j.jeurceramsoc.2013.06.014

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Available online at www.sciencedirect.

com

Journal of the European Ceramic Society 33 (2013) 3403–3410

A kinetic study of the quartz–cristobalite phase transition


Lucia Pagliari a , Monica Dapiaggi a,∗ , Alessandro Pavese a,b , Fernando Francescon c
a Dipartimento di Scienze della Terra, Università degli Studi di Milano, via Botticelli 23, 20133 Milano, Italy
b IDPA-CNR National Research Council, via Botticelli 23, 20133 Milano, Italy
c Ideal Standard International, C.O.E., Ceramic Process Technology, via Cavassico Inferiore 160, I-32026 Trichiana, Belluno, Italy

Received 26 April 2013; received in revised form 30 May 2013; accepted 8 June 2013
Available online 11 July 2013

Abstract
Cristobalite is a common silica polymorph in ceramics, as it can crystallize in SiO2 -rich systems during high temperature processes. Its occurrence
in final traditional ceramic bodies remarkably affects their thermal expansion, thus playing an important role in the shrinkage upon cooling. The
quartz–cristobalite transformation kinetics is investigated by in-situ isothermal X-ray powder diffraction experiments and then correlated to the
average particle size (d) of the starting quartz using a model here developed. An Avrami-like rate equation, i.e. α(t) = 1 − exp(− k × t)n , in which
the n-term is assumed to account for the dependence on the average particle size, has provided the best fitting of theoretical to experimental data,
yielding activation energy values that range from 181 to 234 kJ mol−1 , and exponential n-coefficients from 0.9 to 1.5. Ex-situ observations have
demonstrated that the formation of cristobalite from quartz after 50 min, 2, 4 and 6 h at 1200 and 1300 ◦ C, exhibits a remarkable dependence on d
of quartz, showing comparable behaviours in the case of d equal to 15.8 and 28.4 ␮m, but significant differences for d of 4.1 ␮m. The formation
of cristobalite is boosted remarkably at temperature higher than 1200 ◦ C, with an increase by weight even of 500%, with respect to its content
at lower temperature. The method of sample preparation (dry powder, wet powder and tablet of compressed dry powder) seems to influence the
results only at temperature > 1200 ◦ C and in the case of fine powder.
© 2013 Elsevier Ltd. All rights reserved.

Keywords: Quartz–cristobalite transition; X-ray diffraction; Kinetics

1. Introduction thermal dilations affect the shrinkage upon cooling of the


traditional ceramic bodies, especially in sanitaryware-fireclay
Cristobalite is a high temperature silica polymorph stable technology4 making extensive use of chamotte (calcined clay
in the temperature (T) range 1470–1705 ◦ C, but that also or ball-clay), and plays therefore a relevant role on the quality
occurs out of its field of stability as a metastable phase. The of the final product.5 Moreover, a controlled formation of
theoretical sequence of the silica transformations upon heating cristobalite is also exploited to properly “tune” the thermal
involves ␣-quartz, ␤-quartz (573 ◦ C), ␤-tridymite (870 ◦ C) and expansion of a ceramic body, in order to make it match well
␤-cristobalite, which has a cubic structure (1470 ◦ C). Given that with that of the glaze, thus avoiding the appearance of flaws
the formation of ␤-tridymite requires particular impurities to be due to a mismatch between inner and surface. For this reason,
present,1 ␤-cristobalite can even appear at lower temperature. much interest is paid to cristobalite-involving reactions that
Cristobalite also develops at T ≤ 1000◦ C from SiO2 glass.2 take place in the traditional ceramic industrial processes.6
Upon cooling, ␤-cristobalite turns into its tetragonal ␣ phase at Moreover, the stage of advancement of any high temperature
270 ◦ C, not being able to overcome the activation energy barrier transformation, both in natural and industrial environment,
to change into the stable silica phase. ␣- and ␤-cristobalite depends on the particle-size-distribution (PSD) of the involved
have average bulk thermal expansion values of 91.7 and materials and firing-time experienced.7 Such an aspect becomes
6 × 10−6 ◦ C−1 , respectively.3 Such remarkably different crucial in traditional ceramic manufacturing contexts, in which
the raw materials used have undergone milling treatments to
comply with processing features of the wholesale production,
∗ Corresponding author. Tel.: +39 02 50315605; fax: +39 02 50315597. thus steering the phase composition and texture of the final
E-mail address: monica.dapiaggi@unimi.it (M. Dapiaggi). output. Even though are aware that quartz is not the only source

