750

J. Appl. Cryst. (1997). 30, 750-754

Thermostimulated Sol-Gel Transition in Suspensions of Sulfate--Zirconium Oxyehloride

LEILA A. CHIAVACCI, a CELSO V. SANTILLI, a* SANDRA H. PULCINELLI a AND ALDO F. CRAIEVICH b'c
alnstitute of Chemistry, UNESP, Araraquara, SP, Brazil, bNational Synchrotron Light Laboratory, CNPq, Campinas,
SP, Brazil and Clnstitute of Physics, USP, Sro Paulo, SP, Brazil. E-mail: [email protected]
(Received 23 July 1996; accepted 29 January 1997)

Abstract Thermoreversible gels are two-phase systems which

The sol-gel transition of an aqueous solution of acquire elastic properties under temperature changes as a
zirconium oxychloride modified by sulfuric acid (Zr:S result of the formation of a connected network by
3:1) has been studied by small-angle X-ray scattering physical linkages (Djabourov, 1987). Thermoreversible
(SAXS) performed in situ during one cycle of heating gelation is often observed in biological systems and
and cooling between 298 and 358 K. The experimental mixtures of synthetic polymers, but this behaviour is not
SAXS curves exhibit three regions, at small, medium and usual for inorganic gels. To our knowledge, only one
high q values, characteristic of the Guinier, 'fractal' and article reports the formation of an inorganic thermo-
Porod regimes, respectively. The value of 5.5 A for the reversible gel, namely of aluminium and sodium
radii of the primary particles, obtained from the cross polyphosphate, which was attributed to the formation
over of the Porod and fractal regimes, is consistent with of a network of polycations and polyanions (Lima &
the size of the inner core of the polynuclear Galembeck, 1994).
Zr18Oa(OH)38.8(SO4)12.6.33H20 molecule. These mole-
In this paper we present the first SAXS study
cules aggregate as small clusters (31 A) of fractal concerning the thermoreversible sol-gel transition of
structure, with a dimensionality D = 2.16 in the sol. zirconium oxychloride modified by sulfuric acid. Struc-
The value of the fractal dimensionality, which is tural parameters, such as the fractal dimensionality and
characteristic of ideal branched polymers, decreases the upper and lower cut-off lengths, are determined from
during heating, attaining a value of 2.0, characteristic the SAXS results.
of swollen polymers. Above the critical temperature
(329 K), the average size of aggregates increases abruptly
2. The structure function for fractals
to 200 A and the fractal dimensionality decreases to D =
1.75, this value being characteristic of a cluster--cluster Many investigations have clearly established that the
aggregation process. structures of a number of gels and aerogels can be
described by the concept of fractals (Chaput, Boilot,
Dauger, Devreux & De Geyer, 1990; Freltott, Kjems &
1. Introduction Sinha, 1986; Teixeira, 1988). The structure function and
Zirconia-based ceramic materials have been obtained the SAXS intensity, I(q), which is proportional to the
through the sol-gel process, aiming at improving their structure function, exhibits the following remarkably
thermomechanical, electrical, optical and catalytic prop- simple behaviour:
erties (Bartlett, Woolfrey, Percy, Spiccia & West, 1994). I ( q ) = K q -~' (1)
Zirconia gels are prepared by hydrolysis and condensa-
tion of alkoxides such as zirconium n-propoxide or where q is the modulus of the scattering vector and a and
ethoxide. The zirconium alkoxide precursors are highly K are constants; the exponent c~ = D, where D is the
reactive and these reactions have to be controlled by fractal dimensionality of the mass fractal object. The
modif),ing the precursors using complexing ligands infinite q range for which (1) holds only applies to ideal
(Chatry, Henry, In, Sanchez & Livage, 1994). There is fractals having extremely large size and very small basic
some evidence (Henry, Jolivet & Livage, 1992) that units. In real systems, the potential dependence of I(q)
mineral complexing ligands, like SO4z-, PO43- and [(1)] is limited both at large and at small q values because
N O 3 - , can also control the condensation reactions of of the finite size or correlation length of the aggregates, ~,
aqueous zirconium solutions prepared from metal and because of the non-zero size of the elementary
halides. While studying the possibility of controlling particles, a, respectively. The effect of these cut-offs is
these reactions by adding a mineral complexing ligand, the appearance of two crossovers in the scattering curve:
we discovered that an aqueous solution of zirconium the first at q = 1/~, between the Guinier and the fractal
oxychloride in the presence of sulfate exhibits a regimes, and the second at q = 1/a, between the fractal
thermoreversible sol-gel transition. and the Porod regions.

