Preparation of Ti3SiC2 by electron-beam-ignited solid-state reaction

Article abstract of DOI:10.1111/j.1151-2916.1998.tb02733.x

This paper describes a novel way to prepare the ternary phase Ti₃SiC₂ in a single-step procedure that we call electron-beam-ignited solid-state reaction (EBI-SSR). The preparation route is discussed by means of an isothermal section of the Ti-Si-C phase diagram. Properties such as the Vickers hardness and the electrical resistivity of the resulting samples are presented. Our property data compare well to those that have been published. The main advantages of this preparation method are the controllability of process parameters such as heating rates, temperatures, and times, as well as the short duration of the overall sample preparation. However, a disadvantage is the presence of second phases (typically in amounts of <8%) that must be reduced via further optimization of the process.

Full text of DOI:10.1111/j.1151-2916.1998.tb02733.x

J. Am. Ceram. Soc., 81 [11] 3025–28 (1998)
ournal
J
Preparation of Ti3SiC2 by Electron-Beam-Ignited Solid-State Reaction
Fred Goesmann, Roland Wenzel, and Rainer Schmid-Fetzer†
Technische Universit¨at Clausthal, AG Elektronische Materialien, D-38678 Clausthal-Zellerfeld, Germany
This paper describes a novel way to prepare the ternary
phase Ti3SiC2 in a single-step procedure that we call elec-
tron-beam-ignited solid-state reaction (EBI-SSR). The
preparation route is discussed by means of an isothermal
section of the Ti-Si-C phase diagram. Properties such as the
Vickers hardness and the electrical resistivity of the result-
ing samples are presented. Our property data compare well
to those that have been published. The main advantages of
this preparation method are the controllability of process
parameters such as heating rates, temperatures, and times,
as well as the short duration of the overall sample prepa-
ration. However, a disadvantage is the presence of second
phases (typically in amounts of <8%) that must be reduced
via further optimization of the process.
strated the possibility of producing very-high-purity ␶-phase
material via the hot isostatic pressing (HIP) of powder mixtures
of titanium, SiC, and graphite. The samples showed no porosity
and are the best material that has been produced in bulk form
thus far.
II. Sample Preparation by Electron-Beam-Ignited
Solid-State Reaction (EBI-SSR)
We prepared Ti3SiC2 using an electron-beam furnace that
was initially designed as an evaporation source. This type of
heating has the advantages of having an ultrahigh vacuum
(UHV) environment and very fine power control, while the
samples are constantly visible, which makes it ideal for opti-
mizing the heating procedures with very good reproducibility.
The samples were prepared from titanium (99.9%, 100 mesh,
Johnson Matthey, Ward Hill, MA) and SiC of 3C modification
(99%, SGL Carbon AG, Werk Ringsdorff, Bonn, Germany)
powders, which were weighed so that the amounts of titanium
and carbon corresponded to the required composition in the
ternary phase. This procedure was followed for two reasons: (i)
it is the preparation method that is most similar to our actual
electrical contact formation, where a titanium film on SiC is
annealed, and (ii) it gave the highest purity of the ternary phase.
The silicon composition will be discussed later. The powder
was mixed in a ball mill and cold-pressed into pellets of up to
500 mg in a hydraulic press using a pressure of 430 MPa. The
green density of this unreacted pellet was 73% of the theoret-
ical density of the Ti + SiC mixture, which is not high consid-
ering that it is a mixture of a ductile material (titanium) and
hard grains (SiC). The actual reaction occurred in a high
vacuum chamber (Model 306A, Edwards, Grand Island, NY),
where the sample was heated in an electron-beam furnace
(Model EB1, Edwards) using a graphite crucible. Graphite was
the most readily available and promising candidate as crucible
material, because we wanted to stay within the ternary system.
The pellets did not stick, which indicated negligible reaction
with the crucible. The pellets were heated, to make them glow
homogeneously. At a temperature of 900°C, they ignited and
reacted while the temperature increased above 1700°C (which
was the temperature at which the thermocouple was destroyed)
but below 2617°C (where a small piece of molybdenum foil
should have shown signs of melting). The reaction front moved
through the pellet at a rate of 1 cm/s. The pellets were heated
further by the electron beam at 1500°–1600°C after the reac-
tion occurred, to evaporate excess silicon. The total heat treat-
ment was completed in 5 min.