0955-2219/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jeurceramsoc.2013.06.014
3404 L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410

for cristobalite formation during ceramic production (see for


example,8 and references within), being it the metakaolinite
coming from thermal decomposition of kaolinite, still the kinet-
ics of the phase transition is an important step in the course of
understanding the kinetics and the energetics of the whole high
temperature process that generates a traditional ceramic body.
In the light of the discussion above, it is of interest to get
insight into kinetic mechanisms of the transformations of SiO2 -
based systems leading to ␤-cristobalite, and in particular, as a
fundamental reference, into the phase transition from quartz.
The aim of the present study is to investigate the
quartz–cristobalite transformation paying particular attention
to PSD of the starting quartz powders, firing-time, firing-
temperature and sample preparation. In this view, we planned
two sets of experiments:

Fig. 1. Particle size distribution of the five powder investigated.


(i) those in which measurements are carried out on samples
that have undergone annealing after thermal treatment;
eventual residue found after sifting. PSDs were eventually mea-
(ii) those that are meant to enable the study of the reaction
sured by a CILAS LASER MOD 920L.
kinetic parameters.
2.2. Ex-situ and in-situ high temperature experiments
Such investigations are carried out by ex-situ and in-situ
X-ray powder diffraction experiments; the quartz–cristobalite We carried out three series of high temperature experiments.
transformation kinetics is explored by isothermal runs and then The first one was meant to provide an overview on the
interpreted by a formalism here developed that uses the Avrami behaviour at non-ambient conditions of the available samples.
model. Ex-situ analyses were performed on samples heated in a LKN-
S86 Nannetti muffle furnace at 1200 and 1300 ◦ C (ramp time:
2. Experimental 6 h), and then cooled slowly down to room temperature. Three
sample preparation techniques were used: (i) dry compression
2.1. Samples to form tablets; (ii) mixing sample and water to obtain a paste;
(iii) bare powders. Altogether, 21 samples were produced, three
In the present investigation five powder quartz samples pro- for each PSD. X-ray powder diffraction (XRPD) patterns were
vided by Sibelco were used, with different PSDs, as displayed collected by a θ–2θ Bragg-Brentano parafocusing PANalytical
in Fig. 1. PSD- and micro-structure parameters are set out in X’ Pert PRO diffractometer, equipped with multi-channel X’
Table 1. Celerator detector. A Cu-K␣ radiation with λ = 1.5418 Åwas
Given that the d-values of the quartz powders range from used. The optics were very simple, with a fixed divergence slit
4 to 29 ␮m but none of them lies in the interval 4.5–15 ␮m, of 0.5◦ and an antiscatter slit of 0.5◦ . Samples were ground and
two more mixtures were prepared, by blending the two finest then pressed in a back-loading sample holder; diffraction sig-
raw powders: 50% SA600 (50% SA600 +50% SA250S; nals were collected from 5◦ to 80◦ 2θ, with a step size of about
d = 9.958 ␮m) and 70% SA600 (70% SA600 +30% SA250S; 0.017◦ 2θ, and a counting time of 30 s per step.
d = 7.629 ␮m). The PSDs were measured as follows: first, 7 g In the second series of measurements the coarsest, the
of sample were mixed with 40 ml of de-ionised water, 5 ml of finest and the intermediate grain-size powders (SA10S,
1% Sodium Hydroxide solution and 5 ml of 1% Sodium Hexa SA600 and SA250S respectively) were used to investigate
meta Phosphate (Calgon) solution, in a high speed homogeniser, the quartz–cristobalite transformation as a function of time-
at 10,000 rpm, for 1 min; then, the resultant slurry was screened temperature. Wet samples were chosen for such experimental
through a 125 ␮m sieve and classified, taking into account the runs as they provide conditions closer to those of the industrial

