James R Salvador Jung Young Cho and Jihui YangGeneral Motors Research amp Development

Andrew A Wereszczak and Hsin WangOakridge National Lab

Mechanical and Elastic Property Evaluation of

n- and p-type Skutterudites

jamessalvadorgmcom

GM RampD Thermoelectrics Team Researchers

Gregory P MeisnerJihui YangMike Reynolds

PostdocsXun Shi Jung Cho Zuxin Ye

Engineering OperationsKevin RoberJohn Manole

Gov ContractsEd Gundlach Amanda DemitrishRick Leach

ManagementJan HerbstMark Verbrugge

GMPT Integration amp TestingGreg Prior (Retired)

Joshua Cowgill

CollaboratorsSubcontractors Marlow Industries Jeff Sharp Jim Bierschenk Josh Moczygemba

and Alan Thompson Oak Ridge National Laboratory Hsin Wang Andy Wereszczak University of Nevada Las Vegas Changfeng Chen Yi Zhang

Future Tech Francis Stabler

Heat Technology Inc

Emcon (Faurecia)

Shanghai Institute of Ceramics Lidong Chen

University of Michigan Ctirad Uher

University of South Florida George Nolas

Brookhaven National Laboratory Qiang Li

Michigan State University Don Morelli

General Electric Global Research Todd Anderson Peter DeBock

AcknowledgementsUS Department of Energy Grant DE-FC26-04NT 42278

John Fairbanks (DOE) Carl Maronde (NETL)US Department of Energy Assistant Secretary for Energy Efficiency and Renewable Energy Office of Vehicle Technologies as part of the Propulsion Materials Program under contract DE-

AC05-00OR22725 with UT-Battelle LLC

Skutterudite a CoAs3 mineral found near Skutterud Norway in 1845 and compounds with the same crystal structure (body-centered cubic Im3 Oftedal I (1928) Zeitschrift fuumlr Kristallographie 66 517-546) are known as ldquoskutteruditesrdquo

W Jeitschko D J Braun Acta Crystallogr 1977 B33 3401 Filled Skutterudite Crystal Structure for LaFe4P12

Filled Skutterudites Technologically Important and Scientifically Fascinating Materials

Co As

Co4As12

Empty cages

Fracture and Failure in Ceramics Battling Flaws

Fracture toughness KIC =Yσradicaπ Mode I fracture Y is a crack shape factor and is larger for surface and edge cracks

Small Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 20 40 60 80 100

Flaw Size c (microm)

KIc = 1

KIc = 2

Large Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000

Flaw Size c (microm)

KIc = 1

KIc = 2

1) httpwwwndt-edorgEducationResourcesCommunityCollegeMaterialsMechanicalFractureToughnesshtm2) V Ravi S Firdosy T Caillat B Lerch A Calamino R Pawlik M Nathal A Sechrist J Buchhalter and S Nutt Mechanical Properties of Thermoelectric

Skutterudites Proceedings of the American Institute of Physics Conference Space Technology and Applications International Forum (2008) February 10-14

Albuquerque NM

1)

a is the crack length for edge cracks

or one half crack length for internal cracks

2)KIc = 11 - 22 Mpam12

GM RampD Thermoelectrics Team Researchers

Gregory P MeisnerJihui YangMike Reynolds

PostdocsXun Shi Jung Cho Zuxin Ye

Engineering OperationsKevin RoberJohn Manole

Gov ContractsEd Gundlach Amanda DemitrishRick Leach

ManagementJan HerbstMark Verbrugge

GMPT Integration amp TestingGreg Prior (Retired)

Joshua Cowgill

CollaboratorsSubcontractors Marlow Industries Jeff Sharp Jim Bierschenk Josh Moczygemba

and Alan Thompson Oak Ridge National Laboratory Hsin Wang Andy Wereszczak University of Nevada Las Vegas Changfeng Chen Yi Zhang

Future Tech Francis Stabler

Heat Technology Inc

Emcon (Faurecia)

Shanghai Institute of Ceramics Lidong Chen

University of Michigan Ctirad Uher

University of South Florida George Nolas

Brookhaven National Laboratory Qiang Li

Michigan State University Don Morelli

General Electric Global Research Todd Anderson Peter DeBock

AcknowledgementsUS Department of Energy Grant DE-FC26-04NT 42278

John Fairbanks (DOE) Carl Maronde (NETL)US Department of Energy Assistant Secretary for Energy Efficiency and Renewable Energy Office of Vehicle Technologies as part of the Propulsion Materials Program under contract DE-

