Baixe zanotto 1989 e outras Notas de estudo em PDF para Engenharia de Produção, somente na Docsity! Trends in homogeneous crystal nucleation in oxide
glasses
E. D. Zanotto* & M. C. Weinberg
Department of Materials Science and Engineering, University of Arizona,
Tucson, AZ 85712, USA
Manuscript received 26 September 1988
Calculations are performed to estimate the temperature
of maximum nucleation rate, Tmax for several oxide
glasses. H is found that for one class of glasses for which
Tmas<T, (where T, is the glass transition temper-
ature ), homogeneous crystal nucleation has never been
experimentally observed. The postulate thar the failure
to detect homogeneous nucleation in these glasses is due
to transient nucleation effects is tested for several cases.
Ft is concluded that for some glasses, such as B,0; and
Na,0.41,0,.6SiO, (albite), long transient times ap-
pear to be a contributing factor to the prevention of
homogeneous crystallisation. However, it is demon-
strated that transient nucleation effects are not re-
sponsible for the faci that homogeneous nucleation
would not be observed in Na,0.2Si0; and PbO.SiO;,.
Hence, the failure of glasses with low Tax to crystallise
homogeneous!y cannot in general be attributed to trans-
ient effects and one must conclude that their steady state
nucleation rates are particularly small.
Although a large amount of information exists regard-
ing crystal nucleation in glasses, during the past 10-15
years the realisation has developed that several basic
questions pertaining to nucleation in glasses remain
unresolved.“-* This has come about, in part, due to a
number of experimental studies in which homo-
geneous crystal nucleation rates in simple glass com-
positions have been measured.é 9 Hence there has
been an effort to obtain homogeneous crystal nucle-
ation data in as many systems as possible.
During the course of this endeavour two difficulties
have been encountered. First, homogeneous and het-
erogeneous nucleation are not always unambiguously
distinguishabie, and this problem has been addressed
in several works;º 1 second, a very limited number
of simple glass compositions appear to exhibit homo-
geneous crystal nucleation. Although an explanation
can be given for this with some systems, in general this
finding is not understood. Homogeneous nucleation
seems to appear at rather large undercoolings while
certain compositions show a tendency towards surface
nucleation at relatively small undercoolings and if the
*On sabbatical leave from Universidade Federal de Sao Carlos, Brazil.
186 Physics and Chemistry of Glasses Vol. 30 No. 5 October 1989
surface nucleated crystallites also tend to exhibit rapid
growth, then complete crystallisation of the sample
may ensue at modest undercoolings. Under such
circumstances deep undercooling of the glass is pre-
vented, and hence it would be impossible to seek
homogeneous crystal nucleation in such systems.
For a number of glass compositions, however, deep
undercoolings are possible with a minimal (or at least
tolerable) amount of surface crystallisation in
evidence. Under these circumstances it is feasible to
seek for signs of homogeneous nucleation; a number
of such experiments have been performed, and homo-
geneous nucleation has been detected in only a small
percentage ol the systems studied.
Three conditions must be satisfied for homogeneous
erystal nucleation to be observable (by common
microscopic techniques): (i) the crystal growth rate
must be sufficiently large (in some temperature region)
so that the nuclei can be grown to a size where they are
detectable; (ii) the steady state homogeneous nucle-
ation rate must be sufficiently large (say greater than
10? m”? s-! in order to form a sufficient number of
particles within a reasonable time; (iii) the transient
times cannot be too long in the temperature region of
significant nucieation. In most cases condition (i) is
easily satisfied but there are a few notable exceptions,
two of which will be mentioned herein. Hence, the
detection of homogeneous nucleation usually depends
upon the magnitude of the steady state rate or the
length of transient times in the nucleation region.
