Functionally Graded Materials Between Ferritic and Austenitic Alloys using

Additive ManufacturingJ.S. Zuback, T.A. Palmer and T. DebRoy

AWS Fabtech Conference, Chicago, IL, November 6th, 2017

U.S. Department of Energy

Grant number DE-NE0008280

Advisor: S.A. David, Oak Ridge National Lab

TPOC: Dr. S. Sham, US DOE

In collaboration with:

Dr. W. Zhang, Ohio State University

Dr. Z. Feng, Oak Ridge National Lab

Materials in nuclear power plants

2

T. Allen et al., Materials Today, 13(12) (2010) 14-23.

Dissimilar metal welds between ferritic to austenitic

materials in nuclear power plants

June 6-7th, 2017 3

Laha, Metall. Mater. Trans A, 2001

Problem: Carbon diffuses from the ferritic steel towards the austenitic alloy

Consequence: Carbon depleted zone in steel Poor creep performance

Huang, Metall. Mater. Trans A, 1998

Courtesy of creep testing at ORNL

Solution: Reduce carbon diffusion to

improve creep performance

4June 6-7th, 2017

ApproachThermodynamic and kinetic

models for designing

composition profiles that

minimize carbon diffusion

Fabricate transition joints by

additive manufacturing

Test and characterize

fabricated joints

What causes carbon diffusion?

5

Wt

% C

arb

on

Distance

Constant concentration

2.25Cr-1Mo

SteelAlloy 800H

Car

bon

Chem

ical

Pote

nti

alDistance

2.25Cr-1Mo

SteelAlloy 800H

Driving force

for diffusion

Uniform carbon concentration Chemical potential gradient

𝐽𝑖 = −𝐷𝑖𝑑𝑐𝑖𝑑𝑥

𝐽𝑖 = −𝐿𝑖𝑇

𝑑µ𝑖𝑑𝑥

Depends on

alloying

elements

Fick’s first law of

diffusion

Inconel filler metal

June 6-7th, 2017

Thermodynamic modeling

6June 6-7th, 2017

Distance [mm]

Chem

ical

Pote

nti

al[k

J/m

ol]

0 0.1 0.2 0.3 0.4 0.5-70

-65

-60

-55

-50

-45

-40

2.25Cr-1MoSteel

IN82 FM

T = 773 K

Large thermodynamicdriving force

Goal: Reduce carbon chemical potential gradient

Distance [mm]

Chem

ical

Pote

nti

al[k

J/m

ol]

0 2 4 6 8 10-80

-70

-60

-50

-40

T = 773 K

Gradual change

Low driving force

Dissimilar metal weld Functionally graded material

Kinetic modeling

7June 6-7th, 2017

Goal: Predict carbon migration after years of service

Distance [mm]

wt%

C

0 0.1 0.2 0.3 0.4 0.5

0.1

0.2

0.3

0.4

0.5Initial

2 years

10 years

20 years

2.25Cr-1MoSteel

IN82 FM

0.080 wt% C

T = 773 K

0.446 wt% C

Dissimilar metal weld

Distance [mm]

wt%

C

0 2 4 6 8 100

0.1

0.2

0.3

0.4

0.5T = 773 K

1 2 30.08

0.1

0.12Initial1 year2 years10 years20 years

0.090 wt% C

0.106 wt% C

Graded transition joint

Diffusion model validation

8

Testing calculations against independent experimental data

Comparison

• Good agreement between

calculated and measured

carbon diffusion profiles

• Accumulation of carbon in

austenitic material along

interface

• Carbon depletion in ferritic

material along interface

• Confidence to apply the

model to functionally graded

materials

Experimental data from M.L. Huang, L. Wang, Metall. Mater. Trans A, 29 (1998) 3037-3046

Fabrication of functionally graded

specimens using additive manufacturing

9

Directed energy deposition

Photo courtesy of PSU ARL

Wt% of element

Powder Al Cr Fe Mn Ni Si Ti C O P S

Pyromet®

800 0.38 21.0 Bal. 0.88 34.0 0.62 0.41 0.095 0.015 0.003 0.006

Fe - - Bal. - - - - 0.005 - - 0.016

Cr <0.01 Bal. 0.07 - - <0.01 - 0.004 0.43 0.0013 0.023

Pre-blended powders were

entered into the powder

feeder at increments of 10%

800H

Controlling chemical composition profilesCan composition be controlled using pre-blended powders?

