You are on page 1of 9

Article

pubs.acs.org/JPCC

Bimodal Acidity at the Amorphous Silica/Water Interface


Morgane Pfeiffer-Laplaud,† Dominique Costa,‡ Frederik Tielens,§ Marie-Pierre Gaigeot,†,⊥
and Marialore Sulpizi*,#

LAMBE CNRS UMR8587, Université d’Evry val d’Essonne, Blvd F. Mitterrand, Bât Maupertuis, 91025 Evry, France

Chimie Paristech CNRS, UMR8247, Institut de Recherches de Chimie de Paris, 11 rue P et M Curie, 75005 Paris, France
§
Sorbonne Universités, UPMC Univ Paris 06, CNRS, Collège de France, Laboratoire de Chimie de la Matière Condensé de Paris, 11
place Marcelin Berthelot, 75005 Paris, France

Institut Universitaire de France, 103 Blvd St Michel, 75005 Paris, France
#
Department of Physics, Johannes Gutenberg Universitat, Staudingerweg 7, 55099 Mainz, Germany

ABSTRACT: Understanding the microscopic origin of the


acid−base behavior of mineral surfaces in contact with water is
still a challenging task, for both the experimental and the
theoretical communities. Even for a relatively simple material,
such as silica, the origin of the bimodal acidity behavior is still a
debated topic. In this contribution we calculate the acidity of
single sites on the humid silica surface represented by a model
for the hydroxylated amorphous surface. Using a thermo-
dynamic integration approach based on ab initio molecular
dynamics, we identify two different acidity values. In particular,
some convex geminals and some type of vicinals are very acidic
(pKa = 2.9 and 2.1, respectively) thanks to a special
stabilization of their deprotonated forms. This recalls the behavior of the out-of-plane silanols on the crystalline (0001) α-
quartz surface, although the acidity here is even stronger. On the contrary, the concave geminals and the isolated groups present a
quite high pKa (8.9 and 10.3, respectively), similar to the one of silicic acid in liquid water.

■ INTRODUCTION
Knowledge of the acid−base behavior of mineral surfaces is
pH, where the trimodal behavior is only observed when
tritation is initiated at high pH.10
crucial for the understanding of their chemistry, including not Other techniques such as evanescent wave cavity experi-
only surface dissolution, surface interactions with ligands ments3 have also shown the bimodal character of fused silica,
(especially biomolecules), and heterogeneous catalysis but measuring a ratio 4:1 of pKa, respectively for non acidic/acidic
also the building of charge at the interface with solvent and its sites. See ref 11 and references therein for more description of
consequence on ion transport, and more generally in the experimental works showing the existence of two types of
context of technological applications. Measuring the acid−base silanols in terms of acid−base character at the fused silica/water
character of insulating mineral oxides often relies on indirect interface.
measurements, using pH-sensitive molecules with spectroscopic Beyond the existence of two types of silanols at the surface of
signatures1−3 or potentiometric titration techniques4,5 that only silica and their actual pKa values, one challenge is to unravel the
operate over a limited pH range. Certainly a big step forward microscopic nature of these two populations, which is the goal
was made in 1992 when Eisenthal and co-workers demon- of the present paper using theoretical simulations. Beforehand,
strated that nonresonant second harmonic generation (SHG) one has to specify the silanols nomenclature that will be used
spectroscopy can provide a surface-sensitive and label-free throughout our work. See Figure 1 for a schematic
method for monitoring the acid−base behavior of fused silica.6 representation and also refs 11 and 12 for more details.
The planar silica/water interface has hence been shown by Qn refers to a Si atom bonded to n other Si atoms through
SHG and other methods to exhibit surface sites with two bridging oxygens, forming Si−O−Si siloxane groups. Upon
distinct acidities.1,3,6−9 In particular, the SHG experiments dissociation of water on the surface, (hydroxylated) Si−OH
showed that 19% of silanol groups on fused silica surfaces silanol sites are formed. In the present work we will concentrate
exhibit a pKa of 4.5, whereas 81% exhibit a pKa of 8.5.6 on hydroxylated surface sites. On the hydroxylated silica
Nonlinear SFG spectroscopy experiments on α-quartz also lead surface, a Q3 Si has one single Si−OH silanol group called
to similar conclusions on a bimodal behavior of silanols on
crystalline silica.9 Received: March 25, 2015
Very recent SHG results from Gibbs−Davis points to a Revised: November 9, 2015
trimodal or bimodal behavior of silica depending on the starting