In the present work we observe a trend in the
experimental results gathered to date. Homogeneous
nucleation has been reported for glass compositions
which have relatively low reduced glass transition tem-
peratures, T,/T, and whose (predicted and observed)
temperature at which maximum nuckation occurs,
Tomas above the glass transition temperature. On
the other hand, homogeneous nucleation has not been
reported (at least for inorganic glasses) for those
compositions for which the temperature of maximum
(predicted) homogencous nucleation occurs below the
glass transition temperature. It might be suspected
that the latter compositions have long transient times
and hence that the lack of observable homogeneous
E. D. ZANOTTO & M. C. WEINBERG: HOMOGENEOUS NUCLEATION IN OXIDE GLASSES
nucleation in such systems is due to transient effects;
this hypothesis is analysed in detail for several
compositions.
Trends in the location of the nucleation curve
For seven glass compositions James!" demonstrated
that the experimental values of Tao, are always at or
somewhat above T, and that T,/T is in the range 0:54
to 0:59. In a recent publication! ? one of the present
authors shows that, for homogeneous nucleation, the
calculated values Of Trax are also close to T, in
agreement with James's observations; it was also
demonstrated that the reverse also applies, i.e. if the
predicted Tmas falls below T;, only surface (hetero-
geneous) nucleation is observed.
Here, we calculate Ta, for 12 oxide glasses. It
should be emphasised, however, that the values for
Fº predicted by classical nucleation theory are many
orders of magnitude lower than the actual valuestt”
but the predicted temperature dependence is well
described by theory when the viscosity is used to
estimate the kinetic barrier. Therefore, only the lo-
cation Of Tax Will be computed.
Location of the steady state maximum nucleation rate
The steady state nucleation rate, according to classical
theory, may be written?
cr
f=—"—exp(— W*/R 1
KT p( RT) [89]
where C is a constant, T is temperature, q is the
viscosity, R is the gas constant, and W* is the bulk free
energy required to form a critical nucleus which, for
spherical nuclei, is given by
16n9º
4 =
“rar a
where o denotes the liquid-crystal surface energy and
AG, is the bulk free energy difference between liquid
and crystal per unit volume.
The temperature at which the nucleation rate is
maximum can be obtained by solving the equation
dr/dT=0. a reduced temperature, T,, is defined as
T/Tw where E, is the melting temperature, and f is
the entropy of fusion in units of R, then the temper-
ature of maximum steady state nucleation rate may be
found from
s
too Sicnma=o 63
where
12% +ylnT)
AT) vin T,
h=[ATU =) Tin TI (3
In deriving Equation (3), it has been assumed that the
difference in specific heat betwcen crystal and liquid,
AC,, is constant and y = AC,/AS,, (where AS, is the
(4)
Physics and Chemistry cf Glasses Vol. 30 No. 5 October 1989
entropy of fusion); furthermore, it was assumed that o
is proportional to the enthalpy of fusion with «, the
Turnbull ratio, ranging from 1/3 to 1/2; finally,
AT=1—T. The viscosity was taken to be of the
Fulcher form,
Iny= b 6
nn=a+ qo (6
T
em For
Table 1 shows the thermodynamic and viscosity
data employed in the present calculations and the
results are shown in Figure 1, where the reduced
maximum nucleation temperatures are plotted against
To Tay the predictions are given as lines or geometrical
figures rather than points. The extension along the
Table 1. Thermodynamic and viscosity data for several
glass forming systems
To Aa
Glass (kJ tdimole) A B n
1. Na,0.2C80.3S10, NC;S; 1564 BIO 486 4893 547
2 Lij0.250, 15, 1307 s7%0 E8L 1347 595
3 BaO 250, BS; 1693 37000 183 1702 795
4 GO; Go JI38T ISO 994 17960 O
—680 16393 0
5 C4O.A1,0, 2510, CAS; 1826 135500 —585 6750 738
6. Na,0 2810, NS, 147 35500 064 2315 Sé
45190
7 Li0.P,05 LP 926 6100 —410 2000 462
& PO, Pp 853 21760 687 0071 O
27200
9. PPO.SIO, PS 1037 34000
60420
10. SiO, s 1996 15000
11. NajO.ALO, ÓSIO, NAS, 1380 55000
12 80, B 73 2600
login=4+BAT. T) Pas (1) togty)
2) logtm)
Homogeneous nueleation
Tola
Heterogensous nucleation
os 06 97
E
Figure 1. Plot of the calculated values of reduced temperature of
maximum nucleation frequency against reduced glass transition tem-
perature interval (lines and geometrical figures)
O experimental points for NC;S,, LS,, and BS, glasses
187
E. D. ZANOTTO & M. €. WEINBERG: HOMOGENEOUS NUCLEATION IN OXIDE GLASSES
E
LO poses.
