Micelle structure of nonionic surfactants containing carbon dioxide moieties in protic ionic liquids

Partial substitution of ethylene oxide with carbon dioxide moieties can yield greener nonionic surfactants with comparable functionalities. In water, studies showed that the incorporation of CO2 moieties suppresses the formation of liquid crystalline phases at high concentrations. A similar reduction in solvation and suppression of liquid crystal formation is observed here in the ionic liquids ethylammonium nitrate and propylammonium nitrate. Small-angle neutron scattering is used to study the solvation and packing of micelles in ionic liquids as functions of temperature, concentration, and content of CO2 moieties. By comparing with aqueous solutions, this work shows that while the nature of surfactant-solvent interaction is comparable among water and alkylammonium nitrate ILs, their behaviours in the solvated micelle shell are different. The lack of liquid crystalline phases should be attributed to the small excluded volume of micelles, which can be fine-tuned via ion design and choice of solvent.


Introduction
Over the past two decades, ionic liquids (ILs) have attracted interest as promising candidates for next-generation solvents. They exhibit negligible vapour pressure and tuneable solvent properties [1][2][3]. More than 10 6 different ion pairs have been postulated, greatly expanding on the hundreds of organic solvents currently being used in industry and research. Among these, some ILs were found to be inhomogeneous in the bulk [4]. The delicate balance of directional intermolecular interactions (e.g. hydrogen bonding), electrostatic forces, and Van der Waals forces induces molecular segregation resulting in bicontinuous, sponge-like polar/ apolar networks in pure ILs [5][6][7]. This is exemplified by alkylammonium nitrate ILs, among which ethylammonium nitrate (EAN) is the most extensively studied [8]. A series of structural studies have shown the domain size and correlation length can be controlled through variation of the alkyl length of the cation [1,8]. The formation of this nanostructure enables them to dissolve a diverse range of polar and apolar solutes, and the presence of this rich pattern of intermolecular forces has made them excellent media for self-assembly [5,[9][10][11][12].
Polyoxyethylene n-alkyl ethers (C n E m ) are the most widely used nonionic surfactants present in almost all cleaning products today. They are often added as polydisperse mixtures (a distribution in headgroup and tail length) to engineer desirable phase behaviour and regulate the viscoelastic properties of the solution. Although among the most widely used surfactants, the ethylene oxide-based headgroups are derived from petrochemicals. This motivated a search for greener replacement, and 1 3 carbon dioxide stood out as a promising candidate. Recent advances in catalytic processes have enabled large-scale production of nonionic surfactants with CO 2 groups partly substituting ethylene oxide ( Fig. 1) [27,28]. This can be achieved by simply copolymerising ethylene oxide and CO 2 with a suitable catalyst onto the OH-group of a chosen alcohol. These surfactants with CO 2 -derived ester groups (CO 2 moieties) can greatly reduce petrochemical consumption and deliver comparable functionalities to conventional nonionic surfactants in aqueous systems [29][30][31][32]. More strikingly, the inclusion of even a single CO 2 moiety per surfactant is reported to completely suppress the formation of gel-like liquid crystalline phases in water at high surfactant concentration [29]. This is attributed to reduced headgroup hydration and increasing micelle softness. Ordered packing of these soft micelles is discouraged even at high concentrations, allowing solutions to be much more easily stored and handled, which is of high importance for practical applications.
Previous studies of conventional C n E m surfactants in ethylammonium nitrate (EAN) have shown that they can support a wide range of amphiphilic self-assembly behaviour into micelles, lyotropic liquid crystals, and microemulsions [13,14,33,34]. Systematic trends in micelle morphology and phase behaviour are observed parallel to their aqueous analogues [14]. In this study, we have investigated micelle formation by nonionic surfactants containing CO 2 moieties in two ionic liquids, EAN and propylammonium nitrate (PAN). We used small-angle neutron scattering (SANS) to elucidate how this modification of the headgroup influences its interaction with IL solvents and thereby modulates their aggregation properties. Dodecyl and hexadecyl surfactants with varying CO 2 moiety contents are studied at multiple temperatures and compositions, allowing us to also assess the effect of the alkyl chain length on the aggregation behaviour. The aim of this investigation was to provide insight into the origin of solvophobicity displayed in ILs [35], the interaction between ILs and the modified surfactant headgroups, and to compare this behaviour to that previously observed in water. This will enable rational design of phase behaviours via choice of solvents.

