Temperature accelerated Degradation of GaN HEMTs under High power - - PowerPoint PPT Presentation

▶
temperature accelerated degradation of gan hemts under
SMART_READER_LITE
LIVE PREVIEW

Temperature accelerated Degradation of GaN HEMTs under High power - - PowerPoint PPT Presentation

Dec 10, 2023 518 likes •833 views

Temperature accelerated Degradation of GaN HEMTs under High power Stress: Activation Energy of Drain Current Degradation Yufei Wu, Chia Yu Chen and Jess A. del Alamo Microsystems Technology Laboratory Acknowledgement: DRIFT MURI,

slide-1
SLIDE 1

Temperature‐accelerated Degradation of GaN HEMTs under High‐power Stress: Activation Energy of Drain Current Degradation

Yufei Wu, Chia‐Yu Chen and Jesús A. del Alamo

Microsystems Technology Laboratory

Acknowledgement: DRIFT‐MURI, TriQuint Semiconductor

1

slide-2
SLIDE 2

Outline

  • 1. Motivation
  • 2. High‐power and high‐temperature stress

experiments

  • 3. An improved approach
  • 4. Conclusions

2

slide-3
SLIDE 3

Motivation

3

  • N. Malbert, IRPS 2010
  • Activation energy, Ea :

essential in predicting lifetime

  • Conventionally:

high temperature accelerated life test

slide-4
SLIDE 4

Motivation

4

  • N. Malbert, IRPS 2010

Problems:

  • Requires multiple devices
  • Carrier trapping not properly dealt with
  • Different degradation mechanisms can emerge at different

temperatures

  • Activation energy, Ea :

essential in predicting lifetime

  • Conventionally:

high temperature accelerated life test

slide-5
SLIDE 5

Motivation

  • N. Malbert, IRPS 2010

Desirable: Ea extraction from measurements on a single device

time Tj

Step‐temperature stress

5

  • Activation energy, Ea :

essential in predicting lifetime

  • Conventionally:

high temperature accelerated life test

slide-6
SLIDE 6

Outline

  • 1. Motivation
  • 2. High‐power and high‐temperature stress

experiments

  • 3. An improved approach
  • 4. Conclusions

6

slide-7
SLIDE 7

7

Setup for DC reliability studies

DC/Pulsed Characterization

‐ KeithleySources ‐ Agilent B1500A

Windows‐based PC Accel‐RF System Hardware

MIT RF/DC Characterization Suite ‐ DC FOMs ‐ Current collapse

DUT

Switching Matrix RF/DC Units Accel‐RF Software ‐ RF measurement ‐Temperature control ‐ Stressing

Tbase

Heater

Augmented with:

  • external instrumentation for DC/pulsed

characterization

  • software to control external

instrumentation and extract DC FOMs

Devices: Prototype GaN Power Amplifier MMIC from industry

Accel‐RF AARTS RF10000‐4/S system

slide-8
SLIDE 8

High‐power DC Experiment Flowchart

Start Detrapping Full Characterization End: detrapping + full characterization Inner loop Short Characterization (DC) DC and Temperature Stress

  • Detrapping: Tbase = 250 °C for 7.5 hours
  • Full characterization
  • At Tbase = 50 °C
  • Full DC I‐V sweep
  • Current collapse

8

slide-9
SLIDE 9

High‐power DC Experiment Flowchart

Start Detrapping Full Characterization End: detrapping + full characterization Inner loop Short Characterization (DC) DC and Temperature Stress

  • Detrapping: Tbase = 250 °C for 7.5 hours
  • Full characterization
  • At Tbase = 50 °C
  • Full DC I‐V sweep
  • Current collapse
  • Stress:
  • High‐power condition
  • Base temperature stepped up
  • Short characterization
  • Every 30 minutes at Tbase = 50 °C
  • DC FOMs: IDmax, IGoff, RD , RS , VT , …

9

slide-10
SLIDE 10

High‐power DC Experiment Flowchart

Start Detrapping Full Characterization End: detrapping + full characterization Inner loop Short Characterization (DC) DC and Temperature Stress

10

  • Detrapping: Tbase = 250 °C for 7.5 hours
  • Full characterization
  • At Tbase = 50 °C
  • Full DC I‐V sweep
  • Current collapse
  • Stress:
  • High‐power condition
  • Base temperature stepped up
  • Short characterization
  • Every 30 minutes at Tbase = 50 °C
  • DC FOMs: IDmax, IGoff, RD , RS , VT , …
slide-11
SLIDE 11

