Anions dramatically enhance proton transfer through aqueous interfaces

Abstract

Proton transfer (PT) through and across aqueous interfaces is a fundamental process in chemistry and biology. Notwithstanding its importance, it is not generally realized that interfacial PT is quite different from conventional PT in bulk water. Here we show that, in contrast with the behavior of strong nitric acid in aqueous solution, gas-phase HNO₃ does not dissociate upon collision with the surface of water unless a few ions (> 1 per 10⁶ H₂O) are present. By applying online electrospray ionization mass spectrometry to monitor in situ the surface of aqueous jets exposed to HNO_3(g) beams we found that production increases dramatically on > 30-μM inert electrolyte solutions. We also performed quantum mechanical calculations confirming that the sizable barrier hindering HNO₃ dissociation on the surface of small water clusters is drastically lowered in the presence of anions. Anions electrostatically assist in drawing the proton away from lingering outside the cluster, whose incorporation is hampered by the energetic cost of opening a cavity therein. Present results provide both direct experimental evidence and mechanistic insights on the counterintuitive slowness of PT at water-hydrophobe boundaries and its remarkable sensitivity to electrostatic effects.

Proton transfers (PTs) at water interfaces, such as water boundaries with air (1, 2) or lipid membranes (3), intervene in fundamental phenomena. Arguably the most important PTs are those that take place through and across water boundaries rather than in the bulk liquid. Interfacial PTs participate in the acidification of the ocean (4), the chemistry of atmospheric gases and aerosols (1, 5, 6), the generation of the electrochemical gradients that drive energy transduction across biomembranes (3, 7, 8), and in enzymatic function (9, 10) because the activation of neutral species is most generally accomplished via acid-base catalysis (11). Interfacial PT, in contrast with conventional PT in bulk water, depends sensitively on the extent of ion hydration because the density of water in interfacial layers vanishes within 1-nm (12). The acidity of hydronium at the interface, , is therefore expected to bridge that of , which protonates most nonalkane species in the gas-phase (13), and , which neutralizes only relatively strong bases in solution. Critically controlled by ion hydration in thin yet cohesive interfacial water layers that resist ion penetration, PT “on water” clearly confronts unique constraints. Species that behave as strong acids “in water” may become weak ones on water if dissociation were hindered by kinetic and/or thermodynamic factors in the interfacial region (14, 15).

Herein we address these important issues and report the results of experiments in which we monitor the dissociation of gaseous nitric acid HNO_3(g) molecules in collisions with interfacial water, H₂O_(if), reaction 1 (Eq. 1):

[1]

The Technique

Experiments were conducted by intersecting continuously refreshed surfaces of free-flowing aqueous jets with HNO_3(g)/N_2(g) beams at ambient temperature and pressure. The formation of interfacial nitrate, , was monitored in situ via surface-specific online electrospray mass spectrometry (ESMS) (16, 17) (SI Text and Figs. S1 and S2). ESMS is routinely used to analyze the composition of bulk liquids. However, we have demonstrated that by changing the instrumental configuration and operating parameters it is possible to sample the interfacial layers of the liquid jet. We have previously taken advantage of the high sensitivity, surface selectivity, and unequivocal identification capabilities of our modified electrospray mass spectrometer to investigate fast gas–liquid reactions on the surface of aqueous jets (5, 18). The claim that the mass spectra obtained in our instrument mostly reflect the ion composition of the outermost layers of the jet has been validated by showing that: (i) the relative anion abundances (i.e., the relative mass spectral signal intensities) measured on jets consisting of equimolar solutions of mixed salts are not identical but follow a normal Hofmeister series (as expected at the air–water interface and confirmed by other surface-sensitive techniques), and are specifically affected by surfactants (cationic or anionic) (19); and (ii) mass spectra of jets exposed to reactive gases reveal the presence of species necessarily produced at the interface rather than in the bulk liquid (20, 21).

