Wafer-Level Calibration & Verification up to 750 GHz. Choon Beng Sia, Ph.D. Mobile:

Transcription

1 Wafer-Level Calibration & Verification up to 750 GHz Choon Beng Sia, Ph.D. Mobile:

2 Outline LRRM vs SOLT Calibration Verification Over-temperature RF calibration 110 GHz 4 port Calibration & Validation Sub-THz Wafer Level VNA Calibration PAGE 2 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

3 On-Wafer RF Measurements Air Coplanar Probes Probes with Microstrip Design SHORT Electrical behavior standard dimensions probe pitch probe placement Impedance Standard Substrates (ISS) LOAD THRU PAGE 3 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

4 The Need for VNA Calibration Systematic Errors Imperfections in Network Analyzer and Test setup Define the measurement reference plane in a Test setup Random Errors Vary with time in Random Fashion (Unpredictable) Main Contributors: Instrument noise, switch and connector repeatability. Drift Errors Due to System performance changing after a calibration has been done. Primarily cause by Temperature Variation When to Recalibrate? Perform System Stability Check PAGE 4 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

5 SOLT Calibration Short L short Load L term Open (probes in air) C open Thru Delay Thru Oldest calibration technique All standards must be perfectly known Available every vector network analyzer (CalKit required) Open has capacitance/inductances Short and load have inductances Cal coefficients are defined for a particular probe placement Sensitive to probe placement Looks best after calibration Forces VNA to correct itself based on Cal Coefficients How about over-temperature? PAGE 5 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

6 LRRM Calibration Line Reflect Reflect (probes in air) Match Same Standards as SOLT Industry Standard verified by NIST Line-Reflect-Reflect-Match Calibration Only need to know Thru (line) delay & Match resistance Measurements referenced to trimmed resistor Dynamic Load Inductance Extraction Minimize probe placement sensitivity, providing accurate load inductance extraction Improves the accuracy of reference impedance Robust and Accurate Less sensitive to probe placement errors Requires less information about standards PAGE 6 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

7 SOLT vs LRRM Calibration Experiment: Infinity probes GSG pitch 100 m Probes ISS Calibration Technique SOLT LRRM OPEN Capacitance (ff) -6.5 Not Required SHORT Inductance (ph) 3.3 Not Required Load (Ohm) Load Inductance (ph) -0.4 Automatically Determined Thru Delay 1ps 1ps PAGE 7 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

8 50 Ohm and Thru Standards 50 Ohm Standard Use only precision trimmed loads marked on the ISS maps. ±0.3% accuracy. Untrimmed Loads Trimmed Loads 1ps Thru Standard Use alignment marks to maintain correct probe separation distance. Initial Contact Final Contact PAGE 8 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

9 Measure OPEN & SHORT after LRRM Calibration Probes are calibrated using LRRM and then, 2 measurements are made: Probes lifted in the air (OPEN) Separate SHORT standard Using network & Admittance (S to Y) parameters to extract OPEN cap and SHORT Inductance Port 1 Port 2 A B C Y Y Y Y A B B A = Y 11 + Y 12 C = Y 22 + Y 21 B = -Y 12 or -Y 21 B C B PAGE 9 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

10 OPEN Standard measured with LRRM OPEN standard = Lift up probes SOLT OPEN Cap defined as -6.5 ff LRRM OPEN Cap = -6.5 ff LRRM measures the Standard s real behavior Very repeatable since NO Probe Placement Sensitivity issue PAGE 10 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

11 SHORT Standard measured with LRRM 3.55 ph (Actual Short inductance) 3.3 ph (Cal value for SOLT) SOLT requires probe on SHORT = 3.3 ph LRRM gives actual value Slight deviation could be due to Probe Placement Error Worn-out Standard Is this critical? Small L Gate Inductance RFCMOS (ph) Large L 2nH Inductor, Error 0.01% PAGE 11 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

12 Calibration Verification PAGE 12 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

13 1ps Verification Line PAGE 13 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

14 1ps Verification Line L R = 1ps Port 1 Port 2 B A C Rs LINE +2Probe_R C PAGE 14 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

15 3ps Verification Line PAGE 15 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

16 RF Over-Temperature Calibration PAGE 16 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

17 Example of RF Over-Temperature Device Measurement Inductor (N5W10D120) -35 C 25 C 85 C Same L T Q -35 C 85 C Inductance Quality Factor MUST characterize devices at operating temperatures Device models must predict temperature effects PAGE 17 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

18 Over-temperature RF Probe Calibration 1. Load std = 50 Ohms at all temperatures (Standards must be thermally Isolated) 2. RF probes in thermal equilibrium as devices during cal & measurement. Test Sites PAGE 18 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

19 Recommended Flow for Over-temperature Cal How to check if probe in thermal equilibrium with DUT? Perform Stability Check Lift Probes up in Air (OPEN) Measure and Monitor S11 & S22 OPEN allow probe thermal stability checks anywhere. Fail Fail RF 25 C Ramp to 125 C Heat Probes At 125 C Thermal Soaking Stability Check? Pass RF 125 C Heat Probes Stability Check? Pass Start Measurement How Long should I wait for my probes to achieve thermal equilibrium at 125ºC? Probes Open in Air Did I spend too much time during the RF cal? Example of 125ºC RF Calibration Sequence PAGE 19 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

