414
Langmuir 1985,1, 414-420
dynamic shifta are positive, in agreement with the theory.I8
(6) T h e Pressure-Sensitive Shoulders at -21 10
(CO) a n d 1765 (NO) cm-’. As illustrated in Figure 9, the
shoulders at -2110 (CO) and -1765 (NO) cm-’ are clearly
observed only a t coverages near 0 = 0.5 and are present
also on samples mainly exposing (100)faces. Their frequencies are lower than those of CO and NO on (100)faces,
suggesting for these species a higher electron density on
the s* antibonding orbitals (and hence, following the usual
molecular orbital schemes of the metal carbonyls and
nitrosyls, stronger Ni2+-CO and Ni2+-NO bonds). This
hypothesis is, however, in contrast with their smaller
stability. We suggest that these peaks are associated with
a small fraction of CO and NO molecules adsorbed on
nearest-neighbor Ni2+ ions of the (100) faces and destabilized by repulsive interactions. The minimum Ni2+-Ni2+
distance on (l00l faces is 2.95 A, i.e., smaller than the van
der Waals distance: this implies that CO and NO molecules located a t such low distances (forming compressed
patches) should show some orbital overlap between adjacent molecules.
The most abundant formation of compressed clusters
in NO adlayers (in the vicinity of B = 0.5) is likely associated with a spin pairing process causing the stabilization
of dimeric forms (which of course is not operating for CO).
The whole subject of the hypothetical formation of such
compressed patches will be discussed in a subsequent
paper.25
(7) The Nitrite (NO2-) Species. As illustrated in
Figure 9, two types of NO2- species are formed by NO
adsorption on edges (steps) (monodentate and bidentab):
(25) Escalona-Platero, E.; Zecchina, A., unpublished results.
Their frequencies are influenced by the filling of (100)faces
through inductive effecs, in a way similar to that described
for the high-frequency nitrosyls a t 1740 cm-’. The concentration of the bidentate species definitely decreases
when the NO pressure decreases, while that of the monodentate species shows an opposite trend (and vice versa).
We interpret this reversible behavior as due to the interconversion of the two species caused by the changes of
electronic and steric factors brought about by the filling
of (100)faces. Very similar effects have been observed on
the NiO-MgO solid solutions and have received a similar
e~planation.~
Conclusions
The comparison of the IR spectra of CO and NO adsorbed on a series of NiO samples characterized by decreasing specific surface area and increasing perfection of
the crystallites allows assignment of the bands associated
with species adsorbed on extended (100)faces, on edges
and steps and atomic defects. The maximum CO and NO
coverage on (100)faces is only -0.5: this means that the
formation of carbonylic and nitrosylic adducts involves
only half of the Ni2+ions and that more compressed phases
are unfavored by strong adsorbate-adsorbate repulsions.
At 0 = 0.5, CO and NO form “in registry” well-ordered
surface phases (similar to those observed on metals), where
the CO and NO molecules interact via “dipole-dipole” and
chemical effects.
Acknowledgment. This investigation has been carried
out with the financial support of the Minister0 della
Pubblica Istruzione, Progetti di Rilevante interesse Nazionale.
Registry No. NiO, 1313-99-1;CO, 630-08-0; NO, 10102-43-9.
High-Purity, Monodisperse TiOz Powders by Hydrolysis of
Titanium Tetraethoxide. 1. Synthesis and Physical
Properties+
Eric A. Barringer* and H. Kent Bowen*
Materials Processing Center, Department of Material Science and Engineering, Massachusetts
Institute of Technology, Cambridge, Massacusetts 02139
Received November 15, 1984. I n Final Form: March 15, 1985
Spherical, monodisperse TiOz powders were formed through the controlled hydrolysis of dilute ethanolic
solutions of titanium tetraethoxide. Powders having average diameters in the range 300-600 nm were
synthesized by using ethoxide and water concentrationsof 0.1-0.2 M and 0.3-0.7 M, respectively. Induction
times for precipitation, obtained through measurements of the onset of turbidity, can be discussed in terms
of nucleation kinetics and the hydrolysis and condensation reactions. Chemical and physical properties
such as particle size, surface area, density (3.1 g/cm3),and crystal structure (amorphous to X-rays) were
measured. Of particular interest, the surface structure of the particles, determined through nitrogen
adsorption isotherms, varied from mesoporous to atomically smooth and was dependent on precipitate
washing procedures and suspension aging.
Introduction
~ i t a npowders,
i~
widely used in industrial
as pigments, opacifiers, photocat&sts, and fillers, have
been obtained either directly from titanium-bearing minerals or by precipitation from solutions of titanium salts
‘Research supported by DOE, Contract AC02-80ER10588.