I. Introduction
ITANIUM SILICON CARBIDE (Ti3SiC2) is an interesting com-
pound in the Ti-Si-C ternary system for several reasons.
T
Firstly, Ti3SiC2 is a machinable ceramic material with a low
hardness1 and good oxidation resistance.2 Secondly, it can be
found in titanium–SiC metal-matrix composites (MMCs) as a
reaction phase between titanium and SiC.3 Thirdly, Ti3SiC2 is
the boundary-layer phase that determines the electrical prop-
erties of a titanium contact on semiconducting single-crystal
6H-SiC.4
The purpose of this study was to investigate an inexpensive
and simple way to prepare Ti3SiC2; we call this method elec-
tron-beam-ignited solid-state reaction (EBI-SSR).‡ This study
was undertaken so that this material could be available for bulk
diffusion studies, because it is not available commercially.
Several methods to prepare the ternary phase Ti3SiC2 (␶)
have been suggested. The first method, which was used by
Jeitschko and Nowotny,5 was a solid-state reaction where the
␶-phase crystals were formed inside cavities in bulk samples
and selected by hand. From these specimens, the crystallo-
graphic structure was described for the first time. To prepare
single-crystal Ti3SiC2, chemical vapor deposition (CVD) reac-
tions were used.6,7 A solid-state combustion reaction from el-
emental powders was suggested by Pampuch et al.8 The re-
sulting material contained ␶-phase and a minor but unspecified
quantity of TiC1−y. High-purity powder (less than 5% TiC1−y
)
was produced via a solid-state reaction and subsequent tedious
chemical removal of silicides and carbides by Racault et al.2 A
method to prepare mixtures of SiC and ␶-phase from silicon
and TiC1−y via solid-state reaction was suggested by
Radhakrishnan et al.9 Furthermore, the ␶-phase was found in
reaction zones between SiC and titanium.3,4 Arunajatesan et
al.10 proposed a way to synthesize Ti3SiC2 via arc melting of
compressed mixtures of elemental powders and subsequent an-
nealing, which was difficult because of weight losses during
preparation. Most recently, Barsoum and El-Raghy1 demon-
Some of the reacted pellets were polished metallographically
and examined via optical microscopy as well as scanning elec-
tron microscopy (SEM); others were ball milled to produce
material for X-ray diffractometry (XRD) study. In addition,
chemical analysis and measurements of electrical resistivity
and hardness were performed.
Attempts that began with elemental powders, similar to the
approach used by Pampuch et al.,8 led to very vigorous reac-
tions that left behind the ternary phase and >10% TiC1−y, al-
most regardless of the amount of excess silicon used in the
pellet preparation. Starting with TiSi2, TiC1−y, and SiC led to
reaction products that contained more than two phases.
The reactions can be depicted in an isothermal section of the
A. Carim—contributing editor
Manuscript No. 190526. Received December 4, 1997; approved August 18, 1998.
†Author to whom correspondence should be addressed.
‡German Patent No. 19749050.
3025

3026
Communications of the American Ceramic Society
(2) Weight Loss
Vol. 81, No. 11
Ti-Si-C phase diagram; this section basically has been taken
from Goesmann et al.4 and has been modified here by the
inclusion of a solution range of ␶-phase. Figure 1(a) shows this
section for 1000°C; Fig. 1(b) is an enlarged view of this sec-
tion. This figure is discussed in detail in Section III(5). The
solid-state equilibria is assumed to remain mainly constant over
the temperature range that has been used in the preparation.
The dotted line in the figure from Ti to SiC is the range of
possible starting compositions if only titanium and SiC are
used. The solid square in the figure indicates the actual starting
composition of the pellets. The dotted line from Si through ␶
toward TiC1−y is the possible shift in composition, provided
that only silicon evaporates. We have started from the solid
square and intend to reach the stoichiometric ␶-phase, which
is located at the left end of the broad line, via reaction and
annealing.
If the excess silicon completely evaporates, a 12.5% loss in
weight is expected, to attain the stoichiometric ␶-phase com-
position. On average, the pellets lost 4% of weight during the
heating procedure, which indicates that the silicon bakeout was
incomplete. The total weight loss is low for a UHV environ-
ment; because of the fine powder control and the ability to
observe the samples during preparation, such results can be
achieved.
