Manufacturing Metallic Parts with
Designed Mesostructure
via
Three-Dimensional Printing of Metal
Oxide Powder
Christopher B. Williams
Rapid Prototyping and Manufacturing Institute
http://www.rpmi.marc.gatech.edu
David W. Rosen
Woodruff School of Mechanical Engineering
Georgia Institute of Technology
Atlanta, Georgia
Systems Realization Laboratory
http://www.srl.gatech.edu
August 08, 2007
Manufacturing Metallic Parts with
Designed
Mesostructure
Parts of Designed
Mesostructure:
•
•
•
•
•
What are they?
Why are they of interest? via
How are they manufactured?
What are the limitations of current manufacturing processes?
What are potential
areas for Powder
improvement?
Oxide
Three-Dimensional Printing of Metal
Christopher
B. Williams
• What is our answer?
Rapid Prototyping and Manufacturing Institute
– 3DP of metal-oxide ceramic green part followed
by posthttp://www.rpmi.marc.gatech.edu
David
W. inRosen
processing
a reducing atmosphere
• Why
3DP
metal-oxide
powders?
Woodruff
School
of of
Mechanical
Engineering
• Georgia
Preliminary
- characteristic cellular
Instituteresults
of Technology
Atlanta,
– Thin
wallsGeorgia
– Angled trusses
– Small channels
material geometry:
Systems Realization Laboratory
http://www.srl.gatech.edu
August 08, 2007
3
Low-Density Cellular Materials
Metallic Foams
Lattice Block Material
Benefits:
• High strength
• Low mass
• High stiffness
• Acoustic & vibration
dampening
Linear Cellular Alloys
• Strain isolation
• Energy absorption
• Excellent heat transfer
ability
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
Cellular Material Applications:
Designed Mesostructure
4
(V. Wang, 2006)
Acetabular Cup
(Fleck & Deshpande, 2004)
(C. Seepersad, 2005)
Combustor Liner
(H. Muchnick, 2007)
Blast Resistant Panel
Robot Arm
(V. Wang, 2004)
Georgia Institute of Technology
Systems Realization Laboratory
5
Cellular Material Manufacturing
Stochastic Cellular Material Manufacturing
Ordered Cellular Material Manufacturing
(Hydro / Alcan / Combal Process)
(Honeycomb via Crimping & Stamping)
Existing cellular material manufacturing techniques are severely limited:
1. Part Macrostructure
3. Non-repeatable results
2. Materials
4. Limited mesostructure topology
Williams, C. B., F. M. Mistree, D. W. Rosen, 2005, “Investigation of Additive
Manufacturing Processes for the Manufacture of Parts with Designed
Mesostructure,” ASME IDETC, DETC2005/DFMLC-84832
Georgia Institute of Technology
Systems Realization Laboratory
6
Direct Metal Additive Manufacturing
Limitations
(wrt cellular materials)
x
x
SLS
DMLS
SLM
x
EBM
3DP
Processes
Laser Engineered Net Shaping
MJS
EDSSM
LENS
SDM
x
x
x
x
x
x
x
x
LOM
x
x
x
CAMLEM
x
x
UOC
Electron Beam Melting
x
x
x
x
x
x
x
x
x
x
Williams, C. B., F. M. Mistree, D. W. Rosen, 2005, “Investigation of Additive
Manufacturing Processes for the Manufacture of Parts with Designed
Mesostructure,” ASME IDETC, DETC2005/DFMLC-84832
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Georgia Institute of Technology
Systems Realization Laboratory
7
Direct Metal Additive Manufacturing
Electron
Beam Melting
Direct Metal
Laser Sintering
Selective
Laser Melting
http://www.mcp-group.com/rpt/rpttslm_1.html
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
3DP of Metal Oxide Powder +
Sintering in Reducing Atmosphere
8
Step Two
Additive Manufacturing via 3DP
Step One
Spray-Drying
Paste
Spray
Preparation
Drying
Oxide
Powders
Additives
Binder
Honeycomb Extrusion
Drying
Spraying
Compounding
Step Three
Direct Reduction
H2
Finished Metal Part
© Christopher B. Williams
Direct Reduction
Georgia Institute of Technology
Systems Realization Laboratory
9
Reduction & Sintering of Metal Oxides
Step Two
Shape Fabrication
Step One
Paste Preparation
Oxide
Powders
H2O
Additives
Honeycomb Extrusion
Drying
Fe3O 4 + 4H 2 
 3Fe + 4H 2 O
Co3O4 + 4H 2 
 3Co + 4H 2 O
Compounding
NiO + H 2 
 Ni + H 2 O
Flexible Die Design
Step Three
Maraging Steel: Fe 18.5Ni
8.5Co 5Mo
Direct Reduction
H2
Finished Metal Part
Direct Reduction
J. Cochran, T. Sanders, D. McDowell - GT Lightweight Structures Group
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
10
Reduction & Sintering of Metal Oxides
• Materials:
– Fe, Cu, Co, Cr, Ni, Mo, W, etc.
