NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.

Performance and Reliability of Interface Materials for Automotive Power Electronics

Sreekant Narumanchi National Renewable Energy Laboratory sreekant.narumanchi@nrel.gov (303) 275-4062 NREL Team members: Doug DeVoto, Mark Mihalic, Paul Paret Industry Session Applied Power Electronics Conference Long Beach, CA March 19, 2013 NREL/PR-5400-57964

This presentation does not contain any proprietary or confidential information

2

Outline • Background • State-of-the-art of interface materials/interfaces • Non-bonded thermal interface materials (TIMs)

• Thermal resistance

• Bonded interface materials (BIMs) • Thermal resistance

• Reliability of bonded interfaces (accelerated testing)

• Modeling of BIMs

• Summary

3

Importance of Thermal Management and Reliability • Excessive temperature degrades

the performance, life, and reliability of power electronics and electric motors.

• Advanced thermal management technologies enable

– keeping temperature within limits

– higher power densities – lower cost materials,

configurations and system.

• Improve lifetime/reliability as well as develop new predictive lifetime models.

Reduce cost, improve reliability

Courtesy: Oak Ridge National Laboratory (ORNL)

4

State-of-the-Art of Interfaces/Interface Materials • Interfaces (especially polymeric interface materials) can pose a major bottleneck to heat

removal. • BIMs, such as solder, degrade at higher temperatures and are prone to thermomechanical

failure. • Need for improved reliability as well as predictive lifetime design tools for lowering cost. • Important for configurations employing wide bandgap devices.

Metallized substrate

Base plate/heat exchanger

Device

Bonded interfaces/ materials

Wire/ribbon bonds

Delamination Voiding Cracking Bonded Interface

2

2.5

3

3.5

4

4.5

100 110 120 130 140 150 160Dist

ance

acro

ss th

e Pa

ckag

e (m

m)

Temperature (°C)

New Degraded

Degraded Substrate Attach

132.5°C 156.7°C

5

Thermal resistance of various non-bonded TIMs

0

50

100

150

200

250

300

350

400

0 0.05 0.1 0.15 0.2Thickness (mm)

Ther

mal

resi

stan

ce (m

m2 K

/W)

Arctic Silver 5

Thermalcote-251G

Wacker Silicone P12

Keratherm KP77

Dow Corning TC5022

Shinetsu-X23-7783D-S

Chomerics XTS8010

Chomerics XTS8030

3M AHS1055M

Honeywell PCM45

172 kPa, ~ 75 C sample temperature TIM thickness in all cases is 100 μm

0102030405060708090

100

100 200 300 400 500 600Pressure (kPa)

Resi

stan

ce (m

m2 K/

W)

Graftech graphite - 125 microns (initial)Indium -150 microns (initial)Shinetsu X23-7783D-S (no thickness control)Honeywell PCM45G (no thickness control)Single-sided CNTSingle-sided CNT+Paraffin

• Red dashed line in the two figures above is the target thermal resistance (3 to 5 mm2K/W).

• Most non-bonded TIMs do not come close to meeting thermal specification of 3 to 5 mm2K/W at approximately 100-μm bond line thickness.

0 25 50 75

100 125 150 175 200 225 250 275

Ther

mal

cote

25

1G

Wac

ker

Sili

cone

P12

Arc

tic S

ilver

5

Ker

athe

rm

KP

77

Cho

mer

ics

XTS

8010

Cho

mer

ics

XTS

8030

Hon

eyw

ell

PC

M45

Dow

Cor

ning

TC

-502

2

Shi

nets

u X-

23-

7783

-D

3M A

HS

105

5M

Res

ista

nce

(mm

2 K/W

)

Bulk resistance Contact resistance

6

Thermal Resistance of Sintered Silver and Solder

Samples Thickness (µm)

Resistance (mm2K/W)

Silvered Cu-Cu sintered interface

20 5.8

27 8.0

64 5.4

Cu-Cu soldered interface (SN100C)

80 1.0

150 4.8

200 3.7

• The thermal resistance tests were performed using the NREL ASTM TIM apparatus – Average sample temperature ~ 65°C,

pressure is 276 kPa (40 psi). • The silvered Cu-Cu sintered interface

showed promising thermal performance.

• Results hinted at some problems with the bonding of the silvered Al-Al interface.

• The initial thermal results for a lead-free solder (SN100C) interface were promising.

• Bonded interface resistance in the range of 1 to 5 mm2K/W is possible.

ASTM test fixture

7

Thermal Resistance of Thermoplastics with Embedded Fibers

Sample

Thermoplastics with embedded carbon fibers

• Thermoplastic films (provided by Btech) bonded between 31.8-mm-diameter copper disks.

• Promising thermal results (8 mm2K/W for 100-µm bondline thickness).

• Continuing work at NREL to further decrease contact resistance to approach target thermal performance, as well as characterize reliability.

Sequence of bonding steps

8

Reliability of Bonded Interfaces

• Investigate the reliability of emerging BIMs to meet the thermal performance target of 3 to 5 mm2K/W.

• Identify failure modes in emerging BIMs. • Experimentally characterize their life under known conditions. • Develop lifetime estimation models.

Sample Assembly

Traditional Power Electronics Package

Silicon die

Metalized substrate

Base plate

BIM

Wire/ribbon bonds

9

Approach

BIM Mechanical Characterization

Sample Synthesis

Reliability Calculation

Thermal Testing/ Characterization

Synthesis of samples using vacuum solder reflow station

and hot press

Cycling of samples in a

thermal shock chamber

Characterization of samples via hipot tester, acoustic microscope,

and laser profilometer

Shear tests to extract mechanical characteristics of

BIMs

Number of cycles to crack

initiation/ delamination

Fatigue life prediction

Strain energy density per cycle

Extraction of viscoplastic parameters

Experimental Approach Modeling Approach – Finite Element Analysis/Calculations

10

Approach – Sample Assembly • Five samples of each BIM were synthesized for testing and

included: – Silver plating on the substrate and base plate. – Substrate based on a Si3N4 active metal bonding process;

base plate material is copper. – An interface between substrate and base plate with

50.8-mm x 50.8-mm footprint. • Samples followed manufacturer-specified reflow profiles, and

bonds were inspected for quality.

