Article pubs.acs.org/JPCB
Ammonium Ionic Liquid Solubilities in Water and Micellar Formation Rachaud Keyes and Paul Scovazzo* Department of Chemical Engineering, University of Mississippi, 134 Anderson Hall, University, Mississippi 38677, United States ABSTRACT: A number of proposed applications for ionic liquids (ILs) involve IL/water interfaces, such as chemical separations or drug delivery. Therefore, an understanding of the solubility and micellar behavior ILs in aqueous environments is critical. The anion, bis(trifluoromethanesulfonyl)imide (Tf2N) promotes water stability and forms water immiscible ILs. This study, therefore, paired the Tf2N-anion with three different classes of IL cations. The three classes examined were 1-alkyl-3-methylimidazoliums (Rmim), alkyl-trimethylammoniums (CTA), and bulky ammoniums (BAM). CTAs can be synthesized from inexpensive ammonium surfactants; however, large CTAs are solids at ambient conditions. In contrast, large BAMs remain in the liquid state at ambient conditions. We used total organic carbon (TOC) analysis to determine the IL content in IL saturated water. Surface tension measurements of IL containing water determined if micelles existed in the IL saturated water. We used linear free energy relationship (LFER) semiempirical models to correlate the IL water solubility to the molecular volume and IL cation structure. The reported LFERs can predict the IL solubility in water before the IL is synthesized. Combining the LFER results with surface tension measurements and thermodynamic calculations allowed us to determine that micelle formation is not significant for the tested ILs with molecular weights ≤510.
■
INTRODUCTION Literature contains a number of proposed processes involving interfaces between room temperature ionic liquids (ILs) and aqueous phases. Consequently, the aqueous solubilities of ILs are design parameters for processes such as, chemical separations,1 pharmaceutical separations,2,3 drug delivery,4 and water treatment.5 In addition, ecotoxicity and environmental transport of ILs are related to their aqueous solubilities.6−8 Therefore, an understanding of the solubility and the potential for micellar behavior of ionic liquids (ILs) in water is crucial. Unfortunately, published information on the liquid/liquid behavior between ILs and water is limited or even absent for some classes of ILs.9 Ammonium-based ILs are an economical class of ILs10 with limited published information on their solubility in water.11 Ammonium-ILs potential uses included antibacterial agents, antifungal agents, drug delivery, solvents for drugs, pharmaceutical separations, and other processes.2,3 In contrast to all of the potential uses for ammonium-ILs, of the 12 quaternary ammonium salts that are the topic of this article, we only found three with aqueous phase solubilities previously reported in the literature. We selected a set of ammonium-ILs for testing whose cation hydrophobic regions ranged both in the size and types of alkyl groups. The ILs tested covered a range of alkyl chain lengths, from butyl to hexadecyl. Some of the cations contained branched (isopropyl) alkyl groups. Varying the range and type of the hydrophobic region potentially alters the partitioning of a third species between the IL and aqueous phases; a potential design criteria for future studies related to separations or drug delivery. © XXXX American Chemical Society
All of the cations used bis(trifluoromethanesulfonyl)imide (Tf2N) as their counteranion. Tf2N is a water stable anion that forms water immiscible ILs. Tf2N-based ILs are attractive for organic extraction from aqueous phases due to their relatively low viscosity and high immiscibility with water.12 Figure 1 illustrates the three classes of ILs studied in this article using butyl as the representative alkyl group on the cation. The 1-alkyl-3-methylimidazolium-based ILs (Rmim) were selected since imidazolium-ILs already have aqueous solubilities reported in the literature from different research laboratories using different testing techniques. The inclusion of
Figure 1. IL class groupings showing class names (Rmim, CTA, and BAM) along with sketches of the cations that define the classes. Cation structural based abbreviations given: [C(R)mim], [N(R)111], [N(R)11(i-3)], and [N(y)RRR]. R and y are the number of carbons in the nalkyl groups. Received: May 25, 2017 Revised: June 30, 2017 Published: July 5, 2017 A
DOI: 10.1021/acs.jpcb.7b05109 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Table 1. ILs Used in This Study with Their Abbreviations Used in the Text, Molecular Weights (MW), Liquid Molar Volumes, and Group Contribution Calculated Molecular Volumes, Vix IL abbreviation
IL chemical name
CAS #
MW
liq. molar vol. (mL/mol)a
Vix (mL/mol)f