Table 1
Granulometric characterization of the starting powders. Dv  and RMS strain come from the analysis of the powder diffraction peak broadening (see the experimental
paragraph for the details).
Sample d (␮m) D50 (␮m) Mode (␮m) Dv  (nm) RMS strain (×10−4 )

SA10S 28.38 23.42 50.22 270 (3) 3.81 (5)


SA6S 24.59 17.90 34.58 251 (3) 4.06 (5)
SA12S 23.05 18.24 31.50 256.8 (1) 3.84 (4)
SA250S 15.78 10.90 23.81 230 (3) 3.72 (6)
SA600 4.135 2.689 2.787 167 (1) 3.48 (7)
L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410 3405

environments. The samples were heated at 1200 and 1300 ◦ C simplicity, through a histogram composed of a set of j-bins,
for 50 min, 2, 4 and 6 h, and then annealed. Heating treat- each bearing a ξ j fraction by weight of the original material with
ments and XRPD measurements were carried out with the same average 2Dj size. Let us introduce the following definition for a
instruments and experimental set-ups reported above. generic observable F that takes the values Fj in the jth-bin:
The third set of measurements was oriented to the study 
of the reaction kinetics of the quartz–cristobalite phase tran- F= Fj × ξj . (4)
sition, by in-situ experiments. A θ–θ geometry Philips X’ Pert j
diffractometer was used, equipped with an Anton-Paar heating Assuming that the activation energy Ea is the same for any
chamber (HTK 16 MSW) and a Pt heating strip, which is able to particle size, one then can write the reaction kinetics equation
achieve a maximum temperature of ∼1600 ◦ C. Explorative heat- for the j-fraction as
ing ramps were performed on the finest and coarsest samples  
(SA600 and SA10S, respectively): the maximum temperature Ea
g(αj ; pj ) = Aj exp − ×t (5)
was ∼1300 ◦ C, and data were collected from 19◦ to 75◦ 2θ at RT
room temperature before and after treatment, and from 19.5◦ to where pj is a possible parameter depending on the
27◦ 2θ, every 50 ◦ C, until about 500 ◦ C, and every 5 ◦ C from 500 jth-bin. In some case, it may be convenient to use
to 1300 ◦ C, with a step-size of 0.03◦ 2θ and counting time of 1 ln(g(αj ; pj )) = ln(Aj )(− Ea /RT) + ln(t); the formalism we are
s/step. Such 2θ-intervals allow one to record the most intense developing does not change, and hereafter we assume to proceed
peaks of quartz and cristobalite. Isothermal kinetics experi- with Eq. (5). If one sets αj = α+j and pj = p + δpj , then from
ments were then carried out by the same diffractometer and high Eq. (5) it follows:
temperature device, and using SA600-SA10S-SA250S-samples.
Each sample was studied along three isotherms, chosen as a    
Ea
function of the preliminary ramp experiments. g(α + j ; p + δpj ) × ξj = exp − ×t× A j × ξj
XRPD patterns were collected for all of the samples, before RT
j j
and after thermal treatments. The quantitative analyses and (6)
the determination of the volume-average crystallite coherence
If one Taylor-expands in j and δpj the right-hand member of
domain size (Dv ) and root mean square (RMS) micro-strain9
Eq. (6), and
were performed by the Rietveld refinement technique combined
with the Warren-Averbach10 method (MAUD software11 and  
(i) takes into account that j j ×ξ j = 0 and j δpj × ξ j = 0;
isotropic model). For all the experiments, we were not able to
(ii) neglects the second order terms in j and δpj ;
refine the RMS strain parameter of cristobalite if its content
then it follows
was less than 2 wt.%: in such cases the too low diffraction peak
intensity did not allow the separation of the size and strain profile   
Ea
contributions. g(α; p) × ξj = exp − × t × A. (7)
RT
j
3. Kinetic theory We tried out several kinetic models,12 and chose an Avrami-
like15–17 formalism as it proved to be the most satisfactory one
The fundamental equation describing reaction kinetics, in terms of: (i) matching between observations and theory, (ii)
which is reported below12 : and consistence between the inferred Ea s versus d of quartz. In
dα this light, the reaction rate is described by the general equation
= kf (α), (1) beneath
dt
is often cast into its integral version, namely [− ln(1 − α)]1/n = k × t. (8)

