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.

Reliability of Bonded Interfaces

PI: Douglas DeVoto National Renewable Energy Laboratory May 14, 2013 Vehicle Technologies Program Annual Merit Review Arlington, Virginia NREL/PR-5400-58170 Project ID: APE028

This presentation does not contain any proprietary, confidential, or otherwise restricted information.

2

Overview

Timeline Project Start Date: FY10 Project End Date: FY13 Percent Complete: 85%

Barriers and Targets • Cost • Weight • Performance and Lifetime

Budget Total Project Funding: DOE Share: $2,300K

Funding Received in FY12: $650K Funding for FY13: $625K

Partners • Interactions / Collaborations

– General Motors, Btech, Semikron, Heraeus, Kyocera, Virginia Tech, Oak Ridge National Laboratory (ORNL)

• Project Lead – National Renewable Energy

Laboratory

3

IGBT / Diode Die

Metalized Substrate Substrate Attach

Base Plate

Die Attach

Interconnect Encapsulant

Enclosure Terminal

2

2.5

3

3.5

4

4.5

100 110 120 130 140 150 160Dist

ance

acr

oss t

he P

acka

ge (m

m)

Temperature (°C)

New Degraded

Relevance

• Excessive temperature degrades the performance, life, and reliability of power electronics components.

• Interfaces can pose a major bottleneck to heat removal. • Bonded interface materials (BIMs), such as solder, degrade at higher

temperatures and are prone to thermomechanical failure.

Degraded Substrate Attach

132.5°C 156.7°C

Traditional Power Electronics Package

4

Relevance

• As operating temperatures increase, the coefficient of thermal expansion (CTE) mismatch between the substrate and the base plate causes defect initiation and propagation in the joining layer.

CTE (ppm/°C) Silicon: 2.6 AlN: 4.5 Si3N4: 3.2 Copper: 16.5 Aluminum: 22.7 Sn63/Pb37 Solder: 24.7

Cooling

Heating

5

Objectives

• Overall Objective – Investigate the reliability of emerging BIMs (such as sintered silvers, lead-free solders, and

thermoplastics with embedded carbon fibers) for power electronics applications to meet the thermal performance target of 5 mm2K/W

– Identify failure modes in emerging BIMs, experimentally characterize their life under known conditions, and develop lifetime estimation models

• Address Targets – High-performance, reliable, low-cost bonded interfaces enable:

o Compact, light-weight, low-cost packaging o High-temperature coolant and/or air cooling

• Uniqueness and Impacts

– Thermal performance and reliability of emerging sintered materials and thermoplastics for large-area attachments were characterized

6

Milestones Date Description

October 2012 Completed finite element analysis (FEA) to determine strain energy density in lead-based solder samples with various fillet radius geometries

December 2012 Completed experimental temperature cycling of Btech HM-2 and Semikron sintered silver samples to 2,000 cycles

February 2012 Completed FEA to determine strain energy density in lead-based solder samples under various temperature cycle profiles

June 2013 Complete experimental temperature cycling of lead-based solder samples to 2,000 cycles

July 2013 Complete double lap shear testing of lead-based solder samples and use stress/strain data to revise viscoplastic properties needed for FEA

7

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 high-potential 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 - FEA/Calculations

8

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

included: – Silver plating on the substrate and copper base plate – Substrate based on a Si3N4 active metal bonding process – An interface between 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

Credit: Douglas DeVoto, NREL

9

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)

Credit: Douglas DeVoto, NREL (all photos)

10

Thermoplastic Evaluation Btech HM-2 (Carbon Fibers within Polymer Matrix)

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

195°C – 1 MPa (150 psi) of pressure was applied – Bond line thickness was measured to be 88.9 µm

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

Credit: Douglas DeVoto, NREL (all photos)

500 Cycles

1,000 Cycles

2,000 Cycles

Cold Plate

Sample Hot Plates

Cold Plate

Screw Jack

Hot Press Sample Assembly

11

500 Cycles

1,000 Cycles

2,000 Cycles

Semikron Sintered Silver • Bonding

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

processing temperature; then pressure was applied – Compression testing of substrates at ORNL showed cracking of

substrates required between 30 MPa to 50 MPa of pressure • Reliability Results

– Uniform bonds were obtained – Cohesive fracture initiated at bonding perimeter

Sample Assembly

Sintered Silver Evaluation

Credit: Douglas DeVoto, NREL (all photos)

12

• After 2,000 cycles, perimeter fracturing 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

Sintered Silver Evaluation

Credit: Paul Paret, NREL

Credit: Douglas DeVoto, NREL

Metalized Substrate

Base Plate

Sintered Silver

A

B

A B 91.6 µm

13

Solder Evaluation Lead-based (Sn63Pb37) Solder

• Bonding – Manual stencil was used to apply a 127-µm-thick solder layer to

the substrate and base plate surfaces – The assembled sample was placed in a vacuum solder reflow

oven and raised to its processing temperature • Reliability Results

– Bonds with voiding under 2% were obtained – Cohesive fracture initiated at bonding perimeter