[C(4)mim] [Tf2N] [C(6)mim] [Tf2N] [C(10)mim] [Tf2N] [N(4)111] [Tf2N] [N(6)111] [Tf2N] [N(8)111] [Tf2N] [N(10)111] [Tf2N] [N(12)111] [Tf2N] [N(14)111] [Tf2N] [N(16)111] [Tf2N] [N(4)11(i-3)] [Tf2N]
1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide 1-decyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide butyltrimethylammonium bis(trifluoromethylsulfonyl)imide hexyltrimethylammonium bis(trifluoromethylsulfonyl)imide octyltrimethylammonium bis(trifluoromethylsulfonyl)imide decyltrimethylammonium bis(trifluoromethylsulfonyl)imide dodecyltrimethylammonium bis(trifluoromethylsulfonyl)imide tetradecyltrimethylammonium bis(trifluoromethylsulfonyl)imide hexadecyltrimethylammonium bis(trifluoromethylsulfonyl)imide dimethyl(butyl(i-propyl))ammonium bis((trifluoromethyl)sulfonyl) imide dimethyl(hexyl)(i-propyl))ammonium bis((trifluoromethyl)sulfonyl) imide dimethyl(decyl(i-propyl))ammonium bis((trifluoromethyl)sulfonyl) imide tributyl(methyl)ammonium bis(trifluoromethylsulfonyl)imide triethyl(hexyl)ammonium bis(trifluoromethylsulfonyl)imide
174899-83-3 382150-50-7 433337-23-6 258273-75-5 210230-43-6
419.4 447.4 503.5 396.4 424.4 452.5 480.5 508.6 536.6 564.7 424.4
294.7b 332.5b 393.7c 289.6d 324.5d 356e (20 °C) 393.2d solid solid solid 315.4d
239.90 268.08 324.44 235.29 263.47 291.65 319.83 348.01 376.19 404.37 263.47
452.5
353.1d
291.65
508.6
424.6
d
348.01
480.5 466.5
383.5d 365.8d
319.83 305.74
[N(6)11(i-3)] [Tf2N] [N(10)11(i-3)] [Tf2N] [N(1)444] [Tf2N] [N(6)222] [Tf2N]
405514-94-5
For references 10 and 25, uncertainties in density measurements did not exceed ±0.0005 g/mL, giving molar vol. uncertainties of ±0.2 mL/mol. Reference 25. cReference 26. dReference 10. eReference 27. fThis study. Calculated following the example in the Correlation of Aqueous Solubility Section, eqs 1 and 2. a b
■
the Rmim-ILs allowed comparison of our solubility measurement technique with those in the literature. The alkyltrimethylammonium-ILs (CTA) are relatively inexpensive (compared to Rmim) since they can be made from standard surfactants.10 Bulky ammoniums (BAM), as a class, are liquids at higher molecular weights than CTA. The branch or multiple long-chain alkyl groups in the BAMs may be of use in designing ILs with a desired IL/water partition coefficient of a third component. Literature contains a wide range of methods for measuring the concentration of ILs in water. None of the methods are universally applicable and each have their drawbacks. For aromatic ILs, such as imidazoliums, UV−vis spectroscopy is a proven technique; however, it cannot be used for low IL concentrations nor for ammonium-based ILs.13 The visible method mixes water with the IL until the IL visibly disappears. The production of precise results using the visible method, therefore, is tedious.14 Karl Fischer titrations determine the water content of a mixture. Then the IL concentration results from subtracting the water content from the whole, which may lead to significant errors.13 Gravimetric methods may, also, produce large errors for samples with low IL content.15 Ionselective electrodes require ion-specific membranes; while thermogravimetry and mass spectroscopy are time-consuming and not suitable for routine use.13 We chose to use TOC analysis for determining the IL concentrations in water because of the drawbacks of the other methods listed above. Previously reported methods for determining ammonium-based IL solubilities in water are visible,8 evaporation method,14 and conductivity methods.11,14 The evaporation method used the weight of the aqueous sample residue after water evaporation; a 16 h plus technique.14 Literature contains two different conductivity methods; one involving sample dilution11 and one that measured the saturated concentration without dilution.14
THEORY
Aqueous Solubility and Micelle Formation. ILs in aqueous solutions are, in general, under infinite-dilution-like conditions7 because of their low molar concentrations. For example, Rmim[Tf2N] solubilities in water range from 10−3 to 10−5 mol/mol.16 The IL-anion has the most significant impact on determining the IL solubility in water17 with the Tf2N having some of the lowest solubility values. Because IL solubilities in water are entropically driven,11 relative solubilities between ILs with the same anion mainly depends on the ILmolar volumes.7,11,15,16 Lastly, the molecular structure of the cation can fine-tune the solubility via alkyl chain length,16,18 alkyl chain branching,7 cation symmetry/asymmetry,16 and aromatic vs nonaromatic structures.7 For example, both aromatic and nonaromatic (imidazoliums, pyridinium, pyrrolidinium, and piperidinium) ILs with alkyl chain branching have slightly higher solubility in water compared to their linear isomers.7 This difference in solubility between branched and linear isomers is difficult