scispace - formally typeset
Search or ask a question
Journal ArticleDOI

Accurate and Computationally Efficient Modeling of Nonquasi Static Effects in MOSFETs for Millimeter-Wave Applications

TL;DR: In this article, an improved physical equivalent circuit was derived using a transmission line model, by incorporating the high-frequency longitudinal gate electrode and a channel distributed RC network, which was implemented in a BSIM-BULK MOSFET model and validated with dc and RF data, obtained from technology computer aided design device simulations and experimental data.
Abstract: A lumped-circuit nonquasi-static (NQS) model, that is applicable for both large-signal transient simulations and a small-signal ac analysis, is developed in this paper. An improved physical equivalent circuit, capturing NQS effects in the millimeter waveband, is derived using a transmission line model, by incorporating the high-frequency longitudinal gate electrode and a channel distributed RC network. The proposed model is implemented in a BSIM-BULK MOSFET model and validated with dc and RF data, obtained from technology computer-aided design device simulations and experimental data. The proposed model is in very good agreement with the data up to ${50}{f}_{t}$ . The transient currents, for a gate-voltage switching rate of ${5}\times {10}^{{10}}$ V/s, show excellent match with the data. The dc, transient, and ac simulations using the proposed model are much faster than a 10-segmented MOSFET model. This shows that the proposed model is better than other computationally complex compact models, for most RF applications.
Citations
More filters
Journal ArticleDOI
TL;DR: In this article, a BSIM-based compact model for a high-voltage MOSFET is presented, which has been extended to include the overlap capacitance due to the drift region as well as quasi-saturation effect.
Abstract: A BSIM-based compact model for a high-voltage MOSFET is presented. The model uses the BSIM-BULK (formerly BSIM6) model at its core, which has been extended to include the overlap capacitance due to the drift region as well as quasi-saturation effect. The model is symmetric and continuous, is validated with the TCAD simulations and experimental 35- and 90-V LDMOS and 40-V VDMOS transistors, and shows excellent agreement.

23 citations

Journal ArticleDOI
TL;DR: An improved model of bulk charge effect for both drain current and capacitances and its implementation in the industry standard Berkeley short-channel IGFET model (BSIM)-BULK model is presented.
Abstract: In this brief, we present an improved model of bulk charge effect for both drain current ( ${I}_{\text {DS}}$ ) and capacitances and its implementation in the industry standard Berkeley short-channel IGFET model (BSIM)-BULK model. The proposed model captures all the well-known and important bulk charge effects, as the Abulk term does for BSIM3/BSIM4. The model is validated with the experimental and technology computer-aided design (TCAD) data. The proposed model enhances the fitting accuracy for ${I}_{\text {DS}}$ , and more importantly its derivatives and capacitances too.

10 citations


Cites background from "Accurate and Computationally Effici..."

  • ...It shows excellent convergence for dc [1], small signal, RF, and transient simulations, and also well suited for high-speed analog applications [3]–[6]....

    [...]

Proceedings ArticleDOI
01 Apr 2019
TL;DR: The recent and upcoming enhancements of the industry standard BSIM-BULK model are presented and an analytical model for bulk charge effect, in both current and capacitance, is implemented to improve the model accuracy for transconductance and output conductance.
Abstract: In this work, we present the recent and upcoming enhancements of the industry standard BSIM-BULK (formerly BSIM6) model. BSIM-BULK is the latest body referenced compact model for bulk MOSFETs having a unified core, which is developed by the BSIM group for accurate design of analog and RF circuits. The model satisfies the symmetry test for DC and AC, correctly predicts harmonic slope, and exhibits accurate results for RF and analog simulations. In order to further improve the model accuracy for transconductance $(g_{m})$ and output conductance $(g_{ds})$, an analytical model for bulk charge effect, in both current and capacitance, is implemented. Several other advanced models are added to capture real device physics. These include: parasitic current at the shallow trench isolation edges; leakage current components in zero threshold voltage native devices; new model for NQS to capture the NQS effects up to the millimeter wave regime; self heating effect; and heavily halo implanted MOSFET’s anomalous g m , flicker noise and I DS mismatch. All these enhancements have been implemented to high standards of computational efficiency and robustness.