0743-7463/85/2401-0414$01.50/0
or alkoxides. The most common procedures have been
based on the hydrolysis of acidic solutions of Ti(1V) salts.
In addition, gas-phase oxidation reactions of TiC141-3and
(1) Formenti, M.; Juillet, F.; Meriaudeau, P., Teichner, S. J.; Vergnon,
P. J. Colloid Interface Sci. 1972, 39, 79.
(2) George, A. P.; Murley, R. D.; Place, E. R. Faraday S y m p . Chem.
SOC.1973, 7, 63.
0 1985 American Chemical Society
Langmuir, Vol. 1, No.4,1985 415
High-Furity, Monodisperse TiO, Powders. 1
hydrolysis reactions of titanium alkoxides'n6 have been
employed to generate finely divided, high-purity T i 0 2
powders. However, these powders have generally lacked
the properties of uniform size, equiaxed shape, and unagglomerated state desired for quantitative studies of
colloidal phenomena.
Recently, several techniques have been reported for
synthesizing monodisperse powders through controlled
nucleation and growth proceases in dilute Ti(IV) solutions.
MatijeviE and co-workers studied the controlled acid hydrolysis of Tic4 and formed hydrosoLs of spherical anatase
and rutile particles.6 Visca and MatijeviE7 hydrolyzed
liquid aerosols to produce uniform, spherical anatase
particles from TiCI, and monodisperse, amorphous particles from titanium alkoxides. Yet, the oxide yield for
these processes was very low, and thus these techniques
are not amenable to subsequent property studies.
This paper presents a procedure by which monodisperse,
high-purity, spherical Ti02particles are formed through
the controlled hydrolysis of titanium tetraethoxide. T h e
procedure, the physical properties of the powders, and the
effects of reagent concentration and aging on these prop
erties are discussed. Because the surface structure strongly
influences the interfacial electrochemical properties and
dispersion stability, the topic of part 2 of this research,
nitrogen adsorption behavior is analyzed in detail. Finally,
the hydrolysis and condensation reactions are discussed
in terms of particle nucleation and growth kinetics.
Experimental Section
I. Powder Synthesis. The TiO, powders were prepared by
hydrolysis of dilute ethanoticsolutions of titanium tetraethoride,
Ti(OC,H,),. Two hundred proof anhydrous ethanol and the
ethoride (Alfa Products, Danvers, MA) were used without further
purification: deionized water was prepared by using a Millipore
Continental Water System (Millipore Corp., Medford. MA). Since
homogeneous nucleation of particles was desired. the ethanol and
water were ultrditered through 0.22-pm pore size Millipore fdters
to minimize the level of insoluble impurities. To prevent water
contamination of the ethanol and the ethoxide, all work was
conducted in a glovebox having a dry atmosphere (<lo0 ppm
H20).
For a given synthesis experiment, the ethoxide was dissolved
in ethanol and water in a separate portion of ethanol (-onehalf
of the total volume). The water solution was poured into the
ethoxide solution and rapidly mixed using a magnetic-driven
stirrer, giving a solution with concentrations in the range of 0.1-0.2
M ethoxide and 0.3-0.7 M water; the molar ratio of water to
ethoxide was always greater than 2.5. After mixing for 2-5 8, the
stirring rate was reduced to a minimal level to minimize coagulation of the titania particles during the growth stage of the
renction. Precipitation of amorphous, hydrated powders oocurred
in 2-90 s (T= 298 K);the time for precipitation, measured with
a stopwatch, was the time between solution mixing and the first
visual observation of turbidity. Because of finite mixing times,
the probable error in the measured time was *0.5 8. Slow stirring
was continued for an additional 10 min to provide some equilibration time.
The TiO, almsoh were transferred out of the glovebox and the
powders were washed with deionized water by repeated centrifugation (4wOg force) and ultrasonic (probe) dispersion cycles,
replacing the supernatent with deionized water prior to each
dispersion step; four cycles were usually performed. The powders
were then ultrasonically dispersed in the desired aqueous solutions.
To prevent agglomerate formation in the alcohol by dissolu(3) Suyama. Y.; &to. A. J . Am. Cerom. Soe. 1976.69.146.
(4) Mazdiyasni, K. S.Cerom. Int. 1982,s. 42.
(5) Yan. M.; Rhcdea, W.W.:Springer, L.Am. Cemm. Soe. Bull. 1982,
61,911.
(6) MatiieviE, E.;Budnick, M.; Meites. L. J. Colloid Interface Sei.
1977,61,302.