11

Definitions of Various Figures of Merit

Parameter Definition IDmax ID at VGS =2 V, VDS =5 V IGoff IG at VGS =‐5 V, VDS =0.1 V RD Drain resistance measured with IG = 20 mA/mm RS Source resistance measured with IG = 20 mA/mm VT VGS – 0.5VDS when ID = 1 mA/mm at VDS = 0.1 V Current Collapse Percentage change in IDmax after 1 sec. VDS = 0 V, VGS = ‐10 V pulse

slide-12
SLIDE 12

High‐power DC Experiment

5000 10000 15000 1E-3 0.01 0.1 1 50 220 190 180 170 160

|IGoff| (mA/mm) time (min)

5000 10000 15000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Tbase(°C) = 220 50 210 200 190 180 170 160 150 140 130 120 90

IDmax/IDmax(0) time (min)

  • uter loop data
  • uter loop data

High‐power stress: VDS = 40 V, ID = 100 mA/mm, Tbase = 50 °C – 230 °C, 600 min/step

12

slide-13
SLIDE 13

High‐power DC Experiment

  • |IGoff| increases from Tbase=170 to 190°C; then saturates
  • Significant IDmax degradation for Tbase > 180 °C
  • Thermally activated IDmax degradation rate shown

13

High‐power stress: VDS = 40 V, ID = 100 mA/mm, Tbase = 50 °C – 230 °C, 600 min/step

  • uter loop data
  • uter loop data

5000 10000 15000 1E-3 0.01 0.1 1 50 220 190 180 170 160

|IGoff| (mA/mm) time (min)

5000 10000 15000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Tbase(°C) = 220 50 210 200 190 180 170 160 150 140 130 120 90

IDmax/IDmax(0) time (min)

slide-14
SLIDE 14

14

High‐power DC Experiment

High‐power stress: VDS = 40 V, ID = 100 mA/mm, Tbase = 50 °C – 230 °C, 600 min/step

  • RD increases significantly, consistent with IDmax decrease
  • RS increases much less

5000 10000 15000 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00

RS

R/R(0) time (min)

RD Tbase(°C)= 50 90 120 130 140 150 160 170 180190 200 210 220 230

slide-15
SLIDE 15

Activation Energies of Degradation Rates

25 26 27 28 29 30 31 2 4 6 8 10 12

IDmax: Ea=0.94 eV RD: Ea=0.91 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Outer loop data (device detrapped)

Tchannel obtained from thermal model of MMICs

25 26 27 28 29 30 31 2 4 6 8 10 12

RD: Ea=1.00 eV IDmax: Ea=0.58 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Inner loop data

15

slide-16
SLIDE 16

25 26 27 28 29 30 31 2 4 6 8 10 12

IDmax: Ea=0.94 eV RD: Ea=0.91 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Outer loop data (device detrapped)

  • Inner loop data :

Large difference between Ea for IDmax and RD

16

Tchannel obtained from thermal model of MMICs

Activation Energies of Degradation Rates

25 26 27 28 29 30 31 2 4 6 8 10 12

RD: Ea=1.00 eV IDmax: Ea=0.58 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Inner loop data

slide-17
SLIDE 17

25 26 27 28 29 30 31 2 4 6 8 10 12

IDmax: Ea=0.94 eV RD: Ea=0.91 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Outer loop data (device detrapped)

  • Inner loop data :

Large difference between Ea for IDmax and RD

  • Outer loop data :

Thermally activated behavior

17

Tchannel obtained from thermal model of MMICs

Activation Energies of Degradation Rates

25 26 27 28 29 30 31 2 4 6 8 10 12

RD: Ea=1.00 eV IDmax: Ea=0.58 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Inner loop data

slide-18
SLIDE 18

25 26 27 28 29 30 31 2 4 6 8 10 12

IDmax: Ea=0.94 eV RD: Ea=0.91 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Outer loop data (device detrapped)

  • Inner loop data :

Large difference between Ea for IDmax and RD

  • Outer loop data :

Close Ea values for IDmax and RD common physical origin

18

Tchannel obtained from thermal model of MMICs

Activation Energies of Degradation Rates

25 26 27 28 29 30 31 2 4 6 8 10 12

RD: Ea=1.00 eV IDmax: Ea=0.58 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Inner loop data

slide-19
SLIDE 19

5000 10000 15000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Tbase(°C) = 220 50 210 200 190 180 170 160 150 140 130 120 90

IDmax/IDmax(0) time (min)

5000 10000 15000 1E-3 0.01 0.1 1 50 220 190 180 170 160

|IGoff| (mA/mm) time (min)

Conclusions Drawn from the Experiment

  • IG degradation:
  • Increases fast at first
  • Eventually saturates
  • ID degradation:
  • Significant degradation only

after IG degradation is saturated

  • Thermally activated

19

slide-20
SLIDE 20

Conclusions Drawn from the Experiment

  • IG degradation:
  • Increases fast at first
  • Eventually saturates
  • ID degradation:
  • Significant degradation only

after IG degradation is saturated

  • Thermally activated

20

  • Desirable: separate IG and ID degradation
  • Key idea: short stress to degrade IG without ID degradation, then

long stress to degrade ID

slide-21
SLIDE 21

Outline

  • 1. Motivation
  • 2. High‐power and high‐temperature stress

experiments

  • 3. An improved approach
  • 4. Conclusions

21

slide-22
SLIDE 22
  • Phase 1: degrade IG without significant ID degradation
  • Short stress period
  • Tbase = 50‐220 °C, in 20 °C steps
  • Stress time: 6 minutes