AC05-00OR22725 with UT-Battelle LLC

Skutterudite a CoAs3 mineral found near Skutterud Norway in 1845 and compounds with the same crystal structure (body-centered cubic Im3 Oftedal I (1928) Zeitschrift fuumlr Kristallographie 66 517-546) are known as ldquoskutteruditesrdquo

W Jeitschko D J Braun Acta Crystallogr 1977 B33 3401 Filled Skutterudite Crystal Structure for LaFe4P12

Filled Skutterudites Technologically Important and Scientifically Fascinating Materials

Co As

Co4As12

Empty cages

Fracture and Failure in Ceramics Battling Flaws

Fracture toughness KIC =Yσradicaπ Mode I fracture Y is a crack shape factor and is larger for surface and edge cracks

Small Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 20 40 60 80 100

Flaw Size c (microm)

KIc = 1

KIc = 2

Large Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000

Flaw Size c (microm)

KIc = 1

KIc = 2

1) httpwwwndt-edorgEducationResourcesCommunityCollegeMaterialsMechanicalFractureToughnesshtm2) V Ravi S Firdosy T Caillat B Lerch A Calamino R Pawlik M Nathal A Sechrist J Buchhalter and S Nutt Mechanical Properties of Thermoelectric

Skutterudites Proceedings of the American Institute of Physics Conference Space Technology and Applications International Forum (2008) February 10-14

Albuquerque NM

1)

a is the crack length for edge cracks

or one half crack length for internal cracks

2)KIc = 11 - 22 Mpam12

Skutterudite a CoAs3 mineral found near Skutterud Norway in 1845 and compounds with the same crystal structure (body-centered cubic Im3 Oftedal I (1928) Zeitschrift fuumlr Kristallographie 66 517-546) are known as ldquoskutteruditesrdquo

W Jeitschko D J Braun Acta Crystallogr 1977 B33 3401 Filled Skutterudite Crystal Structure for LaFe4P12

Filled Skutterudites Technologically Important and Scientifically Fascinating Materials

Co As

Co4As12

Empty cages

Fracture and Failure in Ceramics Battling Flaws

Fracture toughness KIC =Yσradicaπ Mode I fracture Y is a crack shape factor and is larger for surface and edge cracks

Small Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 20 40 60 80 100

Flaw Size c (microm)

KIc = 1

KIc = 2

Large Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000

Flaw Size c (microm)

KIc = 1

KIc = 2

1) httpwwwndt-edorgEducationResourcesCommunityCollegeMaterialsMechanicalFractureToughnesshtm2) V Ravi S Firdosy T Caillat B Lerch A Calamino R Pawlik M Nathal A Sechrist J Buchhalter and S Nutt Mechanical Properties of Thermoelectric

Skutterudites Proceedings of the American Institute of Physics Conference Space Technology and Applications International Forum (2008) February 10-14

Albuquerque NM

1)

a is the crack length for edge cracks

or one half crack length for internal cracks

2)KIc = 11 - 22 Mpam12

Fracture and Failure in Ceramics Battling Flaws

Fracture toughness KIC =Yσradicaπ Mode I fracture Y is a crack shape factor and is larger for surface and edge cracks

Small Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 20 40 60 80 100

Flaw Size c (microm)

KIc = 1

KIc = 2

Large Size Range - Allowable Stress vs Flaw Size

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000

Flaw Size c (microm)

KIc = 1

KIc = 2

1) httpwwwndt-edorgEducationResourcesCommunityCollegeMaterialsMechanicalFractureToughnesshtm2) V Ravi S Firdosy T Caillat B Lerch A Calamino R Pawlik M Nathal A Sechrist J Buchhalter and S Nutt Mechanical Properties of Thermoelectric

Skutterudites Proceedings of the American Institute of Physics Conference Space Technology and Applications International Forum (2008) February 10-14

Albuquerque NM

1)

a is the crack length for edge cracks

or one half crack length for internal cracks

2)KIc = 11 - 22 Mpam12

Strength-Limiting Flaw Classification For Brittle Materials

Bi-DimensionalHybrid Flaw

PoresPorous Regions

Large GrainsAgglomerates

Inclusions

Volume Type(3-Dimensional)

Machining DamagePitting

Handling DamageOxidation

Surface Type(2-Dimensional)

Edge chipping

Edge Type(1-Dimensional)

Edge

100 microm

2a

Volume

20 microm

2a

Surface

50 microm

2a

Φ = e-(σσο)^M

Fracture and Failure in Ceramics A Statistical ProblemThe Weibull Modulus (M) is a measure of the scatter of data in a Weibull distribution

