Oxidation of Tin+1AlXn (n = 1-3 and X = C, N) II. Experimental Results

Article abstract of DOI:10.1149/1.1380256

In this, Part II of a two-part study, the oxidation kinetics in air of the ternary compounds Ti₂AlC, Ti₂AlC_0.5N_0.5. Ti₄AlN_2.9. and Ti₃AlC₂ are reported. For the first two compounds. in the 1000-1100°C temperature range and for short times (≈20 h) the oxidation kinetics are parabolic. The parabolic rate constants are k_a (m²/s) = 2.68 × 10⁵ exp - 491.5 (kJ/mol)/RT for Ti₂AlC. and 2.55 × 10⁵ exp - 458.7 (kJ/mol)/RT for Ti₂AlC_0.5N₀₅.. At 900°C, the kinetics are quasi-linear, and up to 100 h the outermost layers that form are almost pure rutile, dense, and protective For the second pair, at short times (< 10 h) the oxidation kinetics are parabolic at all temperatures examined (800-1100°C), but become linear at longer times. The k₁ values are 3.2 × 10⁵ exp - 429 (kJ/mol)/RT, for Ti₄AlN_2.9 and 1.15 × 10⁵ exp - 443 (kJ/mol)/RT for Ti₃AlC₂. In all cases, the scales that form are comprised mainly of a rutile-based solid solution. (Ti_{1 y}Al_y)O_{2 y/2} where y < 0.05, and some Al₂O₃. The oxidation occurs by the inward diffusion of oxygen and the outward diffusion of Al and Ti. The C and N atoms are presumed to also diffuse outward through the oxide layer. At the low oxygen partial pressure side, the Al³⁺ ions dissolve in and diffuse through the (Ti_{1 y}Al_y)O_{2 y/2} layer and react with oxygen to form Al₂O₃ at the high oxygen pressure side. This demixing results in the formation of pores that concentrate along planes, especially at longer times and higher temperatures. These layers of porosity impede the diffusion of Al. but not those of Ti and oxygen, which results in the formation of highly striated scales where three layers, an Al₂O₃-rich, a TiO₂-rich, and a porous layer repeat multiple (>10) times. The presence of oxygen also reduces the decomposition (into TiX₃ and Al) temperatures of Ti₄AlN_2.9 and Ti₃AlC₂ from a T > 1400°C, to one less than 1100°C.

Full text of DOI:10.1149/1.1380256

Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
C551
0013-4651/2001/148͑8͒/C551/12/$7.00 © The Electrochemical Society, Inc.
Oxidation of Tin¿1AlXn „n ϭ 1-3 and X ϭ C, N…
II. Experimental Results
M. W. Barsoum, N. Tzenov, A. Procopio, T. El-Raghy, and M. Ali
Department of Materials Engineering, Drexel University, Philadelphia, Pennsylvania 19104, USA
In this, Part II of a two-part study, the oxidation kinetics in air of the ternary compounds Ti2AlC, Ti2AlC0.5N0.5, Ti4AlN2.9, and
Ti3AlC2 are reported. For the first two compounds, in the 1000-1100°C temperature range and for short times ͑Ϸ20 h͒ the
oxidation kinetics are parabolic. The parabolic rate constants are kx (m2/s) ϭ 2.68 ϫ 105 exp Ϫ 491.5 (kJ/mol)/RT for Ti2AlC,
and 2.55 ϫ 105 exp Ϫ 458.7 (kJ/mol)/RT for Ti2AlC0.5N0.5. At 900°C, the kinetics are quasi-linear, and up to 100 h the outermost
layers that form are almost pure rutile, dense, and protective. For the second pair, at short times ͑Ͻ10 h͒ the oxidation kinetics are
parabolic at all temperatures examined ͑800-1100°C͒, but become linear at longer times. The kx values are 3.2 ϫ 105 exp
Ϫ 429 ͑kJ/mol͒/RT, for Ti4AlN2.9 and 1.15 ϫ 105 exp Ϫ 443 ͑kJ/mol͒/RT for Ti3AlC2. In all cases, the scales that form are
Ͻ
comprised mainly of a rutile-based solid solution, (Ti1ϪyAly͒O2Ϫy/2 where y
0.05, and some Al2O3. The oxidation occurs by the
inward diffusion of oxygen and the outward diffusion of Al and Ti. The C and N atoms are presumed to also diffuse outward
through the oxide layer. At the low oxygen partial pressure side, the Al3ϩ ions dissolve in and diffuse through the (Ti1ϪyAly͒O2Ϫy/2
layer and react with oxygen to form Al2O3 at the high oxygen pressure side. This demixing results in the formation of pores that
concentrate along planes, especially at longer times and higher temperatures. These layers of porosity impede the diffusion of Al,
but not those of Ti and oxygen, which results in the formation of highly striated scales where three layers, an Al2O3-rich, a
TiO2-rich, and a porous layer repeat multiple ͑Ͼ10͒ times. The presence of oxygen also reduces the decomposition ͑into TiXx and
Ͼ
Al͒ temperatures of Ti4AlN2.9 and Ti3AlC2 from a T
1400°C, to one less than 1100°C.
© 2001 The Electrochemical Society. ͓DOI: 10.1149/1.1380256͔ All rights reserved.
Manuscript submitted July 26, 2000; revised manuscript received February 6, 2001. Available electronically July 9, 2001.
In this, Part II of a two-part study,1 we report on the oxidation in
͑balance TiO2͒, followed by a TiO2-rich ͑balance Al2O3͒ layer. This,
in turn, is followed by an O embrittled alloy layer.
air in the 800-1100°C temperature range, of the ternary compounds
Ti2AlC, Ti2AlC0.5N0.5, Ti4AlN2.9, and Ti3AlC2. Since this is the first
report on the oxidation of these compounds, there are no previous
results with which to compare; it is thus instructive to review the
oxidation behavior of some related solids such as Ti, and some
Ti-aluminides such as TiAl, Ti3Al, and ‘‘Ti2Al,’’ which is a two-
Since the ternary compounds studied in this work do not dissolve
oxygen, comparing the parabolic rate constants calculated from
weight gain measurements, kw , with those calculated from oxide
film thicknesses, kx , is problematic. To circumvent this problem, the
kx values reported in this paper taken from the literature were cal-
culated directly from micrographs shown in the various references
and/or when the thicknesses of the oxide layers when explicitly
given. For example, Unnan et al.7 measured the TiO2 film thick-
nesses on commercial Ti after their exposure in air in the 593-760 °C
temperature range and deduced thata
2
phase mixture of the first two. The oxidation of Ti3SiC2 was
briefly reviewed in Part I.1
The oxidation of pure Ti in the 600-1000°C temperature range is
parabolic.3-7 In this temperature range, individual rutile TiO2 layers
form that range in thickness from 1 to 8 ␮m depending inversely on
temperature.3-7 These stratified layers tend to spall off periodically.
