Anda di halaman 1dari 7

Intermetallics 15 (2007) 253e259

www.elsevier.com/locate/intermet

Competitive formation of glasses and glassematrix composites


Z.P. Lu a,*, D. Ma b, C.T. Liu a, Y.A. Chang b
a
Metals and Ceramic Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6115, USA
b
Department of Materials Science and Engineering, University of Wisconsin-Madison, WI 53706, USA
Received 31 March 2006; received in revised form 8 May 2006; accepted 29 May 2006
Available online 9 August 2006

Abstract

By systematically investigating the effect of chemical composition on the competitive formation of glasses in various systems, we attempt to
address two long-standing scientific puzzles upon metallic glasses, i.e., (i) which composition is the best for forming glasses and glassematrix
composites and (ii) what determines the easy glass-forming composition range in a given alloy system. Our findings have led to the construction
of a qualitative microstructure selection map, which is useful for guiding the design of bulkier metallic glasses and glassematrix composites. In
addition, our analysis demonstrates that the classical kinetic treatment of glass formation is insufficient; to analyze glass formation properly, it is
necessary to go beyond simple assumptions of single polymorphic solidification during crystallization.
2006 Elsevier Ltd. All rights reserved.

Keywords: B. Glasses, metallic; B. Phase transformations; B. Phase diagrams

1. Introduction (1) Glass formation occurs even after nucleation of crystalline


phases, essentially due to the suppression of crystal growth
Scientific problems about metallic glasses, including (i) [8], and
which composition is the best for forming glasses and glasse (2) The best glass-forming compositions were found to be off-
matrix composites and (ii) what determines the glass-forming eutectic rather than the eutectic points in some systems
composition range in a given alloy system, have challenged such as GdeFe and TieSi [3].
material scientists ever since metallic glass was first reported
in 1960 [1e4]. Based on the assumption that the nucleation It is Boettinger who first realized the importance of crystal
frequency of a melt scale as 1/h (h, viscosity of the liquid), growth effect on glass formation and qualitatively analyzed
Turnbull [5] has suggested that an alloy would be an easier the relationship between glass formation and the growth veloc-
glass former if its reduced glass transition temperature Trg ities as a function of composition in a regular binary eutectic
( Tg/Tl, where Tg and Tl are the glass transition temperature system [9,10]. More recently, this constrained solidification
and the liquidus temperature, respectively) is not less than 2/ theory has been applied to study the competition between
3. The implication of this theory is that the greatest glass- crystal growth and glass formation in irregular eutectic sys-
forming ability (GFA) is always associated with low-melting tems [11,12]. It was found that the best glass former in an ir-
eutectics (i.e., deep eutectics) [4,6], as has been demonstrated regular eutectic system is located at off-eutectic due to the
in many binary systems (e.g., TieBe [7]). However, this sce- large kinetic difficulty in the growth of the faceted crystalline
nario is strongly challenged by two types of experimental find- phase, as demonstrated in the CueZr system [11]. Despite its
ings over the years: success, however, such analysis appears to be inapplicable to
the cases where solidification of competing crystalline phases
is unconstrained (i.e., with limited nucleation and/or undergo-
* Corresponding author. ing a negative thermal gradient), such as the glass formation in
E-mail address: luzp@ornl.gov (Z.P. Lu). electromagnetically levitated undercooled melts and from the

0966-9795/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.intermet.2006.05.015
254 Z.P. Lu et al. / Intermetallics 15 (2007) 253e259

free side of melt-spun ribbons. In these cases, nucleation ef-


fects on glass formation cannot be neglected. Thus, there is (b)