It is analogous to the standard deviation in a Gaussian distribution

For ceramic fracture and failure M is a measure of the flaw population distribution

Controlling fracture is controlling or eliminating the outliers in the flaw population

Thermal Shock Resistance Parameter For

Thermoelectric Materials in Operation

Q

Q

x κ

T2

T1

Steady State Heat Flow Q

Temperature Gradient as a result of

heat flow

Stress as a result of a temperature

gradient

Maximum allowable heat Flow

Of Concern is the Long Term Survival of the TE Legs Under Normal

Operating Conditions

ECTER Tens

Therm bullminus

=κνσ )1(

RTherm =σTens =

ν =κ =

CTE =E =

Thermal shock resistance parameter (the larger the better)Tensile stress or strengthPoissonrsquos ratioThermal conductivityCoefficient of thermal expansionElastic modulus

Kingery J Am Cer Soc 383-15 (1955)

aYKIc

Tens =σ

KIc =Y =a=

Fracture toughnessCrack shape factorGriffith flaw size

Must seek to minimize cTensile Strength ltlt Compressive StrengthManage tensile stress for conservative design

Griffith Criterion

FEA finds that stress is particularly concentrated at

edges and corners under 500 oC thermal gradient

Thermal Stress Experienced by a TE Module During

OperationThe stresses calculated by FEA for a module are quite

sensitive to

(a)CTE E inputs

(b)Boundary Conditions (these

presented here are very strenuous)

When these mismatched CTErsquos (stainless

steel contacts and Al2O3 insulators are

accounted for tensile stresses in the legs

climb to greater than 300 MPa via FEA

modeling

1) O M Jadaan and A A Wereszczak ldquoProbabilistic Design Optimization and Reliability Assessment of High Temperature Thermoelectric Devicesrdquo Ceramic

Engineering and Science Proceedings 29 [3] 157-172 (2008)

Elastic Modulus and Poissonrsquos ratio were

determined by Resonant Ultrasound Spectroscopy

(RUS)

Elastic Property Measurements

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Φ = e-(σσο)^M

Fracture and Failure in Ceramics A Statistical ProblemThe Weibull Modulus (M) is a measure of the scatter of data in a Weibull distribution

It is analogous to the standard deviation in a Gaussian distribution

For ceramic fracture and failure M is a measure of the flaw population distribution

Controlling fracture is controlling or eliminating the outliers in the flaw population

Thermal Shock Resistance Parameter For

Thermoelectric Materials in Operation

Q

Q

x κ

T2

T1

Steady State Heat Flow Q

Temperature Gradient as a result of

heat flow

Stress as a result of a temperature

gradient

Maximum allowable heat Flow

Of Concern is the Long Term Survival of the TE Legs Under Normal

Operating Conditions

ECTER Tens

Therm bullminus

=κνσ )1(

RTherm =σTens =

ν =κ =

CTE =E =

Thermal shock resistance parameter (the larger the better)Tensile stress or strengthPoissonrsquos ratioThermal conductivityCoefficient of thermal expansionElastic modulus

Kingery J Am Cer Soc 383-15 (1955)

aYKIc

Tens =σ

KIc =Y =a=

Fracture toughnessCrack shape factorGriffith flaw size

Must seek to minimize cTensile Strength ltlt Compressive StrengthManage tensile stress for conservative design

Griffith Criterion

FEA finds that stress is particularly concentrated at

edges and corners under 500 oC thermal gradient

Thermal Stress Experienced by a TE Module During

OperationThe stresses calculated by FEA for a module are quite

sensitive to

(a)CTE E inputs

(b)Boundary Conditions (these

presented here are very strenuous)

When these mismatched CTErsquos (stainless

steel contacts and Al2O3 insulators are

accounted for tensile stresses in the legs

climb to greater than 300 MPa via FEA

modeling

1) O M Jadaan and A A Wereszczak ldquoProbabilistic Design Optimization and Reliability Assessment of High Temperature Thermoelectric Devicesrdquo Ceramic

Engineering and Science Proceedings 29 [3] 157-172 (2008)

Elastic Modulus and Poissonrsquos ratio were

determined by Resonant Ultrasound Spectroscopy

(RUS)

Elastic Property Measurements

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Thermal Shock Resistance Parameter For

