UNIVERSITY OF CALIFORNIA, BERKELEY BSIM4.3.0 Model Enhancements and Improvements Relative to BSIM4.2.1 Xuemei (Jane) Xi, Jin He, Mohan Dunga, Ali Niknejad, Chenming Hu University of California, Berkeley 1 UNIVERSITY OF CALIFORNIA, BERKELEY OUTLINE New Features of BSIM4.3.0 beta’ release vStress effect model vNew temperature model vHolistic noise model enhancement vUnified current saturation model q Velocity saturation q Velocity overshoot q Source injection thermal velocity limit vNew document for multi-layer gate tunneling vForward body bias 2 UNIVERSITY OF CALIFORNIA, BERKELEY Model for Isolation-induced Stress Effects L Trench isolation edge SB SA W LOD SD Instance parameters added: SA, SB, SD SD is neighbour finger distance which is constant throughout all the fingers. Stress effect calculation only if: 1) both SA and SB are given and are larger than 0 for finger number NF=1; 2) SA, SB and SD are all given and are larger than 0 for NF >1 Intermediate geometry definitions : LOD = SA + SB + NF ⋅ L + ( NF − 1 ) ⋅ SD 3 UNIVERSITY OF CALIFORNIA, BERKELEY Mobility Model With STI Stress Define : ρ µeff = ∆µ eff / µ effo = ( µ eff − µ effo ) / µ effo = So, µ eff µ effo µ eff µ effo −1 = 1 + ρ µeff (relative mobility change due to stress) (Vth insensitive to Lod, SA and/or SB) 4 UNIVERSITY OF CALIFORNIA, BERKELEY Stress Effects Model-1/LOD Model v Simple stress distribution function: 1/(SA+L/2), 1/(SB+L/2) v ρ µeff ρ µeff = Inv _ sa = expression with LOD, L, W, and T dependence ku0 ⋅ ( Inv _ sa + Inv _ sb ) Kstress _ u0 1 SA + 0 .5 ⋅ Ldrawn Inv _ sb = 1 SB + 0.5 ⋅ Ldrawn LKU0 WKU0 Kstress _ u 0 = 1 + + LLODKU0 ( W drawn + XW + WLOD )WLODKU0 ( Ldrawn + XL ) + ( Ldrawn + XL ) LLODKU0 PKU0 ⋅ ( Wdrawn + XW + WLOD )WLODKU0 Temperatur e × 1 + TKU 0 ⋅ − 1 TNOM v All data can be fitted well with only one set of parameters (ie. Global model for LOD effect) and do not need extra binning parameters if binning is desired. v For multi-finger device: Inv _ sa = 1 NF NF −1 1 + i ⋅ ( SD + Ldrawn ) drawn ∑ SA + 0. 5 ⋅ L i= 0 Inv _ sa = 1 NF NF −1 ∑ SB + 0.5 ⋅ L i =0 drawn 1 + i ⋅ ( SD + Ldrawn ) v For irregular LOD device: n 1 sw i 1 =∑ ⋅ SAeff + 0.5 ⋅ Ldrawn i=1 Wdrawn sai + 0.5 ⋅ Ldrawn n 1 sw i 1 =∑ ⋅ SBeff + 0.5 ⋅ Ldrawn i =1 Wdrawn sbi + 0 .5 ⋅ Ldrawn 5 UNIVERSITY OF CALIFORNIA, BERKELEY Stress Effect µeff , υ sat Model µ eff = υ sat = 1 + ρ µeff ( SA , SB ) 1 + ρ µeff ( SAref , SB ref ) µeffo 1 + K ⋅ ρ µeff ( SA , SB ) 1 + K ⋅ ρ µeff ( SAref , SB ref ) υ sato Where µ effo , υ sato are low field mobility, saturation velocity at SAref, SBref 6 UNIVERSITY OF CALIFORNIA, BERKELEY Stress Effect Model to VTH0, K2, ETA0 LKVTH0 WKVTH0 + ( Ldrawn + XL )LLODKVTH ( Wdrawn + XW + WLOD )WLODKVTH PKVTH0 + ( Ldrawn + XL )LLODKVTH ⋅ ( Wdrawn + XW + WLOD )WLODKVTH Kstress _ vth0 = 1 + VTH 0 = VTH 0original + KVTH0 ⋅ (Inv _ sa + Inv _ sb − Inv _ saref − Inv _ sbref ) Kstress_vt h0 STK2 ⋅ (Inv _ sa + Inv _ sb − Inv _ sa ref − Inv _ sbref ) Kstress_vt h0LODK2 STETA0 ETA 0 = ETA 0original + ⋅ (Inv _ sa + Inv _ sb − Inv _ sa ref − Inv _ sbref ) Kstress_vt h0LODETA0 K 2 = K 2original + 7 UNIVERSITY OF CALIFORNIA, BERKELEY Stress Effect Model Verification Drain current relative change (%) with SAref=5µm SA=SB 8 UNIVERSITY OF CALIFORNIA, BERKELEY Stress Effect Model Verification 9 