Mrs.M.Muthulakshmi is currently pursuing masters degree program in Embedded Systems Technologies in Anna University of Technology- Tirunelveli, India. E-mail:lakshmingm.2@gmail.comInternational Journal of Scientific & Engineering Research, Volume 3, Issue 5, May-2012 1
ISSN 2229-5518
Parameter Extraction for BSIMSOI4.3 MOSFET Model
Mrs.M.Muthulakshmi, Mrs.L.Sheela
Abstract— This paper provides the extracted values of the parameters affecting the threshold voltage model of SOI MOSFET. The parameter extraction is done for BSIMSOI4.3 MOSFET model. The proposed procedure is designed to give the results based on the device characteristics data. Simulations are performed using the extracted parameters and finally it is compa red for extracted parameters and generic device parameters. The effect of body bias voltage on the threshold voltage is studied by using the extracted val ues. Finally we summarize the work with accurate study of extracted parameters in depicting their effect on the characteristics of device.
Index Terms— Body bias effect, BSIM, BSIMSOI, DIBL, Extraction Methodology, MOSFET, Threshold voltage.
—————————— ——————————
EEMICONDUCTOR devices have various models that satisfy the behaviour of the device at different operating conditions. Each model has its uniqueness that de- scribes the physical phenomena of the device. The process of obtaining parameters by minimizing the difference be- tween the measured parameter and the modelled parame- ter at different bias value is called parameter extraction. Parameter extraction is a crucial and difficult step for cir- cuit and device simulation. The parameters may be tech- nological or electrical or adjusting type. The models need to provide accuracy in reproducing the I-V characteristics. The most compact MOSFET model available considering all the effects due to the reduction in transistor dimensions is BSIMSOI4.3 [5].There are two ways of obtaining these parameters namely ―gradient based method‖ and ―optimi- zation method‖. In the gradient based method the parame- ters are extracted one by one. In the optimization method all the parameters are determined at the same time [2], [8],
[9].
We have considered the least mean square (LMS) me-
thod, because any physical parameter can be studied as a function of technology or operating conditions, which cha- racterize the device. Threshold voltage is an important fac- tor for determining the logical function of the circuit. Hence, the threshold voltage is a fundamental parameter for MOSFET modelling and characterization.
Suayb Yener [1], has designed parameter extraction al-
gorithm from BSIM3v3 MOSFET model equations. BSIM3v3 models the bulk MOSFET characteristics. They have considered the BSIM3v3 complete threshold voltage
————————————————
Mrs.L.Sheela is currently working as an Assistant Professor in Anna
University of Technology-Tirunelveli, India. E-mail:l_sheela79@yahoo.com
model and the Mobility model.
In our current work, we provide an analytical parameter extraction methodology to extract all the parameters in the BSIMSOI4.3 threshold voltage model. Recently in MOSFET fabrication SOI process technology is being used to reduce parasitic bipolar effects or drain induced barrier lowering. Another advantage of SOI process technology is reduction of overlap capacitance of gate with drain, source and body regions.
Session2 provides briefly the complete BSIMSOI4.3 thre- shold voltage model used for parameter extraction. Ses- sion3 presents the parameter extraction strategy and pro- cedure, process input parameters, extraction steps and the derived extraction equations. Session4 tabulates the ex- tracted parameter values. Session5 discusses in detail, the simulation results obtained for general and extracted pa- rameters.
Accurate modelling of threshold voltage is one of the most important requirements for precise description of device elec- trical characteristics. In addition, it serves as a useful reference point for the evaluation of device operation regimes. By using threshold voltage, the whole device operation regime can be divided into three operational regions: strong inversion re- gion, weak inversion region and transition region. Complete threshold voltage model considers all the physical mechan- isms like non-uniform doping effect, short channel and nar- row channel effects and is given by (1).
In the (1) [3], the first term represents the Vth for Vbs=0V, the second and third terms consider the vertical non-uniform doping effect, the fourth term represents the lateral non- uniform doping effect, narrow width effect is considered in the fifth term, the sixth term models the small size effect in devices due to both small channel length and small width, the seventh and the eighth term are related to short channel effect and DIBL and the last term are used to model the pocket for- mation due to the lateral non-uniform doping effect.
