Combined Experimental and Theoretical Study of the Competitive Absorption of CO2 and NO2 by a Superbase Ionic Liquid

A superbase ionic liquid (IL), trihexyltetradecylphosphonium benzimidazolide ([P66614][Benzim]), is investigated for the capture of CO2 in the presence of NO2 impurities. The effect of the waste gas stream contaminant on the ability of the IL to absorb simultaneously CO2 is demonstrated using novel measurement techniques, including a mass spectrometry breakthrough method and in situ infrared spectroscopy. The findings show that the presence of an industrially relevant concentration of NO2 in a combined feed with CO2 has the effect of reducing the capacity of the IL to absorb CO2 efficiently by ∼60% after 10 absorption–desorption cycles. This finding is supported by physical property analysis (viscosity, 1H and 13C NMR, and X-ray photoelectron spectroscopy) and spectroscopic infrared characterization, in addition to density functional theory (DFT) calculations, to determine the structure of the IL-NO2 complex. The results are presented in comparison with another flue gas component, NO, demonstrating that the absorption of NO2 is more favorable, thereby hindering the ability of the IL to absorb CO2. Significantly, this work aids understanding of the effects that individual components of flue gas have on CO2 capture sorbents, through studying a contaminant that has received limited interest previously.


■ INTRODUCTION
Post-combustion CO 2 capture is an important requirement of many industrial processes. The high-temperature combustion of fossil fuels produces large quantities of CO 2 (10−15 vol %), as well as other impurities, such as SO 2 (0.05−0.2 vol %) and NO x (0.15−0.25 vol %), which consists predominantly of NO and NO 2 . 1−3 The reported values are obtained before any desulfurization or denitrification technologies. In particular, NO x is known to have a significant impact on health and the environment, causing the formation of atmospheric ozone and acid rain. 4 It is therefore vital that NO x emissions are regulated (40 μg·m −3 a year), leading to the fitting of NO x scrubbers to power stations, comprising oxidizing and reducing agents responsible for the conversion of NO x to N 2 . 5 Aqueous alkanolamines have been employed as CO 2 capture sorbents, but the presence of NO x was found to result in the irreversible formation of carcinogenic nitrosamines and a decrease in CO 2 capture efficiency. 6−8 Ionic liquids (ILs) have also been widely investigated for the capture of CO 2 as a non-volatile alternative to toxic alkanolamines. However, to date, the effect of NO 2 on the ability of an IL to capture CO 2 in a combined feed has not been investigated.
The interest in ILs stems from the ability to alter their physiochemical properties, such as thermal stability or CO 2 absorption capacity, through changing the combination of cation and anion, which enables the tuning of their properties for specific applications. 9,10 For example, the amount of CO 2 absorbed by a particular IL has been shown to have a strong dependence on the anion, with conventional anions only physically absorbing small quantities of CO 2 , 11,12 compared with task-specific ILs that incorporate amine functionality and chemically absorb up to 1 nCO 2 :nIL. 13,14 Superbase ILs (SBILs) containing an aprotic heterocyclic anion (AHA) were developed to minimize the increase in viscosity observed in amine-functionalized ILs, and they can reversibly capture a greater than equimolar amount of CO 2 . 15−18 Extensive studies into the absorption of other acidic gases such as SO 2 and NO by SBILs have found that irreversible absorption was observed in many cases, often on multiple active sites within the IL, affecting the recyclability of the system. 19−25 The effect of impurities on the CO 2 uptake of the SBIL t r i h e x y l t e t r a d e c y l p h os p ho n i u m b e n z i m i d az ol i d e , [P 66614 ] [Benzim], has previously been investigated in combined feeds. SO 2 was shown to deactivate the IL through binding to the absorption site available to CO 2 , while the presence of NO exhibited little effect on the IL's capacity for the uptake of CO 2 . 26,27 For NO, the co-binding of CO 2 and NO as carbamate and NONO-ate species, respectively, was observed at different N-sites of the benzimidazolide anion. However, competition for the same binding site was reported between CO 2 and SO 2 , which markedly influenced the absorption capacity and recyclability of the IL. The differing effects of SO 2 and NO on the absorption of CO 2 by [P 66614 ][Benzim] highlight the need to assess the components of flue gas impurities, both individually and in combination with CO 2 .
