successful modulation of the electric current that flows between the two sides of the substrate is achieved using electrolyte-gating and electrochemical modulation of the electronic charge transport of the bulk of the transistor channel. In addition to this, the transistors are exhibiting fast switching and high ON/OFF current ratios. [ABSTRACT FROM AUTHOR]

Transistor random threshold voltage variations due to process fluctuations seriously affects the stability of Static Random Access Memory (SRAM). In this paper, a SRAM bit transistors threshold voltage $({Vth})$ deviation monitoring scheme and system is proposed. This scheme ingeniously achieves on-chip measurement of all transistors threshold voltages without altering compact SRAM bit array layout. Control signal strategies and Transistor ${Vth}$ Determination Circuit (TVDC) for different types of Devices Under Test (DUTs) have been proposed. The system is implemented using a 65 nm CMOS process with a core area of 0.01875mm2. Through Monte Carlo analysis, the Weighted Average (WA) difference of the proposed scheme and the direct measurement method is not more than 10mV, and the Root Mean Square Error (RMSE) difference is not more than 3mV. This system can also effectively detect the cell position of the transistor threshold voltage mismatch simulated by modifying the substrate voltage. For SRAM arrays of different scales, the method proposed in this paper has area efficiency and flexible reconfigurability. [ABSTRACT FROM AUTHOR]

In this paper, for the first time, the record high-performance top-gated graphene technology platform is used for electrostatic discharge (ESD) physics exploration while investigating the implications of various design and technology options. Impact of diffusive versus ballistic carrier transport on the failure mechanism in top-gate as well as back-gate graphene FET (GFET) is investigated. A unique contact limited failure in graphene transistors is reported. Physical insights on current saturation in GFET and unique step-by-step failure in dielectric capped transistors are presented for the first time. Moreover, device degradation under ESD timescales and its implications on current saturation are revealed. Finally, the influence of various top-gate designs on the ESD performance is reported. New physical insights and matured GFET technology has eventually enabled record high ESD robustness. [ABSTRACT FROM AUTHOR]

Many methods have been proposed for the experimental determination of the collector resistance in bipolar junction transistors and HBTs. In this paper, the most widely used methods are reviewed and applied to SiGe HBTs with a large variety of device sizes fabricated in different technologies and generations, including high-speed and high-voltage transistors. First, the accuracy of those methods, which are based on an extraction from single-transistor characteristics, is evaluated from simulated data using a sophisticated compact model and, where applicable, also device simulation. This approach allows the origin of observed inaccuracies or failures of certain methods to be identified. Second, the test structure-based methods are reviewed, and third, all methods were applied to experimental data. This paper and its results provide insight into each method’s accuracy; its application limits with respect to a technology, a device size, and an operating range as well as its requirements in terms of equipment and extraction effort. [ABSTRACT FROM AUTHOR]

A transient and static hybrid-triggered active clamp is proposed in this paper. By skillfully incorporating different detection mechanisms, the proposed clamp achieves enhanced static electrical overstress protection capability over the transient one. Furthermore, the proposed clamp achieves improved electrostatic discharge reaction speed in both human body model and charged device model events over the static one. Moreover, the superior transient-noise immunity of the proposed clamp over traditional transient ones is essentially revealed in this paper. The proposed clamp is successfully verified in a 65-nm bulk CMOS process. In addition, the design flexibility of the proposed clamp for other processes is also deeply discussed in this paper. [ABSTRACT FROM PUBLISHER]

The memristor-based memory cell design is getting renewed attention in recent years due to its high speed and low power consumption. This paper aims at designing a novel hybrid memory cell incorporating a minimum number of transistors for a rapid bidirectional write operation. The read stability is one of the key elements for high efficiency and superior performance in memory. The memory cell in this paper was designed undertaking adequate measures for maintaining the stability in successive reads without any refresh operation. Extensive simulation experiments were conducted using 32-nm predictive technology model transistors and the results demonstrate superb performance and sound stability of the proposed memory cell in terms of read/write time as well as switching power consumptions. [ABSTRACT FROM PUBLISHER]

Double-gate devices, like independent-gate (IG) FinFET, have introduced new possibilities and challenges in synthesis of transistor networks. Existing factorization methods and graph-based optimizations are not actually the most effective way to generate optimized IG FinFET based networks because only reducing the number of literals in a given Boolean expression does not guarantee the minimum transistor count. This paper presents two novel methods aiming the minimization of the number of devices in logic networks. The first contribution is a method for defactoring Boolean expressions able to apply the conventional factorization algorithms together with IG FinFET particularities, so improving it. The second contribution is a novel graph-based method that improves even more transistor arrangements by exploiting enhanced nonseries-parallel associations. Experimental results shown a significant reduction in the size of transistor networks delivered by the proposed methods. [ABSTRACT FROM AUTHOR]