Mass spectrometers report the net charge that arrives at the detector per unit time. Therefore, the produced in reaction 1 on the surface of the electroneutral liquid jet can be detected after it has been separated from counterions. Separation is brought about during the pneumatic breakup of the liquid jet by a fast annular N_2(g) nebulizer gas flow, which shears the outermost liquid layers into droplets carrying net charges of either sign. These droplets have size and net charge distributions, and carry more surface and electrostatic energies than the original jet, at the cost of the kinetic energy lost by the nebulizer gas. A key feature of our ESMS instrumental configuration is that the jet is orthogonal to the inlet to the mass spectrometer (Figs. S1 and S2). This geometry overwhelmingly favors the detection of ions emanating from the peripheral layers of the jet. Ions are ultimately ejected to the gas-phase because of severe charge crowding in the nanodroplets that result from extensive solvent evaporation (22). We have presented detailed data analysis (16), based on mass balances and the application of the kinetic theory of gases to fast gas–liquid reactions, which suggests that the thickness of the interfacial layers sampled in these experiments is certainly within a few nm, and most likely approximately 1 nm (see below and SI Text).

Results

Fig. 1 displays mass spectral (m/z = 62) signal intensities, I₆₂, as a function of pH (of the bulk aqueous solution) on liquid jets exposed to HNO_3(g); I₆₂ remains above detection limits on the surface of pH 4.5 to 9.5 jets, but sharply increases both on more basic and more acidic solutions to limiting values, , above pH 11 and below pH 3. Notably, we found that values are uniformly reached at all pH values on > 1-mM NaCl jets. The previously reported uptake coefficient of HNO_3(g) on deionized water (γ > 0.1) (23), reveals that only a small fraction of the HNO_3(g) molecules colliding with the surface of water are incorporated into the bulk liquid, where they fully dissociate [pK_a(HNO_3(aq)) = -1.4]. Therefore, the small signals detected in our experiments on pure water jets indicate that we do sample their outermost interfacial layers (24, 25), and confirms that most of the mass-accommodated HNO₃ diffuses in undissociated form through such layers. The fact that the production of is dramatically enhanced by inert anions on water hints at the possibility that the barrier preventing HNO₃ dissociation at the interface might be kinetic rather than thermodynamic (26, 27). In summary, the results of Fig. 1 and Fig. S3 provide evidence that HNO_3(g) behaves as a weak acid on the surface of water, and extrinsic inert ions can significantly catalyze HNO₃ dissociation therein (15, 28).

Fig. 1.

Electrospray mass spectral nitrate signal intensities (I₆₂) detected on water or 1-mM NaCl microjets exposed to 3 × 10¹² molecules cm^-3 of gaseous nitric acid for approximately 10 μs as functions of pH. Solid, dashed lines are linear regression and 95% confidence limits, respectively, to the data on 1-mM NaCl. Error bars estimated from reproducibility tests. All experiments under 1 atm of N₂ at 293 K.

The air–water interface of electrolyte solutions is preferentially populated by anions. This is borne out by the negative surface potential of most electrolyte solutions (29), by surface-specific spectroscopic studies (30–33), and by theoretical predictions. The adsorption of ions to the surface was surmised long ago from the surface tension minima observed in electrolyte solutions at approximately 1 mM. They were accounted for by electrostatic interactions among ions that saturate the surface of water at approximately 1 mM (34)—i.e., in the concentration range in which we observe an increase of HNO₃ dissociation on water (30, 33). The saturation dependence of production on electrolyte concentration (Fig. S3A) can be ascribed to catalysis by anions A^- adsorbed to identical, noninteracting sites of the air–water interface—i.e., I₆₂ = I₆₂,_max[A^-]/(K_1/2 + [A^-]) (30). We derive K_1/2 = 128 μM (NaCl) and K_1/2 = 77 μM (MgSO₄) (i.e., the concentrations at which the interface would be half-saturated with catalyzing anions), which are commensurate with the values ([NaCl]_max = 400 μM and [MgSO₄]_max = 200 μM) deduced from SHG experiments (30) (Fig. S3B). Although neither Cl^- nor are as surface active as I^- or (32), they should approach the air–water interface far closer than the (R_ion–ion) separations prevalent at the onset of catalytic effects (see below).