The capture of NO 2 by ILs has been the focus of only a few studies, where it was found that an increased uptake is observed for NO 2 compared to NO. 28−30 To date, NO 2 has not been studied in a combined feed with CO 2 , so its direct influence on the ability of a sorbent to capture CO 2 is unknown. The competitive absorption of CO 2 with industrially relevant concentrations of H 2 O, SO 2 , or NO, independently, has been investigated previously in [P 66614 ][Benzim], and this IL was therefore selected for the current study to gain a comprehensive insight into more complex, multi-component feeds. 18,26,27 The use of a recently developed analytical method utilizing mass spectrometry allows the study of this superbase IL under realistic and dry flue gas conditions, with a feed containing 14% CO 2 and 0.2% NO 2 . 26 Further molecular-level information was provided by density functional theory (DFT) calculations and spectroscopic data (NMR, IR, and X-ray photoelectron spectroscopy).
Methods. The gas absorption measurement techniques used in this work were reported in detail previously, and the same protocol was followed in this work. 26,27 To briefly summarize this, the uptake of a single component gas feed (1% NO 2 in argon) by [P 66614 ]- [Benzim] was studied gravimetrically at 22 ± 0.5°C. A mass spectrometer-based method was utilized to study the gas phase concentrations at the outlet after the IL was exposed to a mixed gas feed of 14% CO 2 + 0.2% NO 2 in Ar. A series of cycles were performed consisting of a 2 h absorption period under feed conditions at 22°C, followed by a 2 h desorption period under Ar at 90°C.
Analysis. The viscosity of the IL samples was measured before and after NO 2 absorption using a TA Instruments AR2000. Elemental analysis was carried out using a Thermo Scientific Flash 2000 elemental analyzer. X-ray photoelectron spectroscopy (XPS) was performed with a Kratos AXIS Ultra DLD apparatus, with a monochromated Al Kα radiation X-ray source, charge neutralizer, and hemispherical electron energy analyzer. During data acquisition, the chamber pressure was kept below 10 −9 mbar. The spectra were analyzed by CasaXPS and corrected for charging using the C 1s feature at 284.8 eV.
Attenuated total reflectance-infrared (ATR-IR) spectra were recorded in a modified in situ cell with a ZnSe crystal and a PIKE ATRMax II accessory housed in a Bruker Tensor II infrared spectrometer. A thin film of [P 66614 ][Benzim] (∼250 mg) coated the ZnSe crystal in the cell before the introduction of the gas feed (14% CO 2 in Ar, 0.2% NO 2 in Ar, or a mixed gas feed of 14% CO 2 with 0.2% NO 2 in Ar) with a flow rate of 15 cm 3 ·min −1 at 22°C. Desorption was performed at 90°C under Ar. The background for all spectra was the ZnSe crystal in the cell, and spectra were recorded with 8 scans at 4 cm −1 resolution. The spectrum of the IL before introduction of the gas feed has been subtracted from all the spectra of the IL under gas absorption.
