Sie sind auf Seite 1von 54

Contents of the section:

Theory of glass formation and glassy stae


J. estk,
Termochimica Acta 95 (1986) 459-471


C.A. Queiros, J. estk,
Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B, June 2010, 51 (3), 165172


Boivoj Hlavek, Jaroslav estk, ladislav Koudelka
Journal of Thermal Analysis and Calorimetry, Vol. 80 (2005) 271283


Ana Kozmidis-Petrovic, Jaroslav estk,
J Therm Anal Calorim (2012) 110:9971004


Jaroslav estk, Ana Kozmidis-Petrovic, ivan ivkovi
J. Min. Metall. Sect. B-Metall. 47 (2) B (2011) 229 - 239
~ermochimieo Actu. 95 (1986) 459-471 459
Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands


Institute of Physics of the Czechoslovak Academy of Sciences, Na Slovance 2,
180 40 Prague 8, Czechoslovakia

The classification, identificationand stability of a non-crystallinema-
terial as a glass is discussed. General thermodynamic and compositional con-
sideration are presented. Thermal characterizationof vitrification processes
are specified. A survey is given of the theoretical modelling of thermody-
namic descriptions of glass transformation. The thermodynamic functions of
undercooled liquids during vitrification are explained.

A traditional definition of glasses involves the natural cooling (about

10m2 K s-I) of melts, usually silicates. If other substances are subjected to
sufficiently rapid cooling, for example water cooling of chalcogenides (102 K
s-I), melt spinning of metallic alloys (106 K s-I) or even vapour deposition
of various organic (alcohol)or inorganic (water) compounds (IOIO K s-I),
glass-like materials are similarly obtained, extending tremendously the amount
of data on the formation of glass. Previously such studies mainly have been
of academic interest but now find a wider use because of a growing number of
practical applications. Such studies were previously often related to the
field of the researcher's individual interest but they are now loosing their
original character of academic curiosity because of a grohing number of prac-
tical applications. The different kinds of glassy materials that have been
studied extensively include ionic conductor inorganic salts, semiconductor
chalcogenides,metallic alloys and various organic polymers. The study of
glassy materials focuses on the structural, kinetic and thermodyn~ic aspects,
the latter being the subject of this paper.


A great variety of non-crystalline solids have recently been prepared by a
number of distinct but unconventionalmethods. The methods employed include

0049-6031/86/$03.30 0 1986 Elsevier Science Publishers B.V.


unusually extreme quenching of melts or vapours and/or electrodepositions,

disintegrations, or use of high energies. A disordered distribution of atoms
is obtained by freezing melts or as a result of other disordering processes
involved in the amorphous material formation. The latter processes usually
produce states that are poorly characterized as well as difficult to charac-
terize. That is why non-crystalline solids of an identical composition ob-
tained by various techniques, or by the same technique under different condi-
tions, can differ considerably in their physicochemical properties. A more
precise identification thus becomes difficult since certain parameters, like
local cooling rate and associated gradients, topological and compositional
short-range order, etc., cannot be specified exactly and/or directly. Thus, it
is evident that different names are currently used to describe such materials.
Eckstein (ref.1) classifies non-crystalline solids as vitroids referring to
organic and inorganic materials in the vitreous state, including glasses and
plastics. According to Roy (ref.2) the concept of non-crystallinity (meaning
structurally disordered) is considered hierarchially superior to the terms
glassy and amorphous. (The latter is currently more popular to describe the
newer types of semiconductor and metallic materials.) Many materials are con-

sidered amorphous if they do not exhibit an experimentally detectable glass

transformation point which, however, can be hidden in the early process of
crystallization catalyzed in such cases by excessively active sites or inter-
faces. A small number of vapour-deposited amorphous materials can even exhib-
it a pseudo glass transformation (relaxing to the state obtainable by cooling
of the corresponding liquid to the same temperature) and then immediately cry-

stallize. The specific class of non-crystalline solids, which we herewith

signify by the term glass, is understood to be preparable in a more systematic
and reproducible fashion compared to others, i.e, by cooling melts at differ-
ent quenching rates with the possibility of subsequent heating into the under-
cooled liquid state without overlapping crystallization (refs.3,4).

Another problem is the stability definition of a glassy state. According

to Suga and Seki (ref.';) the glassy state is a thermodynamically unstable one
which can be assumed to be an undercooled metastable state at a certain in-
stant of continuous cooling by the glass transformation process which is con-

sidered here as a general property of all metastable phases. Therefore, they

(ref.5) recommended even further classification of general glasses into glassy
liquids, glassy liquid crystals and glassy crystals formed by the glass trans-
formation of the respective metastable states of liquids (melts) and liquid
and solid crystalline phases. The state of a cooled material can be charac-
terized thermodynamically with respect to its corresponding stable phase down
to the glass transformation temperature, at which a state of internal equilib-

rium is lost, and even lower with some uncertainty due to the irreversible
nature of the glass transformation phenomena. Very thin sections of usually
spalt-cooled glasses are sometimes difficult to characterize precisely because
of the extreme condition of quenching and the irreversibility associated with
their formation. Some of them can even crystallize during reheating before
any glass transformation phenomena are observed. However, they are not to be
excluded from our somewhat strict classification of glasses, even though, the
dividing line between glasses and less respectable amorphous solids is vague,
particularly when compositions exhibiting or not exhibiting a glass transfor-
mation can overlap.
In recent literature, the glassy state is often called a metastable state
which, however; contradicts the common thermodynamic point of view that the
metastable state can exist in the stability region of the other neighbouring

phase and into which the extrapolation of the Gibbs energy function is pos-
sible, i.e., in the vicinity of the first-order transitions, such as melting
or crystallization, where a solid can be superheated or, more likely, a melt
undercooled, see Fig.1. The glass transformation, however, assumes second-
order-like transition, where each phase can exist only on its own stability
side because there is a break in the second derivatives of the Gibbs energy
(e.g. specific heat Cp) while its first derivative (e.g. entropy S) remains
continuous. On the other hand, however, the physical appearance of a glassy
solid looks more stable than that of an undercooled melt, the latter being
more easily transformable to the nearest state of stable crystals by a slight
action to surmount the energy barrier to nucleation. Hence, we come to the
discussion of the term solids in terms of vitroids within the framework of
rheology. This is because such a solid changes with time and observation time
is involved in detecting the extent of change. Reiner (ref.6) introduced the
Deborah number* expressing the ratio between the time of material relaxation
and that of its observation (DN). Conveniently, we can demonstrate the insta-
bility of a glassy state using the textbook case of a. variously positioned
brick, cf. Fig.1. When a brick is standing on its edge (and/or lying in an
unstable position on a very steep slope) its motion is determined by the high
viscosity of the surrounding medium (or by a large coefficient of friction of
the roughened surface, respectively). It follows that such a form of insta-

* Prophetess Deborah's famous song after the victory over the Philistines
includes the lines "The mountains flowed before the Lord". For illustration
we can consider the relaxation time for gases to be in the order of lo-12 s,
while for glasses within their region of glass transformation it is seconds,DN%
1, and approaches I010 s for stable crystals.

bility will also depend on external (procedural) parameters; and in contradic-

tion to the distinct state of metastability, it can be set up in different
levels (positions see Fig.l., dashed and dotted lines).


Let us thus investigate the behaviour of Gibbs energy in Fig.1. At the
melting point the liquid and crystalline phases have an equal Gibbs energy but
differ in enthalpy and entropy content. Upon cooling below the melting temp-
erature and entropy of the undercooled liquid decreases more rapidly than that
of the stable crystalline phase. Examining these different rates of the en-
tropy loss, we can determine a point where the entire entropy of melting would
be diminished, resulting in the entropy of both phases in question becoming
identical at the temperature To, which Kauzmann (ref.7.) called the pseudo-
critical temperature, still above absolute zero. Such a critical trend for
the entropy of an undercooled liquid is not always appreciated enough because
of the prior intersection by the liquid vitrification where the heat capacity
of the liquid changes abruptly to a value close to that of the corresponding

crystalline phase, thus preserving a residual entropy characteristic for the

glassy state alone. However, an unsolved question remains as to what would
happen if the isoentropy temperature of the so-called ideal glass transform-
ation is nevertheless attained by infinitesimally slowing the cooling rate and

thereby avoiding irreversible freeze-in from occurring. Although it is more a

game of imagination, we have to consider the possible existence of some kind
of higher (presumably second) order transition. In this transition the heat
capacity of the undercooled liquid changes under the cond,ition of the internal
equilibrium to a value similar to that of the congruous stable crystal. The
thermodynamic properties of an ideal glass structure thus become very similar
to those of the crystalline state (ref.8). The apparent formation of such an
equilibrium glassy state can even be suggested as a "fourth state of matter",
a promising theme for theoreticians to discuss further, although some mechan-
ical instability arguments imply its occurrence (ref.9).
The viscosity of a liquid can be regarded as a reflection of the relation
between the thermal energy available at a given temperature and the strength
of forces pulling species together and restricting their positions to a given
volume within which any molecular rearrangements can Occur. The possible rate
of these rearrangements rapidly decreases with decreasing volume within which
the species are packed. The volume is determined by the strength of the at-
tractive forces and, in turn, this strength is reflected in the values of the
characteristic temperatures of the boiling, melting and Critical Points. Ac-
cording to Angel1 (ref.10) the basic reason for the "failure" Of a liquid to

FIG.l. Left panel: A dependence of the system Gibbs energy, G, versus the
temperature, T, indicating the characteristic regions of the existence of sta-
ble (melt, crystal - full line), metastable (undercooled liquid, superheated
crystal - dashed line) and unstable (glass - dashed and dotted line) phases.
For comparison, the second plot, the viscosity, shows typical curves represen-
ting the action of a system under the different cooling rates with the glass
transformation interval, Tg, marked while the yet lower plot shows the behav-
iour of the system entropy, S, down to the pseudocriticalpoint, To. The bot-
tom part corresponds to the derivative of the enthalpy change,AH, temperature
dependence (related to the associated phase diagram by the lever rule) whose
typical shape is similar to that of DSC (and also DTA) curves. The arrows
specify the cooling and reheating modes and the temperatureTcr indicates the
point of metastable crystallization (the dotted line in the S vs T diagram).
Right panel: A hypothetical temperature dependence of the reduced heat cap-
acityAcpr (=Ac,/ As) characteristicfor glass formation of metallic alloys
and of inorganic oxides and organic polymer (upper two curves) in comparison
with the approximationsnoted in the text: stepwise, linearly deceasing and
diffuse-lambda shaped (lower three curves) - suitable for extrapolating the
behaviour of a fictive undercooled glass-formingmelt.

crystallize can be attributed to problems in the molecular rearrangements, not

so much in the liquid state but predominantly in the crystalline state during
ordering - the higher the crystalline symmetry of a phase to be formed the
better the glass-forming ability of its mother melt that can be anticipated.
The factors determining the probability of a given substance to freeze-in as a
glass are then related to the problems of finding a suitable solution in the
three-dimensional long-range ordering of the constituent species. The proba-
bility of glass formation increases with the formation of liquid mixtures in
which the Gibbs energy of the non-crystalline arrangement is decreased while
that for the corresponding crystalline phase to precipitate remains
unchanged. The more strongly the mixture of components interact, the more
rapidly the freezing point of the solvent is depressed and the viscosity in-
creased thus slowing the possible nucleation and the consequent growth of nu-
clei. The component interaction, however, should not be so strong as to
generate a new competing crystallizing phase which would strongly decrease the
glass-forming ability of the mother melt. Furthermore, the more stable the
crystal is, the higher the resultant melting point which simultaneously pro-
duces a less viscous melt and increases the probability of its crystalliza-
tion.It is well-known that the most easily vitrificable metallic alloys exhib-
it strong solvent-solute interactions resulting in low eutectic temperatures
but in the vicinity of which the existence of any stable binary or ternary
compound is not exhibited. On reheating such a glassy composition, a metast-
able phase is likely to be produced, but this form always decomposes later to
a combination of stable components. In this light, the theories describing
the stable glass-forming regions in multicomponent systems can be regarded as
dealing with the lowering of the Gibbs energy of a given solid phase on dis-
solution of solutes as a function of solute character (ref.10). The famous
Zachariesen rules (refs.ll,lZ), originally derived for oxide systems, may thus
be understood as rules for predicting low melting points relative to the for-
ces acting between the species, although many of the new inorganic glasses
violate these predictions. Hang et al. (ref.12) recently discussed the pre-
diction of glass formation and devitrification from the point of view of the
thermodynamics of regular solutions of complex oxide systems by examining
their immiscibility behaviour. A quasichemical model for solutions was also
used by Chen (ref.13) who evaluated the alloy formation of transition metal-
based glasses based on the binding energies of constituents in relatiOn to
changes in glass-formation temperature.


By measuring a macroscopic property of a liquid during its continuous cool-

ing we can distinguish certain characteristic points and their corresponding

te~eratures which, however, are related to the procedural para~ters of the
quenching technique employed (refs.4,15). Positioning of these points can be
specified on the basis of thermoanalytical investigations of various physical
properties, such as the heat content (by thermometric-DTA and DCC and heat
compensation-DSC-measurements), viscosity, elasticity, penetration and thermal
expansion (by thermomechancmetry and thermodilatometry), electric resistance,
coefficient of resistance and dielectricity, thermoelectric power, Hall effect
(by thermoelectrometry) and other structural studies (Midssbauer effect, X-ray
diffraction, photoemission, etc.). One of the most informative plots of
enthalpy change with temperature, resembling common recordings from DTA and/or

0% (ref.41, is presented in Fig.1. A practical way to resolve the freeze-in

entropy from the plot of cp vs In T was shown by Angel1 and Rao (ref.16) ter-
med isoentropy glassy state determination.
It is evident that a correlation between the characteristic temperatures

and glass-forming ability has been anticipated. A simple relation between Tg

and Tmelt has already been suggested on a theoretical basis by Kauzmann (ref.
7), assuming Tg to have a behaviour similar to To. Then a reduced glass tran-
sformation temperature Tgr was introduced so that Tgr = Tg/Tm and values of
about 213 are attained. Sakka and Mackenzie (ref.171 examined in detail its
general validity for a great variety of glasses and found it reasonable. In

addition, they reformulated Tgr on the basis of the combination of two for-
mulae relating Tmelt with the thermal expansion coefficient and Tg with the
fractional free volume of glass approaching, however, the value of l/2. The
meaning of reduced temperatures in the fast developing field of metallic glas-
ses was extensively dealt with by Davies (ref.18). Angel1 (refs.19,20) tried
to determine the extent of Tgr values on the basis of the extrapolated To data
evaluated by the Vogel-Tamman-Fulcher equation for viscosityp =po exp kp/(T-

T/J . TP, which has the unit of temperature, renders the viscosity formally
infinite when it equals the temperature of the entire measurements, which sug-
gests a certain rheological limit to the liquidus region from the the~odynam-
ic (Tp= To) and kinetic (Tp= Tg) points of view (ref.19). This showed, how-
ever, that an apparently more appropriate ratio, To/Tmelt, fails to follow the
'two-thirds' and even 'one-half' rules. Hrub$(ref.21) attempted to give a
more practical significance to glass-forming tendencies using the easily

available ratio given by Kg1 = (T,, - Tg) / (Tmelt - Tcr), see Fig.1, and
tested it on various types of chalcogenide glasses. As with Tgr the greater
the value of Kgl, the better the glass-forming ability is approached. In this
context, an interesting feature of glass transformation should be emphasised
again, viz. Tg is displaced to higher temperatures by an increase in the cool-

ing rate which is also reflected in the experimental variability of Tgr and
Kg1 values discussed in more details by Grest and Cohen (ref.22) and Thornburg
(ref.23), respectively. The meaning of the crystallization temperature Tcr as
a material constant is dealt with by Takayma (ref.24).
Many modelistic approaches to understand the various features of vitrifica-
tion have been described in the literature (ref.25). Free volume theory has
laid emphasis on the concomitant decrease in volume and fluidity of glass-
forming melts in the undercooled region. A recent extension of this model was
made by Cohen and Grest (refs.22,27) in conjunction with the percolation con-
cept to impute a first-order character in the glass transformation at T< Tg.
Considering a real glass as microheterogeneous, glass transformation can be
associated with the melting of clusters, recognizing thus the presence of an
intermediate range of structures between the crystalline and non-crystalline
states. Possible limitations of these models were discussed by Rao et al.
(refs.3,25,27). Some. other approaches can be briefly mentioned. Goldstein
(ref.28) has suggested that the configurational state of an undercooled liquid
can be described by an energy hypersurface of position and momentum coordi-
nates and that Tg occurs when the system of particles gets trapped upon cool-
ing into one of the many potential minima which are present on such a hyper-
surface. J'iickle(ref.29) has presented glass transformation as a transition
from ergodic to non-ergodic behaviour and discussed the meaning of the resi-
dual entropy of a glass. Ngai (ref.30) advanced a unified model of low-

frequency dissipation based on the dispersion approach. Edwards (ref.31)

argued that the concepts in the theory of viscoelasticity offer a good basis
for describing the polymeric types of glasses. The degrees of freedom can be
separated by their relaxation times into those that relax faster than the
quenching operation and those that are slower. Harris (ref.32) showed that a
simple model of glass transformation can be described by the statistical mech-
anics of a set of non-interacting particles each of them being able to assume
either of the two energy levels considered. Recognizing a parallel between
metallic, chalcogenide glasses and native solid electrolytes Phillips (ref.33)
related glass formation to directionless units and proposed possible dispro-
portionation on a molecular scale as a dominant barrier to crystallization.
Glass transformation itself is assumed to be a result of increasing cluster
size where its surface/volume ration involved means that the cluster interior
is ordered while the surface layer remains disordered.


Fundamental theoretical studies were made by DiMarzio and Gibbs (ref.34)
and Adam and Gibbs (ref.35) showing that the sluggish relaxation is under-

scored by the existence of an equilibrium glassy state. Experimental glasses

are perceived through their kinetic properties and one should be able to
arrive at a consistent description of the glassy state from the
The glass transformation bears some resemblance to the second-order phase
transition, the latter, however, obeying Ehrenfest's relations dP/dT = AcplTVAa
= Act/ALU, where P,V ,cyand %I!are the pressure, volume, volume expansion and
compressibility,respectively. Due to the kinetic origin of glass
transformation,this relationship is not fulfilled during most experiments and
the degree of its irreversibilityis described by the Prigogine and Defay
factor7r (refs.32-41).
72 = Tg Vg Aa2/bcpA% 31 (1)
Some atttempts to generalize such a description were made using internal vari-
ables 5 , and corresponding affinitiesA having the role of the genertilized
forces and fluxes. The concept of internal parameters was best described in
the frequently cited study by Davies and Jones (ref.36). The essential
difficulties arise, however, when we attempt to associate 5 with physically
identificablequantities. Breuer and Rehage (ref.37) suggested that eq.(l) is
valid only ifaE/aT= 0 so that, for example,
df = 6k -& (~)~~(~~~
Similarly, Staverman (ref.38) atiempted to define K on the basis of the par-
tial derivatives according to the internal parameters in question, DiMarzio
(ref.39) examined the above derivation and resolved that it leads to the equi-
librium conditions -dPfdT= asja~i~~v/dfi=asld5j/3v/c15jfor all pairs of the
order parameters i and j. Because K is always greater than one, he (ref.391
concluded that such a concept is not fully appropriate, Gupta and Moynihan
(ref.4D) proved the validity of then ratio for the systems with multiple or-
der parameters showing that eq.(2) must hold for each member of the set of or-
der parameters not including the assumption (ref.34,39) thata2G&dgj= 0 for
j# j. Therefore, the systems for which only one order parameter is required
to specify the state are trivial special cases because the order parameter, as
the extensive the~od~amic variable, can be arbitrarily subdivided into sub-
sets. If a single order parameter is sufficient to characterize the given
glass, the rate of a selected property change ii , given by dZf/dt =(~Z~~~~~TPX
, is proportional to another one ij. If the rates of different proper'-
ty relaxation,however,are not proportional to each other, this means that
more 5 are to be involved (ref .41). The practical aspects of the pressure ef-
fect on Tg was dealt with in refs. 42 and 43.
Following from the dissipation inequality, Christistensen (ref.44) provides
a the~odyn~i~ admissibilitycriterion for the glass transition temperature
which is characterized as the base temperature at which, for a constrained
sample with no volume change, the stress response to changes in temperature is
~nstant~neo~~,with no explicit evidence of a memory effect (creep response).
In this connection it is notable that the derivation mentioned above reveals a
transitional behaviour when internal variables have a certain relationship.
When a glass is held at a temperature below Tg its structure changes slowly
with time towards a certain equilibrium state characterizableby a fictitious
t~erature lying on the intersectionof extrapolated states of gtass and un-
dercooled liquid, respectively. The temperature difference between the fic-
titious and the actual temperatures thus gives a measure of the departure of
the material from equilibrium which, however, is true only when a single order
parameter is involved. When the description requires multiple order parame-
ters we need instead an adequate number of fictitious t~~ratures to achieve
an equally detailed picture of its actual state, The problem and usefulness
of the concept of fictitious temperatures was well developed by Moynihan et
al. (refs.4547). It should be noted, however, that the very important and
extensively studied field of processes dynamics (refs.48-51) is not dealt with
as it extends from kinetic theories on nucleation growth processes
frefs.18-52)which is a special discipline of th~r~dynamjcs. Here, it is
worth mentioning the illustrative use of hypothetical diagrams, either that of
the enthalpy change versus temperature fref,53), (cf.Fig.l),or that Of the
Gibbs energy versus the concentration frefs,48,4g)making possible the
characterizationof the type of accessible processes that occur in the given


Most thermodynamic studies deal with the behaviour of heat capacity during
vitrification,e.g., Ra~achandrarao et al. (ref54) correlated the discontin-
uous change in thermal expansion and heat capacity with each other using ex-
pressions for the ideal entropy of mixing in substitutionalsolutions of the
components of different size. Similarly, Hillert (ref.50) assumed that the
abnormal heat capacity of the non-crystallinephase above Tg is due to local-
ized defects and describes its configurationalentropy by the model of inter-
stitial solutions. A very illustrativeapproach was developed by Gutzow
(refs.5557) who assumed that the difference between thermal capacities of
liquid cyq * and crystalline, ckr, phases is approximatelyconstant, co, and
that there exists a limiting temperature, To, at which this difference vanis-
hes. Using differences in the molar entropies, s, enthalpies, h, and chemical
potentials,@, well-known frm classical tber~dyn~ics, we can write (ref.%)

Asliqrcr j*melt
= A *melt - T
A c;iq'cr dT/T (3)
A h'iq8cr_ A $,elt - (Tmelt liq,cr dT
A cp (4)

A,,ligncr _, - TJT"'lt Asxiqrcr dT (5)

Introducing dimensionless quantities

Tr - T/Tmelt, TOr - To/Tmelt and Tgr - Tg/Tmelt we obtain

"pr - Acliqrcr/ As,,~~ (6)

Asr- l- lST '=pr dT,/T, (7)

A h,- Ah'&, (
ASmelt Tmelt1 -l- ST '=pr dTr (8)
A% - AP~~"'~/ (AsmeLt Tmelt) - T_fTr Asr dTr 19)

The desired numerical values can be calculated from certain assumptions. We

can predict the behaviour of Acpr as shown in Fig.1, according to the three
gradual approaches: stepwise (Gutzow, ref.%), linearly decaying and diffuse
(Hillert, ref.50) and lambda shaped (refs.3,4) as known from the theory of
broadened phase transitions. In addition we can employ known experimental ev-
idence for the estimation of Tgr as discussed above (l/Z and Z/3). The en-
tropy difference frozen-in due to the glass formation is for a typical glass-
forming melt about l/3 (=ASr %Asg/ Asmelt) but can deviate for those glasses
that are difficult to prepare (l/4 through l/Z). For the simplest case of a
step-wise change we can calculate exploratory data by the introduction of Tgr
and Asr into the equations listed in Table 1 - using the condition of sr = 0
at Tr = Tor. The temperature dependence of the thermodynamic potential dif-
ference read from the above equations (ref.57):

Aur = T f
T9xAs, dTr + T S1(l+ '0 In Tr) dTt -AS rfTgrwTr
r gr I

+ (1 -co) (l-Tgr ) - coTgr In Tgr - a - b Tr

Extension of this model by the introduction of decreasing values ofAcpr does

not changethe limiting values in the Table. It just helps for a better fit
with reality.


where Ati - 1 + CC In Trt Ah(Trj- 1 - CC 1 - T, and

Au(T,)-(1 - CC)( 1 - Tr) - CC T, In TX


The author greatly appreciates the kind interest of Dr Gerd Olofsson of the
Chemical Center of Lund University and thanks Or Bengt Haglund of Sandvik Hard
Materials for making possible the author's participation at the 9 NSTAC.
The study has been devoted to Emeritus Professor Norbert Kreidl of the
University of Albuquerque on the occasion of his 80th birthday celebrated at
the Vienna Congress on Glass in 1984.


