On pulsed RF measurements

Transcription

1 On pulsed RF measurements Mario Weiß, Sébastien Fregonese, Marco Santorelli, Amit Kumar Sahoo, Cristell Maneux and Thomas Zimmer Laboratoire IMS, CNRS - UMR 5218, Université de Bordeaux 1, Cours de la Liberation Talence Cedex, France

2 Motivation Improved SiGe HBT performance within the DOT5 new spectrum of sub-millimeter-wave applications [1] Higher performance due to down-scaling Issue: DC and S-par measurements are more affected by self-heating device temperature variation is generally unknown model parameter extraction difficult Solution: Pulsed DC and S-par measurements Pulse width smaller than thermal time constant of transistor no significant change in temperature during measurement Exploration of save operating area (SOA) [1] [7] E. Ojefors, J. Grzyb, Y. Zhao, B. Heinemann, B. Tillack, and U. R. Pfeiffer, "A 82GHz SiGe chipset for terahertz active imaging applications," in Solid-State Circuits Conference Digest of Technical Papers (ISSCC), 211 IEEE International. IEEE, Feb. 211, pp BipAK 212 Thomas Zimmer 2

3 1. Introduction 2. Experimental Setup 3. System Validation Outline 4. Results and Discussion 5. Extraction of C TH 6. Measurement vs simulation 7. Conclusions BipAK 212 Thomas Zimmer 3

4 ΔT [K] ΔT [K] VBE (V) VBE (V) 1. Introduction 1. Introduction Parameters Small duty-cycle High duty-cycle Duty cycle D Pulse width T w.8.8 Pulse period T p Tp D 1 Tw Generally:.1%<D<1% µ 4.µ 6.µ 8.µ 1.µ 12.µ Time (s) µ 2.µ 3.µ 4.µ 5.µ Time (s) Trade-off isothermal behavior and measurement accuracy µ 4.µ 6.µ 8.µ 1.µ 12.µ Time (s) µ 2.µ 3.µ 4.µ 5.µ Time (s) BipAK 212 Thomas Zimmer 4

5 2. Experimental Setup Pulsed measurement system characteristics: Pulse width Tw min 8ns DC Tw min 1ns RF Τ rise and T fall 2ns Frequency f [5MHz 45GHz] Duty-Cycle D [.1% - 5%] Current resolution [µa] 2. Experimental Setup BipAK 212 Thomas Zimmer 5

6 S11 pulsed (m/s) S22 pulsed (m/s) S11 ref (m/s) S22 ref (m/s) / / / / S11 pulsed (m/s) S22 pulsed (m/s) S11 ref (m/s) S22 ref (m/s) / / / / S11 pulsed (m/s) S22 pulsed (m/s) S11 ref (m/s) S22 ref (m/s) / / / / 3. System Validation 3. System Validation Measurement of an OPEN (4.5fF) with f [5MHz 45GHz] Duty cycle=.1% Duty cycle= 1% Duty cycle= 5% S 11pulsed S 22pulsed S 11ref. S 22ref. S 11pulsed S 22pulsed S 11ref. S 22ref. S 11pulsed S 22pulsed S 11ref. S 22ref. freq Using D of 5% conventional and pulsed S-parameter measurement are in very good agreement freq freq BipAK 212 Thomas Zimmer 6

7 I C [ma] f T [GHz] 3. System Validation Transistor measurement of a SiGe HBT Technology process of STMicroelectronics: f T =23GHz f max =29GHz Device geometry: W E =.23 μm L E = 5. μm 3 25 Standard Pulsed 1µ 25 2 Standard Pulsed 1µs V CE =1.5V 2 V CE =.5V V CE =1.5V V CE =.5V V BE [V] V BE [V] Pulsed I(V) and pulsed RF characteristics in good agreement with conventional measurements BipAK 212 Thomas Zimmer 7

8 I C [A] I C [A] 5. Results and Discussion 4. Results and Discussion Pulsed I(V) for SiGe HBT with W E =.27 μm L E = 5. μm.3 8ns 1ns 15ns 2ns 3ns 5ns 1µs 5µs 1µs.3 VCE=1.5V VCE=1.2V VCE=1V VCE=.5V DC V BE =.95V V CE [V] Collector current I C vs. collector emitter voltage V CE t [ns] Collector current I C vs. pulse width t BipAK 212 Thomas Zimmer 8

9 I C [A] I C [A] 5. Results and Discussion Pulsed I(V) for SiGe HBT with W E =.84 μm L E = 5. μm ns 1ns 12ns 15ns 2ns 3ns 5ns 1µs 1µs 5µs VCE=1.5V VCE=1.2V VCE=1V VCE=.5V.4 DC V CE [V] Collector current I C vs. collector emitter voltage V CE t [ns] Collector current I C vs. pulse width t BipAK 212 Thomas Zimmer 9

10 I C [A] f T [GHz] 5. Results and Discussion Pulsed I(V) and pulsed RF for SiGe HBT with W E =.84 μm L E = 5. μm ns 3ns 1µs ns 3ns 1µs V CE = 1.5 V V CE = 1.5 V 2 15 V CE =.5 V V CE =.5 V V BE [V] Collector current I C vs. base emitter voltage V BE V BE [V] Transit frequency f T vs. base emitter voltage V BE BipAK 212 Thomas Zimmer 1