g(α) = kt, (2) We take that the dependence on each jth-bin is transferred
to the exponential n, i.e. nj , and the equation above changes
where α is a generic coordinate that monitors the new phase accordingly into
formation (hereafter called “coordinate of transformation”, and  
for which lim α(t) = 1), t is time, f and g are some functions 1 Ea
t→∞ [ln(− ln(1 − αj ))] × = ln(Aj ) + − + ln(t). (9)
nj RT
of α, depending on the nucleation-growth model. k is the rate
constant, whose dependence on temperature is described by the Summing over all bins one has and taking into account the
Arrhenius equation: fraction by weight for each jth-bin, one has

Ea
  1
 
Ea

k = A exp − . (3) ln(− ln(1 − αj )) × × ξj = − + ln(t)
RT nj RT
j
The role of the particle size in affecting the reaction kinetics 
+ ln(Aj ) × ξj . (10)
of a powder has earlier been discussed by 13,14 and it is here
j
re-considered. The PSD of quartz is modeled, by the sake of
3406 L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410

Table 2
Chemical and physical characterization of the starting raw powders.
Sample % min % max % max % max % max Loss on Density Specific surface
SiO2 Al2 O3 Fe2 O3 TiO2 K2 O ignition (%) (T/m3 ) (cm3 /g)

SA10S 99.0 0.3 0.03 0.03 0.05 0.2 1.0 2800


SA6S 99.0 0.3 0.04 0.03 0.05 0.2 1.2 1900
SA12S 99.0 0.3 0.03 0.03 0.05 0.2 0.9 3500
SA250S 99.0 0.4 0.05 0.03 0.05 0.3 0.70 5700 (900)
SA600 99.0 0.5 0.06 0.03 0.05 0.3 0.5 13,800 (2800)

From Eq. (7) it finally follows: A decrease of the particle size is accompanied by an increase
of the peak’s broadening, indicating a general reduction of the
  ξj  
Ea
 crystalline domain size and increase of RMS strain. The inverse
ln(− ln(1 − α) × = − + ln(t) relationship between crystallite size and RMS strain is very
nj RT
j well known in literature in materials subjected to mechanical
 treatment.23,24
+ ln(Aj ) × ξj , (11a)
j
4.2. Characterization of the treated powders
and finally
 