© XXXX American Chemical Society A DOI: 10.1021/acs.jpcc.5b02854


J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

A consensus on the microscopic nature of the acidity


bimodal character of silanols at silica surfaces is clearly far from
achieved, showing the need for a theoretical molecular
methodology that would provide a robust way of calculating
pKa values of surface sites in the presence of the solvent and
simultaneously provide a microscopic rationalization of these
values. This can be achieved through ab initio molecular
dynamics (AIMD) simulations of the full silica−water interface,
as already shown in refs 15 and 17. One can indeed generate
controlled surface models on which the pKa of selected sites can
be calculated, taking into account the presence of the liquid at
the interface. Following such a method and using multiple
representative crystalline silica surfaces, Leung has calculated
the deprotonation free energies of silanol groups with different
Figure 1. Scheme of silanols encountered at the surface of silica: structural motifs.15 He showed that a plausible candidate for the
isolated Q3, H-bonded Q3, vicinal Q3, geminal Q2. This scheme does most acidic silanols observed in the experiments (pKa = 4.5)6
not take into account the presence of an aqueous interface.
could be located on locally strained or defected regions with
sparse silanol coverage.15 These AIMD simulations lend
support to the role of interfacial water, finding that silanol
“isolated” because it does not share H-bonds with any other groups on hydrophobically restructured quartz exhibited the
silanol. This occurs whenever the distance between this Si−OH lowest pKa.15 In this study it was also observed that the pKa of
and the closest Si−OH neighbor (d[(Si−O)−H···O−(Si)]) is silanols increased as the number of water layers increased. The
larger than 3.3 Å.11 See Figure 1a for a graphical representation. less acidic sites were shown to exhibit a pKa ≈ 9, which is
A Q3 silanol can also form hydrogen bonds, as shown in Figure similar to that of the monomeric form of silica, i.e., monosilicic
1b. A Q3 Si with one single Si−OH is called “vicinal” whenever acid (pKa(Si(OH)4) = 9.9).18
this silanol shares a common oxygen vertex in its tetrahedra In a recent work17 we have addressed the special case of the
with another silanol, as illustrated in Figure 1c. A Q2 Si has a α-quartz (0001)/water interface. The acidity of the surface
pair of geminal Si−OH, as illustrated in Figure 1d. All Si−OH groups were calculated from AIMD simulations using our
silanols of Q3 type are generally labeled as terminal silanols. developed proton insertion technique.19−22 The (0001) α-
The key issue is to provide a microscopic interpretation of quartz surface certainly represents a special case because the
the two silanol surface acidities measured in experiments in surface exhibits a regular pattern of Q2 silanol distribution and
relation with Q3 and Q2 types of Si silica sites, and in relation an associated regular hydrogen bond arrangement.17 Indeed,
with their local environment (i.e., interactions with the surface when the (0001) α-quartz surface is in contact with water, we
and with water at the wet surface). This is far from being fully showed that half of the silanols maintain an “in-plane”
understood from experiments alone. The main reason is that orientation, donating hydrogen bonds to nearby silanols,
the 4:1 ratio of nonacidic versus acidic sites measured whereas the other half of the silanols adopt an out-of-plane
experimentally3 is concomitant with several microscopic orientation, and act as H-bond donors to interfacial water
structural properties with (i) the average ratio between H- molecules in the first adsorbed layer.17 With our calculations we
bonded and isolated silanols (IR experiments13) but also with could conclude that (0001) α-quartz exhibits a bimodal
(ii) the ratio of Q2 versus terminal silanols (NMR experi- behavior with two different acidity constants, namely calculated
ments11). Whether the 4:1 ratio of nonacidic/acidic sites on at pKa= 5.6 and 8.5, respectively for the out-of-plane and in-
silica corresponds to one or the other population, or to another plane silanols, and relate these pKa values to the H-bond
one, is still unknown. Things are even more complicated, as strengths that the silanols make with the first adsorbed water
Chuang and Maciel showed that these populations overlap, as layer, i.e., strong (out-of-plane) and weak (in-plane) H-bonds.
terminal and Q2 silanols may/may not be H-bonded, From the experimental data it is clear that the bimodal acidity
depending on the local surface topology14 and on these sites’ character is also present at the fused silica/water interface,3 i.e.,
interactions with the solvent at the wet surface. Several a noncrystalline surface. So the question is how do silanols
hypotheses for explaining the bimodal acidity character of behave on an amorphous silica surface in terms of interactions
silica have been put forward. We list here some of them, with a with their environment (amorphous silica being taken as a
special emphasis on the most acidic character. This one has model for fused silica)? Can we predict and explain silanol
been hypothesized to be due to Q3-isolated silanols15 (e.g., near acidities in terms of local environment?
a strained cycle that forms upon dehydration) because they In this paper we provide an explanation for the bimodal
become engaged in interactions with the solvent once the acidity behavior for the amorphous silica/liquid water interface,
surface is wetted; alternative hypotheses are that this more considering an extended model of the amorphous silica surface.
acidic character could be due to any H-bonded site,11 or on the In particular, our model is based on an original structure
contrary, to silanols lacking hydrogen bonds,1 or to silanoliums developed by Garofalini23 and subsequently relaxed via ab initio
(doubly protonated silanols).16 In a previous theoretical work17 simulations.24 This surface has been characterized in both dry
we have provided an explanation for the bimodal character of and microsolvated conditions,24,25 and more recently in contact
silanols at the (0001) α-quartz interface in terms of H-bond with bulk water.26 The model presents a distribution of
strengths, showing that the more acidic silanols (pKa = 4.56) tetrahedral (SiO4) rings in the bulk solid mainly composed of
form the stronger hydrogen bonds to interfacial water whereas four-, five-, and six-membered rings, in good agreement with
the less acidic silanols (pKa = 8.56) form the weaker H-bonds the experimental data.24 It is important to note that besides the
with interfacial water. flexible Si−O−Si bond,27 no strained 3R-rings are present, by
B DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