11 Ic(A) 3.m Isothermal measurement? NBB4TAB3 2.m 1.m Pulsed meas. 1us Pulsed meas. 3ns Pulsed meas. 15ns Pulsed meas. 1ns Sim_R TH = K/W Sim_R TH =42 K/W. 7.m 8.m 9.m 1. V BE (V) For a pulse width of 1ns self-heating has been significantly reduced BipAK 212 Thomas Zimmer 11

12 Magnitude (Y 12 ) 5. Extraction of C TH A) C TH extraction from low freq. Y-parameters Dynamic Self-heating Measured Extracted Without AC self-heating Low frequency S-parameter measurements : 1 khz 3 GHz, V CE = 1.5 V V BE =.95 V DUT : L E *W E = 9.88*.15 µm 2 Frequency (Hz) Normalized Thermal Impedance Z TH DC AC y12 y12 y y I V y V y I V y V y AC DC DC C BE 12 CE 22 C BE 12 CE 22 BipAK 212 Thomas Zimmer 12

13 Phase (Z TH ) Magnitude (Z TH ) L E Frequency domain magnitude and phase of thermal impedance for different geometry of transistor Emitter Length (L E ) m 4.88 m 9.88 m m ~ -1dB/dec L E Frequency (Hz) Symbols measured Lines extracted with Recursive electro-thermal network Emitter Length (L E ) m 4.88 m 9.88 m m Frequency (Hz) BipAK 212 Thomas Zimmer 13

14 Magnitude (Y 12 ) Magnitude (Y 22 ) Measured Extracted with - Single R TH -C TH Network Recursive Network Without AC self-heating Measured Extracted with - Single R TH -C TH Network Recursive Network Without AC self-heating AC+DC Self-heating Frequency (Hz) DC Selfheating AC+DC Self-heating Frequency (Hz) DC Selfheating Comparison of thermal modeling of y-parameters with HiCuM compact model simulation connecting the electro-thermal network at temperature node: 1. Single R-C network, 2. Recursive network and 3. only with Thermal resistance BipAK 212 Thomas Zimmer 14

15 I C (ma) I C (ma) B) From Pulsed I(V) Measurements Comparison between measurements and HiCuM L2 simulation. (A E = 15 x.27 µm 2 ) 12 9 Measured Extracted with single pole network Short time Long time Average C Steady state current TH 12 9 Measured Extracted with recursive network Extracted without self-heating Steady state current 6 3 Transient variation due to self heating 6 3 Transient variation due to self heating Isothermal current Isothermal current Time (s) Thermal modeling with single pole network Time (s) Thermal modeling with recursive network BipAK 212 Thomas Zimmer 15

16 Ic simulation Ic measurement 1ns [E-3] Ic simulation Ic measurement 3ns [E-3] Ic simulation Ic measurement 1ns [E-3] Measurement vs simulation (1/3) PO PO Plot B5TAP/NBB5TAP/sim/Ic_Vb Plot B5TAP/NBB5TAP/sim/Ic_Vb Plot B5TAP/NBB5TAP/sim/Ic_Vb 4 A 4 A vb.m [E+] X Y vb.m [E+] X Y vb.m [E+] VCE=.5V & 1.5V BipAK 212 Thomas Zimmer 16

17 Ic sim. 1ns Ic meas. 1ns Ic sim. 3ns Ic meas. 3ns Ic sim. 15ns Ic meas. 15ns Ic sim. 1ns Ic meas. 1ns [LOG] Ic sim. 1ns Ic meas. 1ns Ic sim. 3ns Ic meas. 3ns Ic sim. 15ns Ic meas. 15ns Ic sim. 1ns Ic meas. 1ns [E-3] Measurement vs simulation (2/3) PO Plot B5TAP/NBB5TAP/sim1/Ic_Vb 1E-1 A Plot B5TAP/NBB5TAP/sim1/Ic_Vb 4 PO A 1E-2 3 VCE=1.5 V 1E E-4 vb.m [E+] X Y vb.m [E+] X Y BipAK 212 Thomas Zimmer 17

18 I CE [A] Measurement vs simulation (3/3) 3.m 25.m 2.m 15.m 1ns S 15ns S 3ns S 1us S 1ns M 15ns M 3ns M 1us M 1.m 5.m V BE [V] V CE BipAK 212 Thomas Zimmer 18

19 6. Conclusions 6. Conclusions State of the Art Pulsed I(V) pulsed RF System validated with conventional measurements Self-heating can be significantly decreased and SOA expanded Pulsed measurements help separating thermal effects from electrical effects (e.g. Self heating Avalanche) Extraction procedure to achieve physical model parameters is much easier BipAK 212 Thomas Zimmer 19

20 Acknowledgment The Authors would like to thank STMicroelectronics for supplying the wafers as well as RF2THZ project. BipAK Mario 212 Weiß Thomas Zimmer 2