(7) V i .M.; MatijsviE. E J . Colloid Interface Sei. 1979, 68. 308.
r t44
I oym
Figure 1. TEM micrograph of a powder syntheaized using 0.15
M Ti(OC2H5),and 0.5 M H a the particles are spherical and have
a narrow size distribution.
tion/reprecipitation pmmawu, the washing pmeedure was initiated
within 30 min of alcoaol removal from the glovebox. Although
this technique was effectivein removingthe alcohol and unreacted
alkoxide, it generated an undesired result. The initial contact
of water with the residual ethoxide on the particle surface caused
the formation of a surface coating of very small 6-10 nm),
crystalline precipitates. As will be discussed. this coating dominated the interfacial properties of the powder.
Alternate washing techniques using ethanol and various
water/ethanol solutions were attempted to avoid the precipitation
problem; however, they were generally not effective. Two ethanol
washes, followed by vacuum drying, and then water washing
prevented the surface precipitate formation. Unfortunately, the
powders could not be completely redispersed (without agglomerates), thus they were useful only in the physical property
characterization studies.
2. Physical Pmpertiea. Physical properties were determined
for all powders. Powder density was measured using a stereo.
pycnometer (QuantachromeCorporation,Syoeset,NY)and helium
gas. Specific surface areas were measured by multipoint BET
(Quantachrome Corporation),using N2gas as the adsorbate and
assuming a cross-sectional area of 0.162 nm2 per molecule.
Powders were vacuum outgas4 for >14 h at 30 "C;this procedure
was found by Yates8 to be effective in removing physically adsorbed water, but not chemisorbed water.
Assessments of particle shape, state of agglomeration, and
qualitative size distributionswere obtained from scanning electron
micrographs. Quantitative number distributions were obtained
for >500 particles from TEM micrographs by using a histogram
method? Particle substructure was investigated by transmission
electron microscopy and X-ray diffraction. Semiquantitative
impurity and carbon analyses of water-washed powder showed
a low cation impurity level and low residual carbon content (0.005 mol fraction ethanol). To confirm the purity, quantitative
chemical analpis was performed using inductively coupled plasma
emission spectroscopy (performed by W. Zamachek, Union
Carbide, Tarrytown. NY); calcium at 40 ppm and silicon at 80
ppm were the only cations detected.
3. Nitrogen Adsorption and Surface Structure. The
presence and type of porosity at the powder surface were determined by measuring complete nitrogen adsorption/desurption
isothem and analyzing them by using the aimethod of Si.""
By use of this technique, porosity was assessed by plotting the
adsorption isotherm for a powder in a reduced form; the volume
of adsorbed N, at a given relative pressure (PIP& V, was plotted
against V,/V. (=a,),where V. was the volume adsorbed by a
nonporous reference powder at the same PIP, and V. was the
volume adsorbed by the reference at a selected value of PIP*
Although u, = 1can be put at any relative pressure, it is usually
located at PIPo = 0.4 because both monolayer coverage and
micropore filling occur for PIP, < 0.4, whereas condensation and
adsorption/desorption hysteresis occur at PIP, > 0.4. Furthermore, high precision is achieved by locating a. = l in the
middle of the range of relative pressures." The major consideration in using the a.-method is the choice of the referencesolid;
(8)Yates, D. E PhD. Thesis, Univeraity of Melbourne, Australia,
1975.
(9) Men, T." P d d eS i Marnusmsat: 3rd ed,C h 4 p " d Hall:
London, 1981.
(10) Sing, K.S. W.Chem. lnd. (London)1%7,829.
(11) Sing, K. S. W.Chem. Ind. (London)1W8, 1520.
416 Langmuir, Vol. 1, No. 4, 1985
250
Barringer and Bowen
c
TiO,
C
0
Anatase Reference
nI
I
200
&
13
150
E
3
=
too
50
01
02
03
04
05
06
07
Diameter (pm)
Figure 2. TEM-histogram size distribution for the TiO, powder
shown in Figure 1. The mean diameter is 390 nm and the geometric standard deviation is 1.09.
Table I. Powder Properties as a Function of Initial
Reagent Concentration (T = 25.5 "C)
mean
number
induction
[Ti(OC2H5)4], [H20], size,
%
density,
time,
M
M
nm yield
cm-3
9
0.1
0.1
0.1
0.1
0.1
0.15
0.15
0.15
0.15
0.2
0.2
0.2
0.3
0.4
0.5
0.6
0.7
0.4
0.5
0.6
0.7
0.5
0.6
0.7
500
440
380
350
310
470
430
380
340
490
440
370
30
47
54
72
74
70
78
-83
-84
>80
>80
>80
1.2 X
2.6 X
4.8 X
8.3 X
1.2 x
4.5 X
6.8 X
1.0 X
1.5 X
1O'O
1O'O
1O'O
1O'O
10"
1O'O
1O'O
10"
10"
-69 (k0.5)
25
12.5
7
4.5
15
1.8
4.7
3
5
3
2
an anatase powder, obtained by heating the precipitated powder
for 20 min at 620 "C, was determined to be an acceptable non-
porous reference.
