Microstructural effect on oxidation kinetics of NbSi2 at 1023 K

Article abstract of DOI:10.1016/j.jallcom.2005.12.015

Poly- and single crystalline NbSi₂ specimens with different microstructures were prepared by arc-melting, spark plasma sintering (SPS) and optical-heating floating zone melting for oxidation experiments at 1023 K. The effects of cracks, pores and grain boundary in the microstructure on the oxidation behavior of NbSi₂ were investigated. For arc-melted poly-crystalline specimens containing micro-cracks, NbSi₂ fully turned into powders after 3 h exposure at 1023 K, which is known as the pesting phenomenon. As a comparison, no pesting was found in the dense SPS poly-crystalline specimens and single crystals after 89 h. The oxide scale consists of Nb₂O₅ and SiO₂. The oxidation kinetics of all specimens follows a linear law. The oxidation rate was higher in the poly-crystalline specimen in comparison to single crystalline. The mechanism of oxidation has been analysed using a kinetic model.

Full text of DOI:10.1016/j.jallcom.2005.12.015

Journal of Alloys and Compounds 422 (2006) 308–312
Microstructural effect on oxidation kinetics of NbSi2 at 1023 K
F. Zhang, L.T. Zhang∗, A.D. Shan, J.S. Wu
Key Laboratory of the Ministry of Education for High Temperature Materials and Testing, School of Materials Science and Engineering,
Shanghai Jiao Tong University, Shanghai 200030, PR China
Received 4 November 2005; accepted 7 December 2005
Available online 24 January 2006
Abstract
Poly- and single crystalline NbSi2 specimens with different microstructures were prepared by arc-melting, spark plasma sintering (SPS) and
optical-heating floating zone melting for oxidation experiments at 1023 K. The effects of cracks, pores and grain boundary in the microstructure
on the oxidation behavior of NbSi2 were investigated. For arc-melted poly-crystalline specimens containing micro-cracks, NbSi2 fully turned into
powders after 3 h exposure at 1023 K, which is known as the “pesting” phenomenon. As a comparison, no pesting was found in the dense SPS
poly-crystalline specimens and single crystals after 89 h. The oxide scale consists of Nb2O5 and SiO2. The oxidation kinetics of all specimens
follows a linear law. The oxidation rate was higher in the poly-crystalline specimen in comparison to single crystalline. The mechanism of oxidation
has been analysed using a kinetic model.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Intermetallics; Crystal growth; Grain boundary; Oxidation
1. Introduction
from melt [10–12]. It is not very likely for brittle materials such
as MoSi2 and NbSi2 due to the fact that the thermal stress may
well be released by forming a number of micro-cracks in the
matrix. In an arc-melted ingot, many cracks are usually seen in
the microstructure of these compounds. Later works [13] noticed
the effects from structural defects and imperfections such as
pores and cracks in the bulk material on the catastrophic frag-
mentation. Berztiss et al. [14] studied the oxidation behavior of
MoSi2 prepared by different processes and found that the oxi-
dation behaviors were a function of microstructure. Only those
materials containing micro-cracks in the microstructure expe-
rienced fragmentation during oxidation at 773 K. Kurokawa et
al. [15] compared low temperature oxidation of fully dense and
porous MoSi2 and found accelerated oxidation and pesting in
porous material only. It is proposed that volume expansion of
oxide formed at cracks results in a wedging effect [16]. In con-
trast, Chou and Nieh [17] found that both as-cast and dense
poly-crystalline MoSi2 underwent severe pesting and MoSi2
single crystal also disintegrated into small pieces when oxi-
dized at 773 K. Whether there is similarity in pesting behavior
between NbSi2 and MoSi2 forms a motivation of the present
work.
Refractory transition-metal disilicides such as MoSi2 and
NbSi2 are promising materials for high temperature structural
applicationsbecauseoftheirlowdensity, highmeltingpointsand
high strength [1–4]. However, the application of Nb–Si alloys is
limited because of their poor oxidation resistance at high temper-
atures. Especially, pesting oxidation of NbSi2 occurs at 1023 K
after a certain period of exposure in air. Pesting oxidation in
NbSi2 was first reported by Rausch [5] and was further con-
firmed by Pitman and Tsakiropoulos [6] and Murakami et al.
