Stellar Wind Contribution to the Origin of Water on the Surface of Oxygen-containing Minerals

The MW discharge experiment consisted of a closed sample cell filled with molecular hydrogen/deuterium gas and a solid mineral sample. A MW discharge was induced in the cell, which produces hydrogen/deuterium atoms both in their neutral and excited states. These atoms then interact with oxygen in the mineral lattice, forming water. The cell was placed in the optical path of a light source, heated to evaporate the surface water, and high-resolution gas-phase FTIR spectrometry was used to detect H₂O in the gas phase of the sample. A schematic drawing of the experiment is shown in Figure 1 and a photo in Figure 15 in Appendix B. In detail, first, we conducted a MW discharge experiment to generate hydrogen/deuterium atoms/ions from molecular hydrogen/deuterium. To achieve this, we utilized a quartz finger connected to a borosilicate glass absorption cell. The cell was equipped with CaF₂ optical windows and a vacuum valve. Connected to the finger was a MW cavity resonant holder, cooled by water, positioned close to the sample surface. A magnetron MW power source was connected with a coaxial cable. This source can be continuously adjusted, and the maximum power used was 75 W at a frequency of 2.45 GHz. We optimized the frequency circuit of the cavity with regard to both MW power and the optimal spatial distribution of the discharge within the quartz tube.

Zoom In Zoom Out Reset image size

At the beginning of the experiment, mineral samples in powdered form (100 mg) were introduced into the quartz finger and evacuated at a temperature of 600 K over a period of 12 hr through a vacuum valve at pressures of 10⁻³ mbar. The cell was then filled with 1 mbar of molecular H₂ or D₂ and the discharge was run for 2 hr. Following this, the gas was evacuated to 10⁻³ mbar and the cell was closed. After that, the finger with the sample was heated (inside an oven) to a temperature of 600 K and the released water from the surface was detected in the gas phase using high-resolution FTIR.

In all experiments the absorption beam from an external radiation source was focused through the absorption glass cell (see Figure 1) to the entrance aperture of the interferometer using a CaF₂ lens. The absorption spectra were recorded using a Bruker IFS 120 spectrometer (Bruker, Germany), equipped with a KBr beamsplitter and an InSb detector. For each measurement, a total of 100 scans were coadded at a spectral resolution of 0.03 cm⁻¹ in the range 800–5000 cm⁻¹. To calculate the partial pressures of each detected species, the collected spectra were fit with the spectr library (A. N. Heays 2022). The library fits HITRAN molecular line parameters (including temperature and pressure-induced spectral effects) to existing data and calculates column densities (I. E. Gordon et al. 2022). Column densities and the optical path of the cell were used to calculate the partial pressure using the ideal gas law. This approach enables rapid data analysis without the need for analytical calibration during the data-evaluation process. Blank measurements (i.e., an empty quartz cell without sample) were performed for all experiments and no formation of water was observed.

An alternative approach for observing the formation of water is the utilization of a sputter gun and subsequent monitoring with TPD. The desorption experiments were carried out in an ultra-high vacuum chamber made of stainless steel. The chamber is schematically shown in Figure 2. The chamber was permanently pumped using a turbomolecular pump. The whole TPD system consists of an ion gun (IonEtch sputter gun, TecTra GmbH), used for irradiation of the sample with hydrogen or deuterium ions, a quadrupole mass spectrometer (RGA-200, QMS), for ion detection, housed in a differentially pumped tube, which is terminated by a nozzle with an aperture of 3 mm on the front end, viewports, and a sample holder. In total, six experiments were performed with a different composition each time. Details are described along with the resulting plots below. All samples were loaded on a sample holder in the vacuum chamber and dried in vacuum at 600 K and <5 × 10⁻¹⁰ mbar. After drying and cooling, the samples were sputtered with hydrogen/deuterium ions first, which should result in water formation on the surface. Bombardment of samples was done at 8.5 × 10⁻⁵ mbar by ions with an energy of 0.5 kV for 20 minutes with a total ion current of ∼10 μA. The samples were then heated by the e-beam method from a tungsten cathode installed 4 mm below the sample holder equipped with a K-type thermocouple. The temperature of the sample was continuously monitored using the K-type thermocouple, which also provided feedback for the TPD computer control system. TPD spectra were obtained at a constant sample heating rate of 2 K sec⁻¹, and the desorption rate of the selected m/z masses was monitored quantitatively using the quadrupole spectrometer. To increase the sensitivity of the mass spectrometer detector during TPD, a nozzle was used that approached the surface of the sample (at a distance of about 1 mm), which allowed collection of the desorbed products from the sample and improved the signal-to-noise ratio. We use isotopic labeling in order to differentiate between water adsorbed from gas residues and water formed as a consequence of ion bombardment.

