Status of the Gallium Nitride High Electron Mobility ... radiation testing and analysis – Reliability test screens for new devices – Guidelines for implementation and testing To

Status of the Gallium Nitride High Electron Mobility Transistor

Radiation Testing for the NEPP ProgramLeif Scheick

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, Ca

Copyright 2017 California Institute of Technology. U.S. Government sponsorship acknowledged. This research was carried out in part by the Jet Propulsion

Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration under the NASA Electronic Parts and

Packaging Program (Code AE). Other data was collected from NASA flight projects.

To be presented at the 8th NASA Electronic Parts and Packaging (NEPP) Program Electronic Technology Workshop June 26, 2017, NASA GSFC, Greenbelt, MD.

Current activities

• Working group discusses best methods for evaluating new wide band gap technologies for infusion into space– GRC, JPL, JSC, GSFC, AFRL– Monthly meeting to share data and resources for radiation

effects testing and reliability analyses• Previous efforts have been broad stroke testing

– Heavy ion testing• Gallium Nitride HEMTs (JPL)• Silicon Carbide MOSFETs (GSFC)

– Reliability screening• Temperature cycling of GaN and SiC

• On going and future efforts– Continues radiation testing and analysis– Reliability test screens for new devices– Guidelines for implementation and testing


GaN Basics

• Current silicon power solutions are at their innate limits for space applications– Silicon devices are at

efficiency limit– Best hi-rel devices are less

then ~400 V drain-to-source• GaN devices are becoming

available– Reliability effects are a

concern– Gate stress is limited (abs max

of Vgs +6, -5 V)– Thermal effects and aging are

under study at GRC

Dielectric

GaN- - - - - - - - - - - - - - -

Si

AlGaN Electron Generating Layer

DGS- - - -


TESTING OF PANASONIC PARTS - CONTINUED

PGA26E19BA


Optical Images

5

Front Back


Accomplishments

6

460µm

300µm

500µm


Heavy Ion Testing

0 20 40 60 80Elasped time [s]

10-7

10-6

10-5

10-4

10-3

10-2

Cur

rent

[A]

_0024 PGA26E19BA Pr@60 MeV.cm2/mg VDS = 200 VVDS = 300 VVDS = 400 V


10-7

10-6

10-5

10-4

10-3

10-2

Cur

rent

[A]

_0031 PGA26E19BA [email protected] MeV.cm2/mg VDS = 300 VVDS = 325 VVDS = 600 V PIGS


10-7

10-6

10-5

10-4

10-3

10-2

Cur

rent

[A]

_0032 PGA26E19BA [email protected] MeV.cm2/mg VDS = 600 V


10-7

10-6

10-5

10-4

10-3

10-2

10-1

Cur

rent

[A]

_0033 PGA26E19BA Cu@20 MeV.cm2/mg VDS = 550 VVDS = 575 V


Heavy Ion Testing

8

20 30 40 50 60

LET [MeV.cm2/mg]

200

300

400

500

600

Dra

in-to

-Sou

rce

(VD

S) [V

]

Average Pass Bias (VSEE) [V]for the PGA26E19BA

VGS=0 V

1 1.2 1.4 1.6 1.8 2

Inverse cosine of angle

0

100

200

300

400

Dra

in-to

-Sou

rce

(VD

S) [V

] Average Pass Bias (VSEE) [V] vs Inv Cosine of Angle

for the PGA26E19BAVGS=0 V

20 30 40 50 60

LET [MeV.cm2/mg]

10-7

10-6

Cro

ss S

ectio

n [c

m2 ] Cross section vs LET

for the PGA26E19BAVGS=0 V

1 1.2 1.4 1.6 1.8 2

Inverse cosine of angle

0

100

200

300

400

Dra

in-to

-Sou

rce

(Vds

) [V]

Average Pass Bias (Vds) [V] vs Inv Cosine of Anglefor the PGA26E19BA

VGS=0 V


Heavy Ion Testing

60 70 80 90 100

Change in range [um]

0

40

80

120

160

200

Cha

nge

in V

see

[V]

20

6044

Effect of ion range on Vseefor the PGA26E19BA

Point lables are LET in MeV-cm2/mg

10 20 30 40 50 60

LET [MeV.cm2/mg]

200

300

400

500

600

Dra

in-to

-Sou

rce

(Vds

) [V]

Average Pass Bias (Vds) [V]for the PGA26E19BA

VGS=0 V @ TAM VGS= 0 V @ BNL

Deeper ion penetration decreases Vsee


Heavy Ion Testing

-10 -8 -6 -4 -2 0

VGS [V]

320

360

400

440

480

Dra

in-to

-Sou

rce

(Vds

) [V]

