Liquid flow along a solid surface reversibly alters interfacial chemistry

See allHide authors and affiliations

Science  06 Jun 2014:
Vol. 344, Issue 6188, pp. 1138-1142
DOI: 10.1126/science.1253793

Monitoring water interfaces in motion

Water behaves differently at interfaces—where it meets the air, or a solid surface—than it does in the middle of the liquid. Past laboratory studies of this phenomenon have mainly focused on still samples, despite the fact that in natural settings such as rivers and rain, the water moves along the surfaces. Lis et al. used a microfluidics apparatus and a spectroscopy technique called sum frequency generation to study the effects of flow on aqueous chemistry at silica and fluorite surfaces (see the Perspective by Waychunas). The flow of fresh water along the surfaces disrupts the equilibrium of dissolved ions, substantially changing the surface charge and the molecular orientation of the water at the interface.

Science, this issue p. 1138; see also p. 1094


In nature, aqueous solutions often move collectively along solid surfaces (for example, raindrops falling on the ground and rivers flowing through riverbeds). However, the influence of such motion on water-surface interfacial chemistry is unclear. In this work, we combine surface-specific sum frequency generation spectroscopy and microfluidics to show that at immersed calcium fluoride and fused silica surfaces, flow leads to a reversible modification of the surface charge and subsequent realignment of the interfacial water molecules. Obtaining equivalent effects under static conditions requires a substantial change in bulk solution pH (up to 2 pH units), demonstrating the coupling between flow and chemistry. These marked flow-induced variations in interfacial chemistry should substantially affect our understanding and modeling of chemical processes at immersed surfaces.

The chemistry taking place at the interface between a solid material and an aqueous solution is relevant for a variety of disciplines, including geology, environmental sciences, and catalysis (13). The local chemical composition at the interface strongly influences the reactivity of the system, as has been demonstrated, for example, in geological studies of weathering (4). Similarly, the abrasion and dissolution of materials immersed in aqueous solutions is at the heart of environmental concerns. Dissolution of ocean organisms’ shells and skeletons stemming from increasing acidification of water could lead to potentially devastating consequences for marine life (5). Accurate knowledge of the composition of both the solid material and the aqueous solution locally at the surface is essential to understand, model, and predict these interfacial chemical processes.

Previous studies have shown that the structure of liquid water at a solid interface is different from that of the bulk phase (68) and can resemble the ice structure (9) because of the specific physico-chemical properties of the surface (e.g., charge, morphology, wetting properties). Generally speaking, interfacial water possesses a more structured hydrogen bonding network than bulk water. Among various factors, the solid surface holding a net electric charge acts to align the static dipole of water molecules at the surface (Fig. 1A, top). The length over which the electric field extends into the solution from the surface is referred to as the Debye length (10) and correlates with the distance from the surface that water retains its preferential alignment (~ 1 to 10 nm).

Fig. 1 Experimental scheme and the effect of flow on the SFG signal at the CaF2-water interface.

(A) (Top) Schematic representation of the molecular arrangement at the charged solid-water interface in presence of electrolytes. For a negatively charged surface, the water dipole moments are aligned away from the interface. Cations (red spheres) form a compact layer adjacent to the surface (Stern layer) with water and partially screen the negative surface charge. Anions (blue spheres) are mostly repelled from the surface. EDL, electrical double layer. (Bottom) Schematic representation of the SFG experimental geometry with respect to the flow cell. Incident VIS and IR beams generate sum frequency light at the solid-liquid interface. v, flow velocity at the center of the channel; h, distance from center of the channel to the surface. (B) SFG spectrum in the OH stretch region of the CaF2-water interface at pH 3 (1 mM HCl) under static (black) and flow (red) conditions. a.u., arbitrary units. (C) SFG spectrum of the CaF2-water interface at pH 12 (10 mM NaOH). (D) Plot of the time dependence of the integrated SFG intensity (black circles) for the water-CaF2 interface at pH 3 under multiple flow on-off cycles (blue curve). The red solid line on the SFG data is a five-point average to guide the eye. In all panels, the SFG intensity is normalized by the maximal intensity of the flow-off SFG curve at pH 12 shown in Fig. 1C. Further information on the OH peak center positions can be found in the supplementary materials (33).

