The effect of surface oxides during hot pressing of TiB2

Article abstract of DOI:10.1111/j.1551-2916.2009.02923.x

Hot pressing of TiB₂ has been investigated with particular emphasis on the evolution of secondary phases originating from the initial surface oxide layer on the TiB₂ powders. Carbothermal reduction of the surface oxides during sinter

Full text of DOI:10.1111/j.1551-2916.2009.02923.x

J. Am. Ceram. Soc., 92 [3] 623–630 (2009)
DOI: 10.1111/j.1551-2916.2009.02923.x
r 2009 The American Ceramic Society
ournal
J
The Effect of Surface Oxides During Hot Pressing of TiB2
Morten Sundheim Jensen, Mari-Ann Einarsrud,* and Tor Grande*w
Department of Materials Science and Engineering, Norwegian University of Science and Technology, N-7491
Trondheim, Norway
Hot pressing of TiB2 has been investigated with particular em-
phasis on the evolution of secondary phases originating from the
initial surface oxide layer on the TiB2 powders. Carbothermal
reduction of the surface oxides during sintering was also inves-
tigated by adding carbon to the TiB2 powder. TiO1ꢀxCx was
shown to be the main secondary phase in hot-pressed TiB2, and
carbon was shown to strongly influence on sintering process and
the amount, composition and distribution of the secondary phase
TiO1ꢀxCx. The formation of TiO1ꢀxCx is discussed in relation
to volatile boron oxide, which reacts with the graphite die to
produce CO gas, which further may cause transport of carbon
into TiB2 during sintering before pore closure. Finally it was
demonstrated that the density could be controlled by addition of
carbon to the TiB2 powder.
ing in intergranular cracking of the material upon relaxation of
residual stresses. The rate of degradation is influenced by the
microstructural features like porosity, cracks, grain size, and
secondary phases along the grain boundaries.10–19
The fabrication of TiB2 ceramics, chemically resistant to-
wards aluminum infiltration along grain boundaries, calls for a
better understanding of the influence of the titanium oxide scale
on TiB2 particles20 on the microstructure evolving during the
sintering process. Residual titanium oxide phases originating
from the oxide scale is thermodynamically not stable in contact
with aluminum and will also dissolve in the cryolite electrolyte.
Here we present an investigation of hot pressing of TiB2 ceram-
ics with particular emphasis on the evolution of secondary
phases originating from the surface oxides. In situ carbothermal
reduction of the oxides has also been studied by addition of
carbon sources, and the observations of secondary phases are
discussed in relation to the thermodynamic stability of the sec-
ondary phases.
I. Introduction
ITANIUM diboride (TiB2) has
a high melting point
T
(B32001C), has a high hardness, electrical conductivity,
II. Experimental Procedure
and thermal conductivity and is chemically stable under many
conditions.1,2 TiB2 is also wetted by liquid aluminum,3 and the
combination of these properties makes TiB2 a promising alter-
native as a cathode material in Hall-Heroult alumina reduction
cells. Molten aluminum wets TiB2 very well and open up for new
cell designs using drained dense polycrystalline TiB2 cathodes,
reducing the anode–cathode interpolar distance with the poten-
tial of both enhancing the performance and the lifetime of the
cell and reducing the energy consumption.4,5
The samples were prepared using different powder compositions
of TiB2 (Grade HCT-F, GE Advanced Ceramics, Cleveland,
OH), carbon black (CB, o100 mm, Merck KGaA, Darmstadt,
Germany), coal tar pitch (CTP), Ti2O3 (o325 mech, 991%,
Alfa Aesar GmbH & Co KG, Karlsrue, Germany) and B2O3 (99
at%, Sigma-Aldrich, St. Louis, MO) as described in Table I.
An oxygen content of 1.7 wt% in the TiB2 powder was mea-
sured by chemical analysis (Leco analyzer, Leco TC-436DR,
Leco Corporation, St. Joseph, MI). Assuming that reaction (1)
is accounting for the reduction of the oxygen impurities present
in the powder during sintering, 1.3 wt% C is necessary in order
to reduce all these oxide impurities.
Despite these promising properties, the utilization of inert
TiB2 cathodes has not been achieved due to the difficulty of
obtaining dense and phase pure TiB2 ceramics with the desired
properties. The low sinterability of TiB2 has been attributed to
the strong covalent bonding resulting in poor bulk and grain
boundary diffusion rates at moderate temperatures.6 High sinte-
ring temperatures are therefore necessary to obtain dense TiB2
materials. Sintering of TiB2 at high temperatures results in sig-
nificant grain growth resulting in grains above a critical size
(B15 mm), which causes spontaneous microcracking of the ma-
terial upon cooling.7 Moreover, oxide impurities present in the
TiB2 powder has been proposed to promote grain growth dur-
ing sintering,8 but the grain growth mechanism is not fully
understood.
