Sunteți pe pagina 1din 22

Journal of Non-Crystalline Solids 76 (1985) 281-302 281

North-Holland, Amsterdam

T H E S Y N T H E S I S OF A M O R P H O U S N i - T i ALLOY P O W D E R S BY
M E C H A N I C A L ALLOYING*

R.B. SCHWARZ**, R.R. PETRICH + and C.K. SAW ++


Materials Science and Technology Division, Argonne National Laborato(v, Argonne, Illinois 60439,
USA

Received 11 March 1985


Revised manuscript received 31 May 1985

Dedicated to the memory of Professor Pol Duwez

Crystalline powders of Ni and Ti were mechanically alloyed by high-energy ball milling in an


inert atmosphere at T < 240 K. The alloyed NixTi I -x powders were characterized by scanning
electron microscopy, X-ray diffraction, and differential scanning calorimetry. For 0.28 < x < 0.72,
the powder is amorphous. This composition range is in good agreement with that deduced from a
free energy diagram for the crystalline terminal solutions and the amorphous phase. Outside this
regime, the powder is a two-phase mixture of an amorphous phase and the crystalline terminal
solution of the major element. After ball milling, the solubility limit of Ti in FCC Ni is
approximately 28%, which is significantly larger than that in annealed Ni. Atomic pair distribution
functions, G(r), were calculated from the X-ray patterns for x = 0.32, 0.4, 0.6, and 0.7. For
x = 0.4, the G(r) of the mechanically alloyed powder is almost identical with that of rapidly
quenched amorphous Ni4oTi60. For Ni33Ti67, the crystallization temperature, the crystallization
enthalpy, and the apparent activation energy for crystallization are 712 K, 0.74 kcal/g at., and 65.8
kcal/g at., respectively. These values are within 10% of those measured by Buschow for rapidly
quenched Ni32Ti6s. The amorphization by mechanical alloying is attributed to mechanical mixing
and to a solid state interdiffusion reaction taking at or near clean boundaries between polycrystal-
line Ni and Ti.

1. Introduction

Amorphous metallic alloys have traditionally been synthesized by the rapid


solidification technique [1], whereby molten alloy is cooled to a temperature
below the crystallization temperature Tx of the amorphous phase at rates
b e t w e e n 10 4 and 1011 K / s . Several new methods of synthesis of amorphous

* Work supported by the US Department of Energy.


** Present Address: Center for Materials Science, Los Alamos National Laboratory, Los Alamos,
NM 87545.
+ Work performed in partial fulfillment of requirements for MS degree, Illinois Institute of
Technology, Chicago, IL 60616.
++ Present address: Celanese Research Co., Summit, NJ 07901, USA.

0022-3093/85/$03.30 © Elsevier Science Publishers B.V.


(North-Holland Physics Publishing Division)
282 R.B. Sehwarz et al. / Synthesis of amorphous N i - Ti alloy

alloys, based on isothermal solid-state reactions which are carried out at


temperatures below Tx, have recently been advanced. These are: (1) amorphi-
zation by solid state interdiffusion reactions [2-5]; (2) amorphization by
mechanical alloying (MA) [6-8]; and (3) amorphization by hydrogen absorp-
tion and desorption [9,10].
Two conditions have been advanced as necessary for a successful solid state
amorphization reaction, hereafter referred as SSAR [4,5]: (a) The two alloying
elements must have a large negative heat of mixing in the amorphous phase
and (b) one of the elements must be an anomalously fast diffuser in the other.
The first condition ensures that a sufficiently large thermodynamic force exists
to drive the crystalline-to-amorphous reaction. The second condition ensures
that the amorphous product will be formed at a reasonable rate. To date,
several systems selected on the basis of these two criteria have been amorphized
by this method: Au and La [4], Au and Y [5], Ni and Zr [11], Hf and Ni [12],
and CO and Zr [13]. In all cases, alternating layers of the pure crystalline
metals, each several hundred nm in thickness, were deposited in vacuum to
form multilayer films. An annealing period of several hours at a temperature
approximately 100 K below Tx was sufficient to react stacks of crystalline films
into a homogeneous alloy film. An added feature of the SSAR method is that
the end products of a given solid state reaction correspond closely to those
deduced from a Gibbs free-energy diagram for the initial (crystalline) and final
(amorphous) phases involved [4,5]. The homogeneity range of the single-phase
amorphous product obtained with the SSAR method is, in general, wide and
centered near the equiatomic composition. In contrast, because of the rapid
cooling rate required for preparation of amorphous alloys by rapid solidifica-
tion of melts, the homogeneity range for such alloys is usually more limited
and is centered near deep eutetics in the binary phase diagram.
Mechanical alloying is a high-energy ball milling technique for producing
composite metal-base powders with controlled microstructures. Thi~ method
has been particularly successful in the synthesis of complex crystalline oxide
dispersion-strengthened alloy powders which are subsequently consolidated by
HIP techniques [14-16]. Koch ]8] first showed that an amorphous Ni60Nb40
powder obtained by MA of pure Ni and Nb powder has an X-ray diffraction
pattern and crystallization behavior similar to that of amorphous liquid-
quenched Ni60Nb40. Little is known about the process of amorphization by
MA. The simplest mechanisms that can be envisaged are based on the rapid
solidification of local melts and on the SSAR. The arguments in favor of each
of these mechanisms are as follows.
Each time two balls collide, the powder particles that are trapped in
between them are subjected to frictional glide and plastic deformation. This
may lead to the formation of melt pools, similar to those observed in the
shock-wave consolidation of powders [17]. These melts would solidify rapidly
by heat conduction into the cooler (less deformed) interior regions of the
particles and into the balls. If this quenching is fast enough, the boundaries
between the particles would become amorphous. The repeated fracture and
R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy 283

bonding of the particles during MA would make the entire powder amorphous.
On the other hand, even though the mixing of the powders is largely a
mechanical process, it is likely that the formation of the amorphous phase
involves solid state reactions similar to those which lead to the amorphization
of multilayered thin films. The conditions SSAR to occur are certainly present
at the clean N i / N b interfaces produced by the ball milling process: (a) Ni and
Nb have a large negative heat of mixing [18], and (b) Ni is an anomalously fast
diffuser in Nb [19]. In the case of bali milling, the interdiffusion should be
further enhanced by the suPersaturation of point defects and the large disloca-
tion density that is generated by plastic deformation. In order to determine the
roles of rapid quenching a n d / o r SSAR in amorphization by MA, this process
must be studied over a wide range of compositions.
The present paper is a study of the synthesis of amorphous Ni,Ti 1 ~ alloy
powders by MA of crystalline Ni and Ti powders. The N i - T i system was
chosen because it satisfies the two criteria for SSAR and because the structural
and thermal properties of amorphous NiTi alloys obtained by the rapid
quenching of melts are well documented. The reaction products obtained by
MA are compared with those previously obtained by the rapid quenching of
melts and with those expected from SSAR.