22

DC Experiment : Improved Approach

slide-23
SLIDE 23
  • Phase 1: degrade IG without significant ID degradation
  • Short stress period
  • Tbase = 50‐220 °C, in 20 °C steps
  • Stress time: 6 minutes
  • Phase 2: degrade ID without further IG degradation
  • Longer stress period
  • Tbase: from 120 °C, increase in steps
  • Stress time: 24 hours

23

DC Experiment : Improved Approach

slide-24
SLIDE 24

5000 10000 15000 0.01 0.1 1

120 Tbase(°C)

|IGoff| (mA/mm) time (min)

After detrapping 150 170 185 200 205 210 215 |IGoff|

5000 10000 15000 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00

Tbase(°C) = 215

IDmax/IDmax(0) time (min)

After detrapping 150 170 185 200 205 210 120 IDmax

A Typical Experiment (Phase 2)

24

High‐power stress: VDS = 40 V, ID = 100 mA/mm, Tbase = 120 °C – 215 °C, 24 hours/step

During phase 1: |IGoff |increases by 2 orders of magnitude; IDmax decreases by 3%

slide-25
SLIDE 25

A Typical Experiment (Phase 2)

25

  • |IGoff | stays at saturated level (~0.5 mA/mm)
  • IDmax degradation shows thermally activated characteristics

During phase 2: During phase 1: |IGoff |increases by 2 orders of magnitude; IDmax decreases by 3%

High‐power stress: VDS = 40 V, ID = 100 mA/mm, Tbase = 120 °C – 215 °C, 24 hours/step

5000 10000 15000 0.01 0.1 1

120 Tbase(°C)

|IGoff| (mA/mm) time (min)

After detrapping 150 170 185 200 205 210 215 |IGoff|

5000 10000 15000 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00

Tbase(°C) = 215

IDmax/IDmax(0) time (min)

After detrapping 150 170 185 200 205 210 120 IDmax

slide-26
SLIDE 26

20.0 20.5 21.0 21.5 22.0 22.5 23.0 23.5 4 6 8 10 12

IDmax: Ea=1.04 eV RD: Ea=0.84 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Outer loop data (long detrapping)

26

Activation Energies of Degradation Rates

slide-27
SLIDE 27

Ea for IDmax close to values reported on similar technologies in conventional long term experiments

20.0 20.5 21.0 21.5 22.0 22.5 23.0 23.5 4 6 8 10 12

IDmax: Ea=1.04 eV RD: Ea=0.84 eV ln(1/|slope|) 1/kTchannel (eV

  • 1)

Outer loop data (long detrapping)

27

Activation Energies of Degradation Rates

slide-28
SLIDE 28

Activation Energy for Drain Current Degradation from Literature

Reference Bias conditions Temperature range Activation energy Ea

  • S. Singhal, et al.

IRPS 2006 VDS=28 V IDS=64 mA/mm Tj=260, 285, 310 °C 1.7 eV

  • P. Saunier, et al.

DRC 2007 VDS=40 V IDS=250 mA/mm Tj=260, 290, 320 °C 1.05 eV

  • E. Zanoni, et al.

Microwave Integrated Circuits Conference 2009 VDS=40 V IDS=167 mA/mm Tj=200, 245, 293 °C 0.68 eV ‐ 1.58 eV

  • N. Malbert, et al.

IRPS 2010 VDS=25 V IDS=417 mA/mm Tj=150, 175, 275, 320 °C 0.8 eV – 1.2 eV

  • J. Joh, et al.

IRPS 2011 VDS=40 V VGS=‐7 V Tj=75, 100, 125, 150 °C 1.12 eV This work VDS=40 V IDS=100 mA/mm Tj=223, 249, 269, 289, 296, 302 °C 1.04 eV

28

slide-29
SLIDE 29

Outline

  • 1. Motivation
  • 2. High‐power and high‐temperature stress

experiments

  • 3. An improved approach
  • 4. Conclusions

29

slide-30
SLIDE 30

Conclusions

  • Two‐phase experiment: separates IG and ID degradation in

GaN HEMTs under high‐power and high‐temperature stress

  • Two mechanisms exist:

‐ IG degrades first and eventually saturates ‐ ID degrades after IG degradation is saturated

  • Demonstrated new technique to extract Ea from

measurements on a single device

  • Eafor permanent IDmax degradation rate : 0.95‐1.05 eV

30