Simultaneously with the formation of a TiO2 scale, substantial
amounts of oxygen dissolve in the Ti substrate. The same is true for
the Ti-aluminides; most, but especially the ones for which the Ti:Al
ratio is around 2:1, dissolve substantial amounts of oxygen ͑see
below͒.
It is well established that the oxidation behavior of Ti-Al inter-
metallics is strongly dependent on the aluminum content.8-13 Typi-
cally, ϳ60 to 70 atom % Al is required in binary Ti-Al alloys to
form continuous alumina, Al2O3 scales in air.9 In pure oxygen, the
kx m2/s͒ ϭ 1.4 ϫ 10Ϫ4 exp Ϫ 231,176 kJ/mol͒/RT
͓1͔
͑
͑
The aim of this paper is to report on the oxidation behavior of
Ti2AlC, Ti2AlC0.5N0.5, Ti3AlC2, and Ti4AlN2.9 in air in the 800-
1100°C temperature range. The results are discussed in light of the
model presented in Part I1 and compared to those of relevant Ti-
aluminide intermetallics.
Experimental
Al content needed is lower.9,10 For concentrations that are less than
50 atom %, both TiO2 and Al2O3 form.8-11 The scale morphologies
that form are similar to the ones observed in this work, namely,
stratified multilayer scales. The layers that form, from the outside in,
are typically TiO2 /Al2O3-rich layer/TiO2 ϩ Al2O3/internal oxides.
In general, the oxidation resistance of Ti3Al in air, or oxygen, is
quite poor at temperatures greater than 850°C.9-13 At lower tempera-
tures and shorter time, the scales, predominately composed of TiO2,
are adherent. Above 850°C, the scales tend to be loosely adherent
and spall off. Roy et al.12 have shown that the oxidation kinetics in
the 725-1025°C temperature range, especially at short times ͑Ͻ5 h͒,
are reasonably well described by a parabolic law. The authors also
noted that the oxidation resulted in a scale in which thin Al2O3-rich
layers alternate with TiO2 layers; a layering that introduced porosity
in the reaction scales. Welsch and Kahveci9 have shown that when a
Ti-26 atom % Al sample is oxidized in dry oxygen, four layers
formed: an external TiO2 layer, followed by an Al2O3-rich layer
All of the samples were fabricated by reactively hot isostatically
pressing,
various
powder
mixture
combinations
of
Ti, TiH2 , Al4C3 , C, AlN, and/or TiN to yield the desired stoichi-
ometries. The sources and purities of the powders used in this work
are listed in Table I. The processing details can be found
elsewhere.14-16
For the Ti2AlC0.5N0.5 and Ti2AlC phases, henceforth referred to
as the 211 phases, the powders were mixed, cold pressed, and sealed
under vacuum in Pyrex tubes. The latter were then placed in a hot
isostatic press, and heated at 10°C/min to 850°C and held at that
temperature for 30 min. The chamber was pressurized with Ar to
Ϸ25 MPa, the heating resumed at the initial rate, at which time the
pressure increased to Ϸ40 MPa. The Ti2AlC0.5N0.5 and Ti2AlC
samples were processed at 1300°C for 15 and 30 h, respectively. The
final grain sizes were of the order of 23 ␮m in both cases. The
a This expression is calculated from Eq. 11 in Ref. 7, and includes the factor of 2
used in Eq. 4 in the definition of kx
.
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C552
Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
Table I. Characteristics and sources of powders used in this work.
Purity
͑%͒
Particle
size
Powder
Source
Composition in wt %a
Ti
99.46
99.3
Ϫ325 mesh
Ϫ325
Ti Powder Specialists,
Sandy, UT
Timet, Henderson,
NV
Aldrich Chemical,
Milwaukee, WI
Cerac,
C ϭ 0.03, Fe ϭ 0.04, Si ϭ 0.02, N ϭ
0.03, O ϭ 0.17, Cl ϭ 0.12, H ϭ 0.034
TiH2
for Ti4AlN3
C
99
99
d
m ϭ 1-2 ␮m
Ϫ325
Al4C3
for Ti3AlC2
Al4C3
Milwaukee, WI
Alpha Aesar,
Ward Hill, MA
99.7
d
m ϭ 10 ␮m
C ϭ 25.57; Ca, Cu, Mn, Ti, Ni, V,
for 211
Ͻ
Zr
0.01; Co ϭ 0.09; Fe ϭ 0.08;
Ͻ
Cr ϭ 0.04; Mg
0.001; Si ϭ 0.1
TiN
AlN
99.8
98.9
Alfa Aesar, Ward
Hill, MA
Alpha Aesar,
Ward Hill, MA
2.5-4 ␮m
N ϭ 32% minimum
d
m ϭ 3.8 ␮m
N ϭ 33.2, O ϭ 1.1,
C ϭ 0.08, Fe ϭ 0.09
a According to manufacturers specifications.
samples were fully dense and predominantly single phase. In both
cases, Ϸ4 vol % impurity phases that contained O and P were de-
tected. More details can be found in Ref. 14.
These reactions assume that the nonmetallic elements diffuse
through the reaction layers and oxidize. The exact nature of the gas
that forms is unknown at this time; the assumption that they are NO2
To fabricate the Ti3Al1.1C1.8 samples, Ti, Al4C3, and C were
mixed to yield a final stoichiometry of Ti3Al1.1C1.8. The green bod-
ies were sealed under vacuum and hot isostatically pressed ͑HIP͒ at
1400°C for 16 h under a pressure of Ϸ70 MPa. The final grain size
was of the order of 30 ␮m. The samples were fully dense, and
predominantly single phase, with Ϸ4 vol % of Al2O3 as an impurity
phase. Further details can be found in Ref. 15. Recent microprobe
analysis of this composition indicated that the final chemistry is
Ti3AlC2 .
The processing details for the Ti4AlN2.9 samples are described
elsewhere.16 In brief powder compacts of TiH2, TiN and AlN were
mixed, dehydrided, and reactively HIP at 1275°C for 24 h under a
pressure of 70 MPa. This was followed by a further anneal in Ar at
1360°C for 168 h. Phase analysis showed TiN and Al2O3 were
present in minor amounts.16 The measured density was Ϸ0.95 of
theoretical ͑4.77 g/cm3͒ and the grains were in the 20 to 30 ␮m
range.6
and /or CO2 is made here in order to balance the reactions.