for thermodynamic property models like COSMOS-RS to capture.7 Molecular simulations of [Tf2N]-based ILs under infinitedilution-like conditions show that the anion and cation “...form ion pairs more often, and for longer times than... if they encountered each other on a purely random basis. This suggests the tendency of these ions to form a second IL-rich liquid phase as soon as their concentration is raised above their (very low) solubility limit in water.”7 Therefore, the potential exists that the ILs in our study could be forming aggregations or even micelles in the aqueous phase. This would be consistent with the behavior of the IL 1-decyl-3-methylimidazolium tetrachloroferrate, [C(10)mim][FeCl4], a known surfactant that forms micelles at aqueous concentrations greater than 40.6 mmol/L.19 We explored this micelle formation potential for the ammonium-IL with the longest alkyl chain. For the IL that we determined formed micelles under test conditions, we measured and reported the critical micelle concentration (CMC), which is the concentration near the solubility limit B
DOI: 10.1021/acs.jpcb.7b05109 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Table 2. Characteristic Atomic and Bond Volume for Group Contribution Calculation of Molecular Volumes in mL/mol20 group
mL/mol
group
mL/mol
group
mL/mol
group
mL/mol
bond carbon hydrogen
−6.56 16.35 8.71
oxygen nitrogen phosphorus
12.43 14.39 24.87
fluorine chlorine bromide
10.48 20.95 26.21
iodine sulfur silicon
34.53 22.91 26.83
in water that results in the formation of the second IL-rich liquid phase predicted by molecular simulations.7 Correlation of Aqueous Solubility. Solubility of ILs in water is a logarithmic function of molar volume.11,16 Logarithmic functionalities with a solute’s liquid molar volume are examples of linear free energy relationships (LFERs). When applied to correlating aqueous solubilities, LFERs are semiempirical models for a pure liquid phase contacting and saturating the water. LFERs correlate the aqueous solubility to the solute’s (IL’s) molar volume and structure. Structure is critical; as illustrated by how COSMO-RS incorrectly describes the differences in water solubility of linear vs branched butyl cation chains.7 The LFER correlation with structure results from grouping the ILs into structural classes so that the IL activity coefficients in the water, within the class, varies only with increasing apolar substituents (−CH2−) added to the cation tracked by the IL molar volumes. Another critical assumption for applying LFER to ILs is that the pure solute phase is liquid or a subcooled liquid for ILs with melting points >25 °C. This means that the enthalpy of solution includes only the excess enthalpies and no enthalpy of phase change. LFERs have been previously applied to ILs in the literature; some applications more rigorously adhered to the underlying necessary assumptions16 than others.11,15 However, these applications still reported good correlations supporting the application of LFER to IL solubilities in water. The solute molar volume used as the independent variable in a LFER can be either the intrinsic molar volume (the molecular weight divided by the experimentally determined liquid density) or a group contribution determined molecular volume. One advantage of the group contribution method is the ability to calculate molecular volumes from chemical structures before synthesis of new ILs. The use of group contribution methods has produced LFERs with strong correlations (R2 = 0.99) for other aqueous solutions of organics, such as n-alkanes, alkanols (primary, secondary, and tertiary), chlorinated benzenes, etc.20 For these two reasons, we used a group contribution method to determine the molecular volumes from adding up the contribution to the molecular volume (reported in molar volume units) from each atomic element and bond in the structure. Table 1 contains both the intrinsic molar volumes and the group contribution molecular volumes, Vix. The group contribution method we used is the one from Abraham and McGowan, 1987; taken from the treatment in Chapter 5 of Environmental Organic Chemistry.20 In this method, all bonds (single, double, triple) are considered equal. The molecular volume (Vix) results from adding up the characteristic atomic and bond volumes given in Table 2. For example, [Tf2N] has two carbons, six fluorides, four oxygens, two sulfurs, one nitrogen, and 14 bonds. Using the values in Table 2 the Vix for the [Tf2N]-anion is
Now the cation in [N(4)111][Tf2N] has seven carbons, one nitrogen, 18 hydrogens, and 25 bonds leading to a Vix for the [N(4)111]-cation of Vix = 7(16.35) + 1(14.39) + 18(8.71) + 25( −6.56) = 121.6 mL/mol
(2)
Adding the group contributions of the cation and anion gives a Vix for [N(4)111][Tf2N] of 235.3 mL/mol.
