In situ XPS of competitive CO2/H2O absorption in an ionic liquid

Superbasic ionic liquids (SBILs) are being investigated as potential carbon dioxide (CO2) gas capture agents, however, the presence of H2O in the flue stream can inhibit the uptake of CO2. In this study a thin film of the SBIL trihexyltetradecylphosphonium 1,2,4-triazolide ([P66614][124Triz]) was deposited onto rutile TiO2 (110) using in situ electrospray deposition and studied upon exposure to CO2 and H2O using in situ near-ambient pressure x-ray photoelectron spectroscopy (NAP-XPS). The molar uptake ratio of gas in the electrosprayed SBIL (n gas :n IL) was calculated to be 0.3:1 for CO2, 0.7:1 for H2O, and 0.9:1 for a CO2/H2O mixture. NAP-XPS taken at two different depths reveals that the competitive absorption of CO2 and H2O in [P66614][124Triz] varies with sampling depth. A greater concentration of CO2 absorbs in the bulk layers, while more H2O adsorbs/absorbs at the surface. The presence of H2O in the gas mixture does not inhibit the absorption of CO2. Measurements taken during exposure and after the removal of gas indicate that CO2 absorbed in the bulk does so reversibly, whilst CO2 adsorbed/absorbed at the surface does so irreversibly. This is contrary to the fully reversible CO2 reaction shown for bulk ionic liquids (ILs) in literature and suggests that irreversible absorption of CO2 in our highly-structured thin films is largely attributed to reactions at the surface. This has potential implications on IL gas capture and thin film IL catalysis applications.


Introduction
The capture and sequestration of carbon dioxide (CO 2 ) has been an important process in the reduction of CO 2 emissions in recent years.Currently, fossil fuel-fired power plants use alkanolamine solvents such as monoethanolamine (MEA) for CO 2 capture.Despite reducing a power plant's CO 2 emissions by as much as 90% [1], these industrial solvents generate problems of their own, with issues such as toxicity and long-term costs challenging their suitability [2].Ionic liquids (ILs) have seen a recent growth of interest as promising green and non-toxic alternatives to current CO 2 capture solvents.ILs are liquid salts consisting of pairs of bulky anions and cations held together by strong coulombic forces [3].The chemical and physical properties of an IL can be fine-tuned via the cation-anion pair.Their high CO 2 capacity, very low volatility, high thermal stability, and lower regeneration temperature compared to MEA make them desirable alternatives [4][5][6].The tunability, low melting points and wide electrochemical windows of ILs have led to their use in a wide range of other applications across electrochemistry (electrolytes for batteries [7], nanostructure growth Reaction with CO2 at N 1 results in the formation of carbamate.Reaction with H2O forms triazole and a hydroxide ion.N 1,r (red, light blue) and N 2,r (green) denote the N atoms following reaction of the anion with CO2 or H2O.(d) Gas exposure regime for the ESD1 [P66614][124Triz] electrosprayed thin film.[8,9], and solar cells [10,11]), tribology (lubricants [12,13] and corrosion inhibitors [14]) and catalysis (solid catalysts with an IL layer (SCILL) and supported IL phase (SILP) catalysts [15,16]).
An important property for ILs in carbon capture is their selectivity for CO 2 .Industrial flue gas contains a complex mixture of gases such as CO 2 , H 2 O, O 2 , N 2 , CO, SO x and NO x [17,18].Therefore, it is vital to understand the effect of a mixture of gases on the IL and how they compete for absorption.For conventional ILs with imidazolium-based cations, the anion has been found to be responsible for the solubility of these gases [19,20], but also dictates some of the physical properties of the IL such as its density, viscosity and melting point [21].Alternatives to conventional physically absorbing ILs came in the way of anime-functionalised task-specific ILs which chemically absorb CO 2 , leading to higher CO 2 capacities [6,22].These ILs, however, suffer increased viscosity upon absorption of CO 2 , which consequently hinders further CO 2 absorption and limits their current applicability [23,24].An emerging class of ILs called superbasic ILs (SBILs) show great promise for CO 2 capture due to their higher CO 2 capacity compared to conventional ILs and the ability to chemically and reversibly absorb CO 2 without significant changes in their viscosity [25].
Commonly, SBILs have phosphonium-based cations and imidazolium-based anions, with anions such as [benzim] − and [124Triz] − yielding the highest CO 2 capacities and solubilities [26,27].In SBILs, the anion chemically reacts with CO 2 in the same way as MEA, forming a carbamate species [22].In fact, the SBIL [P 66614 ][124Triz] is able to both chemically and physically absorb CO 2 [28].Mercy et al used density functional theory (DFT) simulations to study CO 2 capture in the SBIL [P 3333 ][124Triz] [29].They found that while the cation plays no direct role in the chemisorption of CO 2 , the presence of the cation increases the negative charge of the N 1 nitrogen site in [124Triz] − (see scheme 1(b)).This increases the reactivity on N 1 , which is considered the most favourable reaction site in the anion.Unlike conventional physically absorbing ILs, SBILs are capable of achieving greater than equimolar amounts of CO 2 absorption.Taylor et al reported a CO 2 uptake of 1.20:1 (n CO2 : n IL ) in the dry state for the SBIL trihexyltetradecylphosphonium benzimidazolide, [P 66614 ][benzim] [26].SBILs have been found to only experience minor reductions in CO 2 capacity in the presence of water [26], SO 2 [30], and NO [25], and in some cases preferentially and reversibly absorb CO 2 in the presence of other gases [31].
X-ray photoelectron spectroscopy (XPS) is a technique traditionally carried out at ultra-high vacuum (UHV), but advancements in electron optics and the use of synchrotron light and instrumentation have driven development of the technique to near-ambient pressure (NAP) (up to about 20 mbar) which allows investigations of liquid/gas interfaces [32].This technique has been used to study interactions between ILs and gases [33][34][35], including SBILs [31].For example, a study by Henderson et al revealed that the adsorption of water vapour at the surface of ultra-thin films of [C 4 C 1 Im][BF 4 ] induced a reordering of the ions at the IL/gas interface [34].Water vapour remained trapped at the surface for some time even after the gas was evacuated.Greer et al investigated the competitive absorption of CO 2 and NO in [P 66614 ][benzim] [25].Although the absorption of NO resulted in irreversible bonding of NONOate species to the [benzim] − anion, the presence of NO had little effect on the reversibility of CO 2 absorption and the CO 2 capacity of the IL.
There is currently limited information on how interactions between CO 2 molecules and the IL influence structure and ordering at the IL/gas interface, and how these interactions vary through the IL layers.Herein, we report an in situ NAP-XPS study into the ordering and interactions of electrosprayed thin films of the SBIL [P 66614 ][124Triz] on rutile TiO 2 (110) before, during, and after exposure to CO 2 and H 2 O.To the best of our knowledge, this will be the first use of NAP-XPS at two different sampling depths to study CO 2 absorption in ILs.These results provide new insights into competitive absorption of CO 2 and H 2 O in SBILs and the reversibility of these reactions.Results indicate that more CO 2 absorbs in the bulk layers of the thin film, and this absorption is reversible.However, less CO 2 ad/absorbs at the surface and the ad/absorption is irreversible.Water vapour reversibly ad/absorbs at the surface and does so in greater concentrations compared to the bulk layers.