Temperature
Liquid Liquid
an urgent need to extend the principle of competitive forma- Tl
tion of glasses to a more general situation where both the nu-
TE
cleation and crystal growth of competing crystalline phases
play important roles in glass formation.
The first attempt to incorporate the effects of both nucle-
ation and crystal growth on glass formation is the classical ki- Eu End
netic treatment established by Uhlmann [13] and Davies [14]. Tg Start
The recent discovery of multi-component bulk metallic glasses Rc Rcom
which possess an extremely high thermal stability with respect (a) Cooling

to crystallization has made the experimental determination of Composition Log (time)


a timeetemperature-transformation (TTT) diagram for glass-
forming liquids in the undercooled state possible. Direct ob- Fig. 1. (a) A typical binary phase diagram and (b) timeetemperature-transfor-
mation (TTT) diagrams showing the competitive formation between compet-
servation of TTT diagrams for glass formation will indicate
ing crystalline phases and glass. To form a glass, the cooling curve from the
the validity and feasibility of the classical kinetic treatment. liquid state must avoid intersecting the leftmost start curve. Eu stands for
It was found that, however, the measured TTT diagram can eutectic.
be exactly fitted over the entire temperature range by the clas-
sical kinetic treatment in bulk glass-forming Pd40Cu30Ni10P20
(at.%) alloy [13]. In contrast, the same treatment fails to pre- be undercooled to below the glass transition temperature
dict the experimentally determined TTT diagram for the fa- with no formation of both b and the eutectic. In other words,
mous Vitreloy 1 (Zr41.2Ti13.8Cu12.5Ni10Be22.5, at.%) because the liquid must be cooled fast enough from the liquidus tem-
of its significant high nose temperature [16]. Up to date, perature through the glass transition temperature without inter-
this theoretical dilemma remains unresolved. secting the starting TTT curve of any competing crystalline
In this paper, we will present a unique conceptual approach phase (e.g., b and the eutectic for this particular case). The
to predict the formation of glasses and glassematrix compos- minimum cooling rate required to form a full glass (i.e., the
ites. A comprehensive understanding of where glasses and critical cooling rate) is the one needed to bypass the leftmost
glassematrix composites are located in a given eutectic sys- nose of all the TTT curves in the entire undercooling, as de-
tem will be elucidated with considerations of both the nucle- picted by Rc in Fig. 1. Additionally, the composite structure
ation and crystal growth. The insufficiencies of the classical (i.e., the phase b embedded in the amorphous matrix) would
kinetic treatment of glass formation will also be discussed. be formed if the applied cooling rate were in-between Rc
and Rcom. When cooled at rates between Rc and Rcom, the liq-
2. Theoretical analysis uid starts to form the crystalline phase b. However, this trans-
formation is not complete because the cooling curve bypasses
2.1. Basic concept of competitive formation of glass the ending TTT curve for the a-phase formation. Thus, upon
further cooling, the remaining liquid turns into an amorphous
Glass formation is always a competitive process between phase. At cooling rates lower than Rcom, b would form first and
a liquid phase and the competing crystalline phases, as demon- then the remaining liquid will solidify into the eutectic struc-
strated by Lu and Liu [17]. As pointed out by Turnbull [4] and ture at a constant temperature (formation of the eutectic causes
Davies [14], any liquid can form a glass as long as the cooling the system to cease cooling until solidification is complete
rate is fast enough to suppress the crystallization event. In this [19]). Therefore, Rcom is actually the critical cooling rate
sense, glass formation can be well explained by TTT diagrams (i.e., the minimum cooling rate) for forming the composite
[18]. In a typical binary eutectic system [see the phase dia- structure.
gram in Fig. 1(a)], the competitive formation between the
glass phase and its competing crystalline phases during the 2.2. Glass and composite formation in a regular
cooling of an off-eutectic alloy (toward the b-phase side) can eutectic system
be schematically illustrated by Fig. 1(b). The solid curves (de-
noted as the starting TTT curve hereafter) indicate the begin- It is generally believed that glasses form preferentially in
ning of the crystallization with a barely detectable volume systems with deep eutectics [3,4]. In general, eutectic alloys
fraction of crystallites (usually w106) while the dashed can be divided into two main types [20,21]: regular and ir-
curves (denoted as the ending TTT curve hereafter) represent regular. For a regular eutectic system, studies have indicated
the end of the crystallization event. At small undercoolings be- that a symmetrical phase diagram is favored, as illustrated in
tween Tl and TE, formation of primary b phase dominates the Fig. 2(a) [20]. At the eutectic composition Ce, two phases
crystallization event, while, at large undercoolings between TE (e.g., a and b) will grow in a coupled manner to form a stable
and Tg, formation of the eutectic structure dominates (i.e., the eutectic structure of these two phases [19,20]. Therefore, only
coupled growth of b and another equilibrium phase). In this one TTT curve for the eutectic formation in this alloy is seen,
case, in order to form a glass upon cooling, the liquid must as demonstrated in Fig. 2(b). In order to form a fully
Z.P. Lu et al. / Intermetallics 15 (2007) 253e259 255