– Maraging / Stainless steel, Iconel, Super Invar
– No Al or Ti
• Cost effective:
– Metal oxides 10x cheaper than metal counterpart
• Safe:
– Non-carcinogenic
– Chemically stable
• Geometric considerations:
– Need open access to interior
– Minimize thickness variation
– Large shrinkage upon processing
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
11
Reduction & Sintering of Metal Oxides
Step Two
Two
Step
ShapeManufacturing
Fabrication
Additive
Step One
Paste Preparation
Oxide
Powders
H2O
Additives
CAD
file
Create Extrusion
Honeycomb
Patterning
Slice CAD file
into layers
?
data
Store material
Provide
support
Control
Provide new
material
Pattern
Provide energy
Compounding
Drying
Flexible Die Design
Post-Process
Part
Legend
materials
energy
signals
system
boundary
Step Three
Direct Reduction
H2
Finished Metal Part
Direct Reduction
J. Cochran, T. Sanders, D. McDowell - GT Lightweight Structures Group
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
12
Design of an Additive Manufacturing
Process
Design Task:
• To design an AM process for the realization of metal-oxide ceramic green
cellular parts suitable for post-processing in a reducing atmosphere
• Specific requirements of cellular materials:
– Features  250 mm
– Cell sizes of 0.5 – 2 mm
– Multiple materials
– Comparable speed and cost
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
Design and Development of an Additive
Manufacturing Process
Clarification of Task
D/W
D
D
D
D
W
D
W
W
D
W
D
D
D
D
D
W
W
D
W
D
D
W
D
Requirement
Geometry
Able to process any macrostructure geometry
Able to process complex geometry (overhangs
and internal voids)
Able to process small cell sizes (0.5 – 2 mm)
Build small wall thickness (50 – 300 mm)
Minimize amount of effort required to adapt to a
new material
Material
Able to process multiple materials (steel, iron,
aluminum, copper, etc.)
Able to process standard working material
Production
Maximize deposition rate (> 10 cm3/hr)
Store
Build envelope is 305 x 305 x 305 mm or larger
Material
Does not require additional post-processing
Quality Control
Parts are > 98% dense
Pattern
Material properties are comparable to standard
Minimize surface roughness before finishing (<
Provide
0.02 mm Ra)
Energy
Maximize accuracy (> +/- 0.05 mm)
Minimize z-resolution (< 0.1 mm)
Operation
Provide
Does not require special operating environment
New
Minimize operator interaction
Material
Recycling
Minimize environmental impact by minimizing
wasted material
Reusable wasted material
Support
Costs
Sub-Functions
13
Minimize cost of technology
Minimize cost of maintenance
Minimize cost of material
Easily scaled for large applications
Requirements List
Conceptual Design
Solutions
Powder
Two Phase
Powder
Powder
Coated w/
Binder
Material
(1D)
Energy
(1D)
Both
(1D)
Sinter
Melt
Clad
SLS
ECONOMICS
Technology Cost
0
Recoat
Score Recoat by
0 by
Spreading
Normalized
Score Spraying
0.00
TIME
Deposition rate
1
Score
1
Breakable
5-axisNormalized
Material
Score Support
1.00
Deposition
Bed
PERFORMANCE
Material
min. feature size
-1
complex geometry
0
surface finish
-1
Score
-2
Normalized Score
0.33
MATERIALS
Solids Loading
-1
Material properties
-1
Material selection
1
Score
-1
Normalized Score
0.00
Powder /
Binder
Suspension
Wire / Rod
Gas
Material
(2D)
Energy
(2D)
Both
(2D)
SLA
Bind
MJS
Cut
EFF
3DP
Chem.