1 B. Eckstein, Mater. Res. Bull., 3 (1968) 199.

2 R.Ro J. Amer. Cer. Sot., 52 (1969) 344 and J. Non Cryst. Solids, 3
3 C.N.R. Rae and K.J. Rao, Phase Transitions in Solids, McGraw Hill, New
Yorb, 1979.
4 3. Sesthk, Thermophysical Properties of Solids, Elsevier, Amsterdam, 1984.
5 H. Suga and S. Seki, Discuss. Faraday Sot., 69 (1980) 221.
6 M. Reiner, Phys. Today, 17/l (1964) 62.
7 W. Kauzmann, Chem. Rev., 43 (1948) 219.
8 D. Thurnbu~l and M.H. Cohen, J. Chem. Phys., 34 (1961) 120.
9 C.A. Angel1 and J. Donnella, J. Chem. Phys., 67 (1977) 4560.
10 C.A. Angel1 in Preparation and Characterization of Materials, J.M. Honig
and C.N.R. Rao, eds., Academic Press, New York, 1981, p. 449.
11 J.M. Stevels, Philips Tech. Rundsch., 9/10 (1960) 337.
12 J.A. Stanworth, J. Amer. Cer. Sot., 54 (1971) 61.
13 2. Hang, X. Hu and X. Zhao, J. Non Cryst. Solids, 52 (1982) 235.
14 H.S. Chen, Acta Metall., 24 (1976) 153.
16 J. Jest&, Skldr keram., 28 (1978) 321 and in Proc. Czechoslovak Conference
on Glass LIT, Usti n.L. 1982, ESVTS.
16 C.A. Angel1 and K.J. Rao, J. Chem. Phys., 67 (1972) 470.
17 S. Sakka and J.O. Mackenzie, J. Non Cryst. Solids, 6 (1971) 145.

18 H.A. Davies, J. Non Cryst. Solids, 17 (1976) 159 and in Amorphous Metallic
Alloys, F.E. Luborsky, ed., Butterworth, London, 1983.
19 C.A. Angel1 and C.T. Moynihan in Molten Salts, 6. Mamantov, ed., Marcel
Dekker, New York, 1969.
C.A. Angell, J. Amer. Cer. Sot., 51 (1968) 117.
A. Hruby', Czech. J. Phys., 822 (1972) 1187.
G.S. Grest and M.H. Cohen, Phys. Rev., 821 (1980) 4113 and 820 (1979) 1077.
D.D. Thornburg, Mater. Res. Bull., 9 (1974) 1481.
S. Takayama, J. Mater. Sci., 11 (1976) 104.
R. Parthesarythy, K.J. Rao and C.N.R. Rao, J. Phys. Chem., 85 (1981) 3085
and Chem. Sot. Rev., (1984) - in press.
26 M.H. Cohen and G.S. Grest, Ann. N. Y. Acad. Sci., 371 (1981) 199 and Adv.
Chem. Phys., 48 (1981) 455.
K.J. Rao, Proc. Indian Acad. Sci., 93 (1984) 389.
M. Goldstein, J. Chem. Phys., 64 (1976) 4767 and 67 (1977) 2246.
Philosooh. Mao.. 44 (1981) 533.
29 J. Jiickie, Phil&. Msg., 44 (1981) 533.
30 K.L. Ngai, Ann. N. Y. Acad. Sci., 371 (1981) 325 and Comments Solid State
Phys., 9 (1979) 127.
31 S.F. Edwards, Ann. N. Y. Acad. Sci., 371 (1981) 211 and Philos. Mag., 40
(1979) 257.
R. Harris, Solid State Comments 45 (1983) 711.
J.C. Philips, Comments Solid State Phys., 9 (1980) 191 and 10 (1981) 85.
J.H. Gibbs and E.A. DiMarzio, J. Chem. Phys., 28 (1958) 373 and 807.
6. Adam and J.H. Gibbs, J. Chem. Phys., 43 (1965) 139.
R.O. Davies and 6.0. Jones, Adv. Phys., 2 (1953) 370 and Proc. Roy. Sot.,
26 (1953) 217.
H. Breuer and R. Rehage, Kolloid Z.2. Polym., 216, 217 (1967) 159.
J. Staverman, Rheo. Acta, 5 (1971) 713.
A.E. DiMarzio, J. Appl. Phys., 45 (1974) 4143.
P.K. Gupta, T.C. Moynihan, J. Chem. Phys., 65 (1976) 4136.
Ryong-JoonRoe, J. Appl. Phys., 48 (1977) 4085.
A. Jayaraman and M.H. Cohen in Phase Diagrams, A.M. Alpper, ed., Vol. 1,
Academic Press, New York, 1970, p. 281.
43 D.S. Sandinov, Zh. Fiz. Knim. 47 (1973) and 50 (1976) 1653.
44 R.M. Christistensen,
~. ___ ~_ Trans. Sot. Rheol., 21 (1977) 163.
45 M.A. DeBolt, A.J. Easteal, P.B. Macedo and T.C. Moynihan, J. Amer. Cer.
sot., 59 (1976) 16.
46 T.C.-M0ynihan;A.J. Eastel, A.N. Oepolt and J. Tucker, J. Amer. Cer.
Soc.,59 (1976) 12.
47 T.C. Moynihan and K.P. Gupta, J. Non Cryst. Solids, 29 (1978) 143.
48 U. Koster and P. Weiss, J. Non Cryst. Solids, 17 (1975) 359.
49 U. Kbster and U. Herold in Glassy Metals I, H.J. Giitherold and H. Beck,
eds., Topics in Applied Physics, Vol. 46, Springer, Berlin, 1981, p. 225.
50 M. Hillbert in Rapidly Solidified Amorphous and Crystalline Alloys, B.R.
Kear, B.C. Giessen and M. Cohen, eds., Elsevier, Amsterdam, 1982, p. 3.
51 H. Rajczak and F. Stobiecky, Acta Magnetica, 1 (1984) Suppl. 84, 131.
52 D.R. Uhlmann, J. Non Cryst. Solids, 7 (1972) 337 and in Proc. XI. Congress
on Glass, 3. Giitz, ed., DT CSVTS, Prague 1977, Vol. I., p. 43.
53 J. SestLk, Wiss. Zeitschr. Friedrich-Schiller-Univ. Jena: Math. Natur., 32
(1983) 377.
54 P. Ramachandrarao, B. Cantor and R.W. Cahn, J. Mater. Sci., 12 (1977) 2488.
55 I. Gutzow in Amorphous Materials, R.W. Doughlas and B. Ellis, eds., Willey-
Interscience, London, 1972, p. 159.
56 I. Gutzow, J. Non Cryst. Solids, 45 (1981) 301.
57 I. Gutzow, Wiss Zeitschr. Friedrich-Schiller-Univ.Jena: Math. Natur.,32
(1983) 363.
Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B, June 2010, 51 (3), 165172

Aspects of the non-crystalline state

Carlos A. Queiroz*1 & Jaroslav estk2,3
VICARTE, Faculty of Sciences and Technology, New University of Lisbon, P-2829-516 Caparica, Portugal
New Technology - Research Centre in the Westbohemian Region, West Bohemian University, Universitni 8, CZ-
30114 Plze, Czech Republic
Institute of Physics of the Academy of Sciences R, Division of Solid-State Physics, Cukrovarnick 10, CZ-16200
Praha, Czech Republic
Manuscript received 5 March 2010
Accepted 17 April 2010

Disordered matter still presents stringent conceptual difficulties. The often confused concepts of amorphous and glassy
states (and corresponding material substances) were re-examined and previous attempts to trace a distinction are
revised and thoroughly discussed. We examine several rheological and thermodynamical aspects of glasses in the glass
transformation region as well as the dynamics of associated relaxation processes.

1. Introduction: from continuum mechanics Of course, much of the apparent success of the
to a thermodynamic based perspective continuum mechanics approach is actually elusive,
Considering the solid state we can make a distinction as the whole collection of such idealized models may
between perfect and defective crystals, as regards still be unable to perfectly translate the behaviour
metals, ceramics and polymers, when recognizing of some real materials, and intermediate cases may
their structural defects, which are understood as be particularly difficult to encompass within the
deviations from the lowest energy bonding arrange- continuum mechanics theoretical framework. On the
ments.(1) However, we still lack a widely accepted other hand, thermodynamics tries to adopt a unify-
conceptual borderline to distinguish states of matter ing perspective; has a general method that tries to
intermediating between the classic concepts of solid apply to all substances, which is less dependent on
and liquid. It is particularly noticeable that the con- particular modelling hypothesis concerning idealized
cepts of glassy and amorphous solids appear often substances.
confused, having no adjusted convention for their From a classical thermodynamic perspective a
exploitation in the published literature, while still single homogeneous substance in some definite state
lacking more widely accepted associated definitions. of matter is treated as a phase (homogeneous region
Macroscopically, and from a continuum mechan- of matter, not necessarily continuous, which is part
ics perception, which deals with a material body of the thermodynamic system) and we are mostly
response to external contact forces, a solid is a physi- concerned with the evolution of the systems macro-
cal system which resists quasi-statically to external scopic properties as a response to heat flow (heating
stresses applied at the surface, although some kind or cooling). This may be also a convenient perspective
of motion is generally expected. Such stresses are to distinguish and characterize different materials.
specified by the stress tensor and the extent of body Statistical thermodynamics, also named statistical
motion can be theoretically related with the stress mechanics, bridges classical thermodynamics, as a
tensor by constitutive equations. A number of ideal phenomenological global approach, to a particles
materials may be defined to represent particular based approach. It is concerned with the interpreta-
relations between the stress tensor and the motion. tion and prediction of the macroscopic properties
These are theoretical models used to represent and of matter in terms of properties of the microscopic
approach the behaviour of real materials. elementary subsystems that compose it, such as
Although continuum mechanics allow us to molecules, atoms, ions, electrons, etc. Here we are
theoretically treat solids and liquids under a unify- mainly concerned with equilibrium values for certain
ing perspective, comprising intermediate cases as properties (e.g. temperature, pressure, specific heat,
well (e.g. viscoelastic response), while under an viscosity, etc.) and the macroscopic system should
experimental perspective the rheological approach comprise a large number of subsystems, typically of
may be considered as pretty accomplishing the same, about the order of the magnitude of the Avogadro
we still lack a comparable satisfactory unifying ther- number, so that statistical averages can provide a
modynamic treatment. reliable description of the macroscopic system at its
most probable state (thus in equilibrium, in the sense
* Corresponding author. Email that it remains unchanged with time). The statistical

Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010 165
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

thermodynamics approach has been quite successful and pressure. Tool contoured this difficulty by intro-
working with gases (including real gases) and solids, ducing the concept of fictive temperature, Tf,(36) which
but rigorous predictions of the thermophysical prop- corresponds to the extrapolated temperature of the
erties of dense fluids derived after their canonical supercooled liquid. The concept of fictive temperature
ensemble partition function are currently difficult associates itself to the assumption that the glass struc-
to obtain. ture match up approximately to that of a frozen liquid,
Most melts can be undercooled by perhaps overcooled to Tf. The fictive temperature defines the
1020C below the temperature of melting, Tm, but actual structure of the glass and may be generally
the presence of impurities, slight disturbances, or estimated from intersection of the extrapolated su-
even the mere contact with the container, will lead to percooled liquid (equilibrium) line with the glass line,
a prompt crystallization. Low viscosity facilitates the considering a convenient property versus temperature
diffusion of constituent species, which is unavoidable plot (e.g. molar volume).(7) For a complete phenomeno-
to make possible the necessary reconstruction of the logical description of the glassy state, the parameters
fluid liquid web to that of rigid solid. The pre-nucleus necessary and sufficient to describe an equilibrium
(embryonic) sites, often existing in the melt, may state are classified as external, such as temperature
enhance straight forwarded crystallization, which (T), pressure (P) or composition; while Tf is taken as a
is also favoured by their crystallographic orienta- single internal (also called structural) parameter that
tion (e.g. lower symmetry, like monoclinic, which accounts for departure from the equilibrium state, for
can bestow easier configuration and growth of such which the Gibbs free energy is minimized.(8)
nuclei).(2) Considering a molar volume versus temperature
The melt viscosity () tends to increase rapidly plot (see Figure 1; the above plot), and herewith
along with decreasing temperature and reaches the the prolongation of the line of the supercooled liquid
traditionally expected value near Tm in order of 106 towards the equilibrium crystal line below the glass
Pa s.(1) Diffusion dies down speedily and, for that rea- transition, Tg, which, according to the thermodynam-
son, crystal formation turns out to be more and more ic theory proposed by Gibbs & DiMarzio(913) can well
intricate, but enabling, on the other hand, certain be considered as an equilibrium line between the
stability improvement toward the metastability of a supercooled liquid and the glass, it seems acceptable
supercooled liquid, until the development of viscosity that over the extrapolated segment the fictive tem-
near 1013 Pa s conducts to the constrained freeze-in perature equals the real temperature. As a melted
state that characterizes a radically constrained glass. glass under cooling enters the transformation re-
gion, a specific cooling rate may actually be selected
so that a certain fictive temperature, Tf, located over
2. Transformation temperature, fictive
the extrapolated segment range, can be reached by
temperature and Kauzmann temperature
the supercooled liquid. Tf tends to decrease as the
The supercooled liquid constitutes a metastable phase cooling rate decreases. Extrapolation for infinitely
in a state of constrained internal equilibrium, while slow cooling carries out the fictitious temperature
its free Gibbs energy is higher than that at the state towards the line of solid state and that would con-
of stable equilibrium. The free energy of the metast- duct Tf down to a specific temperature, known as
able liquid is thus kept at a local minimum, but the Kauzmann temperature, TK.(14) It is supposed that
not at an absolute minimum (which would imply Tg would then present itself as a sharp elbow, thus
crystallization). without any curvature at all.(15) This suggests that TK
In the transformation region, hereafter assumed represents a lower limit of the glass transition. The
to lie below the melting temperature (Tm) but above glass transition would then lose his dynamic charac-
the Kauzmann temperature (TK), the specific volume ter (its time dependency), thus presenting itself as a
depends on the thermal history of the material (e.g. static structural transition.(15,16) Before TK is actually
annealing time). This dependency cannot be fully reached crystal nucleation might eventually occur
understood within the framework of classic thermo- in differently noticeable manners. However, when
dynamics, were the specific volume is treated as a the crystals growing rate is sufficiently low, it still
function of state, thus depending only on temperature seems reasonable to assume the validity of a smooth
extrapolation trend for the remaining supercooled
There are some important points related to the glassy region revealing the
liquid. It should be stressed that this interpretation
allied characteristic temperatures. Most common is the Kauzmann tempera- depends conceptually on the equilibrium assump-
ture, TK, given as the intersection of extrapolated line for the equilibrium
liquid with that of equilibrium solid. The so called Vogel temperature, Tv,
tion between the supercooled liquid and the glass,
is decisive for the moment where the shear modulus terminates its increase since one needs to consider the extrapolation trend
and the viscosity becomes infinite according to the VogelFulcherTammann
model. This implies a discontinuity where viscous flow ceases (on cooling),
at a certain finite temperature (the Vogel temperature, Tv). For polymers Actually the entropy versus temperature plot should be preferred for obtain-
it customarily lays 52C below Tg but for inorganic glasses is about 100C ing a TK estimate by extrapolation, but since the lines present some curvature
lower. The Kauzmann temperature, TK, may get coincidental with Tv if in such a graph, the extrapolation procedure cannot rely in straightforward
derived from the entropy plot. linear extrapolation procedures being therefore more prone to error.

166Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

extreme of slow cooling, they are factually expected

to vanish at a positive finite temperature (TK).(13,1921)
Recently, Gutzow et al(22) concluded that for glasses
one should expect Cp(T)=0 at T=0, while non zero
values of the zero point entropy (S|T0>0) cannot be
ruled out, what might be interpreted on the basis that
absolute zero cannot in fact be attained, accordingly
the most general formulation of the third principle
of thermodynamics.
Below TK the concept of fictive temperature loses
its significance and any configurationally rearrange-
ments typical of the glass transition are not expected
to occur. The theoretical model proposed by Adam
& Gibbs (AG)(2326) links the structural relaxation rate
to the configurational entropy, which is in a rather
good agreement with laboratory observations for
many orders of magnitude of the structural relaxa-
tion time.(27) Taking it in consideration, while provid-
ing that TK may stand as the transition temperature
of the supercooled liquid for the hypothetical case of an
ideal glass with null-approaching configurational
entropy (Sc=0),(2830) such a transition would corre-
spond to an infinite relaxation time.(9,26,31) However, it
may be admitted that, in these idealized conditions,
glass transition would correspond to a discontinuity
in the first derivative of the molar volume (transi-
tion of second order) while true crystallization
Figure 1. Schematic typical plots for glass transition. Above: occurs with discontinuity of the specific volume
molar volume versus temperature; glasses, A and B, with (transition of the first order). Since one cannot carry
decreasing fictive temperature (TfA and TfB) can be derived by out experiments under such a limiting condition
decreasing the cooling rate; a TK estimate may be obtained as (while taking into account a theoretical point of
the extrapolation limit at null cooling rate; hysteresis effect view), and the glass transition is, in reality, still
is shown for glass B, only. Below: isobaric heat capacity (Cp) not fully understood,(32,33) we must face extrapola-
showing a step-like nonlinear variation with some hysteresis tions with some reservation. Even if not realizable
at the calorimetric Tg (depicted for glass B, on cooling) experimentally, since in practice is expectable that
the usual (kinetic) glass transition always preempts
as coexistence curve, so that a hypothetical ideal such an ideal thermodynamic glass transition, the
glass transition might be assumed.(17) associated limiting notion has provided a stimulat-
Slower cooling rates lead to glasses of lower ing debate and has been included as a key idea in a
enthalpy contents, and providing that crystalliza- number of models of disordered glassy systems that
tion does not occur, the isobaric heat capacity (Cp) is significantly contributed to our current understand-
expected to present a step-like nonlinear variation ing of the glassy state.
with some hysteresis, within a temperature range Accordingly the AG approach, it is worth noting
centred at the calorimetric Tg, decreasing from values that at TK only liberational motion frequencies (often
close to that of the equilibrium liquid to values close associated with angular displacements) are expected
to that of the equilibrium crystal, although a positive to freeze-in. These can be associated to the liquids
but relatively small gap is still expected (Figure 1; configurational heat capacity. The higher frequencies
the plot below). Excess values for entropy and Cp are are still expected to hold at lower temperatures, as
expectable for glasses below Tg, as compared to the they are associated with the vibrational frequency
corresponding crystallized substance. Such excess modes of the amorphous solid. However, the struc-
values are even expected to be retained below TK, tural relaxation of the supercooled liquid is expected
although the observed gaps can somehow decrease
as the temperature drops until they eventually van- It may be possibly argued that point defects can still be expected at T<TK,
ish (at absolute zero or above), as long as Nernsts either for amorphous and/or crystalline structures, and this precludes the
associated entropic contribution to vanish above absolute zero, and hence
heat theorem is obeyed, what is still a question of at TK. This contribution is not strictly accounted for by the original AG
debate for glasses in general [e.g. Ref. 18]. Those model. Noticing that at TK the configurational entropies of supercooled
liquid (extrapolated to the ideal glass hypothetical state) and that of the
gaps are thought to become smaller and smaller for crystal are equal, it seems preferable to identify Sc with the excess entropy
lower cooling rates, and eventually, for a sufficient of the non-equilibrium glassy state with regards to crystal.

Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010 167
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

to be completed at TK. Below TK the glass-like matter dercooled liquid is indistinguishable to that of the
still shows higher specific volume and entropy than congruous stable crystal, it comes to pass defend-
the corresponding the crystalline solid, but, in either able, instead, that the glass transition may be best
case, the contributions for the specific heat are mainly thought as a first-order-like transition, although of
vibrational. a disordered type. Such an approach leaded to the
An ideal glass transition can be interpreted as, random first-order transition (RFOT) theory developed
more or less, resembling a second order phase transi- by Wolynes et al, as a mean-field model approach
tion, according to the Ehrenfest classification.(3436) governed by the assumption of localized excita-
The corresponding theory makes certain that at the tions (entropic droplets).(3945) The thermodynamic
fictive temperature, the two coexisting phases (a properties of such an undercooled highly vitreous
gluey supercooled liquid and an extremely viscous glass) melt possesses novel properties of an ideal glass
should then possess matching entropy, holding, how- structure, becoming therefore very similar to those
ever, distinct specific heat [e.g. Ref. 36]. This implies of the crystalline state, as first noticed by Cohen &
that the two phases should then become isentropic. Turnbull.(46) The meaning of TK for chalcogenide
Besides, the molar volume of these two phases must glasses was studied in more details by ernoek et
also tend to reach a common value. The jump verified al(47) and other statistical, structural, and frequentional
in the specific heat suggests a certain congealment of details of Kauzmann (and other temperatures) are
some kind of structural movements at the Kauzmann given by Hlavek et al.(48)
temperature, which are expected to cease their contri- The Vogel temperature, Tv, taken from the Vogel
bution toward the specific heat. One may interpret the FulcherTammannHesse (VFTH) empirical equa-
rise in the glass transition temperature above TK for a tion(4951) has first been identified with TK under the
real glass transition as related with the period of time framework of the classical AG entropy model,(9,23)
available for the observation of such freezing move- and since then by other theoretical models. The same
ments during typical laboratory experimentation. confluence between Tv, were a kinetic divergence is
So far, the most stringent criticism toward the expected, and the entropy crisis at TK, was found
ideal glass concept (and hence the AG model) seems justifiable within the framework of the RFOT theory.(43)
to be that of Stillinger, who pointed the apparent need This conclusion also was corroborated by the fluctua-
for configurational excitations at such an ideal glass tion theory proposed by Donth,(52,53) by calculations
state to require an unbounded energy.(32) However, based in a ionic potential model presented by Hunt
we should bear in mind that such excitations are as- (for ionic glasses),(54) by a thermodynamic approach
sociated with activation kinetic barriers on the free developed for the model of spin-glasses,(55,56) and
energy surface and with the associated probability of also by means of a stochastic model suggested by
escaping from a particular structural configuration. Odagaki et al.(57) Experimental results obtained by the
The AG assumption of a underlying (kinetically technique of specific heat spectroscopy seem to support
obscured) second order phase transition with an this identification as well.(58)
associated infinite relaxation time seems, at least Taking Tv>0 K, B and ho (a pre-exponential factor)
as a theoretical limiting notion based on simplified as fitting parameters, the equation of VFTH expresses
continuous approach, not consistently invalidated by the glass viscosity as
the argument. For a recent critique of the AG model
the reader is directed to the work of Dyre et al(37) which h(T ) = h0 exp (1)
T - Tv
also questioned,(38) based on dielectric relaxation
data for organic liquids obtained just above Tg, the In the limiting case of Tv0 K, the VFTH equation re-
VogelFulcherTammann prediction for divergence duces to a familiar dependence of the Arrhenius type,
of the relaxation time at a finite temperature, Tv>0 K. where Tv is habitually determined from adjusted
It is also worth noting that a number of authors experimental viscosity data, most often reported for
took into account that the inspired existence of some T>Tg>Tv . For this reason, attempts for the attribution
higher (presumably second order) transition should of physical meaning to the parameter of adjusting Tv
be regarded as somewhat speculative. Accordingly, can be questioned outside the range were the experi-
bearing in mind that at the matching conditions of mental data can be obtained. Viscosity measurements
internal equilibrium, the heat capacity of the un- at T<Tg were seldom reported, and they must be in-
terpreted carefully, owing to the structural relaxation
If the thermodynamic equilibrium may be assumed for the phases, the effect and the possible occurrence of shear thinning
Ehrenfest classification affirm, for a first order transition, that the first
derivatives of de Gibbs free energy with regards to the temperature and
(as reported for silicate melts(26)), since elastic work
pressure present a discontinuity at the phase transition temperature. If the done by a shear stress may generate configurational
first derivatives are continuous but the second ones are discontinuous, the
phase transition is said of second order. For low cooling rates, and when
entropy. The usefulness of the AG model to interpret
considering glass as obtained by undercooling melt, the glass transition may and correlate the temperature dependence of sub-Tg
be seen approximately as a second order phase transition, being however
continuous (doesnt occurs at a definite temperature) and cooling rate
viscosities seems however well established for silicate
dependent, what marks a difference toward a genuine phase transition. melts.(26,59)

168Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

3. Dynamics of the glassy state: non- remains fairly insensitive to the variation of tem-
crystallinity, vitroids and Deborah number perature.(26,59,68,69)
Eckstein(60) classified non-crystalline solids as vitroids, The glass transition occurs when the time of
referring to organic and inorganic materials in the structural relaxation considered reaches the order
vitreous state, including inorganic glasses and or- of magnitude of a reasonable time period needed for
ganic plastics. Vitroid structures are interpreted as the observation of the phenomena with the available
paracrystalline networks (topologically disordered), laboratory equipment (usually ~1103 s), and, as a
thought as intermediary arrangements between melts consequence, for subsequent cooling experimenters
and the corresponding crystals. Such structures can cease to observe structural rearrangements in the
be quantitatively characterized by temperature-like glass network.(70) Pointing in the same direction,
parameters, named configurational temperatures. Nieuwenhuizen(71,72) refers that a system is glassy
Interesting is the use of a terminology that combines when the observation time is much smaller than the
the term solids with the term vitroids, seen within equilibration time. As a glassy system cools through
the framework of rheology. This is because such a the transformation region, the relaxation time of
vitroid changes with time, thus being dependent on slow dissipative phenomena (named -process)
the existent observation time involved in detecting becomes much larger than the time of observation;
the extent of the change (as with viscous flow). while the fast processes (-process) are still observ-
Reiner (61) introduced the so-called Deborah able at equilibrium. There is, apparently, no obvious
number expressing the ratio between the time of the structural signature associated with the steepest
materials relaxation and that of its actual observation increase displayed by the time constant associated
time, i.e. the experimental time scale (De=/). The with the a relaxation process, which is followed by a
name comes from the bible, where the prophetess correspondent viscosity increase. Nevertheless, such
Deborahs song after the victory over Philistines in- a dramatic increase of the relaxation time associated
cludes: the mountains flowed before the Lord. For to the a-process is seemingly connected to a rapid
gases it reaches ~1012; for crystals we have De~1010; fall-off of entropy, as it was first pointed by Gibbs &
and, for glasses close to the transition temperature, DiMarzio.(9)
it lays in an order of units. For an ordinary, as- The distinction between relaxation processes
quenched glass, the experimental relaxation time of and (classification after JohariGoldstein(7375)) is as
a given frozen structure, represented by Tg , is said follows. The leading relaxation process () is charac-
to be proportional to the shear stress relaxation time terized by a time constant that increases from ~1012 s
related to a shear viscosity expediently laying in (~1012 Hz) to ~100 s (~102 Hz) in the glass transition
order of 1012 Pa s. The Deborah number may be used region, accompanying (in an approximately propor-
as a criterion to distinguish solid from liquid-like tional form) the steepest increase of the viscosity,
behaviour.(62) At the glass transition temperature we and deviating from Arrhenius law. A kinetic glass
expect De~1. Thus, when Tg is accessed by dynamic transition temperature Tg,kin is therefore identifiable
mechanical thermal analysis (DMTA), which identi- from dielectric relaxations measurements, ascribed
fies itself with a pronounced peak in the loss tangent to the temperature at which the average (structural)
(tan), the corresponding average relaxation time () relaxation time for the a process approaches 100 s.(76)
may be considered tuned with the angular frequency, Tg,kin often agrees with the temperature at which the
=1/.(63) shear viscosity reaches 1012 Pa s,(77) which is widely
Suga & Seki(64) stressed out that the glassy state is identified with the glass transition temperature,
thermodynamically unstable and can be assumed to e.g. Ref. 78, being also, generally comparable with
subsist as an undercooled metastable state obtained the calorimetric Tg, ascribed to the onset of the heat
at a certain instant of continuous cooling through the capacity jump observed in the transition region. One
glass transformation process, which is considered to may also be able to observe a secondary process of
be a general property of all such metastable phases. relaxation (), in the kHz region (~103 s), which
As Wong & Angell(65) indicated, the distinction obeys to an Arrhenius temperature dependence
between supercooled liquid and glass assumes that and is not significantly altered within the glass
experimental essays are carried out in a time scale of transition region. One may thus consider this last
seconds or minutes, that is, in the usual range of the process as fast in the transition region with respect
Maxwell relaxation time(66) for a glass within the Tg to the alternative process.(55) The process can be
region. This is the relaxation time for tensions when a attributed to reorientation movements in the glass
constant shear is applied, and can be obtained as the network.(79) At higher temperatures (above ~12Tg to
ratio between the viscosity (h) and the shear modulus 14Tg ) the process and amalgamate in a unique
at infinite frequency (G)(67) process.(27,8083) The merging occurs at a temperature
that seems coincidental with the dynamical glass
=G (2)
transition (at Tc, known as crossover temperature)
Remarkably enough is that the shear modulus predicted by mode coupling theory,(84,85) while such

Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010 169
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

a temperature may even be regarded as an upper the nondeterministic chaos theories is thus conse-
bound for the glass transition region, although the quently reflected into the structure of the amorphous
actual melting temperature also seems a reasonable body in glassy state,(48,88,89) which depends on the way
limit to be considered. A deeper connection between of cooling and will have this irregular character.
and relaxation processes seems to exist, as it was Microscale inhomogeneities linked to the glass
recently discussed by Prevosto et al.(86) transition process were also found in frozen aqueous
solutions of organic solutes. Sikora et al(90) found that
as a result of a phase separation processes accompa-
4. Structural features of the glassy state:
nying the freezing process of aqueous solutions of
nanoscale or microscale heterogeneities
sucrose from 120C to 30C, at least two crystalline
The role of the heterogeneities developed within the phases and two amorphous glasses could generally
liquid phase, now a true experimentally sustained fact be clearly identified, while solutions of composition
(largely recognized as a main characteristic feature of close to the eutectic were usually found to freeze
the transition region), was first treated trough the as- directly into glass.
sumption of a semi-crystalline phase by Kauzmann,(14) A feasible intervention of heterogeneity within the
as well as by an assumption of coexistence of gasliquid range of a so called mediumrange order (or modu-
(semi-evaporated) structural units related to numer- lated structures) becomes particularly common
ous works, e.g. Cohen & Turnbull.(46,48) The nonlinear in resolving the liquid and glassy states of various
oscillators contribute also to an abnormal coefficient non-crystalline semiconductors, where the concept
of thermal expansion and the amorphous-like liquid of inhomogeneous random network was most exten-
is thus structurally different from the amorphous-like sively studied.(1) This is also connected with the limits
solids, which casts some light into the explanations where the ordered and disordered states transpire its
regarding the Kauzmann paradox.(14,87) Below Tg factual distinguishing (delimitability). The standard
thermodynamic properties, as the specific heat and way for such observations, based on measuring the
the coefficient of expansion, decrease noticeably and crystallographic characteristics and the amount of
simple extrapolations towards lower temperatures crystalline phases (typically by XRD), is able to detect
lead to the expectancy that the supercooled liquid the amount of crystalline phase down to about 2%
would ultimately decrease to lower energy, entropy, within the glassy matrix (certainly under the restrain
and molar volume, as compared to the crystal, which of a minimum crystal-size detectability).(2,91)
seems unreasonable, hence the paradox. The Kauz- Since we are not concerned here with the capability
mann paradox is based on extrapolation and does to distinguish a minimum of crystallite magnitude
not take into account the fact that the liquid is a (which can be accessed by traditional XRD peak
mechanically heterogeneous substance in which the broadening), neither we account for a specialized
heterogeneity is partially removed as the tempera- diffraction measurement at low diffraction angles
ture goes down below Tg, and the highly nonlinear (to obtain the radial distribution function), we can
oscillators disappear. The liquid phase is structurally direct our attention towards the critical amount of
different from the solid in mechanical sense, viz. by crystalline phase that subsists in the glassy sample.
the appearance and disappearance of highly mobile, This issue is yet befitting the crucial question of how
diffusive, nonlinear oscillators. relevant is to define the limit of yet true glassiness
The blocks (inter-molecular clusters displaying and already nanocrystallinity. There are a few
correlated displacements), and their inter-blocks proposals but the generally accepted figure is, for
(adjacent rearranging clusters)(48) bonding structure, long, the value of 106 vol% (or a less common 103 %)
form the main contributing factor responsible for of crystallites existing within the glass matrix(2) as
dynamical parameters, such as the viscosity or bulk not yet disturbing its non-crystalline categorization
elasticity of the liquid phase. The blocks are also and consequent definition of glassiness. The appro-
responsible for complex relaxation effects, because priateness of this value is difficult to authorize, but,
the interconnected linear oscillators forming their however, keeps its continuance on basis of severe
structures interact. On the other hand, the semi- convenience.
evaporated particles acting as the nonlinear oscilla-
tors can perform their oscillations on several different
5. Concluding remark: on the distinction
amplitudes,(88) and their motions carry with them
between amorphous and glassy materials
elements of uncertainty that can be described, in most
cases, by the nonlinear, non-deterministic theories of According to Roy(92) the concept of non-crystallinity
chaos. Because the initial positional coordinates of (meaning structurally disordered) should be con-
such nonlinear oscillator for its subsequent position sidered as hierarchically superior to the glassy and
and momentum cannot be determined in advance, amorphous concepts, the latter being more popular
the differential changes in initial conditions will bring to describe certain types of semiconductors and met-
completely different trajectories. The general rule for als. These are, however, two related but misleading

170Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

12. Di Marzio, E. A. & Yang, A. J. M. Configuration entropy approach to

controversial concepts as it was noticed by Gupta,(93) the kinetics of glasses, J. Res. Nat. Inst. Stand. Tech., 1997, 102, 135157.
as he demonstrated that non-crystalline solids could 13. Baschnagel, J., Wolfgardt, M., Paul, W. & Binder, K. Entropy theory
and glass transition: A test by Monte Carlo simulation, J. Res. Nat. Inst.
be divided into two distinct classes, amorphous
Stand. Tech., 1997, 102 (2), 159170.
and glassy solids, because they behave differently 14. Kauzmann, W. The nature of the glassy state and the behaviour of
upon heating, and he proposed a structure-based liquids at low temperatures, Chem. Rev., 1948, 43, 219256.
15. Nieuwenhuizen, Th. M. Ehrenfest relations at the glass transition:
definition for distinctions. Accordingly Gupta, a
solution to an old paradox, Phys. Rev. Lett., 1997, 79 (7), 13171320.
non-crystalline solid is a glass if the short range order 16. Nieuwenhuizen, Th. M. Thermodynamic description of a dynamical
of the freeze-in solid below glass transition and that glassy transition, J. Phys. A, 1998, 31, L201L207.
17. Stillinger, F. H. & Debenedetti, P. G. Phase transitions, Kauzmann
of the delivering melt are the same. Glasses are also
curves, and inverse melting, Biophys. Chem., 2003, 105, 211220.
distinctive on exhibiting a glass transition upon cool- 18. Wondracze, L. & Conradt, R. Preface Glass and entropy, J. Non-Cryst.
ing and/or heating. These two conditions are always Solids, 2009, 355 (1012), viiviii.
19. Angell, C. A. Entropy and fragility in supercooling liquids, J. Res. Nat.
satisfied for any non-crystalline solids, which are
Inst. Stand. Tech., 1997, 102, 171185.
obtained by way of melt cooling because the liquid 20. Nieuwenhuizen, Th. M. Formulation of thermodynamics for the glassy
must undergo a glass transition to vitrify (if earlier state: Configurational energy as a modest source of energy, J. Chem.
Phys., 2001, 115 (17), 80838088.
crystallization is prevented). On the other hand, a
21. Pinal, R. Entropy of mixing and the glass transition of amorphous
non-crystalline solid is of an amorphous structural mixtures, Entropy, 2008, 10, 207223.
disposition if the short range order is not equal for 22. Gutzow, I., Petroff, B., Mller, J. & Schmelzer, J. W. P. Glass transition
and the Third Principle of thermodynamics: reconsideration of a clas-
the melt and its solid non-crystalline counter-partner,
sical problem, Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B, 2007, 48
which is exhibited, for instance, by some amorphous (3), 168177.
semiconductors like Si or Ge (showing, e.g. different 23. Adam, G. & Gibbs, J. H. On the temperature dependence of cooperative
relaxation properties in glass-forming liquids, J. Chem. Phys., 1965, 43,
coordination number(94)). Such amorphous solids may
be produced by special techniques such as sol-gel, 24. Scherer, G. W. Use of the Adam-Gibbs equation in the analysis of
precipitation from solutions, sputtering, mechani- Structural Relaxation, J. Am. Ceram. Soc., 1984, 67 (7), 504511.
25. Hodge, I. M. Adam-Gibbs formulation of non-linear enthalpy relaxa-
cal alloying, impact at high pressures (explosion),
tion, J. Non-Cryst. Solids, 1991, 131133, 435441.
milling, bombardment of high energy particles, or 26. Bottinga, Y. & Richet, P. Silicate melt structural relaxation: rheology,
by other modes of crystal disintegration (generally kinetics, and Adam-Gibbs theory, Chem. Geol., 1996, 128, 129141.
27. Richert, R. & Angell, C. A. Dynamics of glass-forming liquids. V. On
termed as amorphization).
the link between molecular dynamics and configurational entropy, J.
Chem. Phys., 108, 1998, 90169026.
28. Avramov, I. Viscosity of glassforming melts, Glastech. Ber. Glass Sci.
Acknowledgement Technol., 71C, 1988, 198203.
29. Shell, M. S., Debenedetti, P. G., La Nave, E. & Sciortino, F. Energy
This work has been supported by the Portuguese landscapes, ideal glasses, and their equation of state, J. Chem. Phys.,
Foundation for Science and Technology (FCT) and by 2003, 118, 88218830.
30. Matsuoka, S. Entropy, free volume, and cooperative relaxation, J. Res.
the Czech Science Foundation (GAR: P204/10/02112,
Nat. Inst. Stand. Tech., 1997, 102 (2), 213228.
MMT: 1M 06031 and MPO: FR-T11/335). 31. Franz, S. & Parisi, G. Effective potential in glassy systems: theory and
simulations, Physica A: Stat. Theor. Phys., 1998, 261, 317339.
32. Stillinger, F. H. Supercooled Liquids, Glass Transitions, and the
References Kauzmann Paradox, J. Chem. Phys., 1988, 88, 78187825.
1. Fritzche, H. What are Non-Crystalline Semiconductors, Vol. 25 in: 33. Louzguine-Luzgin, D. V. & Inoue, A. The outline of glass transition
Fundamental physics of amorphous semiconductors, Ed. F. Yonezawa, phenomenon derived from the viewpoint of devitrification process,
Springer-Verlag, Heidelberg, 1981. Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B, 2009, 50 (1), 2730.
2. estk, J. Glasses: phenomenology of non-isothermal glass formation 34. Ehrenfest, P. Phasenumwandlungen im ueblichen und erweiterten
and crystallization, in: Kinetic Phase Diagrams: non-equilibrium phase sinn, classifiziert nach dem entsprechenden singularitaeten des
transitions, Eds. Z. Chvoj. & J. estk, Elsevier, Amsterdam, 1991, thermodynamischen potentiales, Proc. Koninklijke Nederlandse Akademie
164277. van Wetenschappen, 1933, 36, 153157.
3. Tool, A. Q. & Eichlin, C. G. Variations caused in the heating curves of 35. Jaeger, G. The Ehrenfest classification of phase transitions: introduction
glass by heat treatment, J. Am. Ceram. Soc., 1931, 14, 276308. and evolution,Arch. History Exact Sci.,1998, 53 (1), 5181.
4. Tool, A. Q. Viscosity and the extraordinary heat effects in glass, J. Res. 36. Pippard, A. B. Elements of Classical Thermodynamics, Cambridge
Nat. Bur. Stand., 1946, 37, 7390. University Press, Cambridge, 1981, 136159.
5. Tool, A. Q. Relation between inelastic deformability and thermal 37. Dyre, J. C., Hechsher, T. & Niss, K. A brief critique of the Adam-Gibbs
expansion of glass in its annealing range, J. Am. Ceram. Soc., 1946, 29, entropy model, J. Non-Cryst. Solids, 2009, 355, 624627.
240253. 38. Hecksher, T., Nielsen, A. I., Olsen, N. B. & Dyre, J. C. Little evidence
6. Tool, A. Q. Effect of heat-treatment on density and constitution of for dynamic divergences in ultraviscous molecular liquids, Nature
high-silica glasses of the borosilicate type, J. Am. Ceram. Soc., 1948, Phys., 2008, 4, 737741.
31, 177186. 39. Kirkpatrick, T. R., Thirumalai, D. & Wolynes, P. G. Scaling concepts
7. Badrinarayanan, P., Zheng, W., Li, Q. & Simon, S. L. The glass transition for the dynamics of viscous liquids near an ideal glassy state, Phys.
temperature versus the fictive temperature, J. Non-Cryst. Solids, 2007, Rev. A, 1989, 40 (2), 10451054.
353, 26032612. 40. Wolynes, P. G. Entropy crises in glasses and random heteropolymers,
8. Gupta, P. K. Fictive pressure effects in structural relaxation, J. Non- J. Res. Nat. Inst. Stand. Tech., 1997, 102 (2), 187194.
Cryst. Solids, 1988, 102, 231239. 41. Xia, X. & Wolynes, P. G. Fragilities of liquids predicted from the
9. Gibbs, J. H. & DiMarzio, E. A. Nature of the glass transition and the random first order transition theory of glasses, Proc. Natl. Acad. Sci.
glassy state, J. Chem. Phys., 1958, 28, 373383. USA, 2000, 97 (7), 29902994.
10. DiMarzio, E. A. Equilibrium theory of glasses - an equilibrium theory of 42. Eastwood, M. P. & Wolynes, P. G. Droplets and the configurational
glasses is absolutely necessary, Annals New York Acad. Sci., 1981, 371, 1-1. entropy crisis for random first-order transitions, Europhys. Lett., 2002,
11. Di Marzio, E. A. The entropy theory of glass formation after 40 years, 60 (4), 587593.
Comput. Mater. Sci., 1995, 4, 317324. 43. Stevenson, J. D. & Wolynes, P. G. Thermodynamic-kinetic correlations

Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010 171
C. A. Queiroz & J. estk: Aspects of the non-crystalline state

in supercooled liquids: a critical survey of experimental data and Cryst. Solids, 1991, 131133, 153160.
predictions of the random first-order transition theory of glasses, J. 71. Nieuwenhuizen, Th. M. Thermodynamics of the glassy state: effective
Phys. Chem. B, 2005, 109 (31), 1509315097. temperature as an additional system parameter, Phys. Rev. Lett., 1998,
44. Dzero, M., Schmalian, J. & Wolynes, P. G. Activated events in glasses: 80, 55805583.
The structure of entropic droplets, Phys. Rev. B, 2005, 72, 100201- 72. Nieuwenhuizen, Th. M. Thermodynamic picture of the glassy state
1/100201-4. gained from exactly solvable models, Phys. Rev. E, 2000, 61, 267292.
45. Lubchenko, V. & Wolynes, P. G. Theory of structural glasses and 73. Johari, G. P. & Goldstein, M. Viscous liquids and the glass transition.
supercooled liquids, Ann. Rev. Phys. Chem., 2007, 58, 235266. II. Secondary relaxations in glasses of rigid molecules, J. Chem. Phys.,
46. Cohen, M. & Turnbull, D. Molecular transport in liquids and glasses, 1970, 53 (6), 23722388.
J. Chem. Phys., 1959, 31 (3), 11641169. 74. Johari, G. P. Intrinsic mobility of molecular glasses, J. Chem. Phys.,
47. ernoek, Z., Holubov, J. & ernokov, E. Kauzmann temperature 1973, 58 (4), 17661770.
and the glass transition, J. Optoelectr. Mater., 2005, 7 (6), 29412944. 75. Johari, G. P. Glass transition and secondary relaxations in molecular
48. Hlavek, B., estk, J., Koudelka, L., Moner, P. & Mare, J. J. Forms liquids and crystals, Ann. New York Acad. Sci., 1976, 279, 117140.
of vibrations and structural changes in liquid state, J. Thermal Anal. 76. Wang, L.-M., Tian, Y., Liu, R. & Richert, R. Calorimetric versus kinetic
Calor., 2005, 80, 271283. glass transitions in viscous monohydroxy alcohols, J. Chem. Phys., 2008,
49. Vogel, D. H. Das temperatur-abhngigkeitsgesetz der viskositt von 128 (8), 084503-1084503-8.
flssigkeiten, Physikalische Z., 22, 1921, 645646. 77. Angell, C. A., Ngai, K. L., McKenna, G. B., McMillan, P. F. & Martin, S.
50. Fulcher, G. S. Analysis of recent measurements of the viscosity of W. Relaxation in glassforming liquids and amorphous solids, J. Appl.
glasses, J. Am. Ceram. Soc., 1925, 8 (6), 339355. Phys., 2000, 88 (6), 31133157.
51. Tammann, G. & Hesse, W. Die abhngigkeit der viskositt von der 78. Ojovan M. I. & Lee, W. E. Fragility of oxide melts as a thermodynamic
temperatur bei unterkhlten flssigkeiten, Z. Anorg. Allgemeine Chem., parameter, Phys. Chem. Glasses, 2005, 46 (1), 711.
1926, 156, 245257. 79. Johari, G. P. Localized molecular motions of -relaxation and its energy
52. Donth, E. Characteristic length of glass transition, J. Non-Cryst. Solids, landscape, J. Non-Cryst. Solids, 2002, 307310, 317325.
1991, 131133, 204206. 80. Gtze, W. & Sjgren, L. The mode coupling theory of structural
53. Donth, E. The Glass Transition, Relaxation Dynamics in Liquids and relaxations, Transp. Theory Stat. Phys., 1995, 24, 801853.
Disordered Materials, Springer Verlag, Berlin & Heidelberg, 2001. 81. Roland, C. M. & Casalini, R. Density scaling of the dynamics of vitrify-
54. Hunt, A. An explanation for the Kauzmann paradox and its relation ing liquids and its relationship to the dynamic crossover, J. Non-Cryst.
to relaxation times, J. Non-Cryst. Solids, 1994, 175, 129136. Solids, 2005, 351, 25812587.
55. Leuzzi, L. & Nieuwenhuizen, T. M. Effective temperatures in an 82. Rssler, E. Experimental results revealing a change of diffusion
exactly solvable model for a fragile glass, Phys. Rev. E, 2001, 64, mechanism above Tg, J. Non-Cryst. Solids, 1991, 131133, 242247.
011508-1/011508-24. 83. Kudlik, A., Tschirwitz, C., Benkhof, S., Blochowicz, T. & Rssler, E.
56. Leuzzi, L. & Nieuwenhuizen, T. M. Inherent structures in models for Slow secondary relaxation process in supercooled liquids, Europhys.
fragile and strong glass, Phys. Rev. E, 2001, 64, 066125-1/066125-15. Lett., 1997, 40, 649654.
57. Odagaki, T., Matsui, J. & Hiwatari, Y. On the glass transition singulari- 84. Gtze, W. & Sjogren, L. -relaxation near the liquid-glass transition,
ties and slow dynamics, Physica A, 1996, 224, 7483. J. Phys. C: Solid State Phys., 1987, 20, 879894.
58. Birge, N. O. & Nagel, S. R. Specific-heat spectroscopy of the glass 85. Gtze, W. Aspects of structural glass transitions, in Liquids, Freezing
transition, Phys. Rev. Lett., 1985, 54, 26742677. and Glass Transition, Eds. J. P. Hansen, D. Levesque & J. Zin-Justin,
59. Bottinga, Y., Richet, P. & Sipp, A. Viscosity regimes of homogeneous 1991, Elsevier Science Publishers, Amsterdam, 287503.
silicate melts, Am. Mineral., 1995, 80, 305318. 86. Prevosto, D., Capaccioli, S., Lucchesi, M., Rolla, P. A. & Ngai, K. L. Does
60. Eckstein, B. Vitreous states, Mater. Res. Bull., 1968, 3 (2), 199207. the entropy and volume dependence of the structural -relaxation
61. Reiner, M. The Deborah Number, Phys. Today, 1964, 17, 6262. originate from the JohariGoldstein -relaxation? J. Non-Cryst. Solids,
62. Giordano, M., Russo, M., Capoluongo, P., Cusano, A. & Nicolais, L. 2009, 355, 705711.
The effect of cooling rate on the glass transition of an amorphous 87. Angell, C. A. Thermodynamic aspects of the glass transition in liquids
polymer, J. Non-Cryst. Solids, 2005, 351, 515522. and plastic crystals, Pure Appl. Chem., 1991, 63, 13871392.
63. Hutchinson, J. M. Determination of the glass transition temperature: 88. Hlavek, B., estk, J. & Mare, J. Mutual interdependence of parti-
Methods correlation and structural heterogeneity, J. Therm. Anal. tions functions in the vicinity of glass Tg-transition, J. Therm. Anal.
Calorim., 2009, 98, 579589. Cal., 2002, 67, 239248.
64. Suga, H. & Seki, S. Frozen-in states of orientational and positional 89. estk, J. Thermal dynamics and non-equilibrium phenomena, in:
disorder in molecular solids, Faraday Discuss. Chem. Soc., 1980, 69, Heat, Thermal Analysis and Society, NUCLEUS publ. house, Hradec
221240. Krlov, 2004, 222.
65. Wong, J. & Angell, C. A. Glass Structure by Spectroscopy, Marcell Dekker 90. Sikora A.,Dupanov,V. O., Kratochvl, J. & Zmenk, J. Transitions
Inc., New York, 1976, 23. in aqueous solutions of sucrose at subzero temperatures, J. Macromol.
66. Maxwell, J. C. On the dynamical theory of gases, Phil. Mag., 35, 1868, Sci. B, 2007, 46, 7185.
129145 & 185217. 91. estk, J. Some thermodynamic aspects of glassy state, Thermochim.
67. Dyre, J. C. Colloquium: The glass transition and elastic models of glass Acta, 1985, 95 (2), 459471.
forming liquids, Rev. Mod. Phys., 2006, 78, 953972. 92. Roy, R. Classification of non-crystalline solids, J. Non-Cryst. Solids,
68. Dingwell, D. B. & Webb, S. L. Relaxation in silicate melts, Eur. J. 1970, 3, 3340.
Mineral., 1990, 2, 427449. 93. Gupta, P. K. Non-crystalline solids: glasses and amorphous solids, J.
69. Webb S. & Knoche, R. The glass-transition, structural relaxation and Non-Cryst. Solids, 1996, 195, 158164.
shear viscosity of silicate melts, Chem. Geol., 1996, 128, 165183. 94. Gaskell, P. H. On the structure of simple inorganic amorphous solids,
70. Sjgren, L. & Gtze, W. -relaxation near the glass transition, J. Non- J. Phys. C: Solid State Phys., 1979, 12, 43374368.