In addition to causing water alignment at the surface, the charge present at the interface also attracts ions from the solution to the interface, causing surface-charge screening and leading to the formation of the so-called electrical double layer (10). Accordingly, the composition, as well as the structure, of the fluid adjacent to the surface can vary substantially compared with the bulk (1114). Molecular dynamics simulations have shown, for instance, that the interfacial pH can be quite different from the bulk (11). Although substantial progress regarding interfacial composition and molecular organization has been achieved, in nearly all of these studies on solid-liquid interfacial chemistry, the liquid is not collectively in motion (that is, at rest) (6, 9, 12, 1422). This is despite the fact that macroscopic fluid flow is ubiquitous in nature (e.g., oceans, rivers, and rainfall).

In this work, we used vibrational sum frequency generation (SFG) spectroscopy (9, 13, 16, 23) to demonstrate that liquid flow along an interface substantially, yet reversibly, perturbs the interfacial chemistry and water arrangement at the nanoscale. The observation of this effect on two different types of solid interfaces suggests that this is a generic phenomenon. SFG is a class of second-order nonlinear optical spectroscopy that provides surface-specific vibrational spectra and has been successfully applied to elucidate interfacial water organization at solid interfaces (6, 9, 12, 14, 15, 1922, 24, 25). In an SFG experiment, an infrared (IR) laser pulse is overlapped at the surface in both time and space with a visible (VIS) pulse, leading to the generation of new photons at the sum-frequency of the IR and VIS frequencies (Fig. 1A, bottom). Vibrational information is obtained when the IR frequency is tuned into resonance with a vibrational mode; for example, the OH stretch vibration of water molecules. SFG owes its surface specificity to the selection rule that requires the inversion symmetry (present in isotropic liquids such as water) to be broken to generate a signal, which precisely occurs at an interface. The SFG signal for water is then generated only by those molecules that are preferentially aligned; for example, by the electrostatic field at the solid-water interface (6, 21).

The magnitude of the SFG signal scales with the degree of alignment of those water molecules and is thus tightly related to the surface charge (26, 27). From the SFG vibrational spectrum of interfacial water (plotted as SFG intensity versus IR frequency), it is thus possible to glean information on the orientation of the water molecules and water structure in the interfacial region. We employ SFG to study the interface between a solid and various aqueous solutions within a microfluidic flow cell (Fig. 1A, bottom, and fig. S1). The shear rate in the flow cell is ~104 s−1 (typical Reynolds number of 1000), ensuring a fully developed laminar flow at the center of the cell (fig. S1). All SFG experiments reported here were performed under standard atmospheric conditions (e.g., room temperature and air environment), using s, s, p polarizations for SFG, VIS, and IR fields, respectively. Further details about the experimental setup can be found in the supplementary materials.

The first solid surface that we studied was polycrystalline calcium fluoride (CaF2, Crystran), with a surface RMS roughness of ~1.5 nm over a 10 μm by 10 μm area. Under ambient conditions the CaF2 surface assumes positive, neutral, or negative charge, depending on the pH of the contacting solution (28, 29). The point of zero charge (pzc) of CaF2—that is, the pH at which the surface is charge-neutral—has been shown to vary from pH 6 to 10, depending on the sample preparation and measurement conditions (15, 28, 30). For a neutral CaF2 surface, no electrostatic interaction exists to preferentially align water molecules at the liquid-solid interface. Under this condition, a near-zero SFG intensity has been observed (14, 15, 31). Black traces in Fig. 1, B and C, show the SFG response of water at the positively [~+60 mV surface potential at pH 3 (28)] and negatively [~–30 mV surface potential at pH 12 (28, 30)] charged CaF2 surface, with water molecules preferentially aligned with their dipoles toward and away from the surface, respectively. Figure S2 shows that between these two pH values, an inversion of the molecular alignment occurs when crossing the pzc (32).