construction. The bulk slab is about 10 Å thick (vertical −SiOH + H 2O(aq) → −SiO− + H3O+(aq) (1)
spacing), with an in-plane cell size of 13 Å × 17.5 Å, which
resulted in Si27O67H26 composition (120 atoms as a whole). The transfer process is implemented using the thermodynamic
The slab is about three-layers thick, with thickness locally integration. The charge of the acidic proton of a surface OH
varying between 5 and 8 Å, reflecting the surface rugosity, see group is gradually switched off, transforming the proton into a
Figure 2. The OH density of 5.8 OH/nm2 mimics the behavior neutral “dummy” particle. Simultaneously, a similar dummy
of an hydroxylated amorphous silica surface (4.5/4.9 OH/ proton attached to a water molecule is charged up, creating a
nm2). hydronium in the Eigen form. The fractional charges on the
two groups always add up to unity. Following the approach that
we have discussed in detail in ref 22, the discharge integral is
calculated according to
1
ΔdpA = ∫0 dη⟨ΔdpE⟩r η
(2)
where ΔdpE is the vertical energy gap, defined as the potential
energy difference between product P (MO−+H3O+(aq) in eq
1) and reactant R (−MOH + H2O(aq)) for instantaneous
configurations of a molecular dynamics trajectory. The
subscript rη indicates that the averages are evaluated over the
restrained mapping Hamiltonian
/η = (1 − η)/R + η /P (3)
where η is a coupling parameter that is gradually increased from
0 (−MOH + H2O(aq)) to 1 (−MO− + H3O+(aq)). This
Hamiltonian also contains a harmonic restraining potential
Figure 2. Amorphous silica slab used for calculations. Color code:
yellow, Si; red, O; white, H atoms. The five studied silanol sites are Vrestr keeping the dummy atom close to the equilibrium
highlighted in blue and use a larger size. position of the H+ nucleus in the protonated system. The
Simpson rule (three-point approximation) is used to evaluate
the integral in eq 2:
The acidity of the silanols at the dry surface has been 1 2
characterized in ref 24. The terminal Q3 hydroxyl groups were ΔA TP = (⟨ΔE⟩0 + ⟨ΔE⟩1) + ⟨ΔE⟩0.5
6 3 (4)
found to have a deprotonation energy of ∼650 kJ/mol and the
Q2 ones of ∼600 kJ/mol, with geminal silanols thus slightly This requires generation of three trajectories corresponding to
more acidic. The small 50 kJ/mol difference in the values of η = 0, 0.5, 1. This formula is often a good compromise
deprotonation energies may, however, not be enough to between computational cost and accuracy of the free energy
confidently predict the order of acidity in the presence of an change.19,36 The pKa value of a surface group is obtained from
aqueous phase. Vacuum DFT calculations of the adsorption of the proton transfer integral of eq 2 by adding a thermochemical
lutidine (2,6-dimethylpyridine) on Q2- and Q3-isolated silanols correction. The leading term in this correction adds in the
of this surface furthermore confirm the slightly higher acidity of translational entropy generated by the acid dissociation. This
Q2 sites.28 In the same spirit, we identified that the convex term is missing in reaction 1, which is formally a proton transfer
silanol nest formed around the Q2 site exhibits a higher reaction conserving the number of translational degrees of
chemical reactivity than Q3-isolated silanols, being typically an freedom. Indeed, this translational entropy term is related to
attracting site for the adsorption of glycine, and stabilizing basic the definition of our solvated proton, which is assumed to be in
adsorbates.25 the Eigen form, namely, as H3O+(aq), where instead the proton
Is the higher chemical reactivity of Q2 silanols maintained at is actually free to diffuse around and should be instead
the aqueous interface? represented as H+(aq). The simplification is motivated by the

■ METHODS
Calculation of the Acidity Constants. The acidity
necessity to have a well-defined position for the proton
insertion. A detailed explanation about this issue can be found
in ref 22. The overall formula used for the pKa calculation
constants of surface groups are computed using the reversible becomes
proton insertion/deletion method.19−22 The method was 1
initially developed and tested on a series of aqueous 2.30kBTpK a = ∫0 dη ΔdpE η + kBT ln[c o Λ 3H+]
(5)
compounds,19−22 and then applied to the calculation of the
acidity of surface groups of several oxides in contact with where co = 1 mol dm−3 is the unit molar concentration and ΛH+
water.17,26,33−35 The advantage of this approach is that the solid is the thermal wavelength of the proton. We refer the reader to
(mineral) surface and the solvent (water) are treated at the ref 22 for a detailed derivation of such a formula. The logarithm
same level of theory, and therefore the approach is particularly of the product coΛ3H+ accounts for the liberation entropy of the
suitable for heterogeneous environments such as interfaces. proton and is responsible for a correction of −3.2 pK units to
The acidity constants are calculated from the free energy for the thermodynamic integral.
transferring a proton from a group on the surface (in our case We comment that all the pKa values have been calculated on
the silanol OH) to a water molecule in the bulk solution. The a neutral surface, so at the point of zero charge condition. It is
following reaction is considered: expected that if the overall charge of the surface varies, in
C DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

Figure 3. Representative snapshots of the local structures around the SiOH groups at the water/amorphous silica interface. Hydrogen bonds appear
in white dashed lines.

Table 1. Energies (eV) Obtained in the pKa Calculations of the Geminal, Isolated, and Vicinal Silanols on Amorphous Silicaa
system traj ΔE0 ΔE0.5 ΔE1 ΔA pKa ΔpKa
Q2-Gem. 1 Whole −7.18 ± 1.62 0.35 ± 0.54 7.09 ± 1.03 0.28 2.9 1.6
Half1 −7.45 ± 1.59 0.51 ± 0.53 7.16 ± 0.98 0.36 4.1
Half2 −6.91 ± 1.60 0.19 ± 0.51 7.01 ± 1.06 0.15 0.9
Q2-Gem. 2 Whole −6.34 ± 1.45 0.77 ± 0.52 7.30 ± 1.09 0.67 8.9 0.2
Half1 −6.23 ± 1.49 0.75 ± 0.53 7.30 ± 1.08 0.68 9.0
Half2 −6.45 ± 1.40 0.78 ± 0.50 7.33 ± 1.10 0.66 8.7
Q3-Iso. Whole −5.49 ± 0.67 0.75 ± 0.54 7.24 ± 1.28 0.79 10.3 1.5
Half1 −5.62 ± 0.63 0.66 ± 0.48 7.04 ± 1.31 0.67 8.4
Half2 −5.36 ± 0.69 0.76 ± 0.54 7.45 ± 1.23 0.86 11.3
Q3-Vic. 1 Whole −8.30 ± 2.82 0.63 ± 0.53 7.40 ± 1.53 0.23 2.1 0.7
Half1 −8.62 ± 3.31 0.70 ± 0.57 7.66 ± 1.59 0.31 3.4
Half2 −7.99 ± 2.18 0.56 ± 0.47 7.14 ± 1.43 0.23 2.1
Q3-Vic. 2 Whole −5.24 ± 0.64 1.42 ± 0.50 6.43 ± 1.06 1.15 16.2 0.5
Half1 −5.31 ± 0.72 1.39 ± 0.52 6.56 ± 1.03 1.13 15.9
Half2 −5.18 ± 0.55 1.46 ± 0.49 6.49 ± 1.06 1.19 16.9
a
ΔE represents vertical energy gaps and ΔA is the difference in free energy. Uncertainties on the final values are qualitatively evaluated via an
estimation on the pKa in the first half (Half1) and in the second half (Half2) of the trajectories. This last part confirms the convergence of the
calculations.