100 years
Inri
18 ft year
LS 16 17 15 197 20
TaT
Figure 3, Predicted values of transient times in the temperature range
where the nucleation rates should be highest, ie. between the two
tímits of Tuox
rates are expected to be maximum, ie. between the
two limits for Tnax- The induction periods increase
with increasing T,, (reduced glass transition temper-
ature), being quite short for glasses with low T, and
extremely long for compositions with T, >075.
Hence, onc might conclude that the fact that homo-
gencous crystal nucleation could not be observed
in these systems is due to the suppression of the actual
nucleation rate in comparison to the steady state
rate. This point is illustrated in Table 2 where the
transient times at the temperature of maximum nucle-
ation and the ratio of actual to steady state nucleation
rates are listed for those systems possessing very long
transient times: it may be observed that the nucleation
rates are very small fractions of the steady states rates
even after prolonged heating. These data seem to
support the hypothesis that the absence of homo-
geneous nucleation results from transient efiects and
in the remainder of this work the plausibility of this
argument is analysed in more detail and comparisons
are made with experimental findings.
Table 2. Predicted values for 3 = 1/2
Himas)
System Tum es) Tio After 1 hours)
B sd 39x 1075 16x 10 1º na 0x 1018
NAS, 04% 97 x 101 96x [0-0 n=3x10º
LP 0595 12x 1010 TEXIOTL n=3x 108
NS; 0600 72x 10º 76x 1071 n=175
s 0690 DO x 108 1Oxt08 n=65
Ps 0575 8x 108 30x 1071 n=50
Physics and Chemistry of Classes Vol, 30 No. 5 October 1989
Analysis of transient effects
Governing equations
Here, the influence of transient effects upon the po-
sition and magnitude of the maximum nucleation rate
will be discussed. For times where transient nucleation
effects are not negligible, the position of maximum
nucleation is shifted to higher temperatures (relative
to the temperature of the steady state maximum) and
the magnitude of the nucleation rate is reduced. This
behaviour is analysed for particular choices of the
steady state nucleation rate and the transient time
behaviour.
If Kt, T) denotes the time dependent nucieation rate
and IT) the steady state nucleation rate, then
I=TANHEO (o)
where f(t, 1) is a function which describes the transient
nucleation and depends upon the transient time, r.
The position of the maximum nucleation rate as a
function of time can be found by taking the temper-
ature derivative of Equation (10) and setting it to zero.
One obtains
dinfº/dT = x(dlnf/dx) (dInt/dT) (11)
where x is the time scaled by the transient time. For
the transient time behaviour we choose
NO) =1+25 (exporta, (12)
which is the expression derived by Kashchiev.(16) If
Equation (12) is employed in conjunction with the
following form for the logarithm of the relaxation
time,
Int=2in(T,-ND+2+bMT— T), (13)
then the right side of Equation (11) becomes
x(dlnf/dx) dinc/dT = [AT DAT = Teto)
(14)
and
gog=2x 5 (It nexp(—n?o)/fo. (15)
as
If the reduced temperature, T;, is introduced and
Equation (3) is used for the left side of Equation (11),
then use of Equations (14) and (15) leads to
b 16n2ºBjh
acmmtos
272 b
=etd [a = ir) a
Solutions of Equation (16) for T, for varying x (or t)
give the location of the maximum nucleation temper-
ature as a function of time. The steady state maximum
nucleation temperature is obtained from the solution
of Equation (16) with the right side of this equation
equal to zero.