2. Experimental methods

Pure Ni and Ti powders [20] were mechanically alloyed in a laboratory ball


mill/mixer [21]. Ten-gram charges of a mixture of these powders were loaded
into a hardened-steel vial in an Ar-filled glove box. Sixty steel balls (0.25 inch
diameter, Allegheny Metal H-17, type H-440) were added to the vial. The lid
and the vial were then sealed together with a compressed indium ring. The
mixer agitated the vial at about 20 Hz with an amplitude of 50 mm. Hence, the
balls inside the vial reached velocities of 1 m / s . In the absence of additional
cooling, the energy dissipated by friction raised the temperature of the vial
approximately 30 K above room temperature. In order to favor the fracture of
the particles and thus minimize the formation of ductile coatings on the walls
and balls, the vial was cooled to approximately 240 K by a dripping flow of
liquid nitrogen. The milling was interrupted after each hour in order to rotate
the vial. This further minimized the formation of deposits on the walls. The
synthesis of a homogeneous amorphous powder required approximately 14 h
of MA.
The metal chemical composition of each alloyed powder was analyzed by
atomic-emission spectroscopy on an inductively coupled plasma. The Ni and
Ti contents were within _+2% of the nominal composition. The most signifi-
cant metallic contaminants were Fe ( < 0.5 wt%), Cr( < 0.08 wt%), and Zn( < 0.2
wt%). The accuracy of the analysis is 1-3% of the amount present for Ni and
Ti, 3-10% of the amount present for Cr and Fe, and an order of magnitude for
other contaminants. The Fe and Cr contamination was most likely introduced
284 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

by the erosion of the balls and container during the milling process. The
amounts of H, O, and N in mechanically alloyed Ni32Ti68 were measured by an
inert-gas fusion technique, with NBS-calibrated standards. The results were
< 5 ppm H, 3400 ppm O, and 560 ppm N. The O contamination was most
likely present in the initial Ti powder.
The structure of the mechanically alloyed powders was characterized by
X-ray diffraction with M o - K , radiation and a horizontal GE diffractometer
with a solid state Si(Li) detector. After subtraction of the background intensity,
the diffraction data were Fourier transformed to obtain the reduced total
atomic pair distribution function of the amorphous alloy (section 4.3).

3. Free-energy diagram

It has been shown that the products of SSAR in a stack of crystalline films
can be predicted by a free-energy diagram for the initial and final products
evaluated at the reaction temperature [4,5]. Fig. 1 is such a diagram for the
present Ni-Ti system at 235 K. Some of the approximations adopted to
construct this diagram differ from those used previously, as discussed below.
The free energies of the two starting pure crystalline metals are referred to
the gas state at 0 K, and were obtained from published data [22]. The
calculation of the free energy of the amorphous phase as a function of
composition x, requires knowledge of the free energies of amorphous Ni and
amorphous Ti at the reaction temperature T. The differences in free energy
between the pure crystalline and amorphous phases can, in principle, be
calculated. Consider the cooling of a molten metal in the absence of crystalliza-
tion. As the temperature is decreased from the melting temperature, Tm, the
melt first becomes undercooled and at the glass transition temperature, Tg, it
becomes configurationally frozen in a glassy state. Generally, the heat capacity
of a metal is higher in the undercooled state than in the crystalline state. The
I x
difference in specific heat, A c p (defined as Cp- Cp), increases with falling
temperature until Tg, where ZlCp decreases, smoothly to almost zero. For
T < Tg, Acp = 0. The difference in Gibbs free energy between the amorphous
and crystalline phases of a pure element, A G a - c , is given by*
AG = A H - TAS, (1)
where
Tn~
A H = Z ~ H f - f r ACp d T (2)

and

A S = A S f - - f~m Acp dT
T ' (3)

* We make no distinction in notation between the undercooled state and the glassy or amorphous
state.
R.B. Schwarz et al. / Synthesis of amorphous Ni- Ti alloy 285

/
. L
amorpn.-k ,~
u NixTi, - x /1 - 9 4
lk / 1-9
c~y~t.-t~ ~ 1-98
~-- ~ 4-ioo

~i! "~ ~ ~i -I04


~ \\.~ ,~-

""~ . ~ ~ ' ~ " #.,,~ - omorph.

.IQ
-~--o .~b~. c d

o ' o12' &' o18' i.o


Ni Ti
(I-X)
Fig. 1. Free energy of alloys of Ni and Ti. The heavy solid curve is the free energy of the
amorphous phase. The thin curves are the free energies of the crystalline terminal solid solutions.
The dotted curves are the free energies of the crystalline intermetallic compounds. The dashed lines
are tangents common to the crystalline solid solutions and the amorphous phase. These define
composition regimes a, b, c, and d. Regime (a) is the crystalline Ni-rich terminal solution. Regime
(c) is the single-phase amorphous alloy. Regimes (b) and (d) are two-phase mixtures of amorphous
alloy and the terminal solid solution of the major element. The symbols near the bottom of the
figure denote the products obtained by MA of pure Ni and Ti powders: Solid and half-solid
symbols denote single phase and two-phase products, respectively.

a n d where A H r a n d ASf are the e n t h a l p y a n d the e n t r o p y of fusion, respec-


tively.
Little e x p e r i m e n t a l d a t a on the heat c a p a c i t y of u n d e r c o o l e d liquids is
available. F o r the few metallic systems studied, and over the first 200 K of
undercooling, A c e can be satisfactorily d e s c r i b e d by the linear r e l a t i o n s h i p
Ac e = A T + B, where A a n d B are m a t e r i a l - d e p e n d e n t c o n s t a n t s [23,24]. F o r
these materials, AG"-" can be c a l c u l a t e d directly. F o r materials where no
m e a s u r e m e n t s of Acp are available, a p p r o x i m a t e generic expressions have been
derived. The simplest a p p r o x i m a t i o n that can be m a d e is ,:lcp = 0, which leads
to
~G"-C(T) = ,aHf (Tin - T ) / T o , . (4)

This is the oldest linear a p p r o x i m a t i o n [25], which we used in previous


estimates of the free energy of a pure metal [4,5]. Several i m p r o v e m e n t s on eq.
(4) have been p u b l i s h e d [23,26,27]. Recently, Perepezko a n d Paik [24] evaluated
these a p p r o x i m a t i o n s in terms o f m e a s u r e m e n t s of Zlce in u n d e r c o o l e d melts of
Bi, Sn, In, a n d Hg. T h e y c o n c l u d e d that for u n d e r c o o l i n g s of less than 200 K,
286 R.B. Schwarz et al. / Synthesis of amorphous Ni Ti alloy

their data could be best fitted by the expression of Singh and Holtz [27],
derived by assuming a linear Acp(T ) dependence:

AG ..... ( T ) = A H r T m - T 7T
T~ T,, + 6T" (5)

The extrapolation of eq. (5) to Tg, which is outside the range of the
measurements quoted in ref. [24] (0.65 < T / T m < 1), should be approached
with caution. The Tg of amorphous alloys is usually measured by differential
scanning calorimetry. In alloys, Tg is within 40 K of the crystallization
temperature T~. This proximity reflects the fact that Tg and Tx are determined
by similar kinematic constraints related to rates of thermally activated atomic
jumps. It is further observed that for compositions different to those near deep
eutectics, T g / T m = Tx/T,, = 0.5. For example, for rapidly quenched Ni,.Ti I _,
alloys, TJT,1 = 0.512 for x = 0.5, and T J T m = 0.57 for x = 0.33 [28,29].
The Tg values for pure Ni and pure Ti are not known. Amorphous films of
various pure metals have been prepared by condensation from the vapor phase
onto substrates at cryogenic temperatures. However, upon heating, these films
become unstable and crystallize by a diffusionless mechanism at temperatures
of the order of Tin/100. * This value is far below the value of Tg that would be
expected from the increase in the viscosity of a molten metal with increasing
supercooling. An estimate of the Tg of pure metals can be obtained from (1) the
assumption that in the absence of diffusionless crystallization the pure metal
would crystallize at Tx -- Tg and (2) the observation that for amorphous alloys,
T~ is a weak function of composition. All this means that for pure metals, we
should also expect Tg/Tm --- 0.5. Curve (a) in fig. 2 shows AG "-~ according to
eq. (4), which assumes Acp = 0. Curve (b) in fig. 2 shows a better approxima-
tion to AG a-~. For T/T,, > 0,5, AG a-~ was calculated with eq. (5). For
T / T m < 0.5, AG ~-c was calculated from the value of AG ~-c at T = 0.5T m and
using eqs. (1)-(3) with Acp = 0. From this figure it is apparent that for
T = 0.5Tm, eq. (4) overestimates AG "-~ by 15%.
The free energy of the amorphous phase, AGa(x), is shown as a heavy curve
in fig. 1. The end points of this curve were calculated from curve (b) in fig. 2.
The enthalpy of formation of the amorphous alloy was calculated with
Miedema's model [30,31]. This model is based on the empirical observation
that the heat of mixing of binary alloys includes a negative contribution
proportional to the square of the difference in electronegativity between the
two metals, and a positive contribution proportional to the square of the
difference in the electron density at the Wigner-Seitz cell boundaries. The
model differs from regular solution theory in that (a) it includes an estimate of
the composition-dependent changes in atomic volumes caused by electron
transfer and (b) the number of nearest neighbors of a given atom depends not
only on composition but also on the relative sizes of the two atomic species.

