A kinetic study of the reactions of sulfate radicals at the silica nanoparticle - Water interface

Article abstract of DOI:10.1021/jp027318t

Sulfate radicals, SO₄^.-, were generated using flash photolysis of aqueous S₂O₈^2- solutions and the reactions of the inorganic radicals with the surface of suspended silica nanoparticles (NP) investigated. In the presence of colloidal silica no absorption traces due to SO₄^.- radicals are observed at 100 ??s after the flash of light. However, two transient species with absorption maxima around 320 and 600 nm are formed. A kinetic analysis of the experimental results indicate that SO₄^.- radicals are adsorbed on the NP surface, leading to the formation of an adduct, with ??_max a?? 320 nm (?μ a?? 7000 cm^-1 M^-1), and showing similar reactivity to that observed for the sulfate radical in aqueous solutions. The NP - sulfate radical adducts react with adsorbed water, and with single and geminal SiO^- sites with reaction rate constants of 1.5 ?? 10¹⁴ ?? e^{-(58?±12)kJ/mol)RT} s^-1, < 10³ ?? e^{-(2?±17)kJ/mol/RT} s^-1 and < 10¹¹ ?? e^{-(46?±13)kJ/mol/RT} s^-1, respectively. Two different SiO^. surface defects, showing similar spectra (??_max a?? 600 nm) but different reactivities, are formed from the reaction of NP - sulfate radicals and deprotonated geminal and single silanols.

Full text of DOI:10.1021/jp027318t

J. Phys. Chem. B 2003, 107, 6131-6138
6131
A Kinetic Study of the Reactions of Sulfate Radicals at the Silica Nanoparticle-Water
Interface
†
†
‡
,†,§
Paula Caregnato, Ver o´ nica C. Mora, Galo Carrillo Le Roux, Daniel O. M a´ rtire,* and
M o´ nica C. Gonzalez*,
†,⊥
Instituto de InVestigaciones Fisicoqu ´ı micas Te o´ ricas y Aplicadas (INIFTA), Facultad de Ciencias Exactas,
UniVersidad Nacional de La Plata, C. C. 16, Suc. 4, (1900) La Plata, Argentina, and Escola Politecnica da
USP-Departamento de Engenharia Quimica, S a˜ o Paulo, Brasil
ReceiVed: October 26, 2002; In Final Form: March 28, 2003
•-, were generated using flash photolysis of aqueous S
O
8
2-
solutions and the reactions
Sulfate radicals, SO
4
2
of the inorganic radicals with the surface of suspended silica nanoparticles (NP) investigated. In the presence
•
-
of colloidal silica no absorption traces due to SO
4
radicals are observed at 100 µs after the flash of light.
However, two transient species with absorption maxima around 320 and 600 nm are formed. A kinetic analysis
•-
of the experimental results indicate that SO
4
radicals are adsorbed on the NP surface, leading to the formation
-
1
-1
of an adduct, with λmax ≈ 320 nm (ꢀ ≈ 7000 cm M ), and showing similar reactivity to that observed for
the sulfate radical in aqueous solutions. The NP-sulfate radical adducts react with adsorbed water, and with
-
14
-(58(12)kJ/mol)/RT -1
3
single and geminal SiO sites with reaction rate constants of 1.5 × 10 × e
s , < 10 ×
-
(2(17)kJ/mol/RT -1
11
-(46(13)kJ/mol/RT -1
•
e
s
and < 10 × e
s , respectively. Two different SiO surface defects, showing
similar spectra (λmax ≈ 600 nm) but different reactivities, are formed from the reaction of NP-sulfate radicals
and deprotonated geminal and single silanols.
Introduction
A number of different studies carried out on aqueous suspen-
8-15
sions of silica or with macroscopic silica,
indicate that there
Silica is one of the most common materials on earth. The
surface properties of liquid/silica interfaces and the role played
by the silica surface in interfacial chemical kinetics are important
are at least two types of silanol groups at the water/silica
interface characterized by a dissociation pKa of 4.5-6.5 for
about 15% of the silanol sites and a pKa value of 8.5-9.0 for
the remaining 85%. The pKa values of 4.5-6.5 and 8.5-9.0
were associated with the geminal and single silanols, respec-
1
to understand processes involved in catalysis, environmental
2
a
2b
chemistry, in particular soil chemistry, and chromatogra-
3
,4
phy.
16,17
tively.
Amorphous silica surfaces can be described by the Zhuralev
An experimental approach employed here to study liquid/
solid interfaces involves colloidal particles dispersed at modest
concentrations which provide a high concentration of surface
sites for reaction. Since colloidal silica has a very high specific
surface area and is optically transparent in the near-UV and
visible regions, it is a suitable model surface for investigating
interfacial reaction kinetics involving photochemically generated
5
model, which considers that the factor determining the surface
properties is the presence of silanol groups and siloxane bridges.
Chemical reactions involving the latter moieties lead to modi-
fications of the surfaces, which may be used, on one hand, to
provide information on the structure of the silica surface, such
as nanorugosity, chemical fractality, and distribution of surface
5
silanols. On the other hand, the chemical reactivity of the
•-
species. As a result, sulfate radicals, SO4 , were generated using
surface active groups can be used to yield a solid surface with
entirely different properties (hydrophobic materials, function-
alized derivatives, etc.).6
Silica immersed in an aqueous environment has been shown
to orient its surface silicon atoms in a tetrahedral configuration
with each Si atom holding approximately one OH group. A
total of 4.6-4.9 OH groups per nm is found for all types of
fully hydroxylated silica, independent of its origin. Single,
2-
18,19
flash photolysis (λ > 200 nm) of aqueous S2O8 solutions,
reaction R1. Then the reactions of the inorganic radicals with
the silica nanoparticle (NP) surface were investigated following
the radical production.
5
S2O82- + hν f 2SO4•-
(R1)
2
5
Experimental Section
(mSi-O-)3Si-OH, and geminal, (mSi-O-)2Si(-OH)2, silanols
present in an irregular arrangement, act as centers of adsorption
of molecules capable of forming H bonds or of undergoing
donor-acceptor interactions. Dissociation of surface hydroxyl
Sodium peroxodisulfate (p.a.), sodium perchlorate (99%),
perchloric acid (70-72% aqueous solution), sodium hydroxide
(
99%), and absolute ethanol (99.8%), all from Merck, were used
-
7
groups to SiO ions renders a negatively charged silica surface.