172Physics and Chemistry of Glasses: European Journal of Glass Science and Technology Part BVolume 51Number 3June 2010
Journal of Thermal Analysis and Calorimetry, Vol. 80 (2005) 271283



B. Hlavek1*, J. estk2, L. Koudelka3, P. Moner3 and J. J. Mare 2

Department of Polymers, University of Pardubice, s. Legi 565, 532 10 Pardubice, Czech Republic
Institute of Physics, Academy of Sciences, Cukrovarnick 10, 162 53 Prague 6, Czech Republic
Department of General and Inorganic Chemistry, University of Pardubice, s. Legi 565, 532 10 Pardubice, Czech Republic

The forms of vibrations and displacements of particles in amorphous structures have been investigated. The particles, moving on
highly non-linear amplitude, are responsible for the creation of disordered structures of amorphous bodies. The non-linear oscilla-
tors, even if few in concentration, are characterized by unpredictable trajectories in phase space. The non-linear oscillators are
fully developed in the liquid state above the crossover temperature Tcr and between Tcr and Tg their number decreases. Under Tg
they completely disappear. The interconnection between the linear oscillators in blocks plays the most important role in the charac-
teristic time spectra in liquid state. Using the additive properties of elements polarizibilities, the number of acoustical units in indi-
vidual blocks at Tcr is estimated to be about 600 units. The diameter of blocks at Tcr was estimated to be about 1.8 nm. Even if the
non-linear high amplitude motions disappear at solidification, the remnants of structural irregularity remain and the disordered
structure of glass is formed.

Keywords: crossover temperature, deterministic chaos, glass transition, non-linear oscillators, solidliquid transition

Introduction ter of their motions and are taken, in the present study,
as simple individual units about a monomer size [12].
Substantial fluctuations of liquid density in the area of On the other hand, the blocks are assumed to possess
critical temperature Tc appear as observable phenomena interconnected microstructure and are composed of
visible to naked eye [1]. The facts that liquid phase un- identical elastically bonded particles packed to the
der critical temperature is divided into solid-like struc- high (density) level reaching the density of van der
tures and voids filled up with gas-like molecules (the so Waals liquid phase (Fig. 1).
called wanderers), has been well known for a long The blocks are characterized by a module, which
time [2]. Some of the modern structural theories reaches the level of glassy state and keeps the swarm
(so-called mode coupling theory, MCT) describing the of particles in one block together. A dynamical equi-
structural phenomena of liquid state at lower tempera- librium exists between the block particles and the
tures are also based on a similar idea of local density semi-evaporated particles, as well as between the
fluctuation as well [3]. The assumption of heterogeneity semi-evaporated particles and the gas phase. The ex-
in liquid phase goes back to assumption of semi-crystal- istence of randomly distributed, random size molecu-
line phase by Kauzman [4], as well as to the assump- lar clusters in a viscous liquid has been discussed
tions of coexistence of gas liquid semi-structures related since 1959 till present days [511, 13]. Loosely
to numerous works of Cohen et al. [511]. packed regions separate these clusters. Here we con-
sider that the inter-clusters regions contain molecules
whose vibrations are strongly anharmonic, giving rise
Model structure to non-deterministic vibration movements.
The blocks and their inter-blocks bonding struc-
In this study we take also the liquid structure as a me- ture form the main contributing factor responsible for
chanically divided structure formed by blocks (do- dynamical parameters, such as the viscosity or bulk
mains, or clusters) on one side, and the individual elasticity of liquid sample. The blocks are also re-
semi-evaporated units, which are subjected to sponsible for complex relaxation effects, because the
non-linear anharmonic motions of high amplitudes, interconnected linear oscillators forming their struc-
on the other side. It is assumed that the highly tures interact. On the other hand, the semi-evapo-
non-linear oscillators maintain the individual charac- rated particles acting as the non-linear oscillators are

* Author for correspondence:

13886150/$20.00 Akadmiai Kiad, Budapest, Hungary

2005 Akadmiai Kiad, Budapest Springer, Dordrecht, The Netherlands
HLAVEK et al.

can perform their oscillations on several different am-

plitudes [15, 16], and their motions carry with them
elements of uncertainty described most cases by the
non-linear, non-deterministic theories [1725] of
chaos. Because the initial positional coordinates of
such non-linear oscillator for its subsequent position
and momentum cannot be determined in advance, the
differential changes in initial conditions will bring
completely different trajectories. The general rule for
the non-deterministic chaos theories [1725] is
consequently reflected into the structure of amorp-
hous body in glassy state, which depends on the way
of cooling and will have the irregular character.
For the displacement of block as a whole, the
maximum retardation time will show the level of in-
terconnections inside of block structure, characteris-
tic of a given block size, which changes with tempera-
ture (Fig. 3). It is assumed that the block size, together
with maximum block retardation time max, decreases
with increasing temperature. The block disintegrates
as the critical temperature is approached [15, 16]. In
the opposite case, the blocks bind themselves to a
rigid structure as the temperature decreases below Tcr.

The amplitudes of vibration movements

below the critical temperature

Under the critical temperature Tc we consider the fol-

Fig. 1 Schematic model for solid-like structure just above the lowing most important marking points, as the charac-
Tg temperature, when the first particle starts to act as teristic temperatures for amorphous body, regarding
non-linear oscillator and causes irregularity in structure the vibration amplitudes size:
of linear elements. The separation of blocks causes in-
terference with process, which forms the infinite
the boiling point temperature Tb,
structural network
the crossover temperature Tcr,
the transition temperature of glassy state Tg,
responsible for erratic character of Brownian motion Vogels temperature Tv.
displacements in liquids [14]. The semi-evaporated It can be stated: Tc Tb Tcr Tg Tv (the model
particles are acting against the external pressure and proposed has the following distinct characteristics of
exercise a push aside effect on the individual blocks vibration modes at different temperatures or tempera-
in their vicinity. The opening thus created (Fig. 2) ture areas, which should be mentioned and pinpointed
through straightforward amplitude jump, can be di- beforehand for the text to be clearer).
rected to any direction in amorphous phase. The
non-linear oscillators of semi-evaporated particles

Fig. 2 Sudden amplitude jump of non-linear particle (a), (b) can be performed in any direction (case c) in the sample space

272 J. Therm. Anal. Cal., 80, 2005


be performed successfully by particles in vicinity below

the liquid point Tb, without the destruction of the sample
through the creation of a gas phase in the sample bulk.
The level of bottom of van der Waals curve for iso-
thermal case can be found easily for small volumes by
looking for the intercept of spinodal curve [26] with the
low volume of van der Waals dependence. The intercept
point is at volume fraction of 0.385 of critical volume.
At boiling point Tb, the non-linear movement stops for
some semi-evaporated particles, as the bond of particle
to a strange attractor of Lorenz type [23, 24] enlarges
its basin of attraction and is eventually interrupted as the
particle leaves the oscillatory movement for a gas phase.
Between the crossover temperature and the boiling
point, the high amplitudes of vibration are able to me-
chanically perturb the original bonding of blocks to such
a level, that shear module disappears. In contrast to be-
havior of material at this higher temperature zone, be-
low the crossover temperature, the substantial rise in
shear module G contributes (as a new element) to limita-
tion in vacancy enlargement, as the sample is able to
store the elastic energy W connected to shear module
[27], inside of the vacancy space:
W = 8Gr0 ( r1 r0 ) 2 (1)
The radius r0 represents the radius of a rigid sphere
approximating the particle, and r1 is approximating
the size of radius of the cavity. For example the cavity
of size r1=2r0 formed in solid with modulus G of
about G=1010 N m2 needs about 125 kJ mol1 [27],
while in liquids of van der WaalsEyrings type, such
level of internal void expansion inside of bulk of liq-
uid sample, leads to Troutons rule [2].
The non-zero shear module below the Tcr has
drastic impact on vacancy size and on the sizes of
maximal diffusion jumps, which are depressed or elimi-
nated as the temperature decreases. The presence of
Fig. 3 Schematic picture for a liquid-like structure where the inter-block bonding will bring the shear module to the
linear elements represented in blocks by Voigt series non-zero level and the elastic network will be formed.
are not connected through the elastic network and are The fact of different ways of inter-block bonding has
mobile, being restricted in their mobility by balance of basic influence also on the fragility phenomenas
saturation pressure Psat and by the external pressure Pext [28, 29]. Above the crossover temperature the diffusion
only (Fig. 1). The blocks are a source of cooperative
phenomena of sample time response
process, is driven mainly by few particles, out of the
block areas, which are undergoing the amplitudes
The bottom part of van der Waals curve on the switch. Thus the relation formed by product of viscosity,
liquid side plays the important role for the evaluation size of the diffusing particle and diffusion coefficient,
of non-linear amplitude size. divided by absolute temperature and Boltzmanns
The onset of the spinodal part of van der Waals constant [30, 31] cannot retain its form below Tcr. The
curve seems to pose a sensible restrain on the upper most apparent impact of shear module can be demons-
amplitude of vibration motion. This limit, if exceeded trated on the temperature dependence of the alpha and
has to lead to vacancy enlargement in which the parti- beta relaxation processes. (Also this topic will be
cles in vicinity of high-amplitude oscillator have to treated more closely in the following text). In Tg vicinity
take away the vibration energy of particle, by their own the changes in vibration amplitudes of average vibra-
avalanche like displacement. This type of energy with- tions can be also demonstrated by the experimental
drawal from high amplitude vibrating particle can only measurements of DebyeWaller factor [32].

J. Therm. Anal. Cal., 80, 2005 273

HLAVEK et al.

Earliest experimental justification of the plated as well [15, 16]. For the temperatures T Tg the
structural heterogeneity of liquids entropy S main contributions can be explained as:
X-ray diffraction S k B [ln Wth + ln Wconf ] (2)
The X-ray diffraction studies are the oldest methods and for T Tg
supporting the existence of dual structure in bulk of liq- S k B [ln Wth + ln Wconf + ln Wsemievap ] (3)
uids [33]. These studies have shown that for a liquid,
even above its boiling point, the coordination number
(characterizing the number of particles in the closest vi-
cinity) falls drastically in the area of temperatures be-
tween boiling point and Tc, but the distance characteris-
tic of the first peak of radial distribution g r of particles
in vicinity hardly changes as the temperature ap-
proaches the critical temperature region. On the other
hand, the experimental proof of compact structure, ex-
pressed through unchanged radial distribution g(r), im-
plies the existence of voids and cavities filled with
semi-evaporated particles, to make up for the overall
volume of the sample characterized by average density.

The characteristic of local heterogeneity using the

positron annihilation spectroscopy
The volumes of vacancies are exactly determined and
are reported in numerous papers on positron annihila-
tion spectroscopy PASCA [3438] (Fig. 4). The ex-
periments of PASCA provide the unusually high co-
efficients of thermal expansion in the vacancy area.
The volumes of vacancies are very sensitive to tem-
perature changes above the glassy transition and to
the external pressure changes in the boiling point
area. In the present study, the vacancies in liquid state
contain the semi-evaporated particles without which
the transition from bulk of sample to gas phase at
boiling point would not be possible. As can be seen
from the data of Fig. 4 the coefficient of thermal ex-
pansion in vacancies areas is about ten to a hundred
times as high as that in the block areas. At tempera-
tures T Tg (for the amorphous bodies and the whole
piece of sample) the coefficient of thermal expansion
is 11104 K1. At temperatures T Tg the coeffi-
cient of thermal expansion is 2=4.1104 K1. The
PASCA readings at crossover temperature are higher
than V,14103 K1. The discontinuity in properties
for liquid structure is apparent. It could be contem-
plated that the liquid contains locally expanding
spots, which bear a responsibility for the high coeffi-
cient of thermal expansion of liquids in general.

Fig. 4 a The stepwise rise of vacancy volume in the vicinity

Three types of entropy contributions of Tg=168 K for polybutadiene; the data provided by
Barto et al. [19] are gratefully acknowledged, b The
The entropy contribution connected with the semi-evap- coefficient of volumetric thermal expansion calculated
orated state (Fig. 2), which is created inside the liquid from Fig. 4a, for the vacancies, and the data of ap-
proximately taken from literature [43, 60] quoted for
system in vacancies areas, can potentially be contem-
the sample at glassy state

274 J. Therm. Anal. Cal., 80, 2005


This type of entropy changes composed of three term U0n stands for the basic level of internal energy,
distinct contributions, viz. thermal, conformational f n , gn are characteristic parameters of local potential
and highly non-linear, has been experimentally con- valley, which depend on positions of particles in the
firmed in recent works of Johari [39, 40]. Because the closest vicinity. The equation of motion for particles
vacancies areas, thanks to the reports on PASCA ex- in the form of Eq. (4) is the same for block particles as
periments, have well defined size (which is larger for the semi-evaporated particles. However, if Eq. (5)
than the van der Waals volume, but smaller than is the is substituted into Eq. (4), great differences will ap-
0.385 fraction of critical volume of the particles in- pear in the solutions obtained between the individual
volved), we can estimate the change in enthalpy con- cases for gn0, r0,n(t)=const. (the case of ideal solids)
nected with the semi-evaporated state. As has been or for gn0 (the case of real solids or liquids) or when
shown by Hirschfelder et al. [2], the whole amount of gn(t), f n (t), U0n(t) vary with time. The last example
evaporation enthalpy is needed to produce an expan-
sion of cavity to level of twice as high as is the parti- represents the structural changes in solids and the dif-
cle diameter. So to produce a cavity of size about fusion or flow onset for liquids.
0.385 of fraction of critical volume, a proportionally Beside Eq. (4) in variables and t an additional
smaller part of evaporation enthalpy Hevap can be equation for description of r0,n(t) development in time
used, as the first estimate for the energy of void cre- (for the bottom well displacements) would be needed.
ation. The assumption of the Hsemievap=Hevap/n,
where n is about 24 or even higher, will be in agree- The mathematical treatment of strong non-linearity
ment with the PASCA experiments as well as with the for one oscillator above T g transition
estimate on viscosity by Eyring et al. [41, 42].
To find fast solution to Eq. (4) and provide illustrative
examples that can be easily visualized, the second-or-
The basic phase space behavior of semi- der differential Eq. (4) is usually turned into two sepa-
evaporated particles highly non-linear rate first-order differential equations. This procedure is
oscillator performed in the deterministic chaos theories [1725]
as well as in studies of organized structures or in
studies of non-equilibrium thermodynamics [21, 22].
The equation of motion for particle vibrations By choosing the variables A1 and A2d/dt we
The equation of vibration motion for non-linear oscil- can rewrite Eq. (4) into (two first order differential
lations of nth particle is: equations) the following forms [15, 16]:

2 U n dA1
mn + = Fext, n (4) = 11 A1 + 12 A2 (6)
t 2 dt
where the oscillator external force Fext,n(t) is created = 21 A1 + 22 A2 (7)
through action of many particles in the vicinity, and mn dt
is the particle mass, comparable in size with dimen- Equations (6) and (7) can be analyzed further in the
sions of acoustical unit [12] of about monomer size. Un vicinity of vibration stationary point. The system of
is the potential energy characteristic of the potential Eqs (6) and (7) will have non-zero solution only if
valley in which the nth vibration acoustical unit is per- 11 12
forming its oscillations. The term =rr0,n is the devia- det = 0 = 2 + a 1 + a 2 = 0 (8)
21 22
tion from the bottom of local potential valley.
The vibrations of particle take place inside of the where a1=11+11, 2=11221221.
potential valley Un(t) of the nth oscillator: The changes from a solid to a liquid state, as well
as the other higher temperature transitions, can be in-
U n ( t ) = U 0n ( r0,n ( t ) ) +
vestigated through the change of parameters a1 and a2
(5) within a certain range as shown in our previous com-
+ f n ( r0,n ( t ) ) 2 g n ( r0. n ( t ) ) 3
1 1
2 3 munication [15, 16].
The mathematical reasoning gives the basis to
where in the first approximation the individual pa- the existence of vacancies, which stems from the con-
rameters U0n, f n , gn, r0,n for the blocks particles are siderations of high non-linearity of Eqs (48), sug-
constant and time-independent and vary smoothly gesting the particle trajectories have a non-determin-
with temperature only; r0,n is a vector pointing to the istic character. At solidification process, the end
bottom of potential valley of the nth oscillator. The points of the trajectories in phase space are aiming at

J. Therm. Anal. Cal., 80, 2005 275

HLAVEK et al.

different spots, which forms the irregular structure of The estimate on the size of blocks from the shear
the glassy state. The scheme for the amplitude devel- viscosity data
opment for two coordinates n, dn/dt for different
Numerous predictions regarding the size of the blocks
temperatures in phase space is shown in [15, 16].
are reported in literature by many authors
Following such line of perception, the higher transi-
[13, 5056]. These predictions are based upon differ-
tions such as the boiling or critical point can be con-
ent experimental techniques. Focusing the attention
sidered too. The fraction of vacancies can roughly be
now in a direction not yet used for the block size esti-
estimated through the density changes for the liquid
mate, it will be tried in this study to provide the block
sample [15]. While only few vacancies exist at lower
size approximation, using the techniques of rheology.
temperatures (according to Bueche [43] the ratio of
The answer is searched for two unknown parameters.
vacancies to vibrating particles is equal to 1:40 at Tg),
The first represents the number of linear elements in
at the critical temperature the amount of vacancies
block at which the blocks start to be interconnected in
reaches the level at which the condensed phase disin-
shear flow and start to interfere mutually producing a
tegrates completely [15, 16] and all oscillating parti-
contribution to the elastic shear module response. A
cles are able to reach the upper vibration amplitude
useful source of data, for the domains (blocks) inter-
with addition of very tiny amount of energy.
actions, is the measurements of shear viscosity data
obtained at zero velocity gradients, which have been
known for about forty years for different structures of
The linear cooperative phenomena in blocks
polymer melts with a variety of molecular mass val-
ues [5759]. These data show one common feature in
As a result of separation of the non-linear effects from
the dependence of shear viscosity on the number of
the linear blocks areas, some more than half-century-old
monomer (acoustical [12]) units in chain X sequence.
experimental results and mathematical techniques can
As is considered, as the first example, that the number
successfully be used for the investigation of very topical
X is small and remains below a certain limit, XXc,
time phenomena in liquidsolid transitions (following
where Xc500625, and that the polymer melt is char-
the temperature interval between TV and Tcr). The new
acterized by a constant and gradient-independent vis-
physical explanations about certain phenomena, nowa-
cosity. In this particular case the module of shear elas-
days intensively studied [13, 4456], connected with
ticity is zero, as well as the first normal stress difference
and processes, about the crossover temperature, and
remains zero too. The macromolecules are smaller than
also about the Kauzman paradox [4] can thus be re-
the block size (XXc), and the blocks, if subjected to
vealed, as the system changes its structure as it cools
shear flow, do not exhibit any mutual interference. A
down below the crossover temperature and under Tg, as
different level of viscosity, which rises in linear de-
the particles generating the voids and associated with
pendence on X, characterizes each polymer at this
the higher amplitudes of motion disappear.
particular case. However, at the point when the criti-
cal value Xc is reached in the polymer chains, there
The limiting values of block time characteristic occur changes in the rheological properties. The
block interconnections start to interfere through the
For the time dependence limits for the fastest time in-
overgrown size of macromolecules. For polymers
terval of retardation we use the time interval calcu-
with XXc the melt viscosity in dependence on shear
lated from the sound velocity, which for solid blocks
gradient starts to have the non-Newtonian character
is about 105106 cm s1, and from the minimal length
as the gradient grows and the first normal stress dif-
of acoustic waves, which are determined by the mesh
ference starts to be an important flow characteristic as
size r0,n
. Thus we get: 1/maxmin and vsound=
well. This effect is usually explained in connection
=minmax, where min 2r0,n .
with the chain entanglements. The concept of entan-
Substituting from the estimate of values outlined
glements was introduced originally base upon
above we get max1013 s1for the maximum fre- sketchy pictures [43, 60, 61] and as an evident phe-
quency in sample body and for the fastest time of ma- nomenon while the nature of couplings [60] was left
terial response in block min max1=1013 s. (The esti- as open, with the alternative to cooperative motions
mate on the fastest bottom well displacement based of molecular areas, which is also suited for the low
upon the experiment is quoted by Rossler [47], viz. molecular structures. The later case corresponds more
min=310 s.) For the longest time characteristic for closely to scheme presented in this study, which is
the material we take the characteristic time connected also supported by findings of Achibat et al. [80, 81],
with permanent displacement in series of elements in- reporting a relationship between the domains and the
side of block space and bonded together through elas- entanglements. The model presented attributes the
tic force (Eqs (14), (15)). property of inter-blocks bonding either to the entan-

276 J. Therm. Anal. Cal., 80, 2005


glements (as the individual blocks start to be interwo- as the temperature is increased above Tcr. The inflexi-
ven by polymer network) or to inflexible bonds of van ble blocks interconnection reflects itself into the in-
der Waals type, or to the bonds of chemical nature. ternal energy related shear module contribution under
The inter-block bonding has a consequence in devel- the Tcr. (This is valid for the blocks contacts for the
opment of non-zero shear module. Although different polymers under Tcr temperature too).
polymers have different levels of viscosity or chain The parameter left to be determined is the cross-
flexibility, different molecular mass, or exhibit differ- over temperature itself. For these purposes the mea-
ent levels of normal stresses, the numerical value of surements of the dependence of shear relaxation mod-
Xc for blocks mechanical interference in shear flow ule on temperature in area of main phase transition
has a universal character. Just quoting [59], for poly- can provide this information [62, 63] (Fig. 5). For the
styrene Xc=600, for poly(vinyl acetate) Xc=600, for reasons of clarity of explanations in text, some data of
polyisobutylene Xc=500, for polydimethylsiloxane Schultz [59], Alkonis [63] and Tobolsky [62] were
Xc=625. If the level of critical interconnection is presented to show the idea behind the text more
reached in linear sequence for polymers, the blocks clearly. The crossover temperature Tcr (Fig. 5) is de-
are inter-bonded and exhibit the shear elasticity repre- termined on the basis of the shear elasticity of internal
sented by shear modulus or by the first normal stress energy related relaxation module. The value Xc is de-
difference. Such elasticity created by flexible termined on the basis of the zero gradient viscosity
macromolecules is entropy related. (This type of in- curves. At this level of internal connection the blocks
terconnection has a long distance character rela- start to interfere in flow.
tively to the block size, and one macromolecule can At the temperature Tcr the characteristic domain
over-bridge several blocks too). In contrast to size can be estimated and for the lower temperatures
blocks connected by flexible macromolecules, the TTcr, the part of force responding to external deforma-
blocks can also be interconnected by inflexible chem- tion has internal energy related component (fINT0).
ical bonds. This is the case for the low molecular mass Taking the size of one oscillator unit in interconnection
substances under the crossover temperature. The sequence from the polarizability values [64, 65], for the
blocks connected by inflexible bonds cannot be interconnection number of Xc=600, the characteristic
stretched to higher deformations and have to break up size of a domain of about 6 nm3 (for the molecules of
most typical size of pentane) can be obtained. For the
linear dimension of a domain 61/3=1.81 nm can be ob-
tained. This seems to be in good agreement with the
published values [13, 5056] obtained by different ex-
perimental techniques. As the polarizability of mole-
cules is simple additive properties of individual atomic
polarizabilities, the domain size can be easily estimated,
from the polarizabilities values of individual monomer
unit and from the Xc values, as the Table 1 hints.