Method
Experimental measurements of the SBIL trihexyltetradecylphosphonium 1,2,4-triazolide ([P 66614 ][124Triz], structure shown in scheme 1) were carried out at beamline HIPPIE at MAX-IV synchrotron in Sweden (photon energy range 255-2200 eV) [36].The HIPPIE endstation analysis chamber has a hemispherical Scienta-Omicron HiPP-3 electron energy analyser positioned in the plane of the storage ring at 55 • to the direction of the incoming x-ray beam [36].A NAP cell is docked to the analyser to carry out NAP experiments.All XPS measurements were taken at normal emission (giving a 35 • angle between the incident x-ray beam and the sample surface).
A rutile TiO 2 (110) single crystal (PI-KEM) was cleaned via Ar + sputter/anneal cycles (sputtering at 1 keV for 10 min and annealing at 700 • C for 10 min) until XPS spectra showed no contamination.[P 66614 ][124Triz] was deposited onto the rutile TiO 2 (110) substrate via electrospray deposition (ESD) in vacuum using a Molecularspray UHV4 system.The deposition chamber had a base pressure of 2.0 × 10 −10 mbar and a deposition pressure of 7.0 × 10 −10 mbar.A 0.02 M [P 66614 ][124Triz]/methanol solution was fed into the emitter capillary by a syringe pump delivering a flow rate of 0.3 ml h −1 .The emitter, syringe and tubing were cleaned prior to use by flushing with the solvent.3.0 kV was applied to the emitter with respect to the grounded entrance capillary.Two electrospray thin films were deposited, a 2.3 ± 0.6 nm (ESD1) and a 6.2 ± 1.8 nm (ESD2) thin film.Film thickness calculations are given in the supplementary information (SI).A beam damage study showed no degradation of the IL upon prolonged exposure to the beam, but the sample was still moved ∼0.1 mm between scans to avoid possible charging and degradation of the IL.
The ESD1 thin film of [P 66614 ][124Triz] was characterised using photoemission while exposed to CO 2 , H 2 O and finally a CO 2 /H 2 O gas mixture.The IL film was exposed to each of the gases for about 2 h whilst XPS scans were taken.Gases were leaked into the NAP cell at approximately 1 mbar at room temperature.The molar ratio of CO 2 :H 2 O in the gas mixture was 1.1:1 ± 0.1, calculated using the areas of the gas-phase peaks in the O 1s spectra.Measurements were taken during exposure and after the gas was pumped out of the NAP cell after each stage in order to study the reversibility of absorption.This exposure procedure is detailed in scheme 1(d).An XPS study was carried out at two different depths on the [P 66614 ][124Triz] ESD2 thin film by changing the photon energy as summarised in table 1.These photon energies result in photoelectrons with maximum kinetic energies of 150 and 600 eV, corresponding to surface and bulk sampling depths, respectively.Surface and bulk layers of the ESD2 6.2 nm thin film were probed at sampling depths of 1.6 and 4.0 nm, respectively.Measurements were taken before exposure, during exposure to CO 2 , and during exposure to H 2 O, both at 1 mbar (2 h gas exposure time).A summary of the photon energies used for each photoemission region is presented in table 1.
All XPS peaks [37] have been fitted using 30:70 (Lorentzian:Gaussian) line shapes and a linear background, using the software CasaXPS [38].The binding energy (BE) scale for all regions has been calibrated to the alkyl C 1s signal at 285.0 eV and all fitted peaks are quoted to ±0.1 eV BE [39].

