Sunteți pe pagina 1din 61

Nanocrystalline metals crystallized from amorphous solids:

nanocrystallization, structure, and properties


K. Lu l
State Key Laboratory for RSA, Institute ofMetaL Research, Chinese Academy of Sciences, Skenyang 110015, Peoples
Republic of China

Abstract

Polycrystallinematerialswith nanometer-sizedgrains,termednanocrystalline materials,canbeformedby


crystallizingcompletelyamorphous solidsunderproperheattreatmentconditions.The crystallizednanocrys-
talline materialsexhibit someuniquestructuralcharacteristics andnovel propertieswhich are fundamentally
differentfromthoseof theconventionalcoarse-grained polycrystallinematerials.
Thisarticlereviewsthepresent
stateof the art in this field. The current statusof researchand developmentson the nanocrystallization,
microstructureandpropertiesof the materialswill be summarized. Comparisons of structuralcharacteristics
andpropertiesaremadebetweenthecrystallizednanocrystalline materialsandthosepreparedby othermethods.
Furtherconsiderations of thedevelopmentandapplications of thisnewclassof materialswill alsobepresented.

Keywords: Nanocrystallization;Metals;Structuralcharacteristics

1. Introduction

Nanocrystalline materials are single- or multi-phase polycrystals with grain sizesin the nanometer
region (typically lessthan 100 nm in at least one dimension). Owing to the extremely small dimensions,
nanocrystalline materials are structurally characterized by a large volume fraction of grain boundaries
or interphase boundaries (generally called interfaces) which may significantly alter a variety of
physical, mechanical, and chemical properties with respect to the conventional coarse-grained poly-
crystalline materials. In fact, many properties of nanocrystalline samples are found to be fundamentally
different from, and often superior to, those of the conventional polycrystals and the amorphous solids
[ 1,2]. For example, nanocrystalline materials may exhibit increased strength/hardness, improved
ductility/toughness, reduced elastic modulus, enhanced diffusivity, higher specific heat, enhanced
thermal expansion coefficient (TEC), and superior soft magnetic properties in comparison with
conventional polycrystalline materials. In particular, the nanocrystalline structure might be significant
in the improvement of ductility in brittle materials such as ceramics and some intermetallics [ 3,4],
Therefore, nanocrystalline materials provide us not only with an excellent opportunity to study the
nature of solid interfaces and to extend our understanding of the structure-property relationship in
solid materials down to the nanometer regime, but also present an attractive potential for technological
applications with their novel properties [ 51.
In recent years, nanocrystalline materials have stimulated increasing interests in the fields of
materials science and condensed matter physics. A new journal, entitled Nunostructured Materials
published by Pergamon Press, was started in 1992 [ 61. Many international and regional conferences
or symposia have been held specially on nanostructured materials during the past few years. The first
Onleaveat Max-Planck-Institutfur Metallforschung,Institutfur Werkstoffwissenschaft,
Sees&.92,D-70174Stuttgart,
Germany.
162 K. Lu / Nanoclystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

Fig. 1. A schematicrepresentation
of the atomicstructureof a two-dimensional nanocrystallinematerialdistinguishing
betweentheatomsassociated with the individualgrains(solidcircles) andthoseconstitutingthe grainboundarynetwork
(opencircles) [ 11.

international conference on nanostructured materials was held in Cancun (Mexico) in 1992 [ 71, the
second was held in Stuttgart (Germany) in October 1994 [ 81, and the third is scheduled to be held in
Hawaii (USA) in June 1996.
The basic idea of nanocrystalline materials was proposed by Gleiter in 1981 [ 91; this involved
generating a new class of materials by introducing such a high density of grain boundaries that 50%
or more of the atoms are situated in the grain boundaries. Fig. 1 shows a schematic illustration of a
hard-sphere two-dimensional model of a hypothetical nanocrystalline material [lo], There are two
types of atom in the nanocrystalline structure: crystal atoms with neighbor configuration corresponding
to the lattice and boundary atoms with a variety of interatomic spacings.As the nanocrystalline material
contains a high density of interfaces, a substantial fraction of atoms lie in the interfaces. Assuming
that grains have the shape of spheres or cubes, the volume fraction of interfaces in the nanocrystalline
material may be estimated as 3A ld (where A is the average interface thickness and d is the average
grain diameter) [ 111. Thus, the volume fraction of interfaces can be as much as 50% for 5 nm grains,
30% for 10 nm grains, and about 3% for 100 nm grains.
Nanocrystalline materials can be classified into several categories according to the nanostructure
dimensionality [ 121. They can be zero-dimensionality atom clusters and cluster assemblies, one- and
two-dimensionally modulated multilayers and overlayers respectively, and three-dimensional equiaxed
nanocrystalline structures, as indicated schematically in Fig. 2. The nanocrystalline materials may
contain crystalline, quasi-crystalline, or amorphous phases (sometimes called nanoglasses [ 13,14]),
and can be metals, intermetallics, ceramics, semiconductors, and composites. In recent years, most
attention has been concentrated on the three-dimensional nanocrystalline materials which are expected
to find applications based on their improved mechanical, magnetic, and other properties. Increasing
interest has shifted to the one-dimensional nanostructures for possible electronic applications.
Pioneer explorations to synthesizenanocrystalline samplesartificially were performed in Gleiters
group in the early 1980s [ lo]. They synthesizedultrafine (nanometer-sized) metallic particles using
an inert gas condensation technique and consolidated them in situ into small disks under ultra-high
vacuum (UHV) conditions. Up to now, several different synthesismethods for nanocrystalline mate-
rials have been developed following the classical way of the in situ consolidation of the ultrafine
particles (UFPs), with the starting materials in the solid, liquid, or gaseous states. They include
mechanical attrition (or mechanical alloying) [ 15-181, spray conversion processing [ 191, severe
plastic deformation [ 201, sputtering [ 211, electro-depositing [ 22,231, spark erosion [ 241, rapidly
K. Lu / Nanocrystalline metals crystallized from alnorphottssolids: narzoclystallization, structure, and properties 163

Fig. 2. Schematicof the four typesof nanocrystallinematerialclassifiedaccordingto integralmodulationdimensionality


[121.

quenching (see for example Ref. [ 251) , the complete crystallization of amorphous solids [ 261, and
so on.
Amongst all the above synthesis routes, the gas condensation, mechanical attrition, spray con-
version processing, and crystallization (of amorphous solids) techniques have been most commonly
employed to produce large quantities of nanocrystalline samples up to now. A full description of the
gas condensation method, with a variety of evaporation methods, and its consequent processing of
nanocrystalline materials can be found in some review articles by Gleiter [l] and Siegel [ 12,271.
Comprehensive reviews describing the mechanical attrition process as well as the characteristics and
properties of nanocrystalline materials, are presented by Koch [ 181 and Fecht [ 171.
The basic principle for the crystallization method from amorphous solids is to control the crys-
tallization kinetics by optimizing the heat treatment conditions (annealing temperature and time,
heating rate, etc.) so that the amorphous phase crystallizes completely into a polycrystalline material
with ultrafine crystallites. The amorphous solids can be prepared by means of the existing routes, such
as melt-spinning, splat-quenching, mechanical alloying, vapor deposition, or electrodeposition, etc.
(see for example Ref. [ 281) . Crystallization of amorphous solids has been successfully applied in
producing nanometer-sized polycrystalline materials in various alloy systems,e.g. in Fe-, Ni-, and Co-
based alloys [ 26,291c0], as well as in some elements (e.g. Se, Si) [ 41-431. The complete crystalli-
zation of amorphous solids is being established as a promising method for synthesis of nanocrystalline
materials because it possessessome unique advantages:
(i) it is very simple and convenient to control in preparation procedures. Conventional annealing can
realize the nanocrystallization in most alloy and element systemsproviding they are formed into
the amorphous state,and can produce large quantities of the nanocrystalline samples.Also, variable
grain sizes with a wide size range (from a few nanometers to submicrometers) can be easily
obtained in the as-crystallized nanocrystalline specimens by simply modifying the heat treatment
conditions [ 441;
(ii) the complete crystallization method is an efficient .way to produce porosity-free nanocrystalline
materiak [45]. As no artificial consolidation process is.involved and the nanometer-crystallites
164 K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, arld properties

and their boundaries are formed via a solid statephase transformation, the nanocrystalline sample
is dense and clean in the internal interfaces;
(iii) its capability to create different kinds of interfacial structure (coherent, semicoherent, and
comfiletely incoherent) in the nanocrystalline materials provides more opportunities for explo-
ration of the nature of these nanocrystalline interfaces;
(iv) by use of the crystallization method, some nanocrystalline intermetallics, supersaturated metallic
solid solutions, and composites can be easily synthesized;
(v) the nanocrystallization itself provides us with a unique chance to study the interface formation
process from the amorphous state experimentally. The nanocrystallization kinetics and thermo-
dynamics of the amorphous solids are strongly affected by the presence of plenty of interfaces in
the crystallization products [ 46,471. Consequently, it is possible to reveal some fundamental
features of the interfaces in the nanocrystalline materials from the transformation kinetic and
thermodynamic signals [ 481.
Owing to these features, the complete nanocrystallization from amorphous solids, and the crys-
tallized nanocrystalline materials, have been intensively investigated in the past few years. Brief
summaries on the structure and properties of the nanocrystalline materials were reported in Refs. [49]
and [ 501 respectively. This paper gives an overview to the present statusof research and developments
on the nanocrystalline materials made by crystallization of amorphous solids, including the nanocrys-
tallization process,microstructural characteristics, thermal stabilities, and properties.
Controlled crystallization of amorphous alloys can be used to obtain partially crystallized materials
with nanometer-sized crystallites embedded in the residual amorphous matrix. This special nano-
crystal/amorphous composite structure with appropriate compositions allows the materials to exhibit
excellent mechanical or magnetic properties [ 51,521. The Fe-Cu-Nb-Si-B alloys (or so-called FINE-
MET materials) are a good example of this type of material, which was first investigated by Yoshizawa
and co-workers [ 53,541, and later by Herzer [ 551 and others [ 56-581. These nanocrystal/amorphous
composite materials synthesized via crystallization of amorphous alloys will not be included in the
present paper due to spacelimitations.
This paper will begin with the nanocrystallization process from amorphous solids, including
morphologies and grain sizesof the nanocrystallization products, thermodynamics and kinetics, and
the micromechanism of the transformation. The structural characteristicsof the nanocrystalline material
will be discussed with respect to the interfaces, various defects, and the lattice structure of the
nanometer-sized crystallite. The understanding of the grain size stability in nanocrystalline materials
will be addressed. Research results on some properties of these materials will be summarized and
compared with those of the nanocrystalline samplesprepared by using other methods. Further consid-
eration of the development and applications of this new classof materials will also be presented.

2. Nanocrystallization

The amorphous solids are in thermodynamical metastable statesand they will transfer into more
stable statesunder appropriate circumstances. Crystallization is such a transformation during which
an amorphous phase crystallizes into one or more metastable or stable polycrystalline phases. The
driving force for the crystallization is the Gibbs free energy difference between the amorphous and
the crystalline states.Usually, amorphous solids may crystallize into polycrystalline phases when they
are subjected to heat treatments [ 591, irradiations [ 601, or even mechanical attritions [ 611. Amongst
these techniques, conventional thermal annealing is most commonly utilized in investigations on
crystallization of amorphous solids. The dimension of crystallites ranges from a few micrometers to a
K. Lu/Nanoclystalline tnetals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 165

few nanometers, and is strongly dependent on the chemical composition of the amorphous phase and
annealing conditions. The crystallization of amorphous solids into nanometer-sized polycrystalline
phases (say, less than a few hundred nanometers) is referred as nanocrystallization. Nanocrystalli-
zation can be realized upon either isothermal or anisothermal annealing in various amorphous solids
of metallic alloys and semiconductors. This section is devoted to describing morphologies and grain
sizesof the nanocrystallization products, nanocrystallization thermodynamics and kinetics, as well as
the transformation micromechanism.

2.1. Nanocrystallization products

2.1 .l. Phases and morphologies


The phasesand morphologies of nanocrystallization products are dominated by the transformation
mechanism, which is closely related to chemical compositions of the amorphous state and thermody-
namic properties of the corresponding crystalline phases.During conventional crystallization of amor-
phous solids, three different types of transformation (polymorphous, eutectic, and primary
crystallization) are identified depending on their chemical compositions, as described by Kijster and
co-workers [62,63]. Similarly, all of these three types of transformation have been observed in
nanocrystallization of amorphous solids in different systems.Fig. 3 shows a hypothetical free energy
diagram to illustrate these reactions occurring during nanocrystallization of amorphous solids; the
variation of free energy with composition of the glass and various nanocrystalline phases (in this case,
two nanophases, a solid solution ctland a compound ,B, are included) [64] at a chosen annealing
temperature is represented.
Polymorphous nanocrystallization: an amorphous phase crystallizes into a single nanocrystalline
phase of the same chemical composition. This reaction can occur only in concentration ranges near
the compound ( C1in Fig. 3) or the pure element (C,) . The polymorphous nanocrystallization (reaction
(1) or (2) ) may produce a single compound nanophase (/?) or a supersaturated solid solution
nanophase (cr) .
Eutectic nanocrystallization: an amorphous phase crystallizes into two nanocrystalline phases
simultaneously (e.g. reaction (3) : Q+ /3), during which two nanophases grow in a coupled fashion
analogously to the eutectic crystallization of liquids. The reaction has the largest driving force and the
overall composition of the two phasesremains the same as that of the amorphous matrix. The eutectic
nanocrystallization can occur within aconcentration range around the equilibrium eutectic composition
(e.g. in the Ni-P binary system eutectic nanocrystallization occurs within 18.2-20.0 at.% P [ 651,
where the equilibrium eutectic composition is 19.0 at.%), rather than a specific eutectic composition

, I I

CL C3 Cl
Composiiion -

Fig. 3. Hypotheticalfreeenergydiagramto illustratethenanocrystallization


of amorphous
solids.am,g andp arerespectively
the freeenergycurvesof the amorphous phase,a terminalsolidsolutionandanintermetallicphase.
166 K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

as in the conventional crystallization. A possible reason might be that the nanocrystalline material
contains a large amount of interface that may have higher energetic configurations and thus allow a
relatively wide composition range.
Primby nnnocrystallization: an amorphous phase, with a composition deviating from that for
either the eutectic or the polymorphous reaction (C, in Fig. 3), crystallizes, in the first step, into a
primary nanophase (either a supersaturated solid solution or an intermetallic compound) embedded
in the amorphous matrix (reaction (4) : am( C,) + a f am (C,) ) . The residual amorphous phase
(with a concentration C,) crystallizes, in the second step, into nanophasesin the mechanism of either
the eutectic or the polymorphous nanocrystallization.
Nanocrystallization of an amorphous Ni33Zr67alloy is a polymorphous transformation, of which
the product is composed of only one crystalline phase: a NiZr, intermetallic compound, Fig. 4 shows
transmission electron microscope (TEM) images of the as-crystallized NiZr, nanocrystalline sample
[ 381, There are,irregularly-shaped crystalline regions of about 200-500 nm across. High resolution
electron microscope (HREM) observations (Fig. 4(b) ) indicate that each crystalline region is com-
posed of nanostructured lamellar domains of which the thickness is about a few nanometers and the
length is below 100 nm. Between the neighboring lamellae there are coherent twin boundaries, and
the orientation relationship between the two grains is: [ 0011 I ]I[ 11l] 2 and [ 1lo] r )I[ 1lo] 2 [ 371,
The melt-quenched amorphous pure selenium may crystallize into the nanocrystalline state via a
polymorphous nanocrystallization process [ 421. The crystallization product consistsof lamellae which
are made up by polycrystalline fibers of trigonal nanometer-sized Se grains with random orientations.
The mean grain size of the nanocrystalline Se can be as small as 6 nm. In the amorphous Co13Zre7
alloy [ 313, a polymorphous nanocrystallization process results in equiaxed nanocrystalline CoZr,
grains (with acomplex cubic structure) with random orientations. Similarmorphologies were observed
in polymorphous nanocrystallization products of amorphous Zr50C050 [ 341, (Fe,Co)33Zre7 and
Fe,,ZreY [ 331 alloys, and elemental amorphous Si film [41].
The primary nanocrystallization has been studied mainly in Fe-based amorphous alloys because
of their novel magnetic properties in the nanocrystalline state.Fig. 5 shows a TEM image of a typical
primary nanocrystallization product from an amorphous (Fe&u1)&Si9B r3alloy [ 661. It contains two
crystalline phasesof an cr-Fe(Si) and an Fe2B compound. In the nanocrystallization process, cr-Fe( Si)
crystallizes as a primary nanophase which is randomly distributed in the amorphous matrix and has
random orientations, followed by a complete crystallization of the residual amorphous phase into
nanocrystalline cr-Fe( Si) and Fe2B compound. The shapesof these two nanophasesare approximately
isotropical and their grain sizesare uniform with a narrow size distribution, Crystallographic orienta-
tions of the two nanophases are random and no orientational relationship between them is expected.
As the growth process of the primary phase is controlled by atomic diffusion across the crystal-glass
interface, parabolic growth rate is normally observed during primary crystallization processes [ 671.

Fig. 4. (a) A bright field TEM imageof a nanocrystallineNiZr, samplecrystallizedfrom an amorphous


Ni&r6, alloy
annealed at 753K for 15min; (b) anHREM imageof the fine structureof theNiZr, nanocrystallites.
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 167

(330) a-Fc(Si)
(222) a-Fc(Si)
(332) Fe93
(112)1 220) FC2B

.(200) a-Fc(Si)
$310) a-FcfSi)
,(321) a-Fe(%)

Fig. 5. (a) A bright fieldTEM imageand (b) the correspondingelectrondiffractionpatternof a nanocrystalline


Fe-Cu-Si-
B alloy after a primary nanocrystallization
of theamorphous
alloy [661.

Fig. 6 is an HREM image of a nanocrystalline Ni-P alloy which is a typical eutectic nanocrys-
tallization product of a melt-spun amorphous Ni 80P20alloy [ 261. There are two crystalline phases of
an Ni(P) solid solution (with an f.c.c. structure) and an N&P compound (b.c.t. structure). It is clear
that the two nanometer-sized crystallites are in the form of anisotropically-shaped blocks. The N&P
nanocrystallites are separated by the Ni phase, and orientations of the b.c.t. phase differ slightly from
each other. An orientation relationship between two phases was found: (OO1)b,c,t,Ij(1 10)f,c,c. and
( 1lo),,,,,, ]I( 111)f,c,c.[ 681. Similar morphologies and structures have been observed in other eutectic
nanocrystallization products, such as in the Fe-B binary alloy [ 691 and Fe-Ni-P-B alloy [70,71].
Orientational relationships between the two crystalline phasesare frequently observed, as in the eutectic
crystallization products of metallic glasses [ 591.
The nanocrystallization behaviors and the morphologies of the products are significantly influ-
enced by additions of gaseous contaminations and alloy elements in the amorphous phase. Additions
of foreign elements to the amorphous solids may alter the nucleation behaviors and growth rate of the
nanocrystalline phases, thus changing the microstructural morphologies in the nanocrystallization
products. It was found that in the crystallization of Co,,Zr,, glass, the shape of the crystallized CoZr

Fig. 6. An HREM imageof a nanocrystallineNi-P sample(along [OOl] axis of N&P) crystallizedfrom an amorphous
Ni80P20 alloy [261.
168 K. Lu / Nanocrystalline mefals clystallizedfrom amorphous solids: nanocrystallization, structure, and properties

phase changes significantly when the oxygen contamination of the glass can be reduced [72]. This
observation indicates that the oxygen may change the interface energy and the crystal growth rate,
which is responsible for the morphologies of the crystallization products, Altounian et al, [ 731 reported
a strong dependence of crystallization behavior on the purity of the Zr in Ni-Zr glasses.They assumed
that the metastable f.c.c. phase is stabilized by the oxygen content. Crystallization of Co33Zr67glass
[ 721 was found to be primary crystallization of (CoZr,) I -xO, into a depleted matrix instead of the
assumedpolymorphous reaction due to the existence of a certain amount of oxygen, which may come
from the reaction of the chemically reactive Zr-based alloys with the crucible prior to the melt-spinning
(the CoZrz phase can dissolve only a very limited amount of oxygen [ 741) ,
Annealing conditions may also be an important influencing factor for the morphology and stmc-
tures of the nanocrystallization products. By using an electrical current (pause) flowing along the
amorphous ribbon sample, an ultra high heating rate (as much as lo4 K s- ) can be achieved to anneal
the amorphous sample for a very short period of time, sometimes referred as to flash annealing. It
has been demonstrated that nanocrystalline structures can be obtained by means of the flash annealing
technique in many alloy systems [ 75-791. Nevertheless, nanocrystallization products during the flash
annealing might be very different from those of the conventional furnace annealing. Kulik et al. 1751
found that the product of the first (primary) stageof crystallization in someFe-based glassesis strongly
dependent on the annealing heating rate. For example, the primary crystallization product in an Fe-
Cu-Ta-Si-B glassunder conventional annealing is cr-Fe(Si) solid solution nanophase, while aprimary
intermetallic nanophase of Fe3B is formed in a flash annealing process. In the Fe-Ni-Si-B and Fe-B
amorphous alloys, Kopcewicz et al. [77] noticed that the relative contents of different crystalline
phasesin the flash annealing crystallization products are changed with respect to those during conven-
tional annealing. For instance, in FegOBzO glass, the relative content ratio of the three crystalline phases
in the crystallization product of the conventional furnace annealing is cr-Fe:t-Fe,B:t-F&B = 27:64:5,
but is 22:73:5 for flash annealing. During flash annealing of the Fe-Cu-Nb-Si-B and the Fe-%-B
glasses,the lattice parameter of the cr-Fe nanophase was observed to decrease during crystal growth
[ 78,791, indicating that the chemical compositions of the nanophase formed initially are different
from those formed in the final stage.
The mechanically driven crystallization of amorphous alloys induced by high-energy ball milling
may also result in nanostructures [ 61,801. However, the mechanically driven nanocrystallization
product was found to be very different from that induced by thermal annealing. Comparative studies
[SO] between the mechanically driven and the thermally-induced crystallization processesin Fe-Co-
Si-B glassesshowed that the former process could lead to extremely fine equiaxed crystallites (8-10
nm) with a very narrow size distribution, while the later forms coarse grains as large as a few
micrometers. It was also found that the addition of some Co to the amorphous Fe-Si-B ahoy during
the milling accelerates the crystallization while Ni slows it down. Recent investigations [ 811 on the
mechanically driven crystallization process in an Fe-Mo-Si-B glass revealed that the crystallization
products after milling for 200 h (after complete crystallization) consist of only a single phase of QI-
Fe(Si,Mo,B) b.c.c. solid solution with an average grain size of about 6 nm. This is fundamentally
different from the thermally-induced crystallization product of two phases: an a-Fe{ Si) and a boride.
The mechanically driven nanocrystallization seemsto be apowerful processfor synthesizing extremely
fine stable or metastable nanocrystalline structures that might be very useful for some application
purpose.
2.1.2. Grain size
Grain sizesin the nanocrystallization products of amorphous solids are usually determined either
by direct observations under TEMs with high magnifications (or HREMs) , or by estimations according
to the X-ray diffraction (XRD) profiles. Direct observations under an HREM can give clenr images
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrysfallization, structure, and properties 169