Sample Assembly

Bond Material Type Name Comments

Solder Kester Sn63Pb37 Baseline (lead-based solder)

Sintered Silver Semikron Based on Semikron synthesis process

Adhesive Btech HM-2 Thermoplastic (polyamide) film with embedded carbon fibers

11

Approach – Temperature Cycling • Cycle Profile

– -40°C to 150°C – 5°C/minute ramp rate – 10 minute dwell/soak time

• Failure Mechanisms – BIM: voids and cohesive or

adhesive/interfacial fractures – Substrate: Cu-to-Si3N4 delamination and

Si3N4 cracking

Cohesive Fracture Voids Substrate Delamination and Cracking Adhesive/Interfacial Fracture

Time

Failu

re P

aram

eter

Time

Failure Limit s1 s2

s3 s1<s2<s3

s3 s2

s1

t(s3) t(s2) t(s1)

12

Reliability of Thermoplastic-based BIM • Btech HM-2 (Carbon fibers within polyamide

matrix) – Bonding

o HM-2 was cut to the base plate dimensions o The sample assembly was placed in the hot press and raised to

195°C o 1 MPa (150 psi) of pressure was applied

– Reliability Results o After 2,000 cycles, the bonded interface remained defect-free

Cold Plate

Sample Hot Plates

Cold Plate

Screw Jack

Hot Press Sample Assembly

0 Cycles

1,000 Cycles

2,000 Cycles

13

Reliability of Sintered Silver-based BIM • Semikron Sintered Silver

– Bonding o Si3N4 edges were ground off to match the metallization layer. o The sample assembly was placed in a hot press and raised to its

processing temperature, then pressure was applied. o Compression testing of substrates at ORNL showed cracking of

substrates required between 30 MPa to 50 MPa of pressure. – Reliability Results

o Uniform bonds were obtained. o Adhesive fracture has initiated at the bonding perimeter.

Sample Assembly

0 Cycles

1,000 Cycles

2,000 Cycles

14

• After 2,000 cycles, perimeter fracturing has reached 11% to 21%.

2,000 Cycles

0%

5%

10%

15%

20%

25%

0 500 1,000 1,500 2,000De

lam

inat

ion

Number of Cycles

Sample 1

Sample 2

Sample 3

Sample 4

Reliability of Sintered Silver-based BIM

Credit: Paul Paret

15

BIM Finite Element Modeling (FEM)

-80

-40

0

40

80

120

160

0 2000 4000 6000 8000 10000 12000 14000

Tem

pera

ture

(°C)

Time (s)

Temperature Cycling Profile • Temperature cycling parameters: – -40°C to 150°C – 5°C/minute ramp rate – 10 minute dwell/soak time

• Viscoplastic material model applied to solder layer

• Temperature-dependent elastic material properties incorporated for base plate and substrate

Base plate

Solder layer

Substrate

Quarter Symmetry Model

16

BIM FEM

• Strain energy density in the bonded joint region shown to decrease as fillet radius increases (11.3 MPa and 11.2 MPa for 1.5 mm and 2 mm, respectively).

• Strain energy density versus cycles-to-failure correlation to be obtained for lead-based and lead-free solders.

Solder Layer Top Surface (1.5-mm radius) Solder Layer Top Surface (2-mm radius)

17

Summary • TIMs/BIMs are a key enabling technology for compact, light-weight, low-cost,

reliable packaging and for high-temperature coolant and air-cooling technical pathways.

• Characterization of thermal performance of TIMs/BIMs

– 3 to 5 mm2K/W resistance at 100 μm is a difficult target for non-bonded TIMs

– BIMs can meet this thermal target immediately after bonding – main question is reliability

• Characterization of reliability of BIMs

– Synthesis of various joints between substrates and base plate, thermal shock/temperature cycling, high-potential test and joint inspection (C-SAM), and strain energy density versus cycles-to-failure models

– Thermoplastic BIM is very reliable after 2,000 cycles, sintered silver BIM showing some significant edge delamination

• Initiated FEM for solder-bonded interface geometries – ultimate goal is to develop predictive lifetime model for BIM.

18

Summary

• Current/Future Work

– Complete 2,000 thermal cycles on all selected materials using Si3N4-based substrates

– Report on reliability of each BIM under specified accelerated test conditions

– Derive viscoplastic parameters for lead-based and lead-free solders from double-lap shear test experiments

– Develop strain energy density versus cycles-to-failure predictive lifetime model for lead-based solder

– Expand strain energy density versus cycles-to-failure predictive lifetime model to lead-free solders

– Improve process for large-area sintered silver-based interface, and eventually develop predictive lifetime model

For more information contact: Principal Investigator Douglas DeVoto Douglas.DeVoto@nrel.gov Phone: (303) 275-4256 APEEM Task Leader Sreekant Narumanchi Sreekant.Narumanchi@nrel.gov Phone: (303) 275-4062

Acknowledgments: Susan Rogers and Steven Boyd U.S. Department of Energy Team Members: Mark Mihalic Paul Paret

All photos in this presentation except for one on slide 14, credit: Douglas DeVoto, NREL

Collaborations: Semikron Btech Curamik Delphi GM Virginia Tech ORNL Heraeus Kyocera Shin-etsu Dow Corning 3M Parker Chomerics