particular when a nearby group gets deprotonated, this can have and 2 ps for the isolated SiOH) with a time step of 0.5 fs. On
a strong impact on the site pKa. average, the lengths of the trajectories (true data collection) are
Model Systems. Our AIMD simulations follow the same 10 ps for geminal 1 and vicinal 1, 15 ps for the isolated site, and
general setups used in our previous investigations of solid− 20 ps for geminal 2 and vicinal 2. These sites are described in
liquid interfaces.17,26,33−35 DFT-based Born−Oppenheimer the next section.
molecular dynamics (BOMD) simulations have been per-
formed with the CP2K/Quickstep package,37 using the hybrid
Gaussian and plane wave method. The Perdew−Burke−
■ RESULTS AND DISCUSSION
In our recent work26 we have shown that this amorphous silica/
Ernzerhof (PBE) exchange−correlation density functional was water interface is characterized by a strongly adsorbed layer of
used. Goedecker−Teter−Hutter pseudopotentials,38,39 a dou- water at the interface. In particular, three zones were identified,
ble-ζ DZVP plus polarization Gaussian basis set for the orbitals with increasing “local ordering” of the interfacial water, namely,
and a density cutoff of 400 Ry were used. Only the Γ point was Q3-isolated silanols and concave and convex geminal Q2
considered in a supercell approach. The box dimensions are silanols.
12.77 × 17.64 × 25.17 Å3. Periodic boundary conditions are A picture showing the local environment for a few selected
applied in all directions of space. Dynamics were conducted in groups can be found in Figure 3. For our pKa analysis we have
the canonical NVT ensemble (after 9 ps of equilibration for chosen different types of sites that are representative of
geminal 1, 5 ps for geminal 2 and vicinal 2, 10 ps for vicinal 1 different Qn sites (namely silanols with different number of
D DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

bridging oxygens) and different types of local environment and


solvation. In particular, we chose two different types of
geminals, namely Geminal 1 (panel a in Figure 3), which is a
convex Q2 silanol, and Geminal 2, which is a concave Q2 silanol
(panel b in Figure 3). From our detailed analysis of the
solvation structure at the interface with the amorphous silica,
we could already suggest that the convex Q2 silanol represents
possibly low acidity sites.26 Additional investigated sites include
two Q3 vicinals with very diffferent environments and a Q3- Figure 4. Schematic view of the solvation of the different surface sites
isolated site. The two vicinals are reported in Figure 3, panels c of silica (amorphous and crystalline quartz) obtained from our AIMD
and d, and are characterized by two distinct exposures to simulations: amorphous Q2 convex geminal (red); amorphous Q3-
isolated (black); quartz Q2 out-of-plane (OP) (green). Left: neutral
solvent. Vicinal 1 (panel c) corresponds to a hydroxyl group Si−OH form of the silanol site. Right: deprotonated Si−O− form of
fully immersed into liquid water. It is H-bonded to a the silanol site (conjugated base, silanolate). Labels show the average
neighboring silanol through a bridging water molecule and is number of hydrogen bonds formed between the Si−OH (left)/Si−O−
part of a four-membered H-bond ring. In particular, it receives (right) and their surrounding (silanols and water molecules).
and donates 2 H-bonds with two water molecules. Conversely,
Vicinal 2 (panel d) is a completely buried site, almost Si−OH (geminal 1) donates a strong hydrogen bond to water
inaccessible to solvent. It only shares its proton directly with (1.57 Å, (Si)O−H···Ow intermolecular distance), stronger than
a nearby silanol. Finally, the isolated site is depicted in panel e the one formed by the concave Q2 Si−OH (1.69 Å) and slighly
of Figure 3 and is H-bonded with one water molecule. stronger than that formed by the Q2 out-of-plane silanol from
We provide here a quantitative estimate of these site acidities, the (0001) quartz crystalline surface (1.60 Å).
applying our proton insertion approach19−22 as in ref 17. pKa This is also clear from the (Si−O)H···Ow distances reported
values will be compared to those calculated for the in-plane and in Table 2 taken from the position of the first peak in the radial
out-of-plane Q2 silanols of the crystalline quartz−water distribution functions between silanol H atoms and Ow
interface. oxygens of water (curves reported in Figure 5). Maybe, more
The pKa values found are 2.9 for the convex Q2 silanol and importantly, from these curves one can obtain the information
8.9 for the concave Q2 silanol. This first result suggests that the concerning the position of the second peaks in the H···Ow
acidity is not depending on the type of site only (Q2 for both radial distribution functions and their microscopic interpreta-
cases) but is strongly influenced by the local environment, tion. The second peak of H···Ow for the amorphous Q2 silanol
namely, by the solvent accessibility and H-bond network (Figure 5, red line) corresponds to the interaction of this site
around the OH site. A similar conclusion also holds for the Q3 with three nearby silanols and four water molecules from the
sites. Indeed, vicinals with different local environments can have liquid. A very similar strong coordination from the environment
very different acidities. The Q3 vicinal 1, which is characterized was also found for the Q2 out-of-plane silanol of the crystalline
by steady hydrogen bonds with interfacial water, has a pKa value (0001) quartz, as it is also surrounded by two to three
of 2.1, whereas another vicinal with its hydrophobic environ- neighboring SiOH groups and by two to three water molecules.
ment, namely, our Q3 vicinal 2, has a much higher pKa value of The high computational cost of our AIMD proton-insertion
16.2. Finally, the Q3-isolated silanol has a pKa of 10.3. (A method for pKa calculations prevents us from determining the
summary of the pKa analysis is reported in Table 1). intrinsic acidity of each surface site. However, we could
The (convex) geminal 1 and the vicinal 1 are thus very acidic, envisage estimating acidic heterogeneities within the same type
even more acidic than the Q2 out-of-plane silanols at the α- of silanol group by looking at the local environment around the
quartz/water interface (pKa = 5.6),17 whereas the geminal 2 and protonated and deprotonated forms. The rdfs of all the convex
the isolated sites have a relatively high pKa that is similar to the geminals are reported in Figure 6. The red lines (continuous
one of the silanols in silicic acid (pKa ≈ 9.9 from our previous and dashed) refer to a pair sharing the same Si, and the blue
AIMD simulations17). Finally, the pKa of vicinal 2 is outside the lines (continuous and dashed) refer to another pair sharing the
water window and would possibly not be probed in titration same Si. We only have two concave geminals at the surface,
experiments. Overall, we recover a bimodal behavior at the which is clearly a very limited statistics. A comparison of the
water/amorphous silica with a first acidity around 2−3 pKa rdfs is nonetheless reported in Figure 7. Note that the color
units and a second one around 9−10. code in Figures 6 and 7 to the same code chosen in the
Now the key question is where does the strong/weak acidity previous figures; namely, the continuous red line corresponds
come from? To provide an answer to such a question, we have to the convex geminal 1 and the continuous orange line is for
analyzed in detail the solvation structure of both the protonated the concave geminal 2.
and the deprotonated forms of the silanols. To guide the The position of the first peaks in the H/O rdfs (Figures 6
structural analysis, a scheme is reported in Figure 4, which and 7), corresponding to proton sharing with a nearby hydroxyl
shows a silanol in the protonated form (left panel) and a silanol (from silanol or water), are all shifted to larger distances with
in the deprotonated form (right panel) and their hydrogen respect to the continuous red line of convex geminal 1. From
bond networks obtained from our AIMD simulations. In the considering only the H/O rdfs, the weaker H-bond donor
scheme, the average hydrogen bond distances for the Q2- and character could reflect lower acidity. Considering that for more
Q3-isolated silanols at the amorphous silica/water interface, and acidic sites, the OH bond is more polarized implying a greater
for the Q2 aqueous quartz out-of-plane silanols,17 are reported negative charge on the oxygen, its H-bond acceptor character
for discussion. Note that for clarity, we do not report here all should be enhanced. Indeed, from the O/H rdfs, the more
the sites. acidic convex geminal 1 and the less acidic concave geminal 2
On the basis of the average distances reported in Table 2 for correspond to the extreme H-bond acceptor behaviors, the
the neutral form of the amorphous silica sites, the convex Q2 more acidic site being a very strong acceptor and the less acidic
E DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