40

30
F Se

20

10

0 1

360 380 400 420 440 460 480

Temperature
(Kj
Fig. 7. Variationsof the grain sizeswith the annealingtemperature
along(003)-, (104)- and(210)-zonedirectionsin the
as-crystallizednanocrystalline selenium[42].

of the morphologies and boundaries between the nanometer-sized crystallites, from which crystallite
dimensions in different crystallographic orientations may be accurately measured. Combinations of
the bright field and the dark field TEM images may also provide satisfactory results of crystallite sizes
in a relatively large area of specimen. However, more frequently and more conveniently, the average
gram sizesof nanocrystalline specimens are determined by means of the XRD patterns, from which a
statistic average grain size in the whole sample is obtainable.
The average grain size along a certain crystallographic orientation (h k I), dhk,,may be approxi-
mated by measuring the half maximum breadth of the XRD line according to the Scherrer equation
(see for example Ref. [ 821) : dhkr= CA/ ( Phklcos 0,,,), where h is the wavelength, c is a constant which
can be determined by calibration of the instrument, and /3,iklis the half maximum breadth of XRD line
(hkl).
Measurement results from the XRD experiments using the above equation are in agreement with
TEM or HREM observations in most cases, especially for the nanocrystalline samples containing
uniform equiaxed crystallites. When the crystallite shapes are anisotropic (e.g. in the eutectic or in
some polymorphous nanocrystallization products) the resultant mean grain sizes from the XRD
patterns are dependent on the orientations with which the grain size is calculated. Fig. 7 shows the
variation of the mean grain sizeswith the annealing temperature in the as-crystallized nanocrystalline
selenium in different crystallographic directions. The dimension of grains in the c-axis is evidently
larger than that in u-axis. When the annealing temperature is lower, the difference among these
dimensions diminishes, which agrees with the TEM observations. This result revealed the fact that the
growth rates vary with the orientation during the nanocrystallization process of amorphous selenium
1421.
Grain sizesin the nanocrystallization products are strongly influenced by the annealing conditions
and chemical compositions of the amorphous solids. Upon isothermal annealing, one of the most
important factors dominating the grain sizeis the annealing temperature. The annealing time is usually
determined as the finishing time of the transformation. As shown in Fig. 7, the mean grain sizesincrease
in various orientations with an increasing annealing temperature in the case of Se. Similar variation
tendencies of the increasing grain size at higher annealing temperatures were detected in systems of
Ni-P [44], Fe-Co-Zr [ 321, Fe-Ni-P-B [ 831, and Si [ 411, etc. However, in some alloys, an opposite
tendency is observed, i.e. the mean grain size in the nanocrystalhzation products decreases with an
increment of the annealing temperature, such as in the Co-Zr [ 3 1] and Fe-B [ 841. For the Fe-Si-B
alloy [ 851, the mean grain sizeswere measured within a wide annealing temperature range of 450-
170 K. Lu /Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties

I A A
1 Fe-B l
Co-Zr 0
Ni-P 0
Fe-Co-Zr n
Pd-Cu-Si x
Fe-Ni-P-B A

Fe-B-Si
Si :
\
0

I
0
0,3 0,4 0,s V 67
TITm
Fig. 8. Plotsof themeasured
meangrainsizevs. theannealingtemperature
(dividedby ?,J for differentnanocrystallization
productsfrom amorphous solids.

650 C. They decreasewith increasing temperature in the range of 450-500 C; and reach a minimum
value of about 25 nm at about 500 OC;the grain size increases significantly with an increment of the
annealing temperature above 500 C. For the polymorphous nanocrystallization of the NiZrz glass
[38], however, no significant change in the resultant grain size was detected in a wide temperature
range.
These results indicate the fact that the annealing temperature dependence of grain size is quite
different in different alloy systems.Fig. 8 shows a collection of experimental results of the annealing
temperature dependence of grain size in different alloy systems,from which one may find that the
minimum grain sizesfrequently appear when the annealing temperature is close to about 0.5 T, (where
T, is the melting temperature of the alloy) for these systems [ 861, This phenomenon is certainly a
consequence of the nanocrystallization mechanism which needs further investigation,
The effects of alloying elements on grain size in nanocrystallization products have been checked
only in a few cases.It was found that additions of C and Si in amorphous Fe-based alloys increase the
diffusivity of the metalloids and, therefore, accelerate the growth rate for primary crystallization;
meanwhile, these additions may decrease the number of nuclei and thus create coarser grains [ 871.
For the Fe-B glasses,usually the number of primary crystals decreases with increasing B content
[ 881. Alloying with elements Cu or Au in the Fe-based glassesis known to increase the nucleation
rate of a-Fe by orders of magnitude, while with some slow diffusing elements, e.g. Nb, Zr, or MO, the
crystal growth rate reduces, thus leading to finer microstructures [72] aCr, Co, Ni, or Pd additions
were found to have only a minor influence on the kinetics of primary crystallization in Fe-based glasses
[341*
The resultant grain size of the glass was found to be sensitive to oxygen contamination. In some
Zr-based glasses,oxygen contamination during the melt-spinning process has been observed to result
in primary crystallization of ametastable oxygen-stabilized compound. Existence of oxygen is assumed
to reduce the interface energy, and also reduce the crystal growth rate (by about one order of magnitude
with 1 at.% oxygen), thus leading to an extremely fine grained structure [ 721.
The grain sizeand grain sizedistribution in nanocrystallization products were found to be different
under different annealing conditions. Annealing with an ultrahigh heating rate in a very short period
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 171

of time (i.e. flash annealing) is favorable to formation of finer nanostructures compared with conven-
tional furnace annealing. For an Fe-Cu-Si-B glass [ 751, flash annealing leads to formation of fine a+
Fe( Si) crystallites with an average grain size of about 20 nm, while conventional annealing results in
80 nm cu-Fe(Si) solid solution. In the Fe-Nb and Fe-Zr amorphous alloys, crystallization by using
flash annealing yielded grain sizes in the range 8-20 nm, which are evidently smaller than those
obtained in the furnace treated samples [76].
The effect of the initial structure of the amorphous state on the grain size has been examined in
an Fe-B-,% alloy [ 891. Four different amorphous samples were prepared by using the single roller
spinning technique, where different quenching rates were obtained by changing the rotation speed of
the wheel. It was found that with a decrease of the line speed of the rotating wheel (i.e. .adecrease of
the quenching rate) from 41.5 to 17.0 m s- , the minimum grain size in the crystallization products
increases from 25 to about 70 nm, .while the crystallization products have the same morphology and
the same structure of an Fe(Si) solid solution and an Fe,B compound. An increase in the quenching
rate would enhance the degree of amorphism (or the disordering), which seemsto be a positive effect
for the grain refinement in nanocrystallization products.
Experimental measurement results indicated that the minimum grain sizesare always small (a
few nanometers) for the polymorphous and the eutectic nanocrystallizations, while for the primary
nanocrystallization the minimum grain sizesare larger ( 15 to 30 nm) . Tables 1, 2, and 3 compile the
experimental results of the minimum grain sizesin several systemspublished in the literature for the
polymorphous, eutectic, and primary nanocrystallizations. It should be noted that some data in these
tables, such as in Fe-B alloy [ 841, Fe-Ni-P-B alloy [ 831, are taken from early work dealing with
the interphase spacings in the crystallization products, which are actually a measure of the average
grain size.From these results we may see the minimum grain size is almost independent of the number
of alloy elements. Such a difference in the minimum grain sizes due to different crystallization
mechanisms may provide us with some hints for the nanocrystallization mechanism and the structural
characteristics of the interfaces formed during nanocrystallization processes.Both theoretical calcu-
lations and experimental observations 1931 proved that the microstructures of the amorphous phase

Table 1
Experimentaldataof the minimumgrainsizesd* andthe annealingtemperature
7,in the polymorphous
nanocrystallization
of amorphoussolids

System Crystallinephase(s) d Reference


W-d

Si cr-Si 7.0-8.0 0.50 [411


Se y-Se 6.7-7.8 0.76 [421
Co33Zr67 CoZrz 8.0 0.50 [311
Nldh NiZrz 8.0-10.0 0.49 C381
(Fe, Co)33Zr,, (Fe, Co)Zr, 4.0-5.0 (773 K) [331

Table2
Experimentaldata of the minimumgrain sizesd andthe annealingtemperatureT, in the eutecticnanocrystallizationof
amorphoussolids

System Crystallinephase(s) d* TJTm Reference


(nm)

NlsoPzo Ni,P+Ni(P) 6.0-7.0 0.50 141


Fe8dho Fe,B+Fe(B) 8.0 0.46 [841
Fe4d%oP14B6 (FeNi), +FeNi(PB) 9.0 0.55 L831
172 K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

Table3
Experimentaldataof the minimumgrainsizesd* andthe annealingtemperatureT, in the primary nanocrystallization
of
amorphoussolids

System Crystallinephase(s) d T&m Reference


(nm)
Fe(Si) +Fe,B 21-25 0.51 E851
Fe(Co) + (FeCo),Zr 15 0.51 WI
Fe(Si, MO) + (FeMo),Si + Fe2B 17-20 0.52 [301
Fe(Si) +Fe*B
Pd(Si) + (PdCu),Si 27
19 0.62
0.50 ;;:;
Fe(Si) + Fe,,B, 18 0,53 1921
Fe(Si) + Fe3B 70 (790 K) [751
Fe(Si) +Fe,B 16 (793K) [761

and the interfaces in the nanocrystalline samples play an important role in dominating the grain size
limits: this will be discussedin detail in Section 2.2.1.

2.2. Thermodynamics and kinetics

2.2.1. Nanocrystallization thermodynamics


The conventional crystallization of amorphous solids, i.e. the amorphous-to-crystal transforma-
tion, is usually treated as a solid state phase transformation following the classical thermodynamic
theory of crystallization from the undercooled liquid. Its thermodynamic parameters of the enthalpy
change AH, the entropy change AS, and the Gibbs free energy change AG, are given in the following
equations:
Tm

AH-(T) = -AH,+ (CL-C;) dT


iT
Tm

AS-(T) = -A&/T,+ (C;-C;)/TdT (lb)


1T

AGC-a(~)=AHc-a(~) -T AS-(T) (lc)


where AH, is the enthalpy difference between the amorphous and the crystal at the melting temperature
T,, C; and Cz are the molar heat capacities for the amorphous and crystal respectively.
Similarly, the thermodynamic parameters for a nanocrystallization process of amorphous solids,
or an amorphous-to-nanocrystalline transformation, may be obtained from
T

AH-a(T) =AH"-"(TX) + [(C;-C;) dT (2a)


J
TX

APa =A,S;-a+ (C;-C;)/TdT (2b)


10

AG-(T) =AH-a(~) -T AS-(T) @cl


K. .Lu! Nanocrystalline metals c~ystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 173

I
200.0 300.0 400.0 500.0 600.0 700.0 200.0 300.0 400.0 500.0 600.0 7110.1 3
T (K) T (K)
Fig. 9. Temperature dependences of (a) the entropy changes AS, and (b) the enthalpy changes AH, and the Gibbs free
energy changes AG for the amorphous-to-nanocrystalline (a + nc) and the amorphous-to-crystalline transformation (a + c)
in the Ni-P alloy [ 941.

where AH^(T,) is the enthaipy difference between the amorphous and the nanocrystalline state at
the crystallization temperature TX, which can be measured calorimetrically at a given TX. AS;+ is the
entropy difference between the nanocrystalline and the amorphous stateat 0 K, which may be estimated
according to the free volume concentrations in the two states [ 941. Ci is the molar heat capacity of
the nanocrystalline material.
From the above equations we can see that the three basic thermodynamic parameters for the
nanocrystallization process can be derived providing Ci, Ci, AH-(TX) and AS:+ are available. The
heat capacities and the enthalpy change AH-( TX) can be experimentally measured.For an isothermal
nanocrystallization process in a Ni80P20amorphous alloy, the thermodynamic parameters were meas-
ured [ 941. The heat capacities for the as-quenched amorphous state,the as-crystallized nanocrystalline
state, and the as-castcoarse-grained polycrystalline sample were measured by using differential scan-
ning calorimetry (DSC) in a temperature range of 300-400 K [ 95 J. It was found that the values of
C, for the nanocrystalline state are evidently above those for the amorphous phase with the same
composition (which will be shown in Section 4.1). The heat released or the enthalpy change during
the nanocrystallization at 598 K, was measured to be 3.5 kJ mol-. By means of the density measure-
ment results for both the amorphous and the nanocrystalline samples, the value of AS:+ was estimated
to be about - 0.04kn [94], Figs. 9(a) and 9(b) show the resultant variation relationships of AS-T,
A H-T, and A G-T for the nanocrystallization process. Similarly, these thermodynamic parameters
were also determined for the amorphous-to-crystal transformation in the same system, as shown in
Fig. 9. There are evident differences in these thermodynamic parameters between the two processes.
For the amorphous-to-crystal process, the entropy change is negative; with increasing temperature,
AS decreasesbut AG increases.However, for the nanocrystallization, AS is positive, which means the
entropy is increased during the amorphous-to-nanocrystalline transformation. With increasing tem-
perature, AS and AH increase, but AG decreasesif the grain size is assumed to remain unchanged,
which is the opposite tendency to that in the crystallization process. The fundamental difference
between the conventional crystallization and the nanocrystallization process from amorphous solids
lies in the formation of a large volume fraction of interfaces, if the structure of the nanometer-sized
crystallites is to be regarded as the same as that of the perfect crystal. The apparent difference in the
thermodynamic properties between the two processes may originate from the formation of a large
amount of interfaces, as well as their special microstructural and thermodynamic properties, which
will be discussed in the next section.
174 K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanoctystallization, structure, and properties

The formation of a large amount of interfaces during the nanocrystallization from amorphous
solids has a significant influence on the transformation thermodynamics. Calorimetric measurement
results on the enthalpy change for the nanocrystallization processesrevealed that with a decrease in
grain size, i.e. an increase in the amount of interfaces in the nanocrystallization product, the heat
released during transformation diminishes considerably [48]. Fig. 10 shows the measured data of
enthalpy change during isothermal nanocrystallization processesat different annealing temperatures.
With increasing annealing temperature, the mean grain size increases and the molar fraction of the
interfaces reduces, as plotted in Fig. 11. It is clear that when smaller grains, or more interfaces, are
formed, the enthalpy change becomessmaller, and the difference in AHbetween the nanocrystallization
and the amorphous-to-crystal process is enlarged (see Fig. 10). During anisothermal processes, a
similar variation in behavior was observed; the absolute value of AH decreases with a reduction of

-4.6 I
550.0 575.0 600.0 625.0 650.0 675.0 700.0
Ta (K)
Fig. 10.Measureddataof heatreleased
duringthe amorphous-to-nanocrystalline
transformation
(AM + NC) vs. theannealing
temperatureand the temperaturedependence of the enthalpychangefor the amorphous-to-crystalline
transformation
(AM + C) in the Ni-P alloy [481.

60 - %J .
'0
9 I
- 20
-z 4 a
5 f s
$J 40 - 4 ,-
z P X
I
ba
b
\ - 10
20
!
- 9
:,J-W. Qsi-- ,
-*.
0 0
550 600 650
T, (Kl
Fig. 11,Plotsof the averagegrainsizeof the N&P phasedb,c,,,andthe molarfractionof the interfacess, vs. the annealing
temperaturein thenanocrystallineNi-P samples [48].
K. Lu / Nanocrystalline metals crystallized front amorphous solids: nanocrystallization, structure, and properties 175

heating rate, with which finer nanograins are formed [ 461. These results assumethat the formation of
nanocrystalline interfaces in the amorphous phase is an endothermal process,which reduces the overall
exothermal effect for the transformation. The finer the grains in the nanocrystallization products, the
stronger the effects of interface formation on the enthalpy change for the overall process.
According to the thermodynamic properties of the nanocrystaliization process,the grain sizelimit
for the nanocrystallization of amorphous solids may be estimated. Owing to the large volume fraction
of interfaces in the nanocrystalline samples, the amorphous-to-nanocrystalline transformation may be
considered as a decomposition of the amorphous phase into the nanometer-crystallites and the
interfaces, supposing the interface be regarded as a separated phase in the nanocrystalline sample
[ 961, i.e.
amorphous solid =+ nm-crystallites + interfaces
The Gibbs free energy change for the overall transformation may be expressed as follows if the
interaction effect between the interface and the nanometer-crystallite is negligible:
AG(T)=(l-xi)AGg(T)+xiAGk(T)-AGF(T) (3)
where Xi is the atomic fraction of the interfaces in the nanocrystalline sample, AGf is the formation
Gibbs free energy for different phases,and the superscripts a, c, and i denote amorphous, nanometer-
crystallite, and interfaces respectively. According to the thermodynamic equilibrium condition for a
phase transformation, we may get a maximum value of the interfacial fraction when A G( T) = 0:
AG;(T) -AG;(T)
x,(T) = (4)
AGh(T) -AG;(T)
The atomic fraction of the interface is inversely proportional to the average grain size d, or X,=
ctld (where ct is a constant relative to the thickness of the interface). Assuming the thickness of the
interface is independent of grain size, then the corresponding minimum grain size d is
a[AG;(T) -AG;(T)]
d*(T) = AG;(T) -AG;(T) (5)

It can be seen that the grain size limit is strongly dependent upon the Gibbs free energies of the
three different states:the amorphous, the interfaces in the nanocrystalline sample, and the crystalline
phase(s) . The thermodynamic properties of the nanometer-crystallites might be supposed to be approx-
imately equal to those of the corresponding perfect crystals which are available from the classical
thermodynamics theory, although some recent studies indicated that the microstructure of the nano-
meter-crystallites is more or less different from that of the perfect crystal lattice, as will be discussed
in Section 3.3. Assuming AG(T) =AGi(T) -AGF(T), and AG(T) =AGF(T) -AGF(T), Eq. (5)
simplifies to
AG
dY=aE (6)

Evidently, A G and A G are the excess Gibbs free energies for the interface and the amorphous
phase respectively, related to the corresponding crystalline phase.
The excess Gibbs free energy for the amorphous phase can be approximated according to the
cIassica1thermodynamic theory by assuming the heat capacity difference between the amorphous and
the crystalline state following one of the approximations [97-1021 in the temperature range for
calculations. While the excess Gibbs free energy for the interfaces may be calculated based on the
176 K. Lu /Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structrrre, atrd properties

b
-7

200.0

120.0

0.0

-100.0
0.0 200.0 400.0 600.0 800.0 1000.01200.0
T (K)

Fig. 12.Calculatedresultsof the variationsof the excessGibbsfreeenergy:(a) asa functionof the excessvolumeat three
temperatures(a, 300K; b, 800K; c, 1300K); (b) asa functionof temperature with differentinterfacialexcessvolumes(a,
10%;b, 20%;c, 30%) in Ni [901.

model of the universal equation of state (EOS) [ 103,104], or the quasi-harmonic Debye approximation
[ 105,106].
One of the most important parameters designating the structure of an interface is the excess
volume compared with a single crystal, A Vi = Vi,/ V6- 1 (where Vi and Vb are the specific volumes
for the boundary and the perfect crystal respectively) [ 1071. The thermodynamic properties of an
interface are closely related to its excess volume. Calculations based on the quasi-harmonic Debye
approximation, which has been successfully used in interpreting some thermal properties of nanocrys-
talline metals [ 1061, provided the variations of the interfacial excessGibbs free energy with A Vi and
with temperature in the case of pure nickel [ 901, as shown in Figs. 12(a) and 12(b) . It is seen that
A G increases with decreasing temperature with a given excessvolume, and increases with reduction
of A Vi, at a given temperature.
According to Eq. (6)) the grain size limit d* is determined. It was found that the value of the
ratio A Gil A G is almost independent of the annealing temperature in a wide temperature range (0.3-
0.7T,) with fixed interfacial excessvolume, as shown in Fig. 13(a). However, it is strongly dependent
on the interfacial excessvolume at a given temperature. Fig. 13(b) shows the correlation of d* with
A Vin, which indicates d shrinks with decreasing A Vi,* This result may suggest that for a nanocrys-
tallization from amorphous solids, smaller grains can be obtained when the interfacial excessvolume
is reduced. In other words, finer grains are always associated with denser interfaces with lower
interfacial energy in the nanocrystallization products. This conclusion agrees satisfactorily with exper-
imental observations. As presented in Section 2.1.2, the minimum grain sizes in the polymorphous
and eutectic nanocrystallization products are always small (a few nanometers), while in the primary
one they are larger ( 15-30 nm), as listed in Tables 1-3. This could be interpreted according to their
interfacial structure characteristics in the nanocrystalline materials. In the polymorphous and the
eutectic nanocrystallization products, specific orientation relationships between neighboring nano-
grains are frequently observed. The grain boundaries or interfaces are in low energetic configurations
characterized by coherent or semicoherent boundaries, containing very small excess volumes. How-
ever, the crystallization of a primary phase is controlled by an atomic diffusion process and a com-
positional pileup in front of the growing crystal exists.The primary nanocrystallization products always
contain randomly oriented nanocrystallites, between which there are frequently high-angle grain
boundaries in higher energetic configurations; hence they have larger excessvolumes compared with
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 177

30- a
AV@.30

20 - AVi,=0.20

AVi,=O.lO

a3 0,5 0,7

TiTm

40

: b
30

0
0 10 20 30 40 50 60 70

d' (nm)

Fig. 13. (a) Variation of the ratio ACIAC with temperature


with differentinterfacialexcessvolumes;(b) the calculated
resultof the relationshipbetweenthe interfacialexcessvolume andthe grainsize limit for the nanocrystallinematerials
crystallizedfrom amorphous solids[ 901.

those in the polymorphous and the eutectic cases.These interfacial structural characteristics in different
nanocrystallization products have been detected by microstructural analyses, as will be discussed in
Section 3.1. Therefore, the minimum grain size in the primary nanocrystallization process will be
larger than those in the polymorphous and eutectic nanocrystallization processes.
Another deduction from the calculation is that the higher the excess Gibbs free energy for the
amorphous state, the smaller the value of d, which can be easily seen from Eq. (6). This means that
the higher the degree of amorphism of the original amorphous state, the finer will be the grains
obtained. This is exactly in agreement with the experimental results as presented in Section 2.1.2.