14
The molar volumes of Ti2AlC,14 Ti2AlC0.5N0.5
,
TiO2, and
Al2O3 are 32.86, 32.32, 18.57, and 26.14 cm3/mol. Thus, the volume
changes for Reactions 2 and 3 are ϩ193 and ϩ196%, respectively.
It is thus not surprising that the oxide layers that form are, for the
most part, dense and protective.
The time, t, dependencies of the oxide layer thicknesses, x, that
form on Ti2AlC and Ti2AlC0.5N0.5 at various temperatures are plot-
ted in Fig. 2a and 3a, respectively. A comparison of the two figures
indicates that Ti2AlC is more susceptible to oxidation than
Ti2AlC0.5N0.5. The exponent, n, of the best power law fit to time,
i.e., tn, is shown on the graphs.b From these values of n, one can
conclude that at 900°C, the oxidation kinetics are neither parabolic
nor linear, but that at the higher temperatures, the kinetics are para-
bolic for both compounds. This was verified by plotting the thick-
ness squared vs. t ͑Fig. 2b and 3b͒. The parabolic rate constants, kx ,
were fit to the expression
Prior to the oxidation runs, the samples were rough polished
using 600 grit SiC grinding paper, placed on a Pt mesh, heated to the
oxidation temperature at 10°C/min for a predetermined time, and
furnace cooled. The oxide layer thicknesses were measured in a
scanning electron microscope ͑SEM͒ ͑Amray 1830͒, equipped with
an energy dispersive spectroscopy ͑EDS͒ system for chemical analy-
sis ͑LINK ISIS, Oxford Instruments, England͒. The phases formed
upon oxidation were determined from X-ray diffraction ͑XRD͒
spectra of crushed surface layers.
b The exponents were determined from log-log plots of x vs. t. Once the exponents
were determined, they were used to draw the lines shown in Fig. 2a and 3a.
Results and Discussion
Ti2AlC and Ti2AlC0.5N0.5.—The XRD spectra of the layers that
form as a result of the oxidation of Ti2AlC and Ti2AlC0.5N0.5 in air
at 1100°C, for 4 h are shown in Fig. 1. The major peaks correspond
to TiO2 and Al2O3, with the intensity of the former being greater. No
evidence for titanium aluminate was found even after oxidation at
1100°C for long times. Similar XRD patterns were found at the
other temperatures as well. Based on these results, the most likely
overall oxidation reaction for Ti2AlC is
Ti2AlC ϩ 3.75 O2 → 2TiO2 ϩ 0.5Al2O3 ϩ CO2
Similarly, for the solid solution
͓2͔
Ti2AlC0.5N0.5 ϩ 3.75O → 2TiO ϩ 0.5Al O ϩ 0.5 CO ϩ NO ͒
Figure 1. XRD spectra of reaction layers that form on oxidation of Ti2AlC
and Ti2AlC0.5N0.5 in air at 1100°C for 64 h. ͑᭹͒ rutile peaks, ͑ᮀ͒ Al2O3.
͑
2
2
2
3
2
2
͓3͔
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
C553
Figure 2. Time dependence of oxide thickness, x, formed on Ti2AlC
samples as a function of temperature and plotted as ͑a͒ x vs. t. Solid lines are
plots of x ϭ Ktn, where n was determined from least-squares fits of log x vs.
log t plots; ͑b͒ x2 vs. t. Typical error bars are shown.
Figure 3. Time dependence of oxide thickness, x, formed on Ti2AlC0.5N0.5
samples as a function of temperature and plotted as ͑a͒ x vs. t. Solid lines are
plots of x ϭ Ktn, where n was determined from least-squares fits of log x vs.
log t plots, ͑b͒ x2 vs. t.
x2 ϭ 2kxt
͓4͔
The results are listed in Table I in Part I.1 Arrhenius plots of kx for
Ti2AlC and Ti2AlC0.5N0.5 are shown in Fig. 4a. A least-squares fit of
the experimental results yields
5 ͑denoted by corner arrows͒. The original sharp corners of the
sample are clearly visible and decorated by Al2O3. Four layers can
be identified. An external layer that EDS indicates is TiO2, in which
some Al2O3 stringers are apparent. A thin darker layer that contains
some TiO2, but has a high volume fraction of Al2O3 follows. The
next layer is mostly TiO2, in which some Al3ϩ ions are dissolved,
kx m2/s͒ ϭ 2.55 ϫ 105 exp Ϫ 458,766/RT
͓5͔
͑
for Ti2AlC, and
kx m2/s͒ ϭ 2.68 ϫ 106 exp Ϫ 491,515/RT
͓6͔
͑
for Ti2AlC0.5N0.5. The corresponding activation energies are 458.7
and 491.5 kJ/mol. Also included in Fig. 4a are the kx values for the
oxidation of Ti7 ͑Eq. 1͒, Ti2Al,8,9 and TiAl.9,11 Not surprisingly, the
kx values of the ternaries are comparable to those of the intermetal-
lics.
Ͻ
i.e., (Ti1ϪyAly͒O2Ϫy/2 with y
0.05. An innermost layer, which is
slightly darker and has the appearance of a stain, is once again a
two-phase mixture of Al2O3 and (Ti1ϪyAly͒O2Ϫy/2 , but at a finer
scale that the previous Al2O3-rich layer. A comparison of the order
of the layers and their morphologies indicate that this system is in
stage II, described in Part I. The layers correspond, respectively, to
layers A through D shown in Fig. 1b in Part I. For the sake of
A typical backscattered SEM micrograph of the cross section of
a Ti2AlC0.5N0.5 sample oxidized at 1040°C for 16 h is shown in Fig.
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C554
Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
Figure 5. Secondary SEM micrographs of Ti2AlC0.5N0.5 sample oxidized in
air at 1040°C for 16 h. The arrows mark the original sample dimensions.
Note that original sharp corners of sample are still visible.
3. The outermost A layers increase in thickness from ϳ60 to
ϳ180 ␮m with time. This fact alone unambiguously and conclu-
sively proves the Ti ions are diffusing outward. Furthermore, the Al
signal in these layers increases substantially with time. Again, this is
a result that could not have occurred without the dissolution and
reprecipitation of Al3ϩ ions within the A layers.