* The crystallization temperatures of films of amorphous pure metals have been found to depend
strongly on the purity of these films.
R.B. Schwarz et al. / Synthesis of amorphous Ni- Ti alloy 287

1.0
\MEASUREMENT7
EGIME I~ /
:F
<l
0.5
I Fig. 2. Difference between the free energies of
O
¢,.9
the undercooled and crystalline phases of a pure
metal as a function of the degree of undercool-
irlg. Curve (a) corresponds to equation (4). Curve
(b) for T/Tr. > 0.5 follows from equation (5).
,o Curve (b) for T I T m < 0.5 follows from eq. (5)
T/vm and from eqs. (1)-(3) with Acp = 0.

The entropy of mixing of the amorphous alloy was calculated from regular
solution theory. This neglects the increase in entropy associated with the
destruction of magnetic ordering in Ni upon alloying with Ti.
The thin curves in fig. 1, labeled a-Ni and a-Ti, represent the free energies
of the FCC and H C P terminal solutions. These curves were derived from those
calculated by Saunders and Miodownik [32] for N i - T i alloys at 760 K. Since
our free energy diagram is evaluated at 235 K, we added the term
(760-235)ASm(X), where ZlSm(x ) is the entropy of mixing of a regular
solution. Furthermore, we shifted these curves to take into account the
difference between the energy reference states used in fig. 4 of [32] and in the
present fig. 1. It should be noted that the a-Ni and c~-Ti curves in fig. 1 have
large degrees of uncertainty. In the K a u f m a n n formalism used by Saunders
and Miodownik, thermodynamic data measured near Tm are used to derive
two-parameter descriptions of the free energies of the liquid and crystalline
alloy phases. The a-Ni and c~-Ti curves are extrapolations of this data to
temperatures that are much lower than those at which the numerical fits were
done. This may introduce significant errors [33]. In fig. 1, the free energy
curves of a-Ni and amorphous NixTi I -x intersect at x = 0.25. It will be shown
below that this value agrees rather well with that deduced from the reaction
products obtained through mechanical alloying. This good agreement near
x = 0.25 may be, however, somewhat fortuitous. On the other hand, it is
apparent from fig. 1 that errors in the calculation of the a-Ti curve should have
no effect in the composition ranges of the reaction products predicted by the
free energy diagram.
The dotted, concave curves in fig. 1 represent the free energies of the three
stable crystalline compounds of Ni and Ti [29]. The width of these curves
reflect the solubility of the crystalline phases. For each intermetallic, the
position of the minimum in free energy was calculated as follows. The enthalpy
of crystallization was taken equal to the product of the known melting
temperatures and an assumed entropy of fusion of 2 cal/mol. K. The dif-
ference in free energy between the amorphous and the crystalline phases was
then calculated according to eq. (4).
The two dashed lines in fig. 1 are the common tangents between the free
288 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

Table 1
T h e r m o d y n a m i c p a r a m e t e r s for a m o r p h o u s and crystalline Ni~Ti 1 _ ~ alloys at T = 235 K (Indices
a a n d c d e n o t e a m o r p h o u s and crystalline phases, respectively. Energies are in k c a l / g at.)

Molar Formation Entropic Free energy C r y s t a l l i z a t i o n Calculated M e a s u r e d ¢t


Fraction, e n t h a l p y ~),b) energy a), of e n t h a l p y ~) heat of heat of
of TA Sin, of f o r m a t i o n a~ of f o r m a t i o n a) of
x (Ni~Ti~_x) ~ (NixTi ~ x) ~ (Ni~Ti~_~) ~ (Ni~Ti~ ~)~ (Ni~Ti~_,) ~
0 0 0 -108.7 0 0 0
0.33 -8.68 0.30 -112.5 -2.58 -7.39 -6.40
0.50 -9.44 0.32 -111.5 -3.03 -8.51 -8.10
0.75 - 7.32 0.26 - 105.8 - 3.30 - 6.54 - 8.30
1.00 0 0 - 94.8 0 0 0

~) C a l c u l a t e d a c c o r d i n g to M i e d e m a ' s model [30,31].


b) For the reaction x N i a + ( 1 - x ) T i a ---, (NixTi 1 x) a.
~) For the reaction ( N i , T i l_~,)a __, (Ni~Til ,~)~.
d) F o r the reaction x N i c + ( 1 - x ) T i ~ ~ (NixTi 1 :,)~.
~) F r o m ref. [34].

energies of the crystalline terminal solutions and of the amorphous phase. The
points of contact define four (metastable) reaction products, labeled a, b, c and
d, expected in the absence of crystalline alloy formation.
The solubility limit of a-Ni depends on whether crystalline Ni 3Ti is present
or not. In the absence of c-Ni3Ti, the predicted solubility limit of a-Ni is
approximately 20%. In the presence of c-Ni3Ti, a common tangent between the
a-Ni curve and the dotted curve for c-Ni3Ti (tangent not shown in fig. 1)
would predict a solubility limit for a-Ni of only a few per cent, in agreement
with the value suggested by the binary phase diagram [29]. On the other hand,
the solubility limit of a-Ti in equilibrium with either the amorphous alloy or
with c-NiTi 2 is very small. These differences may play an important role in the
kinetics of formation of the amorphous phase at the boundary between pure
crystalline Ni and Ti (see section 5.2).
Table 1 contains thermodynamic values for amorphous NixTi ~-x evaluated
at x = 0, 0.33, 0.5, 0.75 and 1, corresponding to the pure elements and to the
stable intermetallic compounds NiTi z, NiTi, and Ni3Ti. It is apparent from
this table that the main contribution to the free energy of formation of the
amorphous phase comes from the enthalpy of mixing. The fifth column in
table 1 gives the enthalpies of crystallization of NixTil_ x calculated as the
product of the known melting temperature of the crystalline compounds [29]
and an assumed entropy of crystallization of 2 cal/mol. K. The sixth column
in table 1 gives the heats of formation of crystalline NixTil_ x starting from
pure crystalline Ni and Ti. These values were calculated by decomposing the
reaction as follows:
xNi c + ( 1 - x ) T i c ~ xNi a + ( 1 - x)Ti a, (6)
xNi ~ + (1 - x ) T i ~ ~ (Ni~Ti 1 ~)a, (7)
( NixTi~-x)" ~ ( Nixria x)", (8)
R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy 289

where the superscripts a and c denote amorphous and crystalline phases,


respectively.
The enthalpy change for the reaction in eq. (6) is approximately x H f ( N i ) +
(1 - x)Hf(Ti), where H r are the heats of fusion. The enthalpy changes for the
reactions in eqs. (7) and (8) are given in columns 2 and 5 of table 1,
respectively. Adding these heats of reaction we derived the values in column 6
of table 1.
Column 7 in table 1 gives the heats of formation of the crystalline inter-
metallics measured by Kubaschewski [34]. It should be added that Hultgren et
al. [35] attach considerable uncertainty to these data. Because of this, and in
view of the approximations used here, the differences between the values of
columns 6 and 7 are not excessive.