2.2.2. Apparent nanocrystallization kinetics


The nanocrystallization of amorphous solids is a nucleation and growth process, the overall rate
of transformation will reflect the time and temperature dependence of both. The apparent transformation
kinetics is usually described by using the well-known Johnson-Mehl-Avrami equation [ 1081:
x= 1 -exp[ -&(t-T)] (7)
where x is the transformed volume fraction, T is a time lag, kr is a kinetic parameter dependent on
temperature, and n is called Avrami exponent which reflects the transformation mechanism (the
nucIeation and growth behaviors) [ 1091. The value of n for a transformation can be obtained by
178 K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrytallization, strrtctwe, and properties

plotting In [ - In( 1 -x) ] vs. ln( t- 7) (referred to as the JMA plot), providing the x-t dependence
is experimentally determined.
Kinetics of the eutectic nanocrystallization process in a NiSOP2,,glass has been systematically
investigated by using DSC and the magnetothermal analysis (MTA). Fig. 14 shows the isothermal
DSC measurement curves at different annealing temperatures. The transformed volume fraction x as
a function of time at various annealing temperatures was measured by means of both methods [ 471.
Fig. 15 shows a JMA plot for the nanocrystallization process at 579 K, for which the x(t)-t data were
measured by means of MTA [ 1lo]. It can be seen that there are three distinct stages with different
Avrami exponents in the transformation: at the beginning of the transformation, the average values of
Avrami exponent are about 1.3, corresponding to a surface nanocrystallization with a constant nucle-
ation rate and a one-dimensional growth, which was verified by the metallographic observations [ 1lo] ;
in the middle stage, y1increases up to about 3.5, which assumesa three-dimensional nucleation and
growth process in the bulk of the sample becomes dominant until the whole amorphous sample is
completely crystallized; in the final stage, y1decreases and tends to a constant value of about 0.7,
indicating a process of segregation of solute atoms to dislocations or grain boundaries.
The Avrami exponent is found to be very sensitive to the structure of the amorphous phase. Both
experimental and theoretical results [ 111,112] indicated that the amount of the pre-existing crystalline
nuclei (sometimes called quenched-in crystalline nuclei), which is dependent on the quenching
conditions of the rapid solidifications and/or the after-treatment to the as-quenched glass, affects the

I
0
I

500
I
1000
I

1500 2000
Time (s)
I 1
2500
I
3024
,
3500
I
Fig. 14. DSC measurement curves of the isothermal nanocrystallization process in the Ni-P samples annealed at different
temperatures: (a) 607.4 K, (b) 597.6 K, (c) 587.8 K and (d) 578.0 K. The inset indicates determination of the specific
transformation times: t,, starting time; to, onset time; t,, peak time; t,, finishing time [ 471 I

Fig. 15. A JMA plot of the MTA measurement data for the isothermal nanocrystallization in the Ni-P sample at 579 K [47].
ti is the starting time for the bulk nanocrystallization process and t: is the finishing time for the bulk nanocrystallization
process.
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 179

value of ~1.A pre-anneal treatment to the as-quenched Ni-P glass increases the density of pre-existing
crystalline nuclei [ 1131, and suppressesthe Avrami exponent for the bulk nanocrystallization from
4.0 in the as-quenched state down to 3.0 in the pre-annealed samples, in which the transformation is
dominated by a growth process of nuclei. In the as-quenched NiZr, glass which contains some
quenched-in nuclei, the Avrami exponents at various temperatures stay constant at about 2.9, which
means only a crystal growth process is occurring in the polymorphous nanocrystallization [ 1141.
The value of Avrami exponent is affected by the annealing temperature in several cases.In the
polymorphous nanocrystallization of the Co33Zr67glass, a clear decreasing tendency of n with increas-
ing annealing temperature was detected within a temperature interval of about 100 K [ 3 11. However,
from MTA measurement of the nanocrystallization in the Ni-P alloy, the value of ytshows no depend-
ence on the annealing temperature over a range of 35 K [ 1lo]. In fact, the value of Avrami exponent
may be influenced by the determination procedures, such as the determination of baselines in the DSC
curve [ 1151, the incubation time [ 1161, and the degree of crystallization [ 1161, etc.
Activation energy is one of the important parameters describing the transformation kinetics. The
activation energy for the isothermal nanocrystallization is usually determined according to the Arrhe-
nius relation:

t=t, exp (8)

where t is a characteristic time, to is a time constant, R is the gas constant, and E is the activation
energy for the transformation. Fig. 16 shows the Arrhenius plots for nanocrystallization in the Ni-P
glass using several sets of characteristic times measured from the DSC and MTA experiments [ 471.
It can be seen that straight lines are obtained for each set of data, from which the values of activation
energy can be derived. It was noticed that the value of activation energy for the early stage of
nanocrystallization, being about 173 kJ mol-, is smaller than that for the middle or the final stages,
about 245 kJ mol- . Similar observations were reported in the case of CoZr [ 311 and NiZr [ 1141
alloys.
For the anisothermal nanocrystallization process, the Kissinger relation is frequently used for
determination of activation energy, which is expressed as [ 1171

I I I
1,60 1,65 1,70 1,75 1
l/T x?O-~ (K-1
Fig. 16.Arrheniusplotsfor calculationof theactivationenergiesfor the nanocrystallizatio~
,process
by usingdifferent sets
of the specifictimesmeasuredin the DSC and.the MTA experiments[47]. (& startmgtemperaturefor the surface
nanocrystallization;othersymbolsseeFig. 15.)
180 K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

In 5 = - fT +constant
0
where B is the heating rate and T is a characteristic temperature. By plotting ln(B/T2) against l/T
one can derive the value of E from the slope of the straight line plotted. Fig. 17 shows typical DSC
curves for anisothermal nanocrystallization processes in the Ni-P glass with different heating rates,
Using the characteristic temperatures measured from the DSC, one may make Kissinger plots according
Eq. (9). Fig. 18 shows the Kissinger plots for the nanocrystallization using several setsof characteristic
temperatures measured from DSC and MTA. Well-fitted straight lines were obtained, from which the
activation energies at different stagesof nanocrystallization processescan be derived. It was also found
that the value of E is smaller in the beginning of the transformation (about 225 kJ mol- ) than that
in the later stages (260 k.Jmol-) [ 1lo]. Although the values of activation energy for the nanocrys-
tallization process determined from the Arrhenius plot and from the Kissinger plot are different, a
common behavior of the variation in activation energy at different stagesis obtained.
The activation energy for the crystallization is also found to be related to the nature of the
amorphous state and temperature as well. A decrease of the degree of amorphism in the original
amorphous state will reduce the value of E. E values are found to be larger at higher temperatures in

I I I -I

600 620 640 660 680


T iK)
Fig. 17.DSCmeasurement curvesof thenanocrystallization
process
in the Ni-P samples
heatedlinearlyat differentheating
rates(asindicated) [461.

-14. KA
, I
1.45 19 1.55 1,f-O 1.65 1,70
l/T ~103 (K-l1

Fig. 1%Kissingerplotsfor calculationof the activationenergiesfor nanocrystallization


processusingdifferentsetsof the
characteristictemperatures
(r,, startingtemperature;TO,onsettemperature;Tp,peaktemperature; rr, finishingtemperature)
measuredin the DSC andthe MTA experiments[46].
K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties i8i

1
630 L!s!L
+s +o +f
0 + 0

Fig. 19.The T-T-Tdiagram for the nanocrystallization


processin theNi-P samples
determined
from theDSC andtheMTA
measurements [471; (seeFig. 16).

crystallization of the NGP glass and other glasses [ 1181. These features probably originate from the
crystallization micromechanism [ 1181.
Compared with the conventional crystallization process in amorphous solids, the nanocrystalli-
zation is characterized by formation of a large amount of interfaces from the amorphous matrix. The
effects of interface formation on the transformation kinetics have been experimentally detected. The
temperature vs. time transformation (T-T-T) diagrams for the isothermal eutectic nanocrystallization
in the amorphous Ni-P alloy were determined by means of DSC and MTA. Fig. 19 shows the
measurement results of the T-T-T diagram. It is clear that both measurements yield very close values
for the finishing time of the transformation, but different results for the starting time. The difference
in the starting time increases at lower temperatures. This phenomenon reveals a significant influence
of interface formation on the T-T-T diagram for nanocrystallization. As the nanocrystallization of an
amorphous solid can be regarded as a decomposition process of the amorphous phase into nanometer-
sized crystallites and interfaces, the enthalpy change for the overall transformation may be expressed
as

AH(T)=(l-xi)A.HF(T) +xiAHk(T) -AH;(T) (10)


where AH:, AH;, AH! are the formation enthalpies for the amorphous, the crystalline phase, and the
interface respectively. In the usual crystallization, of which the product consists of coarse-grained
polycrystals with a very small amount of interfaces, the enthalpy change is actually the formation
enthalpy difference between the amorphous and the crystalline state. For a nanocrystallization, how-
ever, the enthalpy change for the overall transformation might be reduced due to formation of a large
amount of interfaces, and it decreaseswith a reduction of grain size, as experimentally detected (see
Fig. 10). Therefore, it may be possible that at the beginning of the transformation, when a very small
amount of crystallites are formed, the energy released from the amorphous-to-nanocrystallite process
is compensated for by that from the amorphous-to-interfaces process; so,the overall heat effect becomes
too small to be detected by the calorimeter. In the MTA experiments, however, because the magnetic
signals are dominated by the formation of magnetic crystalline phase(s), at the very beginning,
formation of magnetic nickel crystallites can be detected despite the interface formation. Therefore,
the effect of the interface formation on the MTA measurements might be much smaller than that for
DSC.
182 K. Lu /Nanocrystalline metals clystallizedfrom amorphous solids: nanocrystallization, strwture, and properties

A similar phenomenon has been observed for the temperature vs. heating rate transformation (T-
HR-T) diagram for the nanocrystallization in the Ni-P glass [46]. The effect of interface formation
on the nanocrystallization kinetics can be clearly seen from the comparison between the measurement
results usilig MTA and DSC. These results suggest that formation of a large amount of interfaces has
a significant influence on the heat effect of the overall nanocrystallization process, which can cause
the calorimetric measurements of the transformation kinetics (especially at the beginning stage) to be
less accurate. Therefore, the service time of amorphous materials estimated based on the thermal
analysis results at elevated temperatures might be an overestimation [47].

2.3. Micromechanism

Crystallization of amorphous solids is generally considered as a nucleation and growth process


of the crystalline phase(s). Following the classical theory of nucleation and growth which was
established for the crystallization process in undercooled melts, formation of ultrafine crystalline
structures during crystallization of amorphous solids can be realized by a combination of high nucle-
ation rate and a slow grain growth process, and/or a small critical nucleus size. In the nucleation
theory the steady state homogeneous nucleation rate Ist is given by

Zst=Zo exp(z) exp( -igGc) (11)

where lo is a pre-exponential factor, L is the Loschmidt number, Q is the activation energy for the
transfer of atoms across the surface of the nucleus which is approximately equal to the diffusion
activation energy, and AG, is the free energy required to form a nucleus of the critical size, which can
be written as

where A G, is the Gibbs free energy difference between the crystal and the matrix amorphous phase,
and y is the interfacial energy of the crystal/glass interfaces. The critical nucleus size r, is

in which AH, is the enthalpy change for the transformation, A T= T, - T is the undercooling.
Evidently, both nucleation rate and the critical nucleus sizeare strongly dependent on temperature.
Theoretical calculations of the temperature dependence of Ist showed that Zsthas a maximum value at
a certain temperature; this was confirmed by the experimental results of Morris in an Fe40Ni40P14B6
glass [ 831, and by Kijster and co-workers in Fe-Ni-B [ 1191 and Co-& glasses [ 341, as shown in
Fig. 20.
Another factor influencing the nucleation rate is the interfacial energy. A decreaseof the interfacial
energy will result in a smaller critical nucleus sizeand a reduction of the nucleation barrier A G, which
favors a high nucleation rate at a given temperature. Therefore, a smaller interfacial energy is favorable
for formation of fine nanocrystalline structures from amorphous solids. This is verified by experimental
evidence [ 901,
For the growth of crystalline nuclei, the growth rate may be expressed by

u=r,voexp (14)
K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 183

temperature T IFc-l
1700 II50 850 650 500 490 300
I I
IO- 1029r;;ol
rY-l -- I \ 7r Pn ^_ L.3
?c
-T I \ -jobp) lo2>
mlo- I \
& - -5
-1
3 IO - Ho* ;>
2 -5 nucleation rate - 17 2
E 10 - -10 5
+ ,$I3 '3
g c9- 3
M - :
-13
10 I II II II II II I d
0,5 0.7 0.9 I.1 1.3 I,5 I,7
reciprocal temperature T -1
Fig. 20. Temperaturedependence
of nucleationrateI andgrowthrateII in crystallizationof Co50Zr50
glasses[341,

where v, is the atomic diameter, Y, is the atomic jump frequency, and Q is the activation energy for
the crystal growth. As the crystallization process of amorphous solids is always with a high under-
cooling (about OST,) , the growth rate increases with an increment of temperature.
According to the crystal growth theory, the driving force for grain growth is proportional to the
interfacial energy. Therefore, the growth rate of crystalline nucleus will be dramatically reduced if the
interfacial energy is lowered. The grain size in crystallization is proportional to the square root of the
ratio (~/1,,).~. It is clear that a decrease in the interfacial energy will result in a significant refinement
of grains. This agrees with the thermodynamic analyses and the experimental results [ 901.
Fig. 21 (a) shows the measurement results of the temperature dependencesof nucleation rate and
growth rate in an Fe-Si-B glass [ 851. It was found that the nucleation rate has a maximum value at
around 500-550 C, while the growth rate increasessteadily with temperature. The resultant grain size
as a function of annealing temperature measured experimentally is plotted in Fig. 21 (b) . A minimum
grain size was detected at 500 C. It is clear that the variation tendency of the grain size with the
annealing temperature can be well correlated with the measurement results of nucleation and growth
rates.
Although the theory of nucleation and growth can be used for interpretation of some experimental
results on the nanocrystallization process, there is some evidence that this theory is still far from
providing a satisfactory explanation. This evidence includes the morphologies of some eutectic nan-
ocrystallization products which consist of discontinuous irregular-shaped blocks (grains), and the
different variation tendencies of grain size with the annealing temperature in different alloy systems,
and so on [ 1201. Furthermore, it was found that the kinetic parameters (such as the pre-exponent and
the activation energy for the crystal growth) during crystallization of amorphous solids are inconsistent
with the theory of nucleation and growth [ 591. Cantor and co-workers [ 121,122] have reported that
the atomic (both metallic and metalloid) diffusion coefficients estimated from the crystal growth rates
during crystallization of amorphous alloys were 2-4 orders of magnitude larger than the results of
direct measurements. This means the crystallization of amorphous solids proceeds much faster than
the atomic diffusion mechanism predicts.
According to the dynamic in situ TEM observations and kinetic analyses of the crystallization
process in amorphous alloys, a new crystallization mechanism for amorphous solids was proposed,
the so-called cluster-deposition (CD) mechanism [ 120,123-1251. In this theory, it is supposed that
crystallization consists of two fundamental processes: (i) formation and growth of the precursor
184 K. Lu /Nanocystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties

Temperature IV 1

lo
450 500 550 600
TEMPERATURE (Cl

Fig. 21. (a) Temperature


dependence of nucleationrateandgrowthrateand(b) thetemperature
dependence
of theresultant
meangrainsizein nanocrystallization
of theFe-Si-B alloy [ 851I

ordered clusters in the amorphous phase,and (ii) the nucleation and growth of crystals, which involves
atomic jumping and ordered clusters shearing deposition. This implies that the transformation of an
amorphous phase to crystalline phases involves not only single atoms jumping (as the classical
mechanism suggests) but also a cluster shearing deposition. Fig. 22 schematically shows the two
processesduring crystal growth. According to the new mechanism, when a crystal grows in a totally
disordered region, atoms in the disordered statejump to the crystal front; when the crystal front touches
a precursor ordered cluster, the cluster prefers to shear (change its orientation to match the growing
crystal) and deposit onto the crystal front.
On annealing the glassy Ni-P sample, for example, two kinds of coordination clusters, Ni-Ni and
Ni-P types, are formed randomly in the amorphous matrix. During crystal growth, the crystal front is
composed of two separated crystalline phases:a b.c.t. N&P and an f.c.c. Ni solid solution, Structurally,
Ni-P coordination clusters have the same structural unit as the N&P phase and the Ni-Ni clusters the
same structural unit asthe Ni solution [ 681. If the crystal front of the Ni phase touches aNi-Ni cluster,
the cluster will be sheared and deposited onto the crystal front which passesthe cluster and continues
growing; if it touches a Ni-P cluster, the Ni front stops growing and a new crystal N&P phase begins.
K. Lu / Nanocrystaltine metals crystallized from amorphous solids: nanocrystallizafion, structure, and properties 185

I crystalline , amorphous

Fig. 22. A schematicdiagramillustratingthe two processes


in crystallizationof amorphous
solidsin onedimension:single
atomicjumping from the amorphousstateto the crystallinestate(AJ) andorderedclustersshearingdepositiononto the
crystalfront (CD) I:1231.

For a N&P crystal front, deposition of a Ni-P cluster will cause the crystal front to continue to grow,
and a Ni-Ni cluster stops it. The combination of the atomic jump process and the shearing deposition
process of the Ni-Ni and Ni-P clusters results in the discontinuous growth of the crystallization
products of Ni and N&P phases,
Growing crystals accompanied by ordered clusters in, and in front of, the growing crystals were
clearly identified in a partially crystallized amorphous Ni-P glass, as shown in Fig. 23. An amorphous
region was left between two growing crystals. From the dark field TEM images it is evident that
ordered clusters or even nanocrystalline particles (bright spots) can be found in the crystals as well
as in the amorphous region. The distribution of the clusters with a certain orientation in the amorphous
region is random, while those in the crystals are specific in arrangement. It can be reasonably considered
that the clusters in the crystals might come directly from the shearing deposition of the clusters in the
amorphous matrix.
Nanocrystallization of amorphous solids seemsto be a complex process, to which many factors
are related. A systematic investigation of the nanocrystallization mechanism is urgently needed for a
fully understanding of the formation process of nanostructures in amorphous solids.

Fig. 23. An amorphous regionleft betweentwo growingcrystalsduringcrystallizationof the N-P glassannealedat 565 K
for 180 min. (c) is a bright field imageand (d)-(f) aredark field imagescorresponding to different diffraction spots
indicatedin the diffraction pattern(a) [ 1231.
186 K. Lu /Nanoctystalline metals crystallized from amorphous solids: nanocrystallization, stnrcture, and properties

3. Structures

In this section, structural characteristics of interfaces and the nanometer-sized crystallites, as well
as of various kinds of defect in nanocrystalline materials crystallized from amorphous solids wil1 be
studied. The thermal stability of the crystallized nanostructures will also be dealt with; this is not only
a crucial feature for applications of this new class of materials, but it also provides a way to fully
understand the microstructures of nanocrystalline materials.

3.1. Interjlxes

With grain refinement down to the nanometer regime in polycrystalline materials, the interface
becomes a significant component in the material, and thus is assumedto be very important in dominating
the macroscopic properties of the sample. Much interest has occurred in exploring the nature of
interfaces in nanocrystalline solids since the appearance of these new materials, Many intensive
investigations have been conducted on the structural characteristics of interfaces in nanocrystalline
materials made by consolidation of UFPs, of which most results were summarized in some previous
review papers [ 1,7,126] .
A number of investigations have been performed on interfaces in nanocrystalline metals using
XRD [ 1271, Mijssbauer spectroscopy f 1281, positron annihilation spectroscopy (PAS) [129,130],
Raman scattering [ 1311, and extended X-ray absorption fine structure (EXAFS) [ 132,133]. Most of
the measurement results were interpreted in terms of grain boundary atomic structures that may be
random, possessing neither the short- nor the long-range order, and having a large excessvolume of
about 10-40% [ 131, The nanocrystalline interface structures were claimed to be different from those
in conventional polycrystals. Recently, some HREM observations of the interface in nanocrystalline
sampleshave shown different results [ 134-1371. In a nanocrystalline Pd, Thomas et al. [ 1341 observed
that the grain boundary planes are basically flat but exhibit some local faceting, and the extended
contrast effects caused by the disordered boundaries (if they exist) could not be detected. From the
HREM experiments, they concluded that the nanocrystalline grain boundaries are in rather low energy
configurations, and rather similar to those of the conventional grain boundaries or the asymmetric tilt
ones. Li et al. [ 1361 observed the fringes stop abruptly in each grain at the grain boundary, indicating
that there is little or no atomic disorder perpendicular to the imaged planes They also reported
observations of grain boundary dislocations and five-fold twins [ 1361,
The interfacial structures in nanocrystalline samples synthesizedby the complete crystallization
method have been investigated in several systemsby use of HREM observations and some interfacial
property measurements.HREM observations of the interfaces in primary nanocrystallization products
revealed that the interfaces appear to have essentially the same structures as those found in coarse-
grained materials. Fig. 24 shows HREM images of the cr-Fe(Si) grain boundary and the cu-Fe(Si)-
Fe2B interface in a nanocrystalline sample crystallized from an amorphous (FessMo1)78Si9B13alloy
[ 1381. It is seen that the cr-Fe( Si) grain boundary is very flat and seemsto be a normal high-angle
grain boundary. The interphase boundary in Fig. 24(b) exhibits flat facets separated by steps.In both
interfaces, no extended contrast effects caused by the disordered boundaries can be detected. As the
orientations of the crystallites are random and there is no orientation relationship between the neigh-
boring grains in this sample, the interfaces are completely incoherent.
Fig. 25 shows an HREM image of the nanostructure containing a number of nanometer-sized
crystallites and interfaces in a nanocrystalline material crystallized from an amorphous Ti70Ni20Si10
alloy [ 1391. It is evident that the contrast between any arbitrary grains is mainly due to the lattice
fringes from one grain terminating abruptly at the interfaces. No highly disordered regions were
K. Lu /Nanoclystalline metals crystallizedfrom amorphous solids: nanoctystallization, structure, andproperties 187

Fig. 24. HREM imagesof (a) the @-Fegrainboundaryand(b) the cr-Fe-Fe,Binterfacein thenanocrystallineFe-Mo-Si-
B samplecrystallizedfrom the amorphous
alloy [ 1381.