1997 International Union of Crystallography Journal of Applied Crystallography

Printed in Great Britain - all rights reserved ISSN 0021-8898 1997
 LEILA A. CHIAVACCI et al. 751

The scattering intensity function l(q) is given by observed if the system is kept at temperatures higher than
the critical temperature for a short period of time (less
I(q)=~oP(q)S(q) (2) than 1 h).
where S(q) is the structure function, P(q) is the form
factor of the elementary units and 99 is the number of
scattering units per unit volume. 4. SAXS measurements
Freltoft, Kjems & Sinha (1986) and Teixeira (1988) The SAXS study was performed using the D24 SAXS
established the structure function for a fractal composed workstation at the synchrotron X-ray source DCI at
of elementary particles of size a having a cut-off for the LURE, France. The workstation is equipped with a bent
correlation represented by a function C(r) oc exp(-r/~). silicon crystal monochromator which yields a mono-
The dependence of S(q) on D, ~ and a is given by chromatic (~. = 1.49 A) and horizontally focused beam. A
set of slits defines the beam vertically. A position-
S(q) = C ( I + {[1/(qa) ] sensitive X-ray detector and a multichannel analyser were
x { [ O F ( O - 1)]/[1 + (1/q2es2)] (D-l)~2} used to determine the SAXS intensity I(q) as a function
of the modulus of the scattering vector q. Because of the
x sin[(D- 1)tan-l(q~)]}) (3) small size of the incident-beam cross section at the
where F is the gamma function and C is a constant. Note detection plane, no mathematical desmearing of the
that this expression becomes S(q) ~_ qD when ~-l << q experimental function was needed. The scattered in-
<< a -1. At small q, S(q) exhibits Guinier-type behaviour tensity was determined as the difference between the
(Chaput, Boilot, Danger, Devreux & De Geyer, 1990). experimental curve recorded with the sample and the
parasitic scattering curve obtained with the solvent
The form factor was approximated by a Gaussian
(water). An ionization chamber, placed downstream from
function assuming spherical-shaped elementary particles:
the sample, was used to monitor the intensity decay of
P(q) "" exp[-(1/5)qZa2]. (4) the transmitted beam and to determine the sample
attenuation. The natural decay intensity of the incident
Equation (3) holds for q < 1/a. For q >> 1/a the SAXS beam was monitored also by recording the electronic
intensity exhibits the classical Porod asymptotic beha- current in the synchrotron source.
viour. In situ SAXS measurements were carded out during
one cycle of heating and cooling (rate 1Kmin --1)
between 303 and 358 K. Each spectrum was recorded
3. Sample preparation during 150 s, which corresponds to a temperature rise of
The colloidal suspensions were prepared by adding an 1.5 K. The sample was put into a cell between two thin
aqueous solution of ZrOCI2 (0.2 mol 1-1, room tempera- films of Kapton, separated by 0.8 mm, placed in a Metier
ture) to a hot (353 K) aqueous solution of sulfuric acid FP82 'hot stage' to control the thermal cycle.
(0.2 mol l-l), drop by drop under magnetic stirring. The
volume of the ZrOC12 solution was adjusted to yield
suspensions containing the following Z r 4 + : H S O 4 - molar 5. Results and discussion
ratios: 1:1, 1:2, 1:3, 2:1, 3:1 and 4:1. Aliquots of 15 ml of The thermoreversibility of this system was evidenced by
the suspensions were put inside acetylcellulose mem- turbidimetric measurements carded out at different rates
brane tubing with molecular weight cut-off smaller than of heating and cooling, as shown in Fig. 1, in which
14 000, and then submitted to static dialysis against gelation during heating is characterized by an abrupt
200 ml of doubly distilled water during 24 h. Dialysis increase in the turbidity at the sol-gel transition
was carried out in order to decrease the ionic strength of temperature. During cooling, the decrease in turbidity
the suspensions by extracting the excess of ions; revealed the liquefaction, the temperature of which (gel-
consequently, peptization of the precipitate was favoured. sol transition) is always lower than the gelation
Peptization occurs only for samples prepared with molar temperature, resulting in a hysteresis loop. This beha-
ratios of 3:1 and 2:1, leading to the formation of viour is associated with differences in the mobilities of
transparent and stable suspensions, which gelify as the the small and large particles. The small particles in the
pH increases from 1.6 to 3.0. Elemental analysis of these sol have high mobility, favouring aggregate growth. On
samples showed that the nominal molar ratio was the other hand, in the large aggregates in the gel, the
preserved after dialysis. We have also noted that the interaction forces between the clusters make liquefaction
isocoric gelation may be obtained by increasing the (disaggregation) more difficult. It is noteworthy that fast
temperature of the sol prepared with a molar ratio of 3:1. rates of cooling prevent the system from returning to the
Moreover, the liquefaction of the system is observed as sol state (Fig. 1b). The in situ SAXS measurements were
the temperature decreases to values smaller than the carried out at 1 K min -~. In order to obtain complete
critical temperature. This thermoreversible behaviour is reversibility, a much slower cooling rate is necessary.
752 CONFERENCE PROCEEDINGS