-1
as received. Distilled water (>18 MΩ cm , <20 ppb of organic
carbon) was obtained from a Millipore system.
*
To whom correspondence should be addressed.
Universidad Nacional de La Plata.
Escola Politecnica da USP-Departamento de Engenharia Quimica.
E-mail: dmartire@inifta.unlp.edu.ar.
For most experiments stock Ludox colloidal silica suspensions
†
(
SM-30, 30 wt % SiO2 in water at pH 10.2 with an average
‡
§
⊥
2
particle diameter of 7 nm, surface area: 345 m /g, Aldrich) were
used. For some experiments colloidal silica was obtained by
E-mail: gonzalez@inifta.unlp.edu.ar.
1
0.1021/jp027318t CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/31/2003

6
132 J. Phys. Chem. B, Vol. 107, No. 25, 2003
Caregnato et al.
suspending fumed silica nanoparticles (silica content, 99.8%;
surface area, 390 ( 40 m /g; particle diameter, 7 nm; Sigma)
in aqueous solutions.
the minimum number of species and on their relative concentra-
tion profiles and absorption spectra.
2
24
The absorbance may be written in a matrical form [a] )
[E][c], where [a], [E], and [c] represent the absorbance, absorp-
tion coefficients, and concentrations matrixes, respectively. The
procedure is based on the alternating calculus of the matrices
[E] and [c] by multiple linear regression (using a least-squares
condition) iteratively until convergence is achieved when the
elements of an error matrix are minimized. From the error
The pH of the samples was measured with a Methrom-
Herisau pH meter model E512. Ludox colloidal silica dilutions
to concentrations of the order of 0.025 mg/mL yields dispersions
of pH 8.0-8.4. Since silica sols at pH 8 were reported to remain
stable even in the presence of salt concentrations of the order
2
0
of 0.15 M, this pH was used in most experiments. When
required, the samples were acidified with HClO4. Ludox
colloidal silica has its greatest tendency to gel at pH 5-6, and
consequently this pH range was avoided. At lower pH, sols
become more stable. Colloidal dispersions of pH > 10.5 were
not used because (i) an efficient scavenging of sulfate radicals
c
matrix, a corrected absorbance matrix, [a ], is obtained. The
number of species is selected as the minimum value that allows
c
a reasonable approximation between [a ] and [a].
Kinetic Computer Simulations. The kinetic model for
computer simulation is based on component balances and
equilibrium equations formulated in terms of a differential
algebraic equations system which is solved by Gear’s Stiff
-
21
by HO ions takes place and (ii) in this pH range colloidal
silica becomes increasingly solubilized and the alkali silicate
thus formed destabilizes the remaining colloids.
25
method and a least-squares estimation criterion. The version
Because of the reduction in sol stability by added salts, the
ionic strength of the nanoparticle suspensions was not adjusted,
except otherwise indicated. When necessary, the ionic strength
was varied by addition of sodium perchlorate up to values e0.15
M because of sol stability.
implemented performs the solution of the differential algebraic
equations system concomitantly to the evaluation of the
parametric sensitivities based on the decoupled direct method.
In consequence, the simulation of the system is robust, such
that it can be carried out with success for very different
parameter values, while the sensitivities are accurate and
obtained without much computational effort. The algorithm for
minimization criterion is based on the rotational discrimination
and provides additional features for assessing the identifiability
of the problem.
The NP concentration used was in the range 0.00025-0.25
g/L. Higher values were avoided, since flash photolysis experi-
ments of persulfate solutions using increasing NP concentrations
in the water-jacket around the flash cell, show that high
concentrations of NP act as a filter to the absorption of light by
S2O82 ions. The light intensity emerging from samples of
persulfate in the absence and presence of NP was coincident
for NP concentrations < 0.25 g/L, λ > 300 nm, and an optical
path length of 20 cm. Therefore, any interference by scattered
light is of no significance under the experimental conditions
used in the photolysis experiments.
-
Global Analysis. The algorithm employed for the computer
simulations (vide supra) was used as a criterion for the goodness
of the fit to a given function and to obtain the best kinetic
parameters (weight coefficients and rate constants) describing
the absorbance profiles of a group of experiments obtained at
different λ and under different experimental conditions.
-5
In experiments with ethanol, [EtOH] g 2 × 10 M was used
•
-
because of the requirement [SO4 ]0 , [EtOH] such that the
Results and Discussion
ethanol concentration remains constant throughout the experi-
-
4
The SO4•- radical ions were generated by photolysis (λ >
ment. The upper limit [EtOH] e 1 × 10 M was chosen in
order to minimize the contribution of the competing reaction
of ethanol with sulfate radicals.
-
3
19,26
200 nm) of (2-7) × 10 M peroxodisulfate solutions
in
the pH range 3-9, reaction R1. In the absence of colloidal silica,
the transient absorption spectrum and the decay rate were
Flash-photolysis experiments were carried out with a con-
•
-
19
22
coincident with those reported for SO4 radicals. Figure 1,
lower inset, shows the signals obtained at 450 nm after
photolysis of 1 × 10 M S2O8 solutions. Flash photolysis
of colloidal silica dispersions in the absence of peroxodisulfate
showed, within our experimental error, no transient absorption.
ventional apparatus (Xenon Co. model 720C) with modified
optics and electronics.23 Two collinear quartz Xenon high-
-
3
2-
intensity micropulse flash tubes, Xenon Corp. P/N 890-1128
(fwhm e 20 µs), with a continuous spectral distribution ranging
from 200 to 600 nm and maximum around 450 nm were used.