Specific surface areas and complete adsorption/desorption
isotherms were determined for four powders: the anatase powder,
an alcohol-washed powder, and two water-washed powders. The
first water-washed powder, designated TiOzA, was precipitated,
water washed, and then air dried. The second powder, designated
TiO, B, was precipitated, water washed, aged in water as a dispersion for 3 months, and then air dried.
Results and Discussion
1. Synthesis and Physical Properties. Figure 1
shows an example of a TiOz powder synthesized using 0.15
M Ti(OC,H,), and 0.5 M H,O; the particles are spherical
and have a narrow size distribution. The surface coating
of fine (5-*,C)nm) precipitates was the result of the water
washing prucedure (discussed later). Figure 2 gives the
TEM-histogram analyzed size distribution for the powder;
the mean diameter was 390 nm and the geometric standard
deviation was 1.09. Dilute TiOz sols were sufficiently
uniform t o exhibit higher order Tyndall spectra and dispersions of some powders settled to form iridescent ordered
arrays (face-centered cubic particle packing). These
powders are the most uniform titania powders reported
in the literature, and the synthesis technique provides a
route to form sufficient quantities of material for further
investigations.
Synthesis data are given in Table I for powders formed
for a matrix of initial reagent concentrations a t T = 25.5
f 0.2 "C. The approximate yields, determined by dry
weight analysis (vacuum dried at 80 "C), were based on
Figure 3. Adsorption isotherms for the anatase reference powder
and an amorphous TiO, powder. Open points are for the adsorption isotherm and solid points are for the desorption isotherm.
the total possible TiOz from the ethoxide, assuming 100%
conversion. The particle number densities were calculated
from the mean size and yield, assuming that only singlets
were present. The data show that the mean size decreased
as the water concentration was increased for fixed ethoxide
concentration and increased as the ethoxide concentration
was increased for fixed water concentration. The yields
increased as either reagent concentration was increased.
Although quantitative size distributions were not obtained for the powders given in Table I, SEM observations
showed a slight increase in distribution width as the concentration of either reagent was increased. Reaction kinetics (Le., multiple nucleation) may have contributed to
the increased width; however, the increased number of
doublets present indicated enhanced coagulation. Particle
coagulation was a significant problem for TiOz synthesis
under the nominally neutral conditions (near the IEP),
especially in early experiments where rapid stirring rates
were employed. The number of doublets and triplets
present in the final dispersion was high for rapid stirring
rates but decreased significantly as the rate during precipitation was reduced. For the reaction series present in
Table I, more rapid stirring rates were required for the
higher reagent concentrations (short reaction times) to
ensure complete mixing. This probably accounts for the
increased distribution width. In addition, the increase in
particle number density with increasing reagent concentrations (Table I) contributed to the increased coagulation
during the growth phase of the reaction. Attempts to
enhance stability by adding acids or bases catalyzed the
hydrolysis reaction and resulted in the uncontrolled precipitation of wide size distribution powders and/or the
formation of gels.
The average measure density for the TiOz powders was
3.1 f 0.1 g/cm3; no variation in density with reagent
concentration and washing procedure was observed within
experimental error. This density was considerably lower
than the densities for anatme (3.84 g/cm3) and rutile (4.26
g/cm3).I2 Within the detection limit for X-ray diffraction
analyses of approximately 1wt % crystallized, the powders
appeared to be amorphous; however, TEM observations
indicated that the surface precipitates were crystalline.
Although the electron diffraction patterns were difficult
to index due to the low number of spots comprising the
rings, the surface precipitates on TiOz A (fresh powder)
(12) "CRCHandbook of Chemistry and Physics''; 56th ed.;CRC Press:
Boca Raton, FL, 1975; pp B-153-154.
Langmuir, Vol. I. No.4, 1985 417
High-Purity, Monodisperse TiO, Powders. 1
Table 11. Nitrogen Adsorption Isotherm nnd a, Values for
the Anatase Refer"
p/pQ
0.03
0.05
0.1
0.15
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Solid
V,c m S h
4
1.127
1.217
1.401
1.461
1.651
0.495
0.534
0.615
0.644
0.725
0.838
1.0
1.122
1.242
1.411
1.117
2.488
1.908
2.278
2.554
2.828
3.351
4.048
5.667
Figure 5. Adsorption (open pointa)/desorption (solid points)
isotherms for Ti02A, Ti02 B,and the amorphous powder. The
hysteresis in the isotherms for Ti02A and Ti02 B indicate capillary condensation.