0

50

100

150

200

250

0 20 40 60 80

Tem

pera

ture

(⁰C)

Time (min)

Reflow Profile 500 Cycles

1,000 Cycles Credit: Douglas DeVoto, NREL (all photos)

14

0%

4%

8%

12%

16%

0 250 500 750 1,000De

lam

inat

ion

Number of Cycles

Sample 1

Sample 2

Sample 3

Sample 4

• After 1,000 cycles, perimeter fracturing reached 11% to 14%

1,000 Cycles

Solder Evaluation

Credit: Paul Paret, NREL

Credit: Douglas DeVoto, NREL

Metalized Substrate

Base Plate

Solder

A

B

A B 96.7 µm

15

BIM Finite Element Modeling

-80

-40

0

40

80

120

160

0 5,000 10,000 15,000

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

-5

-4

-3

-2

-1

0

1

2

3

-0.02 -0.01 -0.01 0.00 0.01 0.01 0.02

Shea

r Str

ess

(MPa

)

Shear Strain

Hysteresis Loop

High DwellRamp DownLow DwellRamp Up

-0.02

-0.01

0.00

0.01

0.02

-40

0

40

80

120

160

0 50 100 150 200 250

Shea

r Str

ain

Tem

pera

ture

(°C)

Time (min)

Temperature and Shear Strain versus Time

Temperature Shear Strain

-5

-3

-1

1

3

-40

0

40

80

120

160

0 50 100 150 200 250

Shea

r Str

ess

(MPa

)

Tem

pera

ture

(°C)

Time (min)

Temperature and Shear Stress versus Time

Temperature Shear Stress

BIM Finite Element Modeling • Stress-strain hysteresis loops help to understand the inelastic behavior of the solder interface • Energy stored in the solder interface region due to deformation during thermal loading is

referred to as the strain energy density

Shear Strain Energy Density

17

BIM Finite Element Modeling • Hysteresis loops for variations in package geometry, dwell

time, and ramp rate were explored – Geometry: 1.5 mm and 2 mm fillet radius – Ramp rate: 5°C/min, 10°C/min, and 15°C/min – Mean temperature: 50°C, 10°C, and 100°C – Dwell time: 10 min, 30 min, and 75 min

• Strain energy density value will be compared to experimental fracture rate to obtain a cycles-to-failure correlation for lead-based solder

-8

-6

-4

-2

0

2

4

6

-0.10 -0.05 0.00 0.05

Shea

r Str

ess (

MPa

)

Shear Strain

Mean Temperature

-40°C to 60°C

0°C to 100°C

50°C to 150°C-6

-4

-2

0

2

4

-0.05 -0.03 -0.01

Shea

r Str

ess (

MPa

)

Shear Strain

Ramp Rate

5°C/min

10°C/min

15°C/min

-5

-4

-3

-2

-1

0

1

2

3

-0.05 -0.03 -0.01

Shea

r Str

ess (

MPa

)

Shear Strain

Geometry

2 mm1.5 mm

-4

-2

0

2

4

-0.05 -0.03 -0.01

Shea

r Str

ess (

MPa

) Shear Strain

Dwell Time

10 min30 min75 min

18

BIM Mechanical Characterization • Strain prediction of solder material is dependent on stress,

temperature, and time – A high enough stress will cause the material to plastically

deform – Solder has a tendency to creep (time-dependent plasticity)

at room temperature; this increases as operating (absolute) temperature approaches the melting temperature

• Viscoplasticity models combine plasticity and creep deformations into one equation to properly define solder in FEA

• Sample testing using a double-lap shear test fixture at various strain rates and temperatures generates the needed data to characterize the viscoplastic nature of solder

Primary Creep

Secondary Creep

Tertiary Creep

Rupture

Time

Stra

in

Initial Strain (elastic + plastic)

Creep Curve

Strain

Stre

ss

Stress-Strain Curve

Elastic Behavior

Plastic Behavior

Permanent Set

Double-Lap Shear Fixture and Sample

Credit: Douglas DeVoto, NREL

19

Collaboration and Coordination

• Partners – Btech (Industry): collaboration on optimizing thermoplastic BIM for

large area attachment – General Motors (Industry): technical guidance – Heraeus (Industry): collaboration on using low-pressure sintered

silver materials before products are commercially available – Kyocera (Industry): provided insight on Si3N4 substrate bonding

process and advantages over AlN substrates – ORNL (Federal): collaboration to determine maximum pressure that

Si3N4 substrates could withstand – Semikron (Industry): provided bonded samples to NREL using

company’s silver sintering process – Virginia Tech (University): collaboration on synthesis of samples using

sintered silver material

20

Proposed Future Work (FY13)

• Derive viscoplastic parameters for lead-based solder from double-lap shear test experiments

• Expand strain energy density versus cycles-to-failure models to lead-free solders