Hydronium, , the counterpart of in reaction 1, was tracked by using hexanoic acid (PCOOH) as a proton scavenger. PCOOH is both a weak acid and a weak base in water: pK_a(PCOOH)_(aq) = 4.8, . However, we have shown that PCOOH is protonated on the surface of mildly acidic water, where it behaves as a stronger base: (17). Fig. 2 displays I₁₁₇ (), I₁₁₈ (PCOOHD⁺), and I₁₁₉ () signal intensities from 1 mM PCOOH in 1∶1/D₂O∶H₂O jets (initially at pH 7) as functions of gas-phase HNO_3(g) or DNO_3(g) concentrations. Fig. 2 (Inset) shows the corresponding I₆₂ and I₁₁₅ (PCOO^-) signal intensities versus HNO_3(g) concentration. It is apparent that: (i) PCOO^- is promptly neutralized upon exposure to the lowest HNO_3(g)/DNO_3(g) concentrations, whereas (ii) the protonation/deuteration (hydronation) of the weaker base PCOOH requires exposure to at least n > 2 × 10¹² molecules cm^-3. The fact that HNO_3(g) dissociates on water containing the anions of either a stronger acid [pK_a(HCl)_(aq) = -7 versus pK_a(HNO₃)_(aq) = -1.4] or a weaker one [pK_a(PCOOH)_(aq) = 4.8] supports the assertion that anions function as catalysts rather than proton acceptors. The appearance of hydronated species [(), (PCOOHD⁺), and ()] reveals that the surface of the jet has been acidified (from pH 7) to pH < 2.5. Because this is achieved under conditions in which the number of hydrons delivered by HNO_3(g)/DNO_3(g) on interfacial layers is much (approximately 10³ times) smaller than those carried by the 50-μL min^-1 1∶1/D₂O∶H₂O aqueous jet, the former must be confined to thin (Δ[cm]) interfacial layers during the lifetime of the jet. The relative abundances of the , PCOOHD⁺, and isotopologues are appreciably different under HNO_3(g) or DNO_3(g) (Fig. 2 A and B) and corroborate that the hydrons delivered by gaseous nitric acid remain (i.e., do not diffuse into and rapidly scramble their isotopic labels with the bulk solvent) in the interfacial layers sampled herein (SI Text and Fig. S4). The assumption that our experiments probe reactive events taking place in interfacial layers of molecular depth is therefore based on substantial evidence. From the frequency of HNO_3(g) collisions with the jet given by the kinetic theory of gases, we estimate that Δ is approximately 1 × 10^-7 cm (17) (SI Text).

Fig. 2.

Electrospray mass spectral signal intensities of protonated isotopologues of hexanoic acid (PCOOH): , m/z = 118(PCOOHD⁺), and , detected on 1-mM PCOOH solutions in 1∶1/D₂O∶H₂O microjets (initially at pH 7) exposed to variable concentrations of gaseous HNO₃ (A) or DNO₃ (B). The inflection point corresponds to (17). Inset shows the evolution of the PCOO^- (m/z = 115) and (m/z = 62) signals detected in negative ion mode. All experiments under 1 atm of N₂ at 293 K.

What is the minimum number of additional water molecules m that renders reaction 1 exoergic? The free energy required to produce a hydrated contact ion pair at the air–water interface, , can be estimated as the sum of the gas-phase process [] (13); plus the electrostatic energy released as the infinitely distant gas-phase point charges reach an approximately 3.3-Å separation in the contact ion pair (E_el = -100 kcal mol^-1); plus the free energy of hydrating []