At both pH 3 and 12, the SFG vibrational signature of the water at rest displays a broad resonance centered at 3160 cm−1 and attributed to the OH stretching mode (14, 15, 23, 33) (fig. S2). For the low-pH solution, we observe a ~100% increase in the SFG intensity (Fig. 1B, red curve) when flowing water along the surface under laminar conditions at a modest shear rate (104 s−1). Conversely, for a basic solution, we observe the opposite behavior—a 50% decrease in the SFG signal—in addition to a shift of the primary SFG spectral peak to 3200 cm−1 (Fig. 1C, red curve). For comparison, to achieve such a drop in the SFG intensity for a static system, a pH 10 liquid is required (fig. S3A); that is, the effect of the flow is equivalent to a 100-fold decrease of the bulk OH concentration. From Fig. 1, B and C, it is clear that flowing the solution along a charged solid surface modifies the vibrational response in opposite ways depending on surface charge. As shown in Fig. 1D, the changes recorded in the integrated SFG intensity are only very weakly dependent on the flow speed within the range of shear rates accessible with our experimental setup (5 × 102 to 104 s−1). The changes in SFG intensity are also independent of the flow direction (fig. S3B). Moreover, the flow-induced changes occur on the ~10-s time scale of the experiment and are reversible when flow is turned off (Fig. 1D). Clearly, a major rearrangement of the interfacial water ensues as a result of the flow.

The observed flow-induced changes in the interfacial water response are not specific to CaF2. Figure 2 shows SFG spectra of a fused silica interface in contact with a 10 mM NaCl solution under flow and at rest for various pH values. Fused silica is glass consisting of silicon dioxide (SiO2), which represents 28% of Earth’s crust and has its pzc at pH ~2 (18). Therefore, a SiO2 surface will be increasingly negatively charged as pH rises above 2. The SFG spectrum of the (negatively charged) fused silica-water interface at pH 6.5 is shown in black in Fig. 2A and displays a maximum intensity at 3200 cm−1 with a shoulder at ~3460 cm−1 (see fig. S4 for peak details) (18, 21). Similar to observations for the negatively charged CaF2, the presence of flow (shear rate = 6 × 103 s−1) causes a very large (50%) decrease in SFG intensity (Fig. 2A, red curve). However, at both low and high pH, the reduction in SFG intensity upon flow is very small (Fig. 2, B and C), in contrast to what is observed for CaF2. It takes tens of minutes for the SFG intensity to reach the static value after stopping the flow (Fig. 2D).

Fig. 2 The effect of flow on the interfacial vibrational spectra of water at the SiO2-water interface.

(A) pH 6.5 (10 mM NaCl), (B) pH 3 (1 mM HCl, 10 mM NaCl), and (C) pH 11 (1 mM NaOH, 10 mM NaCl) at rest (black) and under flow (red). (D) Plot of the time-dependence of the integrated SFG spectrum (black circles) at pH 6.5 as a function of time for a liquid flow cycle (blue curve). The solid red line over the SFG data is a five-point average to guide the eye. All intensity curves are normalized to the flow-off intensity at pH 6.5.

Under specific solution conditions, turning on flow dramatically changes the SFG intensity for both CaF2 and SiO2, albeit at different solution conditions for each surface. As mentioned earlier, the magnitude of the SFG signal reflects the degree of alignment of water molecules due to the electrostatic field at the surface. The large changes in the SFG intensity triggered by the flowing liquid therefore reflect a considerable change in the interfacial water alignment. The water arrangement at the surface is influenced by three primary factors: (i) the magnitude of the surface charge (potential), (ii) the penetration depth of the resulting electric field into the liquid, and (iii) the ability of the surface field to maintain alignment of water dipoles. One could imagine that the flow displaces ions from the surface, thus decreasing the screening of the interfacial charge and increasing the penetration depth of the electric field; that is, increasing the Debye length. However, such a mechanism would always result in an increased water dipole alignment and corresponding increase in SFG intensity upon flow, regardless of the sign of the surface charge, which is clearly not borne out by the data (Figs. 1, B and C, and 2A). It is also plausible that the flow perturbs the field-induced preferential alignment of water at the surface, which would lead to a decrease in the SFG signal but again would not result in different behavior based on surface charge. Furthermore, the reorientational motion of water and the re-equilibration of displaced ions due to diffusion in the near-surface region are both too fast to be consistent with the slow recovery times observed after stopping the flow (Figs. 1D and 2D). These arguments make both hypotheses (ii) and (iii) incompatible with our results and indicate that changes in the surface charge are responsible for the altered water arrangement at the interface when the flow is turned on.