Zoom In Zoom Out Reset image size

The specific adsorption capacity of different minerals used in this study was determined through FTIR measurement. A known mass of a sample (100 mg) was introduced into a sample finger, as shown in Figures 1 and 15 in Appendix B. Then the sample was heated (600 K) under vacuum for 2 hr to remove trace water (attained pressure 0.001 Torr). The dried sample was then cooled to ambient temperature and exposed to a saturated pressure of D₂O (18 Torr) for 1 hr. Subsequently, the sample was evacuated at room temperature until the gas phase contained no detectable D₂O (10⁻³ mbar). Finally, the sample was heated to 600 K to release the adsorbed heavy water. Gas-phase spectra were measured, analyzed, and the adsorption capacity of the mg D₂O/mg sample was calculated. Infrared spectra were collected using a Bruker IFS 120 spectrometer as described above.

The formation of water was observed by irradiation of the sample surface by hydrogen ions produced in a MW discharge. The formation of excited H atomic species was observed, which implies the presence of H⁺ ions as well. An experimental H atomic spectrum is shown in Figure 16 in Appendix C. Fourteen assorted minerals and catalysts were tested for water formation activity. These included synthetic minerals, such as TiO₂ anatase and nanorutile from Aldrich, in-house synthetic TiO₂ P25 anatase, synthetic Fe₂O₃, montmorillonite, kaolinite, and Al₂O₃ (Sigma Aldrich). Two meteorites were also included: Ramlat as Sahmah 445 (RAS 445) and Sayh al Uhaymir 567 (SAU 567). Tested natural samples included diallage (collected in Dzikowies, Poland), olivine (grains in basalt, collected in Smrčí u Semil, Czechia), oligoclase (collected in Vlastějovice, Czechia) and augite (collected in Bořislav, Czechia).

Figure 3 shows spectra of D₂O formed on the surface of mineral samples listed in Table 1 in a MW discharge. The spectra show the IR absorption bands of D₂O. Its formation was observed in all samples except TiO₂-P25, as discussed below. All the samples represent oxygen-containing minerals. Figure 3 shows three IR spectra. The top spectrum shows D₂O reference bands. The middle and bottom spectra depict the formation of D₂O on the surfaces of Fe₂O₃ and Al₂O₃. Other results are shown in Appendix A. Figure 12 shows water formation on the surface of montmorillonite, augite, and kaolinite. Figure 13 shows water formation after irradiation of oligoclase, olivine, and diallage, all of them natural minerals. Finally, Figure 14 shows IR spectra of water formed after irradiation of the two meteoritic samples, RAS 445 and SAU 567.

Zoom In Zoom Out Reset image size

As was shown in Table 1, D₂O molecules were identified in all experiments related to the application of MW discharge, except for TiO₂ P25 {101} anatase. In this case, as shown in Figure 4, even after multiple repetitions of the experiment, we were unable to detect any water formed on the surface of the TiO₂ {101} sample.

Zoom In Zoom Out Reset image size

During the experiment, the sample of the oxide material is exposed not only to a strong current of H, H⁺, and molecular hydrogen in excited Rydberg states (see Figure 16 in Appendix C), but also to intense visible and UV radiation generated in the hydrogen discharge.

The overall photoelectrochemical water splitting reaction (2H₂O ⟶ 2H₂ + O₂) was achieved on a TiO₂ photoanode as early as 1972 (A. Fujishima & K. Honda 1972), with many studies following (R. Li et al. 2015; H. Eidsvag et al. 2021). Studies on the photoelectric activity of TiO₂ show that, in particular, anatase dominated with {101} facets is highly effective for hydrogen production from water splitting (H. G. Yang et al. 2008; K. M. Macounová et al. 2017).