Materials
Octaethylene glycol monohexadecyl ether (C 16 E 8 ) was purchased from Nikkol and used as received. Other surfactants containing CO 2 moieties (Table 1) were developed and produced by Covestro via copolymerisation of ethylene oxide with CO 2 applying an alkyl alcohol as the initiator and a bimetallic catalyst. Electrospray mass spectrometry was employed to verify their alkyl length and headgroup size (see SI). All samples were stored in a freezer at −20 ℃, and the average molecular mass is given in Table 1.
Partially deuterated d 3 -ethylammonium nitrate (EAN) and d 3 -propylammonium nitrate (PAN) were prepared by neutralisation of nitric acid with the corresponding alkylamine, followed by deuteration of exchangeable hydrogens. Ethylamine solution (66-72%, Sigma), propylamine (99%, Sigma), and nitric acid (70%, Ajax) were used as received. Amine solutions were diluted in water to ~20 wt% before being added dropwise to the diluted nitric acid solution (~20 wt%). The mixture was under constant stirring and kept in an ice bath. Temperature was maintained below 10 ℃ and the titration is conducted until pH indicated the reaction mixture was in slight excess of amine. Bulk water was removed using a rotary evaporator (30 mbar, 40 ℃). Deuteration of exchangeable protons was performed by adding 3:1 mol/mol D 2 O to ILs. Bulk water was removed by rotary evaporation, and exchange was repeated 5 times per IL sample. Trace water was removed under a high vacuum over > 48 h (0.2 mbar, room temperature). Final water content is < 0.2 wt% (Karl Fischer titration), and 300 MHz 1 H NMR confirmed that all exchangeable protons had been replaced by deuterium.

Methods
Diffusive interfacial transport experiments were performed to confirm the absence of liquid crystalline phases [36,37]. A Leica DM 2500p microscope fitted with a polariser was used. Here, a drop of molten dry surfactant is added to a  clean glass slide and covered by a thin coverslip. A drop of solvent is added to the corner of the coverslip. Due to the capillary effect, the solvent is drawn into the gap between the coverslip and glass slide. As the solvent meets the surfactant, a continuous concentration gradient is created between pure solvent and pure surfactant. This enables all phases that exist at any composition at the specified temperature to be generated. Crossed polarisers are used to help identify liquid crystalline phases that display birefringence. Images were captured using a QImaging MicroPublisher 3.3 RTV camera with 200 times magnification. Small-angle neutron scattering (SANS) measurements were performed on the Quokka beamline at the Australian Centre for Neutron Scattering with hydrogenous surfactants in partially deuterated ILs (single contrast). Neutrons with average wavelengths of 5.0 and 6.0 Å and a ∆λ/λ up to 15% were selected with a velocity selector. Scattering was collected from 1.0-mm path length samples at two sample-todetector distances of 4 and 14 m, giving a combined range of 0.004 < q < 0.4 Å −1 . Acquisition was performed at 308, 343, and 378 K. Isotropic scattering patterns collected on a 2D detector were reduced to I(q) using modified NIST routines in Igor™.
Model fits to the SANS patterns are shown and described in the SI. In order to facilitate comparison with these systems in water, we have followed the analysis described previously. Forward scattering intensities, I(0), were used to extract the structure factor at the thermodynamic limit (S(0)) in the monodisperse approximation (Eq. 1) [29].
where dry is the dry volume fraction of surfactant calculated from theoretical aggregation number and it is in good agreement with published density of conventional nonionic surfactants (see SI) [38], Δ is the difference in scattering length densities of dry micelles and solvent, and V is the volume of the dry micelle. For the model of hard-spheres, S(0) is accurately described by the Carnahan-Starling model (Eq. 2) [39].