Fig. 25. An HREM imageof a typical areacontainingseveralcrystallitesandinterfacesin the nanocrystallineTi-Ni-Si


samplecrystallizedfrom the amorphous
solidat 873K for I h [ 1391.

detected at the interfaces and triple points. From the lattice fringes of the nanostructures, incoherent
(or sometimes semicoherent) interfaces are frequently found in the samples. Slightly strained appear-
ances were detected in the core regions of interfaces.
In the polymorphous nanocrystallization products of the Ni33Zr67glass, which consist of only one
intermetallic nanophase NiZr,, most grain boundaries are straight and well-fitted in the lattice fringes,
as in Fig. 4(b). The interfaces are coherent twin boundaries with a specified structure. It can be
expected that such boundaries are in a low energetic configuration with very small excessvolumes.
In the Ni-P eutectic nanocrystallization products, most internal interfaces are the Ni(P)-N&P
interphase boundaries. Rather Aat interfaces (with low energy configurations) were observed in the
HREM images [ 681. Owing to the orientation relationship between the two phases, the interfacial
structure should be specified. According to the orientation relationship between the two phases and
their crystal lattice structures, an interface model between the Ni(P) solid solution and the N&P b.c.t.
phase was constructed, as schematically shown in Fig. 26 [ 681. It can be seen that the atoms in the
(717) plane of f.c.c. Ni cannot regularly occupy the trigonal gaps in the (110) plane of the b.c.t. N&P.
Some of the f.c.c. Ni atoms sit right above the atoms of the (110) b.c.t. plane, which causesa defective
interface structure. In other words, according to the orientation relationship determined from the
electron diffraction patterns, the f.c.c. Ni and the b.c.t. N&P phase might be loosely connected.
Grain boundary energy is athermodynamic parameter which represents the nature of the boundary.
Calorimetry measurements during grain growth yielded that the interfacial energy of the Ni-N&P is
188 K. Lu/Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

illO]AI CiliiA

ii12 IA
C

Fig. 26. Schematic diagram showing the interface model between the Ni solid solution (A) and the N&P compound (T) in
the nanocrystallization product of the Ni-P glass, which was determined according to the measured orientation relationship
between the two phases [ 681.

about only 0.155 J mm2 (and is about 0.111 J m-* for the Ni-Ni boundary) [ 1401; this is close to the
value of alow-angle grain boundary. This, however, meansthe interfaces in the nanocrystalline samples
are in rather low energetic configurations, which is not in agreement with the deduction from the
orientation relationship. A possible reason for this disagreement may be that the lattice parameters
have been changed in the nanometer-sized crystallites relative to the perfect crystal lattice, as will be
discussed in Section 3.3, so that the lattice mismatch between the two nanophases will be different
from that between the coarse grains. By means of the thermodynamics of a transformation from the
amorphous to the nanocrystalline state, the interfacial excessenergies in the nanocrystalline sample
might be determined experimentally. As discussed in Section 2.2, the amorphous-to-nanocrystalline
transformation may be considered as a decomposition process of the amorphous phase into the
nanometer-sized crystallites and interfaces which can be recognized as a separated phase. Then, the
molar excessenergy for the interfaces AEi relative to the perfect crystalline state may be obtained by
AEi,( T) = [ AHnMa(T) - AH+(r) ] /Xi, The enthalpy changes, AH- and AH- can be experimen-
tally measured by using DSC.
For the nanocrystalline Ni-P materials with average grain sizesranging from a few nanometers
to 60 nm, the interfacial excessenergies were measured by using DSC based on the above analysis
[48]. It was found that the average interfacial excessenergy decreasessignificantly with a reduction
of grain size in an approximate linear relation, as shown in Fig. 27. In the sample with a grain size of
a few nanometers, the interfacial energy is rather small, close to the energy of a low-angle grain
boundary. The interfacial excessvolume in the nanocrystalline Ni-P sample was estimated by means

I , 1 I I I 0
0
0 IO 20 30 40 50 60
dbct (nm)

Fig. 27. Variations of the interfacial excess energy A&, and the interfacial excess volume A Vi,,with the average grain size
in the nanocrystalline Ni-P alloy [48].
K. LU / NanocrWalline metals cr)Wallized from amorphous solids: nanocrystallization, structure, a&properties 189

of accurate density measurements [ 1411. It was found that the interfacial excess volume decreases
with a reduction of the mean grain sizein a similar tendency to the interfacial excessenergy, as shown
in Fig. 27. This behavior agrees well with the theoretical calculations using the universal EOS by Fecht
[ 1031, and the experimental evidence of (PAS) measurements [ 1421.
Terwilliger and Chiang [ 1431 reported that in the Ti02 nanocrystalline materials made by use of
the UFP consolidation method, a similar decreasing tendency of grain boundary energy with a reduction
of grain size was detected. These results imply that the interface structure in nanocrystalline materials
might be changeable and dependent on grain size.This feature has been utilized to give explanations
of some experimental phenomena such asthe abnormal Hall-Petch relationship [ 1441 and the intrinsic
grain size stability [ 1451.
The above evidences indicate that the interfaces in nanocrystalline materials crystallized from
amorphous solids appear to have essentially the same structure as those found in coarse-grained
materials. No disordered interfaces with high energetic configurations were detected experimentally
in these materials.
Various types of interface can be obtained in the nanocrystalline samples crystallized from
amorphous solids: coherent, semicoherent, and incoherent, depending upon the nanocrystallization
mechanism. Incoherent interfaces are frequently observed in the equiaxed grains controlled by the
kinetics of the assembly process, such as in the primary nanocrystallization products, and coherent
interfaces are found in caseswhere energetics dominates the process, as in some polymorphous and
eutectic nanocrystallization cases.According to the interface morphologies, nanocrystalline materials
may be classified as listed in Table 4 [ 1461, Different interface morphologies correspond to different
interfacial thermodynamic properties and microstructures. It is clear that UFP consolidation and
mechanical attrition can only yield randomly oriented nanometer-sized crystallites with completely
incoherent interfaces. The nanocrystalline materials crystallized from amorphous alloys can have
different types of interface which provide more possibilities for studying the nature of nanocrystalline
interfaces.

3.2. Defects

The PAS technique has been proved to be a powerful method in surveying the characteristics and
distributions of microdefects in solids and has been successfully applied in the study of the defects in
nanocrystalline materials. Earlier investigations by Schaefer and co-workers [ 129,130] indicated that
in the consolidated nanocrystalline metals, there are three kinds of defect which were thought of as
being structural components of nanocrystalline materials: free volumes corresponding to a short
lifetime component; nanovoids with an intermediate lifetime component; pores represented by long
lifetime components. The positron lifetime results of a nanocrystalline Fe sample synthesized by the
consolidation method [ 1291 are listed in Table 5. It is clear there are two long lifetime components,
73and r4, of which the intensity is about lo-30%, besides the short ( rl) and intermediate ( r2) ones.
These long lifetime components were thought of as resulting from the large voids inside the nano-
crystalline sample which were formed during the compaction process of the powders. The existence
of sample porosity inside the nanocrystalline sample was proved by microscopic observations which
showed that there are pores with diameters as large as about 1 pm in the Fe nanocrystalline specimen.
These three types of defect cannot be annihilated by thermal annealing [ 1301.
Systematic PAS investigations on the defects in the nanocrystalline samples crystallized from
amorphous solids were carried out in the Ni-P [ 45,142], Fe-Si-B [ 1471, Fe-Mo-Si-B [ 148,149],
Co-Zr, Fe-Zr, and Fe-Cu-Nb-Si-B alloys [ 1501. Upon analyzing the positron lifetime spectra for
these samples, some common features were found in these materials.
190 K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, strltcture, and properties
K. Lu/ Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 191

Table5
Positronlifetimeresultsfor nanocrystalline
samples
crystallizedfrom amorphoussolids(Ni-P [ 1421,Fe-Si-B [ 1471,Fe-
Mo-Si-B [ 1481))andfor a nanocrystalline Fe samplemadeby consolidationof UFPs[ 1291

Sample 71 72 73 74 11 I,112 13 + I4
(ps) (PS) (PS) (ps) (%I (%)

Ni-P 152.3+ 1.3 350i 20 147Ok96 - 94.4& 1.3 21.0 1.1kO.1


Fe-Si-B 145.5ri: 1.4 376& 23 1845k41 - 89.5kO.9 11.3 2.5+ 0.8
Fe-rMo-Si-B 139.4+ 1.5 240+ 12 1582k42 - 90.35 1.o 10.7 1.2kO.l
Fe 18Oh15 360&-30 1200+200 4000f500 - 0.55 1.0 10+30

Table 6 lists measurement results of the mean positron lifetime in different states of these alloy
samples. It is noticed that the mean positron lifetime of the nanocrystalline samplesis evidently higher
than that in of the amorphous counterparts for most alloys, but no significant difference was detected
in the case of Fe-Zr and Fe-Cu-Nb-Si-B alloys [ 1501. The positron lifetime results of some nano-
crystalline samples are also listed in Table 5. It is seenthat all positrons injected into the nanocrystalline
samples are annihilated with lifetimes clearly longer than those in the delocalized free statesin bulk
crystals of Ni (TV= 110 ps) and Fe (TV= 106 ps) . The short lifetime component TV,which is consid-
erably larger than that of positrons annihilated in the delocalized free state,but shorter than the lifetime
at monovacancies (7i y = 180 ps for the Ni bulk crystals), might be thought of as resulting from the
positron localized and annihilated at a kind of trap at the interfaces which are probably free volumes
with a size smaller than a monovacancy. In addition, the intermediate lifetime component r2 corre-
sponds to the positrons annihilated at the intersections of two or three crystallite interfaces and/or
grain boundaries, or so-called nanovoids, of which the size is about lo-15 vacancies [ 1301. The long
lifetime component 73, with a very small intensity of about l-3%, may be attributed to the formation
of or&-positronium (o-Ps) at the surfaces between the source and the sample, as well as at the
internal surfaces among the ribbon pieces in the stack samples. No large voids were detected in
microscopy observations of the nanocrystalline Ni- and Fe-based samples crystallized from amorphous
states, even under HREMs [ 1391, It was reported that in some crystallized nanocrystalline materials
(Co-Zr, Fe-Zr, etc.), not even an intermediate lifetime component could be detected, signifying the
absence of nanovoids [ 1501. This feature is fundamentally different from that in the nanocrystalline
samples prepared by consolidation of UFPs and those from mechanical attritions [ 1511.
Therefore, the complete crystallization from the amorphous alloys is an efficient way to produce
porosity-free nanocrystalline samples with denser interfaces and less defects than those prepared by
the UFP consolidation method. HREM observations also revealed that the nanocrystalline structure is
less defective than that produced by the consolidation method [ 1391.

Table6
Comparison of themeanpositronlifetimes(ps) for theas-crystallizednanocrystalline,
theamorphous,
andthecoarse-grained
polycrystallinematerialsin severalsystems

Material Nanocrystalline Amorphous Coarse-grained


polycrystalline

Ni-P [ 1421 160 152 110(Ni)


Fe-Si-B [ 1471 164 151 125
Fe-Mo-Si-B [ 1481 166 151 128
Co-Zr [ 1501 191 187 148
Fe-Zr [ 1501 158 158 145
Fe-Cu-Nb-Si-B [ 1501 145 151 114(Fe,Si)
192 K. Lu /Nanocqstalline metals crystallizedfrom amoyhous solids: nanocrystallization, structure, alid properties

160- l Ni-P a
o Fe-3-B

4 iso- ._
-z 11 xIxf 3
e- _ pp P
140-
4p E
130-
400-
P b

r 350-
a .
300-
bH i if * ii ip ip jp

250-
25:
c
2oj l

515: 9 l * .
.
10: 0 .
0
5: 0 O 0 0

0 20 40 60 80 100 120
Average grain size (nm)

Fig. 28. Plots of the positron lifetime results of nanocrystalline Ni-P [ 1421 and Fe-Si-B alloys [ 1471 vs. the averagegrain
size.

Distributions of the microdefects in the nanocrystalline samplesprepared by crystallization of the


amorphous alloys were found to vary with the grain size.Figs. 28 (a)-28( d) respectively show TV,TV,
the intensity ratio II /Z2, and the mean positron lifetime vs. the grain size in the Ni-P [ 1421 and the
Fe--%-B [ 1471 nanocrystalline samples. It is evident that for both nanocrystalline samples, the life-
times 71and r2 increase with a reduction of the mean grain size. A slight increase in the values of TV
indicates their free volumes increase in size slightly with a reduction of grain size, Similarly, the
intermediate lifetime decreaseswith an increasing grain size, revealing that the nanovoids shrink in
size or the nanovoids are agglomerated by less vacancies when the grain size is larger.
It can be seen from Fig. 28(c) that the intensity ratio Z1/12clearly decreases with an increment
of the grain size for both alloys. The mean positron lifetime ?= (~~1,+ 7.12)/(Ii +12) is constant
around 160 ps for the Ni-P nanocrystalline samples (and about 16.5 ps for the Fe-MO-S-B nano-
crystalline samples [ 1481) . However, the mean lifetime is found to decrease with increasing grain
size for the Fe-Si-B alloy, as shown in Fig. 28 (d) .
For the NGP and the Fe-Si-B nanocrystalline alloys, the ratio II /I2 increaseswhen the grain size
is reduced. The ratio II/I2 is proportional to the interface-area ratio of the two kinds of defect in the
sample, or u~C,I~,C~ a Z1/12 (in which Us denotes the specific trapping rates of various types i of
trap, and C is the concentration of defects). Then it is reasonable to believe that the area ratio of
interfaces with the free volume to those with nanovoids is increased with a decreasing grain size.This
means in a unit area of interfaces there are more free volumes and less nanovoids in the samples with
smaller grains. In other words, the interfacial density increases with a decrease in grain size in both
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 193

alloys, which agrees with the measurement results of the interfacial excessvolume and excessenergy
in the nanocrystalline Ni-P (as in Fig. 27) [ 481.
From Fig. 28(c) one may also notice that values of the ratio Ii /I2 for the Ni-P nanocrystalline
sample (with an eutectic nanocrystallization product) are much larger than those for the primary
nanocrystallized Fe-Si-B alloy with the same grain size. Larger values of II/I2 in the Ni-P samples
imply the interfaces in the Ni-P nanocrystalline samples might be denser and contain less excess
volumes in a unit interfacial area than the Fe-Si-B samples. A similar feature is valid for the primary
nanocrystallization of Fe-Mo-Si-B samples, as listed in Table 5. This evidence indicates that the
eutectic nanocrystallization may create much denser interfaces than the primary nanocrystallization,
which is in satisfactory agreement with the structure analysis results presented in Section 3.1.

3.3. Nanometer-sized crystallites

The structure of nanometer-sized crystallites in nanocrystalline materials is always thought of as


the same as that in the perfect crystal lattice and, hence, has seldom drawn attention. However, an
important question is actually open to whether or not the lattice structure is changed relative to the
perfect crystal when the crystallites are refined down to a few nanometers. The lattice structure of
nanometer-sized crystallites in various nanocrystalline samples crystallized from amorphous solids
was studied by means of the quantitative XRD and Mijssbauer spectroscopy. Measurement results
showed that the lattice structure of the nanometer-sized crystallites is evidently deviated from the
equilibrium state. Two kinds of deviation may be classified: (i) supersaturated structures in solid
solutions and (ii) distorted lattice structures in pure elements and stoichiometric line compounds.

3.3.1. Supersaturated solid solutions


Owing to the Gibbs-Thompson effect, solubilities of solutes are expected to be much enhanced
in solid solutions with the grain refinement down into the nanometer regime. Enhanced solubilities
have been experimentally observed in nanocrystalline materials produced by the UFP consolidation
method. For instance, Miitschele and Kirchheim [ 1521 found that the solubility of H in a Pd nano-
crystalline sample (at a concentration of 10m3or below) is increased by a factor of 10 to 100 relative
to a Pd single crystal. A similar effect was reported by Hahn et al. [ 153] for the solubility of Bi in a
nanocrystalline Cu, which reaches about 4% at 100 C, while the equilibrium solubility of Bi in a Cu
single crystal is less than 10m4. Other experimental evidences, such as the formation of Cu-Fe and
Cu-W solid solutions in the nanocrystalline state [ 154,155], imply that an intrinsic enhancement of
the solid solubility is available in nanocrystalline samples.
Solubility enhancements have also been detected in the crystallized nanocrystalline materials
from amorphous solids. In the crystallized nanocrystalline Ni-P alloy, which consists of two nano-
phases of a Ni(P) solid solution and a Ni3P compound, the concentration of P in the f-cc. Ni was
measured as a function of grain size. The P concentration in the Ni(P) solution nanophase was found
to range from 1.86 to 2.81%, which is about 10-15 times the equilibrium solubility [ 1561. The
measured P concentrations in the Ni nanophases are even much larger than the calculated values from
the Gibbs-Thompson equation for the nanometer-sized grains. With a decreaseof the mean grain size,
the P concentration in the Ni nanocrystallites is reduced, rather than increased as expected based on
the Gibbs-Thompson equation. This behavior, which agrees with the variation of magnetic suscepti-
bility with the grain size, could be understood by considering the annealing temperature effect during
formation of the nanocrystalline Ni(P) solid solution from the amorphous phase. Therefore, it was
concluded that the supersaturated nanocrystalline structure is not only an effect of ultrafine grains, but
it is also closely reIated to the formation history of the nanostructures.
194 K. Lu /Nanoctytalline mefals crystallizedfrom amorphous solids: nanocrystnllizntion, structure, ard proncrties

3
Ni3P

r 20 40 60 80 100
Average grain size (nm)
Fig. 29. Variationsof ha andAC with the averagegrain size for the Ni3PandFe2Bnanometer-sized
crystallitesin the
nanocrystallineNi-P andFe-Cu-S-B alloys.

Liu and co-workers [ 157-1591 have investigated the microstructure of a-Fe(Si) solid solution
in the nanocrystalline Fe-Cu-Si-B samples crystallized from the amorphous state by using transmis-
sion Mbssbauer spectroscopy and quantitative XRD. This showed that Si atoms are substitutionally
dissolved in the Fe b.c.c. lattice with a Si concentration of about 13 at.%, and the arrangement of Si
atoms in the cr-Fe( Si) solution nanophase shows short-range order. There are more Fe configurations
with lower coordination numbers in the cu-Fenanophase than in the coarse-grained polycrystals. With
an increasing grain size,the arrangement of Si atoms in the a-Fe lattice was found to be changed.

3.3.2. Distorted lattice structure


Quantitative XRD measurementsof the lattice parameter have been carried out for the manometer-
sized N&P (b.c.t.) compound in the nanocrystalline Ni-P alloy [ 160,161], and for the Fe,B (b.c.t.)
nanophase in the nanocrystalline Fe-Mo-Si-B alloy [ 1621. Both phases are stoichiometric line
compounds having no solubilities of other elements in the equilibrium phase diagrams. Experimental
data revealed the fact that the lattice parameters of these two nanocrystalline compounds evidently
deviated from the equilibrium values of the perfect crystal lattice. The value of a is always larger than
the standard value of a,, while c is always smaller than co for both b.c.t. nanophase compounds.
Meanwhile, the values of a and c are found to change with the grain size; with a reduction of the
average grain size, the value of a increases and c decreasesin both cases.The relative deviations of
the two parameters for the Fe,B nanophase reach as much as An = 0.22% and AC = - 0.24% with a
grain size of about 25 nm, and Aa = 0.21% and AC= - 0.13% for the 6 nm Ni3P. Fig. 29 shows the
variation of lattice parameter changeswith grain sizein the two nanophases.It is clear that the deviation
of lattice parameters in the Fe,B nanophase is greater than that in the N&P with the same grain size,
which might be attributed to the difference in the interfacial structures in the two nanocrystalline
samples.
The unit cell volume of each b.c.t. compound, V= a2c, was found to be enlarged with respect to
the corresponding equilibrium value, i.e. the lattice of the nanometer-sized crystallite is dilated relative
to the perfect crystal structure. With a decreasing grain size, the lattice expansion increases for both
nanophases, as shown in Fig. 30. The variation of the relative unit cell volume change (A V=
V/V, - 1, V, = ao2co) with 1ld may be approximated by a straight line with a positive slope for both
cases,as in Fig. 30.
Miissbauer spectroscopy investigations [ 1621 on the microstructure of the F%B nanophases with
different grain sizesrevealed that the isomer shift (IS) and the half-line-width (HLW) for the Fe,B
K. Lu/Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 195

-,- , , , ( I
w 0,1 02
l/d (nm)
Fig. 30. Plotsof A V vs. 1ld for theNi;P andFe2Bnanocrystallites.

F
E o,16-
.z
3
2 0,15-

09 ~,~,~~,,,,~~~,~~~,,/,~~,.III,III~
20 30 40 50 60 70 60 90
Average grain size (nm)

Fig. 31. Variationsof the Miissbauerspectroscopy


resultsof (a) IS and (b) HLW and(c) the unit cell volumechangeA V
(from the XRD results)with the averagegrainsizefor theFe2BnanocrystalIite[ 1621.

subspectrum increase with decreasing grain size, as shown in Figs, 31 (a) and 31(b) , An increase of
the IS means the density of s-electron of Fe atoms in the Fe,B phase decreases,which reflects a dilated
lattice structure. The XRD measurement results of the unit cell volume change were plotted vs. the
average grain size in Fig. 31 (c). The increasing HLW value indicates that the distribution of the
atomic coordination distances in the Fe,B phase becomes wider and the degree of disordering is
enhanced. This result is in agreement with the XRD measurementsthat the distortion in the nanophase
lattice increases with a reducing grain size.
In nanocrystalline pure selenium crystallized from the melt-quenched amorphous Se (with a
purity of 99.999%)) evident lattice distortion was also detected [ 1631. The lattice parameters a and c
for the nanometer-sized trigonal selenium show systematic changes with grain size, as indicated in
Fig. 32. With a reduction of grain size from 28 to 8 nm, Aa increases from negative to positive, but
196 K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

0,40

0,30
I-

E
.- 0,lO
5
E
J 0,oo -
a,
.ot: -0,lO
4
-0,20
g 0,20 1
-0,30 1) 1 I I / ( I I I / I I
5 10 15 20 25 30

Grain size (nm)

Fig. 32. Variationsof the latticedistortion(Aa andAC) with the meangrainsizein crystallizednanocrystallineselenium
samples[ 1631.