This paper clarifies the correct transmission mechanism, misattributed in the previous research, of a hydrogenated amorphous silicon thin-film transistor (a-Si:H TFT) device. Complete drain current–gate voltage (${I}_{\text {D}}$ – ${V}_{\text {G}}$) transfer characteristics including the forward and backward gate sweeps (gate voltage carried out in the OFF-state $\to $ on-state and the ON-state $\to $ OFF-state) are performed, and the physics models of an energy-band schematic are clearly explained. This paper reveals that the ${I}_{\text {D}}$ – ${V}_{\text {G}}$ curve stretch-out behavior of the subthreshold region is due to the leakage of the Poole–Frenkel region. At the Poole–Frenkel regions, electrons transfer to the drain terminal, and holes remain in the valance band of the a-Si:H bulk. The remaining holes represent the positive voltage and cause the electron barrier height lowering. This causes an increase in subthreshold current of the subthreshold region. Finally, three experiments are performed to prove the correctness of these models, and the a-Si TFTs device improvement methods will be proposed to enhance the stability and performance of product of the a-Si TFTs. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel reconfigurable ferroelectric FET (R-FEFET), which can reconfigure its operation between volatile and nonvolatile modes during run-time by dynamically modulating its hysteresis. The R-FEFET comprises of two gates with ferroelectric (FE) in both the gate stacks. One of these terminals serves as a regular gate, while the other is used as a control to introduce reconfigurability. Employing Landau–Khalatnikov equation-based FEFET model, we extensively analyze the device characteristics in both FinFET and planar technologies. We show that by changing the control voltage between 0 and 1 V, the hysteresis width (HW) can be modulated between 1.1 and 0.3 V. In addition, we show the device characteristics and advantages that R-FEFETs possess to overcome the drawbacks encountered with gate leakage in standard FEFETs. The hold margins in the nonvolatile mode of R-FEFETs increase by ~ $10\times $ with respect to standard FEFETs, and the drive current strengths in the volatile mode show 13% improvements when compared to standard FETs. Using the proposed R-FEFETs, we examine a low-power nonvolatile memory design in Part II of this paper. [ABSTRACT FROM AUTHOR]

In order to better predict the high-frequency switching operation of transistors in power converters, parasitic elements of these devices such as resistances, inductances, and capacitances must be accurately evaluated. This paper reports on the characterization of a gallium nitride (GaN) packaged power transistor using S-parameters in order to extract the device parasitics. Because the transistor is packaged, a calibration technique is carried out using specific test fixtures designed on FR4 printed circuit board (PCB) in order to get the S-parameters in the transistor plane from the measurement. The proposed method is suitable for a wide range of power devices. In this paper, it is applied to an enhancement-mode GaN high electron mobility transistor (HEMT). The impact of junction temperature on drain and source resistances is also evaluated. According to characterization results, equation-based modeling is proposed for the nonlinear parameters. The extracted parasitic elements are compared with reference values given by the device manufacturer. [ABSTRACT FROM AUTHOR]

This paper presents a set of wideband, fully integrated E-band power amplifiers (PAs) designed in 0.12- $\mu \text{m}$ silicon-germanium (SiGe) BiCMOS and working in the range of 60–75 GHz. The single PA is based on common-emitter (CE) driver stages followed by a CE output stage. A three-stage and four-stage versions of the single-PA as well as four- and eight-way combined variants of the four-stage PA are measured and compared. The single, four-way, and eight-way combined PAs achieve saturated output powers of 16, 19.5, and 24 dBm with peak power-added efficiencies (PAEs) of 18%, 11%, and 12%, respectively. Modulated waveforms are passed through each amplifier and data rates as high as 32 Gb/s are demonstrated for all amplifiers when driven in the linear mode. Measurements of how error-vector magnitude (EVM) and PAE degrade versus output power are presented and it is demonstrated that the eight-way combined PA can deliver 17–20 dBm of average power in a 64-quadratic-amplitude modulation, 1-Gbaud waveform (6 Gb/s), with an EVM of −32 to −24 dB and a PAE of 3%–5% without predistortion. To the author’s knowledge, this is the first paper reporting detailed measurements of PA EVM and PAE versus output power at these frequencies. [ABSTRACT FROM AUTHOR]

This paper presents two main contributions toward efficient very large scale integration circuits mapped with simple cells: 1) a complete synthesis flow to provide good-quality circuits mapped only with simple cells; and 2) an area-oriented, level-aware buffering algorithm based on inverter trees to fix cell fanout violations. An effective pattern-based algorithm to identify XORs/XNORs on and-inverter graphs is also presented. We show that efficient implementations in terms of inverter count, transistor count, area, power, and delay can be generated from circuits with a reduced number of both simple cells and inverters, combined with XOR/XNOR-based optimizations. The proposed buffering algorithm can handle all unfeasible fanout occurrences, while: 1) optimizing the number of added inverters and 2) assigning cells to the inverter tree based on their level criticality. When comparing with academic and commercial approaches, we are able to simultaneously reduce the average number of inverters, transistors, area, power dissipation, and delay up to 48%, 4%, 8%, 14%, and 16%, respectively. [ABSTRACT FROM AUTHOR]

This paper presents a new approach for compact modeling of aging and radiation effects for MOS transistors including process variation and environmental influences. This approach can be used in common design flows for analog integrated circuits. Moreover, the aging behavior of whole circuits can be analyzed without the need for abstractions or extrapolation. Only one common physically motivated transistor parameter is used in order to model radiation and degradation mechanisms. [ABSTRACT FROM AUTHOR]

For heterojunction bipolar transistors (HBTs), the noise correlation influences the noise parameters and is therefore relevant for transistor modeling, especially at sufficiently high frequencies. In order to predict the noise correlation, it is necessary to determine the noise transit time in the noise transport model. In this paper, the noise correlation and noise transit time are predicted from the high current model without measured noise data. The four noise parameters are calculated based on the predicted noise transit time and the Y-parameter noise calculation methodology. The extracted noise transit time is validated by the measured result over bias and temperature, which is determined by fitting NFmin of the silicon-germanium (SiGe) HBT between Y-parameter noise calculation methodology and tuner-based noise measurement. It was found that the noise transit time is independent of frequency but dependent on bias and temperature. Also, the results indicate that the calculated four noise parameters based on the extracted noise transit time have a good agreement with the tuner-based noise measurement. [ABSTRACT FROM AUTHOR]