Ea 4.2.1. Ex-situ experiments
(− ln(1 − α))1/n̂ = Â × exp − × t, (11b) The quantity of cristobalite obtained from the ex-situ exper-
RT
 iments at 1200 ◦ C and 1300 ◦ C for 6 h is displayed in Fig. 3, as
 where (1/n̂) = (1/n) = j (ξj /nj ) and ln(Â) = ln(A) = a function of d and sample preparation.
j ln(Aj ) × ξj are the quantities measured by isothermal exper- Figure 3A shows the results related only to experiments
iments. with the maximum temperature of 1200 ◦ C, whereas Figure
3B collates the issues from both temperatures. The formation
4. Results and discussion of cristobalite through such treatments exhibits a remarkable
dependence on both d and sample preparation. Rising tem-
4.1. Characterization of the starting powders perature by 100 ◦ C causes the amount of formed cristobalite to
increase by more than 6 times, for the sample with the small-
Tables 1 and 2 show the results of a characterisation of the est particle size, and by 2–3 times, for the coarsest one. The
initial quartz powders. The volume-average domain size (Dv ) method of sample preparation influences the final quantity of
and the RMS-strain of quartz are presented in Table 1, and can cristobalite only when the firing temperature is 1200 ◦ C, and
be directly compared with the granulometric parameters in the in particular for the finest samples: the wet sample preparation
same table. In particular, one has that the smaller d the smaller boosts cristobalite (Fig. 3A), whereas the bare and dry powders
Dv . The differences in terms of Dv  between starting powders yield the lowest amount; such difference nearly cancels out at
are expectable even by a bare inspection of Fig. 2, which shows 1300 ◦ C (Fig. 3B). This is likely ascribable to that wet prepara-
the portion of diffraction pattern wherein the main peak of quartz tion allows a better particle-particle contact, thus favouring heat
is located. transmission and solid state reactions induced thereby.

Fig. 3. Wt% of cristobalite after the ex-situ test conducted at 1200 ◦ C (A) and
Fig. 2. The main peak of quartz for the five starting powders under investigation. at 1300 ◦ C (B). B shows again results at 1200 ◦ C as a comparison.
L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410 3407

A 15

wt% Cristobalite
10

)
m
4.135


ze
15.78

Si
in
ra
28.38

G
of
n
52min 2h 4h 6h

ea
Time

M
B 60

wt% Cristobalite
40

20

0
)
m

4.135

ze

15.78
Si
in
ra

28.38
G
of
n

52min 2h 4h 6h
ea

Time
M

Fig. 4. Amount of cristobalite formed at 1200 ◦ C (A) and 1300 ◦ C (B) applying
the maximum temperature for different times.

Fig. 4 concerns the cristobalite formation in the finest, coars-


est and intermediate size samples at 1200 ◦ C and 1300 ◦ C, as a
function of time. Fig. 5. Relationship between Dv of quartz and the amount of cristobalite
At 1200 ◦ C, the amount of cristobalite is in general small, but formed at 1200 ◦ C (A) and 1300 ◦ C (B).
a remarkable increase occurs in the finest sample only, and after
6 h of treatment. At 1300 ◦ C the effect of the firing time seems to quartz undergone high temperature treatment. The fine samples
be modest after 52 min, regardless of the average particle size. are those least affected, alike the crystallite size.
Fig. 5 shows the Dv of the samples, i.e. the difference Both Dv and RMS strain do not show any dependence
between the crystalline domain size of quartz before and after on the method of sample preparation.
high temperature treatment, as a function of the amount of cristo-
balite formed at 1200 ◦ C and 1300 ◦ C. At 1200 ◦ C (Fig. 5A), all 4.2.2. Heating ramp experiments
the coarse samples show a negative value for Dv , and therefore In-situ heating ramp experiments on the finest and the coars-
they exhibit a larger average crystallite size after heating (grain est powders were performed, reaching about 1300 ◦ C (details
growth is well known to be promoted by high temperatures). on the procedure can be found in the experimental section). The
However, all samples with a small d (below about 16 ␮m) have coarsest sample (SA10S), of which Fig. 6A shows the diffrac-
a smaller average crystallite size after heating, showing there- tion patterns in the interested thermal range, starts to transform
fore that the reactions occurring on cooling have an influence into cristobalite at 1250 ◦ C, and at 1300 ◦ C about 2.70% of this
on the coherence of the domains that tend to break due to the latter is formed.
volume difference of the involved polymorphs. At the highest The finest sample (SA600) gives at the end of the heating
temperature (1300 ◦ C, Fig. 5B), this is again true, but the sam- ramp about 40.6% of cristobalite, which starts to form at about
ple with a d of about 25 ␮m lies somehow in between the two 1190 ◦ C (see Fig. 6B). As expected, this hints a relevant role is
different behaviours. played by the particle size, in quartz–cristobalite conversion.
The difference between the initial and the final strain of quartz
(RMS) was also determined, but it is not here reported for 4.2.3. Isothermal kinetics analyses
the sake of brevity. Such difference is negative for all of the On the basis of the ex-situ and in-situ experiments dis-
samples explored, proving the occurrence of residual stress in cussed above, we have chosen the temperature values at
3408 L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410