4. Results

4.1. Morphology of the mechanical alloying process

Two distinct stages were observed during the MA of the Ni and Ti powders.
During the first couple of hours, the powder adhered strongly to the surfaces
of the balls and to the walls of the vial, forming a hard coating with a rough
orange-peel texture. As the milling progressed, this coating fractured off. An
indication of the completion of the MA process and of the attainment of a
single-phase amorphous powder was the ease by which the powder could be
removed from the balls. Further MA caused the alloyed powder to form an
agglomerate at one end of the cylindrical vial. However, this product had the
consistency of chalk and could be removed with a spatula.
Intermediate stages of processing were also observed by Benjamin and
coworkers [15,16] during the MA of Ni-based spueralloys. In their work it was
recognized that the formation of a coating on the steel balls is advantageous
since it prevents excessive wear of the balls. But it was further suggested that
this coating be kept to a minimum in order to avoid a heterogeneous final
product. For the present amorphization process, the formation of coatings on
the ball surfaces appears to be a desirable feature since it promotes the
formation of a multilayer system of alternating Ni and Ti bands, similar to the
thin films previously obtained by SSAR.
Fig. 3 is a scanning electron micrograph of a sectioned particle after 5 h of
MA. The scanning images of dispersely scattered X-rays show that in fig. 3 the
dark bands are pure Ni, the light ones are pure Ti, and the gray areas are
amorphous NixTil_x, which is the only other phase present. The scanning
electron micrographs show that the refinement of the layered microstructure is
accompanied by successive fracturing and welding of particles. The welding is
most likely to occur between clean Ni and Ti surfaces, since these have a
strong chemical affinity, as indicated by their large negative heat of mixing. A
similar affinity exists between Fe and Ti and this may favor the initial
290 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

Fig. 3. Scanning electron micrograph of a sectioned particle after 3 h of mechanical alloying. The
light (dark) bands correspond to crystalline Ni (Ti). The gray areas are amorphous NiTi alloy
which, according to the X-ray diffraction pattern, is the only other phase present.

adherence of the Ti powder to the balls and vial. The repeated impact of the
balls causes the ductile Ni and Ti to plastically deform in directions tangential
to the ball surfaces, generating the narrow bands seen in fig. 3. As the Ni and
Ti become alloyed and the material work-hardens, the tendency for cold
welding decreases, while the rate of fracture increases. This results in the
peeling off of the coatings. Some further refinement of the powder size and
microstructure occurs after this stage.

4.2. X-ray diffraction patterns

X-ray diffraction patterns of the powders were taken with a horizontal G E


diffractometer and M o - K , radiation. The powder samples were held in rectan-
gular holders covered by a thin paper whose scattering contributions were later
subtracted. The attenuation of the direct beam by the paper was less than 3%.
The Bragg-Brentano geometry was employed with 0.4 ° incident and 0.2 °
diffracted slits and a vertical diffracted-beam collimation. The diffracted
intensity was measured with a Si(Li) detector. The background correction was
determined by replacing the sample with a lead " b l i n d " box covered by the
same thin paper.
Curve (a) in fig. 4 shows the X-ray diffraction pattern of a mixture of
R.B. Schwarz et aZ / Synthesis of amorphous N i - Ti alloy 291

NixTil.x
I [ i T l q T [ T 1

8 Ni33 Ti67 / ~ ~ X= 010

C
o
o z ~ ~ -- z --O4 -- ,~- X= 0:52
i
z
Ld
j x=oTo
H

X = 0.80
! j --~-~- ,

2 3 4 5 6 7 3 4 5 6 7 8 9 i0
K (.~-I) K (,~-I)

Fig. 4. X-ray diffraction intensity as a function of wavenumber K for mechanically alloyed Ni and
Ti powders. Curves (a) and (b) were obtained after 2 and 5 h of MA, respectively. Curve (c) was
obtained from the powder of curve (b) after thermal annealing for 5 h at 578 K.

Fig. 5. X-ray diffraction intensity for mechanically alloyed Ni~Ti 1 • powders as a function of
wavenumber K and composition x.

crystalline Ni and Ti powders after 2 h of MA. The wavenumber K =


4~r sin 0/)~. Curve (b) in fig. 4 shows the diffraction pattern of the same
powder after an additional 3 h of MA. With increasing M A time, the intensity
under the Bragg peaks is replaced by that under the fewer broad maxima of the
amorphous phase. Curve (c) is the diffraction pattern of the powder of curve
(b) after an anneal in Ar gas for 5 h at 578 K. This anneal enhances the
crystalline-to-amorphous transformation but is insufficient to complete the
reaction.
Fig. 5 shows X-ray diffraction patterns of powders that were mechanically
alloyed for 14 h. For x = 0.1 and 0.2, the patterns show broad primary and
secondary maxima corresponding to the amorphous phase. Superimposed on
these are small Bragg peaks of HCP Ti. For x = 0.32 and x = 0.7, the powders
are single phase amorphous. Similar X-ray patterns were obtained for the
compositions x = 0.4, 0.5, and 0.6. These are not shown in fig. 5. For x = 0.75,
the X-ray pattern indicates a mixture of an amorphous phase and small
crystallites. The Bragg peaks of the crystallites correspond to the intermetallic
Ni3Ti with hexagonal DO24 structure [29]. For x = 0.8, the material is again a
mixture of amorphous and crystalline phases but the Bragg peaks now corre-
spond, as expected from fig. 1, to the F C C Ni-rich terminal solution. The
292 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

lattice parameter of the terminal solution was determined through a


Nelson-Riley plot to be a = 3.62 A, which is 2.7% larger than that for pure Ni,
a = 3.524 ,~. This expansion can be used to estimate the solubility limit of Ti in
a-Ni in the mechanically alloyed powder. For 1.0 > x > 0.96, the lattice
parameter of FCC NixTi ~_~ increases at the rate (1/a)da/dc = 0.092 +_ 0.007
[36,37]. From a linear extrapolation of these measurements it follows that the
solubility limit of a-Ni in the mechanically alloyed powder is 28 at,%. The
solubility limit predicted by the free energy diagram is 20 at%. This contrasts
with the solubility limit in the presence of crystalline Ni3Ti, which is less than 5
at.% [29].
The reaction products identified by the X-ray patterns are shown schemati-
cally at the bottom of fig. 1. A solid symbol denotes a single-phase amorphous
product; a half-solid symbol denotes a two-phase product. As seen in fig. 1, the
products obtained by MA show good agreement with the predictions of the
free-energy diagram, except for x = 0.75. For Ni0.vsTi025, part of the alloyed
powder (less than 10%) shows polymorphous crystallization into Ni3Ti [HCP,
D024 structure]. We notice that this compound is not formed at either x = 0.70
or x = 0.80. In contrast with the behavior at x = 0.75, the X-ray patterns for
x = 0.50 and x = 0.32 do not show crystalline NiTi [BCC, CsCl-type structure]
or NiTi 2 [FCC, Cu 2Mg-type structure], respectively. These observations may
relate to the thermal stability of the amorphous phase. A measure of the
thermal stability is the crystallization temperature TX and the activation energy
of crystallization, A E. Measurements of Tx and A E in rapidly quenched
NixTi~ x are available for 0.23 < x < 0.64 [28] but not for the composition
x = 0.75 of interest here. Preliminary measurements of Tx and AE in mechani-
cally alloyed Ni32Ti68 are presented in section 4.4. Measurements at other
compositions are in progress.