[7]. Al, B and Cr additions are found to improve the oxida-
tion resistance of NbSi2 at 1523 K [8,9]. However, the Al-doped
NbSi2 still displayed pesting at 1023 K. Since most of the atten-
tion for pesting phenomenon has been paid to MoSi2, the study
of NbSi2 is relatively scarce.
According to many observations in MoSi2, it is found that
defects such as cracks, pores and grain boundary in microstruc-
ture are related to pesting. Although early works attributed the
pesting to high residual stresses accumulated during cooling
∗
Corresponding author at: School of Materials Science and Engineering,
In this paper, we first compare different pesting behaviors
of as-cast, spark plasma sintered poly-crystalline NbSi2 and
NbSi2 single crystal at 1023 K. Effects of cracks, pores and
Shanghai Jiao Tong University, 1954 Hua Shan Road, Shanghai 200030, PR
China. Tel.: +86 21 6293 2566; fax: +86 21 5254 0011.
E-mail address: lantingzh@sjtu.edu.cn (L.T. Zhang).
0925-8388/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2005.12.015

F. Zhang et al. / Journal of Alloys and Compounds 422 (2006) 308–312
309
Fig. 1. Scanning electron images of NbSi2 prepared by different processes: (a) arc-melted; (b) sintered by SPS at 30 MPa (back-scattered electron image); (c) sintered
at 60 MPa (hot corrosion image).
grain boundary on oxidation behavior have been investigated
focusing on the oxidation kinetics. A model has been proposed
to describe the observed oxidation kinetics.
3. Results
3.1. Microstructure
The arc-melted poly-crystalline NbSi2 contains quite a num-
ber of pre-existing microcracks and pores (Fig. 1(a)). This may
be caused by the residual thermal stress on cooling during arc-
melting. Microstructure analysis showed that the arc-melted
ingot was of monolithic NbSi2 phase. Laue diffraction con-
firmed transformation from poly-crystalline to single crystalline
NbSi2 after◦the zone melting, whose growth plane was close
NbSi2 prepared under different conditions are shown in Fig. 1(b)
and (c). Compared to the arc-melted ingot, a main feature of the
microstructure is that it contains a few pores but not cracks. A
relative density of more than 90% can be obtained by SPS. The
relative density increased from 93.8 to 95% when the pressure
during sintering increased from 30 to 60 MPa. Besides many
pores, some minor phases such as Nb5Si3 (bright particle in
Fig. 1(b) and SiO2 (dark phase in Fig. 1(b)) can be seen in the
NbSi2 matrix which is mainly due to the oxidation and volatiliza-
tion of silicon during sintering. Similar microstructure was also
reported for sintered MoSi2 [18–20]. Average sizes of the two
NbSi2 specimens were determined to be 2.5 ␮m (30 MPa) and
3.9 ␮m (60 MPa), respectively.
2. Experimental procedure
Poly-crystalline specimens were prepared by arc-melting and spark plasma
sintering (SPS). Stoichiometric amounts of high purity elements (Nb, Si) were
mixed and melted together in an arc-melting furnace on a water-cooled plate
under purified Ar atmosphere to prepare poly-crystalline ingots. The arc-melted
ingots were crushed into powder and ball-milled. The powder was sieved (below
30 ␮m in diameter) and loaded in a graphite crucible for sintering. The sintering
was performed at 1573 K under a load between 30 and 60 MPa. Single crystal
was grown by the optical-heating floating zone method on Asgal FZ-20035WHV
apparatus. The crystal was grown from bottom to top in a vertical direction
applying a growth rate in the range of 5–10 mm/h.
Oxidation behaviors of the specimens were characterized by measuring
their oxidation kinetics at 1023 K in room air. Specimens approximately
5 mm × 5 mm × 1 mm in dimension were cut from the as-cast ingot, sintered
bulk and as-grown single crystals (perpendicular to growth direction). The
specimens were polished to mirror finish and cleaned in acetone and alcohol
consequently prior to observation. Microstructures of the specimens prepared
by different methods were examined using scanning electron microscope (SEM)
and electron probe micro-analyser (EMPA). The orientation of single crystals
was determined by Laue back-reflection method. Average grain size of the sin-
tered poly-crystal NbSi2 was determined by observing the fractural surface or
by hot corrosion method. For those specimens fractured in a transgranular way,
the grains were clearly seen in the fractured surface. Other specimens were first
polished and oxided to reveal their grains. Density of single and poly-crystalline
NbSi2 was measured by the Archimedes method. After oxidation, the phase con-
stitutions of oxide products were characterized by X-ray diffraction (Cu K␣).
