Reversed interfacial fractionation of carbonate and bicarbonate evidenced by X-ray photoemission spectroscopy

The aqueous carbonate system is of central importance in nature and has been studied extensively. The species involved—carbon dioxide, carbonic acid, bicarbonate, and carbonate—are involved in both the global carbon cycle and in physiological buffer and respiration systems. Oceanic carbonate chemistry governs CO₂ uptake by surface waters, and the subsequent saturation of carbonate has a significant role in biomineralization in marine organisms and in ecosystems.^1,2 In mammalian systems, the carbonate buffer system regulates blood pH and is responsible for CO₂ transport across membranes.³ Clearly, understanding aqueous carbonate chemistry is also central to efforts involved in mitigating the effects of climate change (e.g., carbon capture, sequestration, etc.).^4–6 Consequently, much effort has addressed the aqueous carbonate system, dating back over a century⁷ with past experimental studies characterizing the kinetic, thermodynamic, and structural properties of these species in aqueous solution and in ice matrixes.^8–19 Recent theoretical studies, employing molecular dynamics and ab initio quantum calculations, have sought to characterize the thermodynamic and mechanistic details of their hydration and chemical reactions.^20–27 

The nature of the ions and their effects on the water structure at the air/water interface critically influence the uptake of atmospheric gases such as carbon dioxide, which is subsequently hydrolyzed to bicarbonate and carbonate. The aqueous bicarbonate and carbonate ions present at the air/water interface have previously been examined by vibrational sum frequency generation (VSFG), finding that the carbonate ion has a significant effect on the orientation and structuring of interfacial water.^28–30 These VSFG studies probe the interfacial water structure, focusing on the –OH stretching region of the vibrational spectrum, and are thereby an indirect probe of the ion. In contrast, X-ray photoemission spectroscopy (XPS) is an atom-specific probe of a system’s occupied states, which enables both system composition and depth profiling. Depth profiling is achieved through the exploitation of the effective attenuation length (EAL) of the emitted photoelectrons, viz., photoelectrons with energies near 200 eV selectively probe the interface to depths of ∼20 Å, whereas those with higher kinetic energies typically have longer EALs and thereby probe deeper below the surface (e.g., ∼60 Å at 800 eV).³¹ Additionally, the measured binding energies characterize the oxidation state of the probed atoms. The development of ambient pressure photoemission spectroscopy by Siegbahn et al. opened the field to the study of liquid interfaces³² either in the presence of the equilibrium vapor pressure³³ or under vacuum conditions, as in the liquid microjet photoemission spectroscopy experiments first performed by Faubel et al.,³⁴ and has since been applied to the study of a wide range of liquid systems.^35–39 The spatial distribution of K⁺ and $CO32−$ in an aqueous solution was previously investigated using XPS by Brown et al., who concluded that the cation resides slightly closer to the interface than the anion.⁴⁰ 

Here, we present a study of sodium carbonate (Na₂CO₃), sodium bicarbonate (NaHCO₃), and carbonic acid (H₂CO₃) via liquid microjet XPS. We have previously explored the aqueous carbonate system using liquid microjet X-ray absorption spectroscopy (XAS), a complementary technique which probes the unoccupied states of a system; this enabled the detailed characterization of the hydration structure of Na₂CO₃, NaHCO₃, H₂CO₃, and dissolved CO₂.^41–43 Building on those results, the measurement of XPS spectra at different incident photon energies, ranging from 490 eV to 1090 eV, has permitted the characterization of the surface fractionation of the individual ions. Investigating 50:50 mixtures of Na₂CO₃:NaHCO₃ and H₂CO₃:NaHCO₃ reveals that both Na₂CO₃ and H₂CO₃ are present in greater concentrations than that of NaHCO₃ in the probed interfacial region.

Solutions were prepared using 18.2 M$Ω$·cm resistivity water obtained from a Millipore purification system. Concentrated HCl (12.1M) was obtained from Baker. Na₂CO₃ (>99.5% purity) was obtained from Fisher Chemical. NaHCO₃ (>99.7% purity) was obtained from Macron Fine Chemicals.

The solutions probed by XPS are detailed in Table I.

The 0.5M H₂CO₃ solution (III) was generated in situ using our fast-flow liquid microjet mixing system by mixing solutions of 1M HCl and 1M NaHCO₃. This mixing system was previously employed in XAS measurements of aqueous H₂CO₃ and dissolved CO₂.^42,43 Briefly, a dual syringe pump system (Teledyne-ISCO 260D) drives two solutions through a Microvolume Y-connector. The mixed solution then travels through a 50 $μm$ inner diameter fused silica capillary to generate the liquid microjet. In this scheme, the interaction time between the two solutions is ∼0.5 ms, facilitating the observation of short-lived species in solution.