Tt will be of interest, also, to compare the maximum
nucleation ratc at some reduced time, x*, with the
189
E. D. ZANOTTO & M. €. WEINBERG: HOMOGENEOUS NUCLEATION IN OXIDE GLASSES
steady state maximum nucleation rate. If TH is a
solution of Equation (16) at x*, then this ratio of
maximum intensities is given as
HKTEox) JT) *
Ditt le ms, TO. 17
Td mol O TO (9
1f Nº denotes the number of particles nucleated at
some temperature when heated for some time period
under steady state conditions then, since N(1)=
frade, it is possible to show that
N ah a (pt “mn 2
(3) LÊ Cl eetro cla
Nº x [1 nê 12
which is an expression for the reduction in the number
of particles formed at a given temperature due to
transient effects. The number density ratio, which is
given by the ratio between the number formed at T,
and the steady state number formed at Tm, (i.e. the
number density analogue of Equation (17)), is also of
interest and is given by
Ns T) NE) NÁT)
NÃ Tm) NT) NÚ Tuas)
“O
Pas)
2& ar explontx) a?
«[uiien 5 -
n
(19)
Shift in maximum nucleation temperature
Now we employ Equations (13) and (16)(19) to
predict the shift in peak nucleation rate position and
the reduction in the magnitude of the nucleation rate
(and particles formed) due to transient nucleation
effects. Tt is of particular interest to investigate the
behaviour of several systems whose steady state max-
imum nucleation rates were predicted to fall below T,
in order to test the hypothesis that transient nucle-
ation is responsible for the absence of homogeneous
nucleation in such systems. Attention will therefore be
focused primarily upon three of the compositions
(NS,, PS, and B) listed in Table 2; these were chosen
because there is information in the literature regarding
their crystal nucleation and growth behaviour.
First, however, we point out that the results shown
in Table 2 were generated on the assumption that
a=1/2 but if one chooses «= 1/3, the predicted
Table 3. Predicted values for u = 1/3
Tuma)
System Tou (83
B 0705 17x 108
NAS, 0600 26x 108
LP 0640 10x 103
NS, 0660 92x 10!
5 0780 30x10º
Ps 0615 50x10º
190 Physics and Chemistry of Glasses Vol. 30 No. 5 October 1989
values Of Tnax OCcur at much higher values and con-
sequently the transient times at the temperatures of
maximum nucleation rate will be much smaller (com-
pare Tables 2 and 3). In this section we consider the
worst case (a = 1/2) to analyse if transient effects can
obstruct the detection of homogeneous nucleation.
Results
Nas0.280, (T, = 723-746 K)
Figure 4 shows the variation in the temperature of
maximum nueleation rate with heat treatment time for
NS; glass; there is a marked decrease in Toa With
increasingly longer times, reflecting the influence of
transient nucleation in this temperature range
(700-750 K).
The calculated values of 7 and the ratio between the
steady state nucleation rate at temperature T and the
maximum rate 14 Ts) are shown in Figure 5. The
transient period at Trax (688 K) is very long but at
30 K above Trax, it is reduced to about 1 h, while the
steady state rate is lowered by only one order of
magnitude. Hence, if the magnitude of 1 ( Tas) is large
enough, homogeneous nucleation might be measured
in NS, if one chooses an appropriate thermal treat-
ment, say between 710 and 730 K for a few hours.
Crystal nucleation in a NS, glass (with 0:3 mol%
Sb,0,) was sought, exhaustively, by Hishinumal9) at
700 K for 220 h, at 770 K for 250 h, and at 820 K for
7%
Tosa [KO
no
70
gol, : 4
2 1 q 1 2 3
log (time in hours)
Figure 4. Temperature of maximum nucleation rate as a function of
heat treatmen: time for NS» glass
é o
4 2.
2 «log (r in seconds)
mol
oe.
gi
60 70 70 7 750 HO 750
Absolute temperature
Figure 5. Calculated values of transient times and the ratio of the
steady state nucleation rate ar temperature, T, to the maximum rate,
T/Tuax), for NS»
E. D.ZANOTTO & M. €. WEINBERG: HOMOGENEOUS NUCLEATION IN OXIDE GLASSES
IT
-2 “a
-s
º o lo (NR
s + aR i
fo. !