4. Similarly, the layers labeled C and D increase in thickness
from ϳ100 to ϳ400 ␮m, proving that oxygen diffusion inward is
occurring. It is important to point out that layer C does not contain
Al2O3 precipitates. EDS indicates that the Al concentration is low
͑mole fraction Ϸ0.01͒ and more or less constant across this layer.
This is compelling evidence for both the outward diffusion of the
Al3ϩ ions, and their precipitation as Al2O3 particles at the high oxy-
gen partial pressure side, confirming the notion of demixing dis-
cussed in Part I. More compelling evidence for demixing is pre-
sented below.
5. The origin of the porosity observed in the micrographs is
believed to be the continual outward diffusion of the Al3ϩ ions as a
result of demixing. More evidence is discussed below.
Figure 4. Arrhenian plots of parabolic rate constants. ͑a͒ For Ti2AlC and
Ti2AlC0.5N0.5. Also included in the plot are results for commercial Ti,7
Ti2Al,9,10 TiAl.11 The kx values for the intermetallic compositions were for
the most part calculated from micrographs in the original references. ͑b͒
Ti4AlN2.9 and Ti3AlC2. The data at 1100°C were excluded from the least-
Figure 8a is a SEM micrograph of a carbide sample oxidized at
900°C for 96 h. The corresponding Ti and Al maps are shown in Fig.
8b and c, respectively. Qualitatively, there is little difference in the
sequence of the layers formed at 900°C and those formed at 1100°C
͑Fig. 6 and 7͒. The outermost layer at 900°C, however, has a much
weaker Al signal than the ones formed at 1100°C ͑compare Fig. 6c
or 7c with 8c͒. Furthermore, at 900°C, there are no Al2O3 stringers.
Both observations can be taken as indirect evidence that the solubil-
ity and/or diffusivity of Al3ϩ ions in TiO2 is significantly lower at
900°C than at 1100°C. As discussed below, Fig. 5-8 can be used to
estimate the diffusion coefficients of Al in TiO2.
squares fit represented by the bold solid lines. Also included for comparison
9,13
are the kx values obtained from the literature for Ti3Al
and the
211’s shown in a.
consistency and to avoid confusion the labeling of these layers is
identical to the one shown in Fig. 1 in Part I.1
Figures 6a and 7a are backscattered SEM micrographs of two
Ti2AlC0.5N0.5 samples oxidized at 1100°C for 2 and 16 h, respec-
tively. The respective Ti maps are shown in Fig. 6b and 7b; the Al
maps in Fig. 6c and 7c. Here again, four layers, A-D, are clearly
distinguishable. Comparing the two sets of micrographs, the follow-
ing salient points can be made.
The SEM micrographs for the oxidized Ti2AlC samples in the
900-1100°C temperature range are similar to those shown here for
Ti2AlC0.5N0.5 and are not shown for the sake of brevity.
In comparison of the microstructures shown in Fig. 5 to 8, to the
schematic diagram shown in Fig. 1 in Part I,1 it is clear that in all
cases the system is in stage II. In these ternaries, however, the layer
of porosity at shorter times either does not form, or, more likely, has
sintered shut. The fact that the oxidation kinetics are still parabolic
after 80 h ͑Fig. 3͒ at 1040°C indicates that the layer does remain
protective. However, at longer times and at the highest temperatures,
pores do form.
1. Qualitatively, and but for the obvious differences in magnifi-
cations, the two sets of micrographs are quite similar.
2. The width of the Al-rich band, layer B, more apparent in Fig.
7c than in 6c, increases from ϳ20 ␮m to ϳ60 ␮m with increasing
time. The same is true of the Al signal; it is stronger at longer times.
This could not have occurred without the outward diffusion of Al3ϩ
ions through the TiO2-rich layer, viz., the C layer.
Ti4AlN3.—XRD of the reaction scales ͑not shown͒ that form
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
C555
Figure 7. Cross-sectional SEM micrographs of a Ti2AlC0.5N0.5 sample oxi-
dized in air at 1100°C for 16 h. ͑a͒ Backscattered mode, ͑b͒ Ti map, ͑c͒ Al
map.
Here again, the N is assumed to diffuse through the reaction layers
16
15
and oxidize. The molar volumes of Ti4AlN3 and Ti3AlC2 are,
respectively, 61.31 and 45.8 cm3/mol. Hence, the volume changes
upon oxidation are of the order 150% which, theoretically, should
result in a dense and protective scale. The reason that is not the case
is discussed below.
Figure 6. Cross-sectional SEM micrographs of a Ti2AlC0.5N0.5 sample oxi-
dized in air at 1100°C for 2 h. ͑a͒ Backscattered mode, ͑b͒ Ti map, ͑c͒ Al
map.
The time dependencies of x that form on Ti4AlN2.9 at various
temperatures are plotted in Fig. 9. Initially, and up to about ϳ16 h at
800, 900, and 1000°C, and up to ϳ9 h at 1100°C, the kinetics are
reasonably well described by parabolic kinetics. This is shown in
Fig. 9a and b by the dotted lines, which were drawn by first least-
upon the oxidation of Ti4AlN3 are similar to those shown in Fig. 1.
Based on these results, the most likely overall reaction for the oxi-
dation of Ti4AlN3 is
2Ti4AlN3 ϩ 15.5 O2 → 8TiO2 ϩ Al2O3 ϩ 6NO2
͓7͔
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C556
Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
Figure 9. Time dependence of layer thickness, x, formed on Ti4AlN2.9
samples oxidized at ͑a͒ 800 and 900°C, and ͑b͒ 1000 and 1100°C. The
dashed lines represent Eq. 4 in text, plotted using kx values obtained from
least-squares fits of x2 vs. t curves ͑not shown͒ at short times ͑Ͻ20 h͒.
tal scatter, the agreement between the two sets of results is reason-
able, especially at shorter times. At longer times, the kinetics are
clearly no longer parabolic, but are more or less linear. It is worth
noting that it is possible to obtain a better fit of the results shown in
Fig. 9b at longer times, but at the expense of the fits at shorter time.
However, since the microstructural observations are more in tune
with a system in which the kinetics are parabolic at short times, the
analysis described above was adopted. The correction is small, how-
ever, and does not affect any of the conclusions reached herein.
A least squares fit of Arrhenius plots of the kx results yields the
following relationships for Ti4AlN2.9
Figure 8. Cross-sectional SEM micrographs of a Ti2AlC0.5N0.5 sample oxi-
dized in air at 900°C for 96 h. ͑a͒ Backscattered mode, ͑b͒ Ti map, ͑c͒ Al
map.
