Influence of cobalt phase on thermal shock resistance of Al2O3-TiC composites evaluated by indentation technique

Article abstract of DOI:10.1016/j.materresbull.2007.11.001

Cobalt-coated Al₂O₃ and TiC powders were prepared using an electroless method to improve resistance to thermal shock. The mixture of cobalt-coated Al₂O₃ and TiC powders (about 70 wt.% Al₂O₃-Co + 30 wt.% TiC-Co) was hot-pressed into an Al₂O₃-TiC-Co composite. The thermal shock properties of the composite were evaluated by indentation technique and compared with the traditional Al₂O₃-TiC composite. The composites containing 3.96 vol.% cobalt exhibited better resistance to crack propagation, cyclic thermal shock and higher critical temperature difference (ΔT_c). The calculation of thermal shock resistance parameters (R parameters) shows that the incorporation of cobalt improves the resistance to thermal shock fracture and thermal shock damage. The thermal physic parameters are changed very little but the flexure strength and fracture toughness of the composites are improved greatly by introducing cobalt into Al₂O₃-TiC (AT) composites. The better thermal shock resistance of the composites should be attributed to the higher flexure strength and fracture toughness.

Full text of DOI:10.1016/j.materresbull.2007.11.001

Materials Research Bulletin 43 (2008) 2544–2553
www.elsevier.com/locate/matresbu
Influence of cobalt phase on thermal shock resistance of Al O –TiC
2
3
§
composites evaluated by indentation technique
a,b,*
b
a
a
Shi Rui-Xia
, Yin Yan-Sheng , Li Jia , Wang Dong-Zhi
a
School of Materials Science and Engineering of University of Jinan, Jinan, Shandong 250022, China
b
School of Materials Science and Engineering of Ocean University of China, Qingdao, Shandong 266100, China
Received 20 June 2006; received in revised form 31 October 2007; accepted 6 November 2007
Available online 13 November 2007
Abstract
Cobalt-coated Al
2
O
3
and TiC powders were prepared using an electroless method to improve resistance to thermal shock. The
mixture of cobalt-coated Al and TiC powders (about 70 wt.% Al –Co + 30 wt.% TiC–Co) was hot-pressed into an Al O –
2
O
3
2
O
3
2 3
TiC–Co composite. The thermal shock properties of the composite were evaluated by indentation technique and compared with the
traditional Al O –TiC composite. The composites containing 3.96 vol.% cobalt exhibited better resistance to crack propagation,
2
3
c
cyclic thermal shock and higher critical temperature difference (DT ). The calculation of thermal shock resistance parameters (R
parameters) shows that the incorporation of cobalt improves the resistance to thermal shock fracture and thermal shock damage. The
thermal physic parameters are changed very little but the flexure strength and fracture toughness of the composites are improved
greatly by introducing cobalt into Al
attributed to the higher flexure strength and fracture toughness.
2007 Elsevier Ltd. All rights reserved.
2 3
O –TiC (AT) composites. The better thermal shock resistance of the composites should be
#
Keywords: A. Composites; B. Ceramics; C. Mechanical properties; D. Microstructure
1
. Introduction
Ceramic materials possess excellent high temperature resistance. However, the brittleness, higher elastic modules,
bad plastic deformation ability and lower thermal conductivity of ceramic materials lead them to be sensitive to
environments in which the temperature fluctuates rapidly. Therefore, the full exertion of high temperature resistance of
ceramic materials depends on the resistance to thermal shock. Al O –TiC (AT) composites are a kind of important
2
3
engineering ceramics and have been widely used in various engineering fields [1–3]. In many applications the Al O –
2
3
TiC composites are often exposed to rapid temperature changes, which might cause severe thermal stress and thermal
shock damage due to their lower strength and bad thermal properties. Consequently, how to improve the thermal shock
resistance of AT composites has been an important issue in their structural applications.
It has been known that the incorporation of metal phase into alumina matrix can bring about improvement on
mechanical properties, including thermal shock performance [4–7]. Yet, in order to guarantee the higher hardness the
§
Sponsored by the Doctoral Fund of Shandong Province (Grant No. 2006BS04037), National Nature Science Fund of China (Grant No.
5
8
0572034), and the Primary Program of University of Jinan (Y0602).
Corresponding author at: School of Materials Science and Engineering of University of Jinan, Jinan, Shandong 250022, China. Tel.: +86 531
2765983; fax: +86 531 87974453.
*
E-mail address: mse_shirx@ujn.edu.cn (R.-X. Shi).
0025-5408/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2007.11.001