7 citations


Cites background or methods from "Accurate and Computationally Effici..."

  • ...to suppress the higher order harmonics while deigning power amplifiers, and to avoid the overestimation of transistor power gain in RF circuits [14]....

    [...]

  • ...We propose and implement an accurate and computationally efficient transmission line based improved physical equivalent circuit model for NQS in BSIMBULK, to capture the NQS effects [14], [15]....

    [...]

  • ...Furthermore, the left dip in Real(Y21) is due to a gm term in Real(Y21) expression [14]....

    [...]

  • ...b) An improved physical model capturing NQS effects in the millimeter wave band [14], [15]....

    [...]

  • ...11: Drain and source terminal currents for gate switching from 0 to 5V at a rate of 5X10(10)V/s [14]....

    [...]

Proceedings ArticleDOI
29 Jan 2021
TL;DR: In this paper, a fast optimization hyperparameter and sparse support vector machine (FOH-SSVM) algorithm was proposed to solve the problem of signal integrity, which greatly reduced the modeling time and increased the prediction accuracy.
Abstract: Compared with the traditional support vector machine regression (SVR), the SVR hyperparameter fast optimization algorithm can improve the accuracy of the prediction results. However, the data shows that when the training sample is too large, it will increase the complexity of model learning, resulting in too long modeling time. Therefore, we refer to the most effective support vector set search method in the variable selection and sparse support vector machine (VSߝSSVM) algorithm, and appropriately fit the “advantages” of these two algorithms to construct a fast optimization hyperparameter and sparse support vector machine (FOH-SSVM) algorithm. In this work, we use the algorithm to solve the problem of signal integrity. The experimental results show that the modeling time required by the FOH-SSVM algorithm is 1%, which greatly reduces the modeling time. At the same time, the prediction accuracy of the algorithm is increased by 8%, ensuring good prediction performance.

4 citations

Proceedings ArticleDOI
01 Sep 2019
TL;DR: Highlights from Silicon Device Physics, material sciences and electrical engineering are among the first results to be presented from GFs subcontracts in the IPCEI-project, namely a reconfigurable FET compatible with 22-FDX-technology, a CMOS compatible new material Si doped HfO2 for electrocaloric/ pyroelectric effects on chip.
Abstract: Highlights from Silicon Device Physics, material sciences and electrical engineering are among the first results to be presented from GFs subcontracts in the IPCEI-project, namely a reconfigurable FET compatible with 22-FDX-technology, a CMOS compatible new material Si doped HfO 2 for electrocaloric/ pyroelectric effects on chip, modelling of the 22FDX devices in the higher GHz range and first 5G Dual Band transceiver blocks designed in 22FDX

2 citations


Cites background from "Accurate and Computationally Effici..."

  • ...Despite the profound impact of the gate resistance (Rg) modeling on Yparameters, even a more sophisticated distributed RC gate network [16] does not change the results....

    [...]

References
More filters
Journal IssueDOI
TL;DR: In this paper, the EKV3 MOSFET model is compared with DC, CV and RF measurements up to 20 GHz of a 110 nm CMOS technology and scaling behavior over a large range of channel lengths and bias conditions is presented.
Abstract: This article presents a validation of the EKV3 MOSFET compact model dedicated to the design of analogue-RF ICs using advanced CMOS technology. The EKV3 model is compared with DC, CV and RF measurements up to 20 GHz of a 110 nm CMOS technology. The scaling behaviour over a large range of channel lengths and bias conditions is presented. Long-channel devices show significant non-quasi static effects while in short-channel devices the parasitics modelling is critical. This is illustrated with Y-parameters and ft vs. ID in NMOS and PMOS devices, showing good overall RF modelling abilities of the EKV3 MOSFET model. © 2008 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2008. The authors dedicate the present manuscript to the memory of Peter Bendix.