This slow type of cycle was not tried because of the range of the linear region increases toward small q values
limited time available for the experimental run. for higher temperatures, indicating an increase in size of
Fig. 2 shows the log-log plot of the scattered intensity the fractal aggregates. A third regime is observed at small
I(q) as a function of the modulus of scattering vector q, q values where the scattered intensity is approximately
measured at different temperatures during sample heating independent of q, which is characteristic of noninteract-
(Fig. 2a) and cooling (Fig. 2b). For the region of high q ing clusters. This feature is expected for fractal clusters
(q > 0.05A -t) the curves measured at different with finite correlation range which behave as independent
temperatures are approximately overlapping and two objects.
power-law regimes can be distinguished. The first one, at Fig. 4 shows the fitting for the data displayed in Fig. 2
high q, presents a slope of - 4 , in agreement with the with the theoretical function I(q) given by (2) using S(q)
Porod law. This behaviour is confirmed by the plot and P(q) defined by (3) and (4), respectively. The
presented in Fig. 3, which shows that the product I(q)q 4 parameter a was determined from the crossover between
is a constant in the high q region, either for the sol or the the fractal and Porod regions. The other parameters of the
gel, indicating that the primary particles have well theoretical function were determined by a nonlinear least-
defined surfaces. The Porod limit, which is proportional squares refinement procedure. The theoretical equation
to the surface area of the interface, is nearly invariant gives, as expected, a good agreement with the experi-
during the sol-gel transition. This suggests that aggrega- mental observations only below 1/a.
tion occurs by simple branching with local contacts The q value of the crossover of the two power regimes
between the primary particles and without the contribu- remains approximately constant during the sol-gel-sol
tion of any density-increasing or phase-separation transition and corresponds to a = 5.5 A for the size of the
mechanisms, which would produce a noticeable variation primary particles. This value is in good agreement with
in the interface area. the radius of the inner core of a molecule of
The second power-law regime at medium q produces a ZqsO4(OH)38.s(SO4)t2.6.33H20 (Squattrito, Rudolf &
linear behaviour in the double logarithmic plots (Fig. 2). Clearfield, 1987), which is formed under conditions of
The slope is equal to - 2 . 2 at room temperature and synthesis similar to those used in the present study. In
decreases as the temperature is raised. Moreover, the q this structure, the inner core of the molecule is formed by

120-
T(k)
l gel-sol transition 357
~-.-. 80- 352
349
( + ) gel-sol transition

-'"~-~"-~
.... t
333
# 40- 319
sol-gel transition
~o 315

290
I I
300 310
!

Temperature (K)
320
I I
330
1o
Oo o o
o
ooo
OOo o
(b)
311

(a) .~ 0.1 T(k)

357
353
349 (a)
"-" 200 , 347
g 341
0.01 339 ( o ) sol-gel transition
336
100 / 331
326
317
306
........... ) . . J 0.001
0 I I I I I I I I