(a) (b) (c)


L T T
Ce C1

TL

TE TE
Temperature Ce Eu Start

End
Rc Rc Eu
Tg Tg
C2
Log (time) Log (time)
A C2 Ce C1 C3 B
Composition

(f) (d) (e)


T T C3
C2 TL
Am
+ Am Am
+ TL

TE TE
Rcom
Cooling rate

Rc

Eu+ Eu Eu+
Rc Eu Rc Rcom Eu
Tg Tg
C2
Log (time) Log (time)
A C2 Ce C1 C3 B
Composition

Fig. 2. The composition effect on glass formation in a regular eutectic system; (a) phase diagram with glass-forming range, (b)e(e) competitive formation between
glass and primary phases as a function of composition and (f) final microstructure as a function of cooling rate and composition. Eu and Am stand for eutectic and
amorphous structures, respectively.

amorphous structure in this alloy, the applied cooling rate has and (d), the critical cooling rate for full glass formation is still
to be large enough to bypass the nose of the TTT curve (i.e., determined by the starting TTT curve of the eutectic forma-
larger than Rc) so that the formation of the eutectic structure tion, although b can be formed at slow cooling rates. With fur-
can be avoided. As the alloy composition is moved from Ce ther shifting of the composition toward the b-phase side, full
toward C3, the liquidus temperature of this off-eutectic alloy glass formation starts to compete primarily with b-phase for-
is increased and the composition difference between the liquid mation, as shown in Fig. 2(e). C2 is actually the critical com-
and the b phase is decreased [see Fig. 2(a)]. Thermodynami- position beyond which full glass formation competes
cally, the driving force for crystallization of the b phase is en- essentially with the formation of the b phase rather than the
hanced due to the increase of the liquidus temperature. eutectic. It is noted that composite materials, consisting of
Kinetically, the decreased composition difference between a mixture of amorphous matrix and b phase, can be obtained
the liquid and the b phase requires less diffusion of atoms. with cooling rates between Rcom and Rc for compositions
Thus, the formation of the b phase is greatly facilitated larger than C2 [see Fig. 2(e)]. In analogy with the b-phase
when the alloy composition is moved toward the b-phase side, similar results can be obtained when the alloy composi-
side [20]. As a result, for these off-eutectic alloys, there are tion is shifted toward the a-phase side and C02 is the composition
two TTT curves (as the case shown in Fig. 1): one due to limit over which the GFA starts to be determined by the a-
the b-phase formation at small undercoolings (i.e., close to phase formation.
the liquidus temperature) and the other due to the eutectic for-
mation at relatively large undercoolings (i.e., close to the glass 2.3. Glass and composite formation in
transition temperature). In addition, with the increase of the an irregular system
constituent B, the starting TTT curve of the b phase will grad-
ually shift to the top-left (i.e., high-temperature and short-time For an irregular eutectic system, studies indicate that an
regimes), as illustrated in Fig. 2(cee). In order to form a glass asymmetrical phase diagram is encouraged, as shown in
in these off-eutectic alloys, the formation of both b and the eu- Fig. 3(a) [20]. Such kinds of eutectic usually consist of one
tectic has to be completely suppressed. In the case of Fig. 2(c) non-faceted phase (i.e., the a phase) and one faceted phase
256 Z.P. Lu et al. / Intermetallics 15 (2007) 253e259

(a) (b) (c)

L
T T
Ce C1

TL
TE TE

Temperature

Ce
Start

End
Rc Rcom Eu Rc Eu
Tg Tg

C1 Log (time) Log (time)

A Ce C2 C3 B
composition

(f) (d) (e)


T T C3
Am C2
Am+ Am+ TL
TL

TE TE
Cooling rate

Eu+ Eu Eu+
Rc Eu Rc Rcom Eu
Tg Tg
C1
Log (time) Log (time)
A Ce C2 C3 B
Composition

Fig. 3. The composition effect on glass formation in an irregular eutectic system; (a) phase diagram with glass-forming range, (b)e(e) competitive formation be-
tween glass and primary phases as a function of composition and (f) final microstructure as a function of cooling rate and composition. Eu and Am stand for
eutectic and amorphous structures, respectively.