TiO2ðsÞ þ B2O3ðgÞ þ 5CðsÞ ¼ 5COðgÞ þ TiB2ðsÞ
(1)
Because of a carbon yield of B50%, twice the amount of
CTP was added to give the same amount of C. The powders
Table I. Relative Density, Grain Size, Measured and Esti-
mated Weight Loss for the Samples Hot Pressed at 18001C
Weight loss, Weight loss,
Furthermore, although TiB2 has a low solubility in liquid
aluminum9 and is resistant to cryolite,4 previous studies have
shown that the degradation of the material proceeds at rates that
are not satisfactory for application as an alternative cathode
material. The degradation of polycrystalline TiB2 is caused by
penetration of liquid aluminum along grain boundaries, result-
Relative
Grain size
measured
(%)
estimated
(%)
Composition/wt%
density (%)
(mm)
TiB2
93.470.1 5.670.7 1.3870.12
90.170.1 4.370.5 2.3470.07
88.870.1 4.270.4 2.5870.06
84.970.1
97.370.1 5.670.3 2.0970.08
1.5
2.7
3.0
3.0
4.3
TiB210.70 C
TiB212.05 C
TiB214.31 C
TiB212.25
w
—
2.8570.06
Dr. T. Besmann—contributing editor
Ti2O312.75 B2O3
TiB211.67 CTP 97.470.1 7.970.6 3.0970.05
TiB212.55 CTP 97.570.1 7.370.3 3.2070.05
TiB213.31 CTP 96.570.1 3.970.4 3.1770.05
3.6
4.3
4.7
5.6
Manuscript No. 25089. Received August 6, 2008; approved December 5, 2008.
*Member, The American Ceramic Society
This study was supported by The Research Council of Norway through the project
‘‘Thermodynamics Applied to High Temperature Materials Technology.’’
wAuthor to whom correspondence should be addressed. e-mail: tor.grande@mate
rial.ntnu.no
TiB215.1 CTP
93.370.1 3.170.3 4.2470.04
A theoretical density of 4.52 g/cm3 was used for the calculations of the relative
density. wPorous microstructure. CTP, coal tar pitch.
623

624
Journal of the American Ceramic Society—Jensen et al.
Vol. 92, No. 3
containing CB and the powders with oxide additions were mixed
using dry ball milling with WC balls as mixing media in a lab-
oratory plastic jar for 1 h. The powders containing CTP were
prepared by dissolving CTP in tetrahydrofuran followed by ad-
dition of TiB2 powder and some ethanol in order to achieve a
low viscous slurry. This slurry was mixed using ball milling with
WC as mixing media in a laboratory plastic jar for 1 h. The
slurry was then dried for 24 h, before dry-ball milling for 1 h.
The resulting powder was then mortared and sieved.
Thermogravimetrical analysis were performed in inert atmo-
sphere (Argon 5.0) using a thermomicrobalance (Netzch TG 449
C, Selb, Germany) in order to investigate the weight loss of both
pure TiB2 powder and the powder added CB and CTP during
heating to 16001C with a heating rate of 101C/min.
Hot pressing was performed using a pressing die with a di-
ameter of 25 mm made of graphite (G347, Tokai Carbon
Group, Tokyo, Japan). The die was filled with B11.5 g of pow-
der and uniaxially pressed at 50 MPa to precompact the sample.
The pressing die was placed in the hot press (HP50-7010G,
Thermal Technology Inc., Santa Rosa, CA) and sintered for 1 h
in argon atmosphere (Argon 4.0) under 50 MPa. Both the heat-
ing rate and the cooling rate was 101C/min, and the sintering
temperature varied between 13001 and 18001C.
The samples were cut using an automatic cut-off machine
(Struers Accutom-5, Struers, Ballerup, Denmark) and a dia-
mond blade (Diprotex 1A1R, Dia-Team, Trondheim, Norway)
so as to allow the study of the bulk microstructure. The surface
was ground and polished to a mirror-like finish using grinding
and polishing equipment from Struers (RotoPol-31, RotoForce-
4, TegraPol-31, TegraForce-5 and TegraDoser-5) and the pro-
cedure as described in Struers e-Metalog21 for preparation of
TiB2 ceramics. The polished surfaces as well as the powders
were examined using a scanning electron microscope (Hitachi
S-3400N, Hitachi, Ibaraki, Japan) equipped with an electron
dispersive spectrometer (EDS, Oxford Inc., Abingdon, U.K.)
and a digital electron backscatter diffraction (EBSD) detector
(Nordif, Trondheim, Norway). EDS was used for element anal-
ysis, while EBSD was used during orientation image microscopy
for the measurement of grain size. An electron probe microan-
alyzer (EPMA, Jeol JXA-8500F, Tokyo, Japan) was used to
investigate the compositions of the secondary phases. The sam-
ples were also investigated by X-ray diffraction using a Bruker
AXE D8 Advance X-ray diffractometer (Bruker AXE, Karlsrue,
Germany) using CuKa radiation and a VANTEC-1 detector.
A stepsize of 0.021 and a counting time of 1 s were used
in the 2y range 251 to 701. The densities were measured using
Archimedes principle with isopropanol as media.
the addition of CTP up 2.55 wt%, corresponding to B1.3 wt%
C, as shown in Fig. 1(b)). An increase in density with the ad-
dition of oxides is also evident from Table I and Fig. 1(a)). These
results demonstrate that both the presence of oxides and carbon
strongly influence the sintering of TiB2 and thereby the final
microstructure of the hot-pressed materials.
