Content uploaded by Kelsa Benensky Palomares
Author content
All content in this area was uploaded by Kelsa Benensky Palomares on Mar 02, 2019
Content may be subject to copyright.
Nuclear and Emerging Technologies for Space, American Nuclear Society Topical Meeting
Richland, WA, February 25 – February 28, 2019, available online at http://anstd.ans.org/
INVESTIGATION OF PRODUCTION PARAMETER EFFECTS ON SPARK PLASMA SINTERED
MOLYBDENUM CERMET WAFERS FOR NUCLEAR THERMAL PROPULSION APPLICATIONS
James Zillinger1, Becca Segel2, Kelsa Benensky3, 4, Dennis Tucker4, and Marvin Barnes4
1 Department of Chemical and Materials Science Engineering, University of Idaho, Moscow, ID 83844, 208-310-2277,
zill0470@vandals.uidaho.edu
2Department of Chemical and Biomolecular Engineering, Case Western Reserve University, Cleveland, OH 44106, 215-460-
1900, rns59@case.edu
3 Department of Nuclear Engineering, University of Tennessee, Knoxville, TN 37996
4Materials and Process Laboratory, MSFC, Huntsville, Al, 35812, 256-544-2040, kelsa.m.benensky@nasa.gov
This study focused on nuclear fuel fabrication using
powder blending and spark plasma sintering (SPS) of Mo-
ZrO2 surrogate ceramic-metal (cermet) fuels. The study
consisted of two co-projects: a powder
blending/distribution study and an SPS parameter
optimization study. The powder blending study focused on
optimization of the powder processing parameters in
order to fabricate high density cermets with uniformly
distributed ceramic microstructures. The SPS parameter
optimization study focused on the impact of sintering
parameters (temperature, dwell time, and pressure) on
the microstructural properties of a cermet
(density/porosity, grain structure, and hardness). In the
powder blending and distribution study, addition of 0.1
wt% and 0.5 wt% binder resulted in complete coverage
and even distribution of large, spherical ZrO2 particles
with Mo powder for batches of 50 vol% and 60 vol%
ceramic loading respectively. When optimizing SPS
parameters, fuel density (decreased porosity) was directly
related to increase in sintering temperature, pressure,
and time. Optimal sintering parameters suggested from
this study, for Mo-ZrO2 cermet wafers with 60 vol%
ceramic loading, were found to be at a temperature of
1400ºC, at 50 MPa uniaxial pressure, for at least 5
minutes dwell time.
I. INTRODUCTION
Nuclear thermal propulsion (NTP) is based upon the
principle of heating a low mass propellant, such as
hydrogen (H2), via the heat from fission and expanding it
through a nozzle to produce high in-space thrust (15,000 -
250,000 lbf) and specific impulse (750 - 1100 s). NTP is
actively being researched by NASA and other private,
university, and government partners to enable fast transit
times for missions beyond low Earth orbit (LEO). Several
methods enabling this form of propulsion are under
scrutiny, including the use of ceramic-metal (cermet) fuel
systems. In cermet fuel systems, ceramic fissile fuel
particles (such as UO2, UN) are encapsulated by a metal
matrix that is composed of an inert refractory metal, such
as tungsten (W) or molybdenum (Mo). Cermet fuel
wafers are currently being manufactured via spark plasma
sintering (SPS) techniques, which consolidates blended
cermet powders into a net-shape fuel pellet. In this
process, sintering is facilitated by passing a DC current
through a graphite die to heat sample powders to very
high temperatures, while applying uniaxial force. This
study explores the optimal cermet wafer production
methodology for Mo-matrix cermets using ZrO2
microspheres as a surrogate for UO2. Originally
developed by Tucker et al. for W-matrix cermets, a two-
step powder blending/SPS sintering process was utilized
to optimize the fabrication of cermets with desirable
microstructures.3
Fig. 1. The SPS machine presses blended cermet powder
into a dense pellet at high temperatures with uniaxial
force. The powder is inserted into a die (A) which is
placed in the SPS furnace (B).
II. PRE-SINTERING POWDER BLENDING
PROCESS OPTIMIZATION
Prior to sintering, Mo and ZrO2 powders are blended
with a polyethylene binder to create a uniform distribution
of ceramic fuel particles within the metal matrix. In this
study, the powder blending process was evaluated to
identify the impact of binder addition on uniformity of
powder coatings of as blended mixtures and particle
distribution of as-fabricated cermet microstructures, for
variable fuel volume loadings (50, 60 vol% ceramic). It is
desired that the cermet microstructure allows for uniform
particle distributions which eliminate interconnectivity of
2
the ceramic phase. Powder blending is performed via
addition of a polyethylene binder to the initial powders,
powder mixing, and blending via heating of the binder
above the binder drop point to allow for adherence of
metal powders to ceramic microspheres. Binder is then
burned off from the cermet mixture during sintering.
II.A. Experimental
Spherical ceramic ZrO2 particles (~250 µm diameter)
and metal Mo powders (~5 µm) are measured on an
analytical balance to attain the desired volume ratio of
metal to ceramic. A polyethylene binder addition (or lack
thereof) was measured and added to each batch in
quantities spanning between 0 – 1.00 wt%. The mixtures
were transferred to a Turbula® shaker mixer and shaken
for approximately one hour. At this point, the mixed
powders appear homogeneous, so that there was no
obvious localized separation of the larger ZrO2 particles
from the finer metal powders/binder via visual inspection.
Mixtures containing binder addition are transferred into a
beaker, heated on a hotplate above the binder drop point
(170 – 185 °C), and stirred with a stir bar (revolving at
100 - 200 rpm) for approximately 10 minutes until the
powder mixture appeared homogeneous. Heating and
stirring is terminated once all white ZrO2 spheres are fully
coated (unobservable without a visual aid) and the
mixture had begun to agglomerate. Blended powders were
inspected using scanning electron microscopy (SEM) to
characterize metal powder coating uniformity.
In order to analyze the effectiveness of the powder
blending process on the fabrication of cermet wafers with
ideal microstructures, a study analyzing the distribution of
ZrO2 throughout the Mo-matrix was performed. Blended
powders of 50 and 60 vol% cermets with and without
binder addition were fabricated using SPS. For each
mixture, ~13 g of powder was transferred into a 20 mm
diameter Grafoil® lined graphite die and sintered in a
Thermal Technologies DCS-15 SPS furnace using the
sintering parameters (sintering temperature, pressure and
dwell time) of 1500 °C, 5 minutes, 50 MPa. The produced
cermet wafers were ground flat using silicon carbide
(SiC) paper until the surrounding Grafoil® is removed.
Ground samples were cross sectioned using a diamond
saw and mounted in epoxy for polishing and
microstructural characterization of the radial and axial
variation in particle distribution (fig. 2). The samples
were mechanically polished using SiC paper to a 4000
grit (5 μm particles) finish and axial and radial ceramic
distributions observed using optical microscopy (OM).
Fig. 2. Cermet coupon preparation method after SPS. The
sample is then polished to a mirror finish and analyzed via
optical microscopy for a uniform particle distribution.
II.B. Results
II.B.1. Powder Blending Optimization
When optimizing the powder blending process, it is
desired to determine the required addition of binder to
allow for uniform particle coatings, while minimizing
powder clumping. Blended mixtures with varying binder
additions were inspected using SEM to determine which
binder addition yielded the most uniform complete
coverage of metal coatings on ceramic fuel particles (fig.
3). Within 60 vol% cermets experimented on previously,
maximum ZrO2 coating was observed with 0.5 wt%
binder addition. Addition of this much binder to 50 vol%
cermets resulted in self-adherence of the Mo (fig. 3-E &
F). Reduction to 0.1 wt% binder resulted in adequate
coating for 50vol% cermets. (fig. 3-C).
3
Fig. 3. An optimal amount of binder for the 50 vol% Mo-
cermet is 0.1 wt% (C & D). The ZrO2 particle is well
covered in Mo (C & D) and the particles are evenly
spaced allowing for a full ZrO2-Mo coverage (D). A
greater or lesser wt% of binder creates a lack of ceramic
coverage and an uneven particle distribution.
II.B.2. Particle Distribution Study
Upon analysis of radial and axial optical micrographs
taken of the cross sectioned samples, ZrO2 distribution
was found to be much more uniform with addition of 0.5
wt% binder in 60 vol% cermets, and 0.1wt% binder
addition within 50 vol% cermets.
Fig. 4 below is a visual representation of two cermets of
the same composition prepared through blended powders
(Fig. 4-A, B) and without blending (Fig. 4-C, D). The
powder blending process produces a more homogenous
60 vol% sample and prevents ceramic and metal
segregation during loading of powders prior to sintering.
Incorporation of the powder blending process in cermet
fabrication is necessary for optimized microstructures for
high volume loading cermets (>50 vol%).
Fig. 4. (A) & (B) show distribution of ZrO2 in a 60 vol%
cermet with 0.5 wt% binder addition. (C) & (D) show
distribution without addition of binder.
III. SPS PARAMETER OPTIMIZATION
To identify the optimal sintering parameters for Mo-
matrix cermets using SPS, three variables were studied:
dwell time, temperature, and pressure. Dwell times of 5 -
30 minutes and sintering temperatures of 1100 – 1400 °C
were considered. A standard applied pressure for Mo-
cermets was chosen as 50 MPa and was held constant
until the variables of dwell time and temperature are
optimized. Once optimized temperature and dwell time
parameters were established, sintering pressure was
varied (5, 10, 25, 50 MPa). After sintering, the as
fabricated Mo-cermet microstructure was characterized to
understand impact of sintering parameters on cermet
density, matrix density, and particle-matrix bonding.
III.A. Experimental
A standardized blended powder mixture of 60 vol%
Mo-ZrO2 (0.5 wt% binder additions) was utilized for
optimization of SPS parameters. For each sample, 13 g of
blended powder was loaded into a 20 mm diameter
Grafoil® lined graphite die and sintered at temperatures
of 1100, 1200, 1300, 1400, and 1500 °C for 5 minutes.
These dwell times and temperatures were chosen for
initial study due to diminishing return (or a reduced
increase in density) anticipated during the sintering
process. This behavioral trend has been seen previously
for W-CeO2 cermets3 in addition to this project.
Throughout the entirety of the study, ramp rate was held
constant with: 100 °C/minute and 10 MPa/min to
maximum temperature and pressure.
After sintering, fabricated cermets were ground flat
using SiC paper until the Grafoil® layer was removed
from all faces of the sample. Once this sacrificial layer
was removed, samples were weighed and as fabricated
cermet density was determined via the Archimedes
method. A Sartorius Practum 224-1S analytical balance
was used to obtain room temperature weights of each
cermet sample in air and in deionized (DI) water. Once
weighed, samples were prepared for microstructural
analysis using OM and SEM. Samples were polished to a
1µm finish using a Struers LaboForce-100 auto polisher.
The polishing process required the following SiC paper
polishing, followed by diamond suspension polish to
complete the mirror finish varying from 15-25 N force: 80
grit (ground until flat), 180 grit (4 mins), 320 grit (4
mins), 500 grit (4 mins), 600 grit (4 mins), 800 grit (4
mins), 1200 grit (4 mins), 2000 grit (4 mins), 4000 grit (4
mins), 9μm suspension, 3μm suspension, 1μm suspension.
III.B. Results
Using Archimedes principle, the density of cermet
samples was calculated and compared to the theoretical
density (TD) of an ideal 60-vol% Mo-cermet (7.80
g/cm3), as shown below in Table I. The SPS parameters
each correspond to its cermet density, and an optimal
density of 7.57 g/cm2 (97.46%) is chosen. Even though a
greater dwell time and temperature proved a greater
density than the chosen optimal, the selected has a lesser
dwell time and therefore requires less power and funds to
fabricate.
Overall, cermet density did correlate with increasing
pressure for high applied pressures 25 - 50 MPa (Table
II), however porosity was observed to be relatively
independent of sintering pressure at low pressures (< 25
MPa). Minimal density improvements were seen when
varying pressure between 5, 10, and 25 MPa, however a
large shift in cermet density was observed for 25 and 50
MPa applied pressures. For application in production
scale processing of Mo-cermet wafers, when optimizing
the sintering process, applied pressures should not result
in great benefit to the overall wafer density if sintering at
low pressures (5 - 25 MPa).
Polished samples were inspected via SEM to analyze the
effects of applied sintering pressure on sample
microstructure and porosity (fig. 5). In this study, sample
4
porosity was found to increase with reduced sintering
temperature, dwell time, and applied pressure. Porosity,
especially interconnected porosity, acts as an easy
pathway for H2 ingress into the pellets, thereby leaving
the ceramic fuel particles more susceptible to hydrogen
attack during operation. Fig. 5-A,B,D show low levels of
porosity formed when sintering at 1400 °C at varying
dwell times and pressures. Fig. 5-C demonstrates a
representative microstructure formed at the low
temperature extreme where porosity is more prevalent in
the metal matrix and a large void space is also present due
to poor sintering of the metal matrix.
Fig. 6 demonstrates representative microstructures of
cermets with desirable and undesirable ceramic-metal
matrix interfaces. It is desirable for the metal matrix to
have both low porosity and the microstructure to sustain a
diffusional bond of the metal matrix with the ceramic fuel
particles. Fig. 6-A exhibits a cermet microstructure with
diffusional bonding of the ceramic and metal phases, the
perimeter of the ceramic is surrounded completely by the
Mo-matrix and diffusion of the Mo/ZrO2 over the entirety
of the interface occurs. In fig. 6-B, the ceramic particle is
surrounded by the metal matrix, however no diffusion
occurs, therefore particles are held much more loosely
within the metal matrix and are more susceptible to
pullout during preparation or fallout during handling.
A diffusion bonded interface is observed in the
optimally sintered cermets and is anticipated to result in
more desirable thermomechanical properties of the fuel
and better fuel performance. Relatively low porosity was
seen in the metal matrix for all samples, but more pores,
void areas (caused by pullout of ZrO2 particles during
sample preparation), and poor interfacial bonding at the
metal cermet interface (fig. 6-B) were seen in the
microstructure of cermets sintered at 1100 °C (fig. 6-B,
fig. 5-C). The optimally sintered samples in the upper
temperature and time ranges showed not only a lack of
macro-cracks and void space, but also optimal interfacial
bonding (fig. 6-A) and overall higher %TD for these
samples (fig. 6-A, fig. 5A,B,D).
TABLE I. Dependence of as produced cermet wafer
density and %TD with respect to sintering dwell time and
temperature, fabricated with 50 MPa applied pressure
Density
(%TD)
(g/cm3)
Sintering Temperature (°C)
1100
1200
1300
1400
1500
Dwell time (min)
5
90.74
(7.080)
95.06
(7.407)
95.26
(7.494)
97.46
(7.573)
96.72
(7.536)
10
95.74
(7.451)
97.63
(7.569)
97.97
(7.606)
98.1
(7.612)
30
95.90
(7.452)
97.97
(7.612)
98.57
(7.612)
TABLE II. SPS Pressure Analysis: 1400 °C for 5 minutes
Raw Density and %TD
Pressure
(MPa)
Density
(g/cm3 | %TD)
5
7.422 | 95.68
10
7.423 | 95.69
25
7.427 | 95.79
50
7.573 | 97.46
Fig. 5. Back scattered electron image of Mo cermets
produced with varying sintering temperatures, pressures,
and dwell times: (A) 1400°C-5 minutes-50 MPa
(optimal), 7.753 g/cm3 (B) 1400 °C-5 minutes-5 MPa,
7.422 g/cm3 (C) 1100 °C-5 minutes -50 MPa, 7.080
g/cm3 (D) 1400 °C-30 minutes-50 MPa, 7.612 g/cm3.
Fig. 6. Difference between a strong (A) and weak (B)
interfacial bond of the ceramic particle and metallic
matrix for cermets sintered at (A) 1400℃ for 5 minutes at
5 MPa and (B) 1100℃ for 5 minutes at 50 MPa.
IV. CONCLUSIONS
In this study, the impact of powder preparation and
sintering parameters on “as fabricated” Mo-cermet wafer
5
microstructure was explored using ZrO2 as a surrogate for
UO2. The criteria for an optimized cermet microstructure
in this study consisted of:
Fully dense and sintered metallic matrix (low
porosity).
Diffusional interfacial bonding of the metal matrix
and ceramic fuel particles.
