A Novel Method for Transit Time Parameter Extraction Taking into Account the Coupling Between DC and AC Characteristics Dominique BEGE and Didier CELI STMicroelectronics, 850, rue jean Monnet F-38926 Cedex HICUM user s meeting September 30, 2001 Minneapolis, MN 1
Outline ˆ Introduction ˆ Extraction issues Classic extraction strategy DC-AC coupling through main characteristics ˆ A novel extraction strategy Static parameters correction Improvements on current gain and output characteristics ˆ Conclusion ˆ eferences 2
Introduction 7KH#$LP#I#WKLV#SUHVHQWDWLQ#LV To present a single device extraction strategy of the main HICUM model parameters. Î Main parameters are: #DC parameters used for the modelling of the transfer current #AC parameters like transit time parameters modelling both the DC and AC behaviour of the transistor proposed a new extraction strategy to solve this coupling between AC & DC parameters To present the results obtained using measurement on NPN transistor fabricated within a 0.25 µm SiGe BiCMOS technology of STMicroelectronics To present critical points we have encountered with this method 3
Classic Extraction Strategy [1] Depletion capacitances & series resistances 1 1 ----------- = f ---- 2πf I T c Characteristic Transfer current parameters c 10, Q p0, h jci assuming h jei =1 transit time at low bias τ 0, τ 0h, τ bvl critical current ICK ci0, Vlim, Vces, Vpt Base current parameters I EIS, I BEIS m EI =2, m BEI =1 transit time at high current τ Ef0, gτe, ahc, τ hcs 4
f T and Transit time 8e+10 3e-11 7e+10 2.5e-11 6e+10 5e+10 2e-11 V BC ft [Hz] 4e+10 tf [s] 1.5e-11 V BC 3e+10 1e-11 2e+10 1e+10 5e-12 0 0.0001 0.001 0.01 0.1 I C [A] 0 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 I C [A] V BC = -1.5, -1, -0.5, -0.25, 0, 0.25, 0.5 V ˆ The inaccuracy on cx and on the split of depletion capacitance, specially the Base-Emitter capacitance, affect strongly the fit of the f T peak and its low injection behaviour. ˆ Transit time extraction method have been tested from simulated data, the extracted set of parameters corresponds with the initial one validation of the proposed method. 5
Gain and Output Characteristics BETA(IC) name :NN111A128 IC(VCE) name :NN111A128 1.40e+02 1.20e+02 breakdown parameter not yet extracted 2.00e-02 BETA Meas. and Simu. 1.00e+02 8.00e+01 6.00e+01 4.00e+01 IC (A) Meas. and Simu. 1.50e-02 1.00e-02 5.00e-03 2.00e+01 0.00e+00 wrong estimation of Q p0 1e-09 1e-08 1e-07 1e-06 1e-05 0.0001 0.001 0.01 IC (A) VBC (V) V BC = 5.00E-01,-1.00E+00 =-1.5, -1, -0.5, -0.25, 0, step 0.25, = 0.5-5.00E-01 [V] 0.00e+00 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 VCE (V) IB I B = (A) 8e-6, = 8.00E-06, 1.6e-5, 3.2e-5, 3.20E-04 8e-5, 1.6e-4, step = 3.2e-4 6.24E-05 [A] ˆ The lack of accuracy on the onset of high injection effect comes from the inadequate evaluation of the effective knee current I Keff which can be approximated in this region by the ratio: Q p0 I Keff = --------- τ f0 Inadequate Q p0 gives a wrong value for I Keff, τ f0 coming from f T curves. 6