290 300 310 320 330 340 350 360

Temperature (K)
0.01 0.1
(b) q(A-~)
Fig. 1. Temperature evolution of the turbidity of the colloidal Fig. 2. Log-log plot of the experimental SAXS curves measured at
suspension of zirconium oxychloride modified by sulfuric acid different temperatures during (a) heating and (b) cooling. The set of
(Zr:S 3:1 and [Zr] = 0.15 mol 1-I) during a cycle of heating and SAXS curves corresponding to cooling are vertically displaced for
cooling at (a) 0.01 and (b) 1 K min-k clarity.
 LEILA A. CHIAVACCI et al. 753

hydroxy and oxo groups bridging Zr atoms, whereas the the fractal dimensionality from 2.2 to 2.0 and an increase
sulfate ligands are located at the surface of the in the cluster correlation length from 31 to 46/k as the
Zr(OH)xOycore. These molecules are held together by temperature increases from 298 to 328 K. However, the
electrostatic interactions with the remaining water change in ~ seems to be small when compared to the
molecules. large relative variation in the fractal dimensionality,
Fig. 5 shows the temperature evolution of the fractal suggesting that relaxation involves only small and local
dimensionality D (Fig. 5a) and the correlation length of structural modifications.
the aggregates ~ (Fig. 5b) during the sol-gel-sol At about 333 K, the size of aggregates, ~, increases
transition. Note that, during cooling, the values of both abruptly and the fractai dimensionality decreases from
D and ~ are not the same as during heating, indicating a 2.0 to 1.75 and remains almost constant up to 343 K. The
strong hysteresis effect. This was expected from the value of D of 1.75 is close to the value expected for the
turbidimetric measurements at the same heating rate. mechanism of random-walk cluster-cluster aggregation
Both sol and gel can be characterized by a fractal (Chaumont, Craievich & Zarzycki, 1992a,b, and refer-
structure model, the dimensionality D decreasing pro- ences therein). The enhanced mobility of the clusters and
gressively from 2.2 to about 2.0 in the sol phase and the change of the electrostatic interactions as the
remaining constant and equal to 1.75 in the gel phase. temperature increases may play a role in the modification
The value of D of 2.2 at room temperature is of the aggregation mechanism above the gelation
theoretically expected for ideal branched polymers, as temperature.
is D = 2.0 for swollen branched polymers (Chaumont, The thermoreversible behaviour observed for the
Craievich & Zarzycki, 1992a,b, and references therein). colloidal suspensions of zirconium oxychloride modified
This implies that in solution at room temperature, ideal by sulfuric acid is qualitatively similar to that observed in
branching occurs with monomers interlinked by a matrix some polymers in solution with lower critical solution
formed by water, similar to the crystalline structure of temperature (LCST), in which phase separation and
Z r 1 8 O a ( O H ) 3 8 . 8 ( S O n ) 1 2 . 6 . 3 3 H 2 0 . In solution, these inter- gelation occurs above a critical temperature. This
actions relax as the temperature increases due to the behaviour is often assigned to a decrease in the
diminution of the dielectric constant of water. This leads
to swelling of the cluster, which produces a decrease in
1.60

"--= 1.20
I o o o
o o o
o o o o
-~ 0.80 ogo o oo o o o

ao ~0
~ Oo o Oo
o n o o o o o

o o o o
,~ 0.40

0.00 I I

0 2 4 6
q4 10-3(]k .4)
(a)
'a

1.60
0.1 T(k)'
357
"-. 1.20 353
o 349
o
o 347
= 0.80 o o o %o00 o
341
q>o
o~ o
Ooq~
o ~
o o
o o 339
c~%o o o o % o oo o
0.01
336
0.40' 0 o o o
o,, 331
326
0.00' I I I I
317
0 2 4 6 8 306
q4 x 10 3(,~-4)
IE-3
(b) 0.01 0.1
Fig. 3. Porod plots corresponding to (a) the gel phase (T = 343 K) and q(A-b
(b) the sol phase (T= 303 K). In spite of the large statistical errors in
this q range, the Porod asymptotic behaviour seems to be obeyed for Fig. 4. Experimental points and fitting of the structure function S(q) for
both sol and gel. the SAXS curves. The curves are vertically displaced for clarity.
754 CONFERENCE PROCEEDINGS