Fmure 4. TFN micrograph of a water-washed powder (fresh),
designated TiO, A, with a surface coating of fine, crystalline
precipitates.
seemed to consist of a mixture of anatase and rutile,
whereas the surface coating on TiOz B (aged powder)
consisted only of anatase.
The nitrogen adsorption isotherms for the anatase and
amorphous powders are given in Figure 3; the adsorption
data and the a,values for the reference anatase are given
in Table 11. 'The adsorption isotherm for the anatase
agrees well with the isotherms obtained by Parfrtt et aI.I3
for several anatase powders, and the BET surface area of
5.4 mz/g yields an equivalent spherical diameter that is
within experimental error of the actual diameter; thus the
anatase powder is an acceptable nonporous reference. The
isotherm for the amorphous powder appears similar to the
anatase isotherm and the specific surface area of 8.1 mz/g
is less than twice that expected for 400-nm-diameter
particles. These data indicate that the surface of the
amorphous particles are slightly rough but have no porosity
accessible by nitrogen. This does not preclude the possibility of surface microporosity for the powder in the
solvated and dispersed state that either is too small for
nitrogen penetration or is decomposed and sealed during
the drying process, as has been observed for amorphous
iron oxyhydroxides."
The specific surface areas of water-washed powders
ranged from 250 to 320 m2/g. These high values were the
result of the surface coating of fine (5-10 nm), equiaxed
precipitates and were independent of the reaction concentrations. Figure 4 shows an example of such powder,
TiOz A, having a specific surface area of 304 mz/g; small,
well-defined precipitates coat the particle surfaces. The
adsorption/desorption isotherms and the a.plot, given in
Figures 5 and 6, respectively, reveal the presence of both
micropores and mesopores. From the isotherm hysteresis,
the radius, rK. of the mesopores causing capillary condensation can be calculated by using the Kelvin equation:
-2YVm
In P / P o = -cos e
RTrK
(13)
P d % 0.D.: U h . D.;Wissman, T. J. J. Colloid Interface Sei.
IWl,36,211.
(10 Davia. J. A.; Laekie. J. 0. J. Colloid Interface Sei. 1978,67,90.
0
0
10
Amorphous T i 4
2.0
30
as
Figure 6. The a, plots for the amorphous powder, TiO, A, and
TiOz B. The positive deviation from linearity observed for TiO,
A and TiO, B confirms capillary condensation. The micropore
volume, V ,for Ti02 A is given by the intercept V', where V, =
0.001 56 fl.
where y is the surface tension of nitrogen at 77 K (=8.85
erg/cmz), V, the molar volume of nitrogen (=34.7 cm3/
mol), and 8 is the wetting angle (usually assumed to equal
zero).I5 Hysteresis for TiOz A in the range 0.35 < P / P o
< 0.75 corresponds to pore diameters ranging from 1.0 to
6.5 nm,which agrees well with the scale of observed surface
structure (Figure 4).
However, as expected, a more complicated pore structure
is indicated in the a. plot; microporosity leads to a second
linear region at high relative pressures. A similar a. plot
was reported by Bhambhani et a1.16 for a mixture of SiOa
powders, one having significant microporosity, but no other
detectable porosity, and the second having only mesopcrosity. Thus, both micro- and mesoporosity are present
on the TiOz A surface. The micropore volume, V,, calculated from the intercept, V', by using'O
(15) Qmueorb Operation Manual.Qunntachrome Corp.. S-1,
NY. 1981. pp 50-51.
(16) Bhambhmi, M. R: Cuning. P.A,: Sing, K.S.W.:Twk.
Colloid lnrorloce Scr. 1972.38, 109.
D.H.J.
418 Langmuir, Vol. 1, No. 4, 1985
Barringer and Bowen
V, = 0.00156V’
(2)
was 0.19 cm3/g (STP), which agrees well with the estimated pore volume assuming a uniform coating of 10 nm
precipitates on a 400-nm-diameter spherical particle.
Hence, the measured porosity may be attributed to capillary condensation at the precipitate/amorphous surface
contact points and the interprecipitate regions.
The adsorption/desorption isotherms and the a.plot for
the aged powder, TiO, B, are also given in Figures 5 and
6, respectively. Aging for 3 months in water caused a
reduction in the specific surface area to 186 m2/g and a
coarsening of the surface precipitates to form a rough,
ill-defined surface, as shown in Figure 7. The hysteresis
in the isotherms in the range 0.45 < P/Po < 0.9 corresponds (using eq 1)to mesopores with diameters ranging
from approximately 1.5 to 18nm; again, the pore sizes scale
with the observed surface structure (Figure 7). The a. plot
(Figure 6 ) confirms the presence of mesoporosity and also
the absence of any microporosity. Although surface
structure coarsening is evident, no measurable change in
particle size distribution was observed. Hence, a local
dissolution-reprecipitation process may be operative.