-8d |
i
tl
a 190 Eu
Time (h)
Figure 6. Ratio of predicied nucleation rate to iss maximum value at
700 K for NS,. Similar ratio for number of crystals formed in time
140 h. For both one step and two step heat treatments
crystal growth rates would have been sufficiently large
to reveal any internal crystals if they existed but only
surface crystallisation was observed. At 700K, the
predicted nucieation rate would approach its steady
state value after about 100--200 h, even for the worst
possible condition (a = 1/2), as is shown by Figure 6,
and thus homogeneous nucleation should have been
detectable by Hishinuma's experiments, At the other
two temperatures explored, 770 and 820 K, the transi-
ent times are much shorter, but so are the steady state
nucleation rates, rendering internal nucleation
unobservable.
PbO.SiO, (T,=674-695 K)
Similar calculations were carried out for PS glasses,
again for «= 1/2, and Figure 7 shows the transient
times and the ratio between the steady state nucle-
ation rate at temperature T and its maximum valve at
Thasx: in this case the transient times at Taz (597 K) are
very long. HE one chooses a higher temperature, say
650 K, where 7 would be greatly reduced, the steady
state nuclcation rates would be too much decreased,
as is shown by Figure 7. At 612 K, for instance, a heat
mo
A logtrin seconds)
+ fatia
SO SO 610 GM 60 60 60 cão
Absolute temperaturo
Figure 7. Calculated values of transient times and the ratio of the
steady state nucleation rate at temperature, T, to the maximum rate,
P(Toaxh. for PS
Physics and Chemistry of Glasses Vol, 30 No. 5 October 1989
treatment of approximately 2000 h would be required
to bring the nucleation rates (and crystal number
density) to within 2 orders of magnitude of their
steady state values. However, for a = 1/3, Tras = 700K
and the transient times would be only 5s, as is shown
by Table 3,
Hishinumal? subjected a PbO.SiO, glass to
several heat treatments-—at 670 K for 140 h, from 710
to 730 K for 2200 b, at 760 K for 140h, and also at
higher temperatures (780 to 900 K)—-and onty surface
crystallisation was observed. If 1Y Trax) had been
suficientiy high, internal nucleation should have been
detected by Hishinuma's experiments (for « = 1/3); for
a close to 1/2, homogeneous nucleation would not be
measurable.
B,0s (T,= 5530-564 K)
B,O, is an interesting system because it has the
longest induction period among the twelve compo-
sitions studied, with a Toa, Of SIOK for «= 1/3 and
463K for x= 1/2. Both temperatures are well below
T, and the induction periods are extremely long,
46 x 10º and 10!2h, respectively.
For «= 1/2 no thermal path can be found which
will lead to detectable homogeneous nucleation within
a reasonable time period; for the other limiting case,
= 1/3, heat treatments in the vicinity oí 530K for
1000h would be required il nucleation were to be
detected. To the authors" knowledge, however, no one
has been able to observe crystallisation (internal or
surface) in B,O; glass, event when the melt or the
external surface has been seeded with B,O, crystals.
This indicates that the crystal growth rates are much
too low and thus even if internal nucleation occurred,
it would not be observed since the nuclei would not
grow sufficiently. This is the most difficult case to
analyse since 1 is very long, the growth rate is
extremely small, and the nucleation rate is unknown.
Other systems
Albite (NAS,) glass also does not crystaltise, even
when seeded, indicating a very low growth rate, For
other glasses, such as SiO, and P,Os, the predicted
Trax are below T, but the transient times are such that
thermal paths can be found, so that they will show
homogeneous nucieation. However, SiO, glass, which
has been extensively studied due to its commercial
importance, has only shown surface nucleation and
this is also true for P,Os, although this has been
studied much less.
Summary
The fact that homogencous nucleation cannot be
observed in glasses may be related to one or more of
the following causes: low nucleation rate, low growth
rate, and long induction times. It has been shown that
for many oxide glasses which do not exhibit homo-
geneous crystal nucieation the predicted Tra will
occur well below 7, leading to the reasonahle deduc-
191