The spectra characteristic of the block

For the particles forming the block structures, in addi-
tion to the motion in coordinate, which does not lead
to linear particles diffusive displacements, an addi-
tional motion connected with r0,n displacements can be
superimposed on the movements of particles. Besides
the potential valley deformation the additional degree of
freedom, if added to the particle movements, will bring
by itself the increment in specific heat capacity cp.
For the differentially small changes in bottom
Fig. 5 The typical picture of relaxation module behavior of position r0,n (Eq. 5), the coordinated changes of par-
polymer. The sample is undergoing the temperature and
molecular mass change. (According to Tobolsky [62]
ticles positions, ...r0,(n1)r0,(n+1), r0,(n+2)r0,(n+k) in
and Alkonis [63].) The elongation relaxation modulus the vicinity of semi-evaporated participle are needed
of polystyrene, for two different molecular masses as the precondition for permanent displacement of the
Mw =M1M2 as a function of temperature. The relax- nth particle.
ation experiment is performed for a time of 10 s. The
area a stands for glassy zone and b is the rubber-like
zone [60, 61]

J. Therm. Anal. Cal., 80, 2005 277

HLAVEK et al.

Table 1 Calculated domains size from the polarizabilities of

acoustical units
Estimate on Domain
The type of
polarizability of size/nm
acoustical segment /3 for Xc=600
polystyrene 11.6 1.90
polyisobutylene 7.2 1.62
polybutadiene 6.4 1.55
8.0 1.68
9.8 1.95
SiO2 6.7 1.59
GeO2 7.6 1.65
B2O3 4.4 2.05
As2S3 17.3 2.18
glycerol 8.6 1.72
pentane 10.0 1.81

r0,n ( t ) = f ( 0,(n 1) ( t ).... r0,(n + 1) ( t ),

r0,(n + 2) ( t ), ...... r0,(n + k) ( t )

The f stands for functional dependence and is

taken as an unknown function of time t associated
with material characteristic series of discrete retarda-
tion times. The actual positions of r0,n, r0,(n+1) etc. does Fig. 6 A detail of Fig. 1. The individual particles in necklace
not enter into the mathematical formulation. The indi- are numbered according to their positions separating
vidual oscillators can be placed consecutively, in the them from the non-linear particle stress perturbation
space of sample, touching each other in form of neck- shown schematically above of particle '0'
lace (Fig. 6). The mathematical formula will involve
the particle consecutive order in chain time response r0,n
acting on each element f 0nFr expressed sepa-
and also the direction and size of deviations from t
equilibrium positions r0,n, r0,(n+1), etc. rately as shown in Eq. (11). The term Fext represents
In simplistic view of interconnections for one the interaction of non-linear oscillator (the
line of particles connected in the form of an elastic semi-evaporated particle), acting on the ends of chain
necklace such as shown in Fig. 6 we get: (point 0 in Fig. 6), by which the individual blocks
segments are displaced in time. The non-linear oscil-
d 2 r0,n
mn = (10)
lator of semi-evaporated particle acts as the local
dt 2 stress perturbation source on the necklace of elasti-
= k e ( rn, +1 r0,n ) k e ( r0,n r0,n -1 ) + Fext cally connected masses. The blocks play the role of
the sidewalls, and from mathematical point of view,
Through extension of Eq. (10) inside of block we get: they act as the source of exact time response of mate-
d 2 r0,n rial. For the case of Fext=0 we get a homogeneous
mn = equation as a source for the calculation of vibration
dt 2 (11) modes. These modes give the evidence about the ac-
= k e ( r0,n + 1 + r0,n 1 2r0,n ) + f 0nFr + Fext tual distribution of resonance frequencies and time re-
t sponses of material. The shifts in r0,n can be consid-
where Fext represents the external force acting on the ered in any direction inside of sample.
particle through the particles localized in its vicinity. d 2 r0,n
The simplest form of interconnections of elements, mn = k e ( r0,n + 1 + r0,n 1 2r0,n ) (12)
dt 2
without branching or bridging is considered. The term
Fext can involve the effects of local friction force f 0nFr

278 J. Therm. Anal. Cal., 80, 2005


Equations (1012) have many classical solutions The individual retardation times covering the
for many options of Fext and are solved in connection motions inside of blocks, are mutually interdependent
with many different topics of engineering or scientific through Eq. (15) (Figs 7, 8). The time for total block
applications [18, 27, 43]. The typical solution to ho- displacement is thus proportional to N 2 while the fric-
mogeneous Eq. (12) of relaxing block consists of se- tion force for the block movement, which character-
ries of resonance frequencies n, izes the shear viscosity of sample, remains propor-
tional to N only if the total block friction is defined as
4k n dr0,n
2n = e sin 2

mn 2( N + 1) (13)
i =1
f 0n
. It can be assumed that the friction coef-

n = 1, 2, 3, ..... N ficients f0,n of beads inside the solid icebergs vary

and leads to the time response of blocks characterized only mildly with temperature changes. However, it is
by series of relaxation or retardation times (for creep re- expected that the number N increase smoothly with
lated experiments) of interconnected elements [18, 43]. decreasing temperature. At the crossover temperature
(The solutions of Eqs (11)(13) are given in Ap- the inter-chain branching start to appear. The knots
pendix A and B. Please note also that for and bridges between the chains characterizing the
n structure, affect the relaxation times in different way
sin 2 and n the argument of from the viscosity, making the chain sequences
2( N + 1) 2 shorter. For example if two necklaces are connected
n n2 together, then the viscosity rises by factor of two. The
sin 2 const. is valid.)
2N 4N 2 rise in maximal relaxation time will depend on the
Solving Eq. (13) means finding the resonance way how the interconnection is made, leaving the
frequencies n, which determine the mechanical char- maximum increase only for the head-tail interconnec-
acteristics of the block. The block is invariably a lin- tion type. (The last reports on relaxation time and vis-
ear system, and the times of block response are given cosity measurements [13] state that while the viscos-
through simple arithmetic multiplication if constant ity increases with slope of 3.4 with average molecular
local friction is considered. mass M w3.4 , the increase in maximal relaxation time is
less dramatic (M w3 ) [66].
f 0,Frn
n = (14)
2n mn
The alpha and beta slow processes the
and for the constant frictions and masses of each os- mechanical shelters for shear elasticity components
cillator inside of block f 0Fr,n / mn = const' = 1 , we get:
The retardation time 1 is thus the starting point in a
n sequence through which the longer retardation time n
1 (15)
n2 can be created. The longest retardation time N should
grow with decreasing temperature and reflect the ap-
For the sequence of resonance modes in the proximation for the level of interconnections in linear
1 2 1 elements of block size.
sense of Eq. (12) the matrix = 1 2 1

The term N stands for the time necessary for dis-

...... etc...... placement of a block as a whole. The shortest retarda-
of Eq. (12) plays the crucial role in the separations of tion time will reflect the fastest response of bottom of
retardation times and in the separation of resonant fre- potential valley by differentially small displacements
quencies of n. From Eq. (11) for the case of the nth of r0,n. The fastest displacement will involve only
k f Fr the acoustical units in shortest vicinity of the particle
element we get 2n = e = 0,n . considered. Because the acoustical units usually con-
mn mn n tain few atoms and are approximately of monomer
Taking the ratio between the elasticity and size [12], the shortest relaxation time is expected to
friction for each oscillator in block as the same represent connection of about 20100 atoms and to be
k e / f 0Fr,n = const = 1
1 , Eq. (12) gives very simplified in the vicinity of the so-called fast relaxation process
spectra distribution n= 1n2. It can be seen that the [49]. One bead in necklace pictured in Fig. 6 will be
higher retardation times are formed in blocks spaces involved in the friction force of many molecules
through growing sequence of interconnected ele- forming the first level of coordinated motion in the
ments and not through variation of local friction or closest neighborhood of the particle displaced. (This
local elasticity. The friction force f 0,nFr or the external will provide the relaxation time in the vicinity of the
forces F'ext, Fext do not influence the separation of so-called fast relaxation process, which is reported
individual material time responses either. [49, 67] to be about 3.01013 s). It can be seen in Figs

J. Therm. Anal. Cal., 80, 2005 279

HLAVEK et al.

7 and 8 that for the so-called (line A) relaxation the sample elasticity is mainly connected to the parti-
process the interconnection level reaches infinity as cles associated with process as the elastically
soon as the temperature decrease starts to approach interconnected network covering the space of a whole
the Tg temperature and the infinite elastic network is sample is formed. On the other hand, the particles
developed. The so-called slow process (lines connected with slow process are apparently anchored
B, C, D) is different because the parameter N for in space differently, being ultimately bound with very
these processes cannot reach the infinite interconnec- high viscosity coefficient rather than with purely elas-
tion level as the temperature decreases. Searching the tic response to external shear deformation. It can be
N at the crossover temperature for 600 elements we stated that the slow processes are connected with the
get N(T =Tcr)= 16002=3.0101360021.08107 s. loosely anchored or floating parts of a network
Rossler [49] experimentally confirmed such charac- structure.
teristic time level at the crossover temperature. Below Because the area between the Tcr and Tg is con-
the Tcr temperature the term , which means, that nected with sharp increase in shear modulus, it can be
said that the slow processes are shear-module-insensi-
tive. Therefore, the slow processes appear to be local-
ized in sample in different, mechanically sheltered ar-
eas, which certainly transport the shear component of
stress in a way different from what the process does.
Because the number of interconnected elements form-
ing the slow process does not reach infinity (which is
the precondition for the formation of elastic solid
phase with the infinite relaxation time), the elastic
solid phase formation seems to be determined by the
The views presented in this paper are more
closely approaching the description of process pre-
sented in early theories [7176] of Eisenberg [77, 78],
Heijboer [7476] and Bueche [43] which has been
taken as well proofed by experiment and which con-
Fig. 7 The logarithms of characteristic relaxation times determined tradict the recent theories which are involving all par-
by Rossler [49] for variety of organic and inorganic materi- ticles of the system participating in the slow mo-
als. The graphs are testing the sequence shown in
tions [69]. The theories of plasticizers [79] suggest
Eqs (14, 15). Rosslers relaxation times are taken for
also the structural heterogeneity and give example of
the limiting variable interconnection number N. In our
model, 1=31013 s and 600=107 s according to Rossler
materials for which the islands of mobility can be
[49] and according to Eq. (15). At T =Tcr we take the N hardly excluded. (On the other hand, in the case of
value equal to 600 glycerol, the slow process is not present). The other
explanation for the slow processes is connecting them
with changes in coordinates displacements not di-
rectly connected with external shear deformation (the
perpendicular rotation of two different parts of poly-
mer chains around two perpendicular axes has been
described already by Bueche [43] in his discussion of
substance of and slow processes.)
These studies are the subject of our various inter-
disciplinary projects and are the continuation of our
previous and recent interrelated discussions pub-
lished in different sources such as papers [16, 8283]
or books [84, 85].


Fig. 8 The numbers of linear interconnections N for variable The estimate on the size of domain at crossover tem-
relative temperatures obtained from Eq. (15). The lim- perature has been provided based upon the rheologi-
ited interconnection level for slow processes (curves B, cal data of viscosity at zero gradients.
C, D), below Tg are clearly visible

280 J. Therm. Anal. Cal., 80, 2005


The crossover temperature has been estimated the equally distributed masses mn per length
from the relaxation module data. The crossover tem- l=(N +1)=(N +1)x where is a distance between subse-
perature is connected to first appearance of internal en- quently connected masses mn1=mn= =mn+1=mx etc.
Except for the variability of shape of irregular block, the
ergy related shear module.
elastic rod seems to be the geometrically nearest object resem-
The irregular structure of amorphous phase has bling the block structure. If choosing the following substitu-
been explained through the existence of non-deter- tion in Eqs (1012) for the ke=Te/; dx=mn where is a bar
ministic oscillators, which exist in liquid phase. The (spring) density, then Eqs (1214) take the form of wave equa-
disappearance of these particles under the Tg and the tion for F ext=0.
non-predictability of end points of their positions in
mn d r0 ,n =k e (r0 ,n+1 + r0 ,n1 2r0 ,n )
phase space coordinates gives rise to the existence of (16)
irregular structure of amorphous phase. dt 2
The non-linear oscillators have also an abnormal d 2 r0 ,n =Te r0 ,n+1 2r0 ,n + r0 ,n1 =
coefficient of thermal expansion and the amorphous
dt 2 m x
liquid is thus structurally different from the amor-
phous solids, which casts some light into the explana- =Te r0 ,n+1 r0 ,n r0,n r0 ,n1 = (17)
m x x
tions regarding the Kauzman paradox [4]. The
Kauzman paradox is based on extrapolation and does =Te r0 r0
not take into account the fact, that the liquid is a me- m x n +1 x n
chanically heterogeneous substance in which the het-
d 2 r0 =Te d 2 r0 (18)
erogeneity is partially removed as the temperature
dt 2 dx 2
goes down below Tg, and the highly non-linear oscil-
lators disappear. The liquid phase is structurally dif- The connection of difference equation to the wave equa-
ferent from the solid in mechanical sense, viz. by the tion opens further possibilities connected to elasticity and
appearance and disappearance of highly mobile, dif- fracture theories.
fusive, non-linear oscillators.
The and slow processes seem to have a common
origin in the fast process. The process differs from Appendix B: Solution of difference equation
the slow process in numbers of interconnected ele- in form of Eqs (12)
ments, which increase for the process to infinity as
the sample solidifies. The slow process does not pos- The solution of Eqs (12) can be searched in form:r0 e j( t kx )
sess this property and apparently is not fully contrib- where k is a wave vector.
uting to the shear elasticity level. Being localized in- 1

side of blocks, the slow process is insensitive to the k =

inter-blocks bridging which gives rise to shear module Te
and as such, the slow process need not to be present in
In discrete form for the nth particle we get:
some materials, at the point of formation of glassy
r0,n =An e j( t kx ) =An e j( t kan ) since x=na.
state. The characteristic vibration frequencies of linear
Substituting these expressions into Eq. (14) provides:
domains localized outside the non-linear areas provide
the source for discrete characteristic time spectra. The 2n mn =Te (ejka + ejka 2) =
experimental data supporting the double character of a (20)
vibration and double values for amplitude sizes have jk a jk a

been recently brought forward by Medick et al. [46] =Te (e 2

e ) = 4Te sin 2 ka
2 2

a a 2
and by previous work of Rossler et al. [6870].
Substituting from boundary conditions, i.e. that the os-
cillators out of sample with index n=0 and index n=N+1 can
be associated with zero amplitudes An=0, we get the conve-
Appendix A: Differential displacements of nient substitution:
particles in block ruled by wave equation
ka = n (21)
The solution of Eqs (10, 11) in the form of Eqs (1315) has 2 2(N + 1)
many interesting features, and some of them should be men-
tioned. The first interesting feature seems to be the fact that 2n =2Te 1 cos n =4Te sin 2 n (22)
Eqs (1012), relating the individual bottom well displacements mn a N + 1 ma 2(N + 1)
r0,n for many (n) particles do not have to depend on the actual
bottom wells positions of r0,n of interconnected particles. As a
special solution, Eqs (1012) can represent the wave equation
for vibration modes of elastic rod, or the motion of spring with

J. Therm. Anal. Cal., 80, 2005 281

HLAVEK et al.

Acknowledgements 27 J. Frenkel, Kinetic Theory of Liquids, Dover Publ. Inc.,

New Jersey 1955, p.12.
The authors are grateful for the financial support from the re- 28 V. Velikov, S. Borick and C. A. Angel, J. Phys. Chem.,
search project No. 0021627501 (Ministry of Education, Czech B106 (2002) 1069.
Republic) and to Dr. J. Barto from Slovak Academy of Sci- 29 W. Xu, E. I. Cooper and C. A. Angel, J. Phys. Chem.,
ences, Bratislava. B107 (2003) 6170.
30 H. J. Parkhurst Jr. and J. Jonas, J. Chem. Phys.,
63 (1975) 2699.
References 31 R. Zwanzig, J. Chem. Phys., 79 (1983) 4507.
32 U. Buchenau and M. Zorn, Europhys. Lett.,
18 (1993) 523.
1 H. E. Stanley, Introduction to Phase Transitions and Criti-
33 A. Eisenstein and N. S. Gingrich, Phys. Rev.,
cal Phenomena, Oxford University Press, New York 1971.
62 (1942) 261.
2 J. O. Hirschfelder, C. F. Curtiss and R. B. Bird, Molecular
34 J. Barto, Colloid Polym. Sci., 274 (1996) 14.
Theory of Gases and Liquids, J. Wiley, New York 1954,
35 J. Barto, P. Banduch, O. aua, K. Kritiakov,
pp. 122, 291.
J. Kritiak, T. Kanaya and W. Jenniger, Macromolecules,
3 W. Gotze and L. Sjogren, Rep. Prog. Phys., 55 (1992) 241.
30 (1997) 6906.
4 W. Kauzmann, Chem. Rev., 43 (1948) 219.
36 Y. C. Jean, Microchem. J., 42 (1990) 72.
5 M. Cohen and D. Turnbull, J. Chem. Phys.,
37 Y. C. Jean, Macromolecules, 29 (1996) 5756.
31 (1959) 1164.
38 Y. C. Jean, Q. Deng and T. T. Nguyen, Macromolecules,
6 D. Turnbull and M. H. Cohen, J. Chem. Phys.,
28 (1995) 8840.
34 (1961) 120.
39 G. P. Johari, J. Chem. Phys., 116 (2002) 2043.
7 D. Turnbull, in: Thomas J. Hughel, Liquids: Structure,
40 G. P. Johari, J. Chem. Phys., 112 (2000) 10957.
Properties and Solid Interaction, Proceedings of Sympo-
41 R. E. Powell, W. E. Roseveare and H. Eyring, Ind. Eng.
sium, Elsevier, Amsterdam 1965, pp. 624.
Chem., 33 (1941) 430.
8 D. Turnbull, Contemp. Phys., 10 (1969) 473.
42 S. Glasstone, K. J. Laidler and H. Eyring, The Theory of
9 G. S. Grest and M. H. Cohen, Phys. Rev., 21 (1980) 4113.
Rate Processes, McGraw-Hill Book, New YorkLondon
10 M. H. Cohen and G. S. Grest, Phys. Rev.,
B20 (1979) 1077.
43 F. Bueche, Physical Properties of Polymers, Interscience,
11 G. S. Grest and M. H. Cohen, in: I. Prigogine and S. T.
New YorkLondon 1962, p.106.
Rice, Advances in Chem. Phys., Vol. XLVII, J. Wiley,
44 M. Vogel and E. Rossler, J. Phys. Chem.,
New York 1981, pp. 455525.
B104 (2000) 4285.
12 A. Heuer and H. W. Spiess, J. Non-Cryst Solids,
45 G. P. Johari, Phil. Mag., 81 (2001) 1935.
176 (1994) 294.
46 P. Medick, M.Vogel and E. Rossler, J. Magn. Reson.,
13 E. Donth, The Glass Transition, Relaxation Dynamics in
159 (2002) 126.
Liquids and Disordered Materials, Springer Verlag,
47 C. Tschirwitz, S. Benkhof, T. Blochowitz and E. Rossler,
Berlin, Heidelberg 2001.
J. Chem. Phys., 117 (2002) 6281.
14 A. Einstein, Investigation on the theory of the Brownian
48 S. Adichtchev, T. Blochowitz, C. Gainaru, V. N. Novikov,
Movement, Ed. R. Furth, Dover Publications, Inc., New
E. A. Rossler and C. Tschirwitz, J. Phys. Condens. Matter,
York, N.Y. 1956.
15 (2003) 835.
15 B. Hlavek, V. Keslek and J. Souek, J. Chem. Phys.,
49 E. Rossler, Phys. Rev., 69 (1992) 1595.
107 (1997) 4658.
50 E. Donth, J. Non-Cryst. Solids, 53 (1982) 325.
16 B. Hlavek, J. estk and J. Mare, J. Therm. Anal. Cal.,
51 T. Dollase, R. Graf, A. Heuer and H. W. Spiess,
67 (2002) 239.
Macromolecules, 34 (2001) 298.
17 M. I. Rabinowitz, A. B. Ezersky and P. D. Weidman, The
52 J. Barto, O. aua, J. Kritak, T. Blochowity and
Dynamics of Patterns, World Sci. Publ., Singapore 2000,
E. Rossler, J. Phys. Condens. Matter, 13 (2001) 11473.
pp. 239279.
53 M. Beiner, H. Huth and K. Schroter, J. Non-Cryst. Solids,
18 H. J. Paine, The Physics of Vibrations and Waves,
279 (2001) 126.
J. Wiley, New York 1997.
54 E. Hempel, M. Beiner, T. Renner and E. Donth, Acta
19 U. Ueda, New Approaches to Non-linear Dynamics.
Polymer., 47 (1996) 525.
S.I.A.M., Philadelphia 1980.
55 U. Mohanty, G. Diezeman and I. Oppenheim, J. Phys.
20 M. Tabor, Chaos and Integrability in Non-linear Dynam-
Condens. Matter, 12 (2000) 6431.
ics, John Wiley, New York 1989.
56 S. A. Reinsberg, X. H. Qiu, M. Wilhelm, H. W. Spiess and
21 I. Prigogine and I. Stengers, Order out of Chaos, Bantam
M. D. Ediger, J. Chem. Phys., 114 (2001) 7299.
Books, New York 1984.
57 V. R. Allen and T. G. Fox, J. Chem. Phys., 41 (1964) 337.
22 D. Kondempundi and I. Prigogine, Modern Thermody-
58 T. G. Fox and V. R. Allen, J. Chem. Phys., 41 (1964) 344.
namics, J. Wiley, New York 1998, pp. 409452.
59 J. Schulz, Polymer Material Science, Prentice-Hall,
23 O. E. Rssler, Phys. Lett., 57A (1976) 397.
Englewood Cliffs, New Jersey 1974, p. 337.
24 E. N. Lorenz, J. Atm., Sci. 20 (1963) 130.
60 J. D. Ferry, Viscoelastic Properties of Polymers, Wiley,
25 A. Tockstein, Chemick listy, 81 (1987) 569.
New YorkLondon 1961, p. 225.
26 P. G. DeBenedeti, Metastable Liquids, Princeton Univ.
61 F. Bueche, J. Chem. Phys., 20 (1956) 599.
Press, Princeton, New Jersey 1996.