2. Hydrolysis a n d Condensation Reactions. Table
I also gives the measured time for observed tubidity, r, as
a function of reagent concentrations. For the slowest reactions (i.e., lowest concentrations), the slow increase in
observed turbidity caused an appreciable error (-30%)
in the measured time for initial turbidity. However, for
the faster reactions ( r 5 30 s) the turbidity increase was
very rapid; the solutions went from clear (-100% transmission) to turbid (-0% transmission) almost instantaneously. For the calculated number densities and the
reaction vessel used (5-cm path length), this transition
occurred as particles grew from less than 20 nm to greater
than 100 nm? thus indicating a rapid growth process. As
a result, the time required to observe initial turbidity may
be assumed a gocd indication of the induction time for
particle nucleation. By use of these data and evidence
from the literature, a possible reaction mechanism can be
identified.
The overall precipitation reaction for the Ti02 particles
consists of hydrolysis and condensation steps, schematically given a d 7
.
ROH
+
hydrolysis
Ti(OR)4 4H20 Ti(OH), + 4ROH
condensation
Ti(OH)4 TiOZ(s) 2H20
where R = GH,. The hydrolysis reaction has four possible
steps:
1. Ti(OR)4 + H20 Ti(OR),(OH) ROH
-
2. Ti(OR),(OH)
+ H20
3. Ti(OR),(OH),
+ HzO
+
+
Ti(OR),(OH), +ROH
-
Ti(OR)(OH), +ROH
lTi(OR)(OH)sI[ROHl
’3
-
= [Ti(OR)z(OH)2][H20]
4. Ti(OR)(OH),
(17) Lhino, T.;
Minami.
+ H20
Ti(OH),
+ ROH
S. Tech. Rep. Oaoka Uniu. 1953,3,357.
Figure 7. TEM micrograph of powder aged 3 months in water
(Ti02B). The surface precipitates have coarsened to form a rough
ill-defined surface.
where the Bs are the stepwise equilibrium constants. The
overall equilibrium constant for the hydrolysis reaction
through step j , K ,is formed as the product of the stepwise
constants ( K , =
The condensation reaction is the mode by which the
hydrolysis species polymerize to form higher molecular
weight products (nuclei and, subsequently. particles).
Condensation proceeds either through a water elimination
reaction
Ti-OH + HO-Ti
Ti-O-Ti + H 2 0
(4)
or an alcohol elimination reaction
Ti-OR + HO-Ti
Ti-O-Ti + ROH
(5)
fi!m,s,).
-
+
Although the reaction rate for eq 4 is expected to be faster
than that for eq 5, the dominant mechanism depends on
the relative rates of the hydrolysis and condensation reactions and the number of hydroxyl groups on the major
hydrolysis species.
Winterts and BoydIg studied the hydrolysis behavior of
titanium tetrabutoxide as a function of water to butoxide
molar ratio. For ratios 52.5 a variety of mixed (oxide
butoxide) polymers were formed, while TiO, precipitated
for ratios 23.19 Both investigators concluded that the
fourth butoxy group was difficult to remove and that water
was adsorbed into the polymer in addition to being consumed during hydrolysis. Ishino and Minami” also studied
the hydrolysis of Ti(OBu), in butanol and concluded that
hydrolysis was fast and condensation was the slower step.
thus the overall reaction apparently had an equilibrium
state with respect to temperature and initial water concentration. Again, removal of the fourth butoxy group was
found to be difficult.
The kinetic interpretations for the results of the previous
studies are questionable because of the simultaneous
presence of other processes (e.g., evaporation). However,
Bradley et al.2°.21found similar hydrolysis behavior for
Ti(OC,H,), in ethanol. For dilute ethoxide solutions and
low molar ratios of water (<1.5), trimeric ethoxide units
cross-linked to form several mixed polymers, but at higher
water concentrations the ethoxide behaved more as dimers.
Water concentrations >1.7 molar ratio yielded colloidal
precipitates; these precipitates were TiO, for ratios 22.5.
Unfortunately, the investigations were conducted for low
water concentrations and in boiling ethanol (78-80 “C);
thus, the results cannot be extrapolated to the conditions
employed in the present study.
The TiO, powders synthesized in this study are assumed
to form by homogeneous nucleation and growth proteases,
schematically represented in Figure 8 (after Overbeek?.
Significant nucleation occurs when the concentration of
(18) Winter. C. J. Oil Colour Chem. Amx. 1963.36.689: lY63, 3%695.
(19) Boyd, T. J . Polym. Srr. I W , 7,591.