Reaction
IJP-a
Tape / Sheet
IJP-w
EP
LOM
0
1
0Recoat by
1
0.00Dipping
1.00
1
1
1
Direct
Recoat
by 1
1
1
Material
Layer 1.00
1.00
1.00
Addition
1
1
1.00
1
1
1.00
1
1
1.00
0
-1
0 Trusses
-1
Thin
0.50of Build
0.00
Material
0
-1
0
1
0
-1
0
-1
1.00
0.67
-1
-1
-1
-1
-1 Organic
-1
Dissolvable
0.00
0.00
Support 0.00 Support
Material
Material
-1
-1
-1
1
0
1
-1
-1
0
-1
-2
0
0.67
0.33
1.00
1
1
1
1
1.00
1.00
No
Support
1
1
1.00
-1
1
0
0
1.00
-1
1
-1
-1
0.67
-1
-1
-1
-3
0.00
-1
0
1
0
0.25
0
-1
1
0
0.25
1
-1
1
1
0.50
0
0
0
0
0.25
0
-1
1
0
0.25
0
-1
1
0
0.25
1
1
1
3
1.00
1
1
1
3
1.00
Embodiment Design
Morphological Matrix &
Preliminary Selection Decision Support Problem
Three Dimensional Printing
Williams, C. B., F. M. Mistree, D. W. Rosen, 2005, “Towards the Design of a Layer-Based
Additive Manufacturing Process for the Realization of Metal Parts of Designed
Mesostructure,” Solid Freeform Fabrication Symposium, pp. 217-230.
Georgia Institute of Technology
Systems Realization Laboratory
14
Three Dimensional Printing
•
•
•
•
~100 mm feature size
Two-dimensional deposition
Cost effective, scalable technology
50% solids loading in green part
•
•
© Christopher B. Williams
Unable to spread fine particle
sizes
Powder bed leads to trapped
unbound powder
Georgia Institute of Technology
Systems Realization Laboratory
15
Spray Drying
Drying Air
Spray-dried powder
Exhaust
• Fine particles (1-5 mm) in
granule form (30-50 mm)
• Spherical and flowable
• Smaller primitives
• Modular binder/powder
combo
Powder / Binder
Suspension
Granules
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
3DP of Metal Oxide Powder +
Sintering in Reducing Atmosphere
16
1400
Temperature (C)
Step One
Spray-Drying
Paste
Spray
Preparation
Drying
1200
Maraging
Oxide
PVA:
Additives
Binder
Steel
Oxide
Powders
2 wt%
1000
Powder
4 wt%
800
Sintering Step Two
Additive
1350 C Manufacturing via 3DP
Honeycomb Extrusion
ZCorp Z402 printer
Drying
• ZB7 binder
• Layer thickness:
100 mm
Reduction
850 C
600
Spraying
Compounding
• Core saturation: 1.75
400
Step Three
Direct Reduction
Binder burnout
450 C
200
H2
0
0.00
3.54
4.04
6.26 Part
Finished
Metal
14.26
Time (hr)
© Christopher B. Williams
16.76
19.76
24.01
Direct Reduction
Georgia Institute of Technology
Systems Realization Laboratory
17
General Results
•
•
•
•
Fragile green parts
Linear shrinkage:
Relative density:
Internal open porosity:
© Christopher B. Williams
46%
65.3%
29.8%
Georgia Institute of Technology
Systems Realization Laboratory
18
Thin Wall Test
• 400 mm thin wall (sintered)
• Dependent on dpi of 3DP
machine
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
19
Channel Test
(C. Seepersad, 2005)
• 2 mm x 2 mm x 10 mm open channels
• 500 mm channels have been
successfully printed
• Channel size limited by powder removal
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
20
Angled Truss Test
(V. Wang, 2004)
t
LT

x
y
t
LT
x

sin  tan 
L
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
21
Angled Truss Test
•
•
•
•
2 mm diameter truss (1.08 mm sintered)
45o angle
0.328 mm layer overlap
2 mm wall / truss gap (green)
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
22
Angled Truss Test
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
23
Angled Truss Test
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
24
Spray Dried Powder Results
Granule binder
content
Deposited
binder
Relative
density
Open porosity
2 wt%
ZB7
65.3%
29.8%
4 wt%
ZB7
59.2%
36.4%
4 wt%
Solvent
64.3%
33.4%
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
25
Summary: Critical Analysis
•
•
•
•
•
Low sintered density
Poor surface finish
Fragile green part; difficult to de-powder