The slow recovery after stopping the flow suggests the onset of a new chemical equilibrium: Chemical reactions occurring at the interface appear to underlie the change in surface charge and, therefore, water alignment. Although the solubility of CaF2 and SiO2 in water is extremely low, it is not negligible (supplementary materials section 7). For SiO2, the dissolution reaction is dominated by the hydrolysis of the Si-O-Si bond (34)

Embedded Image(pH > pzc) (1)

Equation 1 shows that for pH values above the pzc, where the SiO2 surface is (on average) negatively charged, hydrolysis entails a loss of the negative surface charge. For water at rest, the effective forward and backward reaction rates balance each other when the solution near the surface is nearly saturated with silicic acid. The onset of the flow lowers the silicic acid concentration near the surface by introducing fresh water, not yet saturated with the solute. This provides a driving force for surface hydrolysis that sharply biases the equilibrium in Eq. 1 toward the right-hand side, effectively removing negative charges from the surface and thereby lowering the SFG intensity. However, subsequent deprotonation of the newly formed neutral Si-OH surface state (Si-OH → Si-O + H+) will partially restore the negative surface charge. These two competing reactions create a new steady state with a possibly changed surface charge. As shown in Fig. 2A by the large drop in SFG intensity at pH 6.5 after turning on flow, the interfacial water is not as strongly aligned, indicating that the SiO2 surface has become less negatively charged. The substantial change in the SFG signal upon flow directly implies that deprotonation of the surface Si-OH is slower than silica hydrolysis by water at pH 6.5 (Eq. 1).

After stopping the flow, dissolution of the silica surface results in slow accumulation of silicic acid that gradually tapers off as the near-surface region becomes saturated. This slow saturation, along with the deprotonation of Si-OH surface groups, results in the delayed recovery to the at-rest (more negative) surface charge and thus provides a rationale for the observed slow recovery (Fig. 2D). Equation 1 represents only one elementary step in the dissolution reaction. The Si(OH)3O can be further protonated in solution (depending on the solution pH); however, its global concentration remains too low (compared with the 10 mM NaCl) to substantially change the ionic strength or pH of the solution.

The situation is different at low pH, where at equilibrium, the surface is less negative (more Si-OH rather than Si-O), on average. Equation 2 shows that for this case, silica hydrolysis at the surface is essentially neutral in terms of surface charge

Embedded Image (pH < pzc) (2)

Thus, at low pH, the liquid flow, although still driving away silicic acid, is expected to have minimal effect on the surface charge (and equivalently the SFG signal), consistent with our experimental observation (Fig. 2B).

At high pH, we also observe that flow minimally affects the SFG signal (Fig. 2C). Although the dissolution of silicic acid due to hydrolysis certainly alters the charge (Eq. 1), the subsequent deprotonation of Si-OH following silica hydrolysis is expected to be much faster than at pH 6.5. The combination of hydrolysis and fast subsequent deprotonation does not lead to a substantial shift of the surface charge. Alkaline cleavage of the Si-O-Si bond is also possible at high pH and is consistent with the data we see here (see supplementary materials section 8). Taken together with Fig. 2, this demonstrates that flow-induced changes in chemical dissolution equilibria can modulate the SiO2 surface charge, which is ultimately reflected by changes in the interfacial water alignment, as detected by the SFG measurements.