DFT Calculations. DFT calculations followed a similar protocol to previous work on this system, and while a brief overview will be included here, a more detailed description can be found in the literature. 27 Calculations were performed with the Gaussian09 software package 32 using a combination of the hybrid functional UB3LYP and the triplet-ζ basis set 6-311+G*, as reported in previous works. 33 −35 Starting geometries for [P 3333 ][Benzim] models were also informed by previous molecular dynamical studies, 36 with absorbates manually added using the ChemCraft software package. 37 Minima structures for all possible reaction mechanisms were fully optimized without constraints with transition states located by initially running geometry scans, where only the degree of freedom connecting two minima was fixed. Full transition state optimizations were subsequently performed on the highest energy structures obtained along each reaction coordinate. The verification of both minima and transition states were carried out with the aid of analytical frequencies at 1 atm and 298.15 K, whereby only positive frequencies were observed for each minima, with each transition state possessing a single imaginary frequency for the mode associated with the reaction coordinate. Corrections for long-range interactions were included with the aid of the Grimme D3 dispersion model, 38 while solvent effects were simulated with an implicit model of acetonitrile (ε = 35.688) using a polarizable continuum model (PCM).

■ RESULTS AND DISCUSSION
The absorption of 1% NO 2 in Ar by [P 66614 ][Benzim] was initially examined by a gravimetric technique to allow direct comparison with the literature on the uptake of individual gases. It was found that an average of 4.60 nNO 2 :nIL was absorbed at saturation (Figure S1), with the higher than equimolar capacity indicating a multi-site absorption effect for NO 2 , as was found for NO. 27 The amount of NO 2 absorbed far exceeds that found for 1% NO by the same IL (1.73 nNO:nIL), indicating a different mechanism of absorption for NO 2 . 27 The regeneration of the IL at 90°C under argon was studied (well below the IL's decomposition temperature of 289°C), 18 with 3.64 nNO 2 :nIL remaining after 2 h. NO 2 absorption is evidently not a reversible process, with the absorbed species strongly bound to the IL, which has also been observed for caprolactam-based ILs. 28 The ability of NO 2 to compete with CO 2 for absorption by [P 66614 ][Benzim] was investigated using a combined feed through a series of absorption and desorption cycles, employing a mass spectrometry-based gas absorption rig. 26 Realistic flue gas concentrations of 14% CO 2 and 0.2% NO 2 were selected, with the results shown in Figure 1 (and Table  S1). It demonstrates that after two cycles in the gas absorption rig, the CO 2 capacity of the IL was unaffected by the presence of NO 2 , with 0.78 nCO 2 :nIL still absorbed. After the third cycle, a clear decrease in CO 2 capacity was observed, with 0.66 nCO 2 :nIL absorbed after exposure to a calculated 0.38 nNO 2 :nIL. The ability of [P 66614 ][Benzim] to capture CO 2 in the presence of NO 2 continued to decrease, reaching 0.31 nCO 2 :nIL after 10 absorption/desorption cycles, an ∼60% reduction in capacity. It was evident that the desorption conditions (2 h under Ar at 90°C) were unable to regenerate the original CO 2 capacity. This strong irreversible absorption of NO 2 is in contrast to the behavior observed with the NO cofeed, where 0.72 nCO 2 :nIL was still absorbed after 10 cycles. 27 Following the 10 competitive absorption/desorption cycles with CO 2 and NO 2 , the treated IL was characterized using a number of techniques. Elemental analysis showed an increase in the nitrogen content of [P 66614 ][Benzim] from 6.10 to 7.67 wt % and an increase in the viscosity from 1087 to 1516 mPa·s (at 25°C). These changes were assigned to the irreversible incorporation of nitrogen into the IL through the absorption of NO 2 . Changes in the physical properties of the IL after exposure to NO 2 were more significant than those observed after exposure to NO (nitrogen content increased from 6.10 to 6.48% and viscosity from 1087 to 1235 mPa·s), correlating with the reduced effect of the NO impurity (0.72 nCO 2 :nIL  was still absorbed after 10 cycles under NO and CO 2 ). 27 As the irreversible absorption of NO 2 causes changes in the physical properties of the IL, NMR and XPS of the IL after exposure were performed to characterize the nature of the strongly bound NO 2 species. The structure was initially probed by 1 H NMR ( Figure S2), where it was expected that changes in the spectra upon the absorption of NO 2 would be caused by a change in the environment of the protons in the heterocyclic anion, which was demonstrated by a downfield shift in the peaks at 6.34/ 6.91/7.36 ppm (a) to 6.57/7.10/7.65 ppm (b) after the IL was treated with 1% NO 2 for 24 h, similar to shifts observed with SO 2 and NO. 26,27 The IL from the absorption rig after the final cycle (c) displayed a similar shift, showing that the same absorbed species was formed under different conditions. In both spectra after NO 2 exposure, new peaks were observed above 13 ppm, suggesting the formation of HNO 3 from the reaction of NO 2 with residual water in the IL. The formation of HNO 3 was further confirmed by IR spectroscopy of [P 66614 ]-[Benzim] after 24 h exposure to 1% NO 2 ( Figure S4). The 13 C NMR spectra ( Figure S3) showed analogous changes, where small shifts were noted for the peaks attributed to the benzimidazolide anion (114−148 ppm) due to changes in aromaticity caused by NO 2 absorption.