Table 2. Positions and Heights of the First Two Peaks from the Radial Distributions Functions (rdf) around the Q2-Geminal 1
(Gem. 1), Q2-Geminal 2 (Gem. 2), Q3-Isolated (Iso.), Q3-Vicinal 1 (Vic. 1), Q3-Vicinal 2 (Vic. 2), and Out-of-Plane (OP)
Silanols
first peak second peak
rmax(Å) Imax rmin(Å) Imin rmax(Å) Imax rmin(Å) Imin
H−O Gem. 1 1.57 3.65 2.14 0 3.14 1.40
3.42 2.18 4.10 0.57
H−O Gem. 2 1.69 2.78 2.24 0 3.42 1.76 3.66 0.75
H−O Iso. 1.60 3.83 2.19 0 3.25 1.30 3.47 0.96
H−O Vic. 1 1.74 2.73 2.28 0 3.23 1.44 3.76 0.73
H−O Vic. 2 1.65 3.44 2.16 0 4.03 0.97 4.35 0.44
H−O OP 1.55 4.10 2.03 0 2.82 2.67 3.22 0.66
O−H Gem. 1 1.69 1.42 2.56 0.11 3.10 1.54 3.60 0.31
1.89 1.49
O−H Gem. 2 1.87 0.61 2.46 0.12 2.99 1.08 3.52 0.42
O−H Iso. 1.78 1.34 2.39 0.10 3.22 1.70 3.66 0.76
O−H Vic. 1 2.21 0.40 2.42 0.31 3.07 1.60 4.40 0.53
O−H Vic. 2 3.41 0.69 4.27 0.19
O−H OP 1.58 2.34 2.16 0 3.08 0.93 3.31 0.74
3.53 1.30 3.77 0.64
O−−H Gem. 1 1.57 4.08 2.23 0.06 3.06 1.16 3.79 0.36
O−−H Gem. 2 1.63 3.85 2.11 0.00 3.02 1.41 3.40 0.67
O−−H Iso. 1.70 3.66 2.48 0.23 3.13 1.36 3.34 0.87
O−−H Vic. 1 1.58 3.39 2.36 0.15 3.11 1.47 4.34 0.39
O−−H Vic. 2 1.50 3.36 2.14 0.04 3.90 0.65 4.74 0.35
O−−H OP 1.53 4.77 2.28 0.08 3.54 1.81 4.17 0.49

Figure 5. Radial distribution functions (rdf) around SiOH and SiO− Figure 6. Comparison of the local environment for convex geminal
on amorphous silica or on α-quartz. The black line correspond to the silanols: (left) donated hydrogen bonds from silanol H atoms; (right)
Q3-isolated sites, the red and orange ones to the Q2-convex (geminal accepted hydrogen bonds from water H atoms. The red lines
1) and concave (geminal 2) geminals, respectively. Blue lines represent (continuous and dashed) refer to a pair sharing the same Si, and the
the Q3-vicinals, dark blue for vicinal 1 and light blue for the buried blue lines (continuous and dashed) refer to another pair sharing the
vicinal 2. The green line is used for the out-of-plane silanol on quartz. same Si.
Top panel: rdfs. Lower panel: integrated densities.
If we now turn to the conjugated silanolate bases (Figure 4,
site being a poor one. All the remaining geminals lie in right), we observe that they are strongly and equally stabilized
between. Conclusions on their relative acidities are thus not for the convex Q2 Si−O− of amorphous silica and the Q2 out-
straightforward. Nonetheless, because the two concave sites are of-plane site of crystalline (0001) α-quartz. Note, e.g., that for
not as good H-bond acceptors as the convex ones (maximum 1- the amorphous convex Q2 SiO−, one H-bond is accepted by
H-bond is accepted, Figure 7, right panels) we expect both one water (distance 1.64 Å) and two H-bonds are accepted by a
concave silanols to exhibit the similar less acidic pKa. Note that nearby silanol (1.64 Å). A pictorial view of the local
the higher acidity correlated to a stronger H-bond acceptor environment for the deprotonated convex geminal Q2 can be
character also holds for acidic Q3 vicinal 1 vs basic Q3 vicinal 2 found in Figure 8, panel a.
on one side and for acidic Q2 out-of-plane geminal vs less acidic Similarly for the out-of-plane Q2 SiO− on the quartz
Q2 in-plane geminal on quartz on the other side. crystalline surface: two H-bonds are accepted from water
F DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