Generally, photocatalytic water splitting involves the conversion of water (H₂O) into hydrogen (H₂) and oxygen (O₂) through the use of a catalyst and natural light (in our experiment visible and UV light generated in the hydrogen MW discharge). In the initial step, under irradiation, electron–hole pairs are generated in a semiconductor photocatalyst, prompting the excitation of electrons from the valence band (VB) to the conduction band. In the second step, charge separation occurs as the photogenerated electrons migrate toward the surface. Upon reaching the surface, the photogenerated holes react with water to produce oxygen and H⁺, and the electrons present recombine with the H⁺ to produce H₂. The redox and oxidation reactions on the photocatalyst's surface are represented by the following equations:

$$\begin{eqnarray}&&\mathrm{Oxidation}:\ {{\rm{H}}}_{2}{\rm{O}}+2{{\rm{h}}}^{+}\longrightarrow 2{{\rm{H}}}^{+}+\displaystyle \frac{1}{2}{{\rm{O}}}_{2},\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&\mathrm{Reduction}:\ 2{{\rm{H}}}^{+}+2{{\rm{e}}}^{-}\longrightarrow {{\rm{H}}}_{2},\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&\mathrm{Overall}\ \mathrm{reaction}:\ {{\rm{H}}}_{2}{\rm{O}}+1.23\mathrm{eV}\longrightarrow {{\rm{H}}}_{2}+\displaystyle \frac{1}{2}{{\rm{O}}}_{2}.\end{eqnarray} \tag{ 3 }$$

The process of water splitting is highly endothermic, with a reaction Gibbs free energy of 1.23 eV (corresponding to light with a wavelength 1008 nm). To initiate the reaction, the photocatalyst must have a band gap greater than 1.23 eV, typically between 1.6 eV and 1.8 eV in practice, considering overpotentials. However, the band gap should not be excessively high, as it would diminish the photocatalyst's ability to absorb visible light. In addition to an appropriate band gap, the conductive band minimum must be more negative than the redox potential of H⁺/H₂ (0 V versus normal hydrogen electrode), and the VB maximum must be more positive than the redox potential of O₂/H₂O (1.23 V). Another very important factor for photocatalyst efficiency is the recombination time, i.e., the duration for an electron–hole pair to recombine. If the electron–hole pair recombines before reaching the surface, the reaction is prevented. Other important material properties include crystallinity, dimensionality, pH dependency, and band gap. Besides these properties, TiO₂ is also known to freely exchange oxygen atoms in CO₂ at room temperature (D. C. Sorescu et al. 2014; S. Civiš et al. 2016) and reduce CO₂ through methanogenesis to generate CH₄ or other hydrocarbons on the surface (S. Civiš et al. 2017; S. Civiš 2021). The usage of the isotopically labeled samples in this study was inspired by the referenced papers.

One other aspect of the photocatalytic reduction process concerns the number of required electrons. R. Li et al. (2015) report a lack of O₂ evolution from water splitting on anatase TiO₂ during the initial stage of the reaction. This may be attributed to the existence of trapped states in the bulk and surface regions. The presence of these states reduces the overpotential for water oxidation, making the photogenerated holes more inclined to oxidize the H₂O molecules and produce the OH radical in a two-electron process. This process requires less potential than a four-electron process to directly produce O₂. UV light with high intensity can gradually reduce the deep trapped states near the VB of anatase TiO₂, increasing the overpotential for water oxidation and enabling H₂O oxidation through the four-electron process to yield O₂, which appears in the reaction later on.

Therefore, it is safe to assume that water molecules generated on the surface of TiO₂ in the atmosphere of hydrogen during our discharge experiment are simultaneously decomposed by UV radiation generated by the MW discharge, likely through the simplest reduction process resulting in the formation of H₂ molecules. In our MW discharge experiment, it was technically infeasible to separate the sample bombarded by H⁺ ions from the relatively strong UV radiation generated within the hydrogen MW discharge.

Another approach to observe the experimental formation of water is the utilization of a sputter gun and irradiation of a mineral sample directly by hydrogen/deuterium ions. Six variants of the experiment were performed, as shown in the Table 2.