The 50:50 mixtures of Na₂CO₃:NaHCO₃ (IV) were generated by mixing 1M Na₂CO₃ with 1M NaHCO₃, yielding a 0.5M concentration of both carbonate and bicarbonate. 50:50 H₂CO₃:NaHCO₃ (V) was generated by mixing 1M NaHCO₃ with 0.5M HCl yielding 0.25M concentrations of both species. To generate mixtures of 50:50 H₂CO₃:NaHCO₃ with the same concentration Na₂CO₃:NaHCO₃ as the Na₂CO₃:NaHCO₃ solution (0.5M of each ion), a 2M solution of NaHCO₃ would have been required. This exceeds the solubility limit of NaHCO₃ in water (96 g/L, 1.14M). All mixing was done in situ within the fast-flow liquid microjet mixing system.

Carbon 1s and the corresponding valence band X-ray photoemission spectra were measured at Beamline 11.0.2⁴⁴ at the Advanced Light Source (ALS), Lawrence Berkeley National Laboratory (LBNL) using the Ambient Pressure Photoemmision Spectrometer (APPES-II) endstation, which is based on a NAP Phoibos 150 hemispherical analyzer (Specs Surface Nano Analysis GmbH, Berlin). The sample was introduced via a liquid microjet, into the vacuum chamber, operating at a pressure of ∼10 mTorr. The liquid jet is orientated normal to both the electron optical axis of the input lens of the electron spectrometer and the incident X-ray beam. The carbon 1s photoemission spectra of aqueous Na₂CO₃ (I), NaHCO₃ (II), and H₂CO₃ (III) and 50:50 mixtures of Na₂CO₃:NaHCO₃ (IV) and H₂CO₃:NaHCO₃ (V) were measured using incident photon energies of 490 eV, 690 eV, 890 eV, and 1090 eV, resulting in photoelectrons with ∼200, 400, 600, and 800 eV kinetic energy, respectively. To account for shifts induced by charging of the liquid jet and the detector work function, measured binding energies were aligned to the liquid water 1b₁ binding energy at 6.5 eV.⁴⁵ 

The C(1s) and their respective valence band photoemission spectra, measured with an incident photon energy of 490 eV, are shown in Figure 1 for 0.5M solutions of sodium carbonate, sodium bicarbonate, and carbonic acid (I-III). A large systematic shift to higher binding energies is observed between the various carbonate species, from carbonate to carbonic acid. Measured binding energies are Na₂CO₃ (289.1 eV), NaHCO₃ (290.1 eV), and H₂CO₃ (291.0 eV) with corresponding FWHMs of 1.1 eV, 1.1 eV, and 1.3 eV, respectively. The spectra shown in Figure 1 are background-corrected using a linear baseline subtraction but are otherwise unnormalized. As the photoemission cross sections are not expected to differ significantly between the various carbonate species, the drastic difference observed in the intensity between species is striking. The spectra of both sodium carbonate and carbonic acid exhibit higher intensity than that of the sodium bicarbonate. To account for differences in alignment between the liquid jet, the incident X-ray beam, and the detector, the spectra were scaled, relative to the valence band intensity of sodium carbonate. The measured signal intensity in the valence band spectra primarily corresponds to that of the water valence band photoemission, which is largely unchanged upon the addition of 0.5M solute. The scaled spectra, which also exhibit this anomalous difference in intensity, are shown in Figure 2. Relative to that of sodium bicarbonate, the signals originating from carbonate and carbonic acid solutions are ∼1.5 and ∼10.2 times larger, respectively.

The measured carbonate and bicarbonate spectra were fit to single Gaussians while the carbonic acid spectrum was fit to two Gaussian peaks due to the presence of a small peak at ∼292.8 eV corresponding to a gas phase CO₂ background present in the chamber resulting from the decomposition of H₂CO₃ to form CO₂ and H₂O. The fitted peaks, shown in Figure 3, reproduce the measured spectra well and were used to deconvolute the spectra of Na₂CO₃:NaHCO₃ (IV) and H₂CO₃:NaHCO₃ (V) mixtures.

Figures 4 and 5 show the measured photoemission spectra (black and green markers) for 50:50 mixtures of Na₂CO₃:NaHCO₃ (IV) and H₂CO₃:NaHCO₃ (V) at incident photon energies of 490 eV, 690 eV, 890 eV, and 1090 eV. The differences in intensity that were observed in the single component solutions are clearly maintained in these mixtures, although at slightly different ratios. The measured photoemission spectra of the mixtures were fit by fixing the width and center obtained from the Gaussian fits of the single component solutions shown in Figure 3 and varying only the amplitudes of the Gaussians. For the spectra measured at energies higher than 490 eV, the relative centers of the Gaussians between species were maintained. The fitted spectra are shown in the solid black and solid green lines in Figures 4 and 5, respectively, exhibit excellent agreement with the measured spectra. The individual components of the fit are shown in blue (Na₂CO₃), red (NaHCO₃), black (H₂CO₃), and purple (CO₂ gas).