4. 3. Atornic pair distribution functions

Details on the analytical techniques for the generation of the radial distribu-
tion functions, G(r), are found in refs. [38] and [39]. The absorption correction
to the X-ray data was essentially constant, since the focusing technique was
employed [38]. Corrections for incoherent scattering were made with values
listed in the International Tables of Crystallography (ITC), Vol. 3. Since
neither incident nor diffracted monochromators were used, the polarization
factor is simply (1 + cos(20))/2. Multiple scattering was assumed to be small
and no correction was made. The atomic scattering factors and the anomalous
dispersion factors were also taken from the ITC. The high-angle normalization
method was employed for 12 < K < 14. The diffracted intensity was attenuated
exponentially for K > 9 in order to dampen out the so-called truncation
oscillations that appear in G(r) as a result of a Fourier transformation over
finite K-space [38]. This only produced small differences in G(r) for r < 2
[below the first peak in G(r)].
Fig. 6 shows G(r) curves for mechanically alloyed amorphous Ni,Ti 1 x.
R.B. Schwarz et al. / Synthesis of amorphous Ni- Ti alloy 293

NixT,-x ~ ~ - / ~ T , q~ '-
L] --Ni4oTi6o
x=0.7 (MECHANICALLYALLOYED,
PRESENT WORK) !
---Ni4oTi60 i
t~(RAPIDLY QUENCHED, /
I, IWAGNER AND LEE ~980]) !
(D
,

. . . . . . J

0 2.0 4.0 6 0 80 I0.0 0 2 4 6 8 I0


r(~,) r (A)
Fig, 6, Reduced (total) atomic pair distribution functions of mechanically alloyed amorphous
Ni~Ti1 ~, for0.3<x~<0.7.

Fig, 7. Reducedatomic pair distribution functionsof amorphousNi40Ti6osynthesizedby the rapid


quenching technique[40] and by MA.

With decreasing x(increasing Ti content), the G(r) curves, as a whole, shift to


larger r, as indicated by the slanted line L 1 passing through their second
maximum. This reflects the expected dilation of the structure with increasing
Ti content. In contrast to this, the shoulder on the right side of the second
maximum has a fixed position at r = 5.1 A (Line L2). This shoulder corre-
sponds to an interatomic distance characterizing a group of atoms that do not
change with average composition. We consider this as direct evidence of
chemical short range ordering (CSRO) in the amorphous structure.
Fig. 7 compares the G(r) curves of amorphous Ni40Ti60 prepared by MA
(present work) and by rapid quenching from the melt [40]. The two G(r)
curves are very similar. However, the G(r) for the rapidly quenched alloy has
somewhat sharper features. To investigate whether this was due to a relaxation
in the amorphous state for the rapidly quenched alloy but not the MA powder,
we measured the G(r) of the MA amorphous Ni32Ti6s before and after a
thermal anneal of 3 h at 553 K. The two G(r) curves were similar, except that
the shoulder at the right of the second maximum decreased in size without
changing its position. Another possible explanation for the difference in
sharpness is that the G(r) reported in ref. [40] was deduced from measure-
ments of I ( K ) extending to values larger than K = 14, the maximum K used
in the present experiments.
The shoulder on the right side of the second maximum in the G(r) curve of
mechanically alloyed NiTi, fixed at r = 5.1 .& (figs. 6 and 7), is also visible in
the G(r) curve of the rapidly quenched Ni40Ti60 (fig. 7). It is thus apparent
that the same CSRO is present in both amorphous alloys, independent of the
method of synthesis. Even though the partial reduced atomic pair distribution
functions of rapidly quenched Ni40Ti60 have been measured [40,41], it is not
294 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

simple to deduce - based solely on this information - the atomic arrangement


that gives rise to this shoulder. The observation that the only crystalline
intermetallic detected in the mechanically alloyed powders was Ni3Ti (for
x = 0.75, see fig. 5) suggests that atomic arrangements typical of this structure
may also exist at other compositions, but in the form of CSRO. Indeed, Ni3Ti
has a hexagonal D024 structure (a = 5.101, c = 8.307 ,~) in which the basal
plane is a dense packing of Ni and Ti atoms. The 5.1 ,a, distance equals the
separation between both N i - N i and T i - T i atoms pairs along the {01.0}
directions in the basal (00.1) plane.

4.4. Thermal stabifity of amorphous Ni 32Ti,~a

A complete study of the thermal stability of mechanically alloyed amorphous


NixTil x as a function of composition will be reported separately. Preliminary
measurements are presented here for Ni32Ti68 as supporting evidence for the
amorphicity of the alloyed powder.
The solid curve in fig. 8 shows the heat supplied to amorphous Ni32Ti68
powder for a constant heating rate of 80 K / m i n , obtained with a differential
scanning calorimeter (DSC) [42]. The instrument was calibrated at different
heating rates with lead and K2SO 4 standards. The measurements were made in
an Ar flow of 30 cm3/min. The powder was first annealed in the calorimeter at
500 K for 15 min. This was done to remove water vapor and other gases that
may have adsorbed onto the powder during its loading into the calorimeter.
From 500 to 990 K the DSC trace showed only one exothermic peak. An X-ray
diffraction pattern of Ni32Ti68 powder which was heated to a temperature just
above the crystallization peak showed 28 Bragg lines, all of which were
identified as belonging to NiTi 2. On further heating of this powder to 950 K,
the diffraction lines became sharper but no new Bragg lines were detected. The
lack of other exothermic peaks in the DSC trace and the absence of crystalline
compounds other than NiTi 2 shows that the mechanically alloyed amorphous
Ni32Ti68 crystallizes polymorphously into NiTi 2. This is in agreement with the
free energy diagram (fig. 1) and with the observations of Buschow [43] on
rapidly quenched Ni35Ti65. It is, however, in disagreement with the measure-
ments of Ruppersberg et al. [44] in rapidly quenched Ni35Ti65, for which at
least two exothermic peaks were found.
For a powder that was mechanically alloyed for 17 h and heated in the DSC
at a rate of 20 K / m i n , the onset of crystallization is at (712.5 -T-0.5)K. DSC
traces similar to that in fig. 8 were obtained at heating rates of S = 10, 40, 80,
and 160 K / m i n . The dependence of Tx on S was used to derive an apparent
activation energy of crystallization, AE, for amorphous Ni32Ti68. In the
method of Kissinger-Boswell [45,46] the quantity A E is derived from a plot of
In (STx 1) versus Tx l, as shown in fig. 9. From the slope of the least-squares-
fitted line it follows that A E = 65.4 kcal/g at. The heat of crystallization of the
amorphous Ni32TI68 powder is obtained from the area defined by an extrapo-
lation of the baseline and the exothermic event in the DSC trace. The average
R.B. Schwarz et aL / Synthesis of amorphous N i - Ti alh)v 295

ixlO -3
20

Ni32 Ti68
HEATING RATE 80 K/mil
0 ix~O-'~ I
600 64O 68O 720 76O 800 1.36 1.38 1.40 1.42 1.44
TEMPERATURE (K) IO00/T x ( K -I)

Fig. 8. Heat supplied to mechanically alloyed amorphous Ni32Ti6s to maintain a heating rate of 80
K / m i n . The arrows indicate heating and cooling.

Fig. 9. Plot of In (SZ'~-1) versus (T~-l). T~ was obtained from differential calorimetry tests at
different heating rates S.

of three measurements at S = 20 K / m i n gives A H~r = (0.74 + 0.02) kcal/g at.