R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
2545
Table 1
2 3
Thermal physic parameters of Al O , TiC and cobalt
a (ꢂ10ꢃ6
K )
ꢃ1
k (W/m K)
E (GPa)
n
Materials
Cobalt
TiC
12.5
7.4
8.8
100
22
26
210
462
380
0.31
0.19
0.26
Al
2
O
3
content of metal phase is very low. The few metal phases cannot be homogeneously dispersed in matrix by using
traditional technology, which results in serious metal particles agglomeration. The powder coating technique has been
proved to be an effective method in processing ceramic–ceramic composites [8,9]. It was found to greatly improve the
homogeneity and sinterability of composites, and consequently their mechanical properties and thermal shock
resistance.
Comparing with Al O and TiC, cobalt possesses a unique set of thermal physical properties (see Table 1), e.g.
2
3
higher thermal conductivity and lower Young’s modulus, which contributes to the improvement of thermal shock
resistance of Al O –TiC–Co composites (ATC). In the research, the ATC composites were prepared from the cobalt
2
3
coating powders that were obtained by a newly developed coating technique, which was expected to improve the
mechanical and thermal shock resistance.
The thermal shock resistance of ATC composite was evaluated by indentation-quench technique and parallel
reference experiments were conducted with AT composites. In order to assess specially the effects of cobalt phase on
thermal shock behavior, four thermal shock parameters (R parameters) were introduced for the two composites in
terms of their mechanical and physical properties.
2
. Experimental procedures
2
.1. Material preparation and mechanical properties
The Al O (average particle size ꢀ40 nm) and TiC (average particle size ꢀ130 nm) powders were coated with
2
3
cobalt film (about 3.96 vol.%) by using electroless method, respectively. In the chemical deposition method,
CoSO ꢁ7H O is reduced by NaH PO to form a thin film of cobalt on the surface of Al O and TiC, respectively. The
4
2
2
2
2 3
two kinds of ceramic powder were mixed at a ratio of 70 wt.% Al O –Co to 30 wt.% TiC–Co and homogenized by
2
3
ultrasonic dispersion as the starting powders (provided by copartners of Ningbo Lingri Surface Engineering Co. Ltd.,
Zhejiang, China). Then the starting powders were hot-pressed in vacuum at 1650 8C for 30 min under a pressure of
3
7
0 MPa (Model HIGH MULTI 5000, FUJI DENPA, Japan). As a comparison study, Al O –TiC composites (about
2 3
0 wt.% Al O + 30 wt.% TiC) were prepared in the same sintering process. The hot-pressed bodies were cut into
2
3
bars, ground and polished to 1 mm finish before flexure strength and fracture toughness testing. Five samples were
tested for either flexure strength or fracture toughness. The flexure strengths were measured in three-point test
(
3 mm ꢂ 4 mm ꢂ 36 mm in size); fracture toughness tests were performed by the single edge notched beam (SENB)
method (2 mm ꢂ 4 mm ꢂ 36 mm in size) at a loading rate of 0.5 mm/min with a notch (2 mm in depth and 0.25 mm in
width) and a span of 20 mm; and HD-187.5 Brinell tester was used to test hardness. Relative density was measured by
Archimedes method. The phase composition of the polished samples was identified using X-ray diffractometry (XRD;
Model D/max-rA, Rigaku Co., Japan) using Cu Ka radiation.
2
.2. Indentation thermal shock
The indentation thermal shock technique developed by Andersson and Rowcliffe [10] was used to study the thermal
shock and thermal fatigue behavior of the materials. In this technique, the thermal shock resistance is measured by
studying the propagation of median/radial cracks around a Vickers indentation after single or repeated quenching. The
specimens for thermal shock were disk shaped, with a diameter of ꢀ30 mm and a thickness of 3 mm (Fig. 1a), with
parallel 1 mm polished surfaces. Three specimens were tested for each composite. The four indentations were made
uniformly, using a load of 294 N, on polished surfaces of each specimen. The holding time of each indentation was 5 s,
and the interval between successive indentations was 60 s. Each indentation comes into being four cracks (Fig. 1b).