The analysis source was a high-pressure mercury lamp (Osram
HBO-100 W). The optical path length of the 1 cm -internal
diameter quartz sample cell was 20 cm. The monochromator
collecting the analysis beam (Bausch & Lomb, high intensity)
was coupled to a photomultiplier (RCA 1P28), whose output
was fed into a digital oscilloscope (Leader LBO-5825). Digital
data were stored in a personal computer. The temperature was
controlled to (0.5 K by circulating distilled water from a
thermostat model NB-D8/17 MGW-Lauda, through the jacket
of the reaction cell. The temperature was measured inside the
reactor cell with a calibrated Digital Celsius Pt-100 Ω ther-
mometer.
Bilinear Regression Analysis. For each experimental condi-
tion, several absorbance decay profiles at different detection
wavelengths were taken. Absorbance is thus a function of
wavelength and time. A bilinear regression analysis taking
advantage of the linearity of the absorbance with both concen-
trations and absorption coefficients, was applied to the experi-
mental absorption matrix in order to retrieve information on
-3
Flash photolysis of aqueous (2-7) × 10 M peroxodisulfate
containing (0.00025-0.25) g/L colloidal silica, shows no
•
-
absorption traces due to SO4 radicals in our time window (t
g 100 µs). However, absorption traces with maxima at around
320 nm (Figure 1) and 600 nm (Figure 1, upper inset) with
markedly different decay kinetics are observed, indicating the
presence of at least two absorbing species. The traces adsorbing
in the wavelength range from 480 to 680 nm are formed with
the same rate as the transient with λmax ≈ 320 nm decays (see
Figure 1).
The decay kinetics and absorption spectra of both transients
were found to be independent, within the experimental error,
of the presence of dissolved molecular oxygen in the range
-
4
0 < [O2] < 2.5 × 10 M. Therefore, most experiments were
performed with air-saturated suspensions. The pH decreases with
the number of flashes irradiated on the samples and conse-
quently, freshly prepared solutions were used for all experiments.
A good reproducibility of the transient signals is observed for
all the experimental conditions.

Reactions of Sulfate Radicals
J. Phys. Chem. B, Vol. 107, No. 25, 2003 6133
2-
Figure 3. Dependence of kapp with S
O
2 8
concentration for experi-
ments with suspensions of pH 8 and 0.025 g/L NP (Ludox) at 285.6
(2), 288.7 (4), 293.2 (b), and 298.4 K(O). Open squares correspond
to samples of pH ) 4.2 at 293.2 K. Unless otherwise indicated, error
Figure 1. Absorbance profiles obtained at 320 nm for experiments
-
3
2-
with 0.025 g/L NP (Ludox), pH ) 8 and (a) 5 × 10 M S
2
O
8
-3
at
M
-
3
2-
2
85.2 K, (b) 1 × 10 M S
2
O
8
at 285.2 K, and (c) 1 × 10
2
-
app
bars in k are of the order of the symbol size. Inset: Dependence of
S
2
O
8
at 298.2 K. Upper inset: Absorbance profiles obtained at 600
k
app with NP concentration for experiments with suspensions of pH 8
nm for experiments a and b of the main figure. Lower inset: Curve d,
corresponding to the absorbance profile obtained at 450 nm for
-
3
2-
8
and 5 × 10 M S
2
O
at 298.4 K
-
3
2-
8
experiments with 1 × 10 M S
2
O
at 298.2 K in the absence of
added NP. Solid curves represent computer simulations (refer to text).
-3
Experiments with samples of pH ) 8 containing 5 × 10
2-
M S2O8 and various NP concentrations showed transient decay
rates with pseudo first-order rate constant, kapp, independent of
[
NP] (see Figure 3 (inset)). The initial absorption of the transient
was also found to be independent of [NP], except for [NP] )
.25 mg/mL which shows lower initial absorbance due to an
0
internal filter effect (vide supra).
Irradiation of samples of variable ionic strength in the range
-
3
2-
from 3 × 10 to 0.15 M, pH ) 8, 0.025 g/L NP, and [S2O8
]
-
3
)
10 M at 293.2 K yields traces which, within the experi-
mental error, show identical decay rate constants and initial
absorbance (results not shown). For colloidal dispersions with
2
-
-2
[
S2O8 ] ) 2 × 10 M, variations of the ionic strength were
limited to the range 0.06 to 0.15 M. Under such conditions, no
ionic strength effect on the decay rate was observed within the
experimental error. Therefore, variations in the ionic strength
up to values e 0.15 M is of no significance for the decay of
the transient.
Figure 2. Absorption spectrum of the transient with λmax around 320
nm for experiments at 298.2 K with 0.025 g/L NP (Ludox) suspensions
-
3
2-
8
containing 5 × 10 M S
2
O
at pH (b) 3.0, (1) 4.0, (O) 7.0, and (3)
9
.0. Inset: Plots of kapp vs pH for the experiments in the main figure.
Experiments with samples of pH ) 8 containing 0.025 g/L
2
-
NP and various S2O8 concentrations showed faster transient
Experiments performed with nanoparticles of different origin
Sigma and Ludox suspensions) but otherwise identical experi-
2
-
decay rates with increasing [S2O8 ] (see Figure 1). The traces
could be well fitted to a first-order kinetics with a pseudo first-
order rate constant, kapp, linearly increasing with [S2O82 ], as
shown in Figure 3 for different temperatures in the range 285
to 299 K and also for colloidal dispersions of pH 4 at 293.2 K.
Figure 4 shows the Arrhenius plots of the intercepts, a, and
slopes, b, of the straight lines in Figure 3 (except for that at pH
(
mental conditions, qualitatively show the same behavior. For
quantitative analysis, experiments were mainly performed with
Ludox NP suspensions.
For clarity, the behavior of both observed transients is
described separately.
-
Transient with Absorption Maximum around 320 nm.
Time-resolved absorption curves were obtained from experi-
4
), yielding activation energies of 58 ( 12 and 53 ( 15 kJ/
-
3
2-
-2
mol, respectively.
ments performed with 1 × 10 M e [S2O8 ] e 5 × 10 M,
2-
The initial absorption of the transient increases with S2O8
0
9
.00025 g/L e [NP] e 0.25 g/L, and pH in the range from 3 to
2
-
-3
concentration, as shown in Figure 1 for [S2O8 ] ) 1 × 10
. For each colloidal dispersion, several absorbance decay
-
3
and 5 × 10 M at 285.2 K. A similar behavior is observed for
initial sulfate radical concentrations, formed by reaction R1 in
experiments in the absence of NP but under otherwise identical
experimental conditions, as expected for the dependence of the
profiles were taken at different detection wavelengths within
the range 290-360 nm. For all the experimental conditions used,
the bilinear regression analysis applied to the absorbance matrix
see Experimental Section) shows that the data may be described
by only one absorbing species.