This paper analyzes the performance of mosaic nonimaging passive infrared (PIR) sensors fabricated by the CMOS–SOI–MEMS technology. The elementary sensor, forming a subpixel, is a thermally isolated nanomachined CMOS transistor, dubbed TMOS, operating at subthreshold. The mosaic uncooled PIR sensors are composed of several TMOS subpixels, which are electrically connected, either in parallel or in series as well as a combination of both options. These mosaic sensors, which are manufactured by nanofabrication methods, exhibit enhanced performance and robust manufacturing on wafer level. The overall figures of merit of these sensors, which are modeled in this paper, indicate why they are most suitable for consumer electronics, including smart homes, wearables, Internet of Things as well as mobile applications. [ABSTRACT FROM AUTHOR]

This paper demonstrates a high-performance black phosphorus nanoribbons field-effect transistor (BPNR-FET) and systematically investigates methods for enhancing its anisotropic carrier transport. The BPNR-FET shows a strong dependence on crystal orientation in which the best mobility performance is achieved in armchair-oriented nanoribbons. A downscaling of nanoribbon width is shown to improve the short-channel effect owing to a better electrostatic gate control. Furthermore, hydrogenation is employed to effectively passivate the dangling bonds and heal the nanoribbon edge defects, leading to nearly hysteresis-free transfer properties. By virtue of bandgap and contact-metal workfunction engineering, n-type BPNR-FET is successfully demonstrated, which enables complementary inverter circuits to be simultaneously realized. This paper unravels the superior performance underscores a conceptually new BPNR-FET, paving the way toward the development of non-planar devices and integrated circuits based on 2-D materials platform. [ABSTRACT FROM AUTHOR]

This paper studies the estimation problem for a class of nonlinear tunnel diode circuits with parameter perturbation. The dynamics of the nonlinear circuit is approximated by the Takagi–Sugeno fuzzy model with linear fractional parametric uncertainties. Considering the fuzzy filter with gain uncertainties, a filtering error system can be gotten naturally. The concentration of this paper lies in deriving the design conditions for the desired resilient filter such that the filtering error system preserves asymptotic stability as well as the $\boldsymbol {\mathcal {L}}_{\boldsymbol {\infty }}$ -norm of the transfer function from the external disturbance input to the filtering error output is below a specified bound. The filter design conditions for guaranteeing the prescribed peak-to-peak performance of the filtering error system are provided by a set of linear matrix inequalities. Finally, a simulation result is presented to illustrate the validity and effectiveness of the proposed filtering design for the parameter-controlled tunnel diode circuit. [ABSTRACT FROM AUTHOR]

In this paper, circuit level analysis of the high frequency and low noise performance of an RF CMOS device with L\mathrm{ eff}= 36 nm is performed using various layout schemes. By using the modeling methodology of interconnect metals and vias, it is found that the gate parasitic capacitance from the interconnects mainly affects the degradation of high frequency and noise performance. An optimized layout scheme is proposed to reduce the gate parasitic resistance and capacitance in this paper, and the proposed layout exhibits improved RF behaviors for fT , f\mathrm {\mathrm {MAX}} , and NFmin at 26 GHz up to ~13%, ~24%, and ~18% compared with the reference layout scheme, respectively. [ABSTRACT FROM PUBLISHER]

In this paper, we present extensive characterization of low-frequency noise in advanced silicon germanium heterojunction bipolar transistors. We demonstrate the extraction methodology of base and collector noise spectral densities for a wide range of transistor geometries. In addition to 1/ $f$ noise, generation–recombination (G–R) mechanisms are observed at low bias in the base current noise. Their existence is confirmed by Random Telegraph Signal (RTS) noise measurements. 1/ $f$ and G–R components are extracted from the base current noise spectra and their bias dependencies are studied. Finally, base current noise spectral densities measured at the same base current density in different geometries are compared to study the individual contribution of 1/ $f$ noise from the periphery as well as the intrinsic device. Part II of this paper will discuss the modeling aspects and noise correlation. [ABSTRACT FROM AUTHOR]

The staircase modulation strategy is the best choice for asymmetric cascaded H-bridge multilevel inverter (ACMI) because it makes the output voltage to present a high number of levels and makes one of the ACMI modules to operate at low frequency. However, the staircase modulation is usually employed without giving its deserved attention. This paper presents a clear understanding of the staircase modulation strategy and its applications in both isolated and grid-connected ACMI.  The main novelty of this paper is the obtaining of the switching function for each module. The principle of operation is carefully described, also indicating how to command the ACMI transistors. The output voltage of the two superior modules is evaluated for different values of iterations. Special attention is given to the power distribution among the ACMI modules when the staircase modulation is applied. It is demonstrated on how the power is processed in each module considering an ACMI with three modules. The analysis is presented showing the staircase modulation within a closed-loop control strategy for isolated and grid-connected modes of operation, showing that the content of this paper is valid for both modes. Experimental results clarify and show the efficacy of the content presented in this paper. [ABSTRACT FROM AUTHOR]