Excellent

agreement

Using pre-alloyed powder

Using blended powders

Fe-35Ni-21Cr

Fe-35Ni-21CrFe-35Ni-21Cr

Fe-35Ni-21Cr FeCr

FeFe-35Ni-21Cr

Fe-35Ni-21Cr

???

Microstructure Prediction

11

Creq

Ni eq

0 10 20 30 40

10

20

30

Austenite

Ferrite

Martensite

A + F

A+

M+

F

A + M

F + MF + M

Substrate

29 mm above

substrateSchaeffler Constitution

Diagram

• Commonly used when welding low

alloy steels, austenitic stainless

steels and dissimilar alloys

• Relates the composition to

microstructure based on common

cooling rates found in welding

• Measured chemical compositions

from EPMA were used to calculate

Nieq and Creq

• Here, a martensitic to austenitic

microstructure is predicted

Height [mm]

Vic

ker

sH

ardn

ess

[HV

]

0 10 20 300

100

200

300

400

A + M AusteniteMartensite

Microstructural characterization

12

Significant changes in microstructure and microhardness are observed

Is a full composition gradient necessary?

13

Distance [mm]

Chem

ical

Pote

nti

al[k

J/m

ol]

0 2 4 6 8 10-80

-70

-60

-50

-40

T = 773 K

Full gradient

Fu

ll g

rad

ien

tP

art

ial gra

die

nt No loss in

functionality

Height [mm]

Vic

ker

sH

ardn

ess

[HV

]

0 10 20 300

100

200

300

400

A + M AusteniteMartensite

No changes in

microstructure

No benefits in

compositional

grading once the

microstructure

becomes fully

austenitic

Soft zone formation near baseplate

14

Time [s]

Tem

per

atu

re[K

]

10-1

100

101

102

103

104

105

600

800

1000

1200

10 K/s1000 K/s 100 K/s

Pearlite

Bainite

20 wt% 800H

10 wt% 800H

10 wt% 800H

20 wt% 800H

Continuous cooling diagram

Bainite is the microconstituent most likely to

form during cooling other than martensite

What if the soft zone is excluded?

Increase in the

carbon chemical

potential near the

deposit/baseplate

interface

More carbon

diffusion after

years of service

compared to a

gradient starting

at 10% 800H

Secondary phase formationExperimental observations

SEM with EDS composition maps

Ti-rich particles near grain boundaries/interdendritic regions

Secondary phase formation

Temperature [K]

Mas

sper

cent

of

phas

e

400 600 800 1000 1200 140010

-3

10-2

10-1

100

101

M23

C6

MC

Temperature [K]

Mas

sfr

acti

on

of

phas

e1200 1400 1600 1800

10-4

10-3

10-2

10-1

100

Austenite

MC

Calculations

Equilibrium secondary phases Scheil-Gulliver solidification

Although both M23C6 and MC carbides are stable at operating temperatures (773K),

only MC is predicted to form when considering solidification

Summary

• Carbon diffusion across dissimilar metal welds between ferritic and austenitic material leads to premature failure

• Functionally graded materials drastically reduce carbon migration by reducing the carbon chemical potential

• An appropriately graded transition joint will take approximately five times longer to deplete the same amount of carbon as a dissimilar weld

• Kinetic considerations are essential for the design of functionally graded materials

• Tradeoffs exist between microstructure and designed functionality

June 6-7th, 2017

Acknowledgements

19