This model was then used to determine the effective, concentration-dependent, hard-sphere volume fraction ( hs ) of the dispersed micelles according to Eq. (2). The relationship between measured hs and the actual volume fraction of surfactant added, dry , is modelled by Eq. (3) [29].
Both B and A are parameters that describe two different concentration-dependent effects regarding the effective volume of the micelles. B accounts for the solvation of dry micelles which increases their effective volume fraction, i.e. the volume ratio of solvated to dry micelles (which must be > 1). The parameter A accounts for the interpenetration of neighbouring micelles which occurs with increasing micelle concentration and reduces the effective volume fraction [29].
A is a second-order term and therefore negligible at low concentrations, but being quadratic in the volume fraction becomes increasingly dominant at higher concentrations. In molecular terms, A dry can be interpreted as the fraction of the initial hard-sphere volume (present at infinite dilution, B dry ) that becomes penetrable at a finite concentration. solid-liquid interface of C16E8C3 with EAN displayed a smooth boundary from which any trapped bubbles could easily diffuse away indicating that no cubic phase occurs in C16E8C3-EAN mixtures at any compositions. This is consistent with existing literature in water, where the incorporation of CO 2 moieties can completely suppress the formation of liquid crystalline phases [29]. To understand the origin of this behaviour and the interaction between micelles, SANS experiments were performed. Figure 3 shows representative SANS patterns of selected surfactants in EAN at 308 K, above the Krafft temperature of the hexadecyl surfactants. Figure 3A shows the results found for all surfactants at 5 wt% and 308 K (complete dataset of all compositions and temperatures are shown in Figs. S2−S5), which is consistent with the formation of spherical core-shell micelles above a critical micelle concentration (CMC). Figure 3B shows that with increasing concentration, scattered intensity increases, and an interaction peak (structure factor) develops above 5 wt%. The upturn in intensity at q < 0.01 Å −1 , which is most apparent in C12E8C3, due to the low scattering intensity from the sample, is likely due to microbubbles trapped within the sample cell [14].

Results and discussion
The scattering patterns of C16E9C2 and C16E8C3 at 5 wt% (Fig. 3A) are almost identical, suggesting that the formation of micelles is insensitive to small differences in the extent of CO 2 incorporation. However, both C16 surfactants containing CO 2 moieties show a higher scattering intensity, and the intensity upturn occurs at a lower q than that of C16E8 (near 0.1 Å −1 ), both indicating that surfactants with CO 2 moieties form larger micelles. We also noticed that for the most highly concentrated samples, the correlation peak is much sharper for C16E8 than C16 surfactants containing CO 2 moieties (e.g. Fig. S2). This indicates that the repulsions between micelles at the same concentration are stronger in C16E8 than surfactant containing CO 2 moieties, similar to their behaviour in water [31].
The lower overall scattering intensity of the C12E8C3 solution is due to its much higher CMC. This is clear from the absence of C12E8C3 micelles in EAN below 2 wt% (Fig. S2), and from invariant analysis (see Fig. S7), which yields the CMC values listed in Table 2. Similar dependence of CMC on alkyl chain length is reported in water and EAN [14].
CMCs are much higher in PAN than EAN, which is higher than in water (e.g. C12E8C3 is 0.0036 wt%, C16E x C y CMCs in water are not available due to poor solubility) [29]. C12E8C3 can no longer form well-defined micelles up to 10 wt%, indicating that the minimum alkyl chain length required to stabilise micelles is greater in PAN than in EAN, as also seen in conventional polyoxyethylene surfactants [34]. Similarly, no micelles were observed by SANS in C16 surfactants up to 2 wt% in PAN, and 5 wt% solutions yielded lower-than-expected total coherent scattering and volume fraction, both consistent with elevated CMC.
