Water Phase Diagram Is Significantly Altered by Imidazolium IonicLiquidVitaly V. Chaban†,‡,* and Oleg V. Prezhdo‡

†MEMPHYS - Center for Biomembrane Physics, Syddansk Universitet, Campusvej 55, Odense M 5230, Kingdom of Denmark‡Department of Chemistry, University of Rochester, 120 Trustee Road, Rochester, New York 14627, United States

ABSTRACT: We report unusually large changes in the boiling temperature, saturatedvapor pressure, and structure of the liquid−vapor interface for a range of 1-butyl-3-methyl tetrafluoroborate, [C4C1IM][BF4]−water mixtures. Even modest molar fractionsof [C4C1IM][BF4] significantly affect the phase behavior of water, as represented, forinstance, by strong negative deviations from Raoult’s law, extending far beyond thestandard descriptions. The investigation was carried out using classical moleculardynamics employing a specifically refined force field. The changes in the liquid−vaporinterface and saturated vapor pressures are discussed at the atomistic resolution. Thereported results guide the search for novel scientific and technological applications ofion−molecular systems.

SECTION: Liquids; Chemical and Dynamical Processes in Solution

Ionic and molecular liquids work together for a number offascinating applications.1−6 Over past decades, the field of

room-temperature ionic liquids (RTILs) has been in thespotlight of the scientific and industrial community. RTILs areviewed as a promising alternative to traditional organic solvents.Cations of RTILs are bulky organic molecules with manifoldsubstituents. The aromatic ring contains either positivelycharged nitrogen, sulfur, or phosphorus atoms (e.g., N,N′-dialkylimidazolium, N-alkylpyridinium, alkylammonium, alkyl-phosphonium, alkylsulphonium, tiazolium, etc.). Asymmetry ofcation is an important prerequisite for the ionic compound toremain liquid at room temperature. In turn, anions areinorganic or organic species, such as halides, tetrafluoroborate,hexafluorophosphate, bis(trifluoromethylsulfonyl)imide, ace-tate, dicyanamide, and so on. Many of these solvents arenonflammable, and they exhibit negligible vapor pressure andexcellent thermal stability.6−11 In combination with certainmolecular liquids, including water, RTILs are used forseparation applications,7 as electrolyte solutions,12 and evenfor nuclear fuel reprocessing.13−15

The mixtures of imidazolium RTILs and water have beeninvestigated in detail.16−25 Recent works have been devoted tothe liquid−vapor equilibria of imidazolium RTILs and organicsolvents, such as methanol, ethanol, tetrahydrofuran, acetone,and so on.26−29 Still, experimental data on liquid−vaporequilibria in systems containing ionic liquids are scarce becausethe majority of conventional equilibrium cells are not adequatefor this kind of systems. In the two month old publication,Passos et al. suggested that boiling point elevation stronglydepends on the anion. For instance, [C4C1IM][CF3SO3]elevates the boiling point of water by 6.2 K, while [C4C1IM]-[C1SO3] elevates the boiling point by 27.2 K. The molarfraction of water in both cases equals to 72%. The observed

difference is likely due to fluorination of the first anion. Thedifference between ion sizes likely plays a role as well.In the present work, we report atomistic resolution molecular

dynamics (MD) simulations of several 1-butyl-3-methylimida-zolium tetrafluoroborate, [C4C1IM][BF4]−water mixtures, andcenter the discussion on how the RTIL alters evaporationbehavior of water. According to the implemented molecularmodels, [C4C1IM][BF4] introduces major changes to theliquid−gas phase transition of water. The observed changescannot be described by the classical Raoult’s law. As such, the[C4C1IM][BF4]−water systems must be considered highlynonideal mixtures.Figure 1 depicts the constructed atomistic model for the

liquid−vapor interface of the [C4C1IM][BF4]−water mixtures.

Received: March 19, 2014Accepted: April 22, 2014Published: April 22, 2014

Figure 1. Liquid−vapor equilibrium simulated for the equimolarmixture of [C4C1IM][BF4] and water. [C4C1IM] cations, [BF4]anions, and water molecules are shown using red, yellow, and whiteballs, respectively. The coordinate axes show the orientation of thesimulated system in the MD simulation cell.

Letter

pubs.acs.org/JPCL

© 2014 American Chemical Society 1623 dx.doi.org/10.1021/jz500563q | J. Phys. Chem. Lett. 2014, 5, 1623−1627

Dow

nloa

ded

via

UN

IV O

F SO

UT

HE

RN

CA

LIF

OR

NIA

on

Nov

embe

r 7,

201

9 at

19:

56:5

8 (U

TC

).Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

Each atom of the ionic liquid and water corresponds to oneinteraction site. Therefore, the water model contains 3sites, the1-butyl-3-methylimidazolium cation contains 25 sites, and thetetrafluoroborate anion contains 5 sites.Saturated vapor pressures in all systems plotted as a function

of temperature (Figure 2) obey an exponential dependence.