The absorption spectrum of the observed transient is inde-
pendent of pH in the range from 3 to 9 (see Figure 2). The
decay rate of the transient follows a first order kinetics with a
rate constant slightly decreasing with increasing pH (see Figure
(
2-
absorbed light intensity on [S2O8 ]. These observations further
support that formation of the transient absorbing at these
wavelengths is intimately related to sulfate radical generation
from reaction R1.
Moreover, for each [S2O82-], the initial absorption of the
traces may be assumed to be, within the 20% error dispersion,
2
(inset)).

6
134 J. Phys. Chem. B, Vol. 107, No. 25, 2003
TABLE 1: Reaction Manifold of SO •--NP Adductsa
rate constant
Caregnato et al.
4
activation energy
•
-
•
g 1 × 10 s m-2
7 -1
SO
4
+ NP f NPS
k
k
k
k
2
3
4
5
(R2)
(R3)
(R4)
(R5)
•
2-
) (1.3 ( 0.2) × 10 M s-1
5
-1
NPS + S
2
O
8
f
E
a3 ) (53 ( 15) kJ/mol
•
4
-1
NPS f
) (1.01 ( 0.02) × 10 s
•
) (3.1 ( 0.2) × 10 M s-1
7
-1
NPS + C
H
2 5
OH f
a
Depicted values correspond to colloidal suspensions of pH ) 8 at 298 K.
-
1
2
00 s , respectively. The error bars for these rate constants
and for all other data reported here are the standard devia-
tions, σ.
The observed results indicate that the transient with λmax
•
around 320 nm, denoted as NPS , is formed after the interaction
•
-
of SO4 radicals with the nanoparticle surface, reaction R2.
•
-
The SO4 radicals are depleted within less than 100 µs by NP
concentrations as low as 0.00025 g/L and a total surface area
2
7
of 0.00135 m in the reaction cell. Therefore, k2 is g 1 × 10
-1
-2
s
m
for the reaction of sulfate radical with the silica surface.
The most simple mechanism explaining the observed trends
•
for the decay of NPS in the absence of ethanol considers
reactions R2-R4 in Table 1, with a first-order apparent rate
2-
constant kapp ) k4 + k3[S2O8 ], as expected from the observed
2
-
dependence of kapp on [S2O8 ], vide supra. The observed
transient is also able to react with ethanol as depicted in reaction
R5.
The pH independent spectrum and initial concentration of
•
•
NPS indicate that NPS does not take part of an acid/ base
Figure 4. Arrhenius plots of intercept a (O) and slope b (b) of the
lines in Figure 3, except for that at pH ) 4.2. Error bars in kapp values
are of the order of the symbol size. Inset: Absorbance profiles obtained
•
equilibrium and that NPS formation is not due to a chemical
•-
reaction between SO4 radicals and surface silanols. Otherwise,
differences in the spectrum and initial absorption would arise
within the pH range from 3 to 9, as the pKa of the silanol sites
were reported to be within this range, vide supra.
Both the reactions of NPS with peroxodisulfate and ethanol,
-
3
at 320 nm for experiments with 0.025 g/L NP (Ludox) and 1 × 10
2
-
M S
O
2 8
at pH ) 8 and, from bottom to top: 298.4, 293.2, 288.7,
and 285.9 K. The solid lines stand for first-order fits.
12
•
reactions R3 and R5, show rate constants of the same order of
magnitude as those reported for the sulfate radicals with these
5
-1 -1
substrates in aqueous solutions (k ) 6 × 10 M
s
and 5 ×
7
-1 -1
28
1
0 M
s
, respectively). This observation seems to indicate
that NPS chemically behaves like a sulfate radical. Identification
of the NPS transient with a SiO or SiOO radical on the NP
surface, may be neglected, since these radicals were reported
to absorb around 630 and 230 nm, respectively. Moreover,
•
•
•
•
2
9
•
SiO reactivity would resemble that of the silicate radical,
SiO3 , which shows a much lower rate constant for its reaction
•-
2
8
5
-1 -1
with ethanol (k ) 8.3 × 10 M s ) than that observed for
•
reaction R5. Formation of silicon radicals, Si , may also be
neglected, since these radicals are efficiently quenched by
Figure 5. Plot of kapp vs ethanol concentration for experiments at 298.7
2
9
2
-
-3
molecular oxygen, contrary to our results showing no effect
K with [NP (Ludox)] ) 0.025 g/L, [S
2
O
8
] ) 5 × 10 M and pH 8.
•
The error bars represent standard deviations.
with the dissolved [O2]. Therefore, NPS may be assigned to a
NP-sulfate adduct, as also supported by a low activation energy
independent of temperature as shown by the intercepts of the
first-order plots in Figure 4 (inset). The quantum yield (Φ1) of
sulfate radical formation by reaction R1, and the reciprocal of
the SO4 lifetime increase with increasing temperature. Thus,
the two opposite effects on the amounts of transient formed by
reaction R2 result in the constant initial concentration of the
observed transient, which supports a global small activation
energy for this reaction.
for reaction R2.
•
No effect of the ionic strength on the decay rate of NPS
2
-
-3
was observed for experiments with [S O ] ) 10 M, where
2
8
•-
27
mainly reaction R4 contributes to transient depletion. The latter
observation indicates that reaction R4 should be, either a
chemical reaction between uncharged substrates or a surface
process between adsorbed molecules.