X-ray photoelectron spectroscopy (XPS) was applied to analyze the elemental
composition and chemical environments of Nb, Si and O elements in the oxide
scale.
¯
to (within 5 ) (1 1 2 0). Typical microstructures of the sintered
3.2. Oxidation behavior
The oxidation kinetics curves of different NbSi2 specimens
are plotted in Fig. 2, showing the mass change versus time at

310
F. Zhang et al. / Journal of Alloys and Compounds 422 (2006) 308–312
Fig. 2. Weight change as a function of exposure time for poly- and single crys-
talline NbSi2 at 1023 K.
Fig. 3. XRD profiles of the oxide formed on NbSi2 after exposure at 1023 K
for 89 h in air for: (a) poly-crystalline specimen; (b) single crystalline specimen.
The bars at the bottom of the plotting indicate peak positions of Nb2O5 from
JCPDS card no. 22-1196.
1023 K in air. Arc-melted specimen with many micro-cracks
exhibited relatively rapid weight gain and fully turned into pow-
der after 3 h exposure in air, which is known as pesting. However,
for the dense sintered poly-crystalline specimens and single
crystal, no fragmentation was found at a much longer expo-
sure time, indicating that the pre-existing cracks are crucial to
pest oxidation. Our previous in situ observation of NbSi2 clearly
shows that new cracks initiate from the pre-existing cracking
sites [21]. Since the oxide spall from the matrix during the oxi-
dation, the oxidation kinetics of NbSi2 almost follows a linear
law showing no indication of accelerated oxidation within the
time period of observation. As shown in Fig. 2, the NbSi2 poly-
crystalline specimens show a much larger weight change than
that of the single crystalline specimen, and the poly-crystalline
specimen sintered at 30 MPa shows a higher weight change than
one sintered at 60 MPa. This indicates that grain boundary and
pores may probably accelerate the rate of oxidation reaction and
will be modeled in the discussion part.
3.3. Phase constitution of the oxidation product
Fig. 3 shows XRD profiles of the oxide formed on poly-
and single crystalline NbSi2 after exposure at 1023 K for 89 h.
There is no qualitative difference in the profiles between the
poly-crystal (SPS) and single crystal materials, indicating that
phase constitutions of the oxides formed on the two materials
are almost the same. According to the XRD profiles, the oxide
is identified to be mainly Nb2O5 and a small amount of SiO2.
XPS further determined Nb 3d and Si 2p states in the oxide scale
(Fig. 4). According to Ref. [22], the peaks in Fig. 4(a) can be
assigned as Nb5+. No other states of the elements were identified.
Since only three elements Nb, Si and O were identified in the
XPS spectra, the Nb and Si are probably in the form of oxide, as
shown in the XRD profiles. Since the crystalline SiO2 peaks are
weak in the XRD profiles, most of the silica formed at 1023 K is
probably in the amorphous form. This result is consistent with
other alloys and compounds in the Nb–Si system [23].
Fig. 4. XPS spectra showing the (a) Nb 3d and (b) Si 2p peaks of the oxide scale
formed on NbSi2 after exposure at 1023 K for 89 h.

F. Zhang et al. / Journal of Alloys and Compounds 422 (2006) 308–312
311
Table 1
Microstructural and oxidation kinetics parameters of single and poly-crystalline NbSi2
Specimen
Relative density
Average grain
size (␮m)
Linear oxidation rate constant
Kl (mg/cm2 h) in Eq. (1)
Relative area
during oxidationa
th
(ρ/ρ
) (%)
NbSi
2
Single crystal
Poly-crystal (SPS@30 MPa)
Poly-crystal (SPS@60 MPa)
100
93.8
95.0
–
2.5
3.9
0.4
1.2
1.0
1
2.8
2.1
a
Calculated from Sꢀ/S according to Eq. (3).
4. Discussion
free surface is equal to its nominal free surface area (Fig. 5(a)).