Early in the study it was revealed that the microstructure of
most of the hot-pressed TiB2 samples was inhomogeneous, and
several apparent layers were evident. A porous layer at the sur-
face towards the graphite die was observed for all the samples
except for the samples with the additions of CTP. The thickness
of the porous outer layer increased with carbon addition and
was reduced in thickness with addition of oxides. The thickness
of the porous layer was approximately 300, 500, and 1000 mm
for the powders TiB212.25Ti2O312.75B2O3, TiB2 and TiB21
0.70CB, respectively after hot pressing at 18001C. The interior of
the specimens were dense and had a homogeneous microstruc-
ture. Typical images of a polished material (pure TiB2 powder)
hot pressed at 18001C are shown in Fig. 2. An overview of the
material is shown in Fig. 2(a)), while micrographs of the porous
layer and the bulk material are shown in Figs. 2(b) and (c), re-
spectively. Closer examination of the material revealed an in-
termediate layer (B200 mm) in between the outer porous layer
and the dense interior with a high amount of a secondary phase,
as shown in Fig. 2(d). This intermediate layer, with a high con-
tent of secondary phase, was not found in any of the samples
with C additions. The samples with higher amounts of CB
(TiB212.05CB and TiB214.31CB) did not obtain a full density
and had a porous microstructure throughout the sample. The
materials with addition of CTP were significantly more dense
than the materials added CB, and no distinct porous layer were
present in these samples. This observation points to the impor-
tance of a homogeneous distribution of C in order to improve
the density. Uneven distribution of CP particles or agglomerated
CP particles resulted in uneven distribution of pores which could
be observed by microscopy (not shown here).
Micrographs of samples with different amounts of CTP ad-
ditions as well as a micrograph of the sample without addition
are presented in Fig. 3. In the bulk of the sample without any
additions (Fig. 3(a)), a secondary phase present at the triple
junctions and along some grain boundaries can be seen as well
as some pores. The addition of 1.67 wt% CTP reduced the
amount of the secondary phase, which was present as particles
at triple junctions and in the interior of some grains (Fig. 3(b)).
The total porosity was also reduced. With addition of 2.55 wt%
CTP even smaller amounts of the secondary phase were present
at triple junctions, but some intragranular particles could still be
observed (Fig. 3(c)). Further addition of CTP resulted in a more
porous microstructure and residual carbon in the material
(Fig. 3(d)).
III. Results
The average grain size of the materials sintered at 18001C is
also included in Table I. The addition of CB inhibited both grain
growth as well as densification, while in the case of CTP both the
grain size and the density increased at low levels of addition.
However, residual carbon, originating both from CB and CTP
strongly inhibits grain growth and densification. No exaggerated
grain growth was observed for any samples up to 18001C.
X-ray diffractograms of selected polished specimens as well as
the TiB2 powder are shown in Fig. 4. Besides the strong reflec-
tions due to TiB2, reflections due to the presence of secondary
phases are clearly visible in all the diffractograms. The presence
of TiO1ꢀxCx with rock salt structure is evident in all diffracto-
grams. Calculation of the lattice parameter and estimation of the
The relative densities of the specimens after hot pressing at
18001C are given in Table I, and the relative densities of the
specimens with different additions of CB as a function of tem-
perature are shown in Fig. 1(a). The density was reduced with
increasing amounts of CB, whereas the density increased with
composition of this phase was performed using Vegards law
22
˚
˚
and lattice parameter 4.183 A for TiO and 4.3274 A for TiC
(Pattern: 00-032-1383). A composition of this phase correspond-
ing to TiO0.43C0.57, TiOO.16C0.84 and TiOO.09C0.91 was found for
specimens hot pressed at 18001C from powder TiB2, TiB21
2.05CB and TiB212.55CTP, respectively. Several of the inten-
sities of the reflections for TiB2 are not in accordance with the
ICDD powder diffraction file for TiB2 (Pattern: 00-035-0741),
which is due to preferential grain orientation during hot pressing
Fig. 1. Relative density of the sintered specimens with different addi-
tions of CB as a function of temperature (a) and relative density as a
function of C content for specimens sintered at 18001C with the addition
of CB and CTP (b). CTP, coal tar pitch.

March 2009
The Effect of Surface Oxides During Hot Pressing of TiB2
625
Fig. 2. Micrographs of TiB2 hot pressed at 18001C. (a) Overview of the sample close to the surface (the surface is located exactly at the left edge of the
micrograph), (b) porous layer close to surface, (c) bulk of the sample and (d) intermediate layer.
as recently reported by Jensen et al.23 Reflections due to un-
known phases were also observed in sample TiB212.05C. Here
we have compared with reference patterns of all known phases
in the system Ti–B–O–C, and evidence for other phases con-
taining O than TiO1ꢀxCx could not be found. B2O3 is not ex-
pected to be observed since it is usually amorphous and does not
form any intermediate phases with TiO2.24
ferred as carbothermal reduction of nonvolatile titanium oxide
as further discussed below. A minor loss of humidity near 1001C
and slow increase in weight due to oxidation by oxygen impu-
rities in the inert gas is also evident, while the high weight loss
below 4001C for the powder containing CTP is due to loss of
volatile hydrocarbons.