Extant thermochemical data on (mH₂O·H₃O⁺) clusters (35) show that (i.e., reaction 1 is thermodynamically allowed for m≥4, even if were not hydrated at all) (36). The hydration of will, of course, contribute to the exoergicity of reaction 1. Because HNO₃ is able to interact with at least four water molecules upon impact with the surface of water (28), the nature of the barrier-hindering reaction 1 remains to be elucidated. It has been proposed that acid-base equilibria at the air–water interface are shifted (relative to bulk water) toward neutral species by approximately ± 2 pK_a units (37). In the case of nitric acid, pK_a(HNO_3(aq)) = -1.4, this proposal makes HNO_3(if) a strong acid at the interface: pK_a(HNO_3(if)) of approximately 0, at variance with our observations. We wish to emphasize that in our experiments, in contrast with most other studies (38), HNO₃ approaches the air–water interface from the vapor instead of the water side. Hence, gas-phase ion thermochemistry (13, 35) is a more appropriate framework for analyzing our results.

Against this background, we performed density functional theory calculations on HNO₃ interacting with water decamers W₁₀(W ≡ H₂O) in the absence and presence of Cl^- to ascertain the molecular basis of our experimental observations. Fig. 3 A and B display the calculated Gibbs free energy (ΔG⁰) and enthalpy (ΔH⁰) profiles at 300 K. We confirmed that HNO₃ embedded in W₁₀ clusters dissociates spontaneously, in accordance with common knowledge, thermodynamics, and Car–Parrinello molecular dynamics (CPMD) calculations (14, 15). In contrast, HNO₃ binds as a molecule to the periphery of W₁₀ via two hydrogen bonds with ΔH⁰ = -13.0 kcal mol^-1 and (because of translational and rotational HNO₃ entropy losses) ΔG⁰ = -1.2 kcal mol^-1. The free energy barrier for transferring a proton from adsorbed HNO₃ into the cluster while leaving a on its surface is quite large: ΔG^‡ = 14.1 kcal mol^-1, or 12.9 kcal mol^-1 above the reactants. Weakly bound undissociated HNO₃ is therefore rather stable toward dissociation and highly mobile on the surface of water. Remarkably, HNO₃ not only binds more strongly (ΔH⁰ = -18.6 kcal mol^-1, ΔG⁰ = -6.9 kcal mol^-1) to clusters containing a Cl^-, but the free energy barrier for transferring a proton to W₁₀·Cl^- is dramatically reduced: ΔG^‡ = 1.2 kcal mol^-1.

Fig. 3.

Calculated Gibbs free energies (ΔG⁰) and enthalpies (ΔH⁰) of reactants, adducts, transition states, and products of optimized water clusters in contact with nitric acid in the absence (A) and presence (B) of interfacial chloride. Proton wires highlighted. Energies in kcal mol^-1.

Discussion

Calculations provide significant clues about the origin of the barrier to HNO₃ dissociation on water. HNO₃ binds to W₁₀ both as H-bond donor and acceptor. However, the proton, an intrinsic water ion, cannot readily slip into cluster leaving behind (Fig. 3A). The barrier to PT on the surface of water is therefore associated with the fact that (i) overcoming the electrostatic attraction in a disjoint () ion pair, or (ii) opening a cavity for to follow after the proton into the cluster, entails significant energy costs. Calculations involving larger water clusters do not eliminate such barrier (28). Clearly, a chloride lets H₃O⁺ advance further into the cluster primarily by countering the electrostatic bias imposed on H₃O⁺ by laggardly , rather than binding to it [recall that pK_a(HCl)_(aq) = -7]. We also noticed that the atomic rearrangements involved in binding HNO₃ to W₁₀ clusters are uncorrelated to those required for subsequent PT. In contrast, the stronger interaction between HNO_3(g) and (Cl·W₁₀)^- clusters primes (Cl·W₁₀⋯HNO₃)^- for PT. The reaction coordinate for PT on pure water is a combination of six internal modes involving displacements of heavy O atoms, whereas PT in the presence of chloride proceeds adiabatically along a three-link proton wire between quasi-degenerate solvent states (Fig. S5 A and B).