AC decreasesfrom positive to negative. For the sample with an average grain size of about 8 nm, Aa
is 0.29% and A c is - 0.19%. The unit cell volume increasesby about 0.36%.
In fact, evidences of the lattice distortion in nanocrystalline materials have been detected in
several other cases.In the nanocrystalline element silicon with a mean grain sizeof about 8 nm, made
by high energy ball-milling, the lattice parameter was found to increase by 0.2% with respect to the
single crystal [ 1641. Splat-quenched pure silver microcrystalline samples from melts were found to
exhibit an enhanced lattice parameter of about 0.03% [ 1651, Similarly, in the melt-quenched micro-
crystalline element cobalt, a lattice parameter change of about 0.03% was also noticed [ 1661. In a
granular Bi-(Si02) metal film with a grain size of 13 nm for Bi made by sputtering, the lattice
parameter c increasesby about 1% for the Bi nanophase [ 1671, HREM observations in nanocrystalline
pure Pd synthesized by the consolidation method, also indicated some evidence of lattice distortion
[ 1681. The crystallite structure of pure nanocrystalline Ni with grain size of 6-15 nm prepared by
sputtering was found to exhibit an evident lattice distortion, which is strongly dependent on the grain
size [ 1691. The lattice parameter a of Ni (f.c.c.) is increased by as much as 0.7% when the grain size
is 6 nm, compared with that of the single crystal.
The intrinsic reason for the lattice distortion and expansion in the pure element nanocrystallites
and the stoichiometric line compound nanophases which have no solubility of other elements in the
equilibrium state is still absent. Basically, according to the classical thermodynamics theory, it is
known that the free energy stateof a crystallite with a finite size will be influenced by the presence of
grain boundaries. The overall free energy of the crystallite relative to the perfect crystal lattice will
increase when the grain becomes smaller. Therefore, for the nanometer-sized crystallites, a greatly
enhanced free energy will result in an enhancement of the solubility of other elements and vacancies
(or other types of point defect). In other words, the consequencesof grain refinement into the nanometer
scale may include either a supersaturated solid solution (due to the enhancement of solubilities) or
distorted lattice structures of crystallites (due to a supersaturation of point defects) in pure elements
or line intermetallics [ 1701. The possibility also exists that the stoichiometric line compound would
have a wider solubility range when the grain size is reduced to the nanometer-regime. In fact, it has
been demonstrated by computer simulations of a hypothetically simple multilayered nanostructure that
with a decrease of the modulation wavelength for the sample, the average lattice parameter changes
significantly, and its variation is dependent on the crystallographic orientation [ 1711.
Different interfacial structures may induce different degrees of lattice distortion. In principle, the
higher the interfacial energy, the stronger the effect of interface on the crystallite lattice structure.
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 197

Therefore, one may expect that the degree of lattice distortion will be larger in the randomly oriented
nanometer-crystallites as there are always incoherent interfaces between them (e.g. in the Fe-Mo-Si-
B nanocrystalline sample), and smaller with the low-energy interface configurations, such as in the
case of NiZr, nanostructures which exhibit almost no lattice distortion [ 1141.
For some UFP consolidated nanocrystalline metals, however, the lattice parameters were found
to be unchanged relative to the equilibrium data [ 1721, although very large strains (about 0.7%) and
randomly oriented crystallites are observed in the samples. Different synthesis methods produce
different nanocrystalline materials with fundamentally different microstructures, in their densities,
porosities, interface structures, triple junctions, internal strain, etc. These factors may, more or less,
influence the lattice structure changes. It has been reported that in the UFPs of pure metals [ 173,174]
and oxides [ 17.51,the lattice parameters change with the particle size. The lattice constant variation
shows quite different behavior in different systems;e.g. in the pure Ni the lattice constant increases
with a decrease of particle size [ 1731, but decreases in the case of gold and silver [ 1741. These
evidences indicate that the effect of the free surface on the lattice parameters might be different from
that of the internal interfaces. In order to clarify the nature of lattice structure in the nanometer-sized
crystallites, further comparative investigations are urgently needed using the same materials and the
same measurement but different synthesis routes.

3.4. Grain size stability

The thermal stability of nanocrystalline materials against simultaneous grain growth has both
scientific and technological interests due to the high density of interfaces which may provide a
significant driving force for grain growth. Grain growth occurs in polycrystalline materials to decrease
the total energy of the system. The driving force for the grain growth process in conventional poly-
crystals, according to the well-known Gibbs-Thompson equation (see for example Ref. [ 1761) can
be expressed as

a/L.= 2Yfl
- (15)
r
where fl is the atomic volume, y is the interfacial energy which is assumedto be aconstant independent
of the grain size, and r is the radius of the growing crystallite. Following this relation, one may obtain
that the driving force for the grain growth increases with a reduction of grain size; this might be
extremely large for the nanometer-sized grains even at room temperature.
However, contrary to the expectations, experimental observations indicate that most nanocrys-
talline materials of either metals or compounds, synthesizedby various methods, exhibit inherent grain
size stabilities up to reasonably high temperatures. The grain sizesmay remain rather stable to elevated
temperatures, sometimes as high as about 0.5T, which is comparable with that of the grain growth in
conventional coarse-grained polycrystals. Investigations on the grain size stability have been reported
in various nanocrystalline materials, including pure metals, oxides, compounds, and composites.
Significant grain growth (doubling of the grain size in 24 h) was observed in pure element nanocrys-
talline materials at ambient temperature or below when the melting temperature of the material is
below 600 C [ 21. Higher grain growth temperatures were obtained for the higher T, metals, e.g. for
Cu up to 373 K, for Pd up to 523 K, and for Ti-Mg up to 723 K [ 1771. Grain growth temperature in
a 12 nm Ti02 sample reaches as high as 0.4-0.5T, [ 1781. The inherent grain size stability in nano-
crystalline materials presents a challenge to the classical theory for grain growth.
Grain growth studies in nanocrystalline materials are difficult because the grain size cannot be
accurately determined using conventional methods. Studies of nanocrystalline grain growth can be
198 K. Lu /Nanocrytalline metals crystallized from amorphous solids: nanocrystallization, structure, rind properties

conducted by observing the grain sizeas a function of time at different temperatures using either direct
TEM observations with high magnifications (HRTEM), or estimated from XRD peak broadening
values. In addition, owing to the high density of interfaces inside the nanocrystalline materials, one
may conduct the grain growth process by using a thermal analysis method, such as DSC, The heat
release during growth of nanometer-sized grains is large enough to be detected by present-day calo-
rimeters (which have a measurement accuracy of about 0.04 mJ s- ) . It has been estimated that the
heat release per second can be as much as 3.4 mJ s- for a grain growth process with an initial size of
5 nm, while it decreasesto 0.007 mJ s-* for a 5 ,um grain growth [ 1791. The DSC method has been
successfully applied in studying grain growth in various nanocrystalline samples with both anisoth-
ermal [ 140,180-l 841 and isothermal DSC measurements [ 179,185-1871. The grain growth kinetics
determined from the isothermal DSC measurements agrees well with the XRD results in the Ni-P
nanocrystalline specimens [ 1791.
Grain growth in nanocrystalline materials can also be studied by measuring some macroscopic
properties as a function of temperature or time which are sensitive to grain size, such as electrical
resistivity, magnetic susceptibility, TEC, etc. The variation of grain size during annealing will be
reflected in changes of these properties. The property measurements,in some cases,may provide more
accurate information of the grain growth behavior in nanocrystalline samples than the DSC method
[ 1881. For instance, upon heating the nanocrystalline Ni-P sample at a rate of 10 K min- , the start
of grain growth of the N&P nanophase can be detected much earlier. Compared with the 621 K for
detection of grain growth by a DSC scan [ 1401, using resistivity measurement it can be detected at
about 613 K [ 1891 and at about 611 K by thermal expansion measurement [ 1901I

3.4.1. Intrinsic grain size stability


Grain size stabilities in nanocrystalline materials crystallized from amorphous alloys have been
investigated using TEM, XRD, DSC, and some property measurements.Experimental results indicate
that nanocrystalline samples exhibit intrinsic grain size stability characterized by relatively high grain
growth temperatures and activation energies, Grain growth of the N&P nanophasein the nanocrystalline
Ni-P samples starts at about 647 K (about 0.6?,) at a heating rate of 40 K min- , which is approxi-
mately identical to (or somewhat higher than) the crystallization temperature of the same-composi-
tional amorphous alloy [ 1891. A rather high grain size stability was found in the NiZr, single-phase
nanocrystalline material crystallized from the amorphous phase [ 1141, No significant grain growth
was detected at elevated temperatures as high as 850 K, while the crystallization temperature of the
amorphous NiZr, is only 680 K at a heating rate of 40 K min-.
The thermal stability of a nanocjstalline Fe-Si-B alloy synthesizedby means of the crystalli-
zation method was studied by means of in situ heating TEM observations [ 1911. The sample with an
average grain size of about 30 nm was found to be thermally stable with undetectable grain growth up
to 723 K for 18 ks (the crystallization temperature of the amorphous alloy is about 793 K). On
annealing the nanocrystalline sample at a temperature between 773 to 873 K, the grain size initially
increases rapidly, and after annealing for about 2.5 to 12 ks, tends to a stable value which is essentially
independent of the annealing time, as shown in Fig. 33.
The intrinsic grain size stability in nanocrystalline materials is also manifested by a step-wise
grain growth behavior in some samples.Upon heating a nanocrystalline sample, the grain sizeremains
unchanged until a critical temperature is reached (i.e. the starting temperature for grain growth T,
which is dependent on the heating rate). With increasing temperature above T,, grains grow gradually
but stop growing at a certain sizerather than continuously growing to the submicrometer or micrometer
regime. After the grain growth process, the grain size, which is still in the nanometer regime, will
remain stable even when the sample is heated up to a rather high temperature. On heating the nano-
K. Lu / Nunocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 199

180
-e- 923K
-o- 873K
-m- 823K
-E- T/~K
-+- 723K

0 5 10 15 20

Annealing Time (ks)

Fig. 33. Grainsizeasa functionof theannealingtimefor thenanocrystalline


Fe-Si-B alloy at variousannealing
temperatures
[191].

crystalline Ni-P sample at a constant heating rate, two exothermal peaks appeared in the DSC curve
(see Fig. 34)) corresponding to a grain growth process of the N&P nanophase prior to growth of the
Ni(P) solid solution nanocrystallites. After the two exothermal peaks (the grain growth processes),
the average grain size increasesfrom about 13 nm up to 36 nm, which remains stable at a temperature
as high as 800 K without further detectable grain growth [ 1401.
Similar phenomena of the step-wise grain growth behavior was noticed in nanocrystalline mate-
rials prepared using other methods. This stable grain size was found to be independent of the heating
rate. In the melt-spun nanocrystalline sample with a single phase of HfN&, a step-wise grain growth
process was also observed [ 1921. On heating the as-quenched sample with a mean grain size of 7 nm
at a heating rate of 40 K min- I, significant grain growth occurs within the temperature range of 701-
743 K, ending with a stable size of about 13 nm which remained up to 823 K. In nanocrystalline NbA&
samples prepared by ball-milling [ 1931, the grain size with an initial value of 10 nm, increases
significantly at about 970 K and tends to be about 70 nm at 1273 K. This size was found to remain
unchanged even at 1373 K.
Grain size stabilities in nanocrystalline Ni-P samples with different mean grain sizes ranging
from 7 to 48 nm were studied by means of DSC measurements [ 1451. It was interestingly noticed that
the grain growth starts at a higher temperature in the sample with smaller grains. The peak temperatures

rr
for the two exothermal peaks, which correspond to the growth processes of the two nanophases, are

lTP2
I

TPl
I ;.'
TSl ,:'
.,.~:d.,,
...'
1--- -___ --,L"---II'* -----------
I I I ,

620 660 700 7LO


T (K)
Fig. 34. A DSC curve for graingrowthin a nanocrystalline
NCP sample(with anaveragegrainsizeof 10nm) at a heating
rateof40Kmin-.
200 K. Lu /Nanooystalline metals crystallizedfrom amorphous solids: nanocrystallization, structwe, and properties

3
N
P 722)

720

718

716
\

660- , , , I , I I 1

648-
I-

I f II1~11 I
5 10 50 100
dbc+ (nm)
Fig. 35. Variations of the characteristic temperatures (as indicated in Fig. 34) for grain growth with the average grain size
in the Ni-P nanoqstalline materials [ 1451.

also increased in the samples with smaller grains. Fig. 35 shows the variation of these characteristic
temperatures for the grain growth with the average grain size. Evidently, these grain growth temper-
atures increase (rather than decreaseasexpectedfrom the Gibbs-Thompson equation) with areduction
in the grain size. The activation energies for the two growth processesalso increase with a decreasing
grain size [ 1451.
This abnormal phenomenon of the grain sizedependence of stability in the Ni-P samples seems
to be originated from the nature of the interfaces. As presented in Section 3.1, the interfacial excess
energy was found to decrease with a reduction of grain size.A decrease of the interfacial energy will
reduce the driving force for a grain growth process and then stabilize the grain size. Other possible
reasons may include various kinds of drag due to triple junctions [ 1941, precipitation of a second
phase [ 1951, and the effect of lattice distortion [ 1961.
The grain size stability in the nanocrystalline materials has so far been found to be closely related
to the strnctural characteristics of the material, such as the grain size and its distribution, grain
morphologies, the nature of interfaces and triple junctions, porosities in the sample, and so on. Some
new structural information of the nanocrystalline materials, e.g. the decreasing interfacial ene.rgywith
the grain refinement and the lattice distortion of the nanometer-sized crystallites, may also play an
important role in controlling the grain size stability of nanocrystalline materials. Effects of these
structural characteristics on the grain size stability in nanocrystalline materials need further investi-
gation.

3.4.2. Grain growth process


Grain growth in conventional polycrystalline materials is considered to be controlled by atomic
diffusion in the grain boundary, and its kinetics is frequently described as
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 201

(16)
where d is the grain size at time t, do is the initial grain size,k is a constant, and m is the grain growth
exponent representing the grain growth behavior. m has been found to have values ranging between 2
and 4. The activation energy for grain growth Q can be calculated from the equation

k=&ex

where K0is a frequency term. It is clear that the two parameter Q and m are significant in describing
the grain growth behavior.
Values of the grain growth exponent m have been determined in several nanocrystalline samples
made by using various methods. In nanocrystalline NbA& prepared by ball milling [ 1931, an analysis
of measurement results yielded that m has a value of 3. With some Ti additions which retard the grain
growth of nanocrystalline NbA& phase, the value of m increasesto 4 [ 1971. Spassov and Kijster [ 331
studied the grain growth of crystallized FeZr, and (Fe,Co)Zrz nanocrystalline samples using TEM.
The grain growth exponent m has a value of 3 for both systems,which was found to be independent
of temperature over a wide temperature range (about 200 K) .
However, based on the measurements on nanocrystalline Cu made by the sliding wear technique,
Ganapathi et al. [ 1981 indicated that it is difficult to identify a grain growth mechanism on the basis
of the exponent m alone, as they can get excellent fits for all values of m= 2, 3, or 4. Hijlfer and
Averback [ 1991 noticed a similar evaluation of the exponent for porous nanocrystalline Ti02.
In most of the grain growth studies involving nanocrystalline materials, the value of m is deviated
from the value of 2, which is deduced from the parabolic grain growth. This implies the growth process
of nanometer-sized grains is not only controlled by the Zener drag mechanism. Other mechanisms
such as pinning of grain boundaries by pores or inclusions, by triple junctions, or by the second phase
may also be operative. It has been demonstrated that pores and impurity doping have considerable
effects on the grain growth behavior [ 200,201].
Activation energies for grain growth in some consolidated nanocrystalline oxides [ 2021 and pure
metals [ 181] were found to be close to the value usually observed for normal grain boundary diffusion
in polycrystalline materials. The gaseous contamination in the nanocrystalline pure metals may con-
siderably increase the activation energy for grain growth [ 1811. However, the activation energy for
grain growth of the FeZr, nanophase was found to be 260) 25 kJ mol- for both the FeZr, and
(Fe,Co) Zr, nanocrystalline samples [ 331. This value is close to that for the crystallization process of
amorphous FeZr2.
Activation energies can also be experimentally determined by means of the Kissinger equation
using the grain growth temperatures at different heating rates [ 1401. In the nanocrystalline Ni-P
specimens with a mean grain size of 7 nm, the activation energy was found to be about 176 kJ mol-
for the growth of N&P grains, and about 191 kJ mol- for the growth of the Ni(P) crystallites. These
values are slightly below the activation energy for the crystallization of the amorphous alloy (217 kJ
mol-I), and are evidently higher than that for the grain boundary diffusion in Ni, but close to the
volume diffusion activation energy.
Activation energy for grain growth in a nanocrystalline single-phase HfN& sample prepared by
melt quenching was also found to be close to the value for volume diffusion of Hf [ 1921. It is known
that the usual controlling factor in grain growth is the interface diffusion. The agreement between the
activation energies for nanocrystalline grain growth and the volume self-diffusion implies that there
might be a volume diffusion process occurring during growth of the nanometer-sized crystallites. As
presented in Section 3.3, the lattice structure of nanometer-sized crystallites might be deviated from
202 K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanocjystallization, structwe, and properties

the equilibrium state, exhibiting either supersaturated or distorted structure. Therefore, it might be
reasonably suggestedthat during a growth process from nanometer-sized crystallites to coarser grains,
the lattice structure will be changed to a more stable state approaching the equilibrium configuration
through atomic diffusion not only on the interfaces but also necessarily inside the crystallite lattice.
By using the idea of volume diffusion in the nanocrystalline grain growth, some experimental phe-
nomena of thermal stabilities in the nanocrystalline Ni-Hf samples with different initial structures
have been well explained [ 1961. Nevertheless, attempts should be made to explore the fimdamental
correlation between the lattice structural characteristics of the nanometer-crystallites and their growth
process.

4. Properties

Owing to the ultrafine grains and the high density of interfaces, nanocrystalline materials exhibit
a variety of properties that are often considerably improved in comparison with those of conventional
coarse-grained polycrystalline materials. Properties of nanocrystalline materials are being extensively
investigated to explore possible applications, as well as to extend our understanding of the stmcture-
property relationship in solids.
Measurements in the consolidated nanocrystalline materials indicated that many properties have
been much improved relative to the conventional polycrystalline materials, e.g. the high strength, large
TFCs, enhanced heat capacities, improved ductility, and so on. However, most of thesenovel properties
are found to be very sensitive to the sample preparation procedures [ 203-2051. The structural char-
acteristics which are controlled by the consolidation and processing, such as porosities, internal stress,
etc., may play an important role in properties. It is ambiguous to distinguish the contributions to the
property enhancement from the grain size effect or from other effects such as pores or strain with the
consolidated samples.
Nanocrystalline materials crystallized from amorphous solids, however, can overcome this dif-
ficulty as there is no consolidation process involved in the sample preparation and the as-crystallized
nanocrystalline samples are always porosity-free and of full density. The nanostructures are formed
naturally during a phase transformation. In addition, the crystallized nanocrystalline materials have
other advantages which are very useful for experimental investigations on properties and the structure-
property relationship:
(i) by means of the crystallization method, samples with three different states: the nanocrystalline,
the amorphous, and the coarse-grained polycrystalline, can be made with exactly the same overall
chemical compositions, which is necessary for property comparisons for different solid states
[189];
(ii) large quantities of uniform nanocrystalline samples with regular shapes (ribbons, fibers, or bulk
samples in some cases) are available, which are necessary for repeated property measurements,
especially for mechanical properties;
(iii) as grain sizesin the as-crystallized nanocrystalline samplescan be easily changed within a wide
range (from a few nanometers to submicrometers) by modifying the heat treatment conditions,
the grain size effect on properties may be systematically investigated using these samples.
Owing to these features, property studies have been carried out in a number of nanocrystalline
materials crystallized from amorphous solids. This section will summarize results on the thermal,
electrical, magnetic, and mechanical properties.
K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 203

4.1. Heat capacity

The heat capacity of a solid is directly related to its atomic structure and thermodynamic properties.
Measurements of the heat capacity as afunction of temperature in nanocrystalline materials are essential
in revealing the fundamental difference between the nanocrystalline state and the amorphous or the
conventional crystalline counterparts of the same chemical composition. Heat capacities C, of nano-
crystalline Pd and Cu produced by consolidation [ 2061, as well as ball-milled Ru and AlRu samples
[ 2071 have been measured experimentally, indicating that the heat capacities of the nanocrystalline
state are much higher than those for the coarse-grained polycrystalline ones (by as much as 50% in
the case of Pd) . Values of C, for the nanocrystalline Pd were also claimed to be greater than those for
the amorphous Pd estimated based on the values of an amorphous Pd-Fe-Si alloy. The C, enhancement
in nanocrystalline samples with respect to the single crystal is found to be dependent on temperature
and grain size of the sample [ 11.
By means of crystallization from amorphous solids, one can get a nanocrystalline material that
has exactly the same overall chemical composition as the precursory amorphous solid, which makes
it available for comparing the heat capacities of the three different states. Heat capacities of the
nanocrystalline, the amorphous, and the as-cast coarse-grained polycrystalline Ni-P samples, which
have the same composition of Ni80P20,were measured by using DSC with a step-wise heating procedure
over a temperature range of 200-570 K [ 2081. It is noticed that values of C, for the nanocrystalline
sample are evidently above those for the amorphous and the crystalline alloy, as shown in Fig. 36,
while values of C, for the amorphous sample have no pronounced difference from those for the as-
cast crystalline state. The enhancement in C, for the nanocrystalline sample relative to the crystalline
one increases with temperature in an approximate linear relation. In the crystallized Fe-Si-B nano-
crystalline alloy, a similar result was detected in the temperature range of 350-450 K; the C, of the
nanocrystalline sample is larger than that of the crystalline sample by about 19%, while it is only 6%
for the amorphous sample [ 891. Table 7 shows the comparison of heat capacities in the three states
for different nanocrystalline samples.
The heat capacity of a solid is closely related to its vibrational and configurational entropy, which
are significantly affected by the nearest-neighbor configurations. A large number of grain boundaries
or interfaces inside nanocrystalline materials may certainly contribute to the enhancement in the heat
capacity, as quantitatively calculated based on theoretical modelling [ 1061. Nevertheless, effects of

28

24

300 350 400 450 500

Temperature (K)
Fig. 36.Plotsof measured
heatcapacitiesvs. temperature
for threedifferentstatesof theNi-P alloys [208].
204 K. Lu /Nanoclystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

Table7
Comparison of heatcapacity(J mol- K- ) of thenanocrystalline,
the amorphous,
andthe coarse-grained
polycrystalline
materialspreparedby usingdifferentmethods;the measurement temperatures
aregivenin parentheses

Material Nanocrystalline Amorphous Coarse-grained Reference


polycrystalline

Cu a(2.50K) 26 24 W61
Pd = (250 K) 37 27 d 24 VW
Ni-P b (400 K) 27 26 26 [2@31
Fe-Si-B b (350-450K) 28 25 24 [891
Ru (250 K) 28 23 [2071
aSamplespreparedby meansof the UFPconsolidation method.
bSamplescrystallizedfrom amorphous solids.
Samples
preparedby ball milling.
dExtrapolatedto purePd basedon the valuesfor amorphous
Pd$ilsFeIO.

the sample porosities, grain boundary junctions, as well as the lattice distortion in the nanometer-
crystallites, should also be seriously considered in interpretations of the heat capacity for nanocrystal-
line materials. The result that C, for nanocrystalline materials is larger than that for the
same-compositional amorphous sample implies the entropy of the nanocrystalline statemight be higher
than that of the amorphous one, as discussed in Section 2.1. Such a feature might be significant for
understanding the nature of nanocrystallization and the solid state amorphization process as well.