2546
R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
Fig. 1. (a) The sample for indentation-quench technique. (b) Schematic of Vickers indentation crack.
The quenching experiments were performed in a vertical tube furnace in air. The specimens were inserted into a
preheated furnace at the testing temperatures and then were held there for 30 min before they were quenched into a
bath of water (ꢀ25 8C). The test was conducted at temperature differences, DT, of 100–200 8C. The radial crack
lengths of each indentation were measured using optical microscopy before and after quenching. To study the thermal
fatigue behavior of two composites, repeated thermal shocks (up to 15 cycles) were performed for selected DT value of
1
00 8C. The fracture surfaces and indentation cracks propagation were investigated using a scanning electron
microscope (SEM; Model S-2500, Hitachi).
3
. Results and discussion
3
.1. Mechanical properties and microstructure
Fig. 2 shows the XRD analysis of ATC composite. The final phases are Al O , TiC and cobalt, in which Al O and
2
3
2 3
TiC are major phases. The basic properties of the composites produced are summarized in Table 2. The two
composites made in this study are almost fully dense. A linear intercept method has been used to evaluate the grain size
of ATand ATC composites [11]. The grain sizes are also shown in Table 2. ATC composites have a smaller mean grain
size than AT composites. This suggests that the grain growth in ATC is suppressed by cobalt particles that pin grain
boundaries.
Hardness, flexure strength and fracture toughness of the two composites are also presented in Table 2. Hardness of
ATC composites decreases slightly with addition of the cobalt phase. However, there is significant improvement in
either flexure strength or fracture toughness of the ATC composites over that of AT composites.
Fig. 2. XRD pattern of sintered ATC.

R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
2547
Table 2
Composites date and properties
Composites
Al
2
O
3
TiC
(vol.%)
Cobalt
(vol.%)
Relative
Grain size
(mm)
Hardness
(HRA)
Flexure
strength
(MPa)
Fracture toughness
(MPa m )
1
/2
(vol.%)
density
ꢃ3
(
g cm
)
AT
ATC
74.2
71.28
25.8
24.76
/
98.7
99.5
2.9
2.4
92.8
92.7
412 ꢄ 65
782 ꢄ 60
4.10 ꢄ 0.60
7.81 ꢄ 0.80
3.96
It is well known that the thermal expansion mismatch led to residual comprehensive stress between TiC particles
and Al O matrix, which is considered to be the primary toughening mechanism of AT composites [3]. Fig. 3 shows
2
3
the typical SEM micrographs of the fracture surfaces of ATand ATC composites. The fracture surface of ATC shows a
large proportion of transgranular cracks with some intergranular type (Fig. 3a), unlike the intergranular fracture mode
of AT (Fig. 3b). In addition, cracks have been observed in the fracture surface of ATC composites (Fig. 3c), where TiC
particles play an effective role in crack branching, bridging and deflection. It is suggested that cobalt phase improves
interfacial bonding, leading to the change of fracture mode. This shows that the addition of 3.96 vol.% cobalt to AT
composites can change the fracture mode of the AT from predominantly intergranular to transgranular.
3
.2. Indentation thermal shock
Two parameters, namely fraction of cracks propagated, %P, and crack extension, %E, are defined as follows [12]:
DN
%
P ¼ N ꢂ 100
(1)
where N(=48) is the total number of available radial cracks around the indentation and DN the number of radial cracks
propagating after thermal shock loading.
Dc
%
E ¼ c ꢂ 100
(2)
where c is the average indentation crack length before thermal shock loading and Dc is the average crack length
extended after thermal shock loading; only cracks which extended are considered in calculating %E.
Table 3 and Fig. 4 summarize crack growth behavior in the AT and ATC composites after single quenching into a
bath of water (ꢀ25 8C). The data show that %P and %E are dependent on DT, and the higher DT, the higher %P and
%
E. From temperature difference 100 to 200 8C, the AT composites exhibit poor crack growth resistance after a single
thermal shock, in that %P and %E are higher than for ATC composites. The partly cracks run through the surfaces for
the two composites at temperature difference 200 8C.
Fig. 5 shows the Vickers indentation SEM micrographs of the two composites before and after thermal shock. The
indentation shapes of ATC are more intact than ones of AT no matter what are before or after thermal shock.
Fig. 3. SEM micrographs of the fracture surfaces: (a) AT composites (b) and (c) ATC composites.