hydrogen bonds. The isolated site seems not to fit within this


scheme. Indeed, its conjugated base has a coordination of 3.5;
however, it forms on average the weakest H-bonds with these
water molecules (1.82 Å), the weakest among all the H-bonds
stabilizing the other deprotonated silanols (also true when
looking for the strongest H-bond among the 3.5 possible). This
type of stabilization of the isolated silanolate may resemble that
of a silicic acid molecule where the base stabilization can only
come from solvent molecules, but not from other silanol
groups. A pictorial view of the local environment for the
deprotonated Q3 site can be found in Figure 8, panels c, d, and
e.
In addition to the radial distribution functions we have also
Figure 7. Comparison of the local environment for concave geminal investigated the orientational distribution of the water around
silanols: (left) donated hydrogen bonds from silanol H; (right) the deprotonated silanols. We find that the orientational
accepted hydrogen bonds from water H. Two OH groups sharing the distribution is very similar for all the cases (geminal, vicinal,
same Si are investigated. The continuous line refers to an OH group
isolated) and the maxima are only displaced by a few degrees.
that donates an H-bond to water (similar to convex silanols), and the
dashed line, to an OH group which donates an H-bond to a nearby We can conclude that water orientation around the
silanol. deprotonated oxygen is not a discriminating element character-
izing the conjugated base.
In some recent analysis of a model of amorphous silica
(1.69 Å) and one H-bond is accepted from a nearby silanol surface the presence of hydrophobic−hydrophilic patches has
(1.53 Å). Both the amorphous Q2 SiO− and the quartz out-of- been suggested in connection to the siloxane density.29,30 It
plane Q2 SiO− are strongly stabilized by the environment and thus may be interesting to comment on the eventual correlation
exhibit an overall 3-fold coordination of the deprotonated between the density of siloxane groups in vicinity of the silanol
oxygen. groups and their pKa’s. Our analysis shows that low pKa is
On the contrary, the concave Q2 SiO− site is not so strongly possible for both Q2 (geminal 1) and Q3 (vicinal 1) sites, which
stabilized. It only accepts two hydrogen bonds, which are also means it is independent of the number of siloxane bridges.
weaker than those stabilizing the deprotonated convex site. A Convex and concave geminals have the same number of
pictorial view of the local environment for the deprotonated siloxane bridging groups; however, they exhibit a quite large
concave geminal Q2 can be found in Figure 8, panel b. difference in the pKa. This would suggest that the environment,
Let us now turn to the vicinals and the isolated sites. Here we namely, the local topology and solvent accessibility are the key
investigated different types of environments on the Q3 sites. element to determine silanols acidity. To quantify the local
For the vicinals a key element to interpret the acidity can be hydrophobicity, we have calculated the rdf of the siloxane O
again found in the conjugated base stabilization. Indeed, also around the silanols with the solvent H, or the nearby silanol H.
here the acidic site (vicinal 1) has a conjugated base that is These are reported in Figure 9. We find that the siloxanes are
strongly stabilized by three hydrogen bonds. Instead, for vicinal quite hydrophobic, and no hydrogen bond is formed between
2 (higher pKa) the conjugated base is only stabilized by two the siloxane oxygens and protons from the environment

Figure 8. Representative snapshots of the local structures around the deprotonated SiO− silanols at the water/amorphous silica interface. Hydrogen
bonds appear in white dashed lines.

G DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

■ CONCLUSIONS
The first conclusion from our structural study is that the silanol
acidity does not depend on the actual n value of the Qn site, at
least not solely. We find pKa values of 2.9 and 8.9 for a convex,
respectively, concave Q2 geminal silanol and pKa of 2.1 and
16.2 for Q3 vicinal silanols with more hydrophilic or
hydrophobic environments, respectively. Finally, we find pKa
value of 10.3 for the isolated Q3 site. The local environment has
a strong impact on the silanols acidity and the geminals are a
good example where different environments (local topology
(nano-roughness) and hydrophobicity/philicity) can lead to
very different pKa. Some convex Q2 silanol at the amorphous/
water interface can form strong hydrogen bonds with water, so
exhibiting a major tendency of the proton to be shared with
water, thus pointing to a higher acidity (pKa value 2.9). On the
contrary, the SiO− deprotonated form of this convex geminal
can be more easily stabilized by the presence of nearby OH
groups from other surface silanols, which in this nested
Figure 9. Local environment for siloxane groups. Upper panel: rdf configuration are particularly prone to solvate the silanolate
between the siloxane oxygen and the water hydrogens. Bottom panel: base. Overall, the convex conjugated base is 3-fold coordinated,
integrated numbers. Black curve: siloxane oxygens around the isolated. whereas the concave conjugated base is only 2-fold
Green curve: siloxane oxygens around the quartz geminal. Red curve:
siloxane oxygens around the convex Q2-Geminal 1. Orange curve:
coordinated. The same conclusion seems to hold for the
siloxane oxygens around the concave Q2-Geminal 2. Black: Q3-Iso. vicinals, where the acidic site has a 3-fold coordinated
Dark blue: Q3-Vicinal 1. Light blue. Q3-Vicinal 2. All are at the water/ conjugated base, whereas the more basic site has a 2-fold
amorphous silica interface. Green: siloxane oxygens around the Q2- coordinated base. We believe this special anion solvation is the
out-of-plane site at the (0001) α quartz surface. key element to explain the higher acidity of some of the Q2
geminal and some of the Q3 vicinal silanols at the amorphous
aqueous silica surface. The corollary is that both Q2-geminal 2
and the Q3-isolated silanols form weaker H-bonds with the
(namely protons from other silanols or from water). In all the solvent, and more importantly their conjugated bases are not
cases (geminal, isolated, vicinal, crystalline quartz) the first peak stabilized by a substantial H-bond network. This is the key
in the radial distribution function appears around 3 Å. For the element for the basic character of the Q3-isolated silanol at the
convex geminal (red curve in Figure 9) the broad band around silica amorphous aqueous interface, and the intrinsic flexibility
3.0−3.5 Å is due to protons from the surrounding silanols, of the silica framework that stabilizes the overall H-bond
whereas for the isolated (black line) the peak originates from network on the surface.
water molecules. Our analysis supports and explains the acidity bimodal
Leung et al.15 have also explored the influence of H-bonding character of amorphous (this work) and crystalline silica
between Q2 silanols at the (100) β-cristobalite surface, where silanols (this work and ref 17), as the result of local H-bond
the surface is composed of Q2 silanols sharing intramolecular organization and strength, that can lead to a strong stabilization
H-bonds with bond lengths of 1.70 Å in vacuum.31 pKa = 7.6 of the silanolate base. We have indeed shown that the essential
was calculated. The possibility of forming an H bond was then element resides in the stabilization of the conjugated base of the
eliminated by replacing the OH group by a H atom, so creating silanol site, by local interactions with the surface and water. The
an isolated Q2 silanol. The pKa was consequently found to H-bond network that can be formed around the conjugated
increase to 8.9, so that Leung et al. concluded that H-bonding base is the key factor to understand the ranking in acidity of
was responsible only for a small decrease of the pKa. In silica silanol sites.