429,607 kJ/mol͒
͑
kx m2/s͒ ϭ 3.2 ϫ 105 exp Ϫ
͓8͔
͑
RT
squares fitting x2 vs. t plots to determine kx at times less than 20 h.
Once the kx values were determined, the time dependencies of x
which is plotted in Fig. 4b. For reasons discussed below, namely, the
dissociation of Ti4AlN2.9, the datum point at 1100°C was excluded
from the least-squares fits.
were replotted on Fig. 9 as dashed lines. Considering the experimen-
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
A series of backscattered SEM micrographs all taken at the same
C557
magnification, of samples that were oxidized at 900°C for 0.5, 1,
and 2 h are shown in Fig. 10a-c, respectively. The following obser-
vations are noteworthy.
1. After 0.5 h ͑Fig. 10a͒, the oxidized layer is more or less com-
positionally uniform and is physically separated from the substrate
by a gap extending parallel to the substrate/oxide interface. The
origin of this gap must be the rapid outward diffusion of the cations.
In other words, this gap forms as a result of the coalescence of
vacancies resulting from the Kirkendall effect. This feature, in
agreement with Part I,1 is henceforth referred to as the P layer or
plane.
2. After 1 h ͑Fig. 10b͒, compositional differentiation ͑where the
outermost layer is TiO2-rich, followed by an Al2O3-rich layer fol-
lowed again by a TiO2-rich one followed by a gap͒ is apparent.
These layers correspond, respectively, to layers A, B, C, and P in
Part I.1 The P layer is now wider and has moved inward. The latter
observation alone is compelling evidence that sintering must be oc-
curring during the oxidation process. This is best seen by comparing
Fig. 10a and b. The gap that formed in Fig. 10a and is still partially
visible in Fig. 10b, but is being ‘‘filled’’ by Al2O3 particles must
occur by a reaction between the outward diffusing cations and the
inward diffusing O2Ϫ ions. The rapid outward diffusion of the cat-
ions leads to the formation of another gap roughly 8 ␮m below the
surface ͑Fig. 10b͒.
3. After 2 h ͑Fig. 10c͒, not only are the three layers much more
clearly defined, but they all have increased in thickness. The original
gap formed in Fig. 10a has almost disappeared, and the second gap
has gotten wider, and is no longer a clearly defined gap. Instead it is
an area of high porosity, which is reminiscent of a collection of
particles in the process of sintering. These particles most probably
form as a result of a reaction between the inward diffusing O2Ϫ ions
and the outward diffusing cations. However, since the Al diffusion is
much more rapid, it tends to concentrate in the B layer nearest the
surface. Note that the diffusion of Al in the outermost layer, A, is
significantly slower than its diffusion in the C layer. For example,
from Fig. 10c, the characteristic diffusion distance for the Al3ϩ ions
in the C layer is Ϸ10 ␮m, in, at most, 2 h. In contradistinction, there
is very little Al signal in the A layer in the same time span.
4. After 64 h at 900°C ͑Fig. 10d͒, four layers, A, B, and C
together with a P layer, are now clearly discernable. In contrast to
Fig. 10a-c, Fig. 10d was taken in the secondary mode in order to
emphasize the morphology of the various layers. Layers A, B, and C
to the right of the P plane are relatively dense; the C layer to the left
of the P plane, however, is quite porous. The Al signal, as deter-
mined by EDS scans ͑not shown͒ is strongest at B, and in and near
the P layer.
After 16 h of oxidation at 1000°C, the layers are now signifi-
cantly thicker, ͑Fig. 11a and b͒ but their order is similar to those at
900°C ͑Fig. 10d͒. Note that the P layer is now significantly deeper
within the oxide layer, and much closer to the substrate/oxide inter-
face. Relative to the outermost layer, the other layers are porous
͑Fig. 11a͒. At 1000°C, there is no layer that corresponds to the
porous C layer to the left of the P plane shown in Fig. 10d. Oxidiz-
ing at 1000°C for 64 h results in the morphology shown in Fig. 11c,
where two P planes are obvious. The outer P plane is now a clear
demarcation line between layer C, and a new layer, which is a two-
phase intimate mixture of Al2O3 and TiO2. This layer was labeled D
in Part I. An important characteristic of the transition is that, while
the Ti4ϩ ions and O2Ϫ continue to diffuse across the P layers, for
some reason the Al3ϩ ions do not. The evidence for this is obvious
by comparing Fig. 11a and c: increasing the oxidation time from 16
to 64 h increases the thickness of the A layer from Ϸ100 ␮m to 216
␮m and creates the D layer, but does not change the thickness of the
B layer that remains at Ϸ50 ␮m. The reason for the transition from
stage II to stage III, i.e. Fig. 11b-c, is not clear at this time.
Here, it is useful to think of the B layers as the repository of all
the Al ions that are swept out of the C layers. In all micrographs
Figure 10. Backscattered SEM image of the oxide layers that formed after
oxidation of Ti4AlN2.9 at 900°C for ͑a͒ 30 min, ͑b͒ 1 h, ͑c͒ 2 h. Gap present
in a and b is being filled, i.e., sintering is occurring in c. ͑d͒ Secondary SEM
image of sample oxidized at 900°C for 64 h.
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C558
Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
Figure 11. Cross-sectional SEM micrographs of a Ti4AlN2.9 sample oxi-
dized in air at ͑a͒ 1000°C for 16 h ͑secondary mode͒ showing porosity, ͑b͒
backscattered mode showing layering, ͑c͒ 1000°C for 64 h ͑backscattered
mode͒ showing the formation of more layers.
Figure 12. Cross-sectional SEM micrographs of a Ti4AlN2.9 sample oxi-
dized in air at 1100°C for 64 h. ͑a͒ Backscattered mode, ͑b͒ Ti map, ͑c͒ Al
map. Dark stained area near substrate/oxide interface is dissociated Ti4AlN2.9
͑see also Fig. 17͒.
shown herein, the B layers are always associated with C layers of
comparable thicknesses adjacent to them on the lower pO side ͑i.e.,
closer to the substrate side͒.
paring the Al signal in the outermost layers in Fig. 11c and 12c, it is
obvious that higher temperatures induce the Al2O3 to migrate out-
ward, i.e., toward the air/oxide interface.
In addition, adjacent to the substrate, a stain-like layer is appar-
ent. This layer is a two-phase mixture of TiNx and Al, the origin of
which is the dissociation of Ti4AlN2.9 according to16
2
After 0.5, 1, and 2 h at 1000°C ͑not shown͒, the oxide scales,
while thicker, are morphologically quite similar to those at 900°C.