Coming back to the CaF2 surface, we now similarly relate a change in surface charge triggered by flow with dissolution of CaF2, which was verified by inductively coupled plasma optical emission spectrometry (see supplementary materials section 7). The main equations that govern the ion dissolution from the surface are

Embedded Image (3)Embedded Image (4)

From these reactions, we see that the surface charge does not originate from protonation or deprotonation of surface sites, but rather from dissolution of F or Ca2+ ions from the surface. For CaF2 it has been shown that dissolution of F (Eq. 3) is favored over Ca2+ dissolution (Eq. 4), due to the strong, energetically favored hydration of F (29, 35). Therefore, when the fluid is at rest, a slight stoichiometric excess of F will be present in the near-surface region, which makes the CaF2 surface positively charged (at low pH), according to Eq. 3.

As in the silica case, turning on the flow introduces fresh water near the surface, dispersing near-surface dissolved Ca2+ and F ions into the bulk fluid and biasing the dissolution equilibria toward the release of ions. Our measurements in Fig. 1B show that the SFG signal increases by nearly 100% when pH 3 water is flowed over CaF2, reflecting a more positively charged surface. This observation directly shows that dissolution of F is faster than Ca2+. This explanation implicitly assumes that the change of the fluoride concentration upon flow dominates the magnitude of the change of the surface charge, whereas the change in near-surface Ca2+ concentration has little effect. This is confirmed by control experiments. Figure S5 shows that the flow effect is seen for a solution with a 10-fold higher Ca2+ concentration than is theoretically available from CaF2 dissolution. In contrast, figs. S6 and S7 show that the flow effect is completely suppressed in the presence of excess F. Because the free concentration of F ions in a saturated solution is nearly constant from pH from 3 to 12 (28), the dissolution equilibria are not expected to depend strongly on pH, and the surface should always become more positively charged in the presence of flow, regardless of pH in our experiments.

Although Eqs. 3 and 4 predict the CaF2 surface to be positive regardless of pH (2830), it has been shown that at typical ambient conditions, dissolution of atmospheric CO2 and subsequent formation of CO32– lead to surface carbonation with increasing pH (that is, surface F is replaced by CO32–), eventually making the surface negatively charged above pH 10 (28). Thus, at high pH the contribution of carbonate must also be considered, making the surface reactions more complicated. Despite these complications, we again find that the F concentration mediates the decreased SFG intensity upon flow at high pH, and that carbonate contributions can be excluded as being responsible for the flow-induced changes (for details, see figs. S8 and S9). Our results show that dissolution of F plays the key role for changing the surface charge (Eq. 3) up to at least pH 12. Therefore, upon flow at acidic pH, the surface charge becomes more positive, causing the SFG intensity to increase (Fig. 1B), whereas upon flow at basic pH (above the pzc), the surface becomes less negative, thereby decreasing the SFG signal (Fig. 1C).

The notion that flow modifies the surface charge by altering the near-surface ionic distribution, thereby shifting the chemical equilibrium, has interesting and important consequences near the pzc for CaF2. Because the surface is always predicted to become more positive (or less negative) upon flow, regardless of the pH of the solution, and the effect of flow is similar to reducing the pH by ~2 units (fig. S3), a surface-polarity reversal should be observable near the pzc (Fig. 3A). The polarity of the CaF2 surface charge is expected to flip from negative to positive upon turning on flow if the solution pH is slightly above the pzc. Accordingly, this charge inversion should lead to a corresponding flow-induced flip-flop or orientation reversal of water molecules at the interface (32).

Fig. 3 Flow induces charge inversion at the CaF2 interface.

(A) Trend of the SFG signal intensity as a function of aqueous solution pH for the CaF2 interface. (B) SFG spectra of the CaF2-water interface at pH 9.5, where the surface is weakly negatively charged, with flow off (time t = 0 s). (C) The water flow is turned on, and the SFG signal decreases to essentially zero at t = 10 s. (D) At t = 200 s after turning flow on, the SFG intensity increases in a red-shifted band. The solid lines over the SFG data are five-point averages to guide the eyes. All curves have been normalized by the flow-off maximal intensity.