2548
R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
Table 3
Crack growth data for single thermal shocks
Composites
Crack growth
DT = 100 8C
DT = 125 8C
%P
DT = 150 8C
%P
DT = 175 8C
%P
DT = 200 8C
%P
%P
%E
%E
%E
%E
%E
a
a
ATC
AT
18.8
25.0
11.9
14.5
50
58.3
14.5
23.7
62.5
68.8
19.7
33.3
66.7
77.1
24.3
41.4
100
100
a
Denotes partly cracks run through the whole sample surface.
Fig. 6 shows the indentation crack propagation SEM micrographs of the two composites. The ATC composites
show better crack propagation resistance than AT composite no matter what are before or after thermal shock. The
crack propagation routes of ATC composites are more winding than those of AT composites no matter what are before
or after thermal shock. There are more TiC particles that play a role in crack bridging in ATC composites than in AT
composites. Moreover, there are some TiC particles that play a role in crack branching in ATC composites. As
mentioned above, the presence of the cobalt phase could, to some extent, promote the crack bridging and branching of
TiC particles. The ATC composites show better resistance to thermal shock than AT composites.
Table 4 and Fig. 7 show crack propagation after repeated thermal shocks, at DT value of 100 8C. With cycles
increasing crack propagation accelerated. The all indentation cracks of the two composites propagated after 15 cycles.
The AT composites are more sensitive to repeated thermal shock. The surface of AT composites was spalled (Fig. 8)
under thermal stress, which is the weakness of the kind of material in utilization. The ATC shows better resistance to
repeated indentation fatigue.
Thermal shock of ceramics includes the fracture initiation and cracks propagation. The theory of cracks
propagation is based on Hasselman research [13]. The theory divides the relationship between flexure strength and
thermal shock temperature difference into three ranges. When thermal shock temperature difference (DT) is lower
than critical temperature difference (DT ) the flexure strength keeps the original value. When DT is equal to DT
c
c
the flexure strength degrades quickly. When DT is higher than DT the flexure strength keeps a constant. The
c
thermal shock behaviors of all ceramic materials possess the characterizations above mentioned. The different
ceramic material only shows different critical temperature difference (DT ). %P and %E are dependent on DT and,
c
the higher DT, the higher %P and %E [12]. The critical temperature difference DT can be quantitatively defined
c
with reference to %P and %E as the lowest temperature difference where two criterions are simultaneously
fulfilled. The one is that average extension of the crack is >10% of the original length (this criterion ensures
that the crack extension is larger than any measurement errors). The other is that more than 25% of indentation
cracks propagate (this criterion eliminates chance growth of single cracks and ensures statistically significant
data of crack growth). According to the two criterions, DT is 100 8C for AT and 125 8C for ATC. For a thermally
c
Fig. 4. Crack growth behavior in AT and ATC composites after single thermal shock: (a) fraction of cracks propagated %P and (b) extension in
length of propagated cracks %E.

R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
2549
Fig. 5. SEM images of Vickers indentation crack before and after thermal shock: (a) AT before thermal shock and (b) AT after thermal shock:
DT = 200 8C). (c) ATC before thermal shock and (d) ATC after thermal shock (DT = 200 8C).
(
shocked ceramic material, a critical DT for a sudden temperature changes is given by the following equations
c
[
14]:
s ð1 ꢃ nÞ
1
f
DTc ¼
(3)
Ea
fðbÞ
ꢀ
ꢁ
1
B
¼
1:5 þ b ꢃ 0:5 eꢃC=b
(4)
(5)
fðbÞ
th
b ¼ k
where s is the initial flexure strength, v the Poisson ratio, B and C shape factors of samples depending on the specimen
f
geometry, b the Biot modulus, E the Young modulus, a the thermal expansion coefficient, t a characteristic heat
transfer length (half the thickness of the plate), k the thermal conductivity and h is the surface heat transfer coefficient
acting between the plate and the cooling medium. f(b) is the damping parameter of the thermal shock (ranging from 0
to 1) and is a function of the Biot modulus. For an infinite plate, Manson [15] has derived a semi-empirical equation
relating b and 1/f(b) over the range 0 < b < 5 as follows:
1
3:25
¼
1:5 þ
(6)
fðbÞ
b