After establishing the role of electrostatics in the catalysis of HNO₃ dissociation on small water clusters, we need to understand why catalytic effects are observed on > 30-μM electrolytes—i.e., at (R_ion–ion) < 120-nm interfacial separations (SI Text) that vastly exceed the size of such clusters. On the basis of our calculations we envision that HNO₃, after alighting on water, roams rather freely over its surface as HNO_3(if) until it approaches an interfacial Cl^-, whereupon it falls into a deeper potential well and undergoes prompt dissociation. In SI Text we estimate that average number of hops required by HNO_3(if) to reach a Cl^- on the surface of > 30-μM solutions would take a few nanoseconds (i.e., competitively with back desorption into the gas-phase) (16, 39). Recapitulating, present experimental results substantiate a key role for electrostatics in the mechanism of HNO₃ dissociation at water-hydrophobe interfaces, and suggest that even sparse anions can effectively catalyze this process.

Implications

Our finding that PT across water-hydrophobic media interfaces is catalyzed by anions has important implications in many fields. Whether HNO_3(g) dissociates on aqueous surfaces, for example, bears on various environmental issues. Whereas is a sink for active nitrogen in the atmosphere because it can be removed by dry and wet deposition, undissociated HNO₃ may react via: 2HNO_3(g) + NO_(g) → 3NO_2(g) + H₂O_(g), thereby sustaining the atmospheric impact of nitrogen oxides (40). Adsorption of HNO_3(g) on ice also depends critically on whether HNO₃ dissociates therein—i.e., whether coverage is a function of P or P^1/2 (P ≡ HNO_3(g) partial pressure) (41). Our results suggest that HNO_3(g) will dissociate upon impact on most environmental aqueous surfaces, including premelted films on ice that contain electrolytes impurities at least at millimolar levels. Our demonstration that PT across internal water-hydrophobe interfaces is facilitated by electrostatics related to the concept of anion-mediated water bridges for PT in proteins (42) is at least consistent with the assumption that charge transfer events at water–protein interfaces are driven by electrostatic preorganization (43, 44). It also accounts for the fact that even weakly basic, mobile anions, such as chloride, may enhance proton motion along membrane surfaces without providing localized proton-binding sites (45, 46).

Experimental Methods

In our experiments, continuously refreshed, uncontaminated surfaces of free-flowing aqueous microjets exposed to < 8 × 10¹² HNO_3(g) molecules cm^-3 for approximately 10 μs are monitored by online negative or positive ion ESMS. Fifty μL min^-1 of deionized water or aqueous electrolyte solutions (pH-adjusted using concentrated NaOH or HCl) are injected as a microjet into the spraying chamber of an ES mass spectrometer held at 1 atm, 293 K via an electrically grounded, stainless steel pneumatic nozzle (100 μm) internal diameter (SI Text and Figs. S1 and S2) (47). A high-speed (approximately ) annular nebulizer N_2(g) flow tears up the much slower () microjet into droplets charged with ion excesses of either sign. Ions are eventually ejected to the gas-phase, charge selected by a polarized inlet port orthogonal to the nozzle, and detected by mass spectrometry. We note that the velocity at which the liquid jet emerges from the nozzle is approximately 500 times slower than that required for observing electrokinetic effects in our experiments (48) (SI Text).

Computational Methods

Gibbs free energies (ΔG) at 298 K were computed from calculated enthalpies (ΔH) and entropies (S) according to ΔG = E_elec + ZPE + H_vib - TS_vib. Geometries of energy minima and transition states were optimized using the X3LYP functional (extended hybrid functional combined with Lee–Yang–Parr correlation functional) (49), the 6-31G** basis for light atoms (50), and 6-311G**++ for Cl^- (51). Hessians at these geometries provided harmonic zero-point energies, vibrational enthalpies, and entropies. Neglect of anharmonicity effects (< 1 kcal mol^-1) may not affect the main conclusions. After geometry optimization, the electronic energy E_elec was evaluated with the 6-311G**++ basis on all atoms. The free energies of nitric acid and nitrate at 1 atm were calculated using statistical mechanics for ideal gases (SI Text).