Ex situ XPS was further used to characterize the IL postexposure in the absorption rig. A comparison of the N 1s region showed the presence of two new N 1s photoelectron peaks at 402.0 and 406.1 eV assigned to N−O species ( Figure  S5). The broad peak at 402.0 eV is assigned to absorbed NO 2 , with the peak at a higher binding energy of 406.1 eV indicating a more oxidized N species, which can be attributed to N 2 O 4 absorption. 39,40 Absorption of NO in the same IL resulted in a single photoelectron peak observed at 402.4 eV, which was attributed to N 2 O 2 formation. 27,41 These species are expected to be strongly absorbed to the IL as the high vacuum in the XPS analyzing chamber would negate the detection of any weakly absorbed gases.
ATR-IR spectra of [P 66614 ][Benzim] under separate CO 2 or NO feeds have been reported previously, showing characteristic bands for the absorbed species, with additional changes in the spectra observed due to changes in the aromaticity of the [Benzim] − anion. 27 For CO 2 , reversible absorption and an equimolar capacity for absorption presented a simpler system than that for multi-site absorption, which has been reported here for NO 2 and NO. 27 (Tables S2 and S3). Although the absorption of N 2 O 4 is energetically very favorable, the absorbed geometry together with a detailed comparison of grouped Mulliken charges (Tables S4 and S5) indicates that it is a purely physical process without charge transfer from the anion, in comparison to that of chemically absorbed CO 2 . Instead, there are extremely strong dipole interactions between the two δ + nitrogen centers of the absorbate and the formal negative charge on the anion. Furthermore, the physical absorption of two N 2 O 4 dimers at both N-sites of the [Benzim] − anion was energetically favorable, with an absorption enthalpy of −210.5 kJ·mol −1 (Table 1c). The multi-site absorption and greater than equimolar absorption of NO 2 to the IL correlates with the large gravimetric uptake capacity (4.60 nNO 2 :nIL), and the high absorption enthalpies relate to the drop in CO 2 capacity observed. These findings are also consistent with the experimentally observed results, where only a small decrease in absorbed NO 2 was observed after desorption, probably due to the loss of weaker, physically bound, NO 2 .