The point of intersection of these curves with the horizontalline, standing for the normal atmospheric pressure, defines thenormal boiling point for each mixture. The TIP3P water modelsomewhat underestimates the normal boiling point (362 K).However, as compared with many other water models, theunderestimation by 11 K is not critical for the currentinvestigation. Although it is quite easy to adjust the boilingpoint for small molecules (for instance, by an increase in thedipole moment), this step would also imply a reparameteriza-tion of the RTIL−water interactions in the liquid phase. Suchreparameterization requires additional, time-consuming testingand likely further adjustments of the interaction parameters.This work reports only boiling point differences between purewater and water containing certain fraction of RTIL. Therefore,the absolute boiling point of water is not of primaryimportance. Noteworthy, even modest molar fractions of[C4C1IM][BF4] significantly alter saturated vapor pressures ofwater and, consequently, increase the normal boiling point.RTIL ions were not detected in the vapor phase of any of thesimulated system. While water evaporates, RTIL remains liquid.Hence, the simulated liquid phase is more RTIL-rich than it isimplied by the initial mixture composition.Figure 3 summarizes the boiling points as a function of the

RTIL content. Notably, the boiling point of the equimolar

[C4C1IM][BF4]−water mixture increases by 25 K. Incomparison, Raoult’s law prediction is just 3.26 K, whereascorrection for the van’t Hoff factor increases it to 6.52 K. Bothpredictions are several times smaller than the actual rise in theboiling point. This negative deviation from Raoult’s law reflectsstrongly favorable interactions in the liquid phase of thesimulated mixtures. It is known from experiments thatimidazolium RTILs mix with water in virtually any proportion.Water and imidazolium RTILs possess very different phasediagrams. Indeed, imidazolium RTILs exhibit a very large liquidrange, up to several hundreds of degrees Celsius. Many RTILsdecompose before evaporation. The combination of the vastdifferences in the phase diagrams of water and the RTIL, withstrongly favorable interactions, allows [C4C1IM][BF4] toheavily adjust the temperature-dependent properties of water.There are many molecular liquids, which mix with water in anyor nearly any proportion (for instance, dimethyl sulfoxide,acetonitrile, low molecular alcohols), but none of them altersthe water phase diagram so significantly. The mentioned liquidsevaporate at comparable or lower temperatures than water.Therefore, the discussed effect cannot be observed.The density of liquids at the liquid−vapor interface (Figure

4) provides important information regarding their phasebehavior. The liquid−vapor interfaces appear well-defined forall [C4C1IM][BF4]−water systems at 360 K. The liquiddensities of the mixture components evolve in agreementwith the composition of the simulated systems. Interestingly,the cation density is larger than the anion density, even thoughanions contain heavier atoms (fluorines and no hydrogens). Inturn, cations contain a ring, which always exhibits a higher mass

Figure 2. Saturated vapor pressure as a function of temperature for theequilibrated liquid−vapor interfaces. The percentages on the plotsshow molar fractions of [C4C1IM][BF4] in its mixtures with water.Horizontal dotted lines correspond to the normal atmosphericpressure.

Figure 3. Water boiling point elevation versus molar fraction of theionic liquid in the mixtures. (a) Values obtained from atomisticsimulations (inset: boiling point elevation versus molality of solution).(b) Values obtained from Raoult’s law using the tabulatedebullioscopic constant for water, 0.512 (red solid line), andconsidering the van’t Hoff factor for [C4C1IM][BF4] (green dashedline). Note that although mixtures are simulated, the vapor pressure isgenerated exclusively by water molecules (Figure 1); therefore, onlythe water normal boiling point is changed.

The Journal of Physical Chemistry Letters Letter

dx.doi.org/10.1021/jz500563q | J. Phys. Chem. Lett. 2014, 5, 1623−16271624

density than the aliphatic compound of the same composition.No RTIL ions were detected in the vapor phase (logarithm ofzero was plotted as minus infinity, Figure 4).Vapor densities are summarized in Figure 5 versus molar

fraction and temperature. The higher is the RTIL content, thesmaller is the water vapor density. None of the densitiesexceeds 1 kg m−3. The vapor densities in the equimolar[C4C1IM][BF4]−water mixtures are over two times smallerthan vapor densities in the 10% [C4C1IM][BF4]−water mixtureat the same temperatures.Appropriate sampling is a central issue in all molecular

simulation methods based on the ergodic hypothesis. It iscommonly believed that liquid−vapor interface simulationsrequire relatively short simulation times because MD of thegaseous phase is fast. Figure 6 plots vapor pressures and vapordensities versus simulated time for the most viscous of the

simulated systems. While vapor density equilibrates within 5 nsindeed, vapor pressure requires 30 ns to reach the equilibriumvalue. (See Figure 6.) Until that time, the vapor pressurecontinues to grow instead of fluctuating around an equilibriumvalue. A fluctuation would indicate insufficient sampling but fastequilibration. Other simulated systems require smