Degradation of magnesium aluminum spinel by lithium fluoride sintering aid

Article abstract of DOI:10.1111/j.1551-2916.2005.00209.x

The effect of LiF sintering aid on the degradation of transparent magnesium aluminate spinel during hot-pressing was studied. LiF is used to etch spinel particles during the hot pressing process. The LiF was found to react with the aluminum in the spinel

Full text of DOI:10.1111/j.1551-2916.2005.00209.x

J. Am. Ceram. Soc., 88 [5] 1321–1322 (2005)
DOI: 10.1111/j.1551-2916.2005.00209.x
ournal
J
Degradation of Magnesium Aluminum Spinel by Lithium Fluoride
Sintering Aid
Guillermo R. Villalobos,*,w Jasbinder S. Sanghera, and Ishwar D. Aggarwal
U.S. Naval Research Laboratory, Washington, District of Columbia 20375-0001
The effect of LiF sintering aid on the degradation of transparent
magnesium aluminate spinel during hot-pressing was studied.
LiF is used to etch spinel particles during the hot pressing proc-
ess. The LiF was found to react with the aluminum in the spinel
structure, thereby leaving Mg-rich regions behind that do not
sinter well and result in opaque white regions in the otherwise
transparent matrix.
wt% LiF (Sylvania, Towanda, PA) sintering aid. Ten grams of
powder/sintering aid mixture was loaded into a 25 mm diameter
I.D. graphite die (Poco Graphite, Decatur, TX) lined with gra-
foil (Polycarbon Inc., Valencia, CA) to minimize carbon diffu-
sion and extend die life. The samples were hot pressed in a
vacuum hot press (Electrofuel, Toronto, Canada) at a heating
rate of 101–16001C/min and held for 2 h. There were also two
30 min holds: one at 9501C and the other at 12001C. Pressure
schedule consisted of maintaining 200 psi until the end of the
12001C hold. The pressure was then slowly increased to 4000 psi
and held until the end of the 16001C treatment. The resulting
samples were 25 mm diameter  2 mm thick.
I. Introduction
AGNESIUM aluminate spinel has great potential as a trans-
M
parent armor material and as a visible-infrared window
material due to its good mechanical and optical properties.1 Its
mechanical properties are comparable with polycrystalline al-
uminum oxide and, since it has a cubic structure (i.e. no biref-
ringence), polycrystalline samples can transmit from 200 nm to
5.5 mm with no optical distortion. Although magnesium alu-
minate spinel has been studied on and off since the 1960s,2,3 the
literature has very little information on its sintering beha-
vior, and the material still cannot be sintered to transparency
reproducibly.4,5
Powder samples were mixed in a mortar at a 50/50 weight
ratio of spinel/LiF, Al2O3/LiF, and MgO/LiF and heat treated
at 9501C in a vacuum furnace to identify reactions during the
initial hot press hold temperature. The 50/50 ratio was used to
increase the amount of reacted material and thereby enable de-
tection of the reaction products by X-ray diffraction (XRD).
The 50/50 mixtures were also run in an SDT (TA Instruments
SDT 2960, New Castle, DE) to determine the reaction temper-
atures and weight loss. The SDT was run in flowing argon at a
101C/min rate to 15001C. The powder and densified samples
were also analyzed using optical microscopy, scanning elec-
tron microscopy (SEM)/energy-dispersive spectrum (EDS)
(LEO 1550 LEO Electron Microscopy Inc., Thorwood, NY),
and XRD (Sintag XDS 2000, Sunnyvale, CA).
Spinel is generally densified with the use of sintering aids, the
most common being LiF. Without LiF sintering aid the material
tends to be translucent and gray. Previous researchers have pro-
posed that the LiF etches and removes impurities from the sur-
face of the spinel particles, thereby enhancing diffusion. It is also
believed that the molten LiF can aid initial compaction by lu-
bricating the particles and allowing better packing. The LiF
must be removed from the material before complete consolida-
tion or it will manifest itself as white precipitates.6
III. Results and Discussion
The spinel hot press schedule is designed around two succes-
sive thermal treatments that are used to first allow the LiF to
react with the particle surfaces, and then a higher temperature
treatment to allow the LiF to volatilize and be removed from the
material before the pore structure collapses and traps the LiF
sintering aid.
Although we eventually wish to form reproducibly transpar-
ent spinel shapes, in this portion of the work we were concerned
with identifying the cause of the inconsistent sintering behavior
and understanding the reactions involved during the course of
the hot-pressing schedule. We have therefore analyzed the
opaque white regions and related those to the effect of LiF con-
centration, interactions between the matrix and LiF, and hot-
pressing conditions.
The literature is not clear on why the two specific holds are used
during the hot press schedule. It is implied that the first hold at
9501C is to allow homogenization of the LiF sintering aid, and
the second hold at 12001C to allow LiF to escape before final
consolidation. SDT analysis of LiF in flowing argon was con-
ducted to confirm the assumptions. The SDT trace shows that
LiF melts at 8501C and rapid weight loss begins at 10501C. LiF
is completely gone at 14001C. Based on these results, we fol-
lowed the traditional hot press schedule.
Figure 1 shows three spinel disks that were mixed with 0.5, 2,
and 10 wt% LiF. It is apparent that increasing the amount of
LiF increases the amounts of opaque white regions. This effect is
characteristic of a transparent material with scattering centers
on the order of the wavelength of visible light. The literature
suggests that the white opaque regions are a result of LiF that
has not been removed and is located at the spinel grain bound-
aries. The results shown in Fig. 1 appear to confirm this.
Figure 2 is a high-resolution SEM micrograph of an opaque
area showing that it is composed of small (300–500 nm) crystals
that have not sintered as well as the surrounding material. EDS
analysis shows that while the transparent area has the expected
2:1 atomic ratio of Al and Mg, the small grains have a 1:1 ratio
of Al and Mg. This suggests that the small grains are made up of
an MgO-rich phase. MgO is considerably more refractory than
spinel and would not be expected to sinter as readily at the lower
sintering temperatures used for spinel.
II. Experimental Procedure
Spinel powder was purchased from Ceralox (Tucson, AZ). The
powder was mixed using a mortar and pestle with 0.5, 2, and 10
L. C. Klein—contributing editor
Manuscript No. 11164. Received July 9, 2004; approved October 5, 2004.
*Member, American Ceramic Society.
wAuthor to whom correspondence should be addressed. e-mail: villalobos@nrl.navy.mil
1321