The peak area ratios for 50:50 mixtures of Na₂CO₃:NaHCO₃ (IV) and H₂CO₃:NaHCO₃ (V) are plotted as a function of electron kinetic energy (eKE) in Figure 6. As the electron kinetic energy increases, the peak area ratio between the probed species approaches a limit of unity, as would be expected for an equimolar mixture. However, our results indicate that concentrations of both Na₂CO₃ and H₂CO₃ are significantly enhanced relative to that of NaHCO₃ throughout the probed region. While the enhancement of a neutral acid (H₂CO₃) over a charged ion (⁠$HCO3−$⁠) is not surprising, and has previously been observed for protonated acetic acid,⁴⁶ the significant enhancement of $CO32−$ over $HCO3−$ appears to conflict with recent models for interfacial ion adsorption. Historically, it had been assumed that all ions were repelled from the air/water interface. This repulsion was originally explained by classical electrostatic theory⁴⁷ and supported by surface tension measurements which showed that the surface tension increased as a function of salt concentration.^48,49 More recently, molecular dynamics simulations,^50,51 surface specific second-order nonlinear optical experiments,^52–54 and XPS measurements^33,39 have predicted and observed the enhancement of certain simple ions at the air/water interface. These models have shown that weakly hydrated, charge-diffuse ions are generally enhanced at the air/water interface.^55–57 Our current experiments suggest that carbonate, a strongly hydrated doubly charged anion, is present in higher concentrations than is the singly charged, less strongly hydrated, bicarbonate anion in the probed region. This can be rationalized if carbonate adsorbs to the air/water interface as an ion pair with sodium (Na⁺:$CO32−$⁠). Adsorption to the air/water interface as a contact ion pair has previously been observed, in both experiment and theory, for aqueous solutions of strong acids.^58–60 In previous VSFG measurements, carbonate was shown to exert a much larger effect on the interfacial water than does bicarbonate.^28,29 More recently, Allen et al. employed phase sensitive VSFG measurements, finding that bicarbonate is accommodated by the interfacial region while carbonate is excluded.³⁰ In these experiments, the presence of aqueous carbonate was found to reorient the surface waters so that the water hydrogens point downward into the bulk.

While these VSFG results appear to conflict with the present XPS measurements, they are not necessarily irreconcilable. Second order nonlinear experiments probing the air/water interface are sensitive only to regions of broken inversion symmetry. While the exact thickness of the probed interfacial region is not quantified, the penetration depth of the probe is approximately half the input wavelength. In aqueous systems containing simple ions, the probe depth is likely less than 1 nm. However, in systems with a more complex depth profile a greater depth may be accessible.⁶¹ Theoretical models typically define the interface in terms of the Gibbs dividing surface, wherein the solvent density reaches half of the bulk density. In either case, these measurements are certainly more surface specific than our low-energy XPS measurements, wherein the effective attenuation length of an electron in water is at a minimum (∼2 nm) when the electron has a kinetic energy of ∼200 eV. As such, even with 200 eV photoelectrons, a significant proportion of the signal arises from photoelectrons generated 2–5 nm below the surface. At 800 eV, the EAL is ∼6 nm. Although the EAL is expected to vary slightly with system composition, we do not expect these variations to affect our interpretation as the EAL should always increase as a function of eKE from 200 eV to 800 eV. Our results would therefore indicate that there exists a significant accumulation of $CO32−$ below the depletion region outside of the region probed by the SFG measurements.

We have presented the X-ray photoemission spectra of aqueous solutions of Na₂CO₃, NaHCO₃, and H₂CO₃ which exhibit a systematic shift to higher measured C(1s) binding energies from carbonate to carbonic acid. The measured spectra of 50:50 mixtures indicate that both carbonate and carbonic acid are present at higher concentrations in the probed region than bicarbonate. Further theoretical modeling is required to address the apparent conflict with current models describing ion adsorption to aqueous interfaces, which would suggest that the singly charged anion, bicarbonate, should be present in higher concentrations relative to the doubly charged carbonate anion. This new result could reflect interesting and important differences in the hydration and counterion interactions (i.e., ion pairing) of the carbonate species.

The authors thank the staff at the Advanced Light Source for excellent experimental support. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences and Materials Sciences Division of the U.S. Department of Energy at the Lawrence Berkeley National Laboratory under Contract No. DE-AC02-05CH11231. X-ray photoemission spectra were collected at Beamline 11.0.2 at the Advanced Light Source. The data presented are available upon request to saykally@berkeley.edu.