4.2. Thermal expansion

Measurements of TECs in the consolidated nanocrystalline metals (Pd and Cu) showed that the
values of the linear TEC aL are almost twice the values for the single crystals [ 2091. By contributing
the increments of LYEfor the nanocrystalline sample to the presence of a large number of grain
boundaries, the TEC for the grain boundary was estimated, being about 2-5 times that for the single
crystal [ 2101. It was also found that the TEC of consolidated nanocrystalline metals is sensitive to the
consolidation conditions [ 2051.
Measurements of the as-crystallized nanocrystalline materials from amorphous alloys show that
their linear TECs are much higher than those of the crystalline and the amorphous states,For the Ni-
P [95] and Fe-Si-B [29] nanocrystalline samples, aL are respectively about 56% and 106% larger
than those for the corresponding as-castcrystalline counterparts with coarse grains; for the amorphous
alloys they are only a few percent higher. Table 8 lists the measurement results of cxLfor the nano-
crystalline, the amorphous, and the crystalline statesin several systems.These results show convincing
evidence that the interfaces in nanocrystalline materials have an enhanced TEC relative to the single
Table8
Comparison of TEC (K- ) of the nanocrystalline,
the amorphous,
andthe coarse-grained
polycrystallinematerials
prepared
by usingdifferentmethods;the measurement temperaturesaregivenin parentheses

Material Nanocrystalline Amorphous Coarse-grained Reference


polycrystalline

Cu a ( 1lo-293 K) 31x 10-6 - 16x 1O-6 [2091


Ni-P b (300-400K) 21.4X 1O-6 14.2X lo+ 13.7x 10-6
Fe-Si-B b (300-500K) 14.1x lo+ 7.4x 10-6 6.9x 1O-6
aSamplespreparedby meansof theUFPconsolidation
method.
b Samples
crystallizedfrom amorphous
solids.
K. Lu / Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 205

>

0 25 50 75 100 125 150

Grain size Pm)


Fig. 37. A plat of measurement
resultsof the linearTEC vs. grainsizeof theN&P phasein the nanocrystalline
M-P alloy.
The datafor the as-castcoarse-grained
polycrystalline(C) andthe as-quenched amorphous(Am) M-P samples are also
indicated[ 1901.

crystal lattice (and the amorphous state aswell) as there are no pores inside the nanocrystalline samples
crystallized from amorphous alloys.
The grain size dependence of the linear TECs in nanocrystalline materials with a wide grain size
range were measured in the Ni-P alloy system [ 1901. Fig. 37 shows the measured data using the as-
crystallized nanocrystalline samples with different annealing temperatures, which indicates the value
of q increases significantly with a reduction of grain sizefrom about one hundred to a few nanometers.
The same variation tendency was noticed for the nanocrystalline samplesundergoing different degrees
of grain growth annealed isothermally for different periods of time.
It is normally considered that grain boundary has an enhanced TEC relative to the crystal lattice
due to its excessvolume, although amounts of the enhancement are found to be different from different
investigations. By describing a nanocrystalline material as a two-component system with a crystallite
component and a grain boundary (or interface) one, its TEC can be estimated by appropriate scaling
of the grain boundary contribution. Assuming both the interfaces and the crystallites have the same
chemical composition and elastic modulus, one may get the aL of nanocrystalline samples by an
approximation of
Cf~=Fi~t+ (l-Fi)C?E (18)
where Fi is the volume fraction of interfaces which is a function of grain size,Fi = c/d (c is a constant),
and cut and CX~are linear TECs for the interfaces and the crystallites respectively. Normally, structures
of both the interface and the crystallite in nanocrystalline materials are considered to remain unchanged
when the grain size varies, i.e. both crl and CY:should be independent of grain size; hence, we may
get

Then, one may expect that a plot of A aL vs. l/d should yield a straight line with a slope of
c(c$--a;).
Fig. 38 shows a plot of A CY,/CU,
vs. 1ld using the experimental data for the nanocrystailine Ni-
P samples with various grain sizesobtained by crystallizing the amorphous solid at different annealing
temperatures and by a grain growth process of the nanocrystalline sample annealed at 6.50 K. It can be
seen that the plot is evidently a curve (far from a straight line) with a decreasing slope when grain
size is reducing. This evidence implies the difference of (c& - 0:) does not remain constant in the
whole grain size range, but varies with the grain size. With a decreasing grain size, the difference in
206 K. Lu /Nanocrystallinemetalscrystallizedfromamorphous
solids:nanocrystallization,structure,arldproperties

60 / I

0,oo 0,05 0,lO 0,15


lid (nm-I)
Fig. 38. A plot of Dcr,/c& vs. 1ld for the two setsof experimentalresults.The solidcirclesrepresentthe datafrom theas-
crystallizednanocrystallineNi-P samples with differentgrainsizes,whilethe opencirclesrepresentthe datafrom thegrain
grownnanocrystaliine Ni-P samples armeaIed isothermallyat 650K [ 1901.

the ctL between the interface and the crystallite, (a;- a:) diminishes considerably. The ratio of
c&Ictt decreasesfrom about 12.7 to 1.2 when the mean grain size decreasesfrom about 100 nm to a
few nanometers, i.e. the TEC of the interfaces decreasesfrom a value of one order of magnitude larger
than that for the crystallite down to only about 20% higher. This evidence may imply: (i) a decrease
of cr,, and/or (ii) an increase of cr:. According to theoretical analyses [ 1061, a decrease in the TEC
of an interface means a reduction of the interfacial excessvolume (A V,,) , or in other words, the TEC
of the interfaces decreaseswhen the interface is densified. Similarly, an increase in CY:indicates that
the crystallite structure is dilated or expanded. Both effects, however, have been experimentally
observed by structural analyses and were presented in Sections 3.1 and 3.3 respectively.

4.3. Electrical properties

According to the theory of electron scattering in solids, the electrical resistivity of nanocrystalline
materials is expected to be higher than that in the corresponding coarse-grained polycrystalline ones
due to the increased volume fraction of atoms lying on the grain boundaries. This analysis is found to
be in agreement with the reported measurement results in nanocrystalline metals (Pd, Cu, and Fe)
made by consolidation of UFPs [ 11, nanocrystalline Ni by electrodeposition [ 2111, and nanocrystal-
line alloy samples crystallized from amorphous solids [ 36,212,213].
The electrical resistivity of nanocrystalline materials is also found to be higher than that for the
amorphous solids. The resistivity p at room temperature for an amorphous Ni-P alloy is 363 ,L& cm,
but 622 @l cm for the as-crystallized nanocrystalline sample with a mean grain size of 10 nm [ 2121.
Similarly, Wang and co-workers [36,213] measured p for an amorphous alloy Fe-Cu-Si-B as 102
,u,Lsz
cm; after crystallization to a grain size of 30 nm p was 126 ~0 cm.
Variation of electrical resistivity with temperature and grain size has been determined experi-
mentally in several nanocrystalline samples. Fig. 39 shows the measurement results of electrical
resistivity vs. temperature in nanocrystalline Ni-P samples with different grain sizes.It is evident that
for different grain-sized samples,resistivities are roughly proportional to the absolute temperature in
a wide temperature range from 78 to 300 K. The value of resistivity at a certain temperature increases
with a reduction of grain size. A similar behavior was found in the Fe-Cu-Si-B nanocrystalline
materials with grain sizesranging from 90 to 30 nm [ 361.
The residual resistivity at 0 K was found to decrease with an increasing grain size, as shown in
Fig. 40(a). Based on the electron scattering theory, the electrical conductivity is supposed to be
K. Lu / Nnnocrystalknemetals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties 207

0 100 200 300


T (K)
Fig. 39. Plotsof resistivityvs. temperature
for theNi-P nanocrystalline
samples
with differentaveragegrainsizes[212]: a,
11nm; b, 14nm; c, 30 nm; d, 51 nm; e, 102nm.

approximately proportional to the grain size. As the volume fraction of the interface in the nanocrys-
talline materials is inversely proportional to the grain size, then the dependence of residual resistivity
on grain size can be correlated with that of the interfacial volume fraction, as indicated in Fig. 40(a).
This variation tendency has been verified in the Fe-Cu-Si-B nanocrystalline alloy [ 361.
The temperature coefficient of resistivity (TCR) shows a decreasing tendency with a reduction
of grain size in the nanocrystalline Ni-P samples, as indicated in Fig. 40(b). Such a decreasing
behavior of TCR with grain size is similar to that in the nanocrystalline Pd made by consolidation of
UFPs [ 11. However, in the Fe-Cu-Si-B samples, the TCR was found to increase when the grains
become smaller [36], The value of TCR increases from 1.60X 10d3 to 2.01 X 10m3 K-, with a
decreased grain size from 90 to 30 nm. The conflicting variation tendency of TCR with grain size in
this alloy might result from its microstructural characteristics of the grain boundaries and also the
crystallites.

-250
a + expl 9.
--- CJt'd Fin - 200
';; \\
;\ \
,_lO- , -150p
\\, c
\
\\
l

\
\
.:\,, -KJ

0 I 0 o-
0 50 100 0 50 100
d(nm)
d (nm)
Fig. 40. (a) Variation of the residualresistivityandthe interfacialvolumefractionwith the averagegrainsizefor the Ni-P
nanocrystaIlinematerials.(b) A plot of TCR vs. theaveragegrainsizein the Ni-P sample[212].
208 K. Lu / Nanoctystalline metals crystallized from amorphous solids: nanocrystallization, structwe, and properties

The above experimental results can be understood following the scattering theory of electrons by
grain boundaries, as discussedin Ref. [ 11. If the crystallite size is smaller than the electron mean free
path, grain boundary scattering dominates and hence the electrical resistivity aswell asthe temperature
coefficient are expected to increase. The electrical resistivity is also very sensitive to lattice imperfec-
tions in solids, such as vacancies and dislocations. As pointed out by Bakonyi et al. [ 2141, besides the
crystarlite boundaries, the presence of a large amount of other types of lattice imperfection has also
been found to have an effect on the electrical resistivity in nanocrystalline materials. This suggeststhe
necessity of a more detailed characterization of the lattice microstructure of the crystallites and its
effect on the electrical resistivity in nanocrystalline materials.

4.4. Magnetic properties

Recently, magnetic properties of partially crystallized amorphous Fe-base alloys made by


the rapid solidification method have received extensive investigations since the pioneering work of
Yoshizawa and co-workers [ 53,541. This new class of materials, which have the nanometer-sized
grain structures of b.c.c. cr-Fe solid solution embedded in the amorphous matrix, exhibit excellent
magnetic properties with a good combination of low coercivity, high permeability, almost zero mag-
netostriction, and low core losses.However, for the completely crystallized nanocrystalline materials
from the amorphous alloys, magnetic properties have seldom been reported.
The Curie temperature T, of the nanocrystalline materials crystallized from amorphous solids is
found to be decreased with respect to that for the coarse-grained polycrystalline counterparts. In the
nanocrystalline Fe-Si-B alloys, crystallized completely from the amorphous phase, Tc was measured
as a function of the mean grain size, as plotted in Fig. 41 [ 891. This indicates that T, decreaseswith
a reduction of the average grain size.The depression of T, for the nanocrystalline sample can be more
than 10 K compared with the microcrystalline counterpart. In the nanocrystalline Ni-P samples, the
Curie temperature, which was determined from the C, measurement data [ 2081, is depressed by as
much as about 55 K with respect to that for the as-castcrystalline sample. This evidence agrees with
the experimental observations in other nanocrystalline materials prepared by different methods. For
instance, a reduction of the Curie temperature of Ni of about 40 IS was reported if the crystal size was
reduced to about 70 nm [ 2151. A depressedTc of about 545 K for a 12 nm pure Ni sample was also
reported [ 1511. The depression of Curie temperature indicates the nearest-neighbor coordinations are
essentially changed in the magnetic nanocrystallites. This reflects, to some extent, that there are more
open disordered spacesor the nearest-neighbor coordination distance in the nanometer-sized crystallites
is increased, caused by either supersaturation of other elements or lattice distortion.

700

690
I I I I I
25 50 75 100
3 (ml
Fig. 41. Variation of Curie temperature with the average grain size for the Fe-Si-B nanocrystalline samples [ 891,
K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties 209

The depression in Curie temperature in the Ni-P nanocrystalline samples was attributed to the
lattice structure change due to supersaturation of P (and/or vacancies) in the Ni nanocrystallites. This
idea is supported by the magnetic susceptibility measurement of the as-crystallized samples using the
MTA [ 1101. The value of susceptibility increases with a reduction of the annealing temperature (i.e.
with a reduction of the mean grain size) ; this was correlated with the difference in the supersaturation
of P in the Ni f.c.c. solid solution at different annealing temperatures [ 1561.
The magnetic properties of the amorphous, the nanocrystalline, the coarse-grained polycrystalline,
and the partially crystallized amorphous Fe-Si-B samples were comparatively studied [ 891. It was
found that the partially crystallized amorphous sample (with nanometer-sized crystallites embedded)
exhibits the best magnetic properties with agood combination of a low coercivity, about 0.05 Oe.(0.11
Oe for the amorphous and 4.96 Oe for the nanocrystalline) , and a high permeability of about 4.42 X lo3
(6.02 X lo3 for the amorphous and 10 for the nanocrystalline) .
The saturation magnetization MS and coercivity H, in the as-crystallized nanocrystalline Fe-Si-
B alloy were measured as a function of the mean grain size [ 891. It was observed that with a reduction
of grain size from a few micrometers to 25 nm, the value of MS decreasesfrom 5 X lo3 to 2 X lo3 Gs,
while the coercivity increases from about 3 to 5 Oe. This behavior may be qualitatively explained by
the presence of a plenty of crystallite boundaries in the nanocrystalline samples.

4.5. Mechanical properties

Hardness, which is defined as the extent to which the surface of a material deforms under stress,
is often used as an indication of the mechanical properties of a solid material. Hardness is commonly
measured using a Vickers microhardness tester in which a pyramidal diamond indentor with a square
cross-section is loaded with a specified force and applied to the sample surface for a given period of
time. Like strength, hardness typically derives from the difficulty in creating dislocations and the
impedance of their motion by the development of barriers.
The microhardness of some nanocrystalline samples synthesizedby crystallizing the amorphous
solids has been studied in recent years; such studies have included nanocrystalline pure element (Se
[ 2161) , single phase systems (NiZr, [ 1141)) and multiphase systems (Ni-P [ 441 and Fe-base alloys
[ 29,217,218] ) , One common feature obtained from these experimental measurements was that after
nanocrystallization of amorphous solids, hardness increases,i.e. hardness of the nanocrystalline mate-
rials is higher than that of the amorphous matrix. Table 9 lists some comparisons of hardness meas-
urement results between the crystallized nanocrystalline samples and the same-compositional
amorphous counterparts. The hardness of the nanocrystalline sample was found to have increased by
as much as twice that of the amorphous one. However, it is found that the nanocrystalline samples can
Table9
Comparison of microhardness(GPa) of thenanocrystalline,
the amorphous,andthecoarse-grained
polycrystallinematerials
preparedby crystallizingamorphous
solids;grainsizesgiven in parentheses

Material Nanocrystalline Amorphous Coarse-grained Reference


polycrystalline

Ni-P 10.4 (9 nm) 6.5 11.3(120nm) [Ml


Se 0.98 (8 nm) 0.41 0.34(25 nm) I2161
Fe-Si-B 11.8 (25 nm) 7.7 6.2 ( N pm) WI
Fe-Cu-Si-B 9.8 (30 nm) 7.5 7.5 (250 nm) I2181
Fe-Mo-Si-B 10.0(45 nm) 6.4(200 nm) [2181
Ni-Zr 6.5 (19 nm) 3.8 (100 nm) [I141
210 K. Lu /Nanocrystalline metals cl-ystallizedfrorn amorphous solids: nanocrystallization, structure, and properties

be harder (such as in Se, Fe-Si-B, and Fe-Mo-Si-B) or softer (Ni-P) than the coarse-grained
polycrystalline counterparts, although it is always expected from the classical hardening theory that a
reduction of grain sizewould result in hardening of the materials.
Many experimental measurements indicate that in conventional polycrystalline materials, the
hardness increases with a decreasing grain size, following the well-known empirical Hall-Petch
relationship [ 219,220] :

H, = Ho + kd- *2 cm

where H, is the hardness,d is the grain diameter, and Ho and k are constants,Normally, for conventional
polycrystalline materials, the value of k is positive.
Many theories have been developed to explain the hardnessvariation in terms of dislocations and
other microstructural features. Among them, the first is the Hall-Petch model [ 219,220], which treated
grain boundaries as barriers to dislocation motion and was derived for large-grained materials with
high dislocation densities. With a reduction of the grain size, there are more grain boundaries which
are effective dislocation barriers; thus the material becomesharder. Other theories include the Cottrell
theory [ 2211, Lis model [ 2221, and Conrads model [ 2231. Wyrzykowkski and Grabski [ 2241 have
considered the effect of grain boundary structure on the Hall-Petch constant k in terms of how well
grain boundaries are able to sustain stress concentrations. This model suggests that relatively well
ordered special grain boundaries with low energy configurations are also rigid and strong; however,
general boundaries in a more relaxed state are less effective as dislocation barriers.
Although this hardness dependence on grain sizehas been confirmed in both theory and practice
in many metallic materials with grain sizesas small as micrometers, it no longer becomes operative
for some nanocrystalline materials. In recent years, the grain size dependence of hardness in nano-
crystalline materials prepared by various methods (UFP consolidation [ 225,226], electrodeposition
[ 227,228], mechanical attrition [ 229,230], wearing [ 2311, sputtering [ 2321, and crystallization from
amorphous solids) has been studied in many systems.With a decreasing grain size, both hardening
and softening effects have been observed experimentally in either pure metals or intermetallics and
compounds [ 233,234]. There have been several reports on the variation of hardness with grain size
that deviates from the normal Hall-Petch relationship. These include results on nanocrystalline metals
(Cu [ 2251, Pd [ 2251, Ni [ 23.51,etc.), intermetallics (Nb3AI [ 2361 and TiAl [ 2321, etc.), and alloys
(Ni-P [ 44,194] and Fe-based alloys [ 2181) . Some comprehensive reviews [ 204,234] have sum-
marized most of these studies on hardness in nanocrystalline materials.
The hardness dependence on grain sizein a relatively wide grain size range has been determined
experimentally in the crystallized nanocrystalline materials. Fig. 42 shows a collection of these results
for a variety of samples. Different variation behaviors of hardness with grain size have been noticed.
For the Fe-Si-B and Fe-Cu-Si-B nanocrystalline samples,the Hall-Petch plots yield positive values
of k, i.e. they exhibit a normal Hall-Petch behavior for a grain size larger than 25 nm. The Hall-Petch
plot for the pure nanocrystalline trigonal Se yields two distinct positive slopes in grain size ranges of
9-15 nm and 1.5-25 nm. For the Fe-Mo-Si-B nanocrystalline alloy, a critical grain sizefor the normal-
to-abnormal Hall-Petch transition exists in the plot of H,-d- 12 around d=47 nm. In other words,
the normal Hall-Petch relation is only valid for the grain size larger than this critical size, whereas an
abnormal Hall-Petch relation with a negative k is found for a grain size below the critical size. A
similar behavior was also detected in the NiZr, single phase nanocrystalline intermetallic compound,
with a critical grain sizeof about 19 nm, as in Fig. 42. However, for the Ni-P nanocrystalline samples,
an abnormal Hall-Petch relation was obtained for the whole grain size range from a few nanometers
to 120 nm.
K. Lu / Nanocrytalline metals crystallized from amorphous solids: nanocrystallization, structure, and properties 211

12,5

IO,0

8
-3 7,5

2 50

23

otl
w OfI 02 0,3 84
d-/z (m.1/2)
Fig. 42. Hall-Petchplotsfor differentnanocrystalline
samples
crystallizedfrom amorphous
solids.

Several attempts have been made in order to interpret the experimental results on the variation of
hardness with grain size in nanocrystalline materials. However, none of them can yet provide a
satisfactory explanation to all the observed results. Chokshi et al. [ 22.51explain the softening behavior
in nanocrystalline materials by diffusional creep of the nanocrystalline grain boundaries because the
Coble creep rate will be much enhanced when the grain size is reduced to the nanometer regime.
However, direct experimental measurements failed to detect the enhanced creep with a large creep rate
in the nanocrystalline samples. The creep behavior in a nanocrystalline TiAl sample at room temper-
ature was found to follow the Ashby-Verrall model of creep rather than the Coble creep model [ 2371.
Nieh and Wadsworth [ 2381 predicted a critical grain sizebelow which the dislocation pileups cannot
form, so that softening may begin when the grain is smaller than the critical size. Using a dislocation
network model similar to that of Li [ 2221, Scattergood and Koch [ 2391 proposed a critical grain size
below which the dislocation networks are bypassed by moving dislocations, and the nanocrystalline
material will soften. It has also been reported [ 2331 that the triple junction in nanocrystalline materials
may play an important role in softening of the material, since an increased triple junction volume
fraction was reported to result in softening and enhanced bulk ductility in polycrystals [ 2401. Fougere
et al. [ 2411 noticed that the hardening and softening of nanocrystalline materials can depend upon the
method to vary the grain size. This observation supports the explanation of the abnormal Hall-Petch
relation according to the grain boundary densification [ 1441. It was found that with a decreasing grain
size in the nanometer regime, the interfacial excess volume and excessenergy will decrease [48].
Such a densification of interfaces enables them to be less effective (compared with those in coarse-
grained samples) at presenting an obstacle to dislocation motion, and the nanocrystalline sample
softens.
Apart from the sign of the Hall-Petch slope k, there are also reports that the grain size exponent
can be deviated from -0.5 for nanocrystalline materials. Christman [242] presents data for both
coarse- and nano-grained polycrystalline materials to show that the yield strength/hardness variation
with grain sizecan be equally well explained assuming that the grain sizeexponent is - 1, - 0.5, - l/
3, and - l/4. Other investigations have also modified the normal Hall-Petch equation by modifying
either the grain size exponent and/or the slope of the plot k [ 243,244].
Recently, some results for crystallizednanocrystalline materials showed that the grain sizedepend-
ence of hardness can be well correlated with that of the mean positron lifetime [ 2171, or that of the
intensity ratio of II/I2 from the positron annihilation measurements [ 1481.
All of these interpretations have concentrated solely on the effect of interfaces in the nanocrys-
Mine material on the hardness, but have not taken the lattice structure of the nanometer-crystallites
212 K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanoctystallization, stnrctwe, and properties

into account. In fact, any change in the lattice structure of the crystallites would also contribute to the
hardness and modulus of the nanocrystalline materials. Zhang et al. [ 2161 noticed that the degree of
lattice distortion in the crystallized nanocrystalline pure Se samplescan be well correlated to the Hall-
Petch behtivior. It is supposed that softening of the nanocrystalline sample may originate from the
lattice expansion of the nanometer-crystallites, which can also be reflected by the variation of the
lattice modulus with grain size [ 2161.
Although both softening and hardening effects have been detected in the hardness measurements,
the enhanced tensile strength of nanocrystalline samples relative to the coarse-grained ones has been
reported, which may be expected solely from the difficulty in generation and motion of dislocations
[ 2031. The tensile strength of the crystallized nanocrystalline materials has not yet been measured due
to the limited thickness of the ribbon samples. Instead, the fracture properties of the crystallized
nanocrystalline Ni-P alloys with grain sizes ranging from a few nanometers to about 100 nm have
been investigated by means of the bending test at room temperature [ 961. It was found from the
fracture surfaces that the ductility of the nanocrystalline sample is enhanced compared with the coarse-
grained ones. There are visible tracesof plastic deformation in the nanocrystalline sampleswith a grain
size of a few nanometers, whereas a completely brittle fracture surface is seen in the samples with a
100 nm grain size. The fracture stressand the fracture strain were derived as a function of grain size.
Both parameters are increased in approximately linear relations with a reduction of the mean grain
size [ 961. Variations of the fracture stressand strain with the mean grain size are correlated well with
the volume fraction and density of the interfaces in the samples.
Fracture properties of the consolidated nanocrystalline sampleshave been studied in some ceram-
ics (TiO;, [ 2451) and composites (WC-Co [ 2461). However, owing to the presence of porosities
and interfacial phasesin the samples tested, reported results are quite different from different investi-
gations.
The fine grain sizes and the high rate of diffusivity observed in the nanocrystalline materials
suggest that the creep rate can be much enhanced relative to that in the coarse-grained polycrystals. If
grain boundary diffusion dominates the creep process at low temperatures, the creep rate deldt can be
written as [ 2471

(21)

where c is the applied stress,R is the atomic volume, d is the grain size,kB is the Boltzmann constant,
B. is a constant, Db is the grain boundary diffusion coefficient, and A is the grain boundary thickness.
Thus, a reduction in grain size will lead to much enhancement in the diffusion creep rate of a
polycrystalline material.
The creep behaviors of the nanocrystalline Ni-P samples with an average grain size of 28 nm and
of a coarse-grained reference sample (260 nm) were investigated in the temperature range 543 to 593
K [ 2481. It was observed that under the same test condition, the creep rate of the nanocrystalline
sample was much larger than that of the coarse-grained one, asshown in Fig, 43. The apparent activation
energies of creep for the nanocrystalline and the coarse-grained samples are measured to be 0.71 eV
and 1.1 eV respectively. From theseresults it was suggestedthat a grain boundary diffusion mechanism
dominates in the creep process of the nanocrystalline sample, while lattice diffusion is controlling the
creep in the coarse-grained sample.