particular, the simple removal of a hydrogen bond around the
Q2 silanols at the (100) β-cristobalite surface could not produce AUTHOR INFORMATION
a difference of 4 units between different Q2 sites and therefore
could not explain a bimodal character of that surface. However, Corresponding Author
in comparing our result to the cristobalite, one should also take *M. Sulpizi. E-mail: sulpizi@uni-mainz.de.
into account the role of the surface flexibility. Indeed, Musso et Notes
al.32 pointed out that for (100) cristobalite, the formation of H- The authors declare no competing financial interest.


bonds between surface OH groups induces a large variation in
the flexible Si−O−Si angles with a sizable rearrangement of the ACKNOWLEDGMENTS
structure across the whole layer thickness. It was also found
that the layer thickness strongly influences the number of H- We thank Johannes Luetzenkirchen and Alvaro Cimas for
bonds formed by the silanols on the surface.32 So one could useful discussions and careful reading of the manuscript.
speculate that H-bond formation on cristobalite (100) is Computer time from HRLS (Hermit) is greatly acknowledged
energetically more expensive than on quartz (0001) and (project number 2DSFG).
amorphous silica, leading to a poorer associated base
stabilization. Certainly, the amorphous surface is more flexible
than crystalline surfaces, so that relaxation to maximize H-bond
■ REFERENCES
(1) Dong, Y.; Pappu, S. V.; Xu, Z. Detection of Local Density
formation should be easier. Distribution of Isolated Silanol Groups on Planar Silica Surfaces Using

H DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX
The Journal of Physical Chemistry C Article