The secondary phase in the sample without addition (TiB2)
was examined using EPMA, and the results are presented in
Table II. This phase consists mainly of the elements Ti, O, C and
the composition was different for the secondary phase located in
the bulk compared with that in the intermediate layer, which
had a higher carbon C to O ratio and higher Ti content. The
carbon content of the secondary phase as measured by EMPA is
somewhat lower than indicated from the diffractograms.
The observed weight loss after sintering is included in Table I
as well as the estimated weight loss corresponding to the vola-
tilization and carbothermal reduction of the oxides. In order to
give more insight into the evaporation during the initial sintering
of TiB2, thermogravimetrical analysis of the powders were per-
formed. The weight loss of the pure TiB2 powder, the powder
containing 2.55 wt% CTP and the powder containing 2.05 wt%
CB are shown in Fig. 5. Onset of weight loss of the pure TiB2
powder is evident close to 12001C, which can be attributed to
loss of volatile B2O3(g). Above 12001C the weight loss of the
powders with the addition of CTP and CB are significantly
higher than the weight loss of the pure TiB2 powder. This is in-
IV. Discussion
(1) Thermodynamic Analysis
In order to understand the weight loss and the influence of the
secondary phases on the densification of TiB2, a thermodynamic
analysis of the Ti–B–O–C system has been performed. The pres-
ent minority phases as well as the vapor pressure of volatile
species as a function of temperature and initial composition has
been investigated using FactSage software.25 It should be noted
that the SGTE database used in the thermodynamic calculations
includes thermodynamic data of only pure substances and ther-
modynamic models for liquid and solid solutions are not taken
into account in the analysis. The formation of TiO1ꢀxCx solid
solution is therefore not predicted by the analysis simply due to
the lack of thermodynamic data for the solid solution. The
analysis will still provide an understanding of the evaporation of
the oxide species in the system.
First, a thermodynamic analysis of pure stoichiometric TiB2
powder with the oxide impurities is presented. Figure 6(a) shows

626
Journal of the American Ceramic Society—Jensen et al.
Vol. 92, No. 3
Fig. 3. Micrographs of bulk of samples with different amounts of CTP hot pressed at 18001C. (a) TiB2, (b) TiB211.67CTP, (c) TiB212.55CTP and (d)
TiB215.1CTP. CTP, coal tar pitch.
the minority condensed phases, and Fig. 6(b) shows the gaseous
species in the system 1 mol Ti12 mol B10.07 mol O (1Ti12B1
0.07O11000Ar) as a function of temperature, corresponding to
the TiB2 powder with the initial oxygen content. An amount of
1000 mol Ar was used to simulate the flowing Ar atmosphere.
The initial weight loss at relatively low temperatures is due to
the volatilization of B2O3. The initially B2O3(g) is the main gas
specie, while above approximately 12001C the main volatile
oxide species are B2O2(g) and further BO(g) with increasing
temperature.
A stepwise reduction of titanium oxide is also taking place as
the temperature is increasing. At high temperatures (417501C)
the main secondary phase is TiO(s), which melts to TiO(l) and
finally TiO(g) and Ti(g) starts to evolve. From Fig. 6 it can be
concluded that the amount of condensed oxide species are re-
duced due to the volatile nature of boron oxides (low temper-
atures) and titanium oxide (high temperatures).
Above 0.07 mol C, coinciding with the amount of O, the bo-
ron and titanium oxides is completely reduced to CO(g) and re-
sidual carbon will be present together with TiB2. Residual
carbon was also observed in the materials with high carbon
content.
Since stoichiometric amounts of Ti and B were used, TiC does
not form according to the thermodynamics. Here the analysis
shown that oxidization of TiB2 at low temperature will lead to
formation of both titanium oxide and boron oxide. However, as
shown in Fig. 6, loss of volatile boron oxide is expected and the
oxide scale on TiB2 powders is more likely to be mainly titanium
oxide. The consequence of carbothermal reduction of TiB2 with
mainly titanium oxide scale is shown in Fig. 8. Here the amount
of minority condensed phases as well as the main gaseous species
in the system Ti11.97B10.02O11000 Ar is shown as a function
of carbon addition at 18001C. In this case the carbothermal re-
action can be summarized as
The effect of carbon addition was then analyzed. The minor-
ity condensed phases and the main gaseous species for the same
system as a function of carbon addition at 18001C are displayed
in Fig. 7. The amount of the main secondary phase TiO(l) and
BO(g) is reduced with increasing carbon contents due to carbo-
thermal reduction of the oxides
TiOðlÞ þ CðsÞ þ TiB2ðsÞ ¼ 2TiBðsÞ þ COðgÞ
(3)
Increasing the carbon content further will case the formation
of TiC(s) on the expense of TiB(s)
TiOðlÞ þ 2BOðgÞ þ 3CðsÞ ¼ TiB2ðsÞ þ 3COðgÞ
(2)
2TiBðsÞ þ CðsÞ ¼ TiB2ðsÞ þ TiCðsÞ
(4)

March 2009
The Effect of Surface Oxides During Hot Pressing of TiB2
627
Fig. 5. Thermogravimetry data for (a) TiB2, (b) TiB212.05CB and
(c) TiB212.55CTP. CTP, coal tar pitch.
tually evaporates at temperatures around 20001C. However,
densification of TiB2 takes place far below 20001C and com-
plete volatilization of titanium oxide is not likely to occur in
accordance with the observation of the secondary phases (Table
II). On the other hand, as showed in this analysis, titanium oxide
can be carbothermically reduced at lower temperatures with the
addition of C. It should be mentioned that although both TiO
and TiC are present in this system at different conditions they
are not coexistent according to this analysis since solid solutions
are not taken into consideration in the calculations.