• Complete 2,000 thermal cycles on lead-based solder samples • Report on reliability of each BIM under specified accelerated test

conditions

21

Summary • DOE Mission Support

– BIMs are a key enabling technology for compact, light-weight, low-cost, reliable packaging and for high-temperature coolant and air-cooling technical pathways

• Approach – Synthesis of various bonds between substrates and base plate, thermal

shock/temperature cycling, high-potential test and bond inspection (acoustic microscope), and strain energy density versus cycles-to-failure models

• Accomplishments – Evaluated a number of bonded interfaces subjected to temperature cycling

o Lead-based solder, sintered silver, thermoplastic

– Implemented FEA for solder bonded interface geometries

22

Summary • Collaborations

– General Motors, Virginia Tech, ORNL, Btech, Semikron, Heraeus, Kyocera

• Future Work – Derive viscoplastic parameters for lead-based solder from double-lap shear

test experiments – Expand strain energy density versus cycles-to-failure models to lead-free

solders – Complete 2,000 thermal cycles on lead-based solder samples – Report on reliability of each BIM under specified accelerated test conditions

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

Reviewer-Only Slides

25

Responses to Previous Year Reviewers’ Comments

• “One evaluator described that the up-front analysis addressed potential failure modes and technical issues to date are understood.”

– It is a primary goal to validate modeling tools by comparing them to likely failure modes (adhesive or cohesive failures) observed through accelerated testing.

• “The other person commented that the collaboration appears adequate or better, especially with industry such as Semikron.”

– Collaboration with many industry partners has leveraged their material and bonding experience to ensured that NREL is testing production quality bonds.

• “One person pointed out that it was hard for them to say much about the future plans, but the research seems well-conceived.”

– At the time of the Annual Merit Review, this project was in its last year and hence no long-term future work goals were presented. Since then, we have submitted the draft of the FY2013 Annual Operating Plan to DOE in which we have proposed a project related to reliability of high-temperature attach materials/interfaces. Future work in the area of bonded interfaces will focus on the failure mechanisms of sintered silver materials/interfaces and other high-temperature attach materials/interfaces under thermal cycling conditions, as well as development of physics-of-failure models.

26

Publications and Presentations • Publications

– D. DeVoto, 2012, “Thermal Performance and Reliability of Bonded Interfaces,” FY20112 DOE APEEM Annual Report, October 2012.

– D. DeVoto, 2012, “Thermal Performance and Reliability of Bonded Interfaces,” IMAPS Automotive Microelectronics and Packaging, May 2012.

– S. Narumanchi and D. DeVoto, 2011, “Thermal Performance and Reliability of Bonded Interfaces,” FY2011 DOE APEEM Annual Report, October 2011.

• Presentations – S. Narumanchi, 2013, “Performance and Reliability of Interface Materials for Automotive Power Electronics,” Applied

Power Electronics Conference, Long Beach, CA, March 2013.

– D. DeVoto, 2012, “Thermal Performance and Reliability of Bonded Interfaces,” Advanced Power Electronics and Electric Motors FY13 Kickoff Meeting, DOE Vehicle Technologies Program, Oak Ridge, TN, November 2012.

– D. DeVoto, 2012, “Thermal Performance and Reliability of Bonded Interfaces,” Presented to the DOE Vehicle Technologies Program Electrical and Electronics Technical Team, Southfield, MI, May, 2012.

– S. Narumanchi, 2012, “Thermal Performance and Reliability of Bonded Interfaces,” SAE World Congress, Detroit, MI, April 2012.

– D. DeVoto, 2011, “Thermal Performance and Reliability of Large-Area Bonded Interfaces in Power Electronics Packages,” ASME International Mechanical Engineering Congress and Exposition, Denver, CO, November 2011.

– S. Narumanchi, 2011, “Thermal Performance and Reliability of Bonded Interfaces,” Advanced Power Electronics and Electric Motors FY12 Kickoff Meeting, DOE Vehicle Technologies Program, Oak Ridge, TN, November 2011.

– D. DeVoto, 2011, “Thermal Performance and Reliability of Bonded Interfaces for Power Electronics Packaging Applications,” Accelerated Stress Testing and Reliability Workshop, San Francisco, CA, September 2011.

27

Critical Assumptions and Issues • Large-area bonded interfaces can lead to thermomechanical stresses in the

package and consequently cracks, voids and delaminations. For any proposed solution, it is important to address issues related to thermomechanical reliability.

– The issue of reliability is specifically being addressed in this project.

• Physics-of-failure and degradation mechanisms for several newer bonded interface materials (e.g., silver sintered material) are not well known and have to be addressed.

– We are addressing these aspects to some extent in this project. The hypothesis is that we are developing generalized (i.e., independent of geometry) strain energy density versus cycles-to-failure relations for specific bonding materials (solders).

• The bonded interface solution will have to be low cost and be easily integrated into the manufacturing process.

– Arguably, none of the materials and processes considered in this project are particularly high cost or complex from a manufacturing process integration standpoint.