The thermal properties of our mechanically alloyed Ni32Ti68 are similar to
those of rapidly quenched Ni30Tivo, measured by Buschow [43]. Using a
similar calorimeter, he obtained T~ = 716 K (for S = 50 K/min), AE = 66.5
k c a l / g at., and AHcr = 0.82 kcal/g at. The similarity in both the thermal
stability and in the G(r) curves of mechanically alloyed and rapidly quenched
amorphous Ni32Ti6s suggests that, at least near this composition, the atomic
structures of the alloys prepared by these two techniques are essentially the
same. However, the value of AHcr measured by Buschow and by us is a factor
of three lower than that deduced in table 1, column 5. Further measurements
(to be reported separately) revealed that this discrepancy is caused by the
neglect of two exothermic processes which take place during the heating of the
powder and over broad T regimes: From 400 to 720 K (below Tx) there is a
heat release which we attribute to atomic rearrangements in the amorphous
phase, leading to chemical short range ordering. From 760 to 900 K (above Tx )
heat is released because of grain growth and reduction in the dislocation
density of the crystalline phase. When these slowly released heats are properly
accounted for, the measured (total) heat release and the calculated crystalliza-
tion enthalpy quoted for x = 0.33 in column 5 of table 1 are within 15%. It is
concluded that the area defined by the exothermic peak in fig. 8, which is
customarily identified with the crystallization enthalpy, AHcr, is only part of
the difference in the heat contents of the truly amorphous and the fully
crystallized phases.
296 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

5. Discussion

Two mechanisms were advanced in the Introduction as possible explana-


tions for the amorphization of powders by MA. These mechanisms are
discussed next in terms of the present experimental results.

5.1. Incompatibility with amorphization by rapid solidification from the melt

From a thermodynamic point of view, the amorphous phase of a binary


alloy of two transition metals (TMs) is metastable over a wide composition
range centered near the equiatomic composition [4,5] (see also fig. 1). On the
other hand, amorphous metallic alloys of two TMs prepared by the rapid
quenching of melts usually have compositions near that of deep eutectics in the
equilibrium binary phase diagram. The restricted composition range observed
with this technique should be attributed to the kinetic constraints associated
with the required fast cooling of the melt to a temperature below the crystalli-
zation temperature of the amorphous phase. The typical thicknesses of
amorphous ribbons fabricated by the melt-spinning technique is 30-50 /~m,
which is close to the 50/~m diameter of the present powder particles. Thus, the
thermal relaxation time for a particle trapped between two colliding steel balls
should be of the order of that for the ribbon in contact with the spinning
wheel.
Fig. 1 predicts the homogeneity range of amorphous NixTil_ x to be
0.28 < x < 0.72. This range compares favorably with that of the present
mechanically alloyed powders, 0.3 < x < 0.7. Several authors have synthesized
amorphous Ni:,Til_~ with x < 0.46 by the rapid quenching of melts. The
reported homogeneity ranges are 0.3 < x < 0.4 [47], 0.25 < x < 0.42 [48], and
0.23 < x < 0.46 [28,43]. In addition, Buschow [28,43] has been able to syn-
thesize amorphous alloys with 0.55 < x < 0.64. Rapidly-quenched amorphous
NixTi~_ ~ has not been reported for 0.46 < x < 0.55 (near the congruent
melting at x = 0.5), nor for 0.64 < x < 0.72 (near the congruent melting at
x = 0.75). On the other hand, amorphous powders are readily formed by MA
at both x = 0.5 and x = 0.70. This suggests that amorphization by MA does
not involve the rapid solidification of melts. A comparison of the pressures
achieved at the interfaces of the colliding balls with that known to develop
during the shock-wave consolidation of powders will further show that the
maximum tempei'ature reached by the particles during MA is well below the
melting temperature.
A shock wave traversing a powder dissipates energy irreversibly. During the
short duration of the shock wave rise time, intense localized plastic deforma-
tion near particle boundaries and interparticle friction convert part of the
shock energy into heat, which is dissipated heterogeneously. There is metal-
lurgical evidence that a sufficiently large shock pressure gives rise to localized
melting, preferentially at interparticle boundaries [17], while leaving the par-
ticle interiors at a much lower temperature. The formation of melt pools at the
R.B. Schwarz et al. / Synthesis of amorphous Ni- Ti alh~v 297

shock front is followed by the rapid solidification and cooling of these melts by
heat conduction into the particle interiors [49]. With appropriate shock param-
eters [50], the method can be used to consolidate amorphous metal powders
[49,51,52] which would be crystallized during conventional high-temperature
sintering. The energy deposited into the powder at the shock front is linearly
proportional to the shock pressure. It has been experimentally found in studies
of shock compaction of steel-base powders that the onset of interparticle
melting occurs at shock pressures of a few GPa [17]. These pressures are
achieved with steel projectiles impinging on a stationary powder at speeds of
the order of 1 k m / s . Because the shock pressure is a monotonically increasing
function of the particle velocity [53], and since the relative velocities achieved
by the colliding balls during MA are only 2 m / s (see section 2), we conclude
that the peak pressures obtained during MA are two to three orders of
magnitude below those needed for melting. Furthermore, since melting did not
occur, the peak temperature reached by particles trapped between colliding
balls must be below the crystallization temperature of the amorphous phase.
Otherwise the powder would have crystallized.

5.2. Compatibility with a solid state amorphizing reaction (SSAR)

At an intermediate stage in the MA of pure Ni and Ti powders, the partially


alloyed particles have a "swirl" structure formed by alternating crystalline
layers of the pure starting elements (fig. 3). The interfaces between these layers
are very clean, since the reactive gases that may have been initially present in
the vial were gettered by the first atomically clean Ti surfaces to be formed.
The extensive plastic deformation sustained by the particles generates in them
a high density of dislocations and point defects [54]. In addition, the tempera-
ture of the trapped particles is raised momentarily above that of the balls and
vial. Because these conditions favor interdiffusion, it is likely that the mechani-
cal formation of clean Ni-Ti interfaces is followed by a solid state reaction of
the type observed in thin films [4,5], leading to the formation of a thin band of
amorphous alloy.
Evidence that the N i / T i interfaces created by MA can react by interdiffu-
sion to form amorphous alloy is provided by the X-ray patterns of fig. 4. These
show that the volume fraction of the amorphous phase increases and those of
the crystalline phases decrease when the MA process is stopped and the
powder is given a thermal anneal at a temperature below T,. A combination of
deformation and long thermal anneals has been recently applied to both
powders [55] and rolled foils [56] to obtain thin alloy foils and wires, respec-
tively, which are partially amorphous. The aim of the present work was to
study the amorphization by MA alone, and thus the MA process was per-
formed with the vial kept at a constant temperature.
Few details are known of the nucleation and growth morphology of the
amorphous phase during SSAR. However, significant insight can be obtained
from the free energy diagram of fig. 1. Consider two large crystals of pure Ni
298 R.B. Schwarz et al. / Synthesis of amorphous Ni- Ti alloy