•
The surface reaction of NPS with adsorbed water (reaction
+
Experiments performed at 298.7 K with 0.025 g/L NP
R4a in Table 2) yields H , in agreement with the pH decrease
-
3
2-
suspensions of pH ) 8 containing 5 × 10 M S2O8 in the
presence of various amounts of ethanol, showed transient decay
rates which followed a first-order kinetics with kapp linearly
increasing with [EtOH] (see Figure 5). The slope and the
intercept of the straight line in Figure 5 obtained form a least-
observed after flash irradiation. A similar reaction is observed
in aqueous solutions between sulfate radicals and water mol-
2
8
•
ecules. Reaction of NPS with surface silanols, SiOH, or with
-
their conjugate base, SiO , as shown in reaction R4b for the
latter groups, should also be considered. Both, an H- abstraction
7
-1
-1
•
squares analysis are (4.5 ( 0.3) × 10 s
M
and 10 100 (
of NPS radicals from surface silanols or an electron-transfer

Reactions of Sulfate Radicals
J. Phys. Chem. B, Vol. 107, No. 25, 2003 6135
TABLE 2: Detail of the Surface Reactions R4 of the NP-Sulfate Adducta
activation energy (kJ/mol)
preexponential factor (s-1)
•
.
+
14
[
[
[
NPS + H
2
O(ads)]surface f NP + HO + H
58 ( 12
46 ( 13
2 ( 17
1.5 × 10
<1 × 10
<1 × 10
(R4a)
(R4bg)
(R4bs)
•
-
-
•
2-
11
3
NPS + >Si(-O )
2
]
surface f >Si(O )O + SO
4
•
-
•
2-
4
NPS + mSi-O ]surface f mSi-O + SO
a
Depicted values correspond to suspensions of pH ) 8. The characters g and s stand for geminal and single silanols, respectively.
Figure 7. Absorption spectrum of the transient with maximum around
Figure 6. Absorbance profiles obtained at 600 nm for experiments
2-
-4
6
00 nm for experiments with [S
2
O
8
] ) 2.7 × 10 M and [NP
-
3
with 0.025 g/L NP (Ludox), pH 8 and (from bottom to top) 1 × 10
,
(
7
Sigma)] ) 0.00025 g/L at different pH: (9) 9.2, (0) 8.2, and (O)
-
3
-3
2-
8
4
.5 × 10 , and 7 × 10 M S
2
O
. Inset: Absorbance profiles
.4. Inset: pH dependence: (O) the area below the curves in the main
figure, (4) C , and (2) C , all in arbitrary units. The solid lines stand
for the fraction of dissociated silanols, [SiO ]/[SiOH]total vs pH.
obtained at 600 nm for experiments with 0.025 g/L NP (Ludox), pH )
g
s
-
3
2-
-5
8
and 5 × 10 M S
2
O
8
in the presence of 2 × 10 M (lower trace)
-
-
5
and 5 × 10 M ethanol. These experiments were performed at much
lower light intensities than those in the main figure. The solid lines
stand for a global computer analysis (see text).
•
-
process between NPS and SiO , reaction R4b, yield surface
SiO radicals.
•
Transient with Absorption Maximum around 600 nm.
-
4
Transient formation at 600 nm is fast (half time <1.5 × 10
-1
s ) compared to its decay rate (half time on the order of 0.01
-1
s ), as shown by the traces in Figures 1 (inset) and 6. Therefore,
the formation and decay kinetics of the transient may be
independently analyzed.
Under all the experimental conditions used, the decay of the
traces in this wavelength range may be well fitted to a
biexponential function, eq 1. In this equation the subscripts s
and g stand for the slow and fast decaying components,
respectively.
-
kgt
-kst
∆
A ) Cge
+ Cse
(1)
Figure 8. Arrhenius plot of k
s g
(O) and k (b) for experiments with
2-
-3
[
NP (Ludox)] ) 0.025 g/L at pH 8 and [S
O
2 8
] ) 1 × 10 M. Unless
The transient absorption decreases with pH in the range from 9
to 6 and no signal is observed below pH 5.4 (see Figure 7).
The biexponential analysis of the traces yield kg and ks values
which are, within error, independent of the detection wavelength
and pH. The coefficients Cg and Cs are wavelength- and pH-
dependent. The general behavior observed for each given
wavelength indicates an increase in Cs and constant Cg values
with increasing pH (see Figure 7 (inset)).
otherwise indicated by the corresponding bars, the error in ln k values
are of the order of the symbol size. Inset: Logarithm plots of C
(b), and ∆A (4) obtained at 600 nm vs the inverse of the
temperature for the experiments in the main figure.
s
(O),
C
g
0
-3
Experiments with dispersions of pH ) 8 containing 5 × 10
2
-
M S2O8 and various [NP] in the range from 0.25 to 0.0025
g/L showed transient decay rates with pseudo first-order rate
constants kg and ks linearly increasing with [NP] (not shown).
The absolute rate constant for the reactions of the fast and slow
Experiments in the temperature range from 284 to 301 K
show increasing decay rates and decreasing signal amplitudes
with increasing temperature. Arrhenius plots for kg and ks (see
7
6
transients with NP are (1.9 ( 0.2) × 10 and (1.5 ( 0.8) × 10
-
1
-1
M
s , respectively. The coefficients Cg and Cs were found
-
3
2-
Figure 8) obtained for experiments with 1 × 10 M S2O8
to be independent of [NP].
and 0.025 g/L NP suspensions of pH ) 8 yield global activation
energies Eg ) 30 ( 11 and Es ) 39 ( 16 kJ/mol, respectively.
Both Cg and Cs decrease at higher temperatures (see Figure 8
Experiments with various initial [S2O82-] but otherwise
identical experimental conditions, showed faster transient decay
2-
rates and higher absorbance with increasing [S2O8 ] (see Figure
6). The latter observation is in agreement with the behavior
(inset)).

6136 J. Phys. Chem. B, Vol. 107, No. 25, 2003
Caregnato et al.
TABLE 3: Manifold of Reactions of SiO Surface Defectsa
•
rate constant (M- s-1) for
1
rate constant (M s-1) for
-1
•
-
•
-
SiO generated from geminal SiO
SiO generated from single SiO
•
7
6
SiO + NP f
k
k
k
6g ) (1.9 ( 0.2) × 10
7g ) 2500 ( 300
k
k
k
6s ) (1.5 ( 0.8) × 10
7s ) 630 ( 50
(R6)
(R7)
(R8)
•
2-
SiO + S
2
O
8
f
•
6
4
SiO + C
H
2 5
OH f
8g ) (3.5 ( 1) × 10
8s ) (8 ( 5) × 10
a
Depicted values correspond to suspensions of pH ) 8 at 298 K.