Fig. 4. Diffractograms of (a)TiB2 powder and hot pressed samples of
powder composition (b)TiB2, (c) TiB212.05CB and (d) TiB212.55CTP
( ꢁ , TiB2; ~, TiO1ꢀxCx; ꢂ, unknown phase). CTP, coal tar pitch.
Because TiO and TiC have the same rock salt the reduction
of TiO is proposed to occur by formation of TiO1ꢀxCx solid
solution rather than the formation of TiB as predicted here.
However, as mention previously the thermodynamic data for
the solid solution were not available for the thermodynamic
calculations.
Based on the thermodynamic analysis presented here the
amount of the condensed boron oxide present in the powder
can be reduced due to volatilization at temperatures as low as
12001C in accordance with our observed weight loss (Fig. 5).
Based on the thermodynamics the residual titanium oxide is
gradually reduced as the temperature increases until TiO(l) even-
Table II. Chemical Composition of the Secondary Phase in
TiB2 Hot Pressed at 18001C
Ti (at.%)
O (at.%)
C (at.%)
B (at.%)
Interior
47.370.3 40.370.8
7.570.2 4.970.7
Intermediate 54.470.8 32.071.3 10.470.3 3.270.9
layer
Fig. 6. Minority condensed phases (a) and the gaseous species (b) in the
system 1Ti12B10.07O11000Ar as a function of temperature.

628
Journal of the American Ceramic Society—Jensen et al.
Vol. 92, No. 3
Table III. Estimated Weight Loss Due to Volatilization and
Carbothermal Reduction of Oxide Impurities
Reaction
Weight loss (wt%)
B2O3(s ) 5 B2O3(g)
B2O3(l)13.5 C(s) 5 0.5 B4C(s)13 CO(g)
TiO2(s)13 C(s) 5 TiC(s)12 CO(g)
1.5
1.8
1.2
oxide impurities due to the reducing atmosphere set up by the
graphite interior of the furnace will also contribute to the ob-
served weight loss.
For the TiB2 powder with addition of 2.55 wt% CTP a total
weight loss of 3.6% was found during TG as shown in Fig. 4.
Except for the evaporation of volatile hydrocarbons at temper-
atures below 4001C (B1.5 wt%), a weight loss of 2.1 wt% was
observed above B12001C. The corresponding weight loss for
the sample with the addition of 2.05 wt% CB was 1.9 wt%. The
increased weight loss for these samples at higher temperatures is
due to carbothermal reduction of oxide impurities in the pow-
der. Complete carbothermal reduction of the oxide impurities is
estimated to give a weight loss of 3.0%. According to this, even
though the amount of oxide impurities is considerably reduced,
the carbothermal reduction of the oxide impurities is not com-
plete. This can also explain the difference between the measured
weight loss of the sintered samples and the estimated weight loss
for complete reduction of the oxide impurities (Table I). This
may be explained by the formation of TiO1ꢀxCx(s) taking place
instead of complete reduction of the titanium oxide to TiC(s).
Furthermore, the weight loss is also related to diffusive gas
transport out of the body through the intergranual porosity,
which will be kinetically hindered with increasing densification
and reducing porosity. Previous observation of the correlation
between the evaporation rate and the specific surface area26 can
explain the relatively high weight loss for the low density sam-
ples compared with the samples with higher densities.
Fig. 7. Minority condensed phases and main gaseous species in the
system Ti12B10.07O11000Ar as a function of carbon content at
18001C.
(2) Weight Loss
The estimated weight loss given in Table I are calculated by es-
timating the weight loss corresponding to the volatilization of
B2O3, carbothermal reduction of titanium oxide and carbother-
mal reduction of boron oxide (for the samples containing CTP
the evaporation of volatile hydrocarbons is also taken into ac-
count). The weight loss according to these reactions is given in
Table III. Here the oxygen content (1.7 wt%) in the raw powder
was assumed to be present as TiO2 and B2O3 in a molar ratio
corresponding to TiB2. Depending on the C addition it is as-
sumed that (i) only evaporation of boron oxide takes place,
(ii) volatilization of boron oxide and carbothermal reduction of
titanium oxide takes place or, for the samples with higher C
additions, (iii) carbothermal reduction of both titanium oxide
and boron oxide takes place.
A weight loss of 0.8% was found for the pure TiB2 powder
during thermogravimetry, as shown in Fig. 4. According to the
thermodynamic analysis this weight loss is due to loss of B2O3.