Thermoelectric Materials in Operation

Q

Q

x κ

T2

T1

Steady State Heat Flow Q

Temperature Gradient as a result of

heat flow

Stress as a result of a temperature

gradient

Maximum allowable heat Flow

Of Concern is the Long Term Survival of the TE Legs Under Normal

Operating Conditions

ECTER Tens

Therm bullminus

=κνσ )1(

RTherm =σTens =

ν =κ =

CTE =E =

Thermal shock resistance parameter (the larger the better)Tensile stress or strengthPoissonrsquos ratioThermal conductivityCoefficient of thermal expansionElastic modulus

Kingery J Am Cer Soc 383-15 (1955)

aYKIc

Tens =σ

KIc =Y =a=

Fracture toughnessCrack shape factorGriffith flaw size

Must seek to minimize cTensile Strength ltlt Compressive StrengthManage tensile stress for conservative design

Griffith Criterion

FEA finds that stress is particularly concentrated at

edges and corners under 500 oC thermal gradient

Thermal Stress Experienced by a TE Module During

OperationThe stresses calculated by FEA for a module are quite

sensitive to

(a)CTE E inputs

(b)Boundary Conditions (these

presented here are very strenuous)

When these mismatched CTErsquos (stainless

steel contacts and Al2O3 insulators are

accounted for tensile stresses in the legs

climb to greater than 300 MPa via FEA

modeling

1) O M Jadaan and A A Wereszczak ldquoProbabilistic Design Optimization and Reliability Assessment of High Temperature Thermoelectric Devicesrdquo Ceramic

Engineering and Science Proceedings 29 [3] 157-172 (2008)

Elastic Modulus and Poissonrsquos ratio were

determined by Resonant Ultrasound Spectroscopy

(RUS)

Elastic Property Measurements

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Of Concern is the Long Term Survival of the TE Legs Under Normal

Operating Conditions

ECTER Tens

Therm bullminus

=κνσ )1(

RTherm =σTens =

ν =κ =

CTE =E =

Thermal shock resistance parameter (the larger the better)Tensile stress or strengthPoissonrsquos ratioThermal conductivityCoefficient of thermal expansionElastic modulus

Kingery J Am Cer Soc 383-15 (1955)

aYKIc

Tens =σ

KIc =Y =a=

Fracture toughnessCrack shape factorGriffith flaw size

Must seek to minimize cTensile Strength ltlt Compressive StrengthManage tensile stress for conservative design

Griffith Criterion

FEA finds that stress is particularly concentrated at

edges and corners under 500 oC thermal gradient

Thermal Stress Experienced by a TE Module During

OperationThe stresses calculated by FEA for a module are quite

sensitive to

(a)CTE E inputs

(b)Boundary Conditions (these

presented here are very strenuous)

When these mismatched CTErsquos (stainless

steel contacts and Al2O3 insulators are

accounted for tensile stresses in the legs

climb to greater than 300 MPa via FEA

modeling

1) O M Jadaan and A A Wereszczak ldquoProbabilistic Design Optimization and Reliability Assessment of High Temperature Thermoelectric Devicesrdquo Ceramic

Engineering and Science Proceedings 29 [3] 157-172 (2008)

Elastic Modulus and Poissonrsquos ratio were

determined by Resonant Ultrasound Spectroscopy

(RUS)

Elastic Property Measurements

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

FEA finds that stress is particularly concentrated at

edges and corners under 500 oC thermal gradient

Thermal Stress Experienced by a TE Module During

OperationThe stresses calculated by FEA for a module are quite

sensitive to

(a)CTE E inputs

(b)Boundary Conditions (these

presented here are very strenuous)

When these mismatched CTErsquos (stainless

steel contacts and Al2O3 insulators are

accounted for tensile stresses in the legs

climb to greater than 300 MPa via FEA

modeling

1) O M Jadaan and A A Wereszczak ldquoProbabilistic Design Optimization and Reliability Assessment of High Temperature Thermoelectric Devicesrdquo Ceramic

Engineering and Science Proceedings 29 [3] 157-172 (2008)

Elastic Modulus and Poissonrsquos ratio were

determined by Resonant Ultrasound Spectroscopy

(RUS)

Elastic Property Measurements

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Elastic Modulus and Poissonrsquos ratio were

determined by Resonant Ultrasound Spectroscopy

(RUS)

Elastic Property Measurements

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

CLTE = 115 ppm to 450oC CLTE = 101 ppm

Sample type E (Gpa)

G (Gpa)

B(Gpa)

ν VT(x103 ms)