effectiveness of the solvent, causing a decrease in the obtained atter ageing at 353 K for 3 h (Chiavacci, 1996).
volume entropy due to the density-increase mechanism Therefore, the thermoreversibility seems to be a
of the polymers (Sperling, 1992). Moreover, in the case consequence of an intermediate step involving the
of a transition from the solution to a phase-separated formation of a sulfate complex, in which the short-lived
structure, a significant variation in the surface area is nodes of the network are due to the electrostatic
expected. interaction between the clusters.
In contrast, our system shows swelling of the fractal
cluster with increasing temperature, indicating that the
effectiveness of the solvent increases. Moreover, the 6. Conclusion
Porod limit does not indicate any variation in the surface An aqueous solution of zirconium oxychloride modified
area (Fig. 3) during the sol-gel transition. Therefore, it is by sulfuric acid exhibits a thermoreversible sol-gel
difficult to explain our results from the thermodynamic transition, as evidenced by turbidimetric measurements.
concepts of LCST. This SAXS study, during a cycle of heating and cooling,
It seems more consistent to attribute the sol-gel-sol demonstrates that both the sol and the gel are formed by
transition to a change in the affinity between the solvent clusters or aggregates having a fractal structure. The size
and the clusters. In general, the enthalpy of hydration of the elementary structural unit is similar to that of the
becomes less negative as the temperature increases. As a inner core of the Zr1804(OH)38.8(SO4)12.6.33H20 mole-
consequence, the solvation tends to decrease during cule. The size of the basic building block does not vary
heating, favouring the formation of more efficient along the whole heating-cooling cycle.
electrostatic interactions between the ZrlsO4(OH)38.8 As temperature increases, the initial clusters of perfect
(504)12.6 molecules by means of sulfate bridging. branched polymers (with D = 2.2) in the sol phase
In fact, we have observed by IR spectroscopy the progressively swell (with D decreasing to 2.0).
presence of a bidentate SO4 ligand in the irreversible gel, Gelation is produced by the enhanced electrostatic
interaction between the clusters resulting in a network
with a fractal dimensionality D of 1.75, which is
consistent with the value expected for the mechanism
2.2- of cluster-cluster aggregation.

This work was supported by CNPq, CAPES/COFE-

2.0- CUB and FAPESP (Brazil).

References
1.8- Bartlett, J. R.,Woolfrey, J. L., Percy, M., Spiccia, L. & West, B.
O. (1994). J. Sol-Gel Sci. Technol. 2, 215-220.
I I I ! Chaput, F., Boilot, J. P., Dauger, A., Devreux, E & De Geyer, A.
300 320 340 360 (1990). J. Non-Co~st. Solids, 116, 133-139.
Temperature (K) Chatry, M., Henry, M., In, M., Sanchez, C. & Livage, J. (1994).
(a) J Sol-Gel Sci. Technol. 1,233-240.
Chaumont, D., Craievich, A. & Zarzycki, J. (1992a). J Non-
Co~st. Solids, 1471148, 41--46.
Chaumont, D., Craievich, A. & Zarzycki, J. (1992b). J. Non-
200 Co~st. Solids, 1471148, 127-134.
Chiavacci, L. A. (1996). Masters thesis, UNESP, Brazil.
gel-sol / J Djabourov, M. (1987). Rev. Gen. Therm. Fr. 3061307, 369-373.
o< Freltofi, T., Kjems, J. K. & Sinha, S. K. (1986). Phys. Rev. B,
100 33, 269-275.
Henry, M., Jolivet, J. P. & Livage, J. (1992). Structure and
sol-gel ,,--me
Bonding, Vol. 77, edited by R. Reisfeld & C. K. Jorgensen,
W------~-D'-O-IIt
pp. 153-203. Berlin: Springer-Verlag.
Lima, E. C. O. & Galembeck, E (1994). J. Collect. Interface
i i I I
300 320 340 360 Sci. 166, 309-315.
Temperature (K) Sperling, L. H. (1992). Introduction to Physical Polymer
Science, pp. 127-130. New York: Wiley Interscience.
(b) Squattrito, P. J., Rudolf, P. R. & Clearfield, A. (1987). Inorg.
Fig. 5. Temperaturedependence of (a) the fractal dimensionalityD and Chem. 26, 4240--4244.
(b) the size of aggregates ~, obtained from the fittings using (2). Teixeira, J. (1988). J Appl. Co~st. 21, 781-785.