As 2-D CMOS reaches its fundamental scaling limits due to device, manufacturing, and interconnect bottleneck related constraints at the nanoscale, migration to 3-D provides a possible alternative to continue technology scaling in future. Toward that goal, several 3-D integration approaches with multi-die, multichip, and sequential layers stacking are pursued. However, these approaches only show incremental density benefits and exhibit new challenges, such as lack of thermal management, increasing cost, and reliability issues. In contrast to these, we proposed a radically different fabric concept, called stacked horizontal nanowire-based 3-D CMOS (SN3D), on a single die that can offer a paradigm shift in technology scaling as well as design. Using prefabricated and doped stacked horizontal nanowires as fundamental building blocks, the fabric is assembled using architected connectivity and insulation features. Innovations in circuit style for mapping to SN3D’s physical framework, and bottom-up material filling-based manufacturing techniques are also central to our approach. In this paper, we detail, fabric’s core constructs, logic circuits implementation in SN3D fabric, benchmarking methodology and results, and finally the manufacturing aspects. Our circuit analysis reveals tremendous benefits; for a 4-bit full adder design, SN3D shows $11\times $ , 19%, 18%, and 6.7 $\times $ , 8.69%, 9% benefits over state-of-art 2-D CMOS and transistor-level monolithic 3-D in terms of density, power, and performance, respectively. In addition, our step-by-step TCAD emulation of manufacturing flow establishes feasibility. If realized, the SN3D fabric can be transformative for the semiconductor industry. [ABSTRACT FROM AUTHOR]

This paper presents a unique methodology to calculate the filtering capacitances of the OpAmp already reported in the literature. The methodology aims to show that there is an optimal solution that can be used to increase the electromagnetic interference (EMI) immunity of any OpAmp for a wide range of frequencies. By using this proposed methodology, an OpAmp structure that reduces the EMI effect is designed in standard 0.18\ \mu \textm mixed-mode CMOS technology. Detailed mathematical analyses and simulation results are presented and discussed. Simulation results show that the maximum EMI-induced offset voltage in the frequency range from 1 MHz to 1 GHz for the OpAmp is 4.4 mV. In contrast, the standard Miller OpAmp exhibits a maximum EMI-induced offset voltage of 92.4 mV under the same operating conditions. [ABSTRACT FROM PUBLISHER]

In this paper, we present a surface potential-based explicit continuous model for a metal-ferroelectric-insulator-semiconductor (MFIS) type gate-all-around negative capacitance transistor (GAA-NCFET). Unlike previously reported models, an explicit formulation to calculate the electrical characteristics of GAA-NCFET is proposed. Our model includes the radial dependence of the electric field in the ferroelectric, ignored in the previous works and accurately captures ferroelectric material parameter variations in the nonhysteretic regime. In contrast to bulk NCFETs, GAA-NCFET characteristics show different bias dependence due to the absence of bulk charge. We also present analytical expressions for the terminal charges which are essential to obtain the trans-capacitances for transient simulations. We find that, compared with conventional MOSFETs, the gate charge saturates to a different fraction of its maximum value. Furthermore, the modeling of quantum mechanical effect and overlap capacitances in an MFIS NCFET structure is discussed. Finally, the proposed model has been implemented in Verilog-A and tested for the transient response of ring oscillator in a commercial circuit simulator. [ABSTRACT FROM AUTHOR]

This paper demonstrates the use of doublegate accumulation mode (AM) and junctionless (JL) transistors for dynamic memory applications at 85 °C. The doping dependent assessments of AM and JL devices include an analysis of storage volume, carrier lifetime, and depth of potential well to determine characteristics of Dynamic Random Access Memory (DRAM). This paper shows significant impact of carrier lifetime for channel doping (Nd) ≤ 1018 cm-3 on Retention Time (RT), while the depth of potential well is more critical at higher doping (>1018 cm-3). RT of ~2.5 s at 85 °C and ~4.5 s at 27 °C is achieved for gate length (Lg) of 400 nm with Nd = 1017 cm-3 which reduces to ~90ms at 85 °C for Lg = 25 nm. This paper discusses the storage volume (Lg x film thickness for a fixed volume with width of 1 µm) optimization to attain maximum retention. Insights and guidelines, as a function of doping and device dimensions, are outlined for dynamic memory applications. [ABSTRACT FROM AUTHOR]

In this paper, a physics-based compact model for calculating the semiconductor charges and gate capacitance of III-V nanowire (NW) MOS transistors is presented. The model calculates the subband energies and the semiconductor charges by considering the wave function penetration into the gate insulator, effective mass discontinuity at the semiconductor-oxide interface, 2-D confinement in the NW, and Fermi-Dirac statistics. The semiconductor charge expression proposed in this paper is completely explicit in terms of applied gate voltage, therefore, making it highly suitable for large circuit simulations. The model is also compared with the results from self-consistent Schrödinger-Poisson solver for different NW sizes and materials and found to be accurate over a wide range of gate voltages. [ABSTRACT FROM AUTHOR]

In this paper, we propose a vertical transistor with n-bridge and body on gate (BOG-DRAM) for Low-power 1T-DRAM application. The vertical channel of the device can reduce the short-channel effect and improve scalability. The storage region stacked on the gate leads to the efficient utilization of storage space. The device with junctionless channel layers on three sides can improve writing time. The conventional current bridge device only has one side gatecontrol depletion region, but the proposed BOG-DRAM has triple-side gate-control depletion region which can improve programming window at shorter gate lengths. BOG-DRAM achieves programming window of 33.6 μA/μm when the storage length is 20 nm. In addition, the work function offset is exploited for low-power application. [ABSTRACT FROM AUTHOR]