282 J. Therm. Anal. Cal., 80, 2005


62 A. V. Tobolsky, Properties and Structure of Polymers, 77 B. Cayrol, A. Eisenberg, J. F. Harrod and P. Rocaniere,
J. Wiley, New York 1960 p. 75. Macromolecules, 5 (1972) 676.
63 J. J. Aklonis, W. J. MacKnight and M. Shen, Introduction 78 A. Eisenberg, The Glassy State in: J. E. Mark,
to Polymer Viscoelasticity, Wiley-Interscience, A. Eisenberg, W. W. Graessley, L. Mandelkern, E. T.
New York London 1972, p. 42. Samulski, J. L. Koenig and G. D. Wignall, Physical Prop-
64 B. Hlavek, E. ernokov, L. Prokpek and M. Veea, erties of Polymers, Amer. Chem. Soc., Washington D.C.
Thermochim. Acta, 280/281 (1996) 417. 1993, pp. 6197.
65 D. R. Lide, Handbook of Chemistry and Physics, 79 H. E. Bair, Thermal Analyses of Additives in Polymers,
82nd Ed., 10, CRC Press LLC, Washington D. C. 2002, in: E. A. Turi, Thermal Characteristic of Polymeric Mate-
pp. 160174. rials, Vol. 2, J. Wiley, Academic Press, New York 1997,
66 M. Doi and S. F. Edwards, Theory of Polymer Dynamics, pp. 22632412.
Clarendon Press, Oxford 1986. 80 T. Achibat, A. Boukenter, E. Duval, G. Lorentz and
67 M. Vogel and E. Rossler, J. Phys. Chem., B104 (2000) S. Etienne, J. Chem. Phys., 95 (1991) 2949.
4285. 81 E. Duval, A. Boukenter and T. Achibat, J. Phys. Condens.
68 J. Wiederich, N. V. Surovtsev and E. Rossler, J. Chem. Matter, 2 (1990) 10227.
Phys., 113 (2000) 1143. 82 J. estk: Trends and boundary aspects of non-tradition-
69 M. Vogel and E. Rossler, J. Chem. Phys., ally applied science of heat presented at the seminary
114 (2001) 5802. Present and future of JTAC (Budapest, 2003), abstracted
70 M. Vogel and E. Rossler, J. Chem. Phys., in: J. Therm. Anal. Cal., 75 (2004) 1006.
115 (2001) 10883. 83 J. J. Mare and J. estk, J. Therm. Anal. Cal., (2005) sub-
71 E. Jeckel and H. U. Herwig, Kolloid Z., 148 (1956) 57. mitted
72 R. F. Boyer, J. Macromol. Sci. Phys., B7 (1973) 487. 84 J. estk, book: Heat, Thermal Analysis and Society,
73 R. F. Boyer, J. Polym. Sci., Polym. Symp., 50 (1975) 189. Nucleus, Hradec Krlove 2004.
74 J. Heijboer, Kolloid Z., 148 (1956) 36. 85 Z. Chvoj, J. estk and A. Tska, book Kinetic Phase Di-
75 J. Heijboer, Kolloid Z., 171 (1960) 7. agrams: non-equilibrium phase transitions, Elsevier, Am-
76 J. Heijboer, Ph.D. Thesis, University of Leiden, 1972. sterdam 1991.

J. Therm. Anal. Cal., 80, 2005 283

J Therm Anal Calorim (2012) 110:9971004
DOI 10.1007/s10973-011-1926-6

Forty years of the Hruby glass-forming coefficient via DTA

when comparing other criteria in relation to the glass stability
and vitrification ability
Ana Kozmidis-Petrovic Jaroslav Sestak

Received: 13 July 2011 / Accepted: 16 September 2011 / Published online: 7 October 2011
Akademiai Kiado, Budapest, Hungary 2011

Abstract The revision of the meaning of the famous Introduction

Hruby glass-forming coefficient (as well as of other anal-
ogous coefficients by, e.g., Weinberg and LuLiu) reveals In the 1960s, the former Institute of Solid-State Physics of
some generalized correlations between glass-forming the Academy of Sciences was one of the leading institu-
ability (GFA) and glass stability. The relative change of the tions in solid state physics within the so called Eastern
Hruby parameter is supreme in almost all cases. The Hruby socialist block then dominated by the former Soviet Union.
parameter is more sensitive in relation to the change of They knew how to prepare super-pure single-crystalline as
both the super-cooled region and the reduced glass-transi- well as glassy Si and Ge, were capable of their various
tion temperature. The only exception is the restricted sen- doping and even tackled the secrecy of transistor but did
sitivity respecting the reciprocal reduced glass-transition not realized its technical usefulness yet. The research
temperature in some cases of the bulk metallic glasses. The turned then to the field of amorphous (mostly chalcogen-
correlation of the Hruby coefficient with GFA is agreeable ide) semiconductors recognizing that the band gap does
for oxide glasses and thus can be commonly employed as a exists in spite of the absence of atomic long-range order
reliable and precise glass-forming criterion. Associated and is not empty containing an appreciable amount of
problems are the experimental determination of relevant localized states but having its edges no more sharp.
temperatures, most pertinent that for glass transition which Transport band edges coincide with so called mobility
is dependent to preparative condition of glass formation. edges where the mobility of carriers dramatically changes
while the optical band gap is determined by the extrapo-
Keywords Glass transition  Reduced quantities  lated absorption curve. It was discovered by Jan Tauc
Hruby criterion  Sensitivity  Glass-forming ability  (19222010) [1, 2] who later immigrated to the US (Brown
DTA/DSC determination  Temperatures University, Rhode Island). His coworker Arnost Hruby
(1919) was technologist who synthesized and analyzed
thousands of chalcogenide compounds and glasses and
investigated their thermal behaviors [3, 4]. He proposed the
A. Kozmidis-Petrovic
evaluation of glass-forming tendency by means of DTA
University of Novi Sad, Faculty of Technical Sciences,
Trg D. Obradovica 6, 21000 Novi Sad, Serbia (lately known as the renowned Hruby glass-forming cri-
e-mail: terion), which was published in a less known Czech journal
of physics [4] even though having received up to now
J. Sestak (&)
several hundreds of citations becoming the best cited paper
New Technology, Research Centre in the Westbohemian Region,
West Bohemian University, Universitn 8, Pilsen 30114, in the journal history. Besides giving a tribute to the
Czech Republic masterwork of Arnost Hruby, the purpose of this article is
e-mail: to revise the meaning and usefulness of Hruby criterion in
comparison with the other analogous coefficients and
J. Sestak
Division of Solid-State Physics, Institute of Physics, uncover some generalized correlations between glass-
Cukrovarnicka 10, 16200 Praha, Czech Republic forming ability (GFA) and glass stability (GS).

998 A. Kozmidis-Petrovic, J. Sestak

Glass-forming ability and GS Significantly before the development of any generalized

nucleation theory for condensed systems, Tammann [19,
Glass is a specific matter reflecting its original liquid makeup 20] called attention to a tendency revealing that the higher
and can be prepared by a suitable freeze-in of various, even the melt viscosity at the melting temperature (Tm), the
non-stoichiometric, combinations of elements/compounds. lower its crystallizability. Qualitatively, this tendency can
Such a disordered matter still presents stringent conceptual be explained by an increased inhibition of motion or
difficulties [59]. The often confused concepts of amorphous molecular rearrangement of the basic units of any melt with
and glassy states have recently been re-examined and pre- increasing viscosity. Mentioned by Kauzman [21] but
vious attempts to trace the distinction revised [5, 6]. Cur- particularly stated by Turnbull [22], who early indicated
rently, Queiroz et al. [7] considered thermodynamic aspects that when Trg is larger than 2/3, the homogenous crystal
of glasses focusing to their behavior in glass-transition nucleation will be essentially suppressed due to the slug-
region (GTR). Sunol et al. [8] examined phenomenological gishness of the crystallization kinetics. Therefore, Trg
and atomic methods by which glass transition can be became the earliest criterion to evaluate GFA of a liquid.
described. This new generic and phenomenological However, experimental observations indicated that Trg may
approach describes the kinetics and thermodynamics of fail to truthfully predict GFA in various cases, often doc-
vitrification as a real non-equilibrium process. Volume and umented by conventional relationships between Rc and
enthalpy relaxation in the GTR was analyzed by Svoboda Trg for different sort of glasses, where some good glass
et al. [9]. The finding that the activation energy of viscous formers show a low Trg and vise versa. Referring to a large
flow in the glass-transition range was identical with the set of available experimental data obtained for nucleation
effective activation energy for relaxation process. This of several silicate glasses, Zanotto [14] concluded that
subject received abundant number of citation responses [10]. glasses having Trg higher than *0.580.60 display only
Parameters predicting the GFA and consequent GS of its surface (mostly heterogeneous) crystallization, while
constrained state (of freeze-in glass) are of a substantial glasses showing volume (homogeneous) nucleation have
meaning to all those interested in various applications to Trg \ 0.580.60. The reduced glass-transition temperature
which glassy materials lend themselves. A timetempera- was examined in more details by Sakka and Mackenzie
turetransformation diagram for material annealing can [23], Uhlmann [24], and for metallic glasses also by Davis
provide all necessary data, but such data are rarely available [25]. Determination of theoretical values of reduced tem-
and moreover are often predicted on the assumption of peratures were approached by Angell [26] based on
homogeneous nucleation, which is a rather unlikely event in extension and extrapolation by means of application of the
practice. When a glassy matter turn out to be experimentally VogelFulcherTamman (VFT) viscosity equation. Sestak
accessible upon a suitably fulfilled melt quenching (critical [27] dealt with a greater augmentation of reduced quanti-
cooling rate, Rc) from its melting point (Tm) through the ties, such as a difference between liquid and crystalline
GTR [11, 12] (defined by the mean glass-transition tem- heat capacities DCpr, and tried to estimate approximate
perature, Tg), certain data became accessible for such a changes of entropy DSr, enthalpy DHr and chemical
material identification. A liquid with good GFA exhibits a potential Dlr assuming the values Tor and Trg. (Tor is the
low value of Rc for the glass formation which, however, has reduced Kauzman temperature).
remained a long-standing question as to why one liquid In order to unify notations, which use to vary in the cited
exhibits better GFA than another and how we can portray it. papers, the following text will employ Tx to denote the
According to the standard nucleation theory, a liquid onset crystallization temperature and Tc to indicate the
with a high viscosity between Tg and Tm typically exhibits peak crystallization temperature, which is convenient
a high GFA with a low Rc. Since the viscosity of oxide today. Let us also remind that GFA is noticeably related to
glasses at Tg is nearly constant (%1012 Pa s) [13], it was the ease of the reverse process of devitrification related to
postulated that a high value of the reduced glass-transition the difference between temperatures of re-crystallization
temperature, Trg (=Tg/Tm), would result in a high viscosity and glass transition.
in the undercooled liquid state, and, consequently, lead to a However, other parameters can be included, which can
low Rc. Zanotto [14, 15] showed that glass tendency to be comfortably monitored by DTA/DSC upon certain
homogeneous and heterogeneous nucleation can be dis- precautions (glass heating, etc.). It can provide a choice of
tinguished on bases of Trg. Cabrall et al. [16] pointed out data about the non-equilibrium glass crystallization tem-
that the maximum homogeneous nucleation rates, Imax, of perature laying customary below that for an analogous
silicate glasses strongly diminish with Trg and that Rc for melt. Then the interval between Tm and Tc (and Tx) is
metallic glass formation also drops with rising Trg. Corre- inversely proportional to GFA and the interval between the
lation between maximum growth rates and Trg for silicate onset of crystallization Tx and Tg is directly entitled to
glasses was shown by Fokin et al. [17, 18]. display GFA. A range of examples revealed that this

Forty years of the Hruby glass-forming coefficient 999

difference vary habitually with composition changes and calculated these data and found that the stability of glasses
tends to reach its maximum value in a composition range having parallel viscosity curves g(T) could be qualitatively
which appears to provide best/optimum glasses. evaluated according to the two above expressions, but they
Even a more sensitive interrelation to the glass forma- were not quantitatively reliable and none of them appears
tion peculiarities can be found on a basis of the widespread to be superior. Moreover, the paper also discloses that for
Hruby parameter [4], developed mostly for chalcogenide glasses for which g(T) significantly differs from each other
systems [28], which are typically available only upon in the region of Tg the stability criteria can become
physical preparation of a given type of glass.This criterion delusive.
[3, 29, 30] has an almost matching implication as the dif-
ference (Tx - Tg ) alone varying, however, more rapidly
when crystallization peak is shifted and taking also into Determinability of glass-transition temperature
account ease of melting. This factor may not be of a par-
ticular significance as the values of Tg and Tm are usually Glasses as an important domain of a generalized disordered
correlated [2931]. A wider applicability of KH was dis- matter still present experimentally stringent conceptual and
cussed by Thornburg [31]. thermodynamic difficulties. It was recently reexamined by
In the recent years, various GS parameters related to the Wunderlich [44] who revealed that in a continuous random
three characteristic temperatures have been investigated. web of a macromolecular system (which can properly
Table 1 contains some popular GS parameters. model a glass); the network can be a single amorphous
Aware that the potential correlation between GS and phase, mesophase, molecularly coupled multiphases of
GFA is of a great importance, this issue became the subject different degrees of order and metastability or even a
of various theoretical studies as well as of assorted modulated crystal-like structure (liquid or plastic crystals)
experimentation. [45].
For example, Weinberg [43] made use of the standard Different state of a similar network (such as silicate
theoretical expressions for crystal nucleation and growth quenched melts and/or solution prepared polysialates) may
rates considering homogeneous nucleation and screw dis- show a range of glass-transitions processes observable even
location growth in stoichiometric glasses. In his paper [43], at a simultaneous manner. Perceptible values needed for a
the trends in GFA and GS were compared with systematic correct recognition of the phase transitions lies in the fact
changes in the melting entropy, DSm, and the viscosity that materials must be pliable for being prepared into its
parameters. The resulting conclusion was that GFA and final shape and then experimentally measurable (often by
GS, defined by (Tc - Tg)/Tm, were poorly related [43]. In a means of DTA/DSC), which involves lot of experimental
subsequent work, Weinberg [33] derived the time neces- variability and associated uncertainty. Therefore, the
sary for crystallization of a minimum detectable fraction determination of glass-forming criteria requires sustaining
based again on a classical homogeneous nucleation and certain precautions and data standardizations. There are
screw dislocation growth. Worth mentioning are time cri- numerous publications discussing the theme, e.g., [4649].
teria used to assess GFA and test the reliability for two However, the best analysis was shown by studies of
particular GS parameters, given by the expressions: (Tx - Illekova [5054] who separated various influences in
Tg)/Tm and (Tx - Tg)(Tc - Tx)/Tm. Weinberg [43] alienated areas emphasizing that in metallic glasses the
measured Tg is habitually lower than the true thermody-
Table 1 Some GS parameters based on the characteristic tempera- namic Tg while for other oxide or chalcogenide glasses, the
tures specified in the text true Tg depends also on the inherent fragility of sample
GS parameter Envisaged by glass. Moreover, Tg of metallic glasses is often affected by
aging of glass even at room temperature. There is also a
KH = (Tx - Tg)/(Tm - Tx) Hruby [4] obvious difference for Tg obtained for either during cooling
KSP = (Tc - Tx)(Tx - Tg)/Tg Saad and Poulain [32] and heating, moreover for some cases is sketched by mere
KW = (Tc - Tg)/Tm Weinberg [33] estimation.
KLL = Tc/(Tg ? Tm) Lu and Liu [34, 35] There is a common correlation of both kinetic processes
cm = (2Tx - Tg)/Tm Du [36] of structural relaxation and crystallization. Nevertheless,
/ = Trg(DTxg/Tg)a Fan [37] the crystallization kinetics is of two major types (JMAYK
x = Tg/Tc - 2Tg/(Tg ? Tm) Long [38] and NGG) and may contradictory affect the peak position,
b = TxTg/(Tm - Tx)2 Yuan [39] namely any thermal annealing may shift the exothermic
Some unusual criteria you can find in [4042] involving extra kinetic crystallization peak opposite, i.e., to higher temperatures
factors [40, 41] or a contrary interpretation in terms of glass state for the classical nucleation-growth (JMAYK) and contrary
safeguarding [43] to, lower temperatures for NGG. Therefore, any thermal

1000 A. Kozmidis-Petrovic, J. Sestak

treatment must be avoided during the determination of This is the reason why we shall analyze the individual
glass-forming coefficients unless the study is aimed to the criteria in more detail respecting the three characteristic
resolution of nucleation/growth rates [5560]. temperatures and mentioning assessment in the paper [39],
Staying power of individual criteria evidently depends which stresses that GS parameters (estimating GFA) should
on a correct determination of characteristic temperatures be enough sensitive. The larger the values of the KH, KW,
and further more detailed analysis falls beyond the subject and KLL the higher is stability of a glass against devitrifi-
of this communication. cation [4, 62]. Also, when comparing one glass to another,
it is essential to know how large the relative changes of the
given parameters are and how they can be compared with
Expressing the Hruby parameter using temperature the relative change of other GS. In other words, it is nec-
relations normalized to Tg essary to distinguish which of the GS parameters shows the
fastest change.
After Weinbergs paper [43], Cabral et al. [61] brought into Our starting point is the fact that all the criteria, KH, KW,
play experimental values and found a correlation between and KLL, include the same characteristic temperatures. For
the Hruby parameter of GS and GFA. However, these this reason, we can express them in a somewhat different
results [61] were somehow contradictory in relation to the way based on the intact ratios of the temperatures, namely
theoretical calculations of Weinberg [43]. Extending these m = Tm/Tg and r = Tc/Tg. In doing so, we assume that in
calculations, Avramov et al. [62] decided to test out these defining both KH and KLL it is possible to replace onset
two approaches upon examining different assumptions, crystallization temperature Tx with maximum crystalliza-
which were supported by experimental data. The paper [62] tion peak temperature Tc as was shown in the paper by
demonstrated that GFA and GS follow the same trend and Nascimento [64]. After simple mathematical transforma-
are directly related to each other. tions [70], one obtains relations KH = (r - 1)/(m - r) and
Several other considerable studies have appeared, which KW = (r - 1)/m and KLL = r/(m ? 1).
inveterated correlation between the GS parameters and By using the substitutions of r and m, the GS parameters
critical cooling rate Rc, or between the GS parameters and can be expressed indirectly by means of both the reduced
maximum section thickness, i.e., the diameter Dmax, by glass-transition temperature Trg and the super-cooled
which the GFA are also estimated [34, 3639, 6365]. In region DTxg. It is because the parameter m factually rep-
the papers [34, 62, 64, 66], a rather good correlation is resents the reciprocal value of Trg, and the parameter
shown between GFA and GS parameters, which are based r correlates to DTxg, as was shown in the work of Lu and
on the three characteristic temperatures such as the Hruby Liu [34], Mondal [71], or in a recent work by Zhang et al.
(KH), Weinberg (KW), and LuLiu (KLL) criteria. Nasci- [72]. Namely, in order to enable the comparison for dif-
mento et al. [64] proved that for the oxide glasses a very ferent glasses, the value of the super-cooled region is
good correlation between the Hruby parameter and GFA divided by Tg [34], which gives (Tx - Tg)/Tg = r - 1.
exists. Also, very good correlation between KLL and GFA Mondal et al. [71] used the same normalization and pro-
has been found for the oxide glasses from [64] and for posed that Tx/Tg (in our cases r) can also be considered as a
those from [67]. However, the parameters Rc and Dmax are measure of the thermal stability of glass. In [72], Zhang
hard to measure a in a sufficiently precise way [35, 38, 39, et al. introduced the factor of crystallization resistance Tg/
6669] so that of essential importance become the corre- (2Tx - Tg), which can be auxiliary expressed as 1/(2r -
lations between GFA and GS expressed via characteristic 1). From a mathematical point of view, the introduction of
temperatures [6169]. These temperatures are usually the ratios r and m allows the reduction of the number of
determined by DTA/DSC [3, 30, 37]. A satisfactory degree independent variables by which the parameters KH, KW,
of correlation between the determined GS parameter and Rc and KLL are defined. Instead of the three characteristic
(or Dmax) would allow one to use the given GS parameter temperatures, we use their two ratios (r and m) acting as
to assess GFA. independent variables.
In spite of the criteria discussed above approachable
through the three characteristic temperatures, the entire
reduced glass-transition temperature shows a weaker cor- Relative changes of GS parameters
relation with GFA. The work of Nascimento [64] proves
that Trg has a reduced correlation with GFA for oxide In order to derive expressions for the relative changes of
glasses than KH, KW, and/or KLL. Furthermore Lu and Liu these parameters, as shown in our previous paper [70], we
[34, 35] showed that Trg had the weaker correlation with first employ logarithms of the expressions for KH, KW, and
GFA than their criterion KLL for glasses analyzed in [34]. KLL and then make differentiation

Forty years of the Hruby glass-forming coefficient 1001

dKH dr dm dr dKW dr dm Sensitivity of the glass-stability parameters

KH r1 mr mr KW r1 m to the changes of temperature ratios
dKLL dr dm 1

KLL r m1 The next objective was to find out how these GS parame-
ters are sensitive to the changes of r and m, and whether
It is necessary to point out that we assume the following
this is reflected on the correlation of the given parameter
relations hold for m [ 1, r C 1, m [ r.
with the GFA. Besides the KH and KLL, the newly defined
By comparing the right hand side expressions in Eq. 1
parameters / [37], and x [38], showed also a rather good
[70], it holds that if the condition dr [ dm is fulfilled, i.e.,
correlation with GFA for a variety of glasses [3739].
if d(Tc/Tg) [ d(Tm/Tg) is satisfied, then the following
Thus, it is necessary to include them in our analysis.
relation keeps continuously valid dKH/KH [ dKW/KW [
Simple transformations of the expressions by which they
have been introduced allow us to express the parameters /
Such a relation of the relative changes of the GS
and x, using ratios r and m, instead of using the charac-
parameters will also hold when a less stringent condition is
teristic temperatures [70]. Thus, / = (r - 1)a/m and
satisfied, i.e., when dr/r [ dm/m. If the condition dr/
x = 1/r - 2/(m ? 1). In order to determine the change
r [ dm/m is not fulfilled, then the following relation of the
of the investigated GS in relation to r and m, we take the
relative changes of the GS parameters will maintain: dKW/
derivatives of expressions for KH, KLL, /, and x with
KW [ dKH/KH [ dKLL/KLL. As it can be seen, the relative
respect to both values of r and m. On this way, for KH is
change of the LuLiu parameter will always have the
obtained dKH/dr = (m - 1)/(m - r)2 and dKH/dm = -
smallest value, and will never become greater than the
(r - 1)/(m - r)2. The analogous expressions for changing
relative change of the Hruby parameter. The above theo-
KLL, /, and x in relation to r and m can be found in [76].
retical derivations were successfully tested on two series of
The individual partial sensitivity of the GS parameter to
oxide glasses and one series of chalcogenide glasses [70]
r and m is, however, not related to the R2 coefficient for the
showing a good agreement with the theoretical results.
correlation of the GS parameter and GFA [76]. The balance
The next step was to establish a relation between the
of sensitivity of the GS parameter to r and m, (i.e., the ratio
maximal value of relative changes of the Hruby, Weinberg,
dGS/dr:dGS/dm) implies the magnitude of the R2 coeffi-
and LuLiu parameters. In order to obtain the maximal
cient. At the same time, the degree of correlation of values
value of relative changes notice that the minus signs in
r and m with the GFA is significant.
expressions (1) should be replaced by the plus ones,
Results of the above theoretical derivation were tested on
because for maximal change we should assume that all
series of oxide glasses. The characteristic temperatures Tg,
changes have the same direction. Thus, we obtain
Tc, and Tm were taken from the work of Cabral et al. [65] for
dKH dr dm dr dKW the following seven oxide glasses: Li2O2SiO2 (LS2);

KH max r  1 m  r m  r KW max Na2O2CaO3SiO2 (NC2S3); 2Na2OCaO3SiO2 (N2CS3);
  BaO2SiO2 (BS2); LiO22SiO2 with 0.2% mol OH (LS2OH);
dr dm dKLL dr dm
 2BaOTiO22SiO2 (B2TS2), and 0.44Na2O0.56SiO2
r1 m KLL max r m1
After comparing right hand side of expressions in Eq. 2, Figure 1 shows the ratios dGS/dr:dGS/dm that we cal-
the following relation between the maximal values of culated for these oxide glasses.
relative changes keeps up [73] As it can be seen from Fig. 1, the values of the ratios
      dKH/dr:dKH/dm, dKLL/dr:dKLL/dm and dx/dr:dx/dm are
[ [ 3 always greater than one. This means that KH, KLL, and x
KH max KW max KLL max
are more sensitive to the change in relation to r than to the
The relation between maximal relative changes of GS change in relation to m. For KH, this is easy to predict by
parameters (cf. Eq. 3) has been tested on several series of comparing the appropriate expressions above. We take the
oxide glasses from [64] and chalcogenide glasses from characteristic temperatures and values for critical cooling
[74, 75]. rate from [65] and calculate the values of GS parameters as
The glass having the largest value of r, that is the largest Tc/ well as the values of R2 factors in correlation with critical
Tg ratio, was selected as reference one, with respect to which cooling rate for mentioned oxide glasses. The higher the R2
we calculated the differences. From these results, it follows value, the better the correlation between GS parameter with
that for the considered glasses the relation between the max- critical cooling rate. With the analyzed oxide glasses,
imal relative changes of GS is in accordance with Eq. 3. It is r correlates well (R2 = 0.900), whereas m does not corre-
worth accentuating that the maximal relative change of the late with GFA. Since only r (and not m) correlates satis-
Hruby parameter KH subsist the greatest upshot. factorily, it can be supposed that for KH, KLL, and x the R2