Fitting of the SANS data to a core-shell sphere form factor with hard-sphere structure factor did not yield a unique and consistent set of physically meaningful fits [14,31]. This is likely due to interpenetration of the headgroups, as was found for these systems in aqueous solution (see SI for details).
To further explore the intermicellar interactions and facilitate comparison with aqueous solutions, we instead analysed the forward scattering, I(0), in terms of effective volume  fraction for a hard-sphere interaction as outlined in the "Methods" Section [29]. All scattering curves exhibit plateaus at q < 0.02 Å −1 (see Fig. 3 and Figs. S2−S6), enabling the thermodynamic limit, I(0), to be determined. I(0) first increases with increasing concentration as more micelles are present, and the total scattering volume increases and then decreases as the packing of micelles becomes increasingly ordered. This structure factor effect leads to the development of a peak near 0.05 Å −1 , which becomes more pronounced and shifts to higher q with increasing surfactant concentration as the average separation between micelles decreases. Fits to Eq. (3) and the derived fit parameters are shown in Fig. 4 and Table 3, respectively, assuming that the micelle aggregation number is independent of concentration [14]. Good fits are obtained overall, especially at high surfactant concentrations. The dry micelle volume fraction for C12E8C3 has been corrected for its CMC of 2.5 wt% but neglected for C16 surfactant which may explain the slight discrepancies observed for the lowest concentrations (see SI). This is most pronounced for C16E8, which has a higher CMC than the other two C16 surfactants as shown by the relatively low scattering intensity in Fig. 3A. Nevertheless, this has negligible impact to the fit over the entire concentration range, and the subtle differences in CMCs among C16 surfactants are negligible within the uncertainties of the invariant analysis (Table 2). Table 3 shows that all four surfactant systems yielded a solvation value, B, greater than two in EAN. This means that isolated micelles in dilute solution occupy more than twice their dry volume due to solvation, which are significantly larger than the corresponding values (for C12 surfactants with same headgroups) in water. This means the headgroup layer of the micelles is much more swollen and the headgroups themselves protrude further into their surroundings in EAN. The solvation numbers of each surfactant are smaller, however, as the volume of each EAN ion pair is much larger than a single water molecule [29].
The hydration of C 12 E 14 of 28.8 corresponds well to approximately two water molecules per ether oxygen hydrogen bonding acceptor [41]. In EAN, the ethylammonium cation can donate into three H-bonds, from which the predicted IL/surfactant ratio should be around 0.67 that of water. This is consistent with our results of 17.2 water molecules versus 12.8 EAN per C12E8C3 [29] and suggests that the solvophilic solvent-headgroup interaction of nonionic surfactants with alkylammonium ILs is also governed by hydrogen bonding.
Increasing alkyl length to C16E8C3 reduces the EAN/ surfactant ratio. This is likely due to the formation of larger micelles reducing the curvature of the headgroup shell, so that less volume is available per surfactant to be occupied by solvent species. Despite having the smallest headgroup, C16E8 has the largest solvation in EAN, confirming that the incorporation of CO 2 moieties reduces headgroup solvation in EAN, just as it does in aqueous systems [29].
The main difference to the aqueous system is that the softness parameter A does not increase with increasing content of CO 2 moieties but slightly decreases. This means the pure ethylene oxide headgroup shows a high degree of interpenetration in EAN. This difference may be attributed to the large volume of micelles in EAN (i.e. the B parameter in EAN is substantially bigger than in water). More swollen micelles means the micellar shell is more prone to penetration. Furthermore, water is a stronger H-bond donor relative to the less polar N-H bonds in ethylammonium cations and therefore is more tightly H-bonded to the ethylene oxide groups. In a similar direction, the difference between B and A is 1.7 and 1.3 (Table 3) in water and EAN, respectively [14,29], which means that the relative effect of the interpenetration  of shells (softness) is more pronounced in EAN compared to water. The solvophobic interaction that gives rise to amphiphilic nanostructure in EAN itself has been shown to also cause ethylammonium ions to associate with surfactant aggregates [15]. The IL cation can preferentially orient at apolar interfaces and exhibit co-surfactant-like behaviour by embedding its alkyl tail into the micelle core. These amphiphilic IL cations may slightly swell the micelle cores, also contributing to softer micelles overall.