Fig. 7. Application of the Avrami equation for the intermediate grain-sized


sample (SA250S), heated at 1550, 1575 and 1600 K.

temperatures. As demonstrated, the data can be fitted nicely with


an Avrami-type model; this means that the formation of cristo-
balite starts with an induction time (necessary for the growth of
critical-sized nuclei, which later start to grow into proper crys-
tals), followed by an acceleration period. The induction time is
strongly temperature-dependent: it is longer at lower temper-
ature, but considerably decreases with increasing temperature.
The different slope in the data with the lowest temperature may
therefore be due to this reason.
The kinetic exponent n ranges from 0.89 to 1.49 (see Table 3):
Fig. 6. Diffraction patterns at increasing temperatures collected in the little angle
range which contains the main peaks of both quartz and cristobalite. A refers to
from,18 fractional n-values hint at diffusion-controlled crystal
the coarsest sample (SA10S) and B to the finest provided powder (SA600). growth, or processes involving complex size and shape particle
distributions for reactants and transformation products.
which to carry out isothermal measurements to investigate the In particular, as to n one observes that:
quartz–cristobalite reaction kinetics, using the finest (SA600),
the coarsest (SA10S) and the intermediate (SA250S) grain-size 1 SA250S and SA10S give ns close to each other, with n
samples. around 1;
By way of example, Fig. 7, which refers to SA250S but is rep- 2 SA600 yields ns about 1.5, save for T = 1500 K. This brings
resentative also of the other samples, reports the three isotherms to light that the finest sample exhibits a remarkable sensi-
investigated; they trend linearly and are nearly parallel to each tivity to T, as proven by that a 25 K-shift is sufficient to
other, thus suggesting that the reaction mechanism is indepen- affect the kinetic exponent. Taking into account Eq. 11a, it
dent of temperature, over the explored range, for each sample. is likely that an effect related to the size-classes gives rise
Moreover, the first data points, referred to the lowest tempera- to such discrepancy with respect to what occurs at higher
ture, follow a trend with different slope with respect to the higher temperatures;

Table 3
Results from the kinetics experiments.
Sample Average grain size d (␮m) Working T (K) n n ln(k) (s−1 ) A (s−1 ) Ea (kJ mol−1 )

1500 1.15 −9.37 193(73)


SA600 4.135 1525 1.47 1.37 −9.28 411.83
1550 1.49 −8.87
1550 1.06 −9.22 181(78)
SA250S 15.78 1575 1.04 1.01 −9.16 115.7
1600 0.94 −8.78
1600 1.07 −9.51 234(191)
SA10S 28.38 1650 0.92 0.96 −8.44 3710.79
1680 0.89 −8.78
L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410 3409

Fig. 8. Arrhenius plot for SA250S, starting from the values of k obtained from
Fig. 10. Amount of cristobalite after the isothermal analyses for the three sam-
the application of the Avrami model.
ples investigated.