1002 A. Kozmidis-Petrovic, J. Sestak

9 0.8

6 d /dr:d /dm

5 d /dr:d /dm 0.5

1 0.2

0 0.1








0.1 0.2 0.3 0.4 0.5

Glasses KLL

Fig. 1 Ratios dGS/dr:dGS/dm (GS = KH, KLL, /, x) for oxide Fig. 2 Correlation between the Hruby parameter KH and parameter
glasses using abbreviation for 0.44Na2O0.56SiO2 (44NS); KLL for BMG bulk metallic glasses, adjusted from [38]
Li2O2SiO2 (LS2); LiO22SiO2 with 0.2% mol OH (LS2OH);
BaO2SiO2 (BS2); Na2O2CaO3SiO2 (NC2S3); 2Na2OCaO3SiO2
(N2CS3); and 2BaOTiO22SiO2 (B2TS2) significantly different from those of the linear connected G-
criteria for tested glasses. This does not mean that KH do
coefficient of correlation with GFA will be larger, because not show good correlation with GFA, in cases for example,
the value of r has a greater influence than m. But for /, the for oxide glasses.
influence of m, which does not correlate with GFA, is
significant. Hence, one can expect that R2 for / is smaller Summary
than for KLL, KH, and x. This was confirmed by our cal-
culation of R2 obtaining the following values: KH = 0.829, All the results discussed above show that the initially
KLL = 0.759, x = 0.784, and / = 0.649. These R2 factors proposed Hruby criterion [4] is one of the best parameters
are smaller from the ones obtained for oxide glasses which at all. As can be seen from the theoretical derivations, the
we investigated in [76], but the tendency of changes of relative change of the Hruby parameter KH is the greatest in
correlation of GS parameters in relation to critical cooling almost all cases, and the maximal value of the change of
rate is similar. The difference between oxide glasses in [76] the Hruby parameter occur the greatest too. Also, the
and in this analysis is that KH has the best correlation with correlation of the Hruby parameter with GFA is very
Rc. Also, in this analysis (as a difference from [76]), values apposite considering oxide glasses.
of dGS/dr:dGS/dm are always the biggest for KH if we
compare these relations for KH, KLL, and x. Acknowledgements This study has been carried out with the partial
In this article, we will look (only from the mathematical support of the Serbian Ministry of Science and Technological
point of view), what occurs when the r \ 1 (Tg [ Tx in the Development, within the projects Nos. III 45021 and 172059 and the
grant support in the field of geopolymers No. FR-TI 1/335 provided
ribbon form of metallic glasses). From above expressions, by the Ministry of Industry and Business of the Czech Republic.
it easily follows that in this case dKH/di [ dKH/dm too.
When comparing the values of dGS/dr and dGS/dm for
parameters KLL, x, and /, for which expressions are given
in [76], it follows that dKLL/dr [ dKLL/dm, dx/dr [ dx/ 1. Tauc J, Grigorovici R, Vancu A. Optical properties and electronic
dm and sometimes d//dr \ d//dm in the case that r \ 1 as structure of amorphous germanium. Phys Stat Sol. 1966;15:62737.
well as when r [ 1. 2. Tauc J. Optical properties of amorphous and liquid semicon-
ductors. In: J Tauc, editor. Amorphous and liquid semiconduc-
By simple transformation the Hruby parameter KH can
tors. London: Plenum Press. 1974.
be expressed through KLL, thereby obtaining that 3. Sestak J. Science of heat and thermophysical studies: a general
r1KLL approach to thermal analysis. Amsterdam: Elsevier; 2006.
KH rr1K LL
. As was shown in our paper [77], between
4. Hruby A. Evaluation of glass-forming tendency by means of
KLL and some others GS parameters, there exists a linear DTA. Czech J Phys B. 1972;22:118793.
dependence with a very high R2 factor. It is evident from 5. Varshneya A, Mauro J. Comment on Misconceived ASTM
above expression that KH are not linearly related to KLL. As Definition of Glass by A. C. Wright. Glass Technol Eur J Glass
Sci Technol A. 2010;51:2830.
an example, Fig. 2 shows that KH is not linearly related to 6. Gutzow I, Pascova R, Schmelzer JWP. Glass transition behavior:
KLL, for BMGs from [38]. The value of R2 factor for the a generic phenomenological approach. Int J Appl Int J Appl Glass
correlation of KH with GFA for tested BMGs (0.630) is Sci. 2010;1(3):22136.

Forty years of the Hruby glass-forming coefficient 1003

7. Queiroz C, Sestak J. Aspects of the non-crystalline state. Phys 32. Saad M, Poulain M. Glass forming ability criterion. Mater Sci
Chem Glasses Eur J Glass Sci Technol B. 2010;51:16572. Forum. 1987;1920:118.
8. Sunol J, Bonastre J. Crystallization kinetics of metallic glasses. 33. Weinberg MC. An assessment of glass stability criteria. Phys
Transformation diagrams. J Therm Anal Calorim. 2010;102: Chem Glass. 1994;35:11923.
44750. 34. Lu ZP, Liu CT. A new glass-forming ability criterion for bulk
9. Svoboda R, Malek J, Pustkova P, Cicmanec P. Volume and metallic glasses. Acta Mater. 2002;50:350112.
enthalpy relaxation in the glass transition region of amorphous 35. Lu ZP, Liu CT. Glass formation criteria for various glass-forming
selenium. J Non-Cryst Solids. 2009;355:26472. systems. Phys Rev Lett. 2003;91:55045.
10. Sestak J. Citation records and some forgotten anniversaries in 36. Du XH, Huang JC, Liu CT, Lu ZP. New criterion of glass
thermal analysis. J Thermal Anal Calorim. in print 2011. doi: forming ability for bulk metallic glasses. J Appl Phys.
10.1007/s10973-011-1625-7. 2007;101:61868.
11. Weinberg MC, Uhlmann DR, Zanotto ED. Nose method of 37. Fan GJ, Choo H, Liaw PK. A new criterion for the glass-forming
calculating critical cooling rates for glass-formation. J Am Ceram ability of liquids. J Non-Cryst Solids. 2007;353:1027.
Soc. 1989;351:20548. 38. Long Z, Xie G, Wei H, Su X, Peng J, Zhang P, Inoue A. On the
12. Ray CS, Reis ST, Bow RK, Holand W, Rheinberger V. A new new criterion to assess the glass-forming ability of metallic
DTA method for measuring critical cooling rate for glass for- alloys. Mater Sci Eng A. 2009;509:239.
mation. J Non-Cryst Solids. 2005;351:13508. 39. Yuan ZZ, Bao SL, Lu Y, Zhang DP, Yao L. A new criterion for
13. Yue Y-Z. Characteristic temperatures of enthalpy relaxation in evaluating the glass-forming ability of bulk glass forming alloys.
glass. J Non-Cryst Solids. 2008;354:11128. J Alloys Comp. 2008;459:25160.
14. Zanotto ED. Isothermal and adiabatic nucleation in glass. J Non- 40. Duan RG, Liang KM, Gu SR. A new criterion for the stability of
Cryst Solids. 1987;89:10361. glasses. J Eur Ceram Soc. 1998;18:113144.
15. Zanotto ED. Experimental test of the classical nucleation theory 41. Iqbal T, Shahriari MR, Siegel GG. A new glass forming ability
for glasses. J Non-Cryst Solids. 1985;74:37394. criterion for multicomponent halide glasses. J Mater Sci Forum.
16. Cabral1 AA, Fokin VM, Zanotto ED. On the determination of 1991;6768:22532.
nucleation rates in glasses by nonisothermal methods. J Am 42. Zamecnk J, Faltus M, Sestak J. Safekeeping parameter for safe
Ceram Soc. 2010; 93:243840. storage of biological material at ultralow temperatures based on
17. Fokin VM, Zanotto ED, Schmelzer JWP. Homogeneous nucle- thermal characteristics. In: Cernosek Z, Cernoskova E, editors.
ation versus glass transition temperature of silicate glasses. International Czech and Slovak conference on calorimetry. Uni-
J Non-Cryst Solids. 2003;321:5263. versita Pardubice; 2011. pp. 147150.
18. Fokin VM, Nascimento MLF, Zanotto ED. Correlation between 43. Weinberg MC. Glass-forming ability and glass stability in simple
maximum crystal growth rate and glass transition temperature of systems. J Non-Cryst Solids. 1994;167:818.
silicate glasses. J Non-Cryst Solids. 2005;351:4789. 44. Wunderlich B. Glass transition as a key to identifying solid
19. Tammann G. Uber die wirkung von Silicium MetatitanAuhedrat. phases. J Appl Polym Sci. 2007;105:4959.
Z Elektrochemie. 1904;10:532. 45. Suga H. Propects of material science: from crystalline to amor-
20. Tammann G. Die abhangigkeit der viskositat von der temperatur phous solids. J Thermal Anal Calorim. 2000;60:5764.
bei unterkuhlten flussigkeiten. Zeit Anorg Allgem Chemie. 46. Lasocka M. The effect of scanning rate on glass transition tem-
1926;156:2457. perature of splat-cooled Te85Ge15. Mater Sci Eng. 1976;23:
21. Kauzmann W. The nature of the glassy state and the behavior of 1735.
liquids at low temperatures. Chem Rev. 1948;43:21926. 47. Avramov I, Gutzow I. Heating rate and glass transition temper-
22. Turnbull D. Under what conditions can a glass be formed. Con- ature. J Non-Cryst Solids. 1988;104:14850.
temp Phys. 1969;10:4738. 48. Cernosek Z, Holubova J, Cernoskova E. Enthalpy relaxation and
23. Sakka S, Mackenzie JD. Relation between apparent glass tran- the glass transition. J Optoelectron Adv Mater. 2002;4:489503.
sition temperature and liquidus temperature for inorganic glasses. 49. Hutchinson JM. Determination of the glass transition tempera-
J Non-Cryst Solids. 1971;6:14562. ture: methods correlation and structural heterogeneity. J Therm
24. Uhlmann DR. A kinetic treatment of glass formation. J Non-Cryst Anal Calorim. 2009;98:57989.
Solids. 1972;7:33748. 50. Illekova E. A generalized model of structural relaxation in
25. Davies HA. The formation of metallic glasses. Phys Chem metallic and chalcogenide glasses. Key Eng Mater. 1993;8183:
Glasses. 1976;17:15963. 5418.
26. Angell CA. Oxide glasses in light of the Ideal glass concept. 51. Illekova E. Review of structural relaxation models with the
I. General aspects: ideal and non- ideal transitions. J Am Ceram mutual correlation of their activation enthalpies. Int J Rapid
Soc. 1968;51:11724. Solidif. 1994;8:195224.
27. Sestak J. Some thermodynamic aspects of the glassy state. 52. Illekova E, Clavaguera-Mora MT, Baro MD, Surinach S. Dif-
Thermochim Acta. 1985;95:45971. ferential scanning calorimetry study of structural relaxation of
28. Hruby A. Glass-forming tendency in the GeSx system. Czech J Ge-doped Se85Te15 glasses. Mater Sci Eng. 1994;B22:18190.
Phys B. 1973;53:126372. 53. Illekova E, Cunat Ch. An extended review of structural relaxation
29. Sestak J. Use of phenomenological enthalpy versus temperature model with the mutual correlation of their parameters. J Non-
diagram (and its derivative-DTA) for a better understanding of Cryst Solids. 1994;172174:597600.
transition phenomena in glasses. Thermochim Acta. 1996;280(/ 54. Illekova E, Cunat Ch, Kuhnast FA, Aharoune A, Fiorani JM.
281):17590. Complex behaviour of specific heat during structural relaxation of
30. Sestak J. Glasses: phenomenology of non-isothermal glass for- Fe73Co12B15 glass. Thermochim Acta. 1992;203:44555.
mation and crystallization. In: Chvoj Z, Sestak J, Trska A, edi- 55. Sestak J. Applicability of DTA and kinetic data reliability of non-
tors. Kinetic phase diagrams: non-equilibrium phase transitions. isothermal crystallization of glasses. Thermochim Acta. 1986;98:
Amsterdam: Elsevier; 1991. p. 164. 33958.
31. Thornburg DD. Evaluation of glass formation tendency from rate 56. Chen LC, Spaepen F. Analysis of calorimetric measurements of
dependent thermograms. Mater Res Bull. 1974;9:14815. grain growth. J Appl Phys. 1991;69:67988.

1004 A. Kozmidis-Petrovic, J. Sestak

57. Malek J, Pustkova P, Shanelova J. Kinetic phenomena in non- 68. Sheng WB. Correlations between critical section thickness and
crystalline materials studied by thermal analysis. J Thermal Anal glass-forming ability criteria of Ti-based bulk amorphous alloys.
Calorim. 2003;72:28997. J Non-Cryst Solids. 2005;351:30816.
58. Ranasinghe KS, Wei PF, Kelton KF, Ray CS, Day DE. Verifi- 69. Senkov ON, Scott JM. Glass forming ability and thermal stability
cation of an analytical method for measuring crystal nucleation of ternary CaMgZn bulk metallic glasses. J Non-Cryst Solids.
rates in glasses from DTA data. J Non-Cryst Solids. 2004;337: 2005;351:308794.
2617. 70. Kozmidis-Petrovic AF. Theoretical analysis of relative changes
59. Ranasinghe KS, Ray CS, Day DE. A generalized method for of the Hruby, Weinberg, and LuLiu glass stability parameters
determining the crystal nucleation and growth rates in glasses by with application on some oxide and chalcogenide glasses. Ther-
DTA. J Mat Sci. 2002;37:54755. mochim Acta. 2010;499:5460.
60. Sestak J, Kozmidis-Petrovic A, Zivkovic Z. Crystallization 71. Mondal K, Murty BS. On the parameters to assess the glass
kinetics accountability and the correspondingly developed glass- forming ability of liquids. J Non-Cryst Solids. 2005;351:
forming criteria. J Min Metall B Metall. 2011;47:22939. 136671.
61. Cabral AA, Fredericci C Jr, Zanotto ED. A test of the Hruby 72. Zhang P, Wei H, Wei X, Long Z, Su ZX. Evaluation of glass-
parameter to estimate glass-forming ability. J Non-Cryst Solids. forming ability for bulk metallic glasses based on characteristic
1997;219:1826. temperatures. J Non-Cryst Solids. 2009;355:21839.
62. Avramov I, Zanotto ED, Prado MO. Glass-forming ability versus 73. Kozmidis-Petrovic AF, Lukic SR, Siljegovic M. Theoretical
stability of silicate glasses. II Theoretical demonstration. J Non- analyses of the relative change of glass stability parameters with
Cryst Solids. 2003;320:920. the application on some oxide and chalcogenide glasses from the
63. Inoue A, Zhang T, Masumoto T. Glass-forming ability of alloys. system Bix(As2S3)100-x. Vancouver, Canada: PACRIM8-S23-
J Non-Cryst Solids. 1993;156158:47380. 082-2009; 2009.
64. Nascimento MLF, Souza LA, Ferreira EB, Zanotto ED. Can glass 74. Sharma A, Barman PB. Calorimetric and optical study of amor-
stability infer glass forming ability? J Non-Cryst Solids. 2005;351: phous Se85-xTe15Bix glassy alloy. Thin Solid Films. 2009;517:
3296310. 30203.
65. Cabral AA, Cardoso AAD, Zanotto ED. Glass-forming ability 75. Abdel-Rahim MA, El-Korashy A, Hafiz MM, Mahmoud AY.
versus stability of silicate glasses. I. Experimental test. J Non- Kinetic study of non-isothermal crystallization of BixSe100-x
Cryst Solids. 2003;320:18. chalcogenide glasses. Phys B. 2008;403:295662.
66. Nascimento MLF, Dantas NO. Assessment of glass-forming 76. Kozmidis-Petrovic AF. Sensitivity of the Hruby, LuLiu, Fan,
ability and the effect of La2O3 on crystallization mechanism of Yuan, and Long glass stability parameters to the change of the
bariumleadzinc phosphate glasses. Mater Lett. 2007;61:9126. ratios of characteristic temperatures Tx/Tg and Tm/Tg. Thermo-
67. Lu ZP, Liu CT. A new approach to understanding and measuring chim Acta. 2010;510:13744.
glass formation in bulk amorphous materials. Intermetallics. 77. Kozmidis-Petrovic AF. Which glass stability criterion is the best?
2004;12:103543. Thermochim Acta. 2011;523:11623.

Journal of

J. Min. Metall. Sect. B-Metall. 47 (2) B (2011) 229 - 239 Mining and

CrystAllIzAtIon kInetICs ACCountAbIlIty And the

CorrespondIngly developed glAss-formIng CrIterIA
A personAl reColleCtIon At the forty yeArs

J. estk*,#, A. kozmidis-petrovi** and . ivkovi***

*New Technology - Research Centre in the Westbohemian Region, West Bohemian

University, Universitn 8, CZ-30114 Pilsen, Czech Republic
**University of Novi Sad, Faculty of Technical Sciences,
Trg D. Obradovia 6, 21000 Novi Sad, Serbia
***University of Belgrade, Technical Faculty, VJ 12, 19210 Bor, Serbia

(Received 12 June 2011; accepted 13 July 2011)


The meaning of experimentally determined characteristic temperatures is analysed in terms of

Hruby glass forming criterion. Despite various analysed modifications it reveals that the sensitivity
of original criterion is unsurpassable. The applicability of Sestak-Berggren empirical equation for
the description of crystallisation kinetics is comparable with the classical modelling based on
Johnson-Mehl-Avrami equation. Their employment is certainly dependent to the purpose of
evaluation showing their substitution ability and limits.

1. Introduction leading to the formation of quenched-

constrained metastable phases (metallurgy);
There are numerous important personalities in extreme providing the unusual freeze-in
who paved the scientific road toward a better state of glasses. Associated processes
understanding of new phase formation, which remained in the center of research until today,
is the core of most crystallization processes. It but rich figures have got lost in the extended
became especially important during the past of extensive scientific research. Some
application of fast temperature changes often important scientist are shown in the raw of
# Corresponding author:

230 J. estk / JMM 47 (2) B (2011) 229 - 239

photos below, just to keep in mind illustrious emphasizing their great protagonists, see Fig.
history of exploration of solid-state reactions 1., read from the left.

Fig 1. Gustav H.J. Tammann (1861-1938) called earliest attention to a tendency, which revealed observation
that the higher the melt viscosity at the melting temperature, the lower is its crystallizability [1]. He also invented
the term thermal analysis[2]. Frederik F.H. Zachariasen (1906-1979) considered the principles how the
bonding requirements are met and the nearest neighbor coordination maintained without imposing an exact long
range order [3]. Walter Kauzman (1916-2009) analyzed the behavior of liquids at lower temperatures, pointing
out that the entropy of undercooled liquid decreases rapidly on cooling towards the kinetic glass transition
temperature [4] and extrapolates to the entropy of the crystal at so called Kauzman temperature. Jakov I. Frenkel
(1894-1952), became famous for his fundamental book [5] factually introducing the concept of disorder. David
Turnbull (1915-2007) brilliant material scientists who in 1946 joined the famous General Electric laboratory
performing research into nucleation of structural transformations [6], introducing reduced temperatures and
demonstrating that such complex glass-forming processes could be quantitatively better understood [7]. Below
raw, Robert F. Mehl (1898-1976) was famous with the American Institute of Mining and Metallurgical concerned
with areas of solid-state reactions, diffusion, precipitation, plastic deformation, preferred orientations, and
oxidation. He separated the role of nucleation from that of growth of new phases in solid-state transformations
and developed generalized theories applicable to re-crystallization describing the volume fraction of a solid
transformed in terms of the formation rate and spatial distribution of nuclei and the subsequent growth of the
nuclei [8]. Arnot Hrub (1919-) was an accomplished technologist who synthesized and analyzed thousands of
chalcogenide compounds and glasses identifying and describing their properties and proposing so called Hruby
glass forming criterion [9]. Donald R. Uhlmann (1936-) former director of the Arizona Research Laboratory,
fellow and awarder of the American Ceramic Society, the coauthor of both the recognized nucleation-growth
theory [10] and the famous Kingerys and Kreidls fundamental book on ceramics [11]. Michael C. Weinberg
(1942-2003), architect of various alternative theories on nucleation and glass-formation, [12,13]. Edgard D.
Zanotto (1954-) another representative of emerging generation of new theoretical kineticists responsible for the
advanced nucleation and crystallization theories mostly applied to silicate glasses [14,15], the chair of the ICG
(TC7) committee on glass crystallization and the editor of J Non-cryst. Sol.
J. estk / JMM 47 (2) B (2011) 229 - 239 231