The main purpose of this paper is to introduce an ultracompact device for electrostatic discharge (ESD) protection based on a bipolar metal oxide silicon (BIMOS) transistor merged with a dual back-to-back silicon-controlled rectifier (SCR) for bulk and for ultrathin body box fully depleted (FD)-silicon on insulator (SOI) advanced CMOS technologies in the hybrid bulk thanks to process co-integration. It is well known that ESD protection is a challenge for IC in advanced CMOS technology. In this paper, an optimized solution is described through the concept, design, 3-D technology computer aided design (TCAD) simulation, and silicon characterization in 28-nm FD-SOI in hybrid bulk. Measurements are done thanks to transmission line pulsed (TLP), very fast TLP and dc behavior. Moreover, the overvoltage is investigated through very fast transient characterization system measurements. It demonstrates a promising candidate to protect against ESD event and to develop new ESD network dedicated to system on chip. [ABSTRACT FROM PUBLISHER]

In this paper, we report a physics-based compact model for monolayer graphene field-effect transistors (m-GFETs) based on the 2-D Density of States of monolayer graphene and the drift-diffusion equation. Furthermore, the Ward-Dutton charge partitioning scheme has been incorporated to the model extending its capabilities to AC and transient simulations. The model has been validated through comparison with DC and RF measurements from two different long-channel m-GFET technologies. Moreover, values of parasitic elements included in the model are extracted from measurements on dedicated test structures and verified through electromagnetic simulations (EM). Finally, an EM-SPICE co-simulation has been carried out to assess the applicability of the developed m-GFET model for the design of “balun” circuits. [ABSTRACT FROM PUBLISHER]

Based on time-variant behavior of metal-oxide–semiconductor field-effect transistors in large-signal operations, harmonic translations and their mutual effects are analyzed. Large amplitudes at terminal voltages of these transistors push them into different regions of operation. In this paper, harmonic translations are derived as a result of such changes in operation region of transistors. Operation in triode region for a portion of oscillation cycle results in iterative harmonic translations between fundamental frequency and second harmonic. They boost each other constructively for significantly stronger oscillation, more second harmonic output power, and enhanced dc-to-RF efficiency. Based on this analysis, a 215-GHz signal source, implemented in a TSMC 65-nm CMOS LP is presented. The proposed oscillator achieves a maximum output power of 5.6 dBm and a dc-to-RF efficiency of 4.6%. The measured phase noise is −94.6 dBc/Hz at 1-MHz offset. The proposed oscillator occupies only 0.08 mm2 of chip area. [ABSTRACT FROM PUBLISHER]

The separation of the data capture and analysis in modern vision systems has led to a massive amount of data transfer between the end devices and cloud computers, resulting in long latency, slow response, and high power consumption. Efficient hardware architectures are under focused development to enable Artificial Intelligence (AI) at the resource-limited sensing devices. One of the most promising solutions is to enable Processing-in-Pixel (PIP) scheme. However, the conventional schemes suffer from the low fill-factor issue. This paper proposes a PIP based Complementary Metal-Oxide-Semiconductor (CMOS) sensor architecture, which allows convolution operation before the column readout circuit to significantly reduce the overall power consumption while improving the resource utilization of the succeeding deep learning accelerator. The simulation results show that the proposed architecture could support the computing efficiency up to 3.37 TOPS/W at the 8-bit weight configuration, which is four times as high as the conventional schemes after normalization. The transistors required for each pixel are only 3.5T, significantly improving the fill-factor. [ABSTRACT FROM AUTHOR]

A bond-pad charging protection design for charging-free reference transistor test structures was examined. This paper concludes that truly charging-free reference transistors cannot be realized with the one conventional bond-pad charging protection design of protecting transistor gates only. This, however, can be achieved by simultaneously protecting all terminals of the reference transistors. The simulations, in this paper, reconfirm the earlier important experimental conclusion that placing protection device(s) at transistor gates may inflict severe damage to transistor gate oxides instead of protecting them. The implication of the above suggests that attention may be required in a circuit layout design stage for those transistors which gates begin to connect, at high metal layers, to highly efficient leakage paths, such as protection devices, n-type source/drain diffusion regions, and VSS bus lines, which tend to pull transistor gates to low potentials during a backend integrated-circuit manufacturing process. This paper proposes an optimum bond-pad charging protection design for the truly charging-free reference transistor test structures by considering a minimum usage in a layout space and minimum gate oxide stress in the fuse zap-off process. [ABSTRACT FROM AUTHOR]

The missing capability of DC fault current limitation is an inherent disadvantage in the typical half-bridge modular multilevel converter (MMC), which restricts its application in long-distance high voltage direct current (HVDC) systems. The full-bridge submodule (FBSM) and clamped double submodule (CDSM) topologies allow for fault current self-clearing, but also suffer from additional hardware cost and power losses. In this paper, an improved submodule topology utilizing the new reverse-blocking insulated gate bipolar transistor (RB-IGBT) is proposed to improve system economy without compromising its performance. The operating principles and DC fault characteristics of the proposed topology are analyzed in details, including comparison with various topologies in terms of cost, losses, fault current clearing speed and thermal. Compared with the traditional CDSM topology, the proposed RBSM can clear DC faults faster with reduced cost of switching devices. A DC fault protection strategy for MMC-HVDC system is also presented. Simulation results under normal operating, permanent and transient fault conditions are carried out to verify the effectiveness of the proposed topology. [ABSTRACT FROM AUTHOR]