IJSER © 2012
The research paper published by IJSER journal is about Parameter Extraction for BSIMSOI4.3 MOSFET Model 2
ISSN 2229-5518
)
(1)
Extraction methodology depends on the model and the way the model is being used [4]. Two different strategies are avail- able for extracting parameters: single device and group device extraction strategy. Single device extraction strategy will fit only one device and will not fit devices with different geome- tries and does not consider the physical dependencies. BSIM- SOI4.3 uses group extraction strategy since this overcomes the drawbacks of single device extraction strategy. All devices of the group are measured under same bias conditions.
Here we extract the parameters for a large size device and
two small size devices. Large size devices are used to extract
parameters independent of short channel and narrow channel
effect. The set of devices with a fixed large channel width but
different channel lengths are used to extract parameters re-
lated to the short channel effect. Similarly, the set of devices
with a fixed, long channel length but different channel widths
are used to extract parameters related to the narrow width
effect.
We consider non-uniformly doped silicon channel with gate oxide thickness Tox in order of nm. We consider different channel doping concentration Nch, source and drain doping concentration NDS. Also, the mask level channel length and width are chosen in order of nm.Parameters are extracted for the transistor, operated at specific room temperature. Before the extraction of the model parameters, we need to provide the process parameters. The extracted parameters require above process parameters for its extraction strategy.
TABLE 1
PARAMETER EXTRACTION STEPS
The different device geometries and bias conditions used in the extraction procedure are given in TABLE 1.
The parameters are extracted from the (1), by splitting it ac- cording to the required criteria.
For extracting the parameters representing body bias effect [7]: K1, K2, K1w1, K1W2, LPEB, LPE0 we consider the following (2).
(2)
where K1 and K2 accounts for threshold voltage variation in long channel length and width devices with uniform substrate doping concentration.
In some technologies, [16] the doping concentration near the source or drain is higher than that in the middle of the chan- nel, referred as lateral non-uniform doping. As the channel length becomes shorter, lateral non-uniform doping will cause Vth to increase in magnitude due to increase in the average doping concentration in the channel. The parameter LPEB and LPE0 models the body bias dependence of the lateral non- uniform doping effect. The increase in K1 with decrease in the channel width is captured by K1W1 and K1W2.
(3)
IJSER © 2012
The research paper published by IJSER journal is about Parameter Extraction for BSIMSOI4.3 MOSFET Model 3
ISSN 2229-5518
(4) The actual depletion region in the channel is always larger than one-dimensional analysis (considering channel length alone), due to the existence of the fringing fields. This effect becomes very substantial as the channel width decreases [11] and the depletion region underneath the fringing field be- comes comparable to the classical depletion layer formed from the vertical field. This results in the increase of the Vth with reduction in the channel width, modelled by the parameters K3 and W0. K3b describes the body-bias dependence of K3.
The extraction of parameters for short channel effect: DVT0, DVT1, DVT2 and DIBL effect: Dsub, Eta0, Etab are done using (5) and (6) respectively.
(5)
(6)
In semiconductor fabrication industries the channel length is reduced to increase both the operation speed and the number of components per chip, which in-turn gives rise to short channel effects [10]. Some of the short channel effects are DIBL [6], surface scattering and velocity saturation. In short channel devices the channel length is in the same order of magnitude as the depletion layer widths of the source and drain junction.
(7) is used to extract parameters for non-uniform doping pro- file: DVTP0, DVTP1, DVTP2, DVTP3, DVTP4.
)
In recent technologies, the doping concentration near the source and drain is higher than in the middle of the channel, referred to as lateral non-uniform doping. As the channel be- comes shorter, lateral non-uniform doping will cause Vth to increase in magnitude.
Using the above procedure 22 of unknown BSIMSOI4.3 para- meters are extracted, completely describing the threshold vol- tage of N-type SOIMOSFET by analytical method. The ex- tracted parameter values are given in the (TABLE II).