2550
R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
Fig. 6. SEM images of crack propagation before and after thermal shock: (a) AT before thermal shock and (b) ATafter thermal shock (DT = 200 8C).
(c) ATC before thermal shock and (d) ATC after thermal shock (DT = 200 8C).
Lower a and higher k result in lower thermal stress during thermal shock, consequently leading to the higher value
of DT . The thermal physics parameters of AT and ATC composites (such as v, a, E and h) were assumed to be
c
identical, and this assumption is basically reasonable for ATC composites with only 3.96 vol.% cobalt addition. And B
and C the shape factors of ATand ATC composites are both identical, so the damping parameters of the thermal shock
are also identical (see Eq. (4)). According to Eq. (3) the critical DT depends on flexure strength s . The flexure strength
c
f
sf can be expressed as follows [7]:
KIC
s ¼ pffiffiffi
(7)
f
Y C
where Yis a coefficient depending on shape and dimensions of crack and loading forms, C crack dimensions and K is
IC
fracture toughness. Thus the critical temperature difference of ATC composites is higher than that of AT composites,
Table 4
Cyclic thermal shock crack propagation of AT and ATC composites
Material
DT (8C)
1 cycle
5 cycles
10 cycles
15 cycles
%P
%E
%P
%E
%P
%E
%P
%E
a
a
ATC
AT
100
100
18.8
25.0
11.9
14.5
50.0
59.4
42.3
50.7
72.9
75.0
68.4
81.3
100
100
87.5
95.8
a
Partly cracks run through the surfaces.

R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
2551
Fig. 7. Crack growth behavior in AT and ATC composites after cyclic thermal shock: (a) fraction of cracks propagated %P and (b) extension in
length of propagated cracks %E.
which should be mainly attributed to the higher flexure strength and fracture toughness of ATC composites (see
Table 2).
The thermal shock resistance of brittle materials depends on thermal physics properties and mechanical properties.
Several thermal shock parameters (R parameters) have been defined to relate these materials to their thermal
resistance, considering the crack initiation and crack propagation conditions, respectively. The common used ones are
defined as following [7]:
sfð1 ꢃ nÞ
R ¼
(8)
(9)
aE
R0 ¼ kR
2
Egf
ðK =s Þ
IV
IC
f
R
¼
¼
(10)
(11)
2
ð1 ꢃ nÞsf
1 ꢃ n
ꢀ
ꢁ
1
=2
gf
Ea2
st
R ¼
Fig. 8. Surface spalling of AT composites after cyclic thermal shock.