Average Pass Bias (Vds) [V] vs Vgsfor the PGA26E19BA

LET = 44.6 MeV-cm2/mg and Normal Incidence

Gate-to-source voltage has no effect on Vsee


Optical and Infrared Images with Hot Spots

11

• Hot spot in boxed region

SN42 SN42 SN44 SN44

SN46 SN46 SN48 SN48


Optical and Infrared Images with Hot Spots

12

• Hot spot in boxed region

SN605 SN605 SN609 SN609

SN610 SN610a SN611 SN611


Failed Die Analysis

13

0.5mA @ 11V

1.2mA @ 8V

1uA @ 310V

400nA @ 615V


SEE population

• SEE on this devices are very telling

• They all occur on the gate

• It appears that all are on the drain side although damage is quite large

• Randomly dispersed

14


TESTING OF TRANSPHORMCASCODE HYBRIDS

TPH3208 and TPH3202


Cascode Hybrids

• Cascode configuration allows for a depletion mode HEMT (Vth<0 V) to emulate an enhancement mode FET (Vth> 0V)

• Advantages are high gain, high bandwidth, high slew rate, high stability, and high input impedance

• Disadvantage is being a hybrid the device may be hard to assure

16


Optical Images

17

TPS3202Gate Source Drain

TPH3208Gate Source Drain


Heavy Ion Testing


10-6

10-5

10-4

10-3

10-2

10-1

Cur

rent

[A]

_1334 TPH3208 [email protected] MeV.cm2/mg VDS = 50 V

These devices are very soft to ion damage


Heavy Ion Testing

0 40 80 120 160 200Elasped time [s]

10-11

10-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Cur

rent

[A]

_1333 TPH3208 [email protected] MeV.cm2/mg VDS = 0 VVDS = 50 V

Ions cause damage when unbiased


Heavy Ion Testing

0 100 200 300 400 500Elasped time [s]

10-6

10-5

10-4

10-3

10-2

10-1

Cur

rent

[A]

_0638 TPH3202 Cu@20 MeV.cm2/mg VDS = 50 VVDS = 100 V

No incremental damage was seen


VSEE as a function of LET

20 25 30 35 40 45

LET [MeV.cm2/mg]

0

20

40

60

80

100D

rain

-to-S

ourc

e (V

ds) [

V]Average Pass Bias (Vds) [V]

for the TPH3208VGS=0 V

Vsee is essentially 0 V due to ion incremental damage


Cross-section as a function of LET

Cross-section is dependent on LET

20 25 30 35 40 45

LET [MeV.cm2/mg]

10-5

10-4

10-3

10-2

10-1C

ross

Sec

tion

[cm

2 ] Cross section vs LETfor the TPH3208

VGS=0 V


TESTING OF TI HYBRID POWER CONVERTER

LMG5200


LMG5200

• Texas Instrument’s LMG2500 is a Half-Bridge Power Stage

• Needs a PWM on the inputs to operate at a power converter

• Use two 80V enhancement mode GaN HEMTs for switches


LMG5200

• GaN HEMTs had to be milled down for ion penetration

• SEB type failure seen with Kr at 36.3 MeV-cm2/mg with Vin > 70 V

• SET on the gate maybe the cause

• Piece-part testing is planned


FUTURE WORK


SEE Mechanism

• Since this effect is voltage dependent, we look to the spot with the largest electric field

• This is under the drain side of the gate

• A gate transient occur before an SEE

Dielectric

GaN- - - - - - - - - - - - - - -

Si

AlGaN Electron Generating Layer

DGS- - - -

Ion strike here?

Or here?

-10 0 10 20 30Elapsed Time [us]

-1

0

1

2

3

Cur

rent

[A]

SEE of EPC1012Vds=200 V, Vgs=0 V

Drain CurrentGate Current


SEE Mechanism

• The electric field magnitude extends into the GaN substrate

• Very narrow region of very high field

• Resulting strain on junction will make it easier to have SEE


More testing

• Panasonic parts– Best vehicle for mechanism investigation

• Transphorm– They are so soft, may also yield mechanism clue

• TI parts– TPS53632G and LMG3410 are on deck for SEE

• Freebird Semi– Resping of EPC product line – include higher voltage

parts – worth another look


Conclusion

• SEE in GaN HEMTs are complex– Mechanisms and underlying device physics are still

under study• New devices show similar effects

– Panasonic parts seem more robust– Transphorm devices show that GaN can be very soft

and there is no natural low voltage threshold• Future plans

– Body of Knowledge document with GRC– Testing guideline in works– Collaborate with NASA flight projects and external

customers

Documents

Status of the Gallium Nitride High Electron Mobility ... radiation testing and analysis – Reliability test screens for new devices – Guidelines for implementation and testing To