The gap however, is less pronounced at 1000°C. After 2 h the C
layer is Ϸ50 ␮m thick, quite porous, and similar in morphology to
the C layer shown in Fig. 10c.
Oxidation of Ti4AlN2.9 for 64 h at 1100°C, results in a striated
microstructure ͑Fig. 12͒. The order of the layers, from the air/oxide
interface inward is A ͑with Al2O3 stringers͒/C/B/C/P/C/P/C/D. Com-
Ti4AlN2.9 → TiN0.72 ϩ Al
͓9͔
The oxidation then presumably proceeds by the oxidation of the Al
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
and TiNx separately. The Mnϩ1AXn phases do not melt, but peritec-
C559
tically decompose into the MXx compound and the A group
element.15-19 The decomposition temperature is a function of
impurities.19 In Ar, the dissociation temperature of Ti4AlN2.9 is in
excess of 1400°C.16 Based on the results shown here, oxygen appar-
ently reduces that decomposition temperature to somewhere be-
tween 1000 and 1100°C. The same is also true for Ti3AlC2 ͑see
below͒. This decomposition, for reasons that are not understood, is
believed to reduce kx compared to what it would have been had
there been no decomposition ͑Fig. 4b, top left͒. It is for this reason
that the 1100°C datum point is not included in calculating the kx
values for Ti4AlN2.9
.
Ti3AlC2.—Based on XRD of the reaction scales ͑not shown͒ the
most likely overall oxidation reactions for Ti3AlC2 is
2Ti3AlC2 ϩ 11.5 O2 → 6TiO2 ϩ Al2O3 ϩ 4CO2
͓10͔
The x vs. t curves for Ti3AlC2, subject to an analysis identical to
the one used to generate Fig. 9, are shown in Fig. 13. Here again, the
kinetics appear to change from parabolic early on, to linear at longer
times. The anomalous point at 64 h that falls on the dotted line in
Fig. 2b is noteworthy; its significance is discussed shortly.
A least-squares fit of the Arrhenius plots of kx for Ti3AlC2, yields
443,186 kJ/mol͒
͑
kx m2/s͒ ϭ 1.15 ϫ 105 exp Ϫ
͓11͔
͑
RT
The actual values are listed in Table I in Part I,1 and plotted in Fig.
4b. Here again because of dissociation, the datum point at 1100°C
was excluded from the least-squares fit. From these results it is
apparent that Ti4AlN2.9, is less oxidation resistant than Ti3AlC2,
which in turn, is not as good as the 211 phases. Not surprisingly, and
in total agreement with the Ti-aluminide literature, increasing the Al
content in the substrate enhances the oxidation resistance, presum-
ably because the Al2O3 content in the scales increases. Also not
surprisingly, the activation energies for Ti4AlN2.9 and Ti3AlC2 are
comparable to those of the 211 phases.
Included in Fig. 4b for comparison purposes are the kx values for
Ti3Al available in the literature.9,13 Despite the sporadic nature of
the latter, it is clear that the kx values for Ti4AlN2.9 are comparable
to those of Ti3Al. In contradistinction, the oxidation resistance of
Ti3AlC2 in significantly better than its binary counterpart ͑Fig. 4b͒.
The reason for this state of affairs is speculated on below.
In the 800-1000°C temperature range, the oxide scale thicknesses
that form on Ti3AlC2 are thinner than, but morphologically similar
to, the ones that form on Ti4AlN2.9. At 800°C the oxide scale ͑not
shown͒ is compact and dense, and quite protective. Even after 64 h
at 800°C, the oxide scale is less than 10 ␮m thick. In Fig. 13, the
thickness of the scale that formed on the sample oxidized for 64 h at
1000°C was thinner than the one formed after the significantly
shorter oxidation of 25 h at the same temperature. It is more impor-
tant, however, that this datum point appears to fall on the extension
of the parabolic kinetics line fit to the short time results ͑Fig. 13b͒.
A comparison of the two microstructures, shown in Fig. 14a and b,
respectively, shed some light on the anomalous behavior of the
sample oxidized for 64 h. The simplest explanation is that, for rea-
sons that are not clear, the oxide layers shown in Fig. 14b remained
dense and crack free, while the outermost layers that formed after 25
h ͑Fig. 14a͒ must have microcracked at some time during the run.
Such cracks would introduce oxygen via a parallel path ͑i.e., not
through bulk diffusion͒ and enhance the oxidation kinetics. In other
words, it is postulated that as the layers get thicker, stresses, prob-
ably sintering ones, are sufficient to micro- or even macrofracture
the layers allowing for a parallel path for oxygen to flow into the
interior. Furthermore, and based on this and other evidence pre-
sented herein, it appears that the deviation from parabolic kinetics
coincides with the onset of demixing; if the latter can be suppressed,
Figure 13. Time dependence of layer thickness, x, formed on Ti3AlC2
samples oxidized at ͑a͒ 800 and 900°C, and ͑b͒ 1000 and 1100°C. The
dashed lines represent Eq. 4 in text, plotted using kx values obtained from
least-squares fits of x2 vs. t curves ͑not shown͒ at short times ͑Ͻ20 h͒.
the oxidation resistance of these ternaries and possibly the Ti-
aluminide intermetallics could be greatly enhanced.
A dramatic example of the type of microstructure that the repeti-
tion of such a process yields is shown in Fig. 15. These micrographs
are of a sample heated to 1100°C for 64 h, which resulted in a highly
striated microstructure in which Al2O3-rich, TiO2-rich, and P layers
repeat almost a dozen times. This is an important micrograph be-
cause it unequivocally proves that ͑i͒ demixing is occurring, and is
occurring repeatedly, and ͑ii͒ demixing is occurring by having the
Al3ϩ ions diffuse toward the high pO side ͑in every layer, the Al2O3
2
layer is adjacent to the high pO side͒. The latter is also seen clearly
2
in most of micrographs shown in this work; in most cases, the vol-
ume fraction of the Al2O3 islands or streamers in the outermost A
layers increase with time ͑e.g., Fig. 12c͒. To understand the atomis-
tic mechanisms required for the striations to form refer to Fig. 16,
which is an SEM micrograph of a sample that was oxidized at
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C560
Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
Figure 15. Cross-sectional SEM micrographs of a Ti3AlC2 sample oxidized
in air at 1100°C for 64 h. ͑a͒ Secondary mode showing topography, and ͑b͒
backscattered mode. Three layers, a bright (Ti1ϪyAly͒O2Ϫy/2 layer, a darker
Al2O3-rich layer, and planes of porosities or P planes, repeat over 10 times.
to assume that the outermost layers, with their many fissures and
cracks, are less protective than if they had remained intact.