2552
R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
Table 5
Thermal physics properties and R parameters of AT and ATC composites
a (ꢂ10ꢃ6 K)
k (W/m K)
E (GPa)
n
g
(J/m
2
R (8C)
R (W/m)
0
R
IV
(10
ꢃ6
)
R
st
(m
1/2 ꢃ1
K )
Materials
f
AT
ATC
8.2
8.2
23.9
23.9
401
401
0.24
0.24
21.0
77.4
95
180
2271
4302
130
131
0.88
1.69
0
where g is the surface fracture energy and other parameters were seen above. R and R parameters predict the critical
f
DT in a body under conditions of sharp and moderate heat or cold flow, respectively, which are expressed as the
c
IV
st
resistance to crack initiation. R and R parameters express the ability of a material to resist crack propagation and
further damage and loss of strength with increasing severity of thermal shock. Higher R and R values indicate crack
IV
st
stability once the critical DT is exceeded. The calculated thermal shock parameters (R parameters) are listed in
c
Table 5. As shown in Table 5, the calculated values of R show that the ATC composites possess a higher DT than that
c
of AT composites, with an increase of 85 8C, which is mainly attributed to flexure strength. However, the result is
higher than the experimental cure, which may be associated with the deviation of experimental data. Compared with R,
0
there is a greater increase in R , which shows that the difference of the critical temperature difference between ATC
composites and AT composites will become large when materials are under conditions of moderate heat or cold flow.
IV
IV
R
shows the ability of material to resist thermal shock damage. The calculated values of R indicate that the
st
resistance to thermal shock damage of the two composites is no distinct difference. R of ATC composites increases
around 92.1% as a result of their higher surface fracture energy. According to Eq. (12) [7], material surface fracture
energy is directly proportional to fracture toughness. The higher fracture toughness results in higher surface fracture
energy. Therefore the higher thermal shock parameters are mainly attributed to higher fracture toughness. In addition,
st
R is directly proportional to DT [7], which indicates critical DT of ATC composites should be higher than that of AT
c
c
composites. The flexure strength and fracture toughness for ATC composites are increased by 89.8% and 90.5% in
comparison to those for AT composites. The influence of 3.96 vol.% cobalt on thermal physics parameters of ATC
composites are little, even it can be ignored. However, cobalt phases greatly enhance the flexure strength and fracture
toughness that improves the thermal shock resistance of ATC. In a word, the better thermal shock resistance of ATC
composites is mainly attributed to their higher flexure strength and fracture toughness.
2
K
IC
gf ¼ 2
(12)
E
4
. Conclusions
Indentation techniques have been successfully used to study the thermal shock behavior of ATand ATC composites.
For a single thermal shock in the DT range 100–200 8C and repeated thermal shock in the DT 100 8C, an increase in
thermal shock resistance over AT composites can be observed in ATC composites with an addition of 3.96 vol.%
cobalt. The two composites are both sensitive to repeated thermal shock. The surface of AT composites was spalled
after 15 cycles. The experimental results show that the critical temperature difference for ATC composites (ꢀ125 8C)
is improved about 25 8C compared with that of AT composites (ꢀ100 8C). The calculated results of four thermal shock
parameters show the ability of ATC composites to thermal shock fracture and thermal shock damage is better than that
of AT composites. The thermal physics parameters of ATC composites are changed very little with adding small
amounts of cobalt but the flexure strength and fracture toughness of ATC composites are greatly improved, leading
them to higher thermal shock parameters. So the better thermal shock resistance of ATC composites should mainly be
attributed to the enhanced strength and toughness.
References
[
[
[
1] X.Q. You, T.Z. Si, N. Liu, Ceram. Int. 31 (2005) 33–38.
2] J. Zhao, X. Ai, X.P. Huang, Mater. Process. Technol. 129 (2002) 161–166.
3] J. Deng, T. Can, J. Sun, Ceram. Int. 31 (2005) 249–256.

R.-X. Shi / Materials Research Bulletin 43 (2008) 2544–2553
2553
[
[
[
[
[
[
4] M.K. Aghajanian, N.H. MacMiUan, C.R. Kennedy, J. Mater. Sci. 24 (2) (1989) 658–700.
5] S.-T. Oh, T. Sekino, K. Niihara, J. Eur. Ceram. Soc. 17 (4) (1997) 495–504.
6] O. Sbaizero, G. Pezzotti, Acta Mater. 48 (3) (2000) 985–992.
7] Y. Zhou, Ceramic materials, Harbin, vol. 342, Harbin Unibersity of Technology Press, China, 1995, p. 354.
8] J. Lu, L. Gao, J. Guo, Mater. Res. Bull. 35 (2000) 2387–2396.
9] D. Vollath, D.V. Szabo, J. Haubelt, J. Eur. Ceram. Soc. 17 (1) (1997) 1317–1324.
[10] T. Andersson, D.J. Rowcliffe, J. Eur. Ceram. Soc. 18 (14) (1998) 2065–2071.
[11] M. Li, M.J. Reece, J. Am. Ceram. Soc. 83 (4) (2000) 967–970.
[12] S. Maensiri, S.G. Roberts, J. Am. Ceram. Soc. 85 (8) (2002) 1971–1978.
[13] D.P.H. Hasselman an, J. Am. Ceram. Soc. 52 (6) (1969) 600–604.
[14] M. Collin, D. Rowcliffe, Acta Mater. 48 (2) (2000) 1655–1665.
[15] S.S. Manson, N.C.A. Technical Note 2933, Washington, D.C., 2002.