With increasing temperature, the SANS patterns of all surfactants in both EAN and PAN exhibit less sharp correlation peaks, the peaks are shifted to lower q, and the overall scattering intensity increases. These changes indicate a growth of the micelles with rising temperature and possible increase in polydispersity. SANS patterns of surfactants containing CO 2 moieties in ILs at elevated temperature can no longer be modelled by interacting spheres, requiring the core to become ellipsoidal (see SI) [32]. This transition is also consistent with past studies of polyoxyethylene surfactants in water and EAN. In both solvents, micelle growth occurs as the temperature approaches their cloud point [14,[42][43][44] and confirms that the thermal response of these surfactants is comparable to conventional nonionic surfactants. This is attributed to the reduced solvation with increasing temperature, thereby reducing the effective headgroup area and surfactant packing. The scattering patterns of the PAN samples (Fig. S6) show that this increase of scattering intensity with temperature is more pronounced for pure C16E8 than surfactants containing CO 2 moieties, where most likely an increasing CMC leads to demicellisation.
In water, EAN, and PAN, the minimum alkyl chain length for conventional nonionic surfactants to form well-defined micelles at low concentrations (i.e. low CMC) is C10, C14, and C16, respectively [14,34,41]. Our results confirm that a similar trend applies to nonionic surfactants with CO 2 moieties, indicating that the driving force of self-assembly is largely unaffected by headgroup modification [40]. From the CMC values, we can conclude that the free energy of micellisation, ΔG°m ic , of hexadecyl surfactants in PAN is lower than in EAN by about a kT, which is itself 6 kT smaller than in water.

Conclusion
The incorporation of CO 2 moieties in the headgroup of nonionic surfactants has a negligible effect on the tendency to form micelles. The solvophobic driving force for self-assembly is governed by the alkyl chain length of surfactant and the average polarity and nanostructure of the solvent. However, the presence of CO 2 -derived ester groups in the headgroup can modulate the solvent-headgroup interaction. Higher content of CO 2 moieties is observed to suppress solvent-headgroup hydrogen bonding which reduces solvation in ILs as it does in water.
While the underlying molecular interaction is similar, this has led to different trends in micelle softness in water and EAN. In water, the micelle softness increases (or the extent of interpenetration among nearest micelles, captured by parameter A) with content of CO 2 moieties. This is attributed to the reduced hydration of the micellar headgroup [29]. In contrast, micelles in EAN are inherently more swollen due to the larger EAN molar volume (relative to water), and micelles become less penetrable with increasing content of CO 2 moieties, opposite to water. Alkylammonium IL cations can act as a co-surfactant and undergo dynamic exchange between the solvation and bulk state [15,[45][46][47]. As a result, the relatively large volume of ILs in micellar shell can become easily displaced by chain-chain interpenetration among neighbouring micelles. This allows well-solvated micelles to be softer than poorly solvated ones in ILs.
Overall, the effect of incorporation of CO 2 moieties into nonionic surfactant headgroups has the same effect of decreasing solvation in ILs and in water. The co-surfactant behaviour of ILs has induced a different trend in the ability to interpenetrate among neighbouring micelles as the content of CO 2 moieties increases. Nevertheless, the excluded volume per micelle in ILs for surfactants containing CO 2 moieties is comparable to, if not lower than, that of water. This suppresses the formation of liquid crystalline phases, allowing concentrated solutions to be easily handled as low-viscosity fluids. Differences in H-bonding and amphiphilic character of water and EAN highlight that solvent can be used as a tool to fine-tune micelle properties and inter-micelle interactions.

Data Availability
The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Should any raw data files be needed in another format they are available from the corresponding author upon reasonable request.

Conflict of interest
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.