Envelope tracking (ET) is a promising power amplifier (PA) architecture for current and future communications systems, which uses dynamic modulation of the supply voltage to provide high efficiency and potentially very wide bandwidth over a large dynamic range of output power. However, the dynamic nature of the supply voltage can lead to a problematic variation in transistor gain, particularly in GaN HEMTs. This paper describes and analyzes this behavior and the detrimental effect it can have on ET PAs. Contributing factors and origins of gain variation are described in detail along with how, for the first time, meaningful comparisons can be made between different devices. Using these guidelines, gain variation is shown to be a widespread issue effecting most GaN HEMTs presented in literature. To allow an analysis of the intrinsic device behavior, an extended transistor model is developed that takes the effect of gate and source field plates into account. This model is refined using measurement data and used to demonstrate the fact that the parasitic gate–drain capacitance ($C_{\textrm {GD}}$) is the main contributor to the small-signal gain variation—a significant part of the overall gain variation. Based on this knowledge, possible strategies to reduce gain variation at the transistor technology level are proposed, allowing the optimization of GaN HEMTs specifically for ET PAs. One identified strategy involves reducing the length of the gate field plate and is shown to be a viable approach to reduce the gain variation in GaN HEMTs, albeit at an increased RF/dc dispersion. [ABSTRACT FROM AUTHOR]

Recently, exploiting emerging nonvolatile memories to implement the process-in-memory (PIM) paradigm have shown great potential to address the von Neumann bottleneck and have attracted extensive research and development. In this paper, we present a novel PIM platform—spintronic processing unit (SPU), within spin transfer toque magnetic random access memory (STT-MRAM). This energy-efficient and reconfigurable PIM platform can perform different tasks—data storage and logic computing—using the same physical fabric that is programmable at the finest grain, i.e., the individual memory cell level, without the need to move data outside the memory fabric. The proposed SPU works just like a typical memory and all the logic functions are achieved through regularmemory-like write and read operations with minimal modifications. The functionality and performance are evaluated via hybrid circuit simulations under the 40-nm process technology node. Our proposed SPU is expected to be a feasible PIM platform in the near future, owing to the increasing maturity of STT-MRAM. [ABSTRACT FROM AUTHOR]

Near threshold computing (NTC) has significant role in reducing the energy consumption of modern very large scale integrated circuits designs. However, NTC designs suffer from functional failures and performance loss. Understanding the characteristics of the functional failures and variability effects is of decisive importance in order to mitigate them, and get the utmost NTC benefits. This paper presents a comprehensive cross-layer reliability analysis framework to assess the effect of soft error, aging, and process variation in the operation of near threshold voltage caches. The objective is to quantify the reliability of different SRAM designs, evaluate voltage scaling potential of caches, and to find a reliability-performance optimal cache organization for an NTC microprocessor. In this paper, the soft error rate (SER) and static noise margin (SNM) of 6T and 8T SRAM cells and their dependencies on aging and process variation are investigated by considering device, circuit, and architecture level analysis. Their experimental results show that in NTC, process variation and aging-induced SNM degradation is $2.5{\boldsymbol \times }$ higher than in the super threshold domain while SER is $8{\boldsymbol \times }$ higher. At NTC, the use of 8T instead of 6T SRAM cells can reduce the system-level SNM and SER by 14% and 22%, respectively. Besides, we observe that we can find the right balance between performance and reliability by using an appropriate cache organization at NTC which is different from the super threshold. [ABSTRACT FROM AUTHOR]

This paper proposes a high-efficiency single-phase T-type boundary conduction mode (BCM) microinverter. The conventional full-bridge BCM microinverter has achieved zero voltage switching (ZVS) and thereby improved the efficiency, but it suffers from high switching losses under light load conditions. The proposed T-type BCM microinverter reserves ZVS and uses a multilevel technique to further decrease the switching losses. The BCM operation with multilevel technique will have too low switching frequency when the grid voltage approaches half of the dc link voltage. To solve this problem, this paper adopts a third operation mode for the T-type switching leg to maintain the switching frequency above a minimum value. The corresponding mode transitions are also detailed to ensure a smooth operation. Because of the turn-offdelay of the freewheeling transistor, the actual lower current boundary deviates from the programmed one, which will distort output current. To address this issue, this paper also proposes a boundary compensation method. A prototype has been built for performance verification, which can test both full-bridge and T-type topology. Compared with the full-bridge BCM microinverter, the proposed T-type BCM microinverter has a higher efficiency over the whole load range. [ABSTRACT FROM AUTHOR]

In this paper, a comparison of gate-all-around nanowire-FETs, nanoplate (NP)-FETs, and FinFETs was undertaken for the same areas of not only the gate metal but also the silicon channel, assuming the same conditions regarding the parasitic components. It is known that an NP-FET structure not only improves the delay performance but also enhances the immunity to short-channel effects. In addition, it is found that the use of a dual-k spacer with an NP-FET further improves the on-state as well as the off-state performances. [ABSTRACT FROM AUTHOR]