[P 66614 ][124Triz] NAP-XPS
NAP-XPS measurements were carried out on an electrosprayed 2.3 nm thin film (ESD1) of the IL [P 66614 ][124Triz].These were taken as the IL was exposed to CO 2 , H 2 O, and finally a CO 2 /H 2 O mixture, all at 1 mbar.Measurements were also taken prior to exposure and after the gas was removed between each exposure stage, as summarised in scheme 1(d), in order to investigate the reversibility of absorption.Survey spectra taken at each stage of exposure are shown in figure S1 of the SI.The C 1s region and fitted components are shown in figure 1 for all stages.The C 1s spectra are intensity normalised to the peak at 285.0 eV.The main C 1s peak at 285 eV can be fitted with two components attributed to the [P 66614 ] + cation.These occur at 285.0 and 285.7 eV in all stages and are assigned as C aliphatic and C hetero , respectively (see scheme 1(a)).These agree with previous reports on the [P 66614 ] + cation [39].A third component at 287.1 eV is attributed to carbon atoms in positions 3 and 5 in the [124Triz] − anion (C 3,5 ).
Exposing the electrosprayed IL to CO 2 in Stage 2 results in a gas-phase CO 2 peak at 293.3 eV (not shown) and a carbamate peak (see scheme 1(b)) at 288.3 eV.As expected, there is no carbamate peak present prior to CO 2 exposure in Stage 1.In addition to a 0.2 eV chemical shift down to 286.9 eV, the C 3,5 component increases in intensity during exposure to CO 2 , reverting to its original intensity and BE when the gas is removed in Stage 3. Similarly, the carbamate peak reduces in intensity but does not reduce to zero when CO 2 is removed, indicating that the reaction with CO 2 is not fully reversible.Hereinafter, this will be referred to as residual carbamate.It has been shown that bulk depositions of [124Triz] − ILs reversibly absorb CO 2 by heating the IL, even in the presence of water [26].A thick film of a similar SBIL has been shown to reversibly absorb and desorb CO 2 by pumping out the gas [31].However, as shown here, thin films of [P 66614 ][124Triz] cannot be regenerated through removing the surrounding gas.Only one CO 2 exposure/desorption stage was recorded in our experiment so we do not know how multiple cycles would affect the uptake and reversibility of absorption.However, in previous work, multiple CO 2 exposure/desorption cycles in the similar SBIL [P 66614 ][benzim] were studied using in situ attenuated total reflectance-infrared spectroscopy as shown in figures S2 and S3 in the SI.Results show an increase in the carbamate band over multiple exposure/desorption cycles and the presence of small amounts of residual carbamate after desorption, corroborating our XPS data.
When the IL is exposed to H 2 O in Stage 4 there is no significant change in intensity of the residual carbamate peak.This would suggest that the absorption of H 2 O, which reacts at the same N 1 site as CO 2 , does not remove the irreversibly absorbed CO 2 .The C 3,5 component also shifts down to 286.7 eV and increases in intensity, similar behaviour to that found in Stage 2 (IL + CO 2 ).When the H 2 O is removed in Stage 5, the residual carbamate peak persists, suggesting that the absorption and desorption of H 2 O has little effect on the absorbed residual carbamate.Overall, the H 2 O pump out spectrum closely resembles that of the CO 2 pump out stage, with the C 3,5 components showing similar intensities and BEs (287.1 eV).This suggests that the IL's reaction with H 2 O is reversible.
When the electrosprayed IL is exposed to the CO 2 /H 2 O mixture in Stage 6, the carbamate peak increases again to a similar intensity to that of Stage 2 (IL + CO 2 ).This further suggests that the presence of H 2 O does not inhibit the absorption of CO 2 and formation of carbamate, which is in agreement with other reports for [124Triz] − -based ILs [26].Note that the 1 mbar CO 2 /H 2 O mixture has an approximately equal number of CO 2 and H 2 O molecules, therefore one might expect the intensity of the carbamate peak to be lower than the 1 mbar CO 2 exposure.The fact that these intensities are similar suggests that a much greater pressure is needed to fully saturate the IL with CO 2 .The C 3,5 component (which shifts again to 286.6 eV) also increases to a similar intensity as that in Stages 2 and 4, when the IL is exposed to CO 2 and H 2 O, respectively.The change in BE and intensity of the C 3,5 component during gas exposure Stages 2, 4 and 6 is not fully understood.The small BE shifts may be due a change in chemical environment of C 3,5 upon chemical reaction or physical interaction between the anion and gas.The change in intensity could be explained by a reordering of anions upon exposure to gas.Note that for the CO 2 /H 2 O mixture, reaction between the two gases may form carbonic acid (H 2 CO 3 ) [40].It is possible this may interact with the IL thin film, though we find no evidence of carbonic acid (∼289 eV) in our spectra.Note that in this experiment we expose the IL to CO 2 first then H 2 O.If we were to expose the IL to H 2 O first then CO 2 , the reversibility of the reactions may be affected.This would make an interesting further study.
Figure 2 shows the N 1s region for all six stages.The spectra have been normalised by peak area between 396 and 404 eV.The N 1s peak can be fitted with three components, the first two of which at 399.1 and ).N 1 and N 2 are chemically equivalent due to resonance effects [41], confirmed by molecular orbital calculations using the software ORCA [42,43].The IL film was prepared by electrospray in vacuum so we would not expect any significant higher BE peaks attributed to reaction with gas.However, there is a small peak at 401.1 eV which may be due to reactions with trace amounts of gas remaining in the UHV chamber.This peak is assigned to atom N 2,r in the reacted anion.We would expect another peak at higher BE attributed to N 1,r (atom at which gas reacts), but this cannot be resolved here.It is unclear whether the IL has absorbed CO 2 , H 2 O, or both in this case because the N 2,r component remains at this same BE throughout each of the following exposure stages (the same is also true for the N 4 and N 1,2 components).