However, complete volatilization of B2O3 will give a weight loss
of 1.5%, indicating that the volatilization of B2O3 is not com-
plete at the temperature and time interval of the thermogravi-
metric investigation or that the oxide in the TiB2 powder are
enriched on titanium oxide (lower B2O3 content). Carbothermal
reduction of oxides due to the possible presence of some C im-
purities in the powder (o0.5 wt% according to the powder
specification) may also account for some of the weight loss. A
higher weight loss (1.4 wt%) was found for the TiB2 powder hot
pressed at 18001C (Table I). This can be explained by the higher
temperature compared to the TG. Carbothermal reduction of
(3) Microstructure
TiO1ꢀxCx(s) was observed in all the samples, however, the
amount of TiO1ꢀxCx(s) was reduced with increasing addition
of C. Boron oxide evaporates as the temperature increases, re-
sulting in a titanium oxide rich layer covering the particles. The
volatile boron oxide is reduced by carbon addition or by react-
ing with the carbon surrounding the sample according to the
following reaction
B2O3ðgÞ þ CðsÞ ¼ ðBOÞ2ðgÞ þ COðgÞ
(5)
The formation of CO(g) may reduce the residual titanium
oxide according to the following reaction
TiOðlÞ þ 3ð1 ꢀ xÞCOðgÞ ¼ TiC1ꢀxOxðsÞ þ 2ð1 ꢀ xÞCO2ðgÞ
(6)
This is in agreement of the findings of Walker and Saha27
who have shown that formation of TiC takes place at temper-
atures above 1600 K upon reduction of titanium oxide due to
the reducing nature of the atmosphere in graphite furnaces. Afir
et al.28 have also reported that TiO2 is progressively reduced by
graphite involving CO(g) and producing CO2(g) above 1500 K.
Moreover, TiO1ꢀxCx(s) is known to form before formation
of pure TiC(s) during carbothermal reduction of TiO2(s).29,30
Titanium carboxides have also been identified in TiB2 materials
by Baumgarter and Steiger31 who found minor amounts of a
TiC–TiO solid solution at the triple points in TiB2 fabricated by
using high-purity powder synthesized in a plasma/arc heater.
Because of the complete solid solubility between TiC and TiO,32
liquid TiO1ꢀxCx (low values of x) will be carbothermally re-
duced upon formation of solid TiO1ꢀxCx (higher values of x).
Fig. 8. Minority condensed phases and gaseous species in the system
Ti11.97B10.02O11000Ar as a function of carbon addition at 18001C.

March 2009
The Effect of Surface Oxides During Hot Pressing of TiB2
629
Therefore, the composition of the oxycarbide may affect the
sintering process as it change from being liquid to solid and is
important with respect to distribution of this secondary phase.
The oxide impurities can be carbothermally reduced by both
CB and CTP during the sintering process, as shown by the thermo-
gravimetric investigation (Fig. 4) and reflected by the measured
weight loss (Table I). However, as shown in Fig. 1(b) the addi-
tion of CTP gave an increase in the density of the samples,
whereas the addition of CB lowered the density. The difference
in density of the samples with these different carbon sources can
be attributed to the difference in the distribution of the carbon
source. The CTP was more homogeneously distributed com-
pared with the CB particles (this was observed by the distribu-
tion of the final porosity in the sintered materials), and reduction
of oxide impurities therefore leads to an increase in density. This
is in accordance to earlier studies which concluded that the
oxide impurities lead to an increase in transport mech
anisms which promotes coarsening during sintering.8,33 In the
case of CTP (Fig. 1(b)) there is a peak in the relative density
when the carbon content corresponds close to the level of ox-
ygen impurities in the TiB2 powder. Increasing the carbon con-
tent further resulted in lower densities which indicate that
residual carbon, not consumed in the carbothermal reduction
process, inhibits the densification of the samples. The density of
the sample added extra oxides had the higher density than the
sample without any additions. According to the thermodynamic
analysis, TiO(l) will form well below the sintering temperature
(Fig. 6). The high density in samples containing oxides is ex-
plained by the formation of a liquid titanium oxide which induces
liquid phase sintering. Below the melting point of TiO(s) the vol-
atility of the surface oxides will promote coarsening due to mass
transport by evaporation–condensation and not densification.
The different chemical conditions through the sample from
the interior to the surface affect the sinterability, as reflected by
the micrographs in Fig. 2. In the outer part of the sample, the
environment is more reducing than in the bulk. Therefore it is
likely that the composition and the amount of the secondary
phase is different than the secondary phase in the interior of the
sample, causing differences in the densification rate. As the tem-
perature increases it is likely that all the boron oxide will escape
from the outer region of both the samples without any additions
and the samples containing carbon, leaving behind a surface
layer of titanium oxide which eventually will be reduced upon
formation of TiC as discussed above. However, in the bulk of
the materials the situation is different. In the case of the TiB2
sample, the oxide impurities in the bulk are only partially re-
duced upon formation of TiO1ꢀxCx(s). With this in mind, com-
paring the microstructure of the outer region of the samples and
that of the bulk, it is likely that the TiC present in the outer
region inhibits densification to a larger extent than the second-
ary phases present in the bulk of the material without any ad-
dition. In the case of the samples with the additions of C most of
the titanium oxide and boron oxide in the bulk will be reduced
upon formation of TiB2, resulting in a very low amount of sec-
ondary phase. The carbothermal reduction of the secondary
phase in the samples containing up to 2.55 wt% CTP results
in higher densities. The increase in the thickness of the porous
layer with the addition of CB can be attributed to the increased
diffusive gas transport which depends on the density and the
densification rate as pointed out by Einarsrud et al.26 The in-
termediate layer found in the samples without carbon additions
may form due to the evaporation and diffusion of TiO(g) to-
wards the sample surface where it will be reduced according to a
reaction corresponding to reaction (6). In the samples added
carbon, where no intermediate layer was found, the carbother-
mal reduction of TiO takes place in the interior of the samples.