3 the average grain size influences n so as to lead to two distinct A-pre-exponential coefficients exhibit a remarkable scattering,
kinetic classes: the one including SA250S and SA10S, and without any apparent correlation to d.
the other SA600; After cooling to room temperature, full-range diffraction data
4 n seems to exhibit an inverse dependence on d. were collected and then analysed with the Rietveld refinement.
Fig. 10 shows the amount of cristobalite crystallized after each
Fig. 8, which represents the Arrhenius plot for the inter- isotherm. As seen before, the sample with the smaller grain
mediate particle-sized powder again, shows a reasonably good dimension (SA600) reacts more easily: cristobalite forms at
correlation (R2 = 0.84, which is similar for the other two sam- lower temperatures with respect to the other samples and in
ples) that allows one to determine the pre-exponential parameter larger quantities upon the same applied conditions.
A and the apparent activation energy Ea of the phase transition At a given temperature the amount of cristobalite formed
process. increases upon decreasing the grain size of quartz involved. The
The activation energies change from 181 to 234 kJ mol−1 , more the particle size increases, the more the temperature nec-
which are in keeping with literature values for such a kind of essary to crystallize cristobalite is higher and the less is the
processes (19–22,25 just as few examples). The Ea ’s are similar tendency of this phase to form. The latter aspect is shown by
to each other, and almost undistinguishable if one takes into that the slopes of the linear interpolations of cristobalite versus
account the uncertainties, determined by the errors propagation. T for each sample in Fig. 10 decrease as a function of d.
In this view, the grain size seems not to significantly influence By way of example, at 1250 ◦ C and α = 0.2, the α(t)-curves
the activation energy of the transition process, as shown in of SA600-SA250S-SA10S intersect each other at ∼1000, 700
Fig. 9, in agreement with the assumption of the model here and 1200 s, respectively; at t = 500 s and α = 0.2, they require
used that takes Ea as independent of the size-class. The temperatures as large as 1300, 1290 and 1320 ◦ C, respectively.

5. Conclusions

The present study has been undertaken with the aim to elu-
cidate the kinetics of the quartz–cristobalite phase transition
which occurs during the production of many ceramic materi-
als. In particular, the attention has been focused on the effects
of size reduction on the cristobalite formation at high tempera-
tures, considering both the obtained quantities of that phase and
the kinetic process of the transformation. On the ground of the
results discussed in this paper, the following conclusions have
been achieved:

1 high temperature promotes cristobalite crystallization and the


method of sample preparation influences the amount of its
formation (wet samples show larger amounts of cristobalite),
but only at comparatively low temperature, i.e. 1200 ◦ C. This
Fig. 9. Apparent activation energy as a function of grain size. is reflective of a kinetic effect;
3410 L. Pagliari et al. / Journal of the European Ceramic Society 33 (2013) 3403–3410