5 citations


"Accurate and Computationally Effici..." refers background or methods in this paper

  • ...In order to further validate the accuracy of the proposed model, the dc and RF experimental data are obtained from [14] where they have matched the NQS behavior using an EnzKrummenacher-Vittoz segmentation method....

    [...]

  • ...Symbols: Measured data [14], Lines: Model....

    [...]

  • ...Comparison with experimental data obtained from [14] is presented in Section IV....

    [...]

  • ...The RF data of [14] have frequency measurements for this device in the range of 50 MHz–20 GHz, for VGS = 0....

    [...]

  • ...Symbols indicate the measured data [14], and it shows the NQS effect at very high frequencies....

    [...]

Proceedings ArticleDOI
31 Oct 2000
TL;DR: In this paper, the impact of channel width scaling on the transistor y-parameters for the second order NQS model was investigated for MOSFETs with small channel widths.
Abstract: In this paper, we investigate the impact of nonquasi-static (NQS) effects on the small-signal RF behaviour of MOSFETs. We observed that the transistor power gain G/sub Tmax/ is largely overestimated in the quasi-static model, whereas it is accurately predicted using a second order NQS model. Furthermore, we investigated the impact of channel width scaling on the transistor y-parameters for the second order NQS model. Accurate modelling results are obtained for MOSFETs with small channel widths. However, experimental results indicate that the effective gate resistance, stemming from the distributed nature of the MOS gate, tends to be lower than R/sub gate/=WR/sub /spl square///(3L) at frequencies close to or larger than f=1/(R(/spl square/)C/sub ox/W/sup 2/). This was observed for transistors with a 25 /spl mu/m channel width.

5 citations


"Accurate and Computationally Effici..." refers background or methods in this paper

  • ...However, for the frequency range above ≈ ft/3, accurate NQS models are preferable to avoid the overestimation of a transistor power gain [13]....

    [...]

  • ...Conventional models in [13] and [16] cannot capture the NQS trend in the millimeter-wave range, whereas the segmentation model with segments ≥ 17 can capture the NQS trend, but it increases the computational time by a large amount....

    [...]

Proceedings ArticleDOI
01 Dec 2016
TL;DR: In this article, an accurate physics-based lumped-circuit model, representing the nonquasi-static effect and the distributed polygate electrode resistance in bulk MOSFET is proposed.
Abstract: An accurate physics-based lumped-circuit model, representing the nonquasi-static effect and the distributed polygate electrode resistance in bulk MOSFET is proposed. In our approach, an improved physical equivalent circuit capturing NQS effect is developed by using pole-zero compensation technique. The new model takes into account the inherent delay in the response of channel charges to the applied voltage. The proposed model is implemented in BSIM6 MOSFET model, and it is validated with Y-parameter data obtained from technology computer-aided design device simulations. The proposed model is in very good agreement with the device data up to 50/t, which ensures that it is suitable for most RF applications over any other computationally complex compact models.

4 citations


"Accurate and Computationally Effici..." refers methods in this paper

  • ...Going beyond [27], we have derived the proposed model and explained the physics behind the NQS trends in more detail....

    [...]

Proceedings ArticleDOI
13 Jun 2000
TL;DR: In this paper, an accurate non-Quasistatic SPICE model based on relaxation time is proposed for ac and transient simulation of high speed and radio frequency (RF) circuits, which is also provided a method for extracting the relaxation time.
Abstract: An accurate Non-Quasistatic SPICE model based on relaxation time is proposed for ac and transient simulation of high speed and radio frequency (RF) circuits. A method for extracting the relaxation time is also provided. Its dependence on Vgs and Vds and channel length is based on physics and built in the model. Finally the model is verified with both 2D simulation and measurement.

4 citations