UNIVERSITY OF CALIFORNIA, BERKELEY Drain current relative change (%) with SAref=5µm Stress Effect Model Verification 10 UNIVERSITY OF CALIFORNIA, BERKELEY Temperature Model Enhancement Temperature mode TEMPMOD created: qTEMPMOD = 0: current model with VFB enhancement qTEMPMOD = 1: New format for vsat, prt, ua, ub, uc: PARAM ( T ) = PARAM ( TNOM ) ⋅ [1 + TEMP _ COEFF ⋅ (T − TNOM )] 11 UNIVERSITY OF CALIFORNIA, BERKELEY Holistic Thermal Noise Model Enhancement Refer to Chapter 9 of BSIM4 manual θ tnoi = RNOIB ⋅ 1 + TNOIB ⋅ Leff V ⋅ gsteff E sat Leff 2 = RNOIA ⋅ 1 + TNOIA ⋅ Leff Vgsteff ⋅ E sat Leff 2 β tnoi (9.2.5) (9.2.6) Default RNOIA=0.577; RNOIB=0.37 12 UNIVERSITY OF CALIFORNIA, BERKELEY Unified Current Saturation -Velocity Overshoot Model Price’s approximation to HD model: λ ∂E y ∂n J = qn µE y ( 1 + ) + qD µE y ∂x ∂x Approximate solution of Price’s equation yields unified current expression that includes velocity saturation and velocity overshoot: I DS 0 I DS , HD = V dseff 1+ OV Leff E sat where E OV sat V ds − Vdseff 1 + Esat ⋅ litl LAMBDA = E sat 1 + ⋅ Leff ⋅ µ eff V ds − V dseff 1 + Esat ⋅ litl 2 − 1 2 + 1 Vdseff I DS 0 = I DS ( BSIM 4.2.1 ) ⋅ 1 + L E eff sat 13 UNIVERSITY OF CALIFORNIA, BERKELEY Unified Current Saturation: -Source–end Velocity Limit and Quasi-Ballistic Transport v sHD = HD transport source carrier velocity: v sBT = Ballistic transport source carrier velocity: r= where VTL: thermal velocity, Leff I DS , HD Wq s 1− r VTL 1+ r XN ⋅ Leff + LC XN ≥ 3.0 Unified current expression with velocity saturation, velocity overshoot and source velocity limit: I DS = [1+ (v I DS ,HD / v sBT ) 2 MM sHD ] 1 / 2 MM 14 UNIVERSITY OF CALIFORNIA, BERKELEY Direct Tunneling through MultipleLayer Gate Stacks v Gate Current modeled as JG = Q INV ⋅ fIMP ⋅ T T ∝ exp( −αt oxe ) v For two layer case T ∝ exp( −α new t oxe ) v For a single layer α double = α 1 ⋅ f + α 2 ⋅ ( 1 − f ) + f ⋅ ( 1 − f ) ⋅ V ox 3h where K 1 ⋅ qm 1 − K 2 ⋅ qm 2 2φ B 1 2φ B 2 α is the tunneling attenuation coefficient already modeled in BSIM4, f = Toxe1 / Toxe v Stands for multiple layers(N≥2) as well. v Using new tunneling attenuation coefficient and interpreted with tunneling equations in BSIM4, BSIM4 is now capable of modeling multi-layer gate tunneling. 15 UNIVERSITY OF CALIFORNIA, BERKELEY Verification v Verified with data of existing gate stack of HfO2 and silicon oxynitride. v Very good fit observed using BSIM model. v BSIM4 direct tunneling equation thus models the multi-layer case. Gate 1st layer 2nd layer 16 UNIVERSITY OF CALIFORNIA, BERKELEY Forward Body Bias To ensure a good model behavior of body effect, body bias is usually bounded between (Vbsc, and φs0 where φ s0 = 0.95 φs ). BSIM4.2.1 already has the smooth function for Vbs low bound. Following is the upper bound smooth function: ( ) 2 ' ' Vbseff = 0.95Φs − 0.5 0.95Φs −Vbseff − δ1 + 0.95Φs − Vbseff − δ1 + 4δ1 .0.95Φs Where: Vbseff = Vbc + 0. 