(20) Bradley, D.C.; Care. R; Wardlew. W. J . Chem. Sor. 1 Y U . 721.
(21) Bradley. D. C.; Cam. R.; Wsrdkw. W. J . Chem. Sor. 1166. S977.
(22) Overbeak,J. ‘Ih.C. In ‘Colloidal Dispnions’; Goodwin. J. W..
Ed.; The Royal Society of Chemistry. London. 1982: pp 1-21.
High-Purity, Monodisperse TiOz Powders. 1
Langmuir, Vol. 1, No. 4, 1985 419
I #-•
L
I I
Ti me
Figure 8. Schematic representation of the homogeneous nucleation and growth process assumed for the alkoxide hydrolysis
(precipitation) reaction (after OverbeekZ2).
E: /
I
1
6
5
4
3
2
[Ti(OC,H,),]
Figure 10. Delay time for the initial observation of turbidity as
a function of initial ethoxide concentration for the initial water
concentrations of 0.5,0.6, and 0.7 M. The average slope is 1.23
--!-
&
0.1.
6, may represent the accumulation rate of a species having
the formula Ti(OR)(OH),, for which the corresponding net
hydrolysis reaction is given by
40
Ti(OR)4
+ 3H20
-
Ti(OR)(OH), + 3ROH
where the equilibrium constant (K3=
K3 =
108
6
2
4
[HZOI
Figure 9. Delay time for the initial observation of turbidity as
a function of initial water concentration for the initial ethoxide
concentrations of 0.10, 0.15, and 0.20 M. The average slope is
2.96 f 0.1.
the rate-limiting hydrolysis species, C, becomes greater
than a critical supersaturation concentration, C*; the nuclei
that form grow rapidly toward their final size. Since
particle nucleation approximately coincides kith the initial
phase of the condensation reaction, the induction time
reflects the rate a t which critical supersaturation is
achieved. Hence, the observed induction times can be
directly correlated to the hydrolysis reaction, for which
supersaturation appears to be the rate-limiting process.
Figure 9 shows the plot of log T vs. log [H20],,, for the
three ethoxide concentrations; the average slope is 2.96 f
0.1. The equivalent plot for log [Ti(OC2H5)4]initial
yields
an average slope of 1.23 f 0.1 (Figure 10). These slopes
lead to the following approximate rate equation for the
formation of the rate-controlling hydrolysis species (R =
C2H5):
rate = k[H20]3[Ti(OR)4]
(6)
Although the formation reaction for the rate-controlling
species consists of several steps (eq 3), an apparent
equilibrium is obtained. Thus, the kinetic equation, eq
n~=,pi)
is
(7)
[Ti(0R)(OH),] [ROHI3
(8)
[Ti(OR)d[H2013
The above equations have been written for reactions
involving monomeric species, yet the actual species present
may exist as dimers or trimers.20*21Nevertheless the
stoichiometry expressed in eq 7 would still apply. In addition, condensation was assumed to be insignificant
during hydrolysis; however, the observed reaction order
for the ethoxide of 1.23 indicates that some simultaneous
condensation may have occurred. The presence of this
condensation and its effects on the above analysis cannot
be assessed without more extensive kinetic data.
Equation 7 implies that the fourth alkoxy group remains
in the hydrolysis species, in agreement with previous obser~ations,"-'~which reflects the effect of the solvent on
the reaction equilibrium. Yet this group can be eliminated
during the condensation reaction. The reaction describing
the condpnsation (polymerization) of the hydrolysis species
to form the hydrated oxide is
Ti(OR)(OH), Ti02.xHz0 (1 - x)H,O ROH (9)
-
+
+
In this reaction both water elimination and alcohol elimination reactions, given in eq 4 and 5, occur. The overall
reaction, given by the sum of eq 7 and 9, becomes
Ti(OR)4 + (2 + x)H20 TiOz.xH20 + 4ROH
(10)
-
The value of x for the water of hydration is probably between 0.5 and 1.0, based on the thermal gravimetric
analyses, which showed a residual water content of approximately 10-20 wt %. Since the TGA weight loss depended on the drying and outgassing history and since the
technique could not differentiate between surface and
internal water, no exact value of x can be assigned.
However, the open structure in the amofphous particles
allows coordinative bonding of the eliminated water and
ethanol to the Ti atoms inside the particles. The low
carbon content of the TiOz powder, measured to be approximately 0.18 wt % (0.005 mol fraction C2H50H),
supports the conclusion that the groups on condensing
Langmuir 1985, 1, 420-428
420
hydrolysis species (eq 9) were primarily hydroxyl groups.
Thus, some water was trapped inside the particles and a
value for x of about 0.5 seems reasonable.