Cannot process Ti or Al
Cannot produce powder-filled cells
•
•
•
•
•
Scalable technology (parallel deposition)
Cost-effective (technology and material)
Modular binder / material combination
Able to process several materials and alloys
Successfully fabricated 400 mm walls, angled trusses,
small channels
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
26
Next Steps…
• Materials Characterization
– XRD phase analysis
– Tensile and bending tests
• Primitive formulation
modeling
• Alternatives for further
densification of green part
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
27
Acknowledgements
• NSF DMI-0522382
• NSF IGERT - 0221600
• Mr. Joe Pechin, Aero-Instant Spray Drying
Services
• Dr. Joe Cochran, Georgia Tech, Materials
Science and Engineering Department
• Michael Middlemas & Tammy McCoy
• Dr. Scott Johnston & Ben Utela
• Dr. Carolyn Seepersad
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
Thank you.
NSF Grant DMI-0085136
NSF IGERT-0221600
NSF Grant DMI-0522382
29
Supplemental Slides
Georgia Institute of Technology
Systems Realization Laboratory
30
Classification of Cellular Materials
LOW-DENSITY CELLULAR MATERIALS
Stochastic
Ordered
Periodic
• (Solid) metal foams
• Metal sponges
• Porous metals
• Hollow sphere foams
• Honeycomb (via
crimping/stamping)
• Lattice Block
Materials
Parts of Designed Mesostructure
Designed Mesostructure
• Linear Cellular Alloys
• Truss Structures (via
Additive Manufacturing)
(Mesostructure: 100mm – 10mm)
• a class of cellular structures wherein material is strategically placed
by a designer in order to achieve certain design objectives (i.e., low
mass, high strength, high stiffness, etc.)
• Pertains to a group of manufacturing processes that provide a
designer the freedom to prescribe mesostructure topology for a
design’s intent
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory
31
Addressing the Gap: Manufacturing Parts of
Designed Mesostructure
Primary Research Question:
How to manufacture three-dimensional, lowdensity, cellular metal structures while
maintaining designer freedom in the selection
of the material and the design of the part
It is proposed to design, embody, and analyze a
mesostructure
and
manufacturing
process
thatmacrostructure?
is capable of producing metallic
cellular materials and providing a designer the freedom to
specify material type, material composition, void morphology,
and mesostructure topology for any conceivable part
geometry.
Georgia Institute of Technology
Systems Realization Laboratory
32
Research Hypothesis
Step Two
Two
Step
ShapeManufacturing
Fabrication
Additive
Step One
Paste Preparation
Oxide
Powders
H2O
Additives
CAD
file
Create Extrusion
Honeycomb
Patterning
Slice CAD file
into layers
Primary Research Hypothesis:
?
data
Control
Drying
Three-dimensional,
cellular
Fe3Olow-density

3Fe +metal
4H 2 Ostructures of
4 + 4H 2 
any macrostructure,
material
can be
Co3O4 +mesostructure,
4H 2 
 3Coor
+ 4H
O
2
manufactured via layer-based additive manufacturing of
by
Nipost-processing
+ H 2O
Compounding NiO + H 2 
metal-oxide
ceramics followed
in a
Flexible Die Design
reducing
atmosphere.
Step Three
Maraging Steel: Fe 18.5Ni
8.5Co 5Mo
Direct Reduction
Store material
Provide
support
Provide new
material
Pattern
Provide energy
Post-Process
Part
Legend
materials
energy
signals
system
boundary
H2
Finished Metal Part
Direct Reduction
J. Cochran, T. Sanders, D. McDowell - GT Lightweight Structures Group
Georgia Institute of Technology
Systems Realization Laboratory
33
Linear Cellular Honeycombs
(via extrusion & reduction)
Step Two
Shape Fabrication
Step One
Paste Preparation
Cochran, McDowell, et al.