Upon exposure to CO 2 in Stage 2, a peak appears at 402.0 eV.We can assign this to N 1,r , the atom at which carbamate forms (N 1,r CO2 ).Similar to the C 1s region in figure 1, this carbamate peak reduces in intensity, but not to zero, when the CO 2 is removed in Stage 3, indicating irreversible CO 2 absorption.This residual carbamate peak appears in all following stages, consistent with the C 1s spectra in figure 1.During these stages all other components remain at similar BEs and intensities as those in Stage 1.The C 3,5   1).The N 1,r CO2 peak appears to be more intense after the H 2 O pump out stage compared to the CO 2 pump out stage.This could indicate a reordering of residual carbamate species to the surface (since measurements were taken at a surface-sensitive photon energy).This is discussed in greater detail below (figure 4).
Finally, when the IL is exposed to the CO 2 /H 2 O mixture in Stage 6 the spectrum closely resembles that of Stage 4 (IL + H 2 O) except for a more intense N 2,r component at 401.1 eV.The pressure in Stage 6 rose slightly higher than 1 mbar (1.3 mbar), therefore resulting in a stronger N 2,r component due to a higher relative concentration of reacted anions.Despite an approximately equal CO 2 :H 2 O molecular ratio in the mixture, the N 1,r H2O component at 402.4 eV is much more intense than the N 1,r CO2 component at 402.0 eV.The reason why H 2 O absorption appears to dominate over CO 2 absorption in Stage 6 will be explored in detail below (see discussion for figure 4).Using the N 1s data in figure 2, the molar uptake ratio of the gases (n gas :n IL ) was calculated to be 0.3:1 for CO 2 , 0.7:1 for H 2 O, and 0.9:1 for the CO 2 /H 2 O mixture (each with an uncertainty of ±0.1).For the CO 2 /H 2 O gas mixture, the total molar uptake of 0.9 consists of approximately 0.15 for CO 2 and 0.75 for H 2 O, further suggesting that H 2 O absorption dominates but does not inhibit CO 2 absorption.See the SI for full details of calculations.Looking at the N 1s spectra, we expect less than 1:1 uptake of gas because the unreacted N 1,2 peak is still present in all spectra.If every anion had reacted then the N 1,2 peak would not be visible.The uptake of CO 2 is lower in our experiment than that found by Taylor et al for the same IL at atmospheric pressure (0.54:1) [30].This is to be expected as we are exposing a much thinner film of IL to lower pressures of CO 2 .
Another explanation as to why H 2 O absorption dominates over CO 2 absorption in figure 2 may be that the timescale over which measurements were taken plays an important role.It has been shown that for the IL [BMIM][OAc] the uptake of water into the bulk takes many hours, yet is much quicker at the IL-gas interfacial layers [44].Conversely, it has been found by Wang et al that bulk thicknesses of phosphonium-based SBILs with similar anions absorbed CO 2 very quickly, on the order of minutes [45,46].Admittedly these timescales are likely to change at the mbar pressures used in our study compared to atmospheric pressures used by Wang et al.Our measurements were taken over a 30 min timeframe at a surface sensitive sampling depth.Therefore, if [P 66614 ][124Triz] is similarly behaved to these ILs, then the reason that H 2 O absorption dominates over CO 2 absorption may be that there is an abundance of H 2 O at the surface due to a slower uptake of the gas into the bulk layers of the IL compared to CO 2 .
In the O 1s region in figure 3 there is a common peak at 530.0 eV for all stages, attributed to O atoms in the TiO 2 substrate lattice [47].The electrosprayed IL (black line) can be fitted with two more components at 531.1 and 532.1 eV.Since there are no oxygen atoms in the IL and we do not see any carbamate species in the C 1s region in Stage 1, we assign these to hydroxyls/defects at the TiO 2 surface [48].
When the electrosprayed IL is exposed to CO 2 (gold line) peaks are fitted at 530.1, 531.1, 532.1, 533.2, and 535.6 eV.These are assigned to TiO 2 , carbamate, C=O in protonated carbamate, C-OH in protonated carbamate, and gas-phase CO 2 (not shown), respectively.Carbamate species formed upon reaction with CO 2 can be protonated and it is likely that there is a mixture of protonated and unprotonated carbamate species present here [49].The 531.1 and 532.1 eV peaks in Stages 2-6 will also contain contributions from the hydroxyl species/defects at the TiO 2 surface.All fitted components in the O 1s region are summarised in table 2. When the CO 2 gas is removed in Stage 3 the broad reaction shoulder does not return to the original intensity from Stage 1.This provides further evidence of irreversible CO 2 absorption and reordering in the IL, shown previously in the C 1s and N 1s regions in figures 1 and 2, respectively.
When the IL is exposed to H 2 O in Stage 4, we see a significant increase in intensity of the reaction components.This would support our assumption that components at 532.1 and 533.2 eV are attributed to interactions with water vapour (C=O and C-OH in protonated carbamate species, respectively).Upon exposure to the CO 2 /H 2 O mixture in the final stage, the broad reaction peak has a similar intensity to that when exposed to H 2 O alone.This suggests that the presence of CO 2 does not inhibit the ad/absorption of H 2 O.The reason why H 2 O absorption appears to dominate over CO 2 absorption in figure 3 can be explained by exploring absorption at different depths.