5. Concluding remarks
Crystallization of amorphous solids serves as a unique method for synthesizing nanocrystalline
materials, which is not only scientifically interesting but also exhibits a great potential for varied
K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 213

1.00

Time {min)

Fig. 43. Creepmeasurementcurvesfor a nanocrystalline


NGP alloy (with an averagegrainsizeof 28 nm,curve I ) anda
coarse-grainedNi-P sample(257 nm,curve 2) at 573K and146MPa [248].

applications. The advantages of this method make it very useful and promising in both fundamental
research and technological studies.
The nanocrystallization process of amorphous solids provides a special phase transformation from
which some inherent properties of nanocrystalline materials (especially the interfaces) may be
obtained. More convincing results would be available from studies of the nanocrystallization process
in a single phase or even an element system. Meanwhile, the nanocrystallization micromechanism
needs further investigations in the future in terms of optimization of the transformation kinetic and
thermodynamic parameters. Control of various influencing factors (such as impurities and gaseous
contamination) in the nanocrystallization process would be helpful for this purpose.
Preparation of large quantities of well-characterized bulk nanocrystalline materials by means of
this method is necessary not only for basic studies but also for widespread use and in the search for
possible technological applications. Bulk nanocrystalline materials might be crystallized from bulk
amorphous precursors, which could be synthesized by consolidation of amorphous ribbons (or pow-
ders) in its viscous flow temperature range to reach a full density, or by searching for some alloy
systems that have relatively high glass forming abilities, so that bulk amorphous materials may be
obtained via conventional casting. Consolidation and in situ nanocrystallization of amorphous powders
(made by ball-milling or via gas atomization) under proper circumstances would be another possible
route for obtaining bulk nanocrystalline materials. There is optimism that attempts in this direction
will be successful for the realization of the practical use of nanocrystalline materials.
Nanocrystalline materials crystallized from amorphous solids possess some specific structural
characteristics which differ from those of the conventional nanocrystalline materials made by using
the consolidation method. Different types of interface and different morphologies of the nanostructures
created during the nanocrystallization of various amorphous solids provide more opportunities for
structure investigations. Nevertheless, some common structural characteristics still exist which might
be intrinsic and representative of the nanocrystalline material. The lattice structure of the nanometer-
sized crystallites seems to change with grain size, and apparent deviations from that of the perfect
crystal lattice are detectable. This feature should be taken into account in further structure and property
studies despite the interface effect. The inherent grain size stability in nanocrystalline materials augurs
well for the future of these new materials, and meanwhile has raised a new topic for investigation.
Both the theory for thermal stability and the effective ways of optimization of the grain growth
parameters in nanocrystalline materials should be dealt with. The nature of destabilization of the
214 K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, strrrcture, and properties

nanostructures seems to be significant for understanding the related solid state phase transformation
processessuch as the amorphization.
Nanocrystallization of amorphous solids makes it possible to compare the properties of the three
different states with the same chemical composition: the nanocrystalline, the amorphous, and the
coarse-grained polycrystalline. Most of the research accomplished to date clearly indicates that the
nanocrystalline materials have properties that are different from and often superior to those of the
amorphous alloys and the conventional coarse-grained polycrystalline materials with the same com-
position.
Although different structural characteristics are obtained in the nanocrystalline materials made
by the crystallization and other methods, their various properties seem to be quite similar. The grain
size dependences of some properties are essentially identical. Unfortunately, no direct comparison for
the same material has been made using different methods. However, the general behavior found so far
for all these samples provides a rather consistent picture.
A major advantage for the property study with the crystallized nanocrystalline samples is that
they are free from porosity and have a wide grain size range in the as-crystallized state. These features
are beneficial to a deep investigation of the structure-property relationship in solids. Therefore, the
crystallization of amorphous solids might be a promising route to create model materials for the
structure-property study, as well as for searching for new nanostructured materials with some specific
novel properties. It would be very useful to have more measurements of various properties with these
nanocrystalline materials in the future.
The majority of investigations on the crystallized nanocrystalline materials to date have dealt
with metallic materials. The possibility to synthesizenanocomposites and nanostructured ceramics via
crystallization of amorphous solids may serve as another concern of materials scientists.Explorations
for a wide application of this method (or its basic ideas) to achieve advanced materials with novel
properties should be considered.

Acknowledgements

This work is financially supported by the Chinese Academy of Sciencesand the National Science
Foundation of China. The author thanks many of his colleagues and students for their contributions to
this work. He appreciates the numerous helpful discussions with Professor U. Kiister, Dr. R. Luck,
Professor J.H. Perepezko, Professor S. Ranganathan, Professor H.-E. Schaefer, Professor R.W. Siegel,
and Professor C. Suryanarayana. The author is also grateful for a fellowship from the Max-Planck
Society of Germany and the hospitality of Professor B. Predel and Dr. R. Liick in Max-Planck-Institut
ftir Metallforschung during preparation of the manuscript in Stuttgart. Those authors who provided
results prior to publications are acknowledged.

Appendix A. Nomenclature

a lattice parameter for a nanocrystalline phase (nm)


a0 equilibrium lattice parameter for the prefect crystal (nm)
B heating rate (K min- )
Bo constant
C lattice parameter for a nanocrystalline phase (nm)
CO equilibrium lattice parameter for the perfect crystal (nm)
K. Lu / Nunocrystulline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties 215

chemical composition of phase i (%)


heat capacity (J n-101- K- )
heat capacity of the amorphous phase (J mol- K-l)
heat capacity of the crystalline phase (J mol- K- )
heat capacity of the nanocrystalline phase (J mol- K-l)
grain size (nm)
average grain size (nm)
initial grain size (run)
creep rate (s- )
gram boundary diffusion coefficient ( m2 s- )
activation energy (J mol- )
volume fraction of the interface in a nanocrystalline material (%)
coercivity (Oe)
hardness ( GPa)
hardness constant (GPa)
steady state nucleation rate ( mm3 s- )
pre-exponent factor for the steady state nucleation rate (me3 s- )
intensity of the short lifetime component in the positron annihilation experiment (%)
intensity of the intermediate lifetime component in the positron annihilation experiment
(So)
intensity of the long lifetime component in the positron annihilation experiment (%)
Hall-Petch slope (GPa m12)
Boltzmann constant (1.38 X 1O-23 J K-)
kinetic parameter dependent on temperature
constant
a frequency term
Loschmidt number
grain growth exponent
saturation magnetization (Gs)
Avrami exponent
activation energy for grain growth (J mol- )
radius of the crystallite (m)
atomic diameter (m)
critical nucleus size (m)
gas constant (8.31 J mol- K-)
time (s)
time constant (s)
finishing time for the isothermal nanocrystallization in DSC measurements (s)
onset time for the isothermal nanocrystallization in DSC measurements (s)
peak time for the isothermal nanocrystallization in DSC measurements (s)
starting time for the isothermal nanocrystallization in DSC measurements (s)
starting time for the isothermal bulk nanocrystallization in MTA measurements (s)
starting time for the isothermal surface nanocrystallization in MTA measurements (s)
finishing time for the isothermal bulk nanocrystallization in MTA measurements (s)
temperature (K)
annealing temperature (K)
Curie temperature (K)
216 K. Lu /Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, and properties

T, melting temperature (K)


Cl starting temperature for grain growth (K)
T Pl first peak temperature for grain growth (K)
T P2 second peak temperature for grain growth (K)
Tf finishing temperature for the nanocrystallization (K)
TCI onset temperature for the nanocrystallization (K)
TP
peak temperature for the nanocrystallization (K)
TS starting temperature for the nanocrystallization (K)
TX crystallization temperature (K)
U crystal growth rate (m s-)
V specific unit cell volume for a nanocrystalline phase ( m3)
VO specific unit cell volume for a perfect crystal lattice ( m3)
vin specific volume for the interface (m3)
Vb specific volume for the perfect crystal ( m3)
x transformed volume fraction (%)
xi atomic fraction of the interface in a nanocrystalline material (%)

Greek letters

a constant relative to the interface thickness (nm)


CyL linear TEC (K- )
4 linear TEE for the crystal (K- )
4 linear TEC for the interface (K- )
cur. linear TEE for the nanocrystalline material (K- )
interfacial energy (J rnm2)
;: interface thickness (nm)
Aa lattice parameter change for a nanocrystalline phase ( [ (a - ao) /ao] X 100%)
A.aL enhanced linear TEIC (c$- cr;) (K-l)
AC lattice parameter change for a nanocrystalline phase ( [ (C-Q) /co] X 100%)
A& excessenergy of the interface (J mol- >
AG Gibbs free energy change (J mol- )
AG- Gibbs free energy change for the amorphous-to-crystalline transformation (J mol- )
AG- Gibbs free energy change for the amorphous-to-nanocrystalline transformation (J mol- *)
AG; formation Gibbs free energy for the amorphous phase (J mol- )
AG: formation Gibbs free energy for the crystalline phase (J mol- )
AG; formation Gibbs free energy for the interface (J mol- )
AG excessGibbs free energy for the amorphous phase ( AGF - AGF) (J mol- )
AG excessGibbs free energy for the interface ( AGi- AG;) (J mol-)
AGc critical free energy required to form the critical nucleus size (J mol- )
A6 Gibbs free energy difference between the crystal and the amorphous phase (J mol- )
AH enthalpy change (J mol- )
AH- enthalpy change for the amorphous-to-crystalline transformation (J mol- )
AH- enthalpy change for the amorphous-to-nanocrystalline transformation (J mol- )
AH; formation enthalpy for the amorphous phase (J mol-)
AH; formation enthalpy for the crystalline phase (J mol- )
AH; formation enthalpy for the interface (J mol- )
K. Lu / Nanocrystakne metals crystalked from amorphous solids: nanocrystallization, structure, and properties 217

AH, enthalpy difference between the amorphous and the crystalline phase at the melting tem-
perature (J mol- )
AH, enthalpy difference between the crystal and the amorphous phase (J mol- )
b driving force for grain growth (J)
AS entropy change (J mol- K-)
AS entropy change for the amorphous-to-crystalline transformation (J mol- K- )
AS- entropy change for the amorphous-to-nanocrystalline transformation (J mol- K-)
AS;-a entropy change for the amorphous-to-nanocrystalline transformation at 0 K (J mol- K- )
AT undercooling (T, - T) (K)
AV unit cell volume change for a nanocrystalline phase ( [ (V- V,) /Vo] X 100%)
A Kn excess volume for the interface ( [ (Vi:,, - Vb) /Vb] X 100%)
vo atomic jumping frequency ( s- I)
P electrical resistivity (pa cm)
PO residual electrical resistivity ($I cm)
(7 applied stress (Pa)
7 incubation time (s)
? mean positron lifetime (ps)
71 the short lifetime component in the positron annihilation experiment (ps)
72 the intermediate lifetime component in the positron annihilation experiment (ps)
73 3 74 the long lifetime components in the positron annihilation experiment (ps)
n atomic volume ( m3)

References

[I] H. Gleiter, Prog. Ma&r. Sci., 33 (1989) 223.


[2] R. Bininger, Mater. Sci. Eng. A, 117 (1989) 33.
[3] J. Karch, R. Birringer and H. Gleiter, Nature, 330 (1987) 536.
[4] R. Bohn, T. Haubold, R. Birringer and H. Gleiter, Ser. Mefull. Muter., 25 (1991) 811.
[5] R.P. Andres, R.S. Averback, W.L. Brown, L.E. BNS, W.A. Goddard III, A. Kaldor, S.G. Louie, M. Moscovits, P.S. Peercy, S.J.Riley,
R.W. Siegel, F. Spaepen and Y. Wang, J. Muter. Res., 4 (1989) 704.
[6] B.H. Kear, R.W. Siegel and T. Tsakalakos (eds.), Nunostrrtcf. Mater., 3 (l-2) (1992).
[7] Proceedings published in B.H. Kear, R.W. Siegel and T. Tsakalakos (eds.), Nunostrrrct. Mater., 3 (1993).
[8] Proceedings published in H.-E. Schaefer, R. Wiirschum, H. Gleiter and T. Tsakalakos (eds.), Nunostntct. Mater., 6 (l-8) (1995).
[9] H. Gleiter, in N. Hansen, A. Horsewell, T. Leffers and H. Lilholt (eds.), Deformation ofPolycrystals: Mechunism andMicrostrucfures,
Riso National Laboratory, Denmark, 1981, p. 15.
[lo] R. Birringer, U. Herr and H. Gleiter, Trans. Jpn. Inst. Met. Suppf., 27 (1986) 43.
[ 1 l] T. Miitschele and R. Kirchheim, Ser. Mefall., 21 (1987) 1101.
[ 121 R.W. Siegel, in M. Nastasi, D.M. Pxkin and H. Gleiter (eds.), Mechanical Properties und Deformarion Behavior of Materials Having
Uifrujne Microsrructures, Kluwer, Dordrecht, 1993, p. 509.
[ 131 H. Gleiter, Nunosfruct. Mater., 1 (1993) 1.
[ 141 J. Jing, A. Kriimer, R. Bininger, H. Gleiter and U. Gonser, J. Non-Crysf. Solids, 113 (1989) 167.
[ 151 H.J. Fecht, E. Hellstem, Z. Fu and W.L. Johnson, Mefall. Truns. A, 21 (1990) 2333.
[ 161 J. Eckert, J.C. Holzer, C.E. Krill III and W.L. Johnson, J. Muter. Res., 7 (1992) 1751.
[ 171 H.J. Fecht, in G.C. Hadjipanayis and R.W. Siegel (eds.), Nanophase Materials, AS1 Series E, Vol. 260, Kluwer, Netherlands, 1994,
p. 125.
[ 181 C.C. Koch, Nanosfruct. Muter., 2 (1993) 109.
[ 191 B.H. Kear and L.E. McCandlish, Nanostruct. Muter., 3 (1993) 19.
[20] R.Z. Valiev, A.V. Korznikov and R.P. Mulyukov, Mater. Sci. Eng. A, 168 (1993) 141.
[21] H. Chang, C.J. Altstetter and R.S. Averback, J. Muter. Res., 7 (1992) 2962.
[22] R.O. Hughes, S.D. Smith, C.S. Pande, H.R. Johnson and R.W. Armstrong, Ser. Mefull., 20 (1986) 93.
[23] U. Erb, A.M. El-She& G. Palumbo and K.T. Aust, Nunosrnrct. Muter., 2 (1993) 383.
[24] A.E. Berkowitz and J.L. Walter, J. Mufer. Rex, 2 (1987) 277.
218 K. Lu /Nanocrystallinemetalscrystallizedfromamorphous
solids:nanocrystallization,structure,andproperties

[25] R. Nagarajan and K. Chattopadhyay,Acta Metall. Mater., 4.2 (1994) 947; A. Cziraki, B. Fogarassy, G. Van Tendeloo, P. Lamparter,
M. Tegze and I. Bakonyi, J. Alloys Camp., 210 (1994) 135.
[26] K. Lu, J.T. Wang and W.D. Wei, J. Appl. Phys., 69 (1991) 522.
[27] R.W. Siegel, in F.E. Fujita (ed.), Physics of New Materials, Springer Series in Materials Sciences, Vol. 27, Springer, Berlin, 1994, p.
65.
[28] H.H. Liebermann, in F.E. Luborsky (ed.), Amorphous Metallic Alloys, Butterworth, London, 1988, p. 26.
[29] H.Y. Tong, J.T. Wang, B.Z. Ding, H.G. Jiang and K. Lu, .I. Non-Cryst. Solids, 150 (1992) 444.
[30] X.D. Liu, J.T. Wang and B.Z. Ding, Ser. Metal/. Mater., 28 (1993) 59.
[31] M.M. Nicolaus, H.-R. Sinning and F. Haessner, Mater. Sci. Eng. A, 150 (1992) 101.
[32] H.Q. Guo, T. Reininger, H. Kronmiiller, M. Rapp, V.Kh. Skumrev, Phys. Status Solidi A, 127 (199 1) 519.
[33] T. Spassov and U. Koster, J. Mater, Sci., 28 (1993) 2789.
[ 341 U. Koster and J. Meinhardt, Mater. Sci. Eng. A, 178 ( 1994) 271.
[35] X.D. Liu and J.T. Wang, Nanostruct. Mater., 2 (1993) 63.
[36] X.D. Liu, B.Z. Ding, Z.Q. Hu, K. Lu and Y.Z. Wang, Physica B, 192 (1993) 345.
[37] C. Beeli, H.-U. Nissen, Q. Jiang and R. Luck, Mater. Sci. Eng. A, 133 (1991) 346.
[38] K. Lu, X.D. Liu, F.H. Yuan and W.D. Wei, Physica E, (1996) in press.
[39] N. Zarubova, N. Moser and H. Kronmiiller, Mater. Sci. Eng. A, 151 (1992) 205.
[40] C.U. Maier and H. Kronmiiller, 2. Mefallkd., 83 (1992) 839.
[41] Y.L. He and X.N. Liu, Acta Electron. Sinica, 4 (1982) 70 (in Chinese).
[42] H.Y. Zhang, Z.Q. Hu and K. Lu, Nanostruct. Muter., 5 (1995) 41.
[43] H.Y. Zhang, K. Lu and Z.Q. Hu, Actu Phys. Sinica, 44 (1994) 109.
[44] K. Lu, W.D. Wei and J.T. Wang, Ser. Metall. Mater., 24 (1990) 2319.
[45] M.L. Sui, L.Y. Xiong, W. Deng, K. Lu, S. Patu and Y.Z. He, J. Appl. Phys., 69 (1991) 4451; Mater, Sci. Forum, 105-110 (1992)
1249.
[46] K. Lu, R. Luck and B. Predel, J. Alloys Camp., 201 ( 1993) 229.
[47] K. Lu, R. Luck and B. Predel, Acta Metall. Mater., 42 (1994) 2303.
[48] K. Lu, R. Luck and B. Predel, Ser. Metall. Mater., 28 (1993) 1387.
[49] K. Lu, in M.A. Otooni (ed.), Science and Technology of Rapid Solidt$cation and Processing, Kluwer, Netherlands, 1995, p* 349.
[50] K. Lu, in S. Banerjee and R.V. Ramanujan (eds.), Advances in Physical Metallurgy, Gordon and Breach, New York, 1995, p, 135.
[51 J Y.H. Kim, A. Inoue and T. Masumoto, Muter. Trans. JIM, 32 (1991) 331.
[52] K. Suzuki, N. Kataoka, A. Inoue, A. Makino and T. Masumoto, Muter. Trans. JIM 31 (1990) 743.
[53] Y. Yoshizawa, S. Oguma and K. Yamauchi, J. Appl. Phys., 64 (1988) 6044.
[54] Y. Yoshizawa and K. Yamauchi, Mater. Trans. JIM, 31 (1990) 307.
[55] G. Herzer, IEEE Trans. Magn., 25 (1989) 2227; IEEE Trans. Magn., 26 (1990) 1387; Mater. Sci. Eng. A, 132 (1991) 1.
[56] I. Matko, P. Duhaj, P. Svec and D. Janickovic, Mater. Sci. Eng. A, 179-180 (1994) 557.
[57] C.F. Conde, M. Millian and A. Conde, J. Magn. Mugn. Mater., 138 (1994) 314.
[58] K. Hono, K. Hiraga, Q. Wang, A. Inoue and T. Sakurai, Acta Metall. Muter., 40 (1992) 2173.
[59] M.G. Scott, in F.E. Luborsky (ed.), Amorphous Metallic Alloys, Butterworth, London, 1988, p. 144.
[60] N. Azam, L. Lenaour, C. Rivera, P. Grosjean, P. Sacovyand J. Delaplace, J. Nucl. Mater., 83 ( 1979) 298.
[61] M.L. Trudeau, R. Schulz, D. Dussault and A. Van Neste, Phys. Rev. Lett., 64 (1990) 99.
[62] U. Kijster and P. Weiss, J. Non-Cryst. Solids, 17 (1975) 359.
[63] U. Koster and U. Herold, in H.J. Giintherodt and H. Beck (eds.), Glassy Metals I, Springer, Heidelberg, 1981, p, 225.
[64] The free energy of a nanocrystalline phase might be relatively higher than that of the corresponding equilibrium crystalline state due
to the contribution of the large amount of disordered grain boundaries.
[65] Z.F. Dong, Y.H. Ma and K. Lu, Ser. Metail. Mater., 31 (1994) 81.
[66] X.D. Liu, Ph.D. Thesis, Institute of Metal Research,Academia Sinica, 1994.
[67] U. Koster, U. Herold and A. Becker, in T. Masumoto and K. Suzuki (eds.), Proc. 4th Int. Conf: on Rapidly @enched Metals, Japan
Institute of Metals, Sendai, 1982, p. 587.
[68] M.L. Sui, K. Lu and Y.Z. He, Philos. Mug. B, 63 (1991) 933.
[69] Z.F. Dong, K. Lu, W.D. Wei and B.Z. Ding, Actu Metall. Sinica, 30 (1994) B304 (in Chinese).
[70] T. Watanabe and M.G. Scott, J. Muter, Sci., 15 (1980) 1131.
[71] K. Lu, M.L. Sui and J.T. Wang, J. Muter. Sci. Lett., 9 (1990) 630.
[72] U. Koster, J. Meinhardt and H. Alves, in Proc. of ISMANAN94, Grenobel, France, 1994, in press.
[73] Z. Altounian, E. Batalla and J.O. Strom-Olson, J. Appl. Phys., 61 (1987) 149.
[74] M.V. Nevitt and J.W. Downey, Trans. AIME, 221 (1961) 1014.
[75] T. Kulik, T. Horubala and H. Matyja, Mater. Sci. Eng. A, 157 (1992) 107.
[76] P. Gonia, I. Orue, F. Plazaola and J.M. Barandiaran, J. Appl. Phys., 73 (1993) 6600.
[77] M. Kopcewicz, E. Jackiewicz, L. Zaluski and A. Zaluska, J. Appl. Phys., 71 ( 1992) 3997.
[78] U. KGster, U. Schiinemann, M. Blank-Bewersdorf, S. Brauer, M. Sutton and G.B. Stephenson, Mater. Sci. Eng. A, I33 (1991) 611.
[79] L. Voropaeva, A. Serebryakov, N. Novokhatskaya, Y. Levin and G. Abrosimova, Ser. Metall. Mater., 27 (1993) 1385.
[80] M.L. Trudeau, J.Y. Huot, R. Schulz, D. Dussault, A. Van Neste and G. LEsperance, Phys. Rev. B, 4.5 (1992) 4626.
K. Lu / Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallization, structure, andproperties 219

[81] F.Q. Guo and K. Lu, Nanostrnct. Mater., 3 (1996) in press.