20 Deviation in Probe S11 OPEN (in Air) Cal 25ºC 125ºC (Deviation show SOLT can t support) 5 mins Probes heated on wafer pads 10mins & beyond Ready for Cal! PAGE 20 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

21 Recommended Flow for Over-temperature Cal How to check if probe in thermal equilibrium with DUT? Perform Stability Check Lift Probes up in Air (OPEN) Measure and Monitor S11 & S22 OPEN allow probe thermal stability checks anywhere. Fail Fail RF 25 C Ramp to 125 C Heat Probes At 125 C Thermal Soaking Stability Check? Pass RF 125 C Heat Probes Stability Check? Pass Start Measurement How Long should I wait for my probes to achieve thermal equilibrium at 125ºC? Probes Open in Air Did I spend too much time during the RF cal? Example of 125ºC RF Calibration Sequence PAGE 21 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

22 Checking Probe Thermal Stability after Cal 1. After Cal, measure Probe OPEN S11/S22 2. Reheat probes on the wafer. 3. After heating, measure Probe OPEN S11/S22 4. Compare 1 & 3 5. Ensure probe temperature change did not affect cal. PAGE 22 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

23 110 GHz 4 port Calibration using GSGSG probes PAGE 23 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

24 Cascade & Keysight 4 port 110 GHz Setup PAGE 24 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

25 Cascade & Keysight 4 port 110 GHz Setup PAGE 25 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

26 Unknown Thru Calibration Adopt unknown thru calibration or SOLR calibration. SHORT-OPEN-LOAD for each ports on ISS ( ). For the Thru standard, adjust the probes to complete 4 thru standards. Example of Thru 3-4 PAGE 26 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

27 Port 3-4 Calibration Verification 1ps Line Delay valid up to 100GHz PAGE 27 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

28 Port 1-3 Calibration Verification 2ps Loop-back Line Increase in Delay due to increase in inductance PAGE 28 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

29 Differential Line 300um 2.3ps delay 2.3ps Differential Line PAGE 29 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

30 Differential Line (WincalXE, 4 Port SOLR cal) 2-3 ps variation in delay (vs ps) PAGE 30 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

31 Sub-THz Wafer Level VNA Calibration PAGE 31 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

32 Challenges for Sub-THz Banded measurements Banded Calibration : GHz, GHz, GHz, GHz, GHz and GHz. mmw tests on same device Design TRL standards. Setup Freq extenders, S-bends and waveguide probes for every frequency bands. Perform calibration & measure device for every band - Tedious and time consuming. How to ensure calibration continuity and device measurement integrity for all frequency bands? Frequency Extenders S-bend Waveguides Waveguide Probes PAGE 32 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

33 Example of Data Discontinuity (160um Line) Discontinuity in Phase Gain? PAGE 33 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

34 How to Verify Calibration Continuity? Aligning Probes Magnitude S21 on 0.5ps Line 0.5ps Line PAGE 34 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

35 How to Verify Calibration Continuity? System Drift, waited too long to make measurements Aligning Probes Return Loss on 0.5ps Line 0.5ps Line PAGE 35 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

36 How to Verify Calibration Continuity? Return Loss Measured Immediately after Calibration - System Drift is a big challenge Aligning Probes 0.5ps Line PAGE 36 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

37 Verifying Calibration Continuity Design verification line with alignment markers on wafer. Characterize same line for all bands after TRL cal (or LRRM cal). Plot all bands, Phase S12 or S21 (group delay can be used too) Use plot as reference ensure Measurement Continuity & Integrity. Alignment Markers Virtual Grid (Prober Software) PAGE 37 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

38 Example of 325 GHz cal Discontinuity in Test Data Discontinuity Perform GHz calibration & measurement. After calibration, check with reference plot and found discontinuity. Stop and recheck PAGE 38 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

39 Example of 325 GHz cal Discontinuity in Test Data PAGE 39 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

40 Ensure Continuity in Contact Resistance? Difficult to determine if Probe Rc is good Aligning Probes Extracted 2*Probe Rc + Line Resistance 0.5ps Line PAGE 40 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

41 Ensure Continuity in Contact Resistance? Check DC contact resistance for each set of waveguide probes (20-50 milliohm range) Contacting probes Short Structure Design on Wafer Additional DC probes (for True Kelvin measurement) to measure Rc of the waveguide probes. RF probe HIGH Sense High Low SHORT LOW Sense PAGE 41 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

42 Ensure Continuity in Biasing Voltage Ensure same Voltage is applied to Device terminals. Correct Series Resistance from Cabling, Probe Bias Tee Network, Probe tips contact for every frequency band setup. Use Voltmeter if power supply has no sense capability Kelvin Power Supply (B1500) 1V Sense RF probe (HIGH Force) RF probe (Ground) Waveguide Probe (Force) 0.95V? DC Probe (Sense) Line PAGE 42 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

43 Summary Choose the calibration method which, Requires minimal info of the calibration standards Insensitive to Probe Placement Always verify with known devices after calibration. Over-temperature Calibration Ensure standards are thermally isolated Ensure probes are in thermal equilibrium as Device/Wafer Unknown Thru or SOLR calibration to support 110 GHz 4 port calibration. For Banded Calibration, Check Probe R DC, Design & use alignment markers to ensure consistent probe placement Ensure data integrity and continuity PAGE 43 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG

44 Thank You Questions to PAGE 44 5 JULY 2016, COMPANY CONFIDENTIAL, DR SIA CHOON BENG