•
expected for the 600 nm transient formed from the decay of
NPS . Fit is obtained from a least-squares analysis to eq 1, and
Deeper Insight into the Reaction Mechanism. SiO signals
were not observed in suspensions of pH below 5.4, where all
surface hydroxyls are in their undissociated form, thus indicating
•
the results give kg and ks. The kg and ks values are found to be
linearly dependent on the S2O82 concentration.
-
that the reaction between NPS and surface silanols is of no
•
•
Data obtained from suspensions of pH ) 8 containing 0.025
significance. However, increasing concentrations of SiO radicals
-
3
2-
g/L NP and 5 × 10
M S2O8
and different ethanol
are formed with increasing pH, as the fraction of surface
dissociated silanols also increases. Figure 7 (inset) shows the
good agreement between the pH dependence of the area below
-
5
concentrations in the range (1-10) × 10 M, showed faster
decay kinetics with increasing ethanol concentrations (see Figure
•
6
(inset)). Fit is obtained from a least-squares analysis to eq 1,
the absorption spectrum of SiO radicals and the fraction of
and the results give kg and ks. The kg and ks values are found to
be linearly dependent on the EtOH concentration. The amount
of transient formed decreases with increasing ethanol concentra-
tion, in agreement with a transient precursor which is also
dissociated silanols at different pH calculated taking the
1
5
reported pK values (vide supra). Therefore, the electron-
a
transfer reaction between surface dissociated silanols and the
adsorbed sulfate radical, reaction R4b, is of significance at pH
•
•
-
reactive toward ethanol, as is the case of NPS .
> 5.3. The behavior for NPS surface reactions with SiO and
•
-
The previous results indicate that the observed traces are due
4
SiOH is similar to that of the reactions of SO radicals with
2
-
2-
7
-1 -1
5
-1
to two transient components, both reacting with NP, S2O8
,
SiO
(k ) 2 × 10 M s ) and H SiO (k < 2 × 10 M
3
2
3
-
1
28
and ethanol as shown in reactions R6-R8 in Table 3, where
s ) in aqueous solutions, as for both systems, the electron-
transfer process is more efficient than H-abstraction. The latter
observation further stresses the similar reactivity patterns of the
•
SiO stands for the transients absorbing at 600 nm. The overall
decay rate constants kg and ks are those depicted in eqs 2a and
•
•-
2
b, respectively.
4
NPS transient and SO radicals.
•
The increase in the concentration of SiO surface radicals with
the fraction of total deprotonated silanols indicate that the two
2
8
-
k ) k [NP] + k [S O ] + k [EtOH]
(2a)
(2b)
g
6g
7g
2
8g
types of silanol sites characterized by pKa of 4.5-6.5 and 8.5-
2-
k ) k [NP] + k [S O ] + k [EtOH]
s
6s
7s
2
8
8s
8-15,17
9
.0
participate in reaction R4b. The fast and slow
components of the decay rates of the 600 nm transient may
thus be related to these two silanol sites as indicated by the pH
dependence of the weight factors Cg and Cs. The independence
of Cg with pH in the range from 7 to 9 suggests that this
component is associated with the silanol site with the lower
pKa, while the increase with pH of Cs follows the increase in
the fraction of dissociated silanols due to those sites with
pKa ) 8.5-9.0.
A computer global analysis of the traces observed in the
wavelength range 560-640 nm for experiments with [NP] )
-
3
2-
-2
0
0
.025 g/L, pH ) 8, 10 M e [S2O8 ] e 2 × 10 M and
-4
e [EtOH] e 1 × 10 M, at 298.2 K, was used in order to
retrieve the values of the absolute rate constants in eqs 2a and
b with minimized error. The best fitting values are those
2
depicted in Table 3. The goodness of the global analysis is
shown by the solid lines in Figure 6. The observed values
-1
-1
Since the two types of silanol groups on the surface of a fully
hydroxylated silica have been correlated with isolated and
k6s[NP] < 0.1 s and k6g[NP] ) 5.4 ( 0.6 s , are in agreement
with the expected values for the reaction of the transients with
the NP concentration used (see above), which indicates that a
reaction of the transient with water should have a negligible
contribution to the overall decay rate of this transient.
1
6,17
the fast and slow components of the SiO•
geminal silanols,
transient may be assigned to radicals formed from geminal and
-
single SiO sites, reactions R4bg and R4bs, respectively. The
•
SiO surface defects in both sites show identical absorption
The ionic strength (µ) effect was studied in experiments with
-
2
2-
spectrum. The geminal silanols have been regarded as the
2
× 10 M S2O8 and 0.025 g/L NP at pH ) 8 under
17
reactive sites for adsorption and chemical reaction. We also
conditions where both transient components decay mainly by
reaction R7. The decay rate constant for the slow transient
component does not depend, within the experimental error, on
•
-
observe here that SiO radicals arising from geminal SiO are
-
more reactive than those radicals arising from single SiO . The
measurable ionic strength effect observed for the fast SiO•
component may be explained by the proximity of a negative
charge due to the dissociated geminal silanol. On the other hand,
-
2
the ionic strength in the range from 6 × 10 M to 0.15 M.
1/2
1/2
However, plots of log kg vs µ /(µ + 1) yield straight lines
with a slope value of 1.7 ( 0.3. The observed value is expected
•
-
2
-
SiO radicals arising from single SiO have their nearest
negative charge at farther distances, and consequently, they
are less prone to influence the SiO reacting center.
For experiments with low [S2O82 ], higher SiO signals were
observed to be formed at lower temperatures. Under these
experimental conditions, reaction R3 is of no importance and
mainly reactions R4a, R4bg, and R4bs compete for the NPS
transient. The total concentration of surface SiO formed from
for the reaction of two repelling species, S2O8 ions and a
negatively charged nanoparticle.