However, for the poly-crystalline specimens containing grain
boundaries and pores which are the quick diffusion paths for
oxygen atoms, the effect from the grain boundary and pore can
be considered to a direct increase to be the effective free surface
of the material (Fig. 5(b)). Supposing oxygen atom (ion) pene-
trates at a certain depth h into the matrix (Fig. 5(b)), the effective
reaction area Sꢀ can be estimated as follows:
In general, the oxidation kinetics of a material can be
described by the following equation:
ꢀ
ꢁ
n
ꢀm
S
= Kt
(1)
where ꢀm is the mass change, S the surface area of a specimen,
t the exposure time, K the rate constant of oxidation, and n is
the exponent number (n = 1, linear law; n = 2, parabolic rate law;
n = 3, cubic law) [24]. Fitting the date in Fig. 2 to Eq. (1) shows
that the oxidation of NbSi2 at 1023 K follows a linear law with
the rate constant K being 0.4, 1.0 and 1.2 mg/cm2 h for single
crystal, dense and porous poly crystal, respectively. This indi-
cates that the oxide formed on NbSi2 is non-protective. XRD
and XPS indicate the oxidation product to be Nb2O5 and SiO2,
thus the oxidation of NbSi2 at 1023 K can be described as the
following reaction:
ꢀ
ꢁ
S
1
S 4h + L
Sꢀ =
πLh + S
=
(3)
(π/4)L2
µ
µ
L
where S is the apparent surface area; L the mean grain size; µ is
the relative density of the specimen. (S/((π/4)/L2)) estimates the
number of grains within the region of interest; and the length
of grain boundary is estimated as (S/((π/4)/L2))πL. Thus the
increased surface area due to grain boundary at a penetration
depth h can be estimated as:
S
πLh
(π/4)L2
4NbSi2 + 13O2 = 2Nb2O5 + 8SiO2
(2)
According to a previous investigation on the cross-sectional
microstructure of the oxide scale formed on NbSi2, the pene-
tration depth of oxygen into the matrix is on the order of 1 ␮m
at 1023 K [6]. The calculated ratios of (Sꢀ/S) in Eq. (3) are 2.8
and 2.1, respectively for the two sintered poly-crystalline spec-
imens (Table 1), which coincide well with the relative ratio of
the fitted oxidation rate constants of the three specimens. This
clearly shows that the oxidation rate is determined by the effec-
tive exposure area.
Due to the non-protective nature of Nb2O5, the oxidation rate
of NbSi2, unlike MoSi2, is controlled by the interface reaction.
The micro-cracks in the arc-melted NbSi2 increase the expo-
sure area dramatically. On the other hand, the high stress due to
volume expansion of oxidation formation at cracks’ interior face
leads to fragmentation, which in turn results in increased surface
The loose Nb2O5 spalls from the specimen and prevents
formation of a continuous scale of SiO2 (crystalline and amor-
phous) on the NbSi2 matrix during oxidation. Therefore, the
oxidation of NbSi2 is a simultaneous reaction of oxygen with
freshly exposed matrix and the reaction rate is intrinsically deter-
mined by the reaction at matrix/gas interface.
Table 1 summarizes the average grain size, density and linear
rateconstantofthreedenseNbSi2 specimens, whichclearlyindi-
cates that grain boundary and porosity contribute to an increased
oxidation rate. The non-protective nature of Nb2O5 determines
that there is no rate limiting process during the oxidation of
NbSi2 at 1023 K, whose reaction rate is merely controlled by
the effective free surface exposed to gas atmosphere. For a fully
dense single crystal without any grain boundary, the effective
Fig. 5. Schematic drawing showing the oxidation mechanism of (h is the diffusion depth from the surface to matrix): (a) single crystalline; (b) poly-crystalline MoSi2.

312
F. Zhang et al. / Journal of Alloys and Compounds 422 (2006) 308–312
area and pesting. Our present work illustrates that accelerated
oxidation can be hindered by dense NbSi2 and single crystal. It
was also observed that pesting is hard to happen to dense MoSi2
poly-crystalline specimen [15], this can be explained in a similar
way.
[3] Y. Murayama, S.J. Hanada, Sci. Technol. Adv. Mater.
145.