VL(x103 ms)

n-type 135 55 80 020 269 450p-type 123 50 73 020 256 429

Summary of Elastic Properties

n-type La005Ba007Yb008Co400Sb1202

p-type Ce030Co257Fe143Sb1198

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

An All-Alumina High-Temperature ldquo3-Pointrdquo Bend Fixture Was Developed and

Used For Strength Testing

This is actually a ldquoBall on Two Rollersrdquo 3-Point Bend Fixture The self aligning nature allows high sample throughput

The close contact between the ball and cylinder make for a virtually oxygen free environment

Comparison of the outer-fiber tensile stress as a function of 3-point-bendforce for the analytical case (black) its finite element analysis (blue) andthe case for the bend fixture shown in Fig 3 (red) Difference was small(~3) so the analytical expression was used to estimate failure stresses inthis study

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Preparation of Filled Skutterudites

Induction melting of the

elements Sealed vessel

required for n-type

The melt is annealed at

750 oC for 1 week

Annealed charge was

milled by HEBM and

sieved to 60 microm

Diced by low speed

diamond saw

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Prototypical Graphite Die and to Create Test Specimens

SPS

Sintering

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

bull Edge type (1-D) extrinsic typically a machining effect

ndash eg edge chipping or chamfering (TE elements can be cast as cylinders)

bull Surface type (2-D) can be extrinsic or intrinsic

ndash eg machining damage reaction layer oxidation layer

bull Volume type (3-D) usually intrinsic

ndash eg pores agglomerates large grains etc

ndash Improvement in material processing will lessen their effect

Room Temperature Strength of Diced and NSS Test Coupons

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

The characteristic strength of skutterudites is twice that of the PbTe based materials with similar microstructures that we have tested Samples prepared by

SPS and cut parallel to pressing axisJR Salvador et al Philo Mag 89 1517 (2009)

Strength as a Function of Temperature

0

50

100

150

200

250

0 5 10 15 20 25

Yb035Co4Sb12 Skutterudite N-typeMmFe4Sb12 Skutterudite P-typePbTe (PbI2 doped) N-typePbTe (Na doped) P-type

Cha

ract

eris

tic S

tren

gth

(MP

a)

Weibull Modulus

95 ConfidenceRatio Rings

N-type SkutteruditeP-type SkutteruditeN-type PbTe (PbI2)P-type PbTe (Na)

(n-type)

Yb022Co4Sb1208

Ce086Co102Fe298Sb1197

Results are for small scale reactions (60g) and frac12 in billets

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

~1 kg of Ce030Co252Fe148Sb12

(n-type)

25 25 160 106 154 17200 14 156 67 148 20300 15 154 96 147 16400 15 139 64 130 25500 15 143 88 135 17

(p-type)

25 20 135 242 132 8200 14 133 108 127 21300 15 133 150 129 11400 15 138 284 136 7500 15 147 134 142 12

|------- 2- parameter Weibull -------| |-- Gaussian --|Characteristic Ave Std

Temp of Strength Weibull Strength DevMaterial (degC) Tests (MPa) Modulus (MPa) (MPa)

Values in parenthesis = plusmn 95 confidence interval σ (MPa)110 120 130 140 150 160 170 180

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

500 oC p-type500 oC n-type

σ (MPa)100 120 140 160 180 200

Pro

babi

lity

of S

urvi

val

00

02

04

06

08

10

12

25 oC p-type25 oC n-type

Machined by Al Thompson Marlow Industries

25 oC

500 oC

La005Ba007Yb008Co400Sb1202

Ce030Co257Fe143Sb1198

Incremental Improvement in Characteristic Strength

Results are for large scale (10 kg batches and 30 cm (80 g

billets)

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Volume-Type Flaw (SKN-3)

1 mm

100 microm1 mmT

C

T

T

Surface-Type Flaw (SKN-17)

1 mm

1 mm

Representative Flaws of Large Scale Materials

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

N-TypePolished Surface

Fracture Surface

P-Type

Fracture Surface

Polished Surface

Source of Persistently Lower Weibull Modulus in n-Type Materials

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Conclusions Demonstrated incremental improvement in the fracture strength of

skutterudite materials by flaw mitigation Characteristic strengths for skutterudites are twice that of PbTe and Bi-Te based materials

Fracture strength of both n- and p-type skutterudites is nearly temperature independent and characteristic strength magnitude is encouraging from a durability standpoint

Weibull modulus (increased uncertainty) of the n-type material still lags p-type due to the larger distribution of volume flaws (larger grain distribution)

There is significant room for improvement in the control of microstructure by optimizing powder metallurgical techniques

Thank You

Thank You