The narrow particle size distributions observed for the
spherical Ti02 powders seem to be the result of a single
nucleation regime precipitation process (Figure 8). Upon
nucleation, C drops below C*, and as long as the growth
(condensation) reaction removes the hydrolysis species
faster than they are generated, then C remains less than
C*. Hence, nuclei grow to their final size and few additional nuclei are formed. In addition, the size distribution
for nuclei generated in the finite nucleation regime becomes insignificant as the particle size increases toward
the final value (a 400 nm).
-
Summary and Conclusions
(1)High-purity, uniform-size, spherical titania powders
were synthesized; the average particle size decreased as the
initial concentration of water was increased but increased
as the initial concentration of ethoxide was increased. The
synthesis procedure gave the most uniform titania powders
reported in the literature. In addition, the reaction and
technique were scalable to sufficiently large quantities to
allow further investigation.
(2) The particle surface structure was found to be dependent on the washing procedure and dispersion aging
in water. Fresh, water-washed powders had an amorphous
interior but were coated with fine (5-10 nm), crystalline
precipitates as a result of water washing. The surface areas
of these powders
= 300 m2/g) decreased upon aging
to less than 200 m2/g. The coarsened-surface precipitates
had an anatase crystal structure.
(3) Particles were assumed to form by a homogeneous
nucleation and growth process. A simple kinetic analysis,
based on precipitation data, yielded the approximate rate
equation (R = C2HJ
rate = k[Ti(OR),] [H20l3
This equation was assumed to represent the accumulation
rate of the rate-controlling hydrolysis species for which
supersaturation was required for particle nucleation. The
corresponding hydrolysis and condensation reactions for
this species were
Ti(OR), + 3H20 Ti(OR)(OH), 3ROH
and
Ti(OR)(OH),
-
-
+
Ti02.xH20 + (1- x)H20 + ROH
where x was given as 0.5. Although quantitative nucleation
and growth studies were not conducted, the data and observations appear to be self-consistent. A more detailed
kinetic study is required to further illuminate the precipitation process.
Registry No. TiOz, 13463-67-7;Ti(OCzH5),,3087-36-3.
High-Purity, Monodisperse TiOa Powders by Hydrolysis of
Titanium Tetraethoxide. 2. Aqueous Interfacial
Electrochemistry and Dispersion Stability?
Eric A. Barringer* and H. Kent Bowen*
Materials Processing Center, Department of Materials Scieuce and Engineering,
Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
Received November 15, 1984. I n Final Form: March 15, 1985
Surface charge density, {-potential, and stability ratio were measured for high-purity, monodisperse TiOs
powders, prepared by hydrolysis of titanium tetraethoxide, in aqueous HC1 eIectrolyte solutions as a function
of pH. The surface electrochemistry was significantly affected by a precipitate washing procedure and
suspension aging; fresh, water-washed powders had a point of zero charge (PZC) of 4.0 and an isoelectric
point (IEP) of 4.3, whereas the PZC and IEP shifted to 5.2 and 5.5, respectively, upon aging for 3 months
in water. The enhanced surface acidity relative to rutile (IEP = 6.0) is attributed to the poorly crystallized
surface of the water-washed precipitated powders. Aging caused 'recrystallization of the surface region
to coarsened anatase crystallites. The aqueous electrochemistrywas &@ by the electrolyte complexation
double-layer model developed by James et al.;3JsJ7an extremely high inner capacitance is required to obtain
agreement. The high capacitance is indicative of a highly porous surface. ,Stability ratios for aged powders,
determined by photon correlation spectroscopy,26showed excellent correlation with measured {--potentials;
a Hamaker constant, AIzl, of 2.0 X 10-20 J was obtained by using the b r i n k and OverbeekZ method. Finally,
electrophoretic mobility and stability measurements as a function of pH in aqueous BaC1, electrolyte solutions
indicated strong specific adsorption of Bas+ ions, in agreement with the results of Feurstenau et al.41
Introduction
The behavior of aqueous colloidal dispersions of inorganic oxides is of great importance in the colloid and ceramic sciences and has been the subject of numerous investigations. In particular, the relationships between in'Research supported by DOE, Contract AC02-80ER10588.
terfacial electrochemistry, electrokinetic potential, and
dispersion stability have been studidd for colloidal SiOz,
Ti02, Fe203, A1203,and others.l-15 Surface electrochem(1) Wiese, G.R.;Healy, T. W. J.Colloid Interface Sci. 1975,51,427.
(2) Berube, Y. G.; de Bruyn, P. L. J. Colloid Interface Sci. 1968,27,
305.
( 3 ) James, R. 0.;Parks, G . A. Surf. Colloid Sci. 1982,12, 119-216.
0743-7463/85/2401-0420$01.50/0 0 1985 American Chemical Society