TABLE 2
EXTRACTED VALUES OF 22 BSIMSOI PARAMETERS
IJSER © 2012
The research paper published by IJSER journal is about Parameter Extraction for BSIMSOI4.3 MOSFET Model 4
ISSN 2229-5518
Fig. 1 Large Channel Length and Width devices (Leff=1000nm, Weff=4000nm) Ids-Vgs characteristics, Vbs pa- ra
The Ids-Vgs characteristics of N-type SOIMOSFET is simu- lated using SPICE. We chose MOSFET devices with different geometries to depict the efficiency of the extracted parameters in considering their dimensions into account. The simulations are done by varying body bias voltage Vbs between 0V to -5V. The Vds used for simulation is 50mV. The Ids-Vgs simulation is done for both analytical extracted parameter and default parameter values. The effect of Vgs is alone studied on thre- shold voltage by considering Vds=50mV and results show that Vds does not have impact on the threshold voltage.
The Fig. 1 shows the Ids-Vgs characteristics for large size de-
vices. We have chosen channel length of the device to be
1000nm and the channel width of the device to be 4000nm. We
infer that the extracted parameter values have lead to a lesser
threshold voltage compared with the default parameter values
for both the Vbs values plotted.
Fig. 2 illustrates the transfer characteristics obtained through
SPICE simulation for short channel devices, having channel
length of the device to be 300nm and the channel width of the
device to be 4000nm.
Fig. 2 Short Channel devices (Leff=300nm, Weff=4000nm) Ids- Vgs characteristics, Vbs parameter, VDS=50mV
From the Fig. 2 we infer that the extracted parameter values have lead to a lesser threshold voltage compared with the de- fault parameter values for the plotted Vbs.
Fig. 3 Narrow Width devices (Leff=1000nm, Weff=600nm) Ids- Vgs characteristics, Vbs parameter, VDS=50mV
Fig. 3 shows the characteristics for narrow width devices that have channel length of the device to be 1000nm and the chan- nel width of the device to be 600nm. From Fig. 3 we infer that the extracted parameter values have lead to a lesser threshold voltage compared with the default parameter values for Vbs
=0V and the extracted parameter values have lead to a greater threshold voltage compared with the default parameter values for Vbs=-5V.
For clarity purpose, the simulation results only show the Ids- Vgs plots for Vbs=0V and Vbs=-5V. The (TABLE III) shows the threshold voltage obtained for all the range of the Vbs consi- dered (0V to -5V).
R © 2012
ww.ijser.org
The research paper published by IJSER journal is about Parameter Extraction for BSIMSOI4.3 MOSFET Model 5
ISSN 2229-5518
TABLE 3
Threshold Voltage for Various device geometries
Body bias vol- tage applied | Threshold voltage in V | |||||
Body bias vol- tage applied | Large size devices (large length and large width) | Short channel devices (short length and large width) | Narrow width devices (large length and short width) | |||
Body bias vol- tage applied | Calculated using parameter’s | Calculated using parameter’s | Calculated using parameter’s | |||
Body bias vol- tage applied | Default values | Extracted values | Default values | Extracted values | Default values | Extracted values |
Vbs=0V | 1.2 | 0.57 | 1.2 | 0.61 | 1.4 | 1.21 |
Vbs=-1V | 1.6 | 0.85 | 1.6 | 0.90 | 1.8 | 1.64 |
Vbs=-2V | 1.8 | 1.07 | 1.8 | 1.11 | 2.0 | 1.97 |
Vbs=-3V | 2.0 | 1.24 | 1.8 | 1.28 | 2.2 | 2.23 |
Vbs=-4V | 2.0 | 1.39 | 1.8 | 1.43 | 2.2 | 2.46 |
Vbs=-5V | 2.0 | 1.52 | 2.0 | 1.56 | 2.4 | 2.66 |
For large size devices, the threshold voltage from extracted parameters varies gradually with respect to Vbs, but default parameters have threshold voltage high and constant for Vbs greater than -3V. In short channel devices, the threshold vol- tage varies gradually with Vbs for the extracted parameters. The default parameters respond a higher and constant value of Vth for Vbs from -2V to -4V. Similarly the Vth response of the narrow width devices is studied. The studied results infer an inverse effect as that of short channel and long size devices.