No significant difference in grain size was found between the
different samples sintered at 18001C. This is in contrast to other
studies where oxide impurities were found to have detrimental
effects on densification and especially grain growth due to in-
creasing mass transport by evaporation–condensation at lower
temperatures33 or by increasing surface diffusivity8 at higher
temperatures, both leading to coarsening. Even though
exaggerated grain growth did not occur in the present study,
exaggerated grain growth due to the presence of oxide impurities
at higher temperatures is possible.
V. Conclusion
It has been shown that oxide impurities in TiB2 powders can be
carbothermically reduced during sintering, which may also in-
crease the density of the ceramics. Densities above 97% was
achieved for materials hot pressed at 18001C with the addition
of CTP corresponding to the oxide impurities, whereas only
93% was achieved for the samples without any carbon added.
An increase in the amount of C above the optimal gave a lower
density due to the presence of residual carbon. No sign of ex-
aggerated grain growth was found in any specimens at temper-
atures up to 18001C. A more homogeneous distribution of
carbon as CTP, compared with CB particles, was shown to re-
sult in a more homogeneous and dense microstructure. A lay-
ered microstructure was observed in most of the hot-pressed
materials, with a porous outer surface layer and a dense bulk.
This was attributed to the difference in the composition of the
secondary phase caused by the evaporation, diffusive transport,
and carbothermal reduction of volatile boron oxide. Finally
TiO1ꢀxCx(s), which is a solid solution of TiC and TiO, was
found in most of the materials demonstrating that the carbo-
thermal reduction of the oxide impurities were not completed
before pore closure during sintering.
References
1R. A. Cutler, ‘‘Engineering Properties of Borides’’; pp. 787–803 in Ceramics and
glasses, Engineered Materials Handbook, Vol. 4, Edited by S. J. Schneider. ASM
International, Materials Park, OH, 1992.
2T. Lundstrom, ‘‘Transition Metal Borides’’; pp. 351–76 in Boron and Refrac-
tory Borides, Edited by V. I. Matkovich. Springer, Berlin, 1977.
3D. Weirauch, W. Krafick, G. Ackart, and P. Ownby, ‘‘The Wettability of Ti-
tanium Diboride by Molten Aluminum Drops,’’ J. Mater. Sci., 40 [9] 2301–6 (2005).
4M. S^rlie and H. A. Øye, Cathodes in Aluminium Electrolysis, 2nd edition, Al-
uminium-Verlag, Dusseldorf, 1994.
5L. G. Boxall, A. V. Cooke, and H. W. Hayden, ‘‘Use of TiB2 Cathode Material
Application and Benefits in Conventional vss Cells’’; pp. 573–88 in Light Metals:
Proceedings of Sessions, AIME Annual Meeting. Metallurgical Soc of AIME,
Warrendale, Pennsylvania, 1984.
6R. Telle, L. S. Sigl, and K. Takagi, ‘‘Boride-Based Hard Materials’’; pp. 802–
945 in Handbook of Ceramic Hard Materials, Vol. 2, Edited by R. Riedel. Wiley-
VCH, Weinheim, 2000.
7M. K. Ferber, P. F. Becher, and C. B. Finch, ‘‘Effect of Microstructure on the
Properties of TiB2 Ceramics,’’ J. Am. Ceram. Soc., 66 [1] C2–4 (1983).
8S. Baik and P. F. Becher, ‘‘Effect of Oxygen Contamination on Densification of
TiB2,’’ J. Am. Ceram. Soc., 70 [8] 527–30 (1987).
9N. J. Finch, ‘‘The Mutual Solubilities of Titanium and Boron in Pure Aluminum,’’
Metal. Mater. Trans.. B, Process Metal. Mater. Proc. Sci., 3 [10] 2709–11 (1972).
10M. K. Bannister and M. V. Swain, ‘‘A Preliminary Investigation of the Cor-
rosion of a TiB2/BN/AIN Composite During Aluminium Evaporation,’’ Ceram.
Int., 15 [6] 375–82 (1989).
11H. R. Baumgartner, ‘‘Subcritical Crack Velocities in Titanium Diboride
Under Simulated Hall-Heroult Cell Conditions,’’ Am. Ceram. Soc. Bull., 63 [9]
1172–5 (1984).
12H. R. Baumgartner, ‘‘Mechanical Properties of Densely Sintered High-Purity
Titanium Diborides in Molten Aluminum Environments,’’ J. Am. Ceram. Soc., 67
[7] 490–7 (1984).