Nonlinear Optical Molecular Probes. Anal. Chem. 1998, 70, 4730− (23) Garofalini, S. H. Molecular Dynamics Computer Simulations of
4735. Silica Surface Structure and Adsorption of Water Molecules. J. Non-
(2) O’Reilly, J. P.; Butts, C. P.; I’Anson, I. A.; Shaw, A. M. Interfacial Cryst. Solids 1990, 120, 1.
pH at an Isolated Silica-Water Surface. J. Am. Chem. Soc. 2005, 127, (24) Tielens, F.; Gervais, C.; Lambert, J. F.; Mauri, F.; Costa, D. Ab
1632−1633. Initio Study of the Hydroxylated Surface of Amorphous Silica: A
(3) Fisk, J. D.; Batten, R.; Jones, G.; O’Reill, J. P.; Shaw, A. M. pH Representative Model. Chem. Mater. 2008, 20, 3336.
Dependence of the Crystal Violet Adsorption Isotherm at the Silica- (25) Folliet, N.; Gervais, C.; Costa, D.; Laurent, G.; Babonneau, F.;
Water Interface. J. Phys. Chem. B 2005, 109, 14475−14480. Stievano, L.; Lambert, J.-F.; Tielens, F. A Molecular Picture of the
(4) Lutzenkirchen, J.; Boily, J.; Lovgren, L.; Sjoberg, S. Limitations of Adsorption of Glycine in Mesoporous Silica through NMR Experi-
the Potentiometric Titration Technique in Determining the Proton ments Combined with DFT-D Calculations. J. Phys. Chem. C 2013,
Active Site Density of Goethite Surfaces. Geochim. Cosmochim. Acta 117, 4104.
(26) Cimas, A.; Tielens, F.; Sulpizi, M.; Gaigeot, M.; Costa, D. The
2002, 66, 3389−3396.
Amorphous Silica-Liquid Water Interface Studied by Ab Initio
(5) Lutzenkirchen, J. On Derivatives of Surface Charge Curves of
Molecular Dynamics (AIMD): Local Organization in Global Disorder.
Minerals. J. Colloid Interface Sci. 2005, 290, 489−497.
J. Phys.: Condens. Matter 2014, 26, 244106.
(6) Ong, S.; Zhao, X.; Eisenthal, K. B. Polarization of Water
(27) Tielens, F.; De Proft, i.; Geerlings, P. Density Functional Study
Molecules at a Charged Interface: Second Harmonic Studies of the on the Conformation and Energetics of Silanol and Disiloxane. J. Mol.
Silica/Water Interface. Chem. Phys. Lett. 1992, 191, 327−335. Struct.: THEOCHEM 2001, 542, 227.
(7) Zhao, X.; Ong, S.; Wang, H.; Eisenthal, K. B. New Method for (28) Leydier, F.; Chizallet, C.; Chaumonnot, A.; Digne, M.; Soyer, E.;
Determination of Surface pKa Using Second Harmonic Generation. Quoineaud, A.-A.; Costa, D.; Raybaud, P. Bronsted Acidity of
Chem. Phys. Lett. 1993, 214, 203−207. Amorphous Silica-Alumina: The Molecular Rules of Proton Transfer.
(8) Du, Q.; Freysz, E.; Shen, Y. Vibrational Spectra of Water J. Catal. 2011, 284, 215.
Molecules at Quartz/Water Interfaces. Phys. Rev. Lett. 1994, 72, 238. (29) Hassanali, A. A.; Singer, S. J. A Model for the Water/
(9) Ostroverkhov, V.; Waychunas, G. A.; Shen, Y. R. New Amorphous Silica Interface: the Undissociated Surface. J. Phys. Chem.
Information on Water Interfacial Structure Revealed by Phase- B 2007, 111, 11181.
Sensitive Surface Spectroscopy. Phys. Rev. Lett. 2005, 94, 046102. (30) Hassanali, A. A.; Zhang, H.; Shin, Y. K.; Knight, C.; Singer, S. J.
(10) Darlington, A. M.; Gibbs-Davis, J. M. Bimodal or Trimodal? The Dissociated Amorphous Silica Surface: Model Development and
The Influence of Starting pH on Site Identity and Distribution at the Evaluation. J. Chem. Theory Comput. 2010, 6, 3456.
Low Salt Aqueous/Silica Interface. J. Phys. Chem. C 2015, 119, 16560. (31) Iarlori, S.; Ceresoli, D.; Bernasconi, M.; Donadio, D.; Parrinello,
(11) Rimola, A.; Costa, D.; Sodupe, M.; Lambert, J.; Ugliengo, P. M. Dehydroxylation and Silanization of the Surfaces of -Cristobalite
Silica Surface Features and Their Role in the Adsorption of Silica: An Ab Initio Simulation. J. Phys. Chem. B 2001, 105, 8007.
Biomolecules: Computational Modeling and Experiments. Chem. (32) Musso, F.; Sodupe, M.; Corno, M.; Ugliengo, P. H-Bond
Rev. 2013, 113, 4216. Features of Fully Hydroxylated Surfaces of Crystalline Silica
(12) Sahai, N.; Rosso, K. Linking Molecular Modeling to Surface Polymorphs: A Periodic B3LYP Study. J. Phys. Chem. C 2009, 113,
Complexation Modeling. Interface Sci. Technol. 2006, 11, 359. 17876.
(13) Morrow, B. A.; McFarlan, A. J. Infrared and Gravimetric Study (33) Gaigeot, M.-P.; Sprik, M.; Sulpizi, M. Oxide/Water Interfaces:
of an Aerosil and a Precipitated Silica Using Chemical and Hydrogen/ How the Surface Chemistry Modifies Interfacial Water Properties. J.
Deuterium Exchange Probes. Langmuir 1991, 7, 1695. Phys.: Condens. Matter 2012, 24, 124106.
(14) Chuang, I. S.; Maciel, G. E. A Detailed Model of Local Structure (34) Tazi, S.; Rotenberg, B.; Salanne, M.; Sprik, M.; Sulpizi, M.
and Silanol Hydrogen Bonding of Silica Gel Surfaces. J. Phys. Chem. B Absolute Acidity of Clay Edge Sites from Ab-Initio Simulations.
Geochim. Cosmochim. Acta 2012, 94, 1.
1997, 101, 3052.
(35) Churakov, S.; Labbez, C.; Pegado, L.; Sulpizi, M. Intrinsic
(15) Leung, K.; Nielsen, I.; Criscenti, L. Elucidating the Bimodal
Acidity of Surface Sites in Calcium Silicate Hydrates and Its
Acid-Base Behavior of the Water-Silica Interface from First Principles. Implication to Their Electrokinetic Properties. J. Phys. Chem. C
J. Am. Chem. Soc. 2009, 131, 18358−18365. 2014, 118, 11752.
(16) Duval, Y.; Mielczarski, J. A.; Pokrovsky, O. S.; Mielczarski, E.; (36) Hummer, G.; Szabo, A. Calculation of Free-Energy Differences
Ehrhardt, J. J. Evidence of the Existence of Three Types of Species at from Computer Simulations of Initial and Final States. J. Chem. Phys.
the Quartz-Aqueous Solution Interface at pH 0−10: XPS Surface 1996, 105, 2004.
Group Quantification and Surface Complexation Modeling. J. Phys. (37) CP2K/Quickstep package; CP2K developers group, http://
Chem. B 2002, 106, 2937−2945. www.cp2k.org (2015).
(17) Sulpizi, M.; Gaigeot, M.; Sprik, M. The Silica-Water Interface: (38) Goedecker, S.; Teter, M.; Hutter, J. Separable Dual-Space
How the Silanols Determine the Surface Acidity and Modulate the Gaussian Pseudopotentials. Phys. Rev. B: Condens. Matter Mater. Phys.
Water Properties. J. Chem. Theory Comput. 2012, 8, 1037. 1996, 54, 1703.
(18) Iler, R. K. Chemistry of Silica - Solubility, Polymerization, Colloid (39) Hartwigsen, C.; Goedecker, S.; Hutter, J. Relativistic Sseparable
and Surface Properties and Biochemistry; John Wiley & Sons: New York, Dual-Space Gaussian Pseudopotentials from H to Rn. Phys. Rev. B:
NY, U.S., 1979. Condens. Matter Mater. Phys. 1998, 58, 3641.
(19) Sulpizi, M.; Sprik, M. Acidity Constants From Vertical Energy
Gaps: Density Functional Theory Based Molecular Dynamics
Implementation. Phys. Chem. Chem. Phys. 2008, 10, 5238−5249.
(20) Cheng, J.; Sulpizi, M.; Sprik, M. Redox Potentials and pKa’s for
Benzoquinone from Density Functional Theory-based Moleular
Dynamics. J. Chem. Phys. 2009, 131, 154504.
(21) Sulpizi, M.; Sprik, M. Acidity Constants from DFT-based
Molecular Dynamics Simulations. J. Phys.: Condens. Matter 2010, 22,
284116.
(22) Costanzo, F.; Sulpizi, M.; della Valle, R.; Sprik, M. The
Oxidation of Tyrosine and Tryptophan Studied by a Molecular
Dynamics Normal Hydrogen Electrode. J. Chem. Phys. 2011, 134,
244508.

I DOI: 10.1021/acs.jpcc.5b02854
J. Phys. Chem. C XXXX, XXX, XXX−XXX

You might also like