Trapping vs. Permanent D Degradation in GaN HEMTs d ti i G N HEMT - - PowerPoint PPT Presentation

trapping vs permanent d degradation in gan hemts d ti i g
SMART_READER_LITE
LIVE PREVIEW

Trapping vs. Permanent D Degradation in GaN HEMTs d ti i G N HEMT - - PowerPoint PPT Presentation

Trapping vs. Permanent D Degradation in GaN HEMTs d ti i G N HEMT Jungwoo Joh and Jess A. del Alamo Massachusetts Institute of Technology Acknowledgements: TriQuint Semiconductor ARL (DARPA WBGS program) ONR (DRIFT MURI program)


slide-1
SLIDE 1

Trapping vs. Permanent D d ti i G N HEMT Degradation in GaN HEMTs

Jungwoo Joh and Jesús A. del Alamo Massachusetts Institute of Technology

Acknowledgements: TriQuint Semiconductor ARL (DARPA WBGS program) ONR (DRIFT‐MURI program)

slide-2
SLIDE 2

Motivation Motivation

  • GaN HEMT reliability: big concern

GaN HEMT reliability: big concern

  • Performance degradation at high voltage:

T i l t d ( bl ) – Trapping‐related (recoverable) – Permanent (non‐recoverable)

G S D AlGaN

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

GaN

slide-3
SLIDE 3

Experimental: GaN HEMT Experimental: GaN HEMT

Gate Gate Source Drain SiN GaN Cap 2DEG AlGaN GaN SiC Substrate

Standard device with integrated field plate :

  • L =0 25 um W=2x25 um

3

  • LG=0.25 um, W=2x25 um
  • Fabricated by TriQuint Semiconductor
slide-4
SLIDE 4

Electrical Stress and Characterization Electrical Stress and Characterization

START Comprehensive but fast:

Coarse characterization (<15 sec)

Characterization IDmax, RS, RD, IGoff, VT…

Fine characterization (~30 sec) Trap analysis (30 min)

Frequent:

Coarse characteri ation e er 1 2 mins

Trapping Analysis l l

Coarse characterization: every 1‐2 mins Fine characterization: before and after stress and at important ponits

Benign: Electrical Stress VDS, VGS (or ID) Benign:

Both sets of measurements to produce a change smaller than 2% on any extracted parameter after 100 executions

4

slide-5
SLIDE 5

Permanent vs. Trapping Permanent vs. Trapping

1

stress recovery V =0

0.9 (0)

total (apparent) permanent degradation

VDS=0 VGS=‐30 V

0 8

Dmax/IDmax

degradation t i

0 7 0.8 ID

trapping degradation

IDmax: VDS=5 V VGS=2 V

0.7 30 60 Time (min)

88 days recovery

5

13 % permanent degradation + 15 % trapping degradation

slide-6
SLIDE 6

OFF‐state Stress @ 100 C

1 2 1 E+03

OFF state Stress @ 100 C

VGS=‐5 V, VDS=40 V @ 100 C

1.1 1.2 R/R(0) 1.E+01 1.E+02 1.E+03 mm)

RS RD

100 C

G S D

VGS VDS 0 9 1

ax/IDmax(0), R

1 E-02 1.E-01 1.E+00 IGoff| (mA/m

IGoff

S

AlGaN 2DEG

0.8 0.9 IDma 1.E-04 1.E-03 1.E-02 |

IDmax

Goff

GaN

500 1000 1500 2000 2500 Time (min)

At these points, trapping analysis was performed at 25 C. IGoff: VDS=0.1 V VGS=‐5 V

6

Fast and sharp IG degradation Slower IDmax & RD degradation

slide-7
SLIDE 7

Transient after VDS=0 pulse

9 5

trapping pulse (1 s VGS=‐10 V, VDS=0 V) @ t=0-

Transient after VDS 0 pulse

8 5 9 9.5

stress time=0 min 5 permanent degradation @

uncollapsed IDlin (before) uncollapsedI (after)

7 5 8 8.5 IDin (mA)

5 15 35 current collapse:

uncollapsed IDlin (after)

6 5 7 7.5 75

155 After 315 min trapping degradation 25 C

10

  • 3

10

  • 2

10

  • 1

10 10

1

10

2

10

3

6.5 t (sec)

collapsed IDlin (after)

7

IDlin (VGS=1, VDS=0.5 V) transient after applying trapping pulse Current collapse increases up to 300 min and saturates.

slide-8
SLIDE 8

Trapping & Permanent Degradation Trapping & Permanent Degradation

1 30 zed) 0.98 20 (normaliz apse (%)

trapping degradation Permanent degradation: ll d I

0.96 10 psed IDlin rrent colla

permanent degradation uncollapsed IDlin Trapping degradation: Current collapse

0.94 Uncollap Cu

p

  • uncol. IDlin – col. IDlin

uncollapsed IDlin =

500 1000 1500 2000 2500 Time (min)

8

Trapping degradation mostly saturates after 300 min. Permanent degradation keeps increasing.

slide-9
SLIDE 9

Impact of Temperature Impact of Temperature

OFF‐state (100 C) OFF‐state (150 C) ( ) ( )

0 96 0.98 1 30 40 normalized) pse (%)

trapping degradation

0 96 0.98 1 30 40

n (norm)

pse (%)

trapping degradation

0.92 0.94 0.96 10 20 llapsed IDlin (n Current collap

permanent degradation

0.92 0.94 0.96 10 20 collapsed IDlin Current collap

permanent degradation

0.9 500 1000 1500 2000 2500 Unco C Time (min) 0.9 500 1000 1500 Unc C Time (min)

More degradation in trapping & permanent at higher T  Both degradations are thermally activated.

9

Increase in current collapse saturates faster than at 100 C.

slide-10
SLIDE 10

Hot‐electron Stress Hot electron Stress

VGS=0, VDS=40 V (ID~800 mA/mm) Room T (Tj~235 C)

RD

G S D

VGS VDS

Room T (Tj 235 C)

IGoff RS

AlGaN 2DEG Hot electrons!

IDmax

Goff

GaN 10

Much less degradation in IG

slide-11
SLIDE 11

OFF‐state vs. Hot‐electron Stress OFF state vs. Hot electron Stress

OFF‐state (100 C) Hot‐electron (RT, Tj~235 C)

0.95 1 20 30 (normalized) apse (%)

trapping degradation

0.95 1 20 30 IDlin (norm) apse (%)

permanent

0.9 10

  • llapsed IDlin

Current colla

permanent degradation

0.9 10 Uncollapsed Current colla

t i d d ti p degradation

0.85 500 1000 1500 2000 2500 Unco Time (min) 0.85 1000 2000 3000 4000 5000 Time (min)

trapping degradation

More permanent degradation than OFF‐state. Less current collapse increase for VDS=0 pulse.

11

 less trap formation in high‐power stress

slide-12
SLIDE 12

Summary Summary

  • During all stress modes:

During all stress modes:

  • 1. Very fast IG degradation (few minutes)
  • 2. Trapping‐related degradation mostly saturates in

pp g g y a short time (few hours)

  • 3. Permanent degradation keeps increasing over

time

  • OFF‐state:

– More trapping degradation – Faster & more degradation at higher T.

12

  • Hot‐electron stress:

– More permanent degradation