1
7
•
Nonbridging oxygen hole centers assigned to surface SiO•
radicals in glassy SiO2 and SiO2 nanoparticles show absorption
-
•
29
around 630 nm, with a spectrum similar to that reported for
•
-
30
SiO3 radicals. The transient absorbing in the wavelength
•
•
range from 480 to 680 nm formed from the NPS radical may
thus be assigned to the surface SiO radicals. This assignment
•
•
-
is also supported by the similar reactivity toward ethanol
observed for the 600 nm transient (reaction R8) and that reported
geminal SiO is proportional to Cg. Since the initial concentra-
tion of NPS does not depend on temperature and considering
•
•
-
for SiO3 radicals (vide supra).
that k4a > k4bg, k4bs and k4bg is of the same order of magnitude

Reactions of Sulfate Radicals
J. Phys. Chem. B, Vol. 107, No. 25, 2003 6137
-1
than k4bs and an Arrhenius behavior exists for the three
M ) and showing reactivity similar to that observed for sulfate
radical in aqueous solutions. Adduct formation is independent
of the fraction of dissociated surface silanols and showed small
adsorption energies. Surface reactions of sulfate radical adducts
with adsorbed water and dissociated silanols were observed and
the reaction rate constants and their activation energies deter-
mined. Sulfate radical hydrogen abstraction from SiOH groups
is of no significance. This observation is in agreement with that
competing reactions, eq 3 is obtained.
A4a
E4a - E4bg
C )
× exp
(3)
g
(
)
A4bg
RT
where A stands for the preexponential factors.
•
A similar dependence is observed for surface SiO formed
-
from single SiO . In fact, plots of the logarithm of Cg and Cs
.
reported for gaseous HO radical adsorption over fused quartz
where hydrogen abstraction from silanols is only a secondary
vs the inverse of the absolute temperature yield straight lines
(
(
see Figure 8 (inset)) from which (E4a - E4bg) ) 12 ( 6 and
.
32
factor in HO radical removal from the surface.
Surface reactions of adsorbed sulfate radicals with single and
E4a - E4bs) ) 57 ( 11 kJ mol-1 are obtained. Therefore, the
observed temperature dependence for the apparent rate constant
-
•
geminal SiO yield SiO radicals with very similar spectroscopic
properties but different reactivity. The observed absorption with
λmax ≈ 600 nm is in agreement with that reported for
k4 ) k4a + k4bg + k4bs is mainly due to reaction R4a and the
-
1
estimated activation energies are E4a ) 58 ( 12 kJ mol , E4bg
-
1
-1
)
46 ( 13 kJ mol and E4bs ) 2 ( 17 kJ mol . The different
2
9
nonbridging oxygen hole centers and for aqueous solutions
values observed for E4bg and E4bs indicate different environments
as expected for single and geminal dissociated silanols. The
reaction of sulfate radicals with water solvent shows an
•-
30
•
of SiO3 radicals. SiO surface radicals formed from geminal
SiO sites are more reactive than single ones and their reactivity
toward ethanol resembles that of SiO3 radicals in aqueous
-
•
-
-
1 28
activation energy of 9.2 kJ mol . Since physically adsorbed
water forms hydrogen-bonded networks with surface silanol
groups with desorption energies of the order of 40 to 70 kJ
3
0
solutions.
Formation and decay of adsorbed sulfate radicals on colloidal
silica may lead to surface modification, and therefore, the results
obtained here are of relevance in those areas involving silica/
liquid interface chemistry, as environmental chemistry and
catalysis. It is of interest to investigate whether formation of
-
1 5
-1
mol , the 58 kJ mol energy barrier required for reaction
R4a to take place may be related to the rupture of the hydrogen-
bond network in the activated complex.
Taking the estimated activation energies, assuming k4 ≈ k4a
•
SiO surface defects also takes place from the interaction of
1
4
at 298 K, and considering eq 3, the values 1.5 × 10 , <1 ×
silica NP with other oxidizing radicals. Experiments to address
this question are presently underway.
1
1
3
-1
1
0
and <1 × 10 s are obtained for the preexponential
factors of reactions R4a, R4bg, and R4bs, respectively. There-
fore, reactions R4bg and R4bs are less than 10% efficient than
reaction R4a, in agreement with the experimental results
showing no decrease in the NPS decay rate constant at pH <
.4 where no SiO surface radicals are formed. The fraction of
Acknowledgment. This research was supported by Agencia
Nacional de Promoci o´ n Cient ´ı fica y Tecnol o´ gica, Argentina
•
(ANPCyT) and Consejo Nacional de Investigaciones Cient ´ı ficas
•
5
y T e´ cnicas, Argentina (CONICET), Funda c¸ a˜ o de Ampara a`
Pesquisa do Estado de S a˜ o Paulo, Brasil (FAPESP) and
Fundaci o´ n Antorchas, Argentina. M.C.G. is a research member
of CONICET. D.O.M. is a research member of Comisi o´ n de
Investigaciones Cient ´ı ficas de la Provincia de Buenos Aires
undissociated silanols increases considerably below pH 6 (vide
supra) favoring water adsorption through an H-bonding network.
We may thus expect slightly higher decay rates at the lower
pH, as observed experimentally.
To test the proposed reaction mechanism which accounts for
the experimental results, a detailed kinetic analysis was per-
formed with the aid of computer simulations (for details of the
program see Experimental Section). To this purpose, reactions
(CIC). P.C. thanks ANPCyT for a graduate studentship.
References and Notes
•-
R1-R3, R4a, R4bg, and R4bs, along with the reactions of SO4
radicals with water, hydroxyl ions, and peroxodisulfate ions and
(1) Conner, W. C.; Pajonk, G. M.; Teichner, S. J. AdV. Catal. 1986,
4, 1.
3
(
2) (a) Brandt, C.; van Eldik, R. Chem. ReV. 1995, 95, 119;. (b) Jury,
•
-
28
SO4 bimolecular recombination, were taken into account.
W. A.; Gardner, W. R.; Gardner, W. H. Soil Physics, 5th ed.; Wiley: New
York, 1991.
•
-
The flash emission is considered a ∆ function producing SO4
radicals. Initial [SO4•-] taken as an input parameter, was
estimated from experiments under identical conditions but in
(3) Tripp, C. P.; Hair, M. L. Langmuir 1993, 9, 3523.