[P 66614 ][124Triz] NAP-XPS depth study
NAP-XPS was carried out on an electrosprayed 6.2 nm thin film of [P 66614 ][124Triz] (ESD2) by probing two sampling depths: 4.0 nm to probe the bulk layers of the IL, and 1.6 nm to probe the surface layers.Photoemission measurements were taken for the electrosprayed IL, the IL exposed to 1 mbar CO 2 , and the IL exposed to 1 mbar H 2 O, for both bulk and surface sampling depths.Note that these sampling depths are expected to alter slightly during gas exposure stages due to attenuation of electrons through the gas.Since the ESD2 film is formed by depositing more IL on top of the previously measured ESD1 film, we expect some subtle differences in the spectra of the two films.All components remain fitted at approximately the same BEs as those in the previous section.
Figure 4 shows measurements in the C 1s region, with spectra normalised to the peak at 285.0 eV.For the electrosprayed IL, even before exposure to CO 2 there is a small carbamate feature at 288.3 eV for the surface sampling depth (grey line) which may be due to trace amounts of residual CO 2 in the chamber.Alternatively, it could indicate irreversibly absorbed CO 2 from previous exposures (since the ESD2 film was deposited over ESD1), which would further suggest that irreversibly absorbed CO 2 moves to the surface.There has been [124Triz] thin film before exposure, during exposure to CO2, and during exposure to H2O.The 6.2 nm thick sample was probed at two sampling depths: 4.0 nm to sample the bulk layers (denoted 'Bulk') and 1.6 nm to sample the surface layers (denoted 'Surface').Surface and bulk sampling depths were recorded at photon energies of 435 and 885 eV, respectively.evidence of irreversible CO 2 absorption in multilayer thin films of the similar IL [P 66614 ][benzim] in a previous study [49].
When the IL is exposed to CO 2 , the resulting carbamate peak is more intense for the bulk sampling depth (gold line) compared to the surface (red line).This implies that a greater concentration of carbamate species occurs in the bulk compared to the surface.Lewis et al have reported similar behaviour for aqueous MEA solutions treated with CO 2 [50].Greater concentrations of CO 2 -reacted MEA were found in the bulk of the solution while unreacted MEA was more concentrated at the surface.
When CO 2 is removed and the IL is exposed to H 2 O alone, a smaller residual carbamate peak remains in the IL + H 2 O spectra (blue lines) at 288.3 eV in figure 4. Similar evidence of irreversible CO 2 absorption was seen earlier in figure 1.The residual carbamate peak here is stronger when probed at the surface sampling depth (dark blue line) rather than in the bulk (light blue line), suggesting that more residual carbamate [124Triz] thin film before exposure, during exposure to CO2, and during exposure to H2O.The 6.2 nm thick sample was probed at two sampling depths: 4.0 nm to sample the bulk layers (denoted 'Bulk') and 1.6 nm to sample the surface layers (denoted 'Surface').Surface and bulk sampling depths were recorded at photon energies of 550 and 1000 eV, respectively.resides at the IL surface than in its bulk layers.In fact, the residual carbamate peak at the surface (dark blue line) is a similar intensity to the carbamate peak at the surface when exposed to CO 2 (red line).This implies that most of the CO 2 that absorbs within the surface layers of the IL does so irreversibly, remaining absorbed when the surrounding gas is pumped out.However, carbamate formed in the bulk (gold line) dramatically reduces in concentration when CO 2 is removed and H 2 O introduced (light blue line).This suggests that the irreversible nature of CO 2 absorption in this IL is largely attributed to reactions at the surface.
To summarise, more CO 2 absorption appears to occur in the bulk of the IL, and this absorption is largely reversible.Less CO 2 ad/absorbs at the surface and this is irreversible.These results show that both the concentration of carbamate and the level of reversibility varies with depth into the sample.
XPS measurements taken in the N 1s region are shown in figure 5.The spectra have been normalised by peak area between 396 and 404 eV.For the electrosprayed IL, the N 1s peaks are similar at the two sampling depths.As discussed in figure 2, N 1,2 atoms in the unreacted anion are largely responsible for the main peak centred on 400 eV, implying that the concentration of unreacted anions does not vary significantly through the IL layers prior to exposure.
Upon absorption of CO 2 , a carbamate N 1,r CO2 peak appears at 402.0 eV for both sampling depths.The N 1,r  CO2 peak is of higher relative intensity for the surface sampling depth (red line) compared to the bulk (gold line), however, comparing the two is tenuous given the very noisy nature of the surface sampling depth spectrum.This is most likely due to attenuation of photoelectrons through the CO 2 gas, considering they have low kinetic energies (maximum of 150 eV).
The absorption of H 2 O results in protonation of N 1 in [124Triz] − , denoted N 1,r  H2O , manifesting as a feature at 402.4 eV.Relative to the main N 1,2 peak and the irreversibly absorbed CO 2 N 1,r CO2 peak, the intensity of N 1,r  H2O is greater when probed at the surface (dark blue line) than in the bulk (light blue line).This suggests that there is a greater concentration of protonated N atoms at the surface than in the bulk.This is also supported by the more intense C 3,5 (C atoms in the reacted anion) component at the surface when exposed to H 2 O in figure 4. Results from figures 4 and 5 suggest that more H 2 O-reacted species remain at the surface of the IL film, and more CO 2 -reacted species diffuse through the surface to the bulk.This would explain why there is little evidence of CO 2 absorption in figure 3 (O 1s) and why H 2 O absorption appeared to dominate over CO 2 absorption when the IL was exposed to a CO 2 /H 2 O mixture in figure 2 (N 1s), because these spectra were taken at surface-sensitive photon energies of 680 eV and 550 eV, respectively.
In our previous study of the similar IL [P 66614 ][benzim], it was found that the presence of H 2 O did not significantly inhibit CO 2 absorption in thin films of the IL at near-ambient pressures [31].This behaviour could be explained by the results presented here.Reactions between the IL and the two different gases may occur primarily at different depths in the IL film.Additionally, the timescale over which the gases absorb in the IL may play an important role here, as these timescales have been shown to vary significantly for CO 2 and H 2 O in ILs [44,45] (although, these studies used different ILs and atmospheric pressures as opposed to mbar used in our study).