A special attention should be paid to the its opposite hindrance, i.e., contradictory
Institute of Physics of the former obstruction of glass-forming ability and
Czechoslovak Academy of Sciences, which associated glass stability. The classical
in the turn of 1970s was one of the leading nucleation-growth description became thus
institutions in solid state physics and important and is analyzed in terms of the
chemistry [16-19] within the so called estk-Berggren empirical equation [18] all
Eastern socialist block then dominated by the hints instigated at the occasion of the forty
former Soviet Union. Its employee Arnot years anniversary of the earliest papers
Hrub [9] was a gifted technologist who publication [9,18].
investigating thermal behaviors of
chalcogenides mainly by means of DTA. His hruby glass-forming coefficient and
famous glass-forming criterion published in related criteria
a less known Czech Journal of Physics [9]
nevertheless received abundant citations Disordered matter still presents stringent
responses [19]. His coworker J. estk conceptual difficulties being prepared under
contributed kinetic study of crystallization extreme conditions of either the self-cooling
processes by various means of thermal of compositionally adjusted mixtures [22]
analysis [16,17] showing the way how to (silicates, alloys even in metallurgy) or by
formulate a generalized model of logistic attuned quenching [23] (usually applied to
(autocatalytic) equation [18] effectively the ribbon preparation of metallic glasses).
analogous to nucleation-growth kinetics. All Coupled and repeatedly confusing concepts
these topics got a prospect to be extensively of amorphous and glassy states were re-
cited in the literature, for example, Hruby [9] examined recently by Queiroz and estk
gained as many as 372 and Sestak Berggren [24-26] considering thermodynamic aspects
equation [18,19] 566 responses becoming of glasses in the glass transition region.
thus the best cited papers in the respective Associated parameters endeavoring the
journal [19], which are comparable with the prediction the glass forming ability (GFA)
Jander equation on diffusion [20] with and/or the consequent glass stability (GS) of
551citations. It is still below Mehl paper [8] the constrained states of freeze-in glasses are
on general kinetics of phase changes with of substantial terminology meaning [22,26].
2172 or Kissinger paper [21] on DTA It is often related to the time-temperature-
determined kinetics with 4461 but analogous transformation (T-T-T) diagram [22]
to similar Uhlmann paper [10] on common for the description of material
crystallization kinetic 417 or Zanotto paper annealing creditable for the subsequent
[14] on generalized nucleation theory with occurrence of nucleation-growth kinetics.
190 citations. The soaring output of the However, T-T-T is rarely available and
kinetic papers published during 1970s is thus moreover needs to be predicted on the
worth of attention revising their meaning and assumption of homogeneous nucleation,
uncovering some generalized correlations which is an unlikely event in practice. When
between the easiness of crystallization and a glassy matter does happen to be
232 J. estk / JMM 47 (2) B (2011) 229 - 239

experimentally accessible upon a suitable formation was assumed by Saad and Poulain
melt quenching (critical cooling rate, Rc) [34] which took into their consideration the
down from its melting point (Tm) through the width of a DTA/DSC peak, i.e., the difference
glass transition region (GTR), defined by the between the onset of crystallization Tc and its
mean glass transition temperature (Tg) maximal value Tx [34, 35]. Such a new
certain data became accessible for such a criterion (Tx Tc)(Tc - Tg)/ Tg comprises the
material identification [27-33]. unit of Kelvin, but is capable to become
Consequently a range of following criteria dimensionless if weighted by squared Tg.
was proposed. The other popular glass-forming parameters
was proposed by Weinberg [36] KW = (Tc
Tg)/Tm and in the recent years, various
other parameters have been investigated,
such as the parameter envisaged by Lu and
Liu [37, 38] as KLL = Tc/(Tg + Tm). There are
additional parameters proposed in the papers
[39-42] mostly related to the previously
mentioned characteristic temperatures dealt
in details with allied papers [31-57].
In the paper by Weinberg [43] the trends
in GFA and GS were compared with
systematic changes in the melting entropy,
DSm and the viscosity h concluding that GFA
and GS (defined by (Tc-Tg)/Tm ) are poorly
related. Weinberg [36] also derived the time
necessary to crystallize a minimum
detectable fraction based again on a classical
Fig. 2. Original proposal of reading the
characteristic temperatures in the Hruby glass- homogeneous nucleation and screw
forming criterion dislocation growth in stoichiometric glasses.
Time criteria were used to assess GFA and
A rather sensitive interrelation to the glass test the reliability of two particular GS
formation (GF) particularity can be found on parameters (given by the previously
basis of the widespread Hrub parameter mentioned expressions [34, 43]) observing
[9,31]. It is typically available upon a that the stability of glasses with the parallel
physical preparation of a given type of glass viscosity curves h(T) could be qualitatively
and can be calculated as KH = (Tc - Tg)/(Tm - weighed up. Zanotto et al [44] used
Tc) on basis of experimentally detectable experimental values and found a correlation
quantizes (usually figured out by DTA [9, 31, between the Hrub parameter [9] of GS and
32], such as Tc , which is the onset of GFA being, however, contradictory in
crystallization temperature). A more relation to the theoretical calculations of
receptive parameter toward the glass Weinberg. Avramov et al. [45] extended the
J. estk / JMM 47 (2) B (2011) 229 - 239 233

calculations of Weinberg [43] but tested a that all the coefficients KH, KW and KLL,
different assumption, which is supported by include the identical three characteristic
experimental data demonstrating that GFA temperatures possible to express them in a
and GS follow the same trend and are jointly somewhat distinctive way by mere ratios of
interrelated. Recently, several other respective temperatures, such as m = Tm/Tg
important studies have appeared [46-55], and r = Tc/Tg. where that the following
which demonstrate correlation between the relations always sustain, i.e. : m>1, r1, m>r.
GS parameters and critical cooling rate Rc, or Thus the both KH and KLL is possible to
between the GS parameters and the incept the crystallization temperature Tx with
maximum sample thickness, i.e. the section the maximum crystallization peak
diameter Dmax, by which the glass forming temperature Tc as was shown by Zanotto et al
ability are estimated [37, 39-42, 46, 47]. It [46]. After simple mathematical
again reveals a rather good correlation transformations we obtain KH = (r-1)/(m-r),
between GFA and GS, based on the above KW = (r-1)/m and KLL = r/(m+1). Using the
three characteristics given by the Hrub substituted r and m, the GS parameters can
(KH), Weinberg (KW) and Lu-Liu (KLL). be expressed indirectly via the reduced glass
Zanotto et al [46] proved that for the oxide transition temperature Trg and super cooled
glasses a very good correlation between the region Txg. It is because the parameter m
Hrub parameter and GFA can exists and represents the reciprocal value of Trg, and the
another good correlation between KLL and parameter r can be correlated to Txg, as was
GFA was also established [46, 49]. Another shown in the work of Lu and Liu [37] and
exploit was found even for metallic glasses Mondal [53] or in a recent work by Zhang et
[50, 51]. al [54]. In order to enable the comparison for
The work of Zanotto et al [46] different glasses [55], the value of the super
proves that Trg has the weaker correlation cooled region is divided by Tg [37], which
with GFA for oxide glasses than KH, KW and gives (Tc-Tg)/Tg = r-1. Mondal et al [53]
KLL,. Similarly, Lu and Liu showed that Trg used the same normalization and proposed
has the weaker correlation with GFA than that Tx/Tg (= r) can also be considered as a
their parameter KLL for glasses analyzed in measure of the thermal stability of glass.
[37]. Larger values of the coefficients KH, Zhang et al [54] introduced the factor of
KW and KLL imply higher stability of the crystallization resistance Tg/(2Tx-Tg)
glass in respect to devitrification [45] Also, (=1/(2r-1)).
when comparing one glass to another, it is In order to derive expressions for the
essential to know how large the relative coefficients relative changes, Kozmidis-
change of the given parameter is and how it Petrovic [31, 52, 57] first took logarithms of
can be compared with the relative change of KH, KW and KLL and then differentiated the
other GS parameters. In other words, it is obtained values achieving inequality
necessary to know which of the GS dKH/ KH > dKW/ KW > dKLL/ KLL, which is
parameters shows the fastest change. valid when d(Tc/Tg) > d(Tm/Tg) i.e. if the
Kozmidis-Petrovic [52] pointed the fact condition dr>dm is satisfied. Resulting
234 J. estk / JMM 47 (2) B (2011) 229 - 239

relative changes of the GS parameters will Prout-Tompkin equation (turning thus into its
also hold under a less stringent condition dr/r limiting case [61]). The increasing value of
> dm/m. The relative change of KLL will m often indicates a more important role of
always have the smallest value being never the precipitated phase on the overall kinetics
greater than the relative change of the consequently showing that this two
original Hrub parameter [9, 31, 57], which parameter model preserve its physical
satisfactorily the best sensitivity KH. The meaning for m 1 [61, 62]. It also appears
ensuing relation between the maximal values that a higher value of the second kinetic
of relative changes of the individual K- exponent (n > 1) signifies increasing reaction
parameters approves thus the priority of complexity. Besides the S-B equation
Hruby coefficient [4,57] by (dKH/ KH)max > subsists as a generalized operate based on the
(dKW/ KW)max > (dKLL/ KLL)max. logistic function [59, 61], a{a(1 a)}, which
is customarily exploited to depict the case of
estk-berggren (sb) kinetic equation population growth where a is the
for a generalized crystallization proportional factor of so called attractivity.
It consists of two essential but counteracting
At the anniversary [19] of estk- parts, the first one responsible for mortality @
Berggren (S-B) equation [18], it reads as am (i.e., reactant disappearance and the
da/dt @ am(1a)n, (where a is the product formation) and the other for fertility
dimensionless degree of crystallization, t is @ (1a)n (i.e., a kind of products hindrance
time, da/dt is the crystallization rate and the generally accepted as an counter-parting
omitted proportional Arrhenius constant autocatalytic effect [61]). The non-integral
exp(-E/RT) incorporates temperature, T). S- exponents, m and n, play thus the similar role
B was proposed in a particular association as a broadly assumed non-integral
with a well established kinetic model dimensions common in the natural world of
introduced jointly by Johnson-Mehl-Avrami fractals [62].
(JMA) [8, 18, 58-62] as da/dt @ {-ln (1- The S-B equation can be found credulous
a)}p(1-a). Since 1971, S-B received for serviceability as an alternative to the
abundant citation responses and even classical JMA equation [61-66] and thus
became a part in the titles of various kinetic exploitable for an assortment of model
papers [62-66]. It is worth noting that upon descriptions of both the interface-controlled
the Taylors expansion the function {-ln (1- and the diffusion-limited crystallization.
a)}p can display a certain mathematical Such a modeling starts from an original
inter-convertibility upon the expansion and comprehensive form of ln ( 1- a ) = - kTtr,
recombination to the matching am [58-61]). where the general exponent, r, can be seen as
The two appropriately paired exponents a multipart number of a robust (so called
of the S-B equation (m and n) are suitable for overall) analysis. In terms of DTA
a wide-ranging kinetic assessment, which measurements it reveals that the overall
includes the classical model of autocatalytic apparent values of activation energies Eapp
reaction (where m=1 and n=1) often called [67] can be commonly correlated to the
J. estk / JMM 47 (2) B (2011) 229 - 239 235

partial activation energies of nucleation, EN, enabling, however, certain but limited
growth, EG and/or diffusion, ED by a simple conversion to the traditional geometrical
relation Eapp = (aEN + bdEG)/(a + bd) where models [60-70]. There is a certain correlation
a and b are characteristic multiplying between the S-B exponents m-n and JMA
constants providing that the denominator (a exponents r (and p) [60, 62, 69, 70] see Fig 3.
+ b d) equals to the power exponent, r, of the Another agreeably workable alternative to
original JMA equation, and the value b the JMA crystallization is the mechanism of so
corresponds to 1 or related to the called normal-grain-growth (NGG)
movement of growth front controlled by resembling mode close to the mathematical
either chemical reaction (1) or diffusion () modeling of phase-boundary reactions [59-
[16, 61, 67]. Moreover, the coefficients d and 61], i.e., am(1a)n where m0. This case is
b are associated with the nucleation velocity abbreviated as the Atkinsons NGG model
and the growth dimension, respectively. [72], which has been effectively applied by
However, it endowed with a rather complex Illekova [73, 74]. NGG factually reflects the
pattern so far mathematically soluble but process of coarsening of the nano-crystalline
experimentally problematic in order to reach phase, when grain size falls below 10 nm,
reliable association with (desirable) which is common in some metastable
microscopic observations [58]. metallurgical alloys and constrained glasses.
In 1990 this S-B equation was completed In this instance the DSC/DTA exothermic
[68] by the conjecture of the so called crystallization peaks is converted into different
accommodation function, h(a) which is a shape possessing atypical symmetry, which is
effortless empirical function containing the the result of rationale arising from the
smallest possible number of constant and differences in JMA and NGG mechanisms
enabling certain flexibility in order to [72-74].
sufficiently match mathematically the real
course of a process under study. In such case,
the kinetic model of a heterogeneous reaction
is factually assumed to be a distorted case of a
simpler (ideal) instance of homogeneous
kinetic prototype, i.e., h(a)f(a) h(a) (1-a)n
[68]. It thus gives a credit to the recognition of
a certain defect state (imperfection, non-
ideality, heterogeneity such as the controlling
role of interfaces). The function h(a) can be
equally am and/or [-ln (1-a)]p displaying in a
way, above mentioned inter-convertibility with
the JMA model function, which brought to
kinetics an additional sphere of a more
widespread applicability approaching thus the Fig. 3. Portrayal correlation between JMA
model-free domain of kinetic evaluations (r) and SB (m-n) exponents.
236 J. estk / JMM 47 (2) B (2011) 229 - 239

Conclusions about the actual portrayal of mechanism. If

we assume, however, that the goal of most
The determination of characteristic studies is a more and less pragmatic [59-61,
temperatures is an ordinary task in thermal 71, 76, 77], i.e., the reportable
analysis [59-61, 75] (and likewise derived determination of an unspecific data on
SB-equation) possibly providing a rather reaction mechanism the SB-equation
good mathematical match toward the trial subsist satisfactory. However, if we are
modeling of crystallization, see Fig.4. It seeking a correlation to one of the
should be reminded that somehow confidential physical-geometric models we
inconsequential information is involved have to look for an alternative
representation based on otherwise
observable morphology between the true
reaction image and the theoretical portrait
symbolized by the model assumed.
However, a more in-depth seeking may
correspond with an innovative so called
model-free kinetics (e.g. imon [78]) and/or
congruent dissociative vaporization kinetics
(e.g. Lvov [79]). In the first case [78] the

Fig. 4. Illustrative cases of experimental DTA/DSC traces demonstrating possibility of twofold

kinetics for an analogous crystallization of two brands of similarly vitrified alloys [74] melt-quenched by
a single roll technique to form ribbons (10-20mm thick and 10mm wide). Right is the tangible glass of
composition of Fe75Si15B10 while left is the nano-crystallized finemetal with a comparable stoichiometry
Fe74Si13B9Cu1Nb3 [80, 81] (just with a small doping of Cu and Nb). Both flat samples were treated
correspondingly, freeze-in ribbons were cut in several pieces and pressed into a Pt-cell with the in-weight
of 15mg; heating rate was 40 K/min (applied under nitrogen inert atmosphere to avoid oxidation). The
ribbons were pre-annealed at the fixed temperature of 500oC and annealing time was progressively
prolonged, for the sake of simplicity the particular curves fall within the shaded areas.
Right: the classical JMA nucleation-growth mechanism [8, 58-62], which display the non-integral
values of S-B exponents [16-18, 70], i.e., m<1 and dimension dependent 1<n<3 (in conjunction with a
nucleation short-rage diffusion plausibly involving their radii relation r/ro = (1 a)1/n , where r and ro
are the radii at t = t and t = 0. The resulting sharp peak apexes are typically shifted along with the
increased heating to commonly follow the kinetic evaluation method by Kissinger [21], widely employed
in the literature [59-69].
Left: a more distinctive Atkinson [72-74] normal-grain-growth (NGG) of nano-particles coarsening,
which show signs of phase boundary reactions (controlled by interface propagation). Such a process is
reliant on a longer range diffusion and nano-particles initial and actual size ro and r. For m approaching
zero (am~1) the scheming exponent n holds for the reaction rate in the form of (1 a)n+1/(n ron). In
consequence the experimentally measured DTA peaks (DT vs. T) reveal different symmetry with an
atypical behavior of broadened apexes and their shifting under increased heating rate [74,80], which
may cause certain intricacy in traditional kinetic appraisals.
J. estk / JMM 47 (2) B (2011) 229 - 239 237

kinetic parameters occurring in (T - a-t) [8] W.A. Johnson, R.F. Mehl, Trans. Amer. Inst.
functions are only perceived quantities Min. Metal. Petro. Eng. 135, (1939) 416.
(similarly to the widespread portrayal of [9] A. Hruby , Czech. J. Phys. B 22 (1972) 1187
apparentness), possessing no traditional and B 53 (1973) 1263.
mechanistic interpretation while the [10] D.R. Uhlmann,. J. Non-cryst. Solids 54,
second case [79] provides the temperature (1983) 253.
coefficient of reaction rate identifiable with
[11] D.R. Uhlmann, N.J. Kreidl (edts)
the molar enthalpy (vant Hoff equation) Advances in Structural Analysis and
rather than assuming (never justified) Structure, Microstructure and Properties, both
participation of an activated intermediate by Tokyo Academic, Boston 1990.
(as in the Arrhenius model). [12] M.C. Weinberg, J. Amer. Cer. Soc. 74
(1991) 1905.
[13] M.C. Weinberg, J Min Metal 35 (1999) 197.

The results were developed within the [14] E.D. Zanotto J. Non-Cryst. Sol. 89 (1987)
CENTEM project, reg. no. 361; E.D. Zanotto and M.C. Weinberg, Phys.
Chem. Glasses 30 (1989) 186.
CZ.1.05/2.1.00/03.0088 that is co-funded
from the ERDF within the OP RDI program [15] E.D. Zanotto, P.F. James J. Non-cryst.
of the Ministry of Education, Youth and Solids 74, (1985) 373.
Sports. [16] J. estk, Thermophysical Properties of
Solids: theoretical thermal analysis Elsevier,
references: Amsterdam 1984 and Russian translation
Toretieskij trmieskij analyz Mir, Moscow
[1] G. Tammann, Wied. Ann. 16 (1897) 280-299
[17] J. estk (ed) : Reaction kinetics by
and Aggregatzustnde Translated by R. F.
thermal analysis, special issue of Thermochim.
Mehl as States of Aggregation. New York, Van
Acta, Vol. 203, Elsevier, Amsterdam 1992.
Nostrand 1925.
[18] J. estk, G. Berggren, Thermochim. Acta
[2] G. Tammannn, Anorg. Chem. 45 (1905) 289
3, (1971) 1.
and 701.
[19] J. estak., Citation records and some
[3] W.H. Zachariesen, J. Amer. Cer. Soc. 54
forgotten anniversaries in thermal analysis. J
(1932) 3841.
Thermal Anal calor, in print 2011
[4] W. Kauzmann, Chem. Reviews 43 (1948) 1 (DOI:10.1007/s10973-011-1625-3)
[5] J. Frenkel, Kinetic Theory of Liquids [20] W. Jander, Z. anorg. Chem. 163 (1927) 1.
University Press, Oxford 1946; and Dower, New
[21] H.E. Kissinger, Anal. Chem. 29 (1957)
York 1955.
[6] D. Turnbull, J.C. Fisher, J. Chem. Phys. 17
[22] J. estk Glasses: phenomenology of non-
(1949) 71 and 18 (1950) 769.
isothermal glass-formation and crystallization
[7] Turnbull D Contemporary Physics 10 (1969) in the book Kinetic Phase Diagrams: non-
473. equilibrium phase transitions (Z. Chvoj, J.
238 J. estk / JMM 47 (2) B (2011) 229 - 239

estk, A. Tska, edts.), Elsevier, Amsterdam [34] M. Saad, M. Poulain, Mater. Sci. Forum
1991, pp.169-264. 19&20 (1987) 11.
[23] A. Varschavski and J. estk. Applications [35] M. Poulain, Thermochim. Acta 280/281
of DSC for the study of transformation processes (1996) 343.
in quenched alloys ( pp. 85-111) in the book
[36] M. C. Weinberg, Phys. Chem. Glasses 35
Characterization techniques of Glasses and
(1994) 119.
Ceramics ed. J. M. Rincon and M. Romero,
Springer-Verlag, Berlin (2000) [37] Z. P. Lu, C. T. Liu, Acta Mater. 50 (2002)
[24] C. Queiroz C, J. estk J. Phys Chem
Glasses: Eur J Glass Sci Technol B. 51 (2010) [38] Z. P. Lu, C. T. Liu, Phys. Rev. Lett. 91
165. (2003) 115505.
[25] Queiroz C., estk J. Introducing non- [39] X. H. Du, J. C. Huang, C. T. Liu, Z. P. Lu, J.
crystalline, disordered, amorphous, vitreous and Appl. Phys. 101 (2007) 086106.
glassy concepts in the book Some [40] G.J. Fan, H. Choo, P. K. Liaw, J. Non-Cryst.
thermodynamic, structural and behavior aspects Solids 353 (2007) 102.
of materials accentuating non-crystalline states
(J. estk , M. Holeek. J. Mlek, edts), OPS- [41] Z. Z. Yuan, S. L. Bao, Y. Lu, D. P. Zhang, L.
ZU Plzen, 2009, pp. 8-28. Yao, J. Alloys Comp. 459 (2008) 251.

[26] estk J, Queiroz C, Mares JJ, Holecek M [42] Z. Long, G. Xie, H. Wei, X. Su, J. Peng, P.
Some aspects of vitrification, amorphisation, Zhang, A. Inoue, Mater. Sci. Eng. A 509 (2009)
disordering and the extent of nano-crystallinity 23.
in the book Glassy, amorphous and [43] M. C. Weinberg, J. Non-Cryst. Solids 167
nanocrystalline materials I: thermal physics, (1994) 81.
analysis, structure and properties estk J,
[44] A. A. Cabral, Jr. C. Fredericci, E. D.
Mare JJ, Hubk P, editors. Berlin: Springer;
Zanotto, J. Non-Cryst. Solids 219 (1997) 182.6
[45] I. Avramov, E. D Zanotto, M. O. Prado, J.
[27] R.C. Ruhl, Mat. Sci. Eng. 1, (1967) 313.
Non-Cryst. Solids 320 (2003) 9.
[28] H. Suga. Faraday Discussion 69 (1980) 221;
[46] M. L. F. Nascimento, L. A. Souza, E. B.
J Thermal Anal Calor 60 (2000) 957.
Ferreira, E. D. Zanotto, J. Non-Cryst. Solids
[29] D.D. Thornburg, Mat. Res. Bull. 9 (1974) 351(2005) 3296.
[47] A. A Cabral, A. A. D. Cardoso, E. D.
[30] S. Takayama, J. Mater. Sci. 11 (1976) 104. Zanotto, J. Non-Cryst. Solids 320 (2003) 1.
[31]A. Kozmidis-Petrovic, J. estk Forty years [48] M. L. F. Nascimento, N. O. Dantas, Mater.
of the Hrub glass-forming criterion via DTA Lett. 61 (2007) 912.
figures regarding the vitrification ability and
[49] Z. P. Lu, C. T. Liu, Intermetallics 12 (2004)
glass stability J Thermal Anal Calor, in print
[50] W. B. Sheng, J. Non-Cryst. Solids 351
[32] J. estk, Thermochim. Acta 95 (1985) 459.
(2005) 3081.
[33] J. estk, Thermochim. Acta 280/281
[51] O. N. Senkov, J. M. Scott, J. Non-Cryst.
(1996) 175.
Solids 351(2005) 3087.
J. estk / JMM 47 (2) B (2011) 229 - 239 239

[52] A. F. Kozmidis-Petrovic, Thermochim. Acta [68] J. estk, J Thermal Anal. 36 (1990) 69.
499 (2010) 54.
[69] J. Mlek, Thermochim. Acta 267 (1995)
[53]. K. Mondal, B. S. Murty B.S, J. Non-Cryst. 61.
Solids 351(2005) 1366.
[70] P. imon, Forty years of estak - Berggren
[54] P. Zhang, H. Wei, X. Wei, Z. Long, X. Su, equation Thermochim. Acta, in print 2011.
J. Non-Cryst. Solids 355 (2009) 2183. doi:org/10.1016/j.tca.2011.03.030.
[55] O.N. Senkov, J.M. Scott. J. Non-Cryst. [71] N. Koga and J. estk, J. Am. Ceram. Soc.
Solids 351(2005) 3087. 83 (2005) 1753.
[56] A.F. Kozmidis-Petrovic. Thermochim Acta [72] H.V. Atkinson. Acta Metallurgica 36 (1988)
510 (2010) 137. 469.
[57] A.F. Kozmidis-Petrovic, J. estk, Forty [73] E. Illekov, J. Non-cryst. Sol. 287 (2001)
years of the Turnbull reduced glass-transition 167.
temperature and Hruby glass-forming coefficient
[74] E. Illekova, J. estk Crystallization of
in the book Glassy, amorphous and
metallic micro- and nano-crystalline glasses in
nanocrystalline materials II: reaction kinetics,
the book Some thermodynamic, structural and
thermodynamics and thermal analysis estk J,
behavior aspects of materials accentuating non-
Simon P. editors. Berlin: Springer 2012
crystalline states (J. estk , M. Holeek. J.
[58] M. Avrami. J. Chem. Phys. 7 (1939) 1103. Mlek, edts), OPS-ZU Plzen, 2009.
[59] J. estk Heat, thermal analysis and [75] J. estk, J. Min. Metall. Sect. B-Metall. 35
society Nucleus, Hradec kralove 2004 (1999) 367.
[60] J. estk (ed), Vitrification, Transformation [76] . ivkovi, N. Milosavijevic, J. estk,
and Crystallization of Glasses, special issue of Thermochim. Acta 157 (1990) 215.
Thermochim. Acta, Vol. 280/281, Elsevier,
[77] . ivkovi, J. estk. J, . Therm. Anal. 53
Amsterdam 1996
(1998) 263.
[61] J. estk Science of Heat and Thermo-
[78] P. imon, J. Thermal Anal. Calor. 76 (2004)
physical Studies: a generalized approach to
123; 79 (2005) 703. and 81 (2005)
thermal analysis Elsevier, Amsterdam 2005
[79] B.V. Lvov, Thermal decomposition of
[62] J .Mlek, Thermochim.Acta 355 (2000) 239.
solids and melts - new thermochemical approach
[63] J.M. Criado, J. Mlek, F.J. Gotor to the mechanism, kinetics and methodology.
Thermoch. Acta 158 (1990) 205. Springer, Berlin 2007.
[64] A.K. Burnham , J. Thermal Anal. Calor. 60 [80] E. Illekov, in: Properties and Applications
(2000) 895. of Nanocrystalline Alloys from Amorphous
precursors, NATO Science Series II:
[65] G. Munteanu, E. Segal J. Thermal Anal.
Mathematics, Physics and Chemistry, No. 184.
Calor. 101 (2006) 89.
(B. Idzikowski, P. vec, M. Miglierini, edts),
[66] M.A. A. Rahim, A,Y.A Latief, Kluwer, Dordrecht 2005, p. 421.
A.E.Korashy,M.A. Sabet Mater. Trans. 521
[81] E. Illekov, I. Matko, P. Duhaj, F. Kuhnast,
(2010) 428.
J. Mater. Sci., 32 (1997) 4645.
[67] J. estk, Phys. Chem Glass 15 (1974) 137.