(4) Tripp, C. P.; Hair, M. L. Langmuir 1991, 7, 923.
(
5) Zhuralev, L. T. Colloids Surf. A: Physicochem. Eng. Aspects 2000,
the absence of NP (ꢀ4 (SO4 ) ) 1600 M cm
50
•-
-1
-1 31
), as shown
1
73, 1.
6) Vidal, A.; Papirer, E. Chemical reactivity. In The surface properties
of silica; Legrand, A. P., Ed.; John Wiley & Sons Ltd.: New York, 1998.
•
•
in Figure 1 for experiments c and d. The NPS and SiO radicals
are the only species absorbing at 320 and 600 nm, respectively.
Simulated concentration profiles for these transients were
converted into the corresponding absorbance curves and com-
pared with the experimental data. An excellent agreement
between experimental and simulated profiles is observed taking
(
(
23.
7) Iler, R. K. The chemistry of silica; Wiley: New York, 1979; p
6
(
(
8) Maciel, G. E.; Sindorf, D. W. J. Am. Chem. Soc. 1980, 102, 7606.
9) Morrow, B. A.; Gay, I. D. J. Phys. Chem. 1988, 92, 5569.
(10) Tuel, A.; Hommel, H.; Legrand, A. P.; Sz. Kovats, E. Langmuir
1990, 6, 770.
(11) van Roosmalen, A. J.; Mol, J. C. J. Phys. Chem. 1978, 82, 2748.
3
20
•
-1
-1
600
•
ꢀ
(NPS ) ) 7000 ( 1000 M cm and ꢀ (SiO ) g 2100
-
1
-1
(
500 M cm (see the solid traces in Figure 1). The lower
(12) Allen, L. H.; Matijevic, E. J. Colloid Interface Sci. 1969, 31, 287.
(13) Allen, L. H.; Matijevic, E. J. Colloid Interface Sci. 1970, 33, 420.
(14) Allen, L. H.; Matijevic, E.; Meites, L. J. Inorg. Nucl. Chem. 1971,
600
•
limit in ꢀ (SiO ) values is due to the higher limit values input
for reaction rate constants k4bg and k4bs.
We may therefore deduce that the proposed mechanism and
estimated reaction rates may account for the observed experi-
mental data.
33, 1293.
(
27.
15) Ong, S.; Zhao, X.; Eisenthal, K. B. Chem. Phys. Lett. 1992, 191,
3
(16) Vance, F. W.; Lemon, B. I.; Ekoff, J. A.; Hupp, J. T. J. Phys. Chem.
B 1998, 11, 1845.
Conclusion
(17) Kropp, P. J.; Daus, K. A.; Tubergen, M. W.; Kepler, K. D.; Wilson,
V. P.; Craig, S. L.; Baillargeon, M. M.; Breton, G. W. J. Am. Chem. Soc.
Sulfate radicals are adsorbed on the NP surface leading to
the formation of an adduct with λmax ≈ 320 nm (ꢀ ≈ 7000 cm
1993, 115, 3071.
-1
(18) Dogliotti, L.; Hayon, E. J. Phys. Chem. 1967, 71, 2511.

6138 J. Phys. Chem. B, Vol. 107, No. 25, 2003
Caregnato et al.
(
19) McElroy, W. J.; Waygood, S. J. J. Chem. Soc., Faraday Trans.
990, 86, 2557.
20) Healy, T. W. Stability of Aqueous Silica Sols in The colloid
(26) (a) Choure, S. C.; Bamatraf, M. M. M.; Rao, B. S. M.; Das, R.;
Mohan, H.; Mittal, J. P. J. Phys. Chem. A 1997, 101, 9837. (b) Hug, G. L.
Natl. Stand. Ref. Data Ser. (U. S. Natl. Bur. Stand.) 1981, 69.
1
(
Chemistry of Silica; Advances in Chemistry 234; Bergna, H. E., Ed.,
American Chemical Society: Washington, DC, 1994; p 147.
(27) Brusa, M. A.; Churio, M. S.; Grela, M. A.; Bertolotti, S. G.;
Previtali, C. M. Phys. Chem. Chem. Phys. 2002, 2, 2383.
(
21) M a´ rtire, D. O.; Gonzalez, M. C. Int. J. Chem. Kinet. 1998, 30,
91.
22) (a) Porter, G.; West, M. A. Techniques of Chemistry, Hames, G.,
(28) Ross, A. B.; Mallard, W. G.; Helman, W. P.; Buxton, G. V.; Huie,
4
R. E.; Neta, P. NDRL-NIST Solution Kinetics Database:- Ver. 3.0; Notre
Dame Radiation Laboratory: Notre Dame, IN; National Institute of
Standards and Technology: Gaithersburg, MD, 1998.
(
Ed. Wiley-Interscience: New York. 1974; Vol. VI, Part II, Chapter X. (b)
Rabek, J. F. Experimental Methods in Photochemistry and Photophysics,
Part I; John Wiley & Sons: Belfast, Northern Ireland, 1982. Chapter 3.
(
29) (a) Skuja, L. J. Non-Cryst. Solids 1994, 179, 51. (b) Glinka, Y. D.
Phys. ReV. B 2000, 62, 4733.
30) Kuzmin, V. A.; Chibisov, A. K. High Energy Chem. 1970, 4, 146.
(23) San Rom a´ n, E. A.; Aramend ´ı a, P. F.; Schumacher, H. J. An. Asoc.
(
Qu ´ı m. Argent. 1980, 70, 887.
(24) San Rom a´ n, E. A.; Gonzalez, M. C. J. Phys. Chem. 1989, 93, 3532.
(31) McElroy, W. J. J. Phys. Chem. 1990, 94, 2435.
(
25) M a´ rtire, D. O.; Rosso, J. A.; Bertollotti, S.; Carrillo Le Roux, G.;
(32) Suh, M.; Bagus, P. S.; Pak, S.; Rosynek, M. P.; Lunsford, J. H. J.
Phys. Chem. 2000, 104, 2736.
Braun, A. M.; Gonzalez, M. C. J. Phys. Chem. 2001, 105, 5385.