Discussion
Ordering of IL thin and thick films on solid surfaces has been investigated using a range of other experimental surface-sensitive techniques including angle-resolved NEXAFS [51,52], sum frequency generation [53,54], neutron reflectometry [55], atomic force microscopy [56,57], and x-ray reflectivity [58].Our approach of using XPS at two sampling depths has proven to be another useful tool in analysing the interfacial behaviour of IL thin films on solid surfaces, as well as their interaction with gases.The study presented here is largely qualitative due to limitations of the technique, for example, the high levels of noise as a result of attenuation of photoelectrons through the gas combined with low counts from the use of thin films.
In our thin films, the IL/vacuum and IL/TiO 2 interfaces are likely to have a greater influence than they do in bulk ILs.The ordered structure of ILs at these interfaces is likely to influence specific interactions with CO 2 /H 2 O and affect how these gases diffuse from the vacuum interface into the bulk.N 1 and N 2 are equivalent in isolated anions, and we would still expect N 1 and N 2 to remain equivalent at the IL/vacuum interface, however, we expect different ordering at the interface compared to the bulk.The charged parts of the ions in imidazolium-based ILs have been found to form an underlayer near the IL/vacuum interface, with the alkyl chains of the cations pointing out towards the vacuum [59].SBILs, such as those used in our experiment, have not been studied as widely in this context.In our previous XPS study of the related SBIL [P 66614 ][benzim], C 1s spectra showed a relatively more intense peak associated with the [benzim] − anion at grazing emission compared to normal emission, suggesting a higher concentration of [benzim] − anions at the IL/vacuum interface compared to the alkyl chains in the [P 66614 ] + cation [49].This study provides evidence of SBILs forming an ordered structure at the IL/vacuum interface.
Fundamental studies of ILs at TiO 2 surfaces are sparse so there is very little literature for comparison [16].Wagstaffe et al studied 4 Å and 30 Å thick films of an imidazolium-based IL at the anatase TiO 2 (101) surface using XPS [52].They found that in both films the two nitrogen atoms in the imidazolium ring are chemically equivalent and were assigned to a single peak in the N 1s spectra.However, this shifted by 0.2 eV to a lower BE in the thinner film due to the interaction of the imidazolium cation with the TiO 2 surface.It is possible that the N 1 and N 2 nitrogen atoms in [124Triz] − may react differently with the TiO 2 surface, and this may in turn affect how these anions react with CO 2 or H 2 O.If the anions were to react with the TiO 2 surface this could make them unavailable for reaction with CO 2 /H 2 O.A comprehensive computational study would be required to gain insights into these complex interfacial interactions.
In this study we used ESD to obtain thin films of [P 66614 ][124Triz].It is likely the film is not completely homogeneous and it is possible the IL could form islands on the surface rather than forming layers.However, we are unable to verify the growth mode from our data.Other methods have been used to deposit thin films of ILs, namely physical vapour deposition (PVD) [34,52,60] and dip-coating [57,61].Each have their own advantages and disadvantages; for example, PVD requires thermally stable ILs while ESD does not.ESD allows thin films of large molecules to be deposited in situ in ultra-high vacuum [62].However, electrosprayed thin films can have an unequal concentration of cations and anions compared to PVD due to the increased diffusion of cations and anions with positive and negative tip biases, respectively [63].ESD of ILs is comparatively underutilised, and prior to this study has not been used to deposit large IL molecules or used alongside NAP-XPS to investigate gas capture in thin films of ILs, to the best of our knowledge.
The results presented here have implications for thin film IL-based technologies.For example, thin films of ILs have been used to modify conventional supported catalysts, allowing the selectivity of the solid catalyst to be fine-tuned and improved (SCILL and SILP catalysis) [64].A problem these technologies face is their sensitivity to changes at the IL/gas phase and IL/solid support interfaces [65].Therefore, adsorption of gaseous/liquid reactants, products or contaminants in SCILL/SILP catalysts may induce changes in the surface structure of the IL thin film, potentially impacting the diffusion and selectivity for intermediate products in these catalysts [15].