(3) Chemical Analysis
For the chemical analysis, samples were prepared by omit-
ting the final heating step, to have an intermediate product to
prove the assumption of silicon loss. Hence, the weight loss
was smaller (only 1.2%).
Carbon content was analyzed in a carbon–sulfur analyzer
(Model CSA 2003, Leybold–Heraeus, Cologne, Germany),
whereas silicon content was analyzed via solution in sodium
potassium peroxide and precipitation of SiO2. Titanium content
was analyzed via atomic absorption spectroscopy.
III. Analytical Results and Discussion
(1) Ignition Conditions
The solid triangle shown in Figs. 1(a) and (b) is the compo-
sition as determined by chemical analysis of the samples after
the ignition step (that is, without annealing for an additional 5
min). The composition results are as follows (in at.%): tita-
nium, 46.2; silicon, 21.7; and carbon, 32.1. The composition
apparently has shifted from the starting values along the direc-
tion of silicon loss only. The open circle in Fig. 1(a), which is
half-covered by the solid square, is the theoretical composition
of the ignited samples calculated from the 1.2% weight loss,
which is assumed to be due to silicon loss only.
The ignition temperature of 900°C correlates well with ear-
lier investigations. The ternary phase was only found in tita-
nium/SiC bulk-diffusion couples annealed at temperatures
Ն850°C in these earlier studies.4
(4) X-ray Diffractometry
The reaction products of the completely heat-treated samples
were verified by XRD using CoK␣ radiation and an iron filter
in an X-ray diffractometer (Model D5000, Siemens, Karlsruhe,
Germany). The samples consisted of Ti3SiC2 with small
amounts of TiSi2 or sometimes TiC1−y. A typical diffraction
pattern is shown in Fig. 2, which shows the diffraction intensity
plotted against the 2␪ angle. Most of the peaks are of the
␶-phase and are marked by chevrons (‘‘ ’’). All peaks in the 2␪
Fig. 1. (a) Isothermal section of the Ti-Si-C phase diagram at
1000°C. (b) Enlarged view of Fig. 1(a), showing the details of the
sample compositions and the minimum extent of a proposed homoge-
neity range of the ternary phase Ti3SiC2 toward SiC, as well as the
resulting two-phase regions ␶ + TiC1−y, ␶ + TiSi2, and ␶ + Ti5Si3; the
stoichiometric composition of the ternary phase is at the left end of the
broad line.
Fig. 2. XRD pattern of the prepared material. Most of the peaks are
from Ti3SiC2; these peaks are marked by chevrons (‘‘ ’’). The position
of the tallest TiC1−y peak is marked by a plus (‘‘+’’), and the most
prominent TiSi2 peaks are marked by asterisks (‘‘ ’’). The best
*
samples, which were obtained in 5 min, resulted in an almost-
XRD-‘‘pure’’ ␶-phase that contained no TiC1−y and very little TiSi2.

November 1998
Communications of the American Ceramic Society
3027
range of 20°–120° were identified. A range from 30° to 80°
was selected in Fig. 2 to show possible traces of TiC1−y and
TiSi2 most clearly. This pattern is meant to serve as an ex-
ample; it was not the only source of information. Additional
scans at higher resolutions and wider ranges were performed to
be sure about the respective phases.
No SiC was detected in the XRD patterns of the samples.
The starting material (SiC powder and titanium powder, SiC +
Ti) was investigated via XRD, and the 2␪ ranges of the respec-
tive peaks were individually and clearly identified. By com-
paring these patterns with the reacted samples, the presence of
SiC could be excluded. SiC and TiC show very similar XRD
patterns; however, two peaks outside the depicted range allow
for a proper distinction: the (311) TiC peak, at 86.52° 2␪ (33%
relative height), and the (311) SiC peak, at 85.80° 2␪ (37%
relative height). Because SiC is strictly stoichiometric, its
peaks are not shifted at all.
Analysis of TiC1−y can be performed provided that SiC is
absent. The position of the TiC1−y peak that is used (the (200)
peak, at 48.82° 2␪) is marked by a plus sign (‘‘+’’) below the
diffraction curve in Fig. 2. The amount of TiC1−y in this sample
is estimated to be <1%, because no peak is visible. The addi-
tional complication of TiC1−y not being a stoichiometric phase
was addressed by using several samples of pure TiC1−y with
varying values of y. The spectra were almost identical, with
only slight shifts in the d-values.
necessary for this homogeneity range is indicated by the broad
line in Fig. 1(b); the extents of the proposed two-phase fields
␶ + TiC1−y and ␶ + TiSi2 are marked by dashed lines.