Variability is a challenge for future scaling as process dimensions reduce. The emerging 3-D sequential stacking technology is more than Moore’s scaling alternative. The 3-D design flow requires the partitioning of the netlist between the tiers. This paper presents the variability analysis of circuits partitioned into different levels. A comparison among local and global variations effects on ring oscillators (ROs) and SRAM is demonstrated. The across-chip variations and correlation range are shown as a critical point for the 3-D very large-scale integrated circuits, where the local variability is dominant. The correlations between devices due to the distances or the allocation into different tiers are directly taken into account in the SPICE model due to a statically unified model applied to 3-D circuits based on Monte Carlo simulations. Design wise, the 3-D integration can further decrease the circuit variability as shown in RO output frequency and SRAM static noise margin. [ABSTRACT FROM AUTHOR]

In this paper, a switchless dual-band, dual-mode power amplifier (PA) has been developed for multioctave operation. For the switchless dual-band operation, we proposed a new coupled-line-based diplexer structure. The proposed diplexer operates as a balun at high-band frequencies and the coupled line at low-band frequencies. To better understand the proposed diplexer structure, a detailed analysis is performed based on a coupled-line theory. The interstage and output-stage matching networks of the PA are designed by using the proposed diplexer. To maximize the power-added efficiency (PAE) of the designed PA, turn off the unused band PAs and decide the optimum off-state bias condition of transistors. The designed dual-band, dual-mode PA is fabricated with commercial 0.25- $\mu \text{m}$ GaN HEMT process. The fabricated PA shows over 15-dB small-signal gain at 5–11 GHz in the low-band mode and 9–18 GHz in the high-band mode. The measured average output power and PAE are 35 dBm, 23% in the low-band mode and 37 dBm, 26% in the high-band mode. [ABSTRACT FROM AUTHOR]

This paper presents an analysis and design technique for an RF-input load modulated balanced amplifier (LMBA) with Doherty-like efficiency enhancement characteristics. The LMBA is a recently introduced power amplifier (PA) architecture in which a control signal is injected into the isolation port of the output coupler of a balanced amplifier. This control signal modulates the apparent load impedance of the two main devices. In the RF-input LMBA presented in this paper, the control signal is generated from the single-modulated RF input using a class-C control amplifier, producing an overall Doherty-like response. A complete theoretical analysis of the RF-input LMBA is presented, by assuming generalized networks for the output combiner and input splitter which are then solved in terms of transistor parameters and boundary conditions for high efficiency. Based on this analysis, a generalized design technique is presented. The LMBA is, thereby, shown to have advantages over the Doherty PA, including the device periphery scaling between the main and control devices for a given back-off range. The design technique is demonstrated at 2.4 GHz using packaged GaN devices. The prototype has a peak CW output power of 45.6 dBm, with a power-added efficiency of 60% at peak power and 50% at 6-dB output back-off, including the power dissipated by the control PA. A long-term evolution signal at 2.4 GHz with a 7.5-dB peak-to-average power ratio is demonstrated with an unlinearized ACLR of −27 dBc and an average drain efficiency of 47% for a 38-dBm average output power. [ABSTRACT FROM AUTHOR]

This paper proposes a diode-based multi-level cell spin–orbit torque magnetic random-access memory (SOT-MRAM) for high density memory applications. By stacking a shared diode over two in-parallel magnetic tunnel junctions (MTJs) that share a common heavy metal electrode, our proposed cell requires only one silicon-based transistor to access 2 bits. Having just one transistor to access 2 bits results in at least $2\times $ higher density when compared to various designs in the literature. Moreover, utilizing a high-density MRAM process, a $4F^{2}$ effective 1 bit area can be achieved. Furthermore, our proposed design maintains the high energy efficiency of the conventional SOT-MRAM. In addition, an improvement in figure of merit of at least 30% defined as energy area product is obtained compared to various designs. This paper also discusses the essential requirements to realize our proposed SOT-MRAM, such as the realization of diode MTJ stack and SOT-MTJs, with different critical currents and electrical resistances. [ABSTRACT FROM AUTHOR]

Ultrahigh-speed magnetization switching mechanism is strongly pursued, as it can improve the write performance of the magnetic random access memory (MRAM) and further extend the application area of spintronics. Currently, a well-studied switching mechanism is the spin transfer torque (STT), whose speed, however, is limited by an intrinsic incubation delay. Recently, spin–orbit torque (SOT) was proposed to solve the speed bottleneck of the STT. In this paper, we evaluate the potential of two types of SOT-MRAMs, whose data can be ultrafast written based on the recently discovered field-free SOT mechanisms, respectively. A cross-layer analysis is presented involving device modeling, circuit-level optimization, and architecture-level evaluation. First, the principle of the magnetization switching is analyzed with the macrospin simulation. Then, the optimization strategies at the circuit level are generalized through the SPICE-type simulation. Finally, we build up the memory architecture with the SOT-MRAMs, STT-MRAM, and static RAM. Their read/write performances are evaluated with NVSim software. It is demonstrated that one of the studied SOT-MRAMs shows the promising prospect in the non-volatile memory, especially suitable for high-capacity cache. [ABSTRACT FROM AUTHOR]