2 the formation of cristobalite is always favoured by a small 6 Sao O, Ozgur C. Investigation of a high stable ␤-cristobalite ceramic pow-
grain size of the starting powder, in keeping with that gener- der from CaO–Al2 O3 –SiO2 system. J Eur Ceram Soc 2009;29:2945–9.
7 Bernasconi A, Diella V, Pagani A, Pavese A, Francescon F, Young K,
ally the smaller the particles, the more easily a transformation
et al. The role of firing temperature, firing time and quartz grain size on
takes place; phase-formation, thermal dilatation and water absorption in sanitary-ware
3 the Avrami model proves that the exponential term, n, is vitreous bodies. J Eur Ceram Soc 2011;31(8):1353–60.
related to the average particle size of starting quartz, whereas 8 Ferrari S, Gualtieri A. The use of illitic clays in the production of stoneware
the activation energy, Ea does not seem to depend on d of tile ceramics. Appl Clay Sci 2006;32:73–81.
the powders on study; 9 Snyder RL, Jaroslav F, Bunge HJ. Defect and microstructure analysis by
diffraction; vol. 10 of international union of crystallography, monographs
4 we observed n-values ranging from 0.89 to 1.49, according to on crystallography. Oxford: Oxford Science Publications; 1999.
the rule that the smaller d the larger n. Ea has average value 10 Warren BE. X-ray diffraction. Reading, MA: Addison-Wesley Publishing
of ∼203 kJ mol−1 . Company; 1969.
11 Ferrari M, Lutterotti L. Method for the simultaneous determination of
Therefore, it is unarguable that the particle dimensions of the anisotropic residual stresses and texture by X-ray diffraction. J Appl Phys
1994;76(11):7246–55.
starting powder plays a crucial role in the cristobalite forma- 12 Khawam A, Flanagan DR. Solid-state kinetic models: basics and mathe-
tion. This is reasonable and confirms what one could expect if it matical fundamentals. J Phys Chem B 2006;110(35):17315–28.
is kept in mind that stronger reactivities are usually associated 13 Koga N, Criado JM. Influence of the particle size distribution on the CRTA
to finer materials. The results of the present work have impor- curves for the solid-state reactions of interface shrinkage. J Therm Anal
1997;49(3):1477–84.
tant implications from a practical point of view and so must be
14 Koga N, Criado JM. Kinetic analyses of solid-state reactions with a particle-
taken into account: on the basis of the conclusions here reported, size distribution. J Am Ceram Soc 1998;81(11):2901–9.
industries could indeed choose the most appropriate raw mate- 15 Avrami M. Kinetics of phase change: I. General theory. J Chem Phys
rials and the most convenient conditions at which these powders 1939;7:1103–12.
should undergo, in order to obtain the desired final product. 16 Avrami M. Kinetics of phase change: II. Transformation-time relations for
random distribution of nuclei. J Chem Phys 1940;8:212–24.
17 Avrami M. Granulation, phase change and microstructure. Kinetics of
Acknowledgements phase change. III. J Chem Phys 1941;9:177–84.
18 Galwey AK, Brown ME. Thermal decomposition of ionic solids; vol. 86
This work was supported by Sibelco, which provided the of studies in physical and theoretical chemistry. Amsterdam: Elsevier;
samples for the research. An Acknowledgment is also due to an 1998.
19 Gualtieri A, Bellotto M, Artioli G, Clark SM. Kinetic study of the kaolinite-
anonymous referee who helped improving the paper.
mullite reaction sequence. Part II: Mullite formation. Phys Chem Miner
1995;22(4):215–22.
References 20 Dapiaggi M, Artioli G, Righi C, Carli R. High temperature reactions in
mold flux slags: kinetic versus composition control. J Non-Cryst Solids
1 Palmer DC. Stuffed derivatives of the silica polymorphs. In: Silica: physical 2007;353(30–31):2852–60.
behavior, geochemistry and materials applications; vol. 29 of Reviews in 21 Hu AM, Li M, Mao DL. Crystallization of spodumene-diopside in the
Mineralogy. 1994. p. 83–122. las glass ceramics with CaO and MgO addition. J Therm Anal Calorim
2 Gupta TK, Jau-Ho J. Origin of cristobalite formation during sintering of 2007;90(1):185–9.
a binary mixture of borosilicate glass and high silica glass. J Mater Res 22 Gualtieri AF, Gemmi M, Dapiaggi M. Phase transformations and reaction
1994;9(4):999–1005. kinetics during the temperature-induced oxidation of natural olivine. Am
3 Fei Y. Thermal expansion. In: Ahrens TJ, editor. Mineral physics and crys- Mineral 2003;88(10):1560–74.
tallography: a handbook of physical constants, vol. 2. 1995. p. 29–44 [AGU 23 Criado JM, Gonzales M, Real C. Correlation between crystallite size and
Reference Shelf]. microstrains in materials subjected to thermal and/or mechanical treat-
4 Njoya D, Hajjaji M, Njopwouo D. Effects of some processing factors ments. J Mater Sci Lett 1986;5:467–9.
on technical properties of a clay-based ceramic material. Appl Clay Sci 24 D’Incau M, Leoni M, Scardi P. High-energy grinding of FeMo powders. J
2012;6:5–66, 106–113. Mater Res 2007;22:1744–53.
5 Shackelford JF, Doremus RH. Ceramic and glass materials. New York: 25 Dos Santos DS, Dos Santos DR. Crystallization kinetics of Fe–B–Si metal-
Springer; 2008. lic glasses. J Non-Cryst Solids 2002;304(1–3):56–63.

You might also like