Oxide
Powders
H2O
Additives
Honeycomb Extrusion
Drying
Compounding
Flexible Die Design
Step Three
Direct Reduction
H2
Finished Metal Part
•
•
•
•
•
•
•
•
•
Direct Reduction
Metal-oxide paste is extruded through die and reduced to a metal part
Can process many different materials
Parts have excellent material properties
Oxide powders are cheaper & safer
Predictable, repeatable results
Interchangeable dies can be designed for specific design intent
Cells across cross-section need not be periodic
Excellent for multi-functional design (structural heat-exchangers)
Georgia Institute of Technology
Limited to linear extrusions
Systems Realization Laboratory
34
Metal via Reduction of Metal Oxides
•
•
•
•
•
•
•
•
•
•
•
•
Decouples cell geometry and material composition
Two
OneCo, Cr, N Cu, Mo, W,Step
ProcessedStep
Fe, Ni,
Mn, and
Shape Fabrication
Paste Preparation
Nb
AllowsOxide
for complex
cell shape, precise
cell Extrusion
Honeycomb
H2O Additives
Powdersand thin wall thicknesses (> 50 mm)
alignment,
Oxide particles are cheaper, safer, purer, and more
stable than metal counterparts
No other method can compare to its material
selection or mechanical properties
Drying
Compounding
Paste rheology can limit freedom
Flexiblecan
Die lead
Designto cracking and laminations
Debinding
Step Three
Shrinkage can cause warpage and dimensionalDirect
instability
Reduction
Material must be reducible at T < Tmelt (Al and Ti are difficult to introduce)
H
Structure must have high surface-to-volume ratio and open access to2interior to
survive reduction process; constant web-thickness is preferable
Finishedon
Metal
Part
Direct Reduction
Mechanical properties dependent
porosity
Creates only linear structures
Georgia Institute of Technology
Systems Realization Laboratory
35
Why Reduction of Metal Oxides?
•
•
•
•
•
Metal Oxide Powders vs. Metal Powders
•
•
•
Cheaper
Safer
Purer
Slurry
•
•
Heat Affected Zones
Recoating
Shrinkage
Material properties
Multiple materials
Note: process can only be used for geometry with constant crosssection
Georgia Institute of Technology
Systems Realization Laboratory
Principal Solution Selection:
Fused Deposition Modeling
36
Subperimeter
Voids
Lewis et al., 2003
4 Roads
Agarwala et al., 1996
Williams, C. B., F. M. Mistree, D. W. Rosen, 2005, “Towards the Design of a Layer-Based Additive Manufacturing Process for
the Realization of Metal Parts of Designed Mesostructure,” Solid Freeform Fabrication Symposium, pp. 217-230.
Conceptual Design
Selection
Augmentation
Analysis
Georgia Institute of Technology
Systems Realization Laboratory
Principal Solution Selection:
Stereolithography
37
UV light source
Difference in index of refraction
(n) dominates cure depth
nresin = 1.5
nTiO2 = 2.5
resin surface
Cd
2d no2  Eo 
Cd 
ln  
2
3Q n
 Ec 
nalumina = 1.44
nFe2O3 = 2.5
Where:
S
• d = particle size,
Q
• Q = scattering efficiency;
l
• S = particle spacing, l = wavelength
• Eo = Exposure given
• Ec = Critical exposure of resin
Griffin & Halloran, 1995
Conceptual Design
Selection
Augmentation
Analysis
Georgia Institute of Technology
Systems Realization Laboratory
Principal Solution Selection:
Direct Inkjet Printing
38



m  m o 1 
 max
0 vol%
2 vol%
5 vol%



n
10 vol%
Re
1
10
1/ 2
We
Re 
D0V0
m
We 
D0V02

Seerden, Reis, Evans, Grant, Halloran, Derby, 2001
Conceptual Design
Selection
Augmentation
Analysis
Georgia Institute of Technology
Systems Realization Laboratory
39
http://www.niroinc.com/images/chem/spray_dryer_typen.jpg
Georgia Institute of Technology
Systems Realization Laboratory
40
Closure
© Christopher B. Williams
Georgia Institute of Technology
Systems Realization Laboratory