(i.e., the b phase) which has anisotropic growth characteristics Fig. 3(c)], glass-forming ability is controlled no longer by
[20,21]. It is well known that a fully eutectic structure cannot the a phase but solely by the formation of the eutectic. C2 is
be obtained when an alloy with the eutectic composition [i.e., the critical composition where the critical cooling curve for
Ce in Fig. 3(a)] is rapidly solidified because the a phase can be full glass formation just barely intersects the noses of both
formed prior to the eutectic and the b phase at large undercool- TTT curves for the eutectic and primary b phase.
ings [20,21]. Therefore, there exist two TTT curves for the At compositions above C2 [e.g., C3 in Fig. 3(e)] the tendency
undercooled liquid in this alloy [as schematically demon- for forming the faceted b phase, is much stronger than that of
strated in Fig. 3(b)]: one due to the formation of the coupled the eutectic, and their glass-forming ability is thus much de-
eutectic and the other caused by the formation of a at rela- creased as compared with that of compositions C1 and C2. As
tively large undercoolings. In order to form a full glass without shown in Fig. 3(e), in order to form a glass, the applied cooling
a, the applied cooling rate has to be fast enough without inter- rates have to be large enough to bypass the nose of the starting
secting both TTT curves, while the composite will be obtained TTT curve for b-phase formation. Similarly, composite mate-
at cooling rates in-between Rc and Rcom. Similar to the case of rials, consisting of b phase embedded in the glass matrix, will
the regular eutectic, as the alloy composition is moved from Ce be obtained with cooling rates ranging from Rcom to Rc.
toward the b-phase side, the tendency for forming the b phase
would become greater (concurrently, the formation of a is 2.4. Composition effects on glass formation
relatively retarded) [20]. As a result, the TTT curve for the
a-phase formation at low temperatures is moved to the As a result of the above analysis, we have attempted to con-
bottom-right (i.e., long-time and low-temperature regimes) struct a qualitative microstructure selection map, as shown in
and eventually vanishes, while the TTT curve for b-phase for- Fig. 2(f), for a typical regular eutectic system in terms of com-
mation is gradually shifted toward the top-left (i.e., short-time position and cooling rate. As can be seen, the best composition
and high-temperature regimes) in the temperatureetime coor- for fully glass formation in such a system is located exactly at
dinate as schematically demonstrated in Fig. 3(cee). Thus for the eutectic composition. Nevertheless, the formation of full
alloys whose compositions are between C1 and C2 [see glass for all other compositions between C02 and C2 also
Z.P. Lu et al. / Intermetallics 15 (2007) 253e259 257

compete primarily with the eutectic, and the critical cooling 1000
rate for the full glass formation is just slightly larger than
950 Tl=970K
that of the eutectic composition Ce. The discrepancies in Rc
of these alloys are caused by the variations in their liquidus
temperatures. Outside the composition interval between C02 900 Te=885K

and C2, the critical cooling rates for forming a fully glassy
850
structure are increased sharply due to the strong competition Primary Ni3P