and Ti placed in contact at a temperature lower than the crystallization


temperature of amorphous NiTi. The free energy of this couple decreases if
interdiffusion can take place. If this occurs in the absence of either amorphous
or crystalline alloy phases, the terminal solubilities are given by a common
tangent traced between the a-Ni and the a-Ti curves in fig. 1. This predicts
that the solubility of Ti in FCC Ni is very large (approximately 65 at.%) while
that of Ni is HCP Ti is extremely small. However, the formation of FCC
NixTil -x alloy with (1 - x) > 0.2 is not likely to occur because of the nuclea-
tion and growth of an amorphous phase, which decreases the free energy of the
system even more. The nucleation of the amorphous phase should require a
low activation energy because for ( 1 - x ) > 0.2 it can take place polymor-
phously (no need for long range diffusion) and in regions of the a-Ni lattice
that are already disordered, such as dislocations and grain boundaries. This is
in contrast to the nucleation of crystalline intermetallics which, on account of
their higher degree of order, should require a larger activation energy.
Several interdiffusion mechanisms which operate at T < Tx need to be
considered. In an earlier work [4] we proposed that the fast mixing of Au and
La films observed at 70°C, leading to the formation of amorphous A u - L a
alloys films, occurs because of the anomalously fast diffusion of Au in La. No
specific microscopic model was proposed for this reaction. Because Ni is an
anomalously fast diffuser in Ti [57], a similar mechanism may be operating
here. However, since the solubility of Ni in a-Ti is very low, the amorphous
alloy cannot be formed polymorphously and thus, for this mechanism to
operate, the Ni must precipitate at already nucleated amorphous alloy. Recent
works suggest that additional diffusion mechanisms may be involved in the
SSAR, especially during the onset of the reaction. These will be discussed next.
The interdiffusion of Ni and Ti at temperatures below Tx may occur
through short circuit paths such as dislocations and grain boundaries. Consider
first the diffusion of Ti along lattice defects in FCC Ni. The grain boundary
diffusion of Ti in a-Ni is not known but it has been estimated that the
diffusion of Ti along grain boundaries in Au, Pd, and Pt is about 103 times
faster than that in the bulk [58]. Thus, the grain boundary diffusion of Ti in
a-Ni should be similarly high and this should lead to interdiffusion and to
nucleation and growth of the amorphous phase. It is important to notice that
in this mechanism the amorphous phase forms as the result of the diffusion of
Ti (the bigger atom) into Ni (the smaller atom). There is some evidence that
this type of reaction does indeed take place at the interfaces between an
early-TM and a late-TM. SchriSder et al [13] used cross-sectional transmission
electron microscopy to study the SSAR in multilayer films of C o - Z r annealed
at 210°C. They observed that after about 2 h of annealing, the amorphous
layers formed at the C o - Z r interfaces had reached a thickness of 10 nm and
had the composition Cos0Zr20. Because the C o - Z r system is characterized by a
large negative heat of mixing, the free energy diagram for this system is similar
to that for Ni-Ti, shown in fig. 1. Because (a) the composition CosoZr20 is
near the Co end of the solubility range of the amorphous phase and (b) the free
R.B. Schwarz et al. / Synthesis of amorphous Ni- Ti alloy 299

energy of formation of amorphous C080Zr20 is certainly less than that of the


equimolar amorphous alloy, it is likely that the C080Zr20 amorphous layer
forms as the result of Zr (larger atom) diffusion into o~-Co (smaller atom).
Since the reaction temperature is low (210°C), the diffusion occurs most likely
along dislocations and grain boundaries, as proposed above. Furthermore, if
the amorphous alloy nucleates first at grain boundaries in c~-Co, the width of
the initial amorphous layer should be of the order of the grain size in c~-Co.
Indeed, the thickness of the amorphous Co80Zr20 layer, 10 nm, is of the order
of the grain size in the (annealed) Co film [13].
A SSAR mechanism similar to that described above may take place by
diffusion of the smaller atom (Ni or Co) along defects in the lattice of the
larger atoms (Ti or Zr). One difference, though, is that solutes smaller than
lattice atoms are attracted to the compressed regions of edge dislocations, and
this may not favor a fast diffusion. Further TEM work on sectioned samples is
needed to verify these processes.
Following the formation of amorphous alloy by diffusion along lattice
defects, the SS,AR must necessarily involve diffusion of either Ni or Ti across
the amorphous alloy already formed [59]. In the case of SSAR in thin films,
this should slow down the reaction kinetics [60]~ However, in the present case
of the amorphization of powders by MA, the amorphous layers formed by the
SSAR are continuously broken and dispersed by the fracturing of the particles.
This creates fresh N i / T i interfaces at which SSAR can take place anew.
Furthermore, the interdiffusion is most likely enhanced by the stress-assisted
migration of the lattice and point defects created by plastic deformation.
In conclusion, the SSAR in the present MA powders may occur by any of
the following mechanisms, or combinations thereof: (a) diffusion of Ti through
lattice defects in c~-Ni followed by a polymorphous nucleation of amorphous
Ni72Ti28, (b) nucleation and growth of an amorphous layer at the N i / T i
interfaces, followed by diffusion of Ni a n d / o r Ti across this layer [59], and (c)
anomalously fast diffusion of Ni through o~-Ti into already nucleated
amorphous alloy [4]. TEM observations [13] suggest that mechanism (a) is the
first to operate.
It should not be surprising that dislocations and grain boundaries play an
important role in the initiation of SSAR because these regions are known to
attract solutes and, in some cases, act as nucleation centers for the precipita-
tion of crystalline phases in supersaturated solid solutions [61]. It has also been
observed that grain boundaries and dislocations act as nucleation sites for the
growth of amorphous NiTi in electron-irradiated crystalline NiTi [62].
One of the requirements for the SSAR is a large negative heat of mixing
between the two elements in the amorphous phase [4,5]. The binary systems
that have been amorphized by MA to date, namely, NiNb [8], the present
NiTi, and GdNi and AgGd [63], fulfill this requirement. We have also used the
MA method to prepare Cu55Ni45 alloy powder (constantan) starting from pure
Cu and Ni powders [64]. Cu and Ni have a positive heat of formation [18].
X-ray diffraction patterns taken as a function of MA time show a smooth
300 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti allc~'

transition from a mixture of FCC Cu and Ni to an FCC Cu55Ni45 alloy. These


few known examples suggest that the requirement of a negative heat of mixing
for SSAR [4,5] also applies to amorphization by MA.

6. Conclusion

Amorphous Ni,Ti 1_ ~ alloy powders synthesized by MA of pure crystalline


Ni and Ti powders have a homogeneity range of 0.3 < x < 0.7. This range is
broader than that measured in amorphous alloys prepared by the rapid
solidification of melts. The homogeneity range of the mechanically alloyed
amorphous powders corresponds to that predicted by a free-energy diagram
for the crystalline terminal solutions and the amorphous NixTil_.,. alloy,
constructed from known thermodynamic data and Miedema's derivation of the
heat of formation of the amorphous alloy.
Amorphous Ni,Ti 1-x powders with x -- 0.35 prepared by the rapid quench-
ing of melts and by mechanical alloying have the same atomic structure and
thermal properties. This is evidenced by the identities of their reduced total
atomic pair distribution functions, G(r), crystallization temperatures, crystalli-
zation enthalpies, and apparent activation energies for crystallization.
The shoulder on the right side of the second peak in G(r), seen in
amorphous NiTi alloys prepared both by MA and by the rapid quenching of
melts, is attributed to CSRO in the amorphous phase. It is suggested that this
order consists of thetrahedra of Ni and Ti atoms in an arrangement similar to
that found in ther basal plane of HCP Ni3Ti.
The peak temperature reached by the particles trapped between colliding
steel balls is lower than the crystallization temperature of the amorphous
phase. The amorphous alloy is formed by mechanical mixing and by a solid
state interdiffusion reaction near clean boundaries between polycrystalline Ni
and Ti. The mixing is enhanced by the diffusion of Ti through dislocations and
grain boundaries in c~-Ni and by the migration of lattice and point defects
generated by plastic deformation. Furthermore, the ball milling disperses the
amorphous material already created by interdiffusion and brings yet unalloyed
Ni and Ti into contact.
The solubility limit of Ti in a-Ni following deformation by ball milling at
low temperatures is approximately 0.25, which is many times larger than the
solubility limit in the presence of crystalline Ni3Ti. This suggests that MA can
be used to manufacture crystalline super saturated alloy powders.
The authors thank Dr J. Faber Jr for the use of his X-ray diffractometer and
Professor C.N.J. Wagner for providing us with the computer program for
generating the atomic pair distribution functions.