Transfer current parameters extraction issue ˆ The classic transfer current extraction must be review, specially the assumption h jei =1. c ˆ 10 Q p0 h jci This direct extraction provides the ratios --------, --------- and -------- which allow to define the saturation h jei h jei h jei c 10 h jei c 10 current I s = -------- --------- = --------- I s is independent of h jei. h jei Q p0 Q p0 V BC =0V I C (V BE ) characteristic V BEi c 10 ------------ V BCi exp exp ------------ V T V T I T = ------------------------------------------------------------------------------------- Q p0 + h jei Q + h JEi jci Q + Q JCi FT V BE =0.7V I C (V BC ) characteristic from V BEi exp ------------ c V 10 T Q p0 Q JEi = -------- --------------------------- --------- h jei I T h jei from Q p0 h jci --------- + Q + -------- Q h jei JEi h JCi jei = VBEi VBCi ------------- ------------- c VT VT 10 e e -------- ------------------------------------- h jei I T gives c 10 -------- = r1 h jei & Q p0 --------- = r2 h jei gives h jci -------- = r3 h jei 7
Static Correction from Transit time parameters tf0 = 2.039E-12 s V BC = 0.000E+00 BETA(IC) name :NN111A128 2.5e-11 1.60e+02 2e-11 1.40e+02 1/(2* /Symbol=p*f T ) [s] 1 ----------- [s] 2πf T 1.5e-11 1e-11 BETA Meas. and Simu. 1.20e+02 1.00e+02 8.00e+01 6.00e+01 20 50% 5e-12 4.00e+01 2.00e+01 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 1/I C [A-1] 0.00e+00 1e-10 1e-09 1e-08 1e-07 1e-06 1e-05 0.0001 0.001 0.01 IC (A) VBC (V) = 0.00E+00 τ 0 Q p0 = I Keff τ 0 I Keff h jei new = Q p0 --------- r2 c 10 new = h jei new r1 h jci new = h jei new r3 8
Forward current gain and output characteristics BETA(IC) name :NN111A128 IC(VCE) name :NN111A128 1.20e+02 1.00e+02 2.00e-02 BETA Meas. and Simu. 8.00e+01 6.00e+01 4.00e+01 IC (A) Meas. and Simu. 1.50e-02 1.00e-02 2.00e+01 Better estimation of Q p0 5.00e-03 1e-07 1e-06 1e-05 0.0001 0.001 0.01 IC (A) VBC (V) V= BC 5.00E-01,-1.00E+00 =-1.5, -1, -0.5, -0.25, 0, step 0.25, = 0.5-5.00E-01 [V] 0.00e+00 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 VCE (V) IB I B (A) = 8e-6, = 1.00E-07, 1.6e-5, 3.2e-5, 3.20E-04 8e-5, step 1.6e-4, = 3.2e-4 5.33E-05 [A] Avalanche breakdown parameters are not extracted. The fit in the circle range is better than the previous comparisons shown in slide 6 The I Keff current is arbitrary determined. The value of h jei can be discussed (physical point of view : h jei =3.5). 9
New proposed extraction strategy Depletion capacitances & series resistances Base current parameters [1] I EIS, I BEIS m EI =2, m BEI =1 Transfer current ratio c 10 ---------, Q h p0 jci ----------,--------- h jei h h jei jei Transit time at low bias [1], [2] τ 0, τ 0h, τ bvl Effective knee current from β(i C ) fall off critical current ICK [1], [2] ci0, Vlim, Vces, Vpt transit time at high current from [1], [2] τ Ef0, g τe, a hc, τ hcs NO Transfer current parameters c 10, Q p0, h jei, h jci ICH I C (V CE ) fit YES END 10
Conclusion We propose an improved extraction strategy to solve the coupling between DC and AC parameters (c 10, Q p0, h jei, h jci,τ 0,...) by estimating Q p0 at high density of currents from the τ 0 value and defining an effective knee current I Keff. A more accurate method to define the effective knee current must be developed. Criterium to be found in order to avoid extraction loops between f T and output characteristics. The physical meaning of h jei can be discussed but the following extraction strategy gives a suitable fit between theory and measurements for DC & AC characteristics. These comparisons could be certainly improved by a best evaluation of the serie resistance (cx) and of the split of the depletion capacitances. The scalability of the extracted parameters need to be verified. 11
EFEENCE [1] D. Berger, D. Céli, N. Gambetta, T. Burdeau: HICUM Parameters Extraction Methods, HICUM workshop, June 14/15 2001, Dresden. [2] B. Ardouin, et al.: Transit Time Parameters Extraction for the HICUM Bipolar Compact Model, Proceedings of the Bipolar Circuits and Technology Meeting, Minneapolis, 2001. 12