Temperature, K
from the much easier formation of primary phases. In other
800
words, the composition band between C2 and C02 as indicated
in Fig. 2(a), actually represents the easy glass-forming range 750
for regular eutectic systems. Outside the easy glass-forming
range, in-situ composite materials (i.e., primary phase of either 700
a or b embedded in an amorphous matrix) can be obtained.
Similarly, a microstructure selection diagram can also be 650
Eutectic
established for an irregular eutectic system, as shown in
Fig. 3(f) which schematically illustrates the resultant phase/ 600
Tg=580K
microstructure as a function of cooling rate and alloy compo-
sition. For compositions below C1 (e.g., Ce), the critical cool-
103 104
ing rate for forming a full glass is very high due to the much
Time, t/s
easier formation of the a phase, although composite materials
(i.e., primary non-faceted a phase embedded in the amorphous Fig. 4. Experimentally determined TTT curves of the B2O3 fluxed Pd40Ni40P20
matrix) can be formed at relatively lower cooling rates. For alloy during the entire undercooling.
compositions ranging from C1 to C2, the critical cooling rate
for full glass formation is much reduced and shows a minimum for the formation of primary phase Ni3P and the other is for
at the composition C1 (i.e., composition C1 has the largest the formation of eutectic structure. Glass competes with the
GFA in the entire system). Nevertheless, the difference be- Ni3P phase at high temperatures (i.e., small undercoolings)
tween the critical cooling rates for all the compositions in and with the eutectic at low temperatures (i.e., large under-
the range from C1 to C2 is not striking. For compositions coolings). This particular experimental observation clearly
above C2, the critical cooling rate is steeply increased again verifies that glass formation is really a competing process
due to the easier formation of the faceted b phase. Thus, the and our analysis is well ground. More recently, work
composition interval between C1 and C2 is the easy glass- on non-contact measurement of TTT curves using the
forming range for the irregular eutectic system, as indicated electrostatic levitation method for glass-forming liquids
in Fig. 3(a). Composite materials can also be formed at com- Zr55Al20Co25, Zr55Al22.5Co22.5 and Zr55Al19Co19Cu7 without
positions above C2 (e.g., C3) but the primary phase is switched doubt further validate our analysis; double-nose TTT
from the non-faceted a to the faceted b. curves were actually experimentally observed in the under-
It is necessary to point out that the microstructure selection cooled Zr55Al20Co25 and Zr55Al19Co19Cu7 alloys [23]. For ex-
maps as a function of cooling rate shown in Figs. 2(f) and 3(f) ample, the alloy Zr55Al20Co25 simultaneously crystallizes into
are slightly different from those as a function of growth veloc- a mixture of two crystalline phases Zr6Al2Co and ZrAl0.6Co1.4
ity obtained with sole consideration of crystal growth [12]. (i.e., the eutectic formation) at small undercoolings while sin-
Our analysis successfully incorporates the nucleation effects gle phase Zr6Al2Co is formed at relatively large undercool-
on glass formation and can thus be applied to various solidifi- ings. In this alloy, the critical cooling rate for glass
cation processes. In addition, actual glass-forming range that formation can be determined by avoiding the leftmost nose
is a function of cooling rate as shown in Figs. 2(f) and 3(f) of the eutectic TTT curve [23]. On the other hand, only one
could be enlarged with high-cooling capacity vitrification single TTT curve of the alloy Zr55Al22.5Co22.5 is seen during
techniques. Nevertheless, the shaded area in Figs. 2(a) and the undercooling below the melting temperature TE. The alloy
3(a) represents the composition range for easy glass formers was found to crystallize also simultaneously into a mixture of
that have relatively larger GFA (i.e., lower critical cooling Zr6Al2Co, ZrAl0.6Co1.4 and other unidentified phases, indicat-
rate) than the rest of the alloys in the system. ing that the glass phase in this alloy competes primarily with
the eutectic structure and can be formed if the nose of the eu-
3. Direct experimental validation of the current approach tectic TTT curve is bypassed [23].
It is important to point out that the microstructure selection
Our current analysis has been directly confirmed by a few maps determined from our current analysis is also consistent
experimental observations in literature. As one example, with experimental data reported in various glass-forming
Fig. 4 shows continuous cooling transformation curves of eutectic systems. Fig. 5 presents GFA data as a function of
B2O3 fluxed Pd40Ni40P20 alloy reported previously in literature composition in two representative eutectic systems, i.e.,
[22]. As shown, there indeed exist two curves for phase com- Ti2CueTiCu and bTieTi5Si3. As can be seen, Fig. 5(a) is
petition over the temperature range between Tl and Tg, one is a typical regular eutectic in the TieCu binary system, which
258 Z.P. Lu et al. / Intermetallics 15 (2007) 253e259

(a) (b)
2130C

1005C 982C 1670C

43 13.5
Temperature

Temperature
1330C

Ti5Si3
790C

TiCu
Ti2Cu

Ti 40-45 Ti 15-20

Fig. 5. Experimental data in two typical eutectic systems clearly verifying the current analyses; (a) regular eutectic Ti2CueTiCu system with glass-forming range
[24] and (b) irregular bTieTi5Si3 system with glass-forming range [24].