■
Vix = 2(16.35) + 6(10.48) + 4(12.43) + 2(22.91) + 1(14.39) + 14( −6.56) = 113.7 mL/mol
EXPERIMENTAL SECTION
Instruments and Materials. Ultrapure water was dispensed from a Millipore Direct-Q 3UV dispenser system (resistivity 18.2 MΩ-cm, 25 °C). The aqueous IL concentrations were measured using a Shimadzu TOC-L/TNM-L total organic carbon/total nitrogen analyzer. In addition, this study used a Sigma 70 interfacial tensiometer by KSV Instruments capable of accurately measuring interfacial tensions using the ring method. Table 1 lists the 15 ILs used in this study along with their names, abbreviations used in the manuscript, CAS numbers (when available), molecular weights (MW), intrinsic molar volumes, and group contribution calculated molecular volumes, Vix. Using the abbreviations in Table 1, [N(4)111][Tf2N] was acquired from IoLiTec Inc., Tuscaloosa, AL, USA, with a purity ≥99% (Cat. No. IL-0032-HP-50g, Lot: 03). Oak Ridge National Laboratories (ORNL, Dr. Gary A. Baker; now at University of Missouri) supplied the [N(4)11(i-3)][Tf2N], [N(6)11(i-3)][Tf2N], and [N(10)11(i-3)][Tf2N] ILs. [N(1)444][Tf2N] was obtained from the Department of Chemistry and Biochemistry at the University of Mississippi (Dr. Charles Hussey). The remainder of the ILs were synthesis in the Department of Chemical Engineering, University of Mississippi, using the following reagents: lithium bis(trifluoromethylsulfonyl)imide (Li[Tf2N], CAS 90076-65-6, TCI-American #B2542), 1-butyl-3-methylimidazolium chloride (CAS 79917-90-1, Sigma-Aldrich #94128, ≥ 98%), 1-hexyl-3methylimidazolium chloride (CAS 171058-17-6), 1-decyl-3methylimidazolium chloride (CAS 171058-18-7, Sigma-Aldrich #690597, ≥ 96%), hexyltrimethylammonium bromide (CAS 2650-53-5, TCI-America #H0534, > 98%), octyltrimethylammonium bromide (CAS 2083-68-3, TCI-America #O0163, > 98%), decyltrimethylammonium bromide (CAS 2082-84-0, TCI-America #D1467, > 99%), dodecyltrimethylammonium bromide (CAS 1119-94-4), tetradecyltrimethylammonium bromide (CAS 1119-97-7), hexadecyltrimethylammonium bromide (CAS 57-09-0), and triethyl(hexyl)ammonium iodide that was synthesized at the University of Mississippi from trimethylamine and 1-iodohexane.21 The Procedures Section, below, summarizes the synthesis methods for these chemicals while Kilaru et al.21 give the details on the synthesis methods for the in-house and ORNL ILs. The ORNL ILs had nondetectable halide content (25 °C). For the ILs acquired from others, the water samples resulted from combining 1 g of pure IL with 38 mL of ultrapure water in 40 mL septa vials. The septa vials then were placed in a
fc =
12.01 × Cn MW
(4)
where fc is the fraction of carbon in the compound, Cn is the number of carbons in the compound, and MW is the total molecular weight of the compound (g/mol). Leading to sat C i,w =
TOC fc
(5)
where TOC is the total organic carbon in mg/L and Csat i,w is the solubility of the IL in water (mg/L). Finally, the mole fraction of IL in water results from sat x i,w =
sat 18 × C i,w
MW
(6)
The precision of the TOC Analyzer was ±4 μg-carbon/L. We also assessed the procedural random error via triplicate analysis of [N(6)222][Tf2N]. In this triplicate analysis we ran three D
DOI: 10.1021/acs.jpcb.7b05109 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 2 identifies the CTA-ILs that are solids in their pure state at the test conditions (25 °C). Error Analysis. This error analysis discusses three types of errors related to our data: procedural random error, instrument precision, and systematic error. As stated in the Total Organic Carbon Testing of Water Saturated with IL Section, the procedural random error determined via triplicate analysis was 0.5%, which is significantly greater than the instrumentation error of the TOC analyzer (4 μg-carbon/L or