Conclusion
The competitive absorption between CO 2 and H 2 O in electrosprayed thin films of the superbasic IL [P 66614 ][124Triz] has been characterised using in situ NAP-XPS.To the best of our knowledge, [P 66614 ] + is the largest IL ion to be successfully deposited via electrospray.Results suggest that both reacted and unreacted [124Triz] − anions reorder and diffuse through the IL thin film upon exposure to CO 2 and/or H 2 O.The NAP-XPS depth study revealed that greater concentrations of CO 2 -reacted species appear in the bulk layers of an electrosprayed IL thin film, reversibly forming carbamate on the anion.However, fewer CO 2 -reacted species appear at the surface layers, and this reaction is irreversible.H 2 O vapour ad/absorbs in greater concentrations at the surface rather than the bulk but does not inhibit the absorption of CO 2 .The molar uptake ratio of gases (n gas :n IL ) in the electrosprayed IL was calculated to be 0.3:1 for CO 2 , 0.7:1 for H 2 O, and 0.9:1 for the CO 2 /H 2 O mixture (each with an uncertainty of ±0.1).To the best of our knowledge, this is the first use of a NAP-XPS at different depths to study gas absorption in ILs.Reordering of IL thin films upon contamination with air and water vapour may also affect the performance of IL thin film-based technologies such as SCILL/SILP catalysis, IL lubricants and corrosion inhibitors.