consists of two intermetallic phases Ti2Cu and TiCu with sim- Zr55Al22.5Co22.5 [23] and Pd40Cu30Ni10P20 [15], phase compe-
ilar melting points. The glass-forming composition range de- tition during undercooling is similar to the cases of either Ce in
termined using a melt-quenching technique (the cooling rate Fig. 2 or C1 in Fig. 3. Glass formation in these alloys is pri-
is w 106 K/s) was found to be 40e45 at.% Cu, almost equally marily limited by the eutectic formation during the tempera-
distributing on both sides of the eutectic (i.e., 43 at.% Cu) ture interval TE and Tg. The eutectic formation could be
[24]. This experimental finding is exactly consistent with the treated as a polymorphic transformation because the melting/
analysis shown in Fig. 2(a) for a regular eutectic system. In solidification behavior of eutectics is similar to that for
the irregular eutectic system like bTieTi5Si3, the melting a pure metal, i.e., congruently melting without significant
points of the two phases differ greatly and the phase diagram change of average composition [19,20]. Therefore, the classi-
is asymmetrical [see Fig. 5(b)]. The glass-forming range ex- cal kinetic treatment of glass formation is valid and suitable
perimentally determined using the melt-spinning technique is for predicting the GFA in these alloys.
from 15 to 20 at.% Si, which is skewed toward the hypereutec- As mentioned earlier, however, multi-component glass-
tic (Ce 13.5 at.% Si) [24]. This experimental observation is forming liquids such as the alloy Zr55Al20Co25 could have
also in excellent agreement with the analysis shown in a non-C-shaped TTT curve with multiple noses. In this
Fig. 3(a). As pointed out earlier, actual glass-forming range case, description of glass-formation kinetics using the classical
in a system is dependent on the used vitrification technique. crystallization theory is insufficient. The main reason is that
With high-cooling capacity, for example, even intermetallic different phases could be kinetically selected to form at differ-
compositions could be amorphized in the TieCu system ent undercoolings. The classical kinetic treatment could only
[25]. Nevertheless, easy glass-forming range in this particular be applied to predict the eutectic TTT curve, but composition
system is indeed around the eutectic [24]. effects on the nucleation and crystal growth for formation of
other crystalline phases have to be taken into account. Desre
and his co-workers [26e28] have attempted to discuss the
4. Insufficiency of the classical kinetic treatment of roles of concentration fluctuations in multi-component sys-
glass formation tems on nucleation. Wu [29] and others [30,31] have investi-
gated barrier crossing in multi-component systems which
As mentioned earlier, the classical kinetic treatment of involves correlated fluctuations of composition and topologi-
glass formation was established based on the assumption cal nucleus growth. Nevertheless, for a multi-component
that glass formation occurs via the avoidance of the polymor- glass-forming liquid with n competing crystalline phases, the
phic crystallization over the entire undercooling, i.e., the trans- time, t, for forming a small volume fraction, X, of crystal at
formation from liquid to one type of crystal which is temperature, T, can be simply expressed as follows:
congruently melting without a change of composition [5,14].
Therefore, two prerequisite conditions for applying this partic- tT; X minteutectic T; X; t1 T; X; t2 T; X; .; tn T; X
ular theory are as follows: 1

(1) The crystallization event is a polymorphic transformation, where teutectic(T, X ) is the time needed for simultaneously form-
and ing a small portion X of the mixture of all n crystalline phases
(2) Only single type crystallization phase is kinetically fa- (i.e., the eutectic structure) and tn(T, X ) is the time required
vored during the entire undercooling. for forming the small fraction X of individual crystalline phase
at the temperature T. In order to form glass, crystallization must
As a result, a single C-shaped TTT curve is always con- be avoid at all undercoolings in the time scale of t(T, X ). With
structed using the classical kinetic treatment. For the alloys modern computation approaches, crystallization dynamics and
Z.P. Lu et al. / Intermetallics 15 (2007) 253e259 259