[4] D.M. Shah, D.L. Anton, S. Chin, D.P. Pope, Mater. Sci. Eng. A 192/193
3 (2002)
(1995) 658.
[5] J.J. Rausch, ARF-2981-4, Armour Research Foundation NSA, 1961, 15-
31171.
[6] S.H. Pitman, P. Tsakiropoulos, in: J. Horton (Ed.), High-Temperature
Ordered Intermetallic Alloys VI, J. Mater. Res. Symp. Proc., (364), Mate-
rial Research Society, Boston, 1995, p. 1321.
[7] T. Murakami, S. Sasaki, K. Ichikawa, A. Kitahara, Intermetallics 9 (2001)
621.
[8] T. Murakami, C.N. Xu, M. Kawahara, Y. Takahashi, H. Inui, M. Yamaguchi,
Intermetallics 7 (1999) 1043.
[9] T. Murakami, S. Sasaki, K. Ichikawa, A. Kitahara, Intermetallics 9 (2001)
629.
[10] J.H. Westbrook, D.L. Wood, J. Nucl. Mater. 12 (1964) 208.
[11] J. Schlichting, High Temp. High Press. 10 (1978) 241.
[12] E. Fitzer, W. Remmele, Proceedings of the Fifth International Conference
on Composites Materials, AIME, 1985, 515.
[13] C.G. Mckamey, P.F. Tortorelli, J.H. DeVan, C.A.J. Carmichael, Mater. Res.
7 (1992) 2747.
[14] D.A. Berztiss, R.R. Cerchiara, E.A. Cerchiara, F.S. Pettit, G.H. Meier,
Mater. Sci. Eng. A 155 (1992) 165.
[15] K. Kurokawa, H. Houzumi, I. Saeki, H. Takahashi, Mater. Sci. Eng. A 261
(1999) 292.
5. Conclusions
The oxidation behaviors of single and poly-crystalline NbSi2
were observed in air at 1023 K. The results show that:
(1) The pre-existing cracks in the microstructure are main rea-
son to pesting phenomenon. For arc-melted poly-crystalline
specimen, NbSi2 fully turned into powders after 3 h. No
pesting was found in the dense SPS poly-crystalline speci-
mens and single crystals after 89 h.
(2) The oxidation rate is intrinsically determined by the reaction
rate between the matrix and the oxygen in air. Nb2O5 scale
spalls from the specimen during oxidation. Thus the oxida-
tion kinetics of all the NbSi2 specimens at 1023 K follows
a linear law.
(3) Grain boundary and pores in the microstructure increase the
effective area of oxidation reaction. The oxidation rates of
the NbSi2 samples from high to low are loose poly-crystal,
dense poly-crystal and single crystal.
[16] T. Maruyama, K. Yanagihara, Mater. Sci. Eng. A 239–240 (1997)
828.
[17] T.C. Chou, T.G. Nieh, J. Mater. Res. 8 (1) (1993) 214.
[18] A.D. Shan, F. Wei, H. Hashimoto, Y-H. Park, Scripta Mater. 46 (2002)
645.
[19] M. Sannia, R.J. Orru, E. Garay, G.Z. Gao, A. Munir, Mater. Sci. Eng. A
345 (2003) 270.
Acknowledgements
[20] S. Hiroyuki, Y. Masaru, H. Ken, Y. Osamu, Mater. Res. Bull. 37 (2002)
1557.
[21] F. Zhang, L.T. Zhang, A.D. Shan, J.S. Wu, Scripta Mater. 53 (2005)
653.
[22] Z. Song, D.L. Tan, F. He, X.H. Bao, Appl. Surf. Sci. 137 (1999)
142.
This work was supported by the National Natural Science
Foundation of China under a contract #50571067. Provision of
high purity raw elements from Kyoto University is also appre-
ciated.
References
[23] J.W. Zhou, Master Dissertation, Shanghai Jiao Tong University,
2004.
[1] B.P. Bewlay, J.J. Lewandowski, M.R. Jackson, JOM 40 (8) (1997) 44.
[2] W.Y. Kim, H. Tanaka, S.J. Hanada, Intermetallics 10 (2002) 625.
[24] M.S. Li, High Temperature Corrosion of Metals, Metallurgical Industry
Press, Beijing, 2001, p. 36.