1322
Communications of the American Ceramic Society
Vol. 88, No. 5
S
L
S
L
L
S
S
S
Fig. 1. Three magnesium aluminate spinel disks with (left to right) 0.5,
2.0, and 10.0% LiF additions showing that the amount of cloudiness
increases with increasing amounts of LiF.
M
L
L
L
S
L
L
L
L
S
10
20
30
40
50
60
Since increasing the amount of LiF resulted in an increase in
the formation of the opaque regions, it has been assumed that
the white regions were due to LiF that was not removed during
the course of the sintering operation. However, this effect alone
cannot explain the removal of Al necessary to create Mg-rich
areas. We hypothesized that LiF does react with the particle
surfaces as is generally thought, but it preferentially reacts with
the Al2O3, thereby leaving MgO-rich areas behind. This reaction
likely occurs when LiF is molten and not yet vaporizing at a
great extent. To test the hypothesis, 50/50 wt% mixtures of
spinel/ LiF, Al2O3/ LiF, and MgO/LiF were prepared and heat-
treated in vacuum for 2 h at 9501C (the first hold temperature
routinely used when hot pressing). Figure 3 shows the XRD
patterns of the spinel/LiF mixture. The XRD patterns show the
presence of spinel, LiAlO2, and MgO Á LiAlO2 was also present
in the Al2O3/LiF sample. There was no evidence of any reaction
in the MgO/LiF sample.
Fig. 3. X-ray diffraction pattern showing MgAl2O4 spinel phase (S),
and the LiAlO2 (L) and MgO (M) compounds that formed after heat-
treating a 50/50 wt% mixture of MgAl2O4 and LiF at 9501C for 2 h in
vacuum.
ever, it is apparent that the LiF is reacting with the spinel
particle and the reaction products are adversely impacting
the transparency of the hot-pressed disks made by the tradition-
al means.
We propose that the opacity is due to the presence of the Mg-
rich phase and the associated porosity that it introduces. In ad-
dition, the LiAlO2 that produces the MgO phase more than
likely remains in the structure, further contributing to the opac-
ity of the spinel. The randomness of the opacity is attributed to
the random presence of LiF in the powder mixture as a direct
result of the mechanical mixing process prior to hot pressing.
These results match well with a thermodynamic study of the
LiF–Al2O3–MgO system that shows that while MgO does not
appreciably react with LiF, Al2O3 can react to form various
lithium–aluminum-containing compounds. Of all the Li–Al-
containing compounds listed in the JANAF tables,7 only
LiAlO2 appears to be a solid throughout the temperature range
of the hot press schedule; the other reaction products are vapors
at the hot pressing temperature and help to explain the results of
the 50/50 mixture experiments and also the presence of the Mg-
rich regions in the hot-pressed disks. Unfortunately, it is difficult
to identify LiAlO2 regions by EDS because EDS cannot detect
lithium, and it is difficult to distinguish Al that belongs to the
matrix from Al that would belong to LiAlO2 inclusions. How-
IV. Conclusion
Although LiF is necessary to densify spinel to transparency, it
also reacts with the aluminum in the spinel matrix to create Mg-
rich areas that cause opacity in the densified material due to
poor sintering. It is also possible that LiAlO2 precipitates from
that same reaction contribute to the formation of opaque re-
gions. Based on these results, it is necessary to study the times
required to adequately homogenize the LiF sintering aid with-
out allowing sufficient time for it to react with the aluminum in
the spinel matrix. It may also require that the LiF be more
evenly distributed in the spinel powder prior to hot pressing. A
more homogeneous initial distribution should decrease the total
amount of LiF needed, and also reduce the hold times, there-
by producing high optical quality spinel samples with high
reproducibility.
Acknowledgments
The authors would like to thank Don W. Roy, Consultant, (Sun City, AZ) and
Richard Brown, MVA Scientific Consultants (Norcross, GA).
References
1M. C. L. Patterson, J. E. Caiazza, D. W. Roy, and G. Gilde, ‘‘Transparent
Spinel Development’’; SPIE 45th International Symposium on Optical Science and
Technology, 30 July–4 August 2000, San Diego, CA.
2D. W. Roy, M. C. L. Patterson, J. E. Caiazza, and G. Gilde, ‘‘Progress in the
Development of Large Transparent Spinel Plates’’; Eighth DoD Electromagnetic
Windows Symposium, April 2000, Colorado Springs, CO.
3D. W. Roy, D. R. Johnson, and D. L. Mann, ‘‘Fabrication and Properties of
Transparent MgAl2O4,’’ Am. Ceram. Soc. Bull., 52 [4] 372–3 (1973).
4M. C. L. Patterson, D. W. Roy, and G. Gilde, ‘‘Improvements in Optical
Grade Spinel’’; Ninth DoD EM Windows Symposium, May 2002, Huntsville, AL.
5M. Shimada, T. Endo, T. Saito, and T. Sato, ‘‘Fabrication of Transparent
Spinel Polycrystalline Materials,’’ Mater. Lett., 28 [8] 413–5 (1996).
6D. W. Roy, Private Communications, 2002.
7M. W. Chase ed., NIST-JANAF Thermochemical Tables, 4th edition, NIST,
Fig. 2. Small-grained, partially sintered Mg-rich phase in MgAl2O4
matrix.
Gaitherburg, MD, 1998.
&