solidification kinetics of each phase or structure is more readily References


available than before. It can thus be foreseen that prediction of
kinetics of glass formation in metallic alloys in a quantitative [1] Cahn RW. Nature 1975;257:356.
manner is highly possible. [2] Cahn RW. Nature 1978;274:848.
[3] Cahn RW, Greer AL. In: Cahn RW, Haasen P, editors. Physical metal-
lurgy. 4th ed., vol. 2. North-Holland; 1996. p. 1739.
[4] Turnbull D, Cohen MH. Nature 1961;189:131.
5. Conclusions [5] Turnbull D. Contemp Phys 1969;10:473.
[6] Highmore RJ, Greer AL. Nature 1989;339:363.
In this paper, it was found that the competition between pri- [7] Tanner LE, Ray R. Acta Metall 1979;27:1727.
mary phases, the eutectic and the amorphous phase at different [8] Greer AL. Key Eng Mater 1987;13e15:293.
[9] Boettinger WJ. In: Proceedings of the fourth international conference on
undercoolings determines the best glass formers, possible easy rapidly quenched metals. Sendai; 1981. p. 99.
glass-forming range and glassematrix composite formation in [10] Boettinger WJ. In: Kear BH, Giessen BC, Cohen M, editors. Rapidly so-
any given systems. Our analysis has led to the construction of lidified amorphous and crystalline alloys. Elsevier Science; 1982. p. 15.
the microstructure selection map, which can be used as guide- [11] Wang D, Li Y, Sun BB, Sui ML, Lu K, Ma E. Appl Phys Lett 2004;84:4029.
lines for developing new metallic glasses and composites in [12] Ma D, Tan H, Wang D, Li Y, Ma E. Appl Phys Lett 2005;86:191906.
[13] Uhlmann DR. J Non-Cryst Solids 1972;7:337.
various material systems. In addition, our analysis has demon- [14] Davies HA. Phys Chem Glasses 1976;17:159.
strated that the classical kinetic treatment of glass formation [15] Loffler JF, Schroers J, Johnson WL. Appl Phys Lett 2000;77:681.
can be applied only to the cases where the glass phase com- [16] Kim YJ, Busch R, Johnson WL, Rulison AJ, Rhim WK. Appl Phys Lett
petes primarily with the eutectic phase. For the other cases, 1996;68:1057.
however, it is necessary to go beyond simple assumptions of [17] Lu ZP, Liu CT. Phys Rev Lett 2003;91:115505.
[18] Gilman JJ. Science 1980;208:856.
single polymorphic crystallization. [19] <http://www.msm.cam.ac.uk/doitpoms/tlplib/CD4/cooling.php>.
[20] Kurz W, Fisher DJ. In: Fundamentals of solidification. Uetikon-Zurich,
Switzerland; 1998. p. 93.
Acknowledgements [21] Elliott R. Eutectic solidification process: crystalline and glassy alloys.
London: Butterworths; 1983. p. 55e88.
[22] Inoue A, Nishiyama N, Kimura H. Mater Trans JIM 1997;38:179.
This research was sponsored partly (Lu and Liu) by the
[23] Mukherjee S, Kang HG, Johnson WL, Rhim WK. Phys Rev B 2004;
Division of Materials Sciences and Engineering, Office of Basic 70:174205.
Energy Sciences, US Department of Energy under contract [24] Dubey KS, Ramachandrarao P, Lele S. Thermochim Acta 1996;280/281:25.
DE-AC05-00OR-22725 with UT-Battelle, LLC. Two authors, [25] Massalski TB, Woychik CG. Acta Metall 1985;33:1873.
D. Ma and Y.A. Chang, gratefully acknowledge financial sup- [26] Desre PJ, Eustathopoulos N. Philos Mag A 2002;82:2965.
[27] Desre PJ, Cini E, Vinet B. J Non-Cryst Solids 2001;288:210.
port for this work from the Defense Advanced Research Pro-
[28] Cini E, Vinet B, Desre PJ. Philos Mag A 2000;80:955.
jects Agency (DARPA), under ARO Contract No. DAAD [29] Wu DT. J Chem Phys 1993;99:1990.
19-01-1d525, and with Dr. Leo Christodoulou as Program [30] Trinkhaus H. Phys Rev B 1983;27:7372.
Manager. [31] Reiss H. J Chem Phys 1950;18:840.

Anda mungkin juga menyukai