13R. C. Dorward, ‘‘Aluminum Penetration and Fracture of Titanium Diboride,’’
J. Am. Ceram. Soc., 65 [1] C6–6 (1982).
14C. B. Finch and V. J. Tennery, ‘‘Crack Formation and Swelling of TiB2–Ni
Ceramics in Liquid Aluminum,’’ J. Am. Ceram. Soc., 65 [7] C100–1 (1982).
15K. K. Nord-Varhaug, ‘‘TEM Investigation of Impurity Phases and the Pen-
etration of Liquid Aluminum in Hot Isostatically Pressed TiB2 Compacts,’’ J. Am.
Ceram. Soc., 79 [5] 1147–54 (1996).
16W. A. Zdaniewski, ‘‘Role of Microstructure and Intergranular Phases in Stress
Corrosion of TiB2 Exposed to Liquid Aluminum,’’ J. Am. Ceram. Soc., 68 [11]
309–12 (1985).
17W. A. Zdaniewski, ‘‘Degradation of Hot-Pressed TiB2–TiC Composite in
Liquid Aluminum,’’ Am. Ceram. Soc. Bull., 65 [10] 1408–14 (1986).
18W. A. Zdaniewski, ‘‘Effect of Segregated Cr on Degradation of (Ti,Cr)B2
Exposed to Liquid Aluminum,’’ J. Electrochem. Soc., 133 [9] 1777–81 (1986).
19M. S. Jensen, M. Pezzotta, Z. Zhang, M.-A. Einarsrud, and T. Grande,
‘‘Degradation of TiB2 in Liquid Aluminum,’’ J. Eur. Ceram. Soc., 28 3155–64
(2008).
20S. H. Kang and D. J. Kim, ‘‘Synthesis of Nano-Titanium Diboride Powders
by Carbothermal Reduction,’’ J. Eur. Ceram. Soc., 27 [2–3] 715–8 (2007).

630
Journal of the American Ceramic Society—Jensen et al.
Vol. 92, No. 3
21. ‘‘e-Metalog.’’ In, Vol. 2007. Struers http://www.struers.com/modules/emeta
28A. Afir, M. Achour, and A. Pialoux, ‘‘Carbothermic Reduction of Titanium-
Dioxide by High Temperature X-Ray Diffraction with Controlled Pressure,’’
J. Alloys Compd., 210 [1–2] 201–8 (1994).
log/, 2007.
22A. Afir, M. Achour, and N. Saoula, ‘‘X-ray Diffraction Study of Ti–O–C
System at High Temperature and in a Continuous Vacuum,’’ J. Alloys Compd.,
288 [1–2] 124–40 (1999).
29R. Koc, ‘‘Kinetics and Phase Evolution During Carbothermal Synthesis of
Titanium Carbide from Ultrafine Titania/Carbon Mixture,’’ J. Mater. Sci., 33 [4]
1049–55 (1998).
23M. S. Jensen, M.-A. Einarsrud, and T. Grande, ‘‘Preferential Grain Orienta-
tion in Hot Pressed TiB2,’’ J. Am. Ceram. Soc., 90 [4] 1339–41 (2007).
24V. N. Pavlikov, V. A. Yurchenko, and S. G. Tresvyatskii, ‘‘B2O3-TiO2
system,’’ Russ. J. Inorg. Chem., 21 [1] 126–7 (1976).
30A. Maitre and P. Lefort, ‘‘Carbon Oxidation at High Temperature During
Carbothermal Reduction of Titanium Dioxide,’’ Phys. Chem. Chem. Phys., 1 [9]
2311–8 (1999).
25C. W. Bale, P. Chartrand, S. A. Degterov, G. Eriksson, K. Hack, R. Ben
Mahfoud, J. Melan@on, A. D. Pelton, and S. Petersen, ‘‘FactSage Thermochemical
Software and Databases,’’ Calphad, 26 [2] 189–228 (2002).
31H. R. Baumgartner and R. A. Steiger, ‘‘Sintering and Properties of Titanium
Diboride Made from Powder Synthesized in a Plasma-Arc Heater,’’ J. Am. Ceram.
Soc., 67 [3] 207–12 (1984).
26M. A. Einarsrud, E. Hagen, G. Pettersen, and T. Grande, ‘‘Pressureless Sinte-
ring of Titanium Diboride with Nickel, Nickel Boride, and Iron Additives,’’ J. Am.
Ceram. Soc., 80 [12] 3013–20 (1997).
G. Neumann, R. Kieffer, and P. Ettmayer, ‘‘Uber Das System TiC–TiN–
TiO,’’ Monatshefte fur Chemie/Chemical Monthly, 103 [4] 1130–7 (1972).
32
.
¨
33C. B. Finch, P. F. Becher, P. Angelini, S. Baik, C. E. Bamberger, and
J. Brynestad, ‘‘Effect of Impurities on the Densification of Submicrometer TiB2
27J. K. Walker and C. K. Saha, ‘‘Formation of a Surface Carbide Layer During
Sintering of Titanium Diboride,’’ J. Am. Ceram. Soc., 71 [4] 207–9 (1988).
Powders,’’ Adv. Ceram. Mater., 1 [1] 50 (1986).
&