Uniform ZrO2 distribution throughout a Mo-matrix
without agglomeration (minimal interconnectivity)
of the ceramic particles.
In this study, the applications and intricacies of the
powder blending process, previously demonstrated for the
fabrication of W-UO2 cermet wafers, were explored for
the fabrication of Mo-cermets with desirable
microstructures. It was found that optimal binder addition
is dependent on ceramic volume loading and lower
ceramic volume loadings (< 60 vol%) required less binder
addition than high fuel volume loading cermets. As-
sintered wafer fuel distribution was independent of the
powder blending distribution for low volume loadings
(40, 50 vol%). When excessive binder was added to the
mixture, metal powders were observed to agglomerate
resulting in more segregated sintered microstructures.
However, for 60 and 70 vol% fuel loadings, much more
desirable microstructures were fabricated using blended
powders (with optimal binder coating) rather than the as
mixed powders.
As a second part to this study, the impact of SPS
parameters on 60 vol% loading cermet microstructures
were explored. Processed cermets were formed into near
TD compacts consisting of a metal matrix with closed
porosity and with a diffusion bonded interface to the fuel
surrogate particles at sintering temperatures exceeding
1200 °C. Density was found to increase, and porosity
decreased with increasing process temperature and dwell
time, with >97% TD achieved at 1400 °C, 5 minutes, 50
MPa pressure. Processed cermet density was relatively
independent of pressure in the low-pressure range (5 - 25
MPa). A large shift in cermet density was observed for 50
MPa applied pressures (> 97 %TD) compared to low
pressure (~95 %TD). The optimized sintering parameters
(1400 °C, 5 min., 50 MPa) suggested in this study that
high TD in wafer fabrication corresponds to a high
sintering pressure (50 MPa). If very high cermet densities
(>97 %TD) and low processing pressures (< 25 MPa) are
desired, it is suggested future optimization studies focus
on increased sintering temperatures or dwell times.
ACKNOWLEDGMENTS
We would like to give thanks to NASA internship
mentors: Kelsa Benesky and Marvin Barnes. Also, to
Ellen Rabenberg, Dr. Omar Rodriguez and William
Carpenter for helping obtain images (ER, WC), gather
hardness values (OR), as well as prepare powder samples
for experiments (WC). This work was funded in part by
the Ohio Space Grant Consortium, the NTP Project at
NASA MSFC, the Center Innovation Fund at NASA
MSFC, and NASA Grant #NNX15AQ35H (KB). Finally,
we would like to thank all members of our MSFC EM32
and EM31 for their support during our internship.
REFERENCES
[1] D. S. Tucker, M. W. Barnes, L. Hone, and S. Cook,
“High density, uniformly distributed W/UO 2 for use
in Nuclear Thermal Propulsion,” J. Nucl. Mater., vol.
486, pp. 246–249, Apr. 2017.
[2] Y. Zhong et al., “Spark Plasma Sintering of Fuel
Cermets for Nuclear Reactor Applications,” MRS
Proc., vol. 1383, Jan. 2012.
[3] J. A. Webb and I. Charit, “Fabrication of Cermets via
Spark-Plasma Sintering for Nuclear Applications,”
JOM, vol. 66, no. 6, pp. 943–952, Jun. 2014.
[4] D. S. Tucker, A. O’ Connor, and R. Hickman, “A
Methodology for Producing Uniform Distribution of
UO2 in a Tungsten Matrix,” J. Phys. Sci. Appl., vol.
5, no. 4, Apr. 2015.
[5] C. Haertling and R. J. Hanrahan, “Literature review
of thermal and radiation performance parameters for
high-temperature, uranium dioxide fueled cermet
materials,” J. Nucl. Mater., vol. 366, no. 3, pp. 317–
335, Jul. 2007.
[6] U. K. Vashi, R. W. Armstrong, and G. E. Zima, “The
hardness and grain size of consolidated fine tungsten
powder,” Metallurgical Transactions, vol. 1, no. 6,
pp. 1769–1771, Jun. 1970.
[7] R. C. O’Brien and N. D. Jerred, “Spark Plasma
Sintering of W–UO2 cermets,” Journal of Nuclear
Materials, vol. 433, no. 1–3, pp. 50–54, Feb. 2013.
