Practical Considerations and Solutions for Temperature-Dependent S-Parameter Measurement for Accurate Parameter Extraction of

Practical Considerations and Solutions for Temperature-Dependent S-Parameter Measurement for Accurate Parameter Extraction of Advanced RF Devices Gavin Fisher, Application Engineer Andrej Rumiantsev, Product Marketing Manager

Agenda Why on-wafer thermal? Investigation background RF characterisation systems Mechanical effects of temperature transition probe and chuck growth Measured electrical effect of thermal transition Suggested practical calibration approach with real life data Effect of load resistance variation How to use WinCal XE to calibrate, de-embed and measurement data Conclusion

Why On-Wafer Over-Temperature Test? Devices are increasingly temperature dependent RF device characterization and modeling need S-parameters at several temperature and bias points Multiple temps between -40 and 125 C common ITRS predicts further extension of the temperature range "Radio frequency and analog/mixed-signal technologies for wireless communications," International Technology Roadmap for Semiconductors, ITRS, p. 36, 2011.

Example of FoM: f T (T) and f MAX (T) P. Chevalier, N. Zerounian, B. Barbalat, et al., "On the use of cryogenic measurements to investigate the potential of Si/SiGe:C HBTs for terahertz operation," in Bipolar/BiCMOS Circuits and Technology Meeting, 2007. BCTM '07. IEEE, 2007, pp. 26-29.

Motivation: Three Contradictions Temperature variation typically requires re-calibration at every temperature Calibration standards are temperature dependent Calibration kit (ISS) must be treated appropriately* Continuously increasing demand for higher characterization frequencies (e.g. mmw range) and measurement accuracy New challenges hard to deal with "High-frequency, over-temperature measurements and modeling," in Application Note Beaverton, OR, USA: Cascade Microtech, Inc. A. Rumiantsev and R. Doerner, Verification of wafer-level calibration accuracy at high temperatures in ARFTG Microwave Measurements Conference-Spring, 71st, 2008, pp. 103-106.

Agenda Why on-wafer thermal? Investigation background RF thermal test issues Mechanical effects of temperature transition probe and chuck growth Measured electrical effect of thermal transition Suggested practical calibration approach with real life data Effect of load resistance variation How to use WinCal XE to calibrate, de-embed and measurement data Conclusion

Intention of Investigation Re-visit application note of with modern equipment and wider measurement range

Calibration / Measurement Issues and Solutions Problem Solution RF Standards temperature dependency Probes misplacement w.r.t standards on transition Probe characteristics vary during the calibration process When is the system stable Maintain ambient or measure load resistance *) Re-adjust Calibration duration to be kept as short as possible. Use of WinCal stability measurements automated reports possible *) A. Rumiantsev, G. Fisher, R. Doerner. Sensitivity Analysis of Wafer-Level over Temperature RF Calibration, to be presented ARFTG-80th

Typical on Wafer Test System - Open

Typical on Wafer Test System - Closed PNA-X vector network analyzer from Agilent Probe positioners Probe station with -60 C to 300 C thermal system Calibration software SMU evue digital vision system In a temperature controlled lab

Engineering on Wafer Test System...

Typical on Wafer Test System 110 GHz Infinity probes 104-783 W Band ISS 1 mm 24 cm Cables Anti-moding absorber Torque wrench 13

Typical on Wafer Test System Dry, frost-free environment Auxiliary chucks Roll-out chuck Stable repeatable platen TopHat 14

Mechanical Effects of Growth Probes grow / retract with temperature chuck in Z and for E/W orientation in X Some movement in Y but comparatively minimal For significant thermal changes evaluate theta also

Mechanical Effects of Growth Initial contact set to 19370 at ambient Initial separation 474 um

Mechanical Effects of Growth Position of probes not varied on positioner Chuck temperature -40 C

Mechanical Effects of Growth Probe contact remade by adjusting chuck position only Probes marks are now 550 um apart 76 um delta Chuck needs to be raised 37 um higher

Temperature / Electrical Evaluation Work Probe equipped with K thermocouples and data logger Sensors attached to probe body, connector and mount

System Stability Instrumentation itself must be stable Use math / memory at selected IF bandwidth Temperature control essential For Agilent PNA-X Error-corrected range 23 C ±3 C with less than 1 C deviation from calibration Thermal system transitioning can effect room temperature

System Stability System stability can also be measured by conducting a cal and using a the WinCal monitoring function or by sequentially conducting open re-measurements Example below takes a monitoring measurement every 5 minutes

System Stability Controlled 27 C System was stable for 1 hour at 27 degrees

Measured Electrical Effect of Thermal Transition Following calibration another method used WinCal sequencing can be used to take repetitive measurements in normal measurement window Of primary interest is to be able to see end of change to determine if the system is stable

Measured Electrical Effect of Thermal Transition -40 C to -20 C

Measured Electrical Effect of Thermal Transition WinCal is also capable of showing differences from a particular trace Similar to monitoring measurements

Measured Electrical Effect of Thermal Transition -40 C to -20 C

Measured Electrical Effect of Thermal Transition % 8 7 6 5 4 3 2 1 0 Difference in S11 at 95 GHz with Time During -40 C to -20 C Transition, % Difference in S11 at 95 GHz 0 10 20 30 40 50 60 Time (Minutes)