As is evident from Fig. 1, there is no ␶ + TiSi2 + TiC1−y
three-phase equilibrium, and one would expect either ␶ + TiSi2
or ␶ + TiC1−y only. The best samples obtained—that is, the ones
with the lowest overall content of non-␶ phases—had traces of
both TiC1−y and TiSi2, which indicates that the samples cannot
be fully equilibrated in such a short processing time.
No Ti3Si5 was observed. For the sample composition to
reach the ␶ + TiC1−y + Ti3Si5 three-phase field, a greater weight
loss than that which has been observed would be necessary.
There may be several reasons why TiC1−y was found in the
samples, which contradicts the assumption of silicon loss only.
First, energy-dispersive X-ray (EDX) analysis via electron mi-
croscopy gave the result that TiC1−y grains were always sur-
rounded by pure ␶-phase, never by Ti3Si5; these may be locally
overheated areas that are not in equilibrium with the rest of the
sample. Secondly, an incorporation of carbon from the crucible
may have occurred.
(6) Microstructure
The pellets contain the ␶-phase in the form of interlocked
needles that are 3 ␮m in diameter and 10–20 ␮m long. Figure
3 shows a BSE image where crystallites of the ␶-phase are
clearly visible. The slightly darker material between the grains
is TiSi2. This statement is confirmed by EDX analysis (spot
and map analysis) of various samples. The volume fraction of
TiSi2, as determined by image analysis, is 6%–8%, which is in
agreement with the XRD results. The structure is much finer
than that observed by Barsoum and El-Raghy.1 If TiC1−y grains
occur in ‘‘overbaked’’ samples, they are surrounded by ␶-
phase. The material is porous (not shown in Fig. 3), and it is
conceivable that using higher pressures at the pelleting step
would help to reduce the porosity of the finer material.
A similar procedure can be performed for TiSi2. The three
tallest peaks are marked by asterisks (‘‘ ’’). The peaks are as
*
follows: the (311) peak, at 46.06° 2␪ (100%); the (004) peak,
at 49.47° 2␪ (40%); and the (022) peak, at 50.61° 2␪ (74%).
Comparison of the third TiSi2 peak with the 100% ␶-phase
peak ((104), at 46.06° 2␪) leads to a TiSi2 concentration of
<8%, assuming perfect isotropy, which is justified for a ball-
milled sample. The other two TiSi2 peaks are less suitable,
because they are on the shoulders of the ␶-phase peaks. The
presence of amorphous carbon or even graphite in small
amounts cannot be excluded, judging by the XRD spectra
alone. However, because such a phase would appear as being
of very low order number in a backscattered electron (BSE)
image and nothing of that sort was observed, we have con-
cluded that such phases are absent.
(7) Electrical Resistivity
The electrical resistivity of the ternary phase was measured
on a bar cut from a pellet using a four-point probe technique.
The resistivity is <100 ␮⍀иcm; however, this value is only an
upper limit, because of the pores in the material. This value is
a factor of 4 greater than that cited in the literature.1 The
resistivity for pure titanium is 50 ␮⍀иcm, and that of TiC1−y
is 60 ␮⍀иcm. We agree with Barsoum and El-Raghy1 in that
pure and dense ␶-phase is likely to have a lower resistivity than
TiC1−y and titanium.
The final composition of the sample, as determined by XRD,
is dependent on the heating procedure. If the silicon bakeout
step is omitted, the amount of silicide increases; if the tem-
perature is too high or the bake is too long, the TiC1−y amount
increases as expected, according to Gottselig et al.3 In both cases,
the reduction of one phase occurs at the expense of the growth
of the other; in some samples, traces of both phases were
found.
(5) Phase Diagram
Obviously, definitive conclusions about thermodynamic
equilibrium properties in the Ti-Si-C system cannot be made
from our experiments, because our samples were not allowed
to equilibrate properly. However, the results still cause some
tentative changes of the simple phase diagram,4 where ␶ is
assumed to be a stoichiometric line compound. We are aware
of more-recent publications10 where a more-elaborate homo-
geneity range of the ternary phase has been proposed. How-
ever, because we did not perform extended investigations on
this topic, our conclusions are an attempt to describe the mini-
mum deviations from a stoichiometric compound suggested by
our experimental results. The absence of SiC in the ignited
sample (the open circle in Fig. 1(b)) indicates a homogeneity
range of the ternary phase. The sample points in Fig. 1 are
clearly located in a three-phase region (SiC + TiSi2 + ␶) if the
␶-phase is stoichiometric. Because only the ␶-phase and TiSi2
were found in the sample, the homogeneity range of ␶ must
have a component perpendicular to the original ␶–TiSi2 tieline.