3.3.1. Formation of H₂ ¹⁸O with Sputter Gun Monitored with TPD on Ti¹⁸O₂

First, a Ti surface was cleaned by bombardment with Ar⁺ ions (4 kV, 10 μA, 10 minutes) at 300 K. After purification, TPD spectra were recorded for control. On the cleaned Ti surface TiO₂ oxide was formed by bombardment with ¹⁸O⁺ ions (500 eV, 6 μA, 10 minutes); the layer of the formed oxide with isotopically labeled oxygen was again checked by TPD. In the next step, the surface of the formed oxide was bombarded with H⁺ ions (500 eV H₂, 10⁻⁶ Torr, 6 μA, 10 minutes). The water formed by the reaction of hydrogen ions with titanium dioxide was finally released during the final TPD measurement. The result is shown in Figure 5.

Zoom In Zoom Out Reset image size

3.3.2. Formation of D₂ ¹⁶O and H₂ ¹⁸O with Sputter Gun Monitored with TPD on Ti^16/18O₂/Au

Here, 50 mg of Ti¹⁶O₂/Ti¹⁸O₂ was finely ground in a friction pan. Then a suspension was prepared by adding approximately 1 ml of acetonitrile. The acetonitrile suspension was dripped onto an Au plate mounted on a tantalum sample holder. After evaporation of the solvent, the sample was placed in a vacuum chamber and evacuated to 10⁻⁷ mbar. After 24 hr, the sample was transferred from the load lock chamber to the TPD chamber (10⁻⁹ mbar). The sample was initially checked for contamination by TPD ramp. The temperature was measured using a thermocouple attached to the sample holder.

In the next step the surface was bombarded by D⁺ or H⁺ ions (500 eV D₂ or H₂ 10⁻⁶ Torr, 6 μA, 10 minutes) to create water molecules. The exposed sample surface was approximately 7 mm in diameter, and the spot size of the incident ion beams was >5 cm. Finally, TPD spectra of the created heavy water D₂O ions m/z 20 and HDO ions m/z 19 were recorded.

Two experiments were conducted using this setup. First, Ti¹⁶O₂ was irradiated with D⁺ and the formation of D₂ ¹⁶O was observed. The result is shown in Figure 6, which shows the released D₂ ¹⁶O in millibars.

Zoom In Zoom Out Reset image size

The second experiment in this setup was the formation of H₂ ¹⁸O. Here, in-house synthesized Ti¹⁸O₂ was used and irradiation of H⁺ was employed. As expected, the formation of H₂ ¹⁸O was observed. The resulting TPD spectrum is shown in Figure 7.

Zoom In Zoom Out Reset image size

3.3.3. Formation of D₂ ¹⁶O with Sputter Gun Monitored with TPD on Mineral Oxides

In a manner similar to the experiment described above, acetonitrile solutions of Al₂O₃, Fe₂O₃, and ilmenite were dripped onto an Au foil and irradiated with D⁺ ions. Again, the formation of D₂O was observed, further confirming the formation of water on the tested minerals. The resulting spectra are shown in Figures 8–10.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The isotopic composition of water on Earth and other terrestrial planets can provide clues about its origin and persistence. The ratio of deuterium to hydrogen (D/H) can indicate, for example, the degree of water loss or fractionation that occurred during planetary evolution. The D/H ratio of Earth's oceans is about 1.6 × 10⁻⁴, while that of Venus's atmosphere is about 1.9 × 10⁻², suggesting that Venus lost most of its water through atmospheric escape. The D/H ratio of Mars's atmosphere is about 4.5 × 10⁻⁴, indicating that Mars also lost a significant amount of water over time.

The amount of water on each terrestrial planet is also uncertain and may have changed over geological history. Earth has about 2.3 × 10²¹ kg of water, which is about 0.04% of its mass. Venus may have had at least 0.3% of Earth's hydrosphere in the past, but now has only about 200 mg kg⁻¹ of water vapor in its atmosphere. Mars may have had a global ocean covering about a third of its surface in the past, but now has only about 3 × 10¹⁶ kg of water ice at its poles and subsurface. Mercury and the Moon appear to be very dry, but may have some water ice trapped in permanently shadowed craters near their poles.