ATR-IR spectra recorded during the absorption of 0.2% NO 2 in Ar by [P 66614 ][Benzim] are shown in Figure 2. It is evident that upon the introduction of NO 2 , a series of ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Research Article overlapping bands in the 1300−1400 cm −1 region increased in intensity. Interestingly, upon initial exposure to the feed, a band at 1335 cm −1 was the most intense in this region, but with increasing exposure time, a band at 1315 cm −1 dominated the spectra. In addition, bands at 1620, 1233, and 1036 cm −1 increased in intensity with exposure to the NO 2 feed. These changing bands could indicate a change in the nature of the absorbed species or absorption of a NO 2 species at another Nsite on the [Benzim] − anion, as was observed with NO. 27 Calculated vibrational spectra for one or two moles of physisorbed N 2 O 4 to N-sites on the [Benzim] − anion showed indistinguishable spectral profiles, with theoretically derived bands at 1426, 1308, and 847 cm −1 for one mole of physically absorbed N 2 O 4 ( Table 2) Table 2). Bands at 1764/68 and 1825/24 cm −1 were predicted for the v asym (O−N−O) vibration when one/two moles of N 2 O 4 were physically absorbed, but these bands were not observed experimentally, as described by the other authors. 42−44 While physisorbed N 2 O 4 was observed initially (eq 1), the spectra showed changes with time on stream, which were not simply an increase in the intensity of the bands assigned to N 2 O 4 , thus suggesting the evolution of the absorbed species with an increasing NO 2 concentration. The N−N bond of absorbed N 2 O 4 has been reported to undergo heterolytic cleavage, forming [NO 3 ] − and [NO] + ionic species due to perturbation by an external charged species (eq 2 and 3). 45,46 The energy diagrams in Table 1 show the most favored reaction pathway following the physisorption of one and two moles of N 2 O 4 . In these reaction schemes, the heterolytic cleavage of the N−N bond of absorbed N 2 O 4 is favorable (b), but the activation barrier for the disproportionation reaction to NO + /NO 3 − was above 100 kJ mol −1 and was considered too high ( Figure S6) (Table 2b). The neutralization of the [Benzim] − anion removes the active site for absorption, causing deactivation of the anion/IL, and provides evidence for the decreased CO 2 capacity. The [NO 2 ] − ion stabilizes the positively charged [P 3333 ] + cation by orienting the negatively charged oxygen atoms into a position to counter the positively charged phosphorus (Table 2c). It could be proposed that the [NO 2 ] − ion reacts with, or deprotonates, the acidic α-protons close to the phosphonium cation center (P−CH 2 −), but this pathway was not observed (see Table 2c, which shows the lowest energy structure). 47 Additionally, Table 1c,d shows that when two moles of N 2 O 4 are physically bound, which is more likely at increased exposure times, the heterolytic cleavage of the N−N bonds has little driving force due to similar initial and final state energies, as well as a higher barrier (51.9 kJ· mol −1 ). Thus, the formation of the neutral [Benzim-NO 2 ] complex (Table 1d) is expected to be equilibrium-limited, leading to the gradual decrease in the CO 2 capacity of the IL.
To probe the evolution of the bands in Figure 2, the spectrum after 1 min under the NO 2 feed was subtracted from all subsequent spectra (Figure 3 (Table 2). A new band was observed at 1233 cm −1 , attributed to the formation of a [NO 2 ] − anion, and related to vibrations at 1405 and 828 cm −1 . Evidence of a new N−N bond was observed at 1036 cm −1 , from the formation of the neutral [Benzim-NO 2 ] complex, associated with bands at 1620, 1308, and 868 cm −1.48−50 The DFT calculated species correlated well with the observed ATR spectra, and additional features were due to changes in the aromaticity of the [Benzim] − anion, as observed with CO 2 / SO 2 /NO. 26,27 Interestingly, the changes in the anion aromaticity are observed during the physical absorption of N 2 O 4 , showing that the strong initial interaction distorts the charge density of the anion. Combining both the DFT calculations and the FTIR spectroscopic study allowed the determination of the absorbed species and how the speciation changed with time. The DFT results indicate the thermodynamically stable intermediates, while FTIR provides evidence for the proposed intermediates and the kinetics of absorption. Together, the results indicate the strong physical absorption of N 2 O 4 followed by the deactivation of the IL through the formation of [Benzim-NO 2 ]/[P RRRR ][NO 2 ] complexes at increased exposure times, which would be expected to result in a loss of absorption sites for CO 2 and the co-absorption of CO 2 was therefore studied as well.
