Secure Mobile Solutions 06162004 Arasch Lagies Jrg Berkner

Secure Mobile Solutions 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 1 DC Temperature Behavior of Hicum Arasch Lagies (IFX), Jörg Berkner (IFX), Ramana M. Malladi (IBM), Kim M. Newton (IBM), Scott M. Parker (IBM) June 15/16, 2004, Bordeaux, France n Temperature Dependent Modeling Problems n Reason for the Modeling Problems n Solution for the Modeling Problems n Results Never stop thinking.

DC Measurement Results and Behavior at T=-40°C, 25°C and 125°C n Sketch of the Measurement Setup: Common-Emitter c Ib +V - ce b e Ib [u. A] = 1, 2, 5, 8, 10, 15, 20 Vce = 0 V to 3 V n 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 2 T↓ T↓ Results: – Slope of saturation region rises with decreasing T. – Active region current rises with decreasing T. Measurements at T=-40°C (blue), 25°C (black), 125°C(red)

DC Simulation Results with Hicum in Spectre at T=40°C, 25°C and 125°C n Sketch of the Simulation Setup: Common-Emitter c Ib +V - ce b e Ib [u. A] = 1, 2, 5, 8, 10, 15, 20 Vce = 0 V to 3 V n 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 3 Results: – The saturation region is shifted to higher VCE for T < 0°C. – The active region current rises with decreasing T. Simulations at T=-40°C (blue), 25°C (black), 125°C(red)

Reverse Gummel Measurements at T=-40°C, 25°C and 125°C n In reverse Gummel IS cannot be fitted for all temperatures. T↑ • Measurements show: Ic, Ie, Ib and Is are all T-dependent. • This is true for reverse Gummel and in forward active mode. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 4 • How can the Is temperature dependence be fitted? Measurements at T=-40°C (blue), 25°C (black), 125°C(red)

Reason for Fitting Difficulties n The saturation current of the parasitic pnp substrate transistor ITSS is temperature-independent modeled. n If ITSS is fitted for T=25°C the current through the pnp substrate transistor becomes too big for T < 0°C, compared to the base current of the npn transistor. n This leads to 1. The shift for the DC output curves for T< 0°C toward higher VCE. 2. Bad fitting of IS in the reverse Gummel and in forward active mode over temperature. 3. Bad convergence behavior for T< 0°C. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 5

Solution for the Fitting Difficulties (1) n For good results ITSS shows an exponential behavior with the temperature. n By using a similar function as used for the saturation current of the parasitic pnp transistor, ISP(T) in Vbic, this problem can be solved. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 6 n Here is

Solution for the Fitting Difficulties (2) n The exponential function for ITSS(T) leads to optimal agreement with measurements (ITSS’ = ITSS/area): ITSS‘(nom) [A]= XIS= EAP [e. V]= 6. 67 E-10 3 1. 1 k [JK-1] = 1. 38 E-23 q [C]= 1. 60 E-19 NFP= 0. 87 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 7 Fitted ITSS’ values from measurements compared to the ITSS’(T) function

Solution for the Fitting Difficulties (3) n Some possibilities for the introduction of ITSS(T): 1. If a scaling file is available ITSS(T) can be implemented there (work around). 2. Generally ITSS(T) should be implemented in the model code. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 8

Measurement vs. DC Simulation Results with Hicum in Spectre at T=-40°C, 25°C and 125°C using ITSS = ITSS(T) n With ITSS=ITSS(T) there is no shift of the saturation region to higher V CE. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 9 Meas. (dash-dotted) vs. sim. (solid) at T=-40°C. Meas. (dash-dotted) vs. sim. (solid) at T=25°C. Meas. (dash-dotted) vs. sim. (solid) at T=125°C.

Measurement vs. Reverse Gummel Results with Hicum in Spectre at T=-40°C, 25°C and 125°C using ITSS = ITSS(T) n With ITSS=ITSS(T) the low current region of IS in the reverse Gummel plot can be fitted accurately. Meas. (dash-dotted) vs. sim. (solid) at T=25°C. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 10 Meas. (dash-dotted) vs. sim. (solid) at T=-40°C. Meas. (dash-dotted) vs. sim. (solid) at T=125°C.

Conclusions n The temperature independence of ITSS for the parasitic pnp substrate transistor leads for some temperatures to a strong current imbalance compared to the base current of the npn main transistor. n This imbalance leads to – a shift of the saturation region in the DC output curves for T < 0°C, – a bad fitting of IS in the reverse Gummel and in forward active mode over T and – difficulties in the simulator convergence for some T. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 11 n The problem can be solved by using a T-dependent function for ITSS similar to the saturation current ISP(T), as used in Vbic. n ITSS(T) can be implemented in a scaling file or better in the model code itself.

Extra Foil (1) Reverse Gummel Measurement and Simulation with Hicum in Spectre at T=-40°C, 25°C and 125°C n In reverse Gummel IS cannot be fitted for all temperatures. 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 12 Measurements at T=-40°C (blue), 25°C (black), 125°C(red) Simulations at T=-40°C (blue), 25°C (black), 125°C(red)

Extra Foil (2) IS Reverse Gummel Measurement and Simulation with Hicum in Spectre 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 13 n Measurement and simulation in a technology of Infineon. n In reverse Gummel IS cannot be fitted for all temperatures. n Here the fitting was done for T=25°C (green). – For T=125°C (blue) the simulation deviates significantly. – For T=-25°C (red) no simulator convergence was achieved. Measurement solid lines, simulation dashed lines. For T = -25°C (red), 25°C (green), 125°C (blue)

Extra Foil (3) DC Output Characteristics Simulation with Hicum in ADS n Here the same parameter set was used as in Spectre. n The default parameter set gives similar results. n For T < 0°C a shift of the saturation region to higher Vce can be observed. c Ib 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 14 b e +V - ce Ib [u. A] = 1, 2, 5, 8, 10, 15 Vce = 0 V to 3 V Simulation plot of ADS. For T = -40°C (blue), 25°C (black), 125°C (red).

Extra Foil (4) DC Output Characteristics Simulation with Hicum in ADS Temperature Sweep n 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 15 Here the same parameter set was used as in Spectre. Simulation plot of ADS with Ib=10 m. A. Temperature sweep with T = -40°C to 20°C in steps of 10°C.

Extra Foil (5) Forward Gummel Characteristic Hicum Simulation with ITSS=const. , ITSS=ITSS(T), compared with measurements 06/16/2004 Arasch Lagies Jörg Berkner Ramana M. Malladi Kim M. Newton, Scott M. Parker Page 16 n In forward Gummel the change from ITSS=const. to ITSS=ITSS(T) leads to slight changes at T<0°C for higher VBE. n This comes closer to the measured values. Simulations and Measurements at T = -40°C (blue), 25°C (black), 125°C (red). Forward Gummel results with ITSS=const. (solid-symbol) and ITSS=ITSS(T). Included measurements (dashdotted lines).