Measured Electrical Effect of Thermal Transition -40 C to 25 C

AR2 Temperature Variation S11 with Chuck / Probe -40 C to 25 C

Slide 31 AR2 It will be nice if you could make a better graph out of here: font size 24pt, better readable... Andrej Rumiantsev, 10/24/2012

Transition +27 C to +75 C S11 Variation

AR3 Transition +27 C to +75 C S11 Variation

Slide 33 AR3 It will be nice if you could make a better graph out of here: font size 24pt, better readable... Andrej Rumiantsev, 10/24/2012

Variation in Phase +25 C to +75 C S11, Degrees

AR4 Phase Change +27 C to +75 C Transition

Slide 35 AR4 It will be nice if you could make a better graph out of here: font size 24pt, better readable... Andrej Rumiantsev, 10/24/2012

Variation in ISS Temperature AR5

Slide 36 AR5 It will be nice if you could make a better graph out of here: font size 24pt, better readable... Andrej Rumiantsev, 10/24/2012

AR6 Variation in ISS Temperature

Slide 37 AR6 It will be nice if you could make a better graph out of here: font size 24pt, better readable... Andrej Rumiantsev, 10/24/2012

jn1 Temperature Variation of Probe Body / Mount Calibrations 27 C 0 C -20 C -40 C

Slide 38 jn1 Andrej Rumiantsev 24-Oct-12 It will be nice if you could make a better graph out of here: font size 24pt, better readable... jnakaya, 10/31/2012

Thermal transient 40 C Calibration Calibration duration approximately 2 min, 32 frequency points, 100 Hz IF bandwidth

Over-Temperature Calibration Approach

Device Measurement Procedure Reduced set of -40,-20,0,27,75,125 C measurements Instrument set up for speed 100 Hz IF (20 or lower preferred) 32 points segmented Biasing done manually for single device 0.85 Vb and 1 Vc for all temp points (peak f max ). Automation possible with sequencing LRRM calibration was used

Device Measurement Procedure Devices measure cold and biased Open and Short de-embedding structures measured for all temperatures Measurements done in raw and corrected Raw measurements allow post calibration manipulation with different calibration sets

Probe Adjustment Load - Probe geometry adjusted for each temp point ISS pre-aligned before adjustment Best results Probes at temperature Move to ISS and adjust Move back to wafer until stable once again Full-auto cal at ISS Time delay can be frustrating...

Device Measurement Procedure Make sure LRRM is set to calculate Load inductance at 110 GHz

WinCal Speed Improvements Repeatability and Validation tests turned off Validation can be carried out over wafer using post-correct reflect no measurement required Monitoring was left off but this could be done as soon as probes over wafer or correct Raw cal open

Multi-Temperature Calibration Validation All calibrations were valid Use view data items and drag renamed % to new report

Dealing with Probe Movement Probe movement during the cal can effect thru delay If open validation fails thru delay can be adjusted

Failed Monitoring after DUT Measurements (125 C) Likely calibration was just too slow Calibration period less that 1:30 a must

Good Monitoring before and after DUT Measurement (75 C) If only measuring 1 device monitor after DUT measurement

Technique to Reduce Drift Effects Calibrate at higher than required temperature to suit expected temperature drop Re-monitor at desired temperature set point Simulated here - chuck temp was at 113 C for 125 C calibration

Actual device Masons Gain Measurements Correct Calibration

Actual Device Masons Gain Measurements Ambient Calibration Applied to Raw Data

Actual Device data Effect of Different Calibration Applied to the Same Data

Using WinCal to View Differences in Measurements with the Same Calibration

Using WinCal to Compute

Using Wincal to Compute Differences in Measurements from the Same Calibrations

Calibration Error Bound Variation

Effect of Load Resistance Variation 27 calibrations loaded Error terms compared by varying load resistance by 1% At 125 C load measured 50.75 Ώ using 4156

Variation in Error Bounds with RL Variation for 1% and 2% Load Variation

A Different Approach Load at DUT Temp Assumes that worst variation is load resistance Currently the subject of ARFTG paper development Sensitivity Analysis of Wafer-Level over Temperature RF Calibration" to be submitted to 80 th ARFTG Load resistance measured using parametric instrument

A Different Approach Load at DUT Temp Temperature still changes Absorber raised ISS from chuck Further experiments to evaluate this approach required

Use of WinCal for PPR and Parameter Extract Measurements of DUT various, Open pads and Short pads to be in same Report WinCal can perform PPR on the fly for all DUT measurements if required

Use of WinCal for PPR and Parameter Extract Preferred is to use Math ScratchPad to product individual data items useful for swapping with other reports

Conclusion Accuracy of on-wafer over-temperature RF measurements up to 110 GHz can be improved High-speed calibration with monitoring after stabilisation is the recommended method Different calibration required per temperature Reduce instrument IF bandwidth to highest tolerable, use low number of points WinCal repeatability / verification is not needed during calibration process over ISS WinCal greatly aids the calibration process

Any Questions? Thank you for attending. For questions, please contact: Gavin Fisher, Application Engineer gavin.fisher@cmicro.com +44-121-2860170