The threshold voltage increases with increase in body bias voltage Vbs and it increases with the shrinking dimensions. The proportionality of increase is given accurately by the ex- tracted parameters. VT0, the threshold voltage at Vbs=0V, is obtained accurately from the extracted values. All the parame- ters have significant effect on Vth, which are also interdepen- dent. Default Vth values deviates a lot from the accurate re- sults, which clearly implies the significance of parameter ex- traction for circuit simulation and modeling of device.
[1] Suayb Yener , Hakan Kuntman, "Algorithm on the Extraction of BSIM MOSFET Model Parameters via Measurement data", IEEE,2010;55(1):233-
244
[2] Yiming Li, ―An automatic parameter extraction technique for advanced
CMOS device modeling using genetic algorithm‖, Microelectronic Engi- neering 2007;84 :260–272
[3] Dr. Wenwei Yang, Dr. Chung-Hsun Lin, Dr. Tanvir Hasan Morshed, Mr.
Darsen Lu, Prof. Ali Niknejad, Prof. Chenming Hu,"BSIMSOI4.3 MOSFET
Model Users Manual", 2009,UC Berkeley, Department of EECS.
[4] Dr. Xuemei(Jane) Xi, Dr. Jin He, Mr. Mohan Dunga, Prof. Ali Niknejad, Prof. Chenming Hu,"BSIM4.3.0 MOSFET Model Users Manual", 2003,UC Berkeley, Department of EECS.
[5] Bing J.Sheu, Donald L.Scharfetter, Ping-Keung Ko, Min-Chie Jeng, ―BSIM:
Berkeley Short-Channel IGFET Model for MOS Transistors‖, IEEE Journal
of Solid-State Circuits, 1987; 4: 22
[6] Shunpei Abe, Yoshiyasu Miyazawa, Yoshikata Nakajima, Tatsuro Hanajiri,
Toru Toyabe, Takuo Sugano, ―Suppression of DIBL in deca nano SOI MOSFET by controlling permittivity and thickness of BOX lay-
ers”, IEEE 2009;50:174-178
[7] Hyeokjae Lee, Hyunchul Nah, Jong-Ho Lee, Dae-Gwan Kang, Young June Park, Hong Shick Min, ―Analysis of body bias effect with PD-SOI for ana- log and RF applications‖, Solid-State Electronics 2002;46:1169-1176
[8] Rodrigo Picos, Oscar Calvo, Benjamin Iniguez, Eugeni Garcia Moreno,
Rodolfo Garcia, Magali Estrada, ― Optimized parameter extraction using fuzzy logic‖, Solid-State Electronics 2007;51:683-690
[9] R.A.Thakker, M.B.Patil, K.G.Anil, ―Parameter extraction for PSP MOSFET
model using hierarchical particle swarm optimization‖, Engineering appli- cations of Artificial Intelligence 2009;22 : 317-328
[10] Jatmiko E.Suseno, Munawar A.Riyadi, Razali Ismail, ―Short Channel Effect
of SOI Vertical Sidewall MOSFET‖, ICSE 2008 Proceedings, 2008.
[11] Arash Daghighi, Azam Asgari Khoshooie, ―A width dependent body vol-
tage model to obtain body resistance in PD SOI MOSFET technology‖, Sol- id-IEEE 2010;53 :1246–1251
[12] G.Hyvert, F.Calmon, T.Nguyen, A.Poncet, C.Plossu, ― Accurate electron mobility extraction in nMOSFETs by RF split CV‖, Microelectronics Engi- neering 2009;86:1665-1667
[13] Ortiz-Conde, A.D. Latorre Rey a, W. Liu b, W.-C. Chen c, H.-C. Lin c, J.J.
Liou b,d, J. Muci a,F.J. García-Sánchez ,"Parameter extraction in polysilicon
nanowire MOSFETs using new double integration-based procedure",Solid- State Electronics 2010;54(6):635–641
[14] Yang Tang, Li Zhang, Yan Wang,"Accurate small signal modeling and extraction of silicon MOSFET for RF IC application", Solid-State Electronics
2010; 54 (11):1312–1318
IJSER © 2012