[ 821 H.P. Klug and L.E. Alexander, X-ray Diffraction Proceduresfor Polycrystalline and Amorphous Materials, Wiley, New York, 1974.
[83] D.G. Morris,ActaMetull., 29 (1981) 1213.
[ 841 A.L. Greer, Actu Metall., 30 (1982) 171.
[85] H.Y. Tong, B.Z. Ding, H.G. Jiang, K. Lu, J.T. Wang and Z.Q. Hu, J. Appl. Phys., 75 (1994) 654.
[ 861 K. Lu, Acta Merall. Sin&, 30 ( 1994) Bl (in Chinese).
[87] U. KGster, R. Abel and H. Blanke, Glustech. Ber., K56 (1983) 584.
[88] U. Ktiter, in G.W. Lorimer (ed.), Proc. on Phase Transformation 87, Institute of Metals, Cambridge, 1987, p, 597.
[89] H.Y. Tong, MSc. Thesis, Institute of Metal Research, Academia Sinica, 1993.
[90] X.D. Liu, K. Lu, B.Z. Ding and Z.Q. Hu, Chin. Sci. Bull., 39 (1994) 217.
[91] P.G. Boswell and G.A. Chadwick, So: Metall., 10 (1976) 509.
[92] F. Zhou and K.Y. He, J. Mater. Sci. Techn., 10 (1994) 430.
[93] K. Lu, Phys. Review B, 51 (1995) 18.
[94] K. Lu, R. Liick and B. Predel, J. Non-Cryst. Solids, 156-158 (1993) 589.
[95] K. Lu, J.T. Wang and W.D. Wei, Ser. Metall. Mater., 25 (1991) 619.
[96] M.L. Sui, S. Patu and Y.Z. He, Ser. Me&l. Muter., 25 (1991) 1537.
[97] F. Sommer, Mater. Sci. Eng. A, 133 (1991) 434.
[98] D. Tumbull, J. Appl. Phys., 21 (1950) 1022.
[99] D.R.H. Jones and G.A. Chadwick, Philos. Mug., 24 (1971) 995.
[ 1001 C.V. Thompson and F. Spaepen, Acta Metall., 27 (1979) 1855.
[ 1011 L. Battezzati and E. Garrone, Z. Metallkd., 75 (1984) 305.
[102] KS. Dubey and P. Ramanchandrarao, Acta Metall., 32 (1984) 91.
[ 1031 H.J. Fecht, Acta Metall. Mater., 38 (1990) 1927.
[ 1041 H.J. Fecht, Phys. Rev. Lett., 65 (1990) 610.
[ 1051 L.A. Gerifalco and V.G. Weizer, Phys. Rev., 114 (1959) 687.
[ 1061 M. Wagner, Phys. Rev. B, 45 (1992) 635.
[ 1071 D. Wolf, Philos. Mug. B, 59 (1989) 667.
[ 1081 M. Avrami, J. Chern. Phys., 7 (1939) 1103; J. Chern. Phys., 8 (1940) 212; J. Chern. Phys., 9 (1941) 177.
[ 1091 J.W. Christian, The Theory of Transfortnation in Metals and Alloys, Pergamon, Oxford, 1975.
[ 1lo] R. Ltick, K. Lu and W. Frantz, Ser. Metall. Muter., 28 (1993) 1071.
[ 11 l] K. Lu and J.T. Wang, Ser. Metall., 21 (1987) 1185.
[ 1121 A.L. Greer, in S. Steeb and H. Warlimont (eds.), Rapidly Quenched Met&, North-Holland, Amsterdam, 1985, p. 215.
[ 1131 K. Lu and J.T. Wang, Mater. Sci. Eng., 97 (1988) 399.
[ 1141 F.H. Yuan, ML%. Thesis, Institute of Metal Research, Academia Sinica, 1994.
[ 1151 K. Lu and J.T. Wang, .I. Non-Crysr. Solids, 117-118 (1990) 716.
[ 1163 I.R. Gabel and M.M. Nicolaus, Thertnochim. Acta, 188 (1991) 95.
[ 1171 H.E. Kissinger, Anal. Chem., 29 (1957) 1702.
[ 1181 K. Lu and J.T. Wang, Sci. China A, 35 (1992) 1266.
[ 1191 U. KGster and U. Schiinemann, in H.H. Liebermann (ed.), Rapidly Solidified Alloys, Marcel Dekker, New York, 1993, p. 303.
[ 1201 K. Lu, Ph.D. Thesis, Institute of Metal Research, Academia Sinica, 1989.
[ 1211 B. Cantor, in S. Steeb and H. Warlimont (eds.), Rapidly Quenched Metals, North-Holland, Amsterdam, 1985, p. 595.
[ 1223 B. Cantor and R.W. Cahn, in F.E. Luborsky (ed.), Amorphous Metallic Alloys, Butterworth, London, 1988, p. 487.
[ 1231 K. Lu and J.T. Wang, J. Crystal Growth, 112 (1991) 525; K. Lu, M.L. Sui and J.T. Wang, J. Crystal Growth, 113 (1991) 242.
[ 1241 K. Lu and J.T. Wang, Muter. Sci. Eng. A, 133 (1991) 504.
[ 1251 K. Lu and J.T. Wang, J. Crystal Growth, 97 (1989) 448.
[ 1261 C. Suryanarayana and F.H. Froes, Metall. Trans. A, 23 (1992) 1071.
[ 1271 X. Zhu, R. Birringer, U. Herr and H. Gleiter, Phys. Rev. B, 35 (1987) 9085.
[ 1281 U. Herr, J. Jing, R. Birringer, U. Gonser and H. Gleiter, Appl. Phys. Lett., 50 ( 1987) 472.
[ 1291 H.E. Schaefer and R. Wiirschum, Phys. Lett. A, 119 (1987) 370.
[ 1301 H.E. Schaefer, R. Wiirschum, R. Birringer, H. Gleiter, Phys. Rev. B, 38 (1988) 9545.
[131] A.P. Webb, J. Phys. C:, 14 (1981) 259.
[132] T. Haubold, R. Bininger, B. Lengeler and H. Gleiter, J. Less-Cornm. Met., 145 (1988) 557.
[133] T. Haubold, R. Birringer, B. Lengeler and H. Gleiter, Phys. Lett. A, 135 (1989) 461.
[ 1341 G.J. Thomas, R.W. Siegel and J.A. Eastman, Ser. Metall. Mater., 24 (1990) 201.
[ 1351 W. Wunderlich, Y. Ishida, R. Maurer, Ser. Metall. Mater., 24 (1990) 403.
[136] D.X.Li, D.H. Ping,H.Q. Ye,X.Y. QinandX.J. Wu, Mater.Lett., 18 (1993) 29.
[ 1371 S.K. Ganapathi and D.A. Rigney, Ser. Metall. Mater., 24 (1990) 1675.
[138] D.H. Ping, D.X.Li,H.Q. Ye,X.D.LiuandZ.Q. Hu,Mater. Sci.Eng.,A194 (1995) 211.
[ 1391 D.H. Ping, D.X. Li and H.Q. Ye, Nanostruct. Muter., 5 (1995) 457.
[ 1401 K. Lu, W.D. Wei and J.T. Wang, J. Appl. Phys., 69 (1991) 7345.
[ 1411 M.L. Sui and K. Lu, Actu Metall. Sinica, 30 (1994) B121 (in Chinese).
220 K. Lu /Nanocrystalline metals crystallized from amorphous solids: nanocrystallization, strwture, and properties

[ 1421 M.L. Sui, K. Lu, W. Deng, L.Y. Xiong, S. Patu and Y.Z. He, Phys. Rev, B, 44 (1991) 6466,
[ 1431 CD. Terwilliger and Y.M. Chiang, MRS Sytnp. Proc., 289 (1993) 15.
[ 1441 K. Lu and M.L. Sui, Ser. Metall. Mater., 28 (1993) 1465.
[ 1451 K. Lu, Nanostruct. Mater., 2 (1993) 643.
[ 1461 R. Liick and K. Lu, J. Mater. Sci. Technol., 11 (1994) 157.
[ 1471 H.Y. Tong, B.Z. Ding, J.T. Wang, K. Lu, J. Jiang and J. Zhu, J. Appl. Phys., 72 (1992) 5124.
[ 1481 X.D. Liu, J.T. Wang, D.H. Ping and D.X. Li, J. Appl. Phys., 74 (1993) 4501.
[ 1491 X.D. Liu, J. Zhu, J. Jiang and J.T. Wang, J. Alloy Camp., 198 (1993) 85.
[ 1501 R. Wiirschum, W. Greiner, R.Z. Valiev, M. Rapp, W. Sigle, 0. Schneeweiss and H.E. Schaefer, Ser. Metall. Mater,, 25 (1991) 2451.
[ 1511 H.-E. Schaefer, in M. Nastasi, D.M. Parkin and H. Gleiter (eds.), Mechanical Properties and Deformatiorl Llehavior of Materials
Havirzg Ultra@ Microstructures, Kluwer, Dordrecht, 1993, p. 81.
[152] T. Miitschele and R. Kirchheim, Ser. Metall. Mater., 21 (1987) 135.
[ 1531 H. Hahn, H.J. HGfler and R.S. Averback, Dejecto@ Fonrm, 66-69 (1989) 549.
[ 1541 J. Eckert, J.C. Holzer, C.E. Krill III and W.L. Johnson, J. Mater. Sci., 7 (1992), 1751.
[ 1551 E. Gaffet, C. Louison, M. Harmelin and F. Faudet, Mater. Sci. Eng. A, 134 (1991) 1410.
[ 1561 K. Lu, M.L. Sui and R. Liick, Nanostrnct. Mater., 4 (1994) 465.
[ 1571 X.D. Liu, K. Lu, Z.Q. Hu, B.Z. Ding, J. Zhu and J. Jiang, J. Appl. Phys., 75 (1994) 3365.
[ 1581 X.D. Liu, J. Zhu, K. Lu, Z.Q. Hu and B.Z. Ding, Nanostrrtct. Mater., 2 (1993) 571.
[ 1591 X.D. Liu, J.T. Wang, K. Lu, J. Zhu and J. Jiang, J. Phys. D:, 27 (1994) 165.
[160] M.L. Sui and K. Lu, Mater. Sci. Eng. A, 179-180 (1994) 541.
[ 1611 M.L. Sui and K. Lu, Acta Metall. Sinica, 30 (1994) B413.
[ 1621 X.D. Liu, K. Lu, Z.Q. Hu and B.Z. Ding, Nanostrrtct. Mater., 2 (1993) 581.
[ 1631 H.Y. Zhang, Z.Q. Hu and K. Lu, Nanostntct. Mater., 6 (1995) 489; J. Phys.: Condens. Matter., 7 (1995) 5327,
[ 1641 E. Gaffet and M. Harmelin, J. Less-Comm. Met., 157 (1991) 201.
[ 1651 D. Kunstelj, A. Kirin and A. Bonefacic, in S. Steeb and H. Warlimont (eds.), Rapidly Quenched Metals, North-Holland, Amsterdam,
1985, p. 899.
[ 1661 N. Shen, I.P. Jones and J.N. Pratt, in T. Masumoto and K. Suzuki (eds.), Proc. 4th Int. Cor$ on Rapidly &crxhed Metals, Japan
Institute Metals, Sendai, 1981, p. 1553.
[ 1671 B.M. Patterson, K.M. Unruh and S.I. Shah, Nanostruct. Mater., 1 (1992) 65.
[ 1681 M. Jose-Yacaman, unpublished results, (1993).
[ 1691 X.D. Liu, K. Lu, H.Y. Zhang and Z.Q. Hu, J. Phys.: Condens. Matter, 6 (1994) L497.
[ 1701 K. Lu and M.L. Sui, J. Mater. Sci. Technol., 9 (1993) 419.
[ 1711 D. Wolf and J.F.Lutsko, Phys. Rev. Lett., 60 (1988) 1170.
[ 1721 J.A. Eastman, M.R. Fitzsimmons and L.J. Thompson, Philos. Msg., 66 (1992) 667.
[ 1731 M.Ya. Gamamik, Phys. Stalus Solidi B, I68 (1991) 389.
[ 1741 M.Ya. Gamarnik and Y.Y. Sidorin, Phys. Starus Solidi B, 156 (1989) Kl.
[ 1751 D. Schroeer and R.C. Nininger, Jr., Phys. Rev. Let?., 19 (1967) 632.
[ 1761 P.G. Shewmon, Transformation in Metals, McGraw-Hill, New York, 1969, p. 300.
[ 1771 C. Suryanarayana and F.H. Froes, Nanosttxct. Mater., 1 (1992) 196.
[ 1781 R.W. Siegel, S. Ramasamy,H. Hahn, Z. Li, T. Lu and R. Gronsky, J. Mater. Res., 3 (1988) 1367.
[ 1791 K. Lu, Ser. Metall. Mater., 25 (1991) 2047.
[ 1801 B. Giinther, A. Kumpmann and H.-D. Kunze, Ser. Metall. Mater., 27 (1992) 833.
[ 1811 A. Kumpmann, B. Giinther and H.-D. Kunze, Mater. Sci. Eng. A, 168 (1994) 165.
[182] P. Knauth, A. Charai and P. Gas, Ser. Mefall. Mater., 28 (1993) 325.
[ 1831 A. Cziraki, Zs. Tonkovics, I. Gerocs, B. Fogarassy, I. Groma, E. Toth-Kadar, T. Tarnoczi and I. Bakonyi, Mater. Sci, Eng. A, 179-
180 (1994) 531.
[184] J. Eckert, J.C. Holzer and W.L. Johnson, J. Appl. Phys., 73 (1993) 131.
[ 1851 L.C. Chen and F. Spaepen, J. Appl. Phys., 69 (1991) 679.
[ 1861 T. TschGpe, R. Bininger and H. Gleiter, J. Appl. Phys., 71 (1992) 5391.
[ 1871 C.H. Moelle and H.J. Fecht, Nanostruct. Mater., 6 (1995) 421.
[ 1881 K. Lu and Z.F. Dong, in preparation.
[ 1891 K. Lu, J.T. Wang and W.D. Wei, J. Phys. D:, 25 (1992) 808.
[ 1901 K. Lu and M.L. Sui, Acta Metall. Mater., 43 (1995) 3325.
[ 1911 H.Y. Tong, B.Z. Ding, H.G. Jiang, Z.Q. Hu, L. Dong and Q. Zhou, Mater. Lett., 16 (1993) 260.
[ 1921 K. Lu, Z.F. Dong, I. Bakonyi and A. Cziraki, Acta Metall. Mater., 43 (1995) 2641.
[ 1931 K. Isonoshi and K. Okazaki, J. Mater. Sci., 28 (1993) 3829.
[ 1941 G. Palumbo, U. Erb and K.T. Aust, Ser. Metall. Mater., 24 (1990) 2347.
[ 1951 K. Boylan, D. Ostrander, U. Erb, G. Palumbo and K.T. Aust, Ser. Metall. Mater., 25 (1991) 2711.
[ 1961 Z.F. Dong, K. Lu and R. Ltick, unpublished results, (1995).
[ 1971 S. Kawanishi, K. Isonishi and K. Okazaki, Mater. Trans. JIM, 34 (1993) 49.
[ 1981 S.K. Ganapathi, D.M. Owen and A.H. Chokshi, Ser. Metall. Mater., 25 (1991) 2699.
K. Lu/Nanocrystalline metals crystallizedfrom amorphous solids: nanocrystallizarion, structure, andproperties 221

[ 1991 H.J. Hiifler and R.S. Averback, Scr: Metall. Mater., 24 (1991) 2401.
[200] H. Hahn, J. Logas and R.S. Averback, 1. Muter. Rex, 5 (1990) 609.
[201] R.S. Averback, H.J. Hofler and R. Tao, Murer. Sci. Eng. A, I66 (1993) 169.
[202] J.A. Eastman, Y.X. Liao, A. Narayanasamy and R.W. Siegel, MRSSymp. Proc., 155 (1989) 1246.
[203] G.W. Nieman, J.R. Weertman and R.W. Siegel, J. Mater. Res., 6 (1991) 1012.
[204] R.W. Siegel and G.E. Fougere, in G.C. Hadjipanayis and R.W. Siegel (eds.), Nanophase Materials, Kluwer, Netherlands, 1994, p.
233.
[205] H. Gleiter, Nunostruct. Mater., 6 (1995) 3.
[206] J. Rupp and R. Bininger, Phys. Rev. B, 36 (1987) 7888.
[207] E. Hellstem, H.J. Fecht, Z. Fu and W.L. Johnson, J. Appl. Phys., 65 (1989) 305.
[208] K. Lu, R. Luck and B. Predel, 2. Metullkd., 84 (1993) 740.
[209] R. Birringer and H. Gleiter, in R.W. Cahn (ed.), Advances in Materials Science, Encyclopedia of Materials Science andEngineering,
Pergamon, London, 1988, p. 399.
[210] H.J. Klam, H. Hahn and H. Gleiter, Acra Metall., 35 (1987) 2101.
[211] I. Bakonyi, E. Toth-Kadar, T. Tarnoczi, L. Varga, A. Cziraki, I. Gerocs and B. Fogarassy,Nunostrncf. Muier., 3 (1993) 155.
[212] K. Lu, Y.Z. Wang, W.D. Wei and Y.Y. Li, Adv. Cjyog, Mater., 38 (1991) 285.
[213] Y.Z. Wang, G.W. Qiao, X.D. Liu, B.Z. Ding and Z.Q. Hu, Muter. Left., 17 (1993) 152.
[214] I. Bakonyi, E. Toth-Kadar, J. Toth, A. Cziraki and B. Fogarassy,in G.C. Hadjipanayis and R.W. Siegel (eds.), Nunophase Murerials,
ASI Series E, Vol. 260, Kluwer, Netherlands, 1994, p. 423.
[215] R.Z. Valiev, R.R. Mulyukov, K.Y. Mulyukov, V.I. Novikov and L.I. Trusov, Pisma Sz. Teckhn. Fiz., 15 (1989) 78.
[216] H.Y. Zhang, Z.Q. Hu and K. Lu, J. Appl. Phys., 77 (1995) 2811; K. Lu, H.Y. Zhang, Y.D. Zhong and H.J. Fecht, in preparation.
[217] B.Z. Ding, H.Y. Tong, H.G. Jiang, J.T. Wang and W.D. Wei, Ser. Metall. Mater., 28 (1993) 1107.
[218] X.D. Liu, J.T. Wang, Z.Q. Hu and B.Z. Ding, Mater. Sci. Eng. A, 169 (1993) L17.
[219] E.O. Hall, Proc. Phys. Sot. London Sect. B, 64 (1951) 747.
[220] N.J. Petch, J. Iron Sreel Inst., 174 (1953) 25.
[221] A.H. Cottrell, Trans. Metal/. Sot. AIME, 212 (1958) 192.
[222] J.C.M. Li, Trans. Metall. Sot. AIME, 227 (1963) 239.
[223] H. Conrad, Acta Metall., 11 (1963) 75.
[224] J.W. Wyrzykowkski and M.W. Grabski, Philos. Mug. A, 53 (1986) 505.
[225] A.H. Chokshi, A. Rosen, J. Karch and H. Gleiter, Ser. Mefall. Mater., 23 (1989) 1679.
[226] G.W. Nieman, J.R. Weertman and R.W. Siegel, Ser. Metall. Mater., 23 (1990) 2013.
[227] A.M. El-She& U. Erb, G. Palumbo and K.T. Aust, Ser. Metall. Muter., 27 (1992) 1185.
[228] C. Cheung, G. Palumbo and U. Erb, Ser. Mefall. Muter., 31 (1994) 735.
[229] J.S.C. Jang and CC. Kock, So: Metall. Muter., 24 (1990) 1599.
[230] P. Le Brun, E. Gaffet, L. Froyen and L. Delay, Ser. Metall. Mater., 26 (1992) 1743.
[231] SK. Ganapathi, M. Aindow, H.L. Fraser and D.A. Rigney, MRSSynzp. Proc., 206 (1991) 593.
[232] H. Chang, H.J. HGAer, C.J. Altstetter and R.S. Averback, Ser. Metall. Muter., 25 (1991) 1161.
[233] C. Suryanarayana, D. Muknopsfhysy, S.N. Psysnkstand F.H. Froes, J. Muter. Res., I7 (1993) 2114.
[234] R.W. Siegel and G.E. Fougere, Nunostrnct. Mater., 6 (1995) 205.
[235] T. Christman and M. Jain, Ser. Metull. Muter., 24 (1990) 1599.
[236] K. Kim and K. Okasaki, Muter. Sci. Forrun, 88-90 (1992) 553.
[237] H. Chang, C.J. Altstetter and R.S. Averback, J. Mater. Res., 7 (1992) 2962.
[238] T.G. Nieh and J. Wadsworth, Ser. Metull. Mater., 2.5 (1991) 955.
[239] R.O. Scattergood and CC. Koch, So: Metall. Mater., 27 (1992) 1195.
[240] V.B. Rabukhin, Phys.Met. Metall., 61 (1986) 149.
[241] G.E. Fougere, J.R. Weertman, R.W. Siegel and S. Kim, Ser. Metall. Mater., 26 (1992) 1879.
[242] T. Christman, Ser. Mend/. Muter., 28 (1993) 1495.
[243] V.G. Gryaznov, M.Y. Gutkin, A.E. Romanov and L.I. Trusov, J. Muter. Sci., 28 (1993) 4359.
[244] S. Li, L. Sun and Z.G. Wang, Nunostrnct. Mater., 2 (1993) 653.
[245] J. Karch and R. Bininger, Cerum. Int., I6 (1992) 291.
[246] L.E. McLandish, B.H. Kear and B.K. Kim, Nanostruct. Mater., 1 (1992) 119.
[247] R.L. Coble, J. Appl. Phys., 34 (1963) 1679.
[248] D.L. Wang, Q.P. Kong and J.P. Shui, Ser. Mefall. Mafer., 31 (1994) 47.

S-ar putea să vă placă și