2. The overall volume fraction of the Al2O3 in each segment
contained between two P layers, is Ϸ20 vol % which is comparable
to the theoretical vol fraction ͑Ϸ20 vol %͒ expected from Reaction
10. The volume fraction in the Al2O3-rich segment of each striation
Figure 14. Cross-sectional backscattered SEM micrographs of a Ti3AlC2
sample oxidized in air at 1000°C for ͑a͒ 25 h, and ͑b͒ 64 h. The oxide layer
thickness for b is less than that for a, despite the significantly longer oxida-
tion time for the former ͑see text for details͒.
1100°C for 16 h. The top few layers are indistinguishable from the
ones described so far. However, in this sample, there is a fairly thick
͑Ϸ300 ␮m͒ D layer, comprised of a fine two-phase mixture of Al2O3
and TiO2. The Al2O3 particles are not visible at the magnification of
Fig. 16, but are discernable at higher magnifications ͑not shown͒. As
the oxidation proceeds, the pO within the P layers must increase to
2
some value at which demixing can occur ͑i.e., the condition given
by Eq. 2 in part I1 is reached͒, at which time the Al3ϩ ions diffuse
toward the high pO side and precipitate as Al2O3. To do so, the
2
Al2O3 particles in layer D must dissolve at the low pO side and
2
eventually reprecipitate at the high pO side. This process must also
2
form vacancies that diffuse inward and ultimately precipitate along a
plane, viz. the P planes. As noted above, these P planes prevent the
further migration of Al3ϩ outward, which is why the Al2O3 particles
are invariably clustered at the high pO side of each segment con-
2
tained between two P layers. The evidence for this hypothesis is
multifold, and includes
1. The change in oxidation kinetics at around 16 h ͑Fig. 13͒ from
parabolic to a faster oxidation rate. Based on Fig. 15, it is reasonable
Figure 16. Cross-sectional backscattered SEM micrographs of a Ti3AlC2
sample oxidized in air for 16 h at 1100°C.
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
C561
Figure 17. Cross-sectional backscattered SEM micrographs of a region near
the oxide/substrate interface of a Ti3AlC2 sample oxidized in air for 64 h at
1100°C. The topotactical nature of the decomposition that results in TiCx and
Al is obvious in some of the grains.
Figure 18. Arrhenius plots for the diffusion of Al in rutile determined from
this work. Also included are the results of Yan and Rhodes.22
is 50 vol%. In other words, after demixing, the 20 vol % Al2O3
within each lamella is redistributed and concentrates at the high pO
side.
Repeating the same exercise using some of the micrographs
shown for the 211 phases yields a second series of points that are
lower than the first ͑Fig. 18͒. At face value, these results are some-
what surprising, since the diffusivity is ostensibly occurring in the
same matrix, i.e., Al2O3-saturated TiO2, in all cases. The most likely
explanation for the discrepancy is that in the case of Ti4AlN2.9 and
Ti3AlC2, the resulting matrix is not fully dense and other parallel
diffusion paths, in addition to bulk diffusion, such as grain bound-
ary, surface, are operative. This is an important observation, and is
the most probable explanation for the differences in kx between the
various ternaries. For reasons that are not clear, the scales that form
on the 211 phases must be denser than those that form on Ti3AlC2,
which in turn are denser than the ones that form on Ti4AlN2.9. This
conclusion is corroborated by much of the microstructural evidence
shown in this work. Further evidence for the fact that the aforemen-
tioned porosity affects the oxidation kinetics can be gleaned by com-
paring the oxidation behavior of Ti, the Ti-Al intermetallics, and the
ternary compounds. The oxygen diffusivities determined from tracer
experiments on fully dense rutile samples are about two orders of
magnitude lower than those determined from oxidation experiments,
be they on Ti, the ternaries studied here, or Ti-aluminide intermetal-
lics ͑see Fig. 3, Part I1͒.
2
3. Such a hypothesis is the simplest way to explain how the
microstructure shown in Fig. 16 is transformed into the one shown
in Fig. 15. It is difficult to conceive of another scenario whereby a
homogeneously distributed two-phase mixture ͑layer D in Fig. 16͒ is
transformed to the heterogeneous striated structure ͑Fig. 15͒.
The same decomposition reaction observed at 1100°C in
Ti4AlN2.9 also occurs in Ti3AlC2 ͑the decomposition temperature in
Ar is Ͼ1400°C15͒. Once again, this somewhat complicates the se-
quence of events leading to the formation of the oxide layers, but in
this case does not seem to greatly affect the kx values. Nevertheless,
as noted above, the least squares fit of kx ͑Eq. 11͒ exclude the datum
point at 1100°C. It is important to note, however, that the end phases
are identical, with or without dissociation. Furthermore, the decom-
position occurs topotactically by having the Al diffuse out of the
basal planes. This is shown in Fig. 17 in which some grains are
clearly dissociated while maintaining their original shape. Given
that this is one of the preferred modes of reaction for Ti3SiC2,20,21
these results are not too surprising.
A collateral benefit of this work, and important independent evi-
dence that Al is indeed diffusing in TiO2, is the possibility of esti-
mating the diffusivity of Al, DAl, in TiO2, by measuring the width of
the C layers, i.e., (Ti1ϪyAly)O2Ϫy/2 , after a given oxidation time.
Since the B layers form from the C layers, the width of the latter can
be used to crudely estimate DAl.c For example, in Fig. 11b, the width
of the C layer after 16 h at 1000°C is Ϸ120 ␮m. This translates to a
DAl of (ϳx2/2t) of ϳ1.25 ϫ 10Ϫ13 m2/s. An Arrhenius plot of the
DAl values calculated from the appropriate micrographs at other
temperatures and times, for both Ti3AlC2 and Ti4AlN2.9 is shown in
Fig. 18. Also included in Fig. 18, are the literature results for the
diffusivity of Al in rutile samples.22 The agreement between the two
sets of results is excellent, especially considering the crude method
used to estimate the DAl used here. And while this does not prove
that Al is diffusing through TiO2, it certainly is consistent with that
notion.