Indeed, this is borne out experimentally. Starting with water at pH 9.5, the CaF2 surface charge is slightly negative under our experimental conditions (due to the presence of carbonate). The water molecules are expected to be weakly aligned because of the small surface charge, with their dipoles directed away from the substrate, on average. The SFG spectrum shown in Fig. 3B displays a broad resonance with a peak intensity at 3220 cm−1 that is 10-fold weaker than at pH 12 (Fig. 1C) because of the proximity to the pzc. Furthermore, observing the primary peak blue-shifted (with respect to its location for pH 12) to 3220 cm−1 in the SFG spectrum is only possible for CaF2 when approaching the pzc from a highly alkaline solution (fig. S10A).

Upon starting the flow, the SFG intensity drops in a matter of seconds in response to generation of a less negative surface, according to Eq. 3. This results in the total surface charge approaching zero, at which point the SFG signal has essentially vanished (Fig. 3C). At this moment, the surface reaches the pzc and is electrically neutral, so it does not induce any preferential water alignment (15, 32). As the flow continues, the SFG signal evolves and increases to give a very different SFG spectrum with a peak intensity shifted to 3080 cm−1 (Fig. 3D). This peak location in the SFG spectrum, combined with the more than 100-fold weaker SFG intensity than at pH 3 (Fig. 1B), is only observable when approaching the pzc from acidic pH (fig. S10B). Figure 3D is thus the SFG spectrum from a CaF2-water interface that is now slightly positively charged. This spectrum therefore reflects water dipoles pointing, on average, slightly toward the surface. Figure 3, B to D, directly demonstrates that flow along the CaF2-water interface generates sufficient positive charge as a result of fluoride dissolution to invert the polarity of the surface and the average direction of water dipoles. The flip-flop behavior predicted based on flow-induced inversion of the surface charge via the dissolution reaction in Eq. 3 is clearly observed in our experiments. Upon switching off the flow, the surface again changes polarity back to negative, and the reversibility can be seen in fig. S11.

The coupling of liquid flow with interfacial chemistry has clear implications for interpretation of flow-based electrokinetic measurements. Streaming potential experiments (36) will likely present a surface charge influenced by the flow conditions, if surface reactions are not suppressed. Besides flow modulating the electrostatics at the interface, our results additionally imply that (electro-) chemical reactions are inevitably affected as a result of the modified surface potential. Therefore, this work is expected to have a substantial impact on the understanding, modeling (37), and sensing (38, 39) of (geo-) chemistry and physics at immersed interfaces.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S11

Reference (40)

References and Notes

  1. Note that due to the presence of the surface field, the SFG signal does not originate solely from a second-order nonlinear process, but also from a third-order process (27). However, both the second- and third-order contributions reflect the surface potential, which is relevant to the discussion of the present results.
  2. Because the experiments were performed in total internal reflection geometry, it is challenging to record a reliable reference of the IR spectral profile transmitted through the prism. Therefore, the spectra shown here have not been normalized by the IR spectral profile. However, because only relative comparison between vibrational bands is made, the lack of normalization does not affect the conclusions. Details are provided in the supplementary materials.
  3. Acknowledgments: We gratefully acknowledge D. Bonn, H.-J. Butt, and G. Auernhammer for fruitful discussions, as well as the anonymous referees for their insightful comments that resulted in a more accurate interpretation of our results. We thank P. Lambin and F. Cecchet for reading the manuscript, H. Burg for measuring the surface roughness of the CaF2 prism, and M. Steiert for the inductively coupled plasma optical emission spectrometry measurements. D.L. is supported by the Belgian Fund for Scientific Research—Fonds de la Recherche Scientifique (F.R.S.-FNRS). S.H.P. and E.H.G.B. are supported by the Marie Curie Foundation, with grants CIG322284 and CIG334368, respectively.
View Abstract

Navigate This Article