References
[1] P. Duwez, R.H. Willens and W. Klement Jr, J. Appl. Phys. 31 (1960) 1136.
[2] J.J. Hauser, J. de Phys. Colloq. 42 (1981) C4-943.
R.B. Schwarz et al. / Synthesis of amorphous Ni-Ti alloy 301

[3] S.R. Herd, K.N. Tu and K.Y. Ahn, Appl. Phys. Lett. 42 (1983) 597: Thin Solid Films 104
(1983) 197.
[4] R.B. Schwarz and W. Johnson, Phys. Rev. Lett. 51 (1983) 415.
[5] R.B. Schwarz, K.L. Wong, W.L. Johnson and B.M. Clemens, J. Non-Crystalline Solids 61-62
(1984) 129.
[6] R.L. White, PhD Dissertation, Stanford University (1979).
[7] A.E. Ermakov, E.E. Yurchikov and V.A. Barinov, Fiz. Metal. Metalloved. 52 (1981) 1184; 54
(1982) 935.
[8] C. Koch, O.B. Cavin, C.G. McKamey and J.O. Scarbrough, Appl. Phys. Lett. 43 (1983) 1017.
[9] S.K. Malik and W.E. Wallace, Solid St. Commun. 24 (1977) 283.
[10] X.L. Yeh, K. Samwer and W.L. Johnson, Appl. Phys. Lett. 42 (1983) 242.
[11] B.M. Clemens, W.L. Johnson and R.B. Schwarz, J. Non-Crystalline Solids 61-62 (1984) 817.
[12] M. van Rossum, M.-A. Nicolet and W.L. Johnson, Phys. Rev. B29 (1984) 5498.
[13] H. SchrOder, K. Samwer and U. K~3ster, Phys. Rev. Lett. 54 (1985) 197.
[14] J.S. Benjamin, Sci. Ame. 234 (1976) 40.
[15] J.S. Benjamin and M.S. Bamford, Metall. Trans. A8 (1977) 1301.
[16] P.S. Gilman and J.S. Benjamin, Ann. Rev. Mater. Sci. 13 (1983) 279.
[17] D.G. Morris, Met. Sci. 15 (1981) 116.
[18] A.R. Miedema, Philips Technical Review 36 (1976) 217.
[19] D. Ablitzer, Phil. Mag. 35 (1977) 1239.
[20] Johnson Matttley Inc., Seabrook, NH 03874.
[21] Model 8000 from Spex Industries, Edison, NJ 08820.
[22] R.C. Weast, CRC Handbook of Chemistry and Physics, 55th ed. (Chemical Rubber Co.,
Cleveland, 1974) p. D-55.
[23] C.V. Thompson and F. Spaepen, Acta Metall. 27 (1979) 1855.
[24] J.H. Perepezko and J.S. Paik, J. Non-Crystalline 61-62 (1984) 113.
[25] D. Turnbull, J. Appl. Phys. 21 (1950) 1022.
[26] J.D. Hoffman, J. Chem. Phys. 29 (1958) 1192.
[27] H.B. Singh and A. Holz, Solid St. Commun. 45 (1983) 985.
[28] K.H.J. Buschow, J. Appl. Phys. 56 (1984) 304.
[29] M. Hansen, Constitution of Binary Alloys, 2nd ed. (McGraw-Hill, New York, 1958).
[30] A.R. Miedema, J. Less-Common Metals 32 (1973) 117; 46 (1976) 67.
[31] A.R. Miedema, R. Boom and F.R. DeBoer, J. Less-Common Metals 41 (1975) 283.
[32] N. Saunders and A.P. Miodownik, Ber. Bunsenges. Phys. Chem. 87 (1983) 830.
[33] P. Nash, personal communication (1985).
[34] O. Kubaschewski, Trans. Faraday Soc. 54 (1958) 814.
[35] R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser and K.K. Kelley, Selected Values of the
Thermodynamic Properties of Binary Alloys (American Society for Metals, Metals Park,
Ohio, 1973).
[36] T.H. Hazlett and E.R. Parker, Trans. ASM 46 (1954) 701.
[37] D.M. Poole and W. Hume-Rothery, J. Inst. Met. 83 (1954) 473.
[38] G.S. Cargill lII, in: Solid State Physics, Vol. 30 eds., F. Seitz, D. Turnbull and H. Ehrenreich
(Academic Press, New York, 1975) p. 227.
[39] C.N.J. Wagner and H. Ruppersberg, Atomic Energy Review. Suppl. No. 1 (International
Atomic Energy Agency, Vienna, 1981).
[40] C.N.J. Wagner and D. Lee, J. de Phys. 41 (1980) C8-242.
[41] T. Fukunaga, N. Watanabe and K. Suzuki, J. Non-Crystalline Solids 61-62 (1984) 343.
[42] Model DSC 2, Perkin Elmer, Norwalk, CT 06856.
[43] K.H.J. Buschow, J. Phys. F13 (1983) 563.
[44] H. Ruppersberg, J. Phys. F14 (1984) 323.
[45] H.E. Kissinger, Anal. Chem. 29 (1957) 1702.
[46] P.G. Boswell, J. Thermal Anal. 18 (1980) 353.
[47] D.E. Polk, A. Calka and B,C. Giessen, Acta Metall. 26 (1978) 1097.
302 R.B. Schwarz et al. / Synthesis of amorphous N i - Ti alloy

[48] A. Calka, E. Blachnia and H. Matyja, in: Metallic Glasses: Science and Technology, Vol. 2.
Proc. Intl. Conf. held in Budapest (June 30-July 4, 1980) eds., C. Hargitai, 1. Bakonyi and T.
Kemeny (Kultura, Hungarian Foreign Trading Co., Budapest, 1980) p. 171.
[49] D.G. Morris, Met. Sci. 14 (1980) 215.
[50] R.B. Schwarz, P. Kasiraj, T. Vreeland Jr and T.J. Ahrens, Acta Metall. 32 (1984) 1243.
[51] C.F. Cline and R.W. Hopper, Scr. Metall. 11 (1977) 1137.
[52] P. Kasiraj, D. Kostka, T. Vreeland Jr and T.J. Ahrens, J. Non-Crystalline Solids, 61 62
(1984) 967.
[53] R.G. McQueen, S.P. Marsh, J.W. Taylor, J.N. Fritz and W.C. Carter, in: High Velocity
Impact Phenomena, ed., R. Kinslow (Academic Press, New York, 1970) p. 293.
[54] Jin-lchi Takamura, in: Physical Metallurgy, 2nd ed., ed.,R.W. Cahn (North-Holland, New
York, 1970) p. 857.
[55] M. Atzmon, J.D. Verhoeven, E.D. Gibson and W.L. Johnson, Appl. Phys. Lett. 45 (1984)
1052.
[56] L. Schultz, in: Proc. Fifth Int. Conf. on Rapidly Quenched Metals, Wi~rzburg, W. Germany,
eds. S. Steeb, H. Warlimont (North-Holland, Amsterdam, 1985) p. 1585.
[57] G.M. Hood and R.J. Schultz, Phil. Mag. 26 (1972) 329.
[58] T.C. Tisone and J. Drobek, J. Vac. Sci. Technol. 9 (1972) 271.
[59] W.L. Johnson, B. Dolgin and M. van Rossum, Proc. NATO Advanced Study Institute, April
1984, Teneriffe, Canary Islands, to be published.
[60] J.C. Barbour, F.W. Saris, M. Nastasi and J.W. Mayer, Phys. Rev. B32 (1985) 1363.
[61] J.W. Christian, in: Physical Metallurgy, 2nd ed., ed., R.W. Cahn (North-Holland, New York,
1970) p. 471.
[62] H. Mori, H. Fujita and M. Fujita, Jpn. J. Appl. Phys. 22 (1983) L94.
[63] R.R. Petrich, C. Zimm, R.B. Schwarz and J.A. Barkley, unpublished data (1985).
[64] P. Kasiraj, R.B. Schwarz and T. Vreeland Jr, unpublished data (1985).

S-ar putea să vă placă și