One option to achieve this relation is an extension of ␶ as a line
compound along an existing tieline toward SiC, so that the
open circle is within a two-phase region. The minimum extent
Fig. 3. BSE image of the microstructure of the sample; the lighter
areas are Ti3SiC2, and the darker areas are TiSi2.

3028
Communications of the American Ceramic Society
Vol. 81, No. 11
(8) Hardness
in the literature.1 For low-porosity and high-purity applica-
tions, use of the HIPed material would be more likely.
The Vickers hardness (HV), as measured on our single-phase
grains, was 570. Reported HV values include 400 for high-
purity HIPed material,1 600 for single crystals,7 and 750 in
less-pure form.8 Our values agree well with the literature data
that was obtained for pure material.
References
1M. W. Barsoum and T. El-Raghy, ‘‘Synthesis and Characterization of a
Remarkable Ceramic: Ti3SiC2,’’ J. Am. Ceram. Soc., 79 [7] 1953–56 (1996).
2C. Racault, F. Langlais, and R. Naslain, ‘‘Solid-State Synthesis and Char-
acterization of the Ternary Phase Ti3SiC2,’’ J. Mater. Sci., 29, 3384–92 (1994).
3B. Gottselig, E. Gyarmati, A. Naoumidis, and H. Nickel, ‘‘Joining of Ce-
ramics Demonstrated by the Example of SiC/Ti,’’ J. Eur. Ceram. Soc., 6,
153–60 (1990).
IV. Conclusion
Preparing Ti3SiC2 that contains <8% of secondary phases
via electron-beam-ignited solid-state reaction (EBI-SSR) is
possible. Three temperature levels are essential in the proce-
dure: the preanneal (for homogeneous glowing), at 800°C; ig-
nition, at 900°C; and the final silicon evaporation reaction, at
1500°–1600°C. It is conceivable that our preparation method
works in a simple furnace under vacuum or maybe in a pro-
tective atmosphere too, because only three temperature levels
are required. The process could be increased in scale easily
and, therefore, may be applicable for the production of inex-
pensive bulk quantities; this condition would foster its use as
refractory material, because this production method is faster
than the hot isostatic pressing process that has been suggested
4F. Goesmann and R. Schmid-Fetzer, ‘‘Temperature-Dependent Interface
Reactions and Electrical Contact Properties of Titanium on 6H-SiC,’’ Semicond.
Sci. Technol., 10, 1652–58 (1995) (and correction, 11, 461 (1996)).
5W. Jeitschko and H. Nowotny, ‘‘Die Kristallstruktur von Ti3SiC2—Ein
Neuer Komplexcarbid-Typ,’’ Monatsh. Chem., 98, 329–37 (1967).
6J. J. Nickl, K. K. Schweitzer, and P. Luxenberg, ‘‘Gasphasenabscheidung im
System Ti-Si-C,’’ J. Less-Common Met., 26, 335–53 (1972).
7T. Goto and T. Hirai, ‘‘Chemically Vapour Deposited Ti3SiC2,’’ Mater. Res.
Bull., 22, 1195–201 (1987).
8R. Pampuch, J. Lis, L. Stobierski, and M. Tymkiewicz, ‘‘Solid Combustion
Synthesis of Ti3SiC2,’’ J. Eur. Ceram. Soc., 5, 283–87 (1989).
9R. Radhakrishnan, C. H. Henager Jr., J. L. Brimhall, and S. B. Bhaduri,
‘‘Synthesis of Ti3SiC2/SiC and TiSi2/SiC Composites Using Displacement Re-
actions in the Ti-Si-C System,’’ Scr. Mater., 34, 1809–14 (1996).
10S. Arunajatesan and A. H. Carim, ‘‘Synthesis of Titanium Silicon Car-
bide,’’ J. Am. Ceram. Soc., 78 [3] 667–72 (1995).
ᮀ