5 ⋅ (Vbs − Vbc − δ1 ) + (Vbs − Vbc − δ 1 )2 − 4δ1 ⋅Vbc Is the low bound smooth function. d 1 = 0.001V, and Vbc is the maximum allowable Vbs and found from dVth/dVbs= 0 to be K 12 Vbc = 0 .9 Φ s − 4 K 22 17 UNIVERSITY OF CALIFORNIA, BERKELEY Gate Current Partition Bugfix From Original: Igcs = Igc ⋅ and Igcd = Igc ⋅ To: Igcs = Igc ⋅ Igcd = Igc ⋅ PIGCD ⋅ Vds + exp(− PIGCD ⋅Vds ) −1 + 1.0e − 4 2 PIGCD 2 ⋅ Vds + 2.0e − 4 1 − (PIGCD ⋅ Vds + 1)⋅ exp(− PIGCD ⋅Vds ) + 1.0e − 4 2 PIGCD 2 ⋅Vds + 2.0e − 4 PIGCD ⋅Vdseff + exp(− PIGCD ⋅Vdseff ) − 1 + 1.0e − 4 PIGCD 2 ⋅Vdseff + 2.0e − 4 2 1 − (PIGCD ⋅Vdseff + 1)⋅ exp(− PIGCD ⋅ Vdseff )+ 1.0e − 4 PIGCD 2 ⋅ Vdseff + 2.0e − 4 2 18 UNIVERSITY OF CALIFORNIA, BERKELEY Gate Current Partition Bugfix -11 2.0x10 Igc with bugfix Original Igc -11 1.8x10 -11 IGC(A) 1.6x10 -11 1.4x10 -11 1.2x10 VDS increase -11 1.0x10 -12 8.0x10 -12 6.0x10 1.00 1.02 1.04 1.06 1.08 1.10 1.12 VGS(V) Effect of gate current bug fix Comparison with experimental data 19 UNIVERSITY OF CALIFORNIA, BERKELEY New Parameters in BSIM4.3.0 -Stress Effect Parameter Name Description Default Value Binnab le? Note SA INSTANCE parameter: Distance between OD edge to Poly from one side 0.0 If not given or (≤0.0), stress effect will be turned off SB INSTANCE parameter: Distance between OD edge to Poly from the other side 0.0 If not given or (≤0.0), stress effect will be turned off SD INSTANCE parameter: Distance between neighbour fingers 0.0 for NF >1: if not given or (≤0.0), stress effect will be turned off saref Reference distance between OD edge to poly of one side Reference distance between OD edge to poly of the other side 1.E-06[m] no >0.0 1.E-06[m] no >0.0 wlod Width parameter for stress effect 0.0 [m] no ku0 Mobility degradation/enhancement coefficient for stress effect 0.0 [m] no kvsat Saturation velocity degradation/enhancement parameter for stress effect 0.0[m] no tku0 Temperature coefficient of ku0 0.0 no lku0 Length dependence of ku0 0.0 [m llodku0] no sbref -1 ≤ kvsat ≤ 1 20 UNIVERSITY OF CALIFORNIA, BERKELEY New Parameters in BSIM4.3.0 -Stress Effect Parameter Name Description Default Value Binnable ? Note wku0 Width dependence of ku0 0.0 [m wlodku0] no pku0 Cross-term dependence of ku0 0.0[m llodku0+wlodku0] no llodku0 Length parameter for u0 stress effect 0.0 no >0 wlodkuo Width parameter for u0 stress effect 0.0 no >0 kvth0 Threshold shift parameter for stress effect 0.0[V*m] no lkvth0 Length dependence of kvth0 0.0[V*mllodku0] no wkvth0 Width dependence of kvth0 0.0[V*mwlodku0] no pkvth0 Cross-term dependence of kvth0 0.0[V*mllodku0+wlodku0] no llodvth Length parameter for Vth stress effect 0.0 no >0 wlodvth Width parameter for Vth stress effect 0.0 no >0 stk2 K2 shift factor related to Vth0 change 0.0[m] no lodk2 K2 shift modification factor for stress effect 1.0 no steta0 eta0 shift factor related to Vth0 change 0.0[m] no lodeta0 eta0 shift modification factor for stress effect 1.0 no >0 >0 21 UNIVERSITY OF CALIFORNIA, BERKELEY New Model Parameters in BSIM4.3.0 -Unified Current Saturation Parameter Name Description Default Value Binnable ? Note LAMBDA Velocity overshoot coefficient 0.0 yes If not given or (≤0.0), velocity overshoot will be turned off VTL Thermal velocity 2.0e5 [m/s] yes LC Velocity back scattering coefficient 0.0[m] no If not given or (≤0.0), source end thermal velocity limit will be turned off ~5e-9(m) at room temperature XN Velocity back scattering coefficient 3.0 yes Temperature Model TEMPMOD Temperature mode selector 0 no If=0, original model will be used If=1, new format will be used Holistic Thermal Noise RNOIA Thermal noise coefficient 0.577 no RNOIB Thermal noise coefficient 0.37 no 22