There are two end-member possibilities for the presence of water during accretion of planets. One possibility is that temperatures were too high in the inner solar system for hydrous phases to exist in the accretion disk, so the terrestrial planets accreted "dry." The other possibility is that the terrestrial planets accreted "wet" from hydrous materials that formed or migrated in the inner solar system. Similarly, the possible sources of water can be divided into endogenous and exogenous categories. Endogenous sources include direct adsorption of water from gas onto grains in the accretion disk and accretion of hydrous minerals forming in the inner solar system. Exogenous sources include delivery of water by asteroids, comets, and planetary embryos from beyond the snow line. Explanation of the origin and isotopic composition of Earth's volatiles based on isotopic ratios of major elements, such as C/H, C/N, C/S, etc., "involves a combination of parent body processing, core formation, catastrophic atmospheric loss, and partial replenishment by late veneer" (M. M. Hirschmann 2016).

Through observations and laboratory experiments, it has been established that when rocky materials are exposed to solar wind irradiation, a chemical reaction occurs between hydrogen ions and silicate minerals, resulting in the formation of hydroxyl (OH) and water molecules (K. L. Laferriere et al. 2022; P. G. Lucey et al. 2022). This has been investigated by M. J. Schaible & R. A. Baragiola (2014), who studied hydrogen implantation on amorphous SiO₂ and olivine with a proton beam. Although they observed the formation of OH (in SiOH complexes), they found no evidence of molecular water formation. We used olivine in our experiment, as well, and in principle one could expect to observe the same results from both experiments, even though the proton sources were different. Another striking difference in the experiments, though, is the detection method. M. J. Schaible & R. A. Baragiola (2014) use transmission FTIR on a solid sample and observe Si–OH absorption bands in their sample. Naturally, this detection technique is limited to a lower resolution (4–8 cm⁻¹). Our analysis is based on evaporating the water that forms on the surface and the subsequent measurement of gas-phase FTIR absorption spectra with a resolution 0.03 cm⁻¹. This gives us a much lower detection limit. On the other hand, we are unable to observe the complexes and adsorbates on the solid sample, giving us no insight into the formation mechanism. A similar experiment was performed by Q. Jiang et al. (2024), who studied hydrogen implantation on the surfaces of olivine, orthopyroxene, and quartz. They found that the hydrogen saturation level by implantation is much higher than the bulk solubility limit and also that the saturation capacity depends on the nature of the mineral, very much in line with our results. The samples were analyzed with the nuclear resonance reaction analysis technique, and the authors observed hydrogen depth profiles in the samples. Even though the authors did not directly observe any formation of water, they estimate the production of water on Earth through this process upon the assumption that all implanted hydrogen is converted into molecular H₂O, which is between 40 and 400 ocean masses for 0.1–1 μm dusts in about 100 yr. As the authors discuss, this calculation utilizes multiple simplifications and constraints, but can be regarded as an upper limit. In contrast, we observe directly the water formation and can relate the observed value to the sample loading. On the other hand, we have no direct insight into the surface formation mechanism and, even though we do not need as many parameters for the calculation, our values are rather more empirical.

The water observed in our experiment and formed by solar wind can become entrapped within an upper layer, typically 20–200 nm in thickness (J. P. Bradley et al. 2014). This offers a plausible explanation for the presence of water in regoliths on airless celestial bodies like the Moon. Additionally, the implantation of solar hydrogen could elucidate why nominally anhydrous minerals in asteroids exhibit water abundances of several hundred parts per million by weight (J. L. Bandfield et al. 2018; L. Piani et al. 2021).

Compelling evidence for these reactions is further corroborated by the detection of OH bonds and the inferred water contents, which can reach up to 1 mol% on the surfaces of IDPs that have been exposed to solar wind irradiation. These findings are also substantiated by the production of water in minerals during laboratory experiments simulating space weathering (J. P. Bradley et al. 2014). This body of evidence suggests that a volatile reservoir, characterized by isotopic similarity to the solar wind, might have existed within the solar system and potentially contributed to the formation of Earth's oceans.