Scheme 1 .
Scheme 1.(a) Chemical structure of the [P66614] + cation with C hetero atoms highlighted in magenta and C aliphatic in black.The reaction scheme for [124Triz] − with (b) CO2 and (c) H2O.Atoms N 1 and N 2 (blue) are equivalent due to resonance effects.Reaction with CO2 at N 1 results in the formation of carbamate.Reaction with H2O forms triazole and a hydroxide ion.N 1,r (red, light blue) and N 2,r (green) denote the N atoms following reaction of the anion with CO2 or H2O.(d) Gas exposure regime for the ESD1 [P66614][124Triz] electrosprayed thin film.

Figure 1 .
Figure 1.C 1s region recorded at a photon energy of 435 eV for an electrosprayed thin film of [P66614][124Triz] with various exposure regimes of CO2 and H2O at 1 mbar.The highlighted lines show common fitted components.The C 3,5 component shifts downwards in binding energy throughout the gas exposure stages.The inset shows how the carbamate and C 3,5 peaks vary in intensity throughout these stages.

Figure 2 .
Figure 2. N 1s region recorded at a photon energy of 550 eV for an electrosprayed thin film of [P66614][124Triz] with various exposure regimes of CO2 and H2O at 1 mbar.The grey lines show common components fitted through various stages of exposure.

Figure 3 .
Figure 3. O 1s region recorded at a photon energy of 680 eV for an electrosprayed thin film of [P6614][124Triz] with various exposure regimes of CO2 and H2O at 1 mbar.The grey lines show common components fitted through various stages of exposure.The inset shows how the broad reaction peak at ∼533 eV varies in intensity throughout these stages.

Figure 4 .
Figure 4. XPS of the C 1s region for an electrosprayed [P66614][124Triz] thin film before exposure, during exposure to CO2, and during exposure to H2O.The 6.2 nm thick sample was probed at two sampling depths: 4.0 nm to sample the bulk layers (denoted 'Bulk') and 1.6 nm to sample the surface layers (denoted 'Surface').Surface and bulk sampling depths were recorded at photon energies of 435 and 885 eV, respectively.

Figure 5 .
Figure 5. XPS of the N 1s region for an electrosprayed [P66614][124Triz] thin film before exposure, during exposure to CO2, and during exposure to H2O.The 6.2 nm thick sample was probed at two sampling depths: 4.0 nm to sample the bulk layers (denoted 'Bulk') and 1.6 nm to sample the surface layers (denoted 'Surface').Surface and bulk sampling depths were recorded at photon energies of 550 and 1000 eV, respectively.

Table 1 .
Photon energies used for each XPS region.ESD1 and ESD2 correspond to the two electrosprayed thin films of [P66614][124Triz], 2.3 and 6.2 nm thick, respectively.Measurements of ESD2 were taken at sampling depths near the surface and in the bulk of the film, achieved by changing the photon energy.
1,r H2O peak at 402.4 eV disappears completely, further indicating that this peak is due to a reversible reaction with H 2 O.This supports previous evidence of reversible H 2 O absorption discussed above (figure

Table 2 .
Assignments and corresponding BEs of fitted components in the C 1s, N 1s and O 1s regions for [P66614][124Triz] for various gas exposure stages.The arrows (→) denote chemical shifts of the C 3,5 component between the following exposure stages: IL → IL + CO2 → IL + H2O → IL + CO2 + H2O.
a C3,5component returns to this BE during the CO2 and H2O pump out stages.b Within the protonated carbamate group.c These peaks are assigned to hydroxyl species and defects at the TiO2 surface in Stage 1.