A co-feed of 14% CO 2 and 0.2% NO 2 in Ar was investigated, simulating conditions in the gas absorption rig (Figure 3ii), to study whether bands assigned to the neutral complex would form and influence the CO 2 absorption capacity of the IL (as observed in the absorption rig results). The chemical absorption of CO 2 results in the formation of a carbamate species, with bands at 1709 (CO) and 1281 cm −1 (N− COO − ) quickly increasing during the first minute of exposure to the feed, which is caused by the greater concentration of CO 2 in the feed. 26 Simultaneously, bands associated with the physical absorption of N 2 O 4 grow (1373 and 1334 cm −1 ) due to the strong absorption enthalpy of the species (−107.8 kJ· mol −1 ) ( ]. This correlates with the stepped reduction in CO 2 capacity observed in the gas absorption rig. Subtracted data in Figure S7 show that bands due to the heterolytic cleavage of physisorbed N 2 O 4 are observed to form with and without the presence of CO 2 . The absorption of CO 2 does not hinder the absorption and subsequent heterolytic cleavage of N 2 O 4 , which, however, deactivates the IL to further absorption of CO 2 . The physical absorption of N 2 O 4 at one site drastically changes the kinetics and thermodynamics of absorption at the remaining site, where the barrier for CO 2 activation is low (6.1 kJ·mol −1 ) (Table 1e,f). Calculations show that the coabsorption of N 2 O 4 and CO 2 to separate N-sites is thermodynamically stable (−147.3 kJ·mol −1 ), aided by the low concentration of NO 2 . After an extended period, both the deactivation of the [Benzim] − anion and the larger absorption enthalpy of the binding of two moles of N 2 O 4 (−210.5 kJ· mol −1 ) suggest that decreases in CO 2 capacity are observed due to the loss of absorption sites. The reaction mechanism between the IL and CO 2 /NO 2 is depicted in Figure S9 These results differ from those when NO is in the co-feed, where the co-absorption of both CO 2 and NO was observed. 27 The results obtained previously with NO were rationalized by the reaction landscape depicted in Figure S10, where the reduction of CO 2 is a barrierless process, and the activation barrier to form the NONO-ate complex is relatively high, showing a preference for the absorption of CO 2 . After the formation of the carbamate, slow deactivation of the IL is observed in the presence of NO via a CO 2 bond intermediate (which enables NO to co-bind). Using in silico techniques, this observation was explained by the relatively high kinetic barriers for the formation of the NO x species bound to the IL from NO, as opposed to NO 2 . Therefore, it is evident that an alternative absorption mechanism occurs, accelerating the effect that NO 2 has on the uptake of CO 2 .
A summary of the effect of the different waste gas stream impurities on [P 66614 ][Benzim] studied here is shown in Figure  4. It is clear that SO 2 has the greatest effect on reducing the CO 2 uptake in ILs followed by NO 2 and finally NO, in clear accord with the calculated absorption energies of −123.9, 27  Spectroscopic results in combination with DFT calculations showed that NO 2 was predominately strongly physically absorbed on multiple sites in the form of N 2 O 4 . This species was then found to undergo heterolytic cleavage resulting in the deactivation of the IL. The ability to investigate the effects of flue gas impurities, individually and competitively, on the uptake of CO 2 , is an important experimental tool in the development of new sorbents and highlights the need for rigorous experimental methods, preferably in tandem with theoretical studies and the investigation of more complex, realistic multi-component feeds. Overall, these results show the importance of investigating the effect of flue gas contaminants. Further studies into the feasibility and optimization of ILs in such a process are still required, but a consideration of impurities and sorbent recyclability is an essential factor. Significant improvements in tuning the selectivity and absorption enthalpies of ILs to reversibly capture NO x or SO 2 at high efficiencies are challenging, and alternative methods of removing acidic impurities will still be required (gas scrubbers and traps). Additionally, tuning the basicity of the IL by selecting anions with different pK a values presents an interesting opportunity for the pretreatment of waste gas feeds, where contaminants such as NO x could be reversibly captured.