In this paper, for the first time, we demonstrate the improvement in power capability and safe operating area of silicon-on-insulator laterally double-diffused MOS (SOI-LDMOS) transistors for power amplifier applications by the introduction of a partial n+ buried layer (PNBL). The power capability of a transistor can be evaluated by ${P}_{\text {max}}/A$ , which is the maximum power per unit area that can be delivered by the transistor and is an important parameter for power amplifiers. ${P}_{\text {max}}$ is dependent on the snapback voltage (${V}_{\text {sb,QS}}$), OFF-state breakdown voltage (${V}_{\text {bd}, \mathrm{\scriptscriptstyle OFF}}$), and maximum current (${I}_{\text {max}}$) in the quasi-saturation regime of an LDMOS transistor. Increase of ${P}_{\text {max}}/A$ by the introduction of a PNBL in the SOI-LDMOS transistors is reported in this paper. The effects of variation of the length, thickness, and doping concentration of the PNBL on ${V}_{\text {sb,QS}}$ and ${P}_{\text {max}}/A$ are analyzed in detail. It is shown that by optimizing the doping and length of the PNBL layer, the maximum power output from the transistor can be made significantly higher than that of a conventional device without PNBL. A procedure to design the optimized structure is also presented. [ABSTRACT FROM AUTHOR]

This paper presents an explicit analysis of the bandwidth and port imbalance of a subquarter-wavelength transmission-line (t-line) transformer and verifies this with the design of a two-stage D-band power amplifier (PA). Series power combining techniques incorporate both stacked heterojunction bipolar transistors (HBTs) and power combining using an 8-way sub-quarter-wavelength t-line transformer above 100 GHz. The extremely compact power combining methodology leads to a small die area of 0.62 mm2 and a record 254-mW/mm2 output power per unit die area. The PA has been fabricated in a 90-nm silicon germanium BiCMOS technology and produces more than 21-dBm output power over the frequency range of 114–130 GHz with a peak output power of $160~{\mathrm{mW}}\approx 22~{\mathrm{ dBm}}$ at 120 GHz and a 3-dB small-signal bandwidth of 35 GHz. This output power is 32% higher than the highest prior art, while the chip area is 77% smaller. [ABSTRACT FROM AUTHOR]

In this paper, a new circuit model is proposed to simulate plasma-induced damage (PID). It includes the two sheaths of the plasma, the blocking capacitor, and the RF power source coupled with the victim device. As a result, this model accounts for both the ac behavior of the plasma and the local voltage imbalance induced by differences in shading factors. The advantage of this approach, compared with the previous works, is that it can yield the waveform impinged by the plasma on the transistor gate, which is known to influence the amount of reliability degradation experienced by the device. Finally, this model is also utilized to elucidate the impact of a parasitic capacitance connected to the transistor gate on the extent of the PID inflicted on the victim device. [ABSTRACT FROM AUTHOR]

A general equivalent circuit model for the metal/organic semiconductor (OSC)/liquid/metal system is presented. Each circuital element, representative of a physically observable phenomenon associated with the device working principle, is analyzed and discussed. Two case studies of electrochemical impedance spectroscopy of devices featuring NaCl (concentration of 0.1 M) and MilliQ water as liquid are reported, showing that both cases can be considered as a particular case of the general model presented in this paper. Experimental verification of the two practical cases is achieved by performing measurements onto electrodes coated by either an n-type or a p-type OSC deposited by sublimation or drop casting, respectively. The good agreement with the experimental data makes our model a useful tool for the characterization and failure analysis of electronic devices, such as water-gated transistors, electrophysiological interfaces, fuel cells, and others electrochemical systems. [ABSTRACT FROM AUTHOR]

In this paper, we report on the assessment of self-amplified silicon–oxide–nitride–oxide–silicon (SONOS) memory device architecture for sub-50-nm gate length (${L}_{g}$) through calibrated simulations. Self-amplification (SA) effect in tunnel field-effect transistor-based SONOS (T-SONOS) memory device has been analyzed. Results show that memory window ($\Delta {W}$) in T-SONOS cell increases as buried oxide thickness increases due to capacitive coupling between the front and back gates. Although the enhanced $\Delta {W}$ can also be achieved in inversion-mode SONOS (I-SONOS) device, its performance is deteriorated when the gate length is scaled down. We have compared the performance of I-SONOS and T-SONOS memory devices for ${L}_{g}$ varying from 100 to 25 nm. Results highlight that I-SONOS device cannot be programmed at ${L}_{g} ={25}$ nm and thus deteriorate the memory operation. However, SA T-SONOS at ${L}_{g} = {25}$ nm achieves ${W} \sim {6}$ V. In addition, the effect of underlap on the performance of T-SONOS cell has been analyzed, and it is shown that memory operation of 25-nm T-SONOS device can further improved with a drain side underlap of 20 nm. This paper provides new opportunities to design SA T-SONOS memory device for the next-generation nonvolatile memories. [ABSTRACT FROM AUTHOR]

The application of capacitive deep trench isolation (CDTI) as a shared vertical transfer gate (VTG) in a back-side-illuminated CMOS image sensor pixel is investigated using 3-D device-level simulations. The parasitic capacitance existence between CDTI and deeply buried pinned photodiode (BPD), and also between CDTI and floating diffusion (FD) region, makes the charge transfer process more difficult. In order to design a lag-free pixel and obtain complete charge transfer from BPD to FD, various considerations regarding the device-level design should be taken into account which is discussed in this paper. A CDTI neighboring two pixels can be functionalized as a shared VTG. Using CDTI as shared VTG facilitates pixel miniaturization and can result in more circuit integration at the pixel surface. This paper proposes a ${2}\,\,\mu \text{m} \times {2}\,\,\mu \text{m}$ pixel with CDTI as shared VTG, an equilibrium full-well capacity of 4605 e−, and a complete charge transfer from BPD to FD. [ABSTRACT FROM AUTHOR]