Some final comments are in order. The Mnϩ1AlXn phases have
several exceptional and mechanical properties, especially at higher
temperatures.14-16 In some cases, they rival the commercially avail-
able Ni-based superalloys, especially considering that the densities
of the ternaries are roughly half the latter. The results presented in
this series of papers, however, make it clear that, before their poten-
tial can be fulfilled, their oxidation resistance has to be enhanced,
Ͼ
especially at T
1000°C. In this respect, they are not unlike the
Ti-aluminide intermetallics.8 The key here, as it is for the interme-
tallics, is to so tailor the chemistry and, possibly microstructure,
such that an Al2O3 protective layer forms on the surface. In contrast
to the Ti-aluminides, however, one advantage that the Mnϩ1AlXn do
possess is that, at least up to 100 h, there is little evidence that they
dissolve any measurable concentrations of oxygen, and conse-
quently, are immune from the embrittlement such dissolution engen-
ders in the Ti-aluminide intermetallics.
c Not all micrographs can be used to estimate DAl. Two criteria used were that there
be only one C layer, and that the oxidation had proceeded for a while. A good example
is Fig. 11a, in which a wide C layer that formed after 16 h is clearly visible. This layer
must have formed in less than 16 h, and thus, the values reported here are slightly lower
than actual. It is important to note that even if the time was, for example, halved to 8 h,
that would alter D by a factor of 2, which considering the scale of Fig. 18, and the crude
approximation made to calculate DAL is almost within the noise level.
Conclusions
The oxidation of reaction HIP polycrystalline samples of Ti2AlC,
Ti2AlC0.5N0.5, Ti4AlN2.9, and Ti3AlC2 in air, in the 800-1100°C
range occurs by the inward diffusion of oxygen and the outward
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C562
Journal of The Electrochemical Society, 148 ͑8͒ C551-C562 ͑2001͒
diffusion of Al and Ti ions through a rutile-based solid solution, viz.,
References
Ͻ
(Ti1ϪyAly)O2Ϫy/2 in which y
0.05. The fate of the C and/or N is
1. M. W. Barsoum, J. Electrochem. Soc., 148, C522 ͑2001͒.
2. M. W. Barsoum, T. El-Raghy, and L. Ogbuji, J. Electrochem. Soc., 144, 2508
͑1997͒.
3. P. Kofstad, Nonstoichiometry, Diffusion and Electrical Conductivity in Binary
Metal Oxides, p. 142, R. E. Krieger Publications Company, Malabar, FL ͑1983͒.
4. P. Kofstad, High Temperature Oxidation of Metals, Wiley Interscience, New York
͑1966͒.
5. K. Hauffe, Oxidation of Metals, Plenum Press, New York ͑1965͒.
6. G. Bertrand, K. Jarraya, and J. M. Chaix, Oxid. Met., 21, ͑1983͒.
7. J. Unnam, R. N. Shenoy, and R. K. Clark, Oxid. Met., 26, 231 ͑1986͒.
8. M. P. Brady, B. A. Pint, P. F. Tortorelli, I. G. Wright, and R. J. Hanrahan, Jr., in
Corrosion and Environmental Degradation, Vol. II, R. Cahn, P. Haasen, and J.
Kramer, Editors, Wiley-VCH, Wienheim, Germany ͑2000͒.
9. G. Welsch and A. I. Kahveci, in Oxidation of High Temperature Intermetallics, T.
Grobstein and J. Doychak, Editors, Minerals and Materials Society, Warrendale, PA
͑1989͒.
10. G. H. Meier and D. Appalonia, in Oxidation of High Temperature Intermetallics, T.
Grobstein and J. Doychak, Editors, p. 185, Minerals and Materials Society, War-
rendale, PA ͑1989͒.
11. Y. Shida and H. Anada, Mater. Trans., JIM, 34, 236 ͑1993͒.
12. T. K. Roy, R. Balasubramaniam, and A. Gosh, Metall. Mater. Trans. A, 27, 3993
͑1986͒.
13. J. S. Wu, L. T. Zhang, F. Wang, K. Jiang, and G. Qiu, Intermetallics, 8, 19 ͑2000͒.
14. M. W. Barsoum, M. Ali, and T. El-Raghy, Metall. Mater. Trans. A, 31, 1857 ͑2000͒.
15. N. Tzenov and M. W. Barsoum, J. Am. Ceram. Soc., 83, 825 ͑2000͒.
16. A. T. Procopio, T. El-Raghy, and M. W. Barsoum, Metall. Mater. Trans. A, 31, 373
͑2000͒.
17. M. W. Barsoum, G. Yaroschuck, and S. Tyagi, Scr. Mater., 37, 1583 ͑1997͒.
18. L. Farber and M. W. Barsoum, J. Mater. Res., 14, 2560 ͑1999͒.
19. N. Tzenov, M. W. Barsoum, and T. El-Raghy, J. Eur. Ceram. Soc., 20, 801 ͑2000͒.
20. M. W. Barsoum, T. El-Raghy, L. Farber, M. Amer, R. Christini, and A. Adams, J.
Electrochem. Soc., 146, 3919 ͑1999͒.
unknown, but is presumed to diffuse outward through the layers and
oxidize. If the pO is relatively low, both oxides can coexist. How-
ever, as the oxidation proceeds and the pO increases, demixing oc-
2
2
curs with the Al2O3 migrating to the high pO side. This process
occurs by the dissolution and reprecipitation2 of Al2O3 particles
through the (Ti1ϪyAly)O2Ϫy/2 matrix and results in the formation of
layers of high porosity. The final microstructure is characterized by
the presence of multiple striated layers in which layers of TiO2,
almost totally denuded of Al2O3, alternate with Al2O3-rich layers
and layers of porosity. The onset of demixing coincides with the
deviation of the oxidation kinetics from parabolic to linear. The
microstructural observations are in agreement with the model pre-
sented in Part I,1 in which it was concluded that the diffusion of O
and/or Ti through the (Ti1ϪyAly)O2Ϫy/2 layer is rate limiting.
Both Ti4AlN2.9 and Ti3AlC2 decompose in air at 1100°C into Al
and the TiNx or TiCx just below the oxide layers that form. This
decomposition, which occurs topotactically, does not affect the final
morphology or the sequence of the layers that form, but slightly
affects the kx values, especially for Ti4AlN2.9
.
Acknowledgment
This work was partially supported by the U.S. Army Research
Office ͑DAAD19-00-1-0435͒ and partially by the National Science
Foundation ͑DMR-0072067͒.
21. T. El-Raghy and M. W. Barsoum, J. Appl. Phys., 83, 112 ͑1998͒.
22. M. F. Yan and W. W. Rhodes, J. Appl. Phys., 53, 8809 ͑1982͒.
Drexel University assisted in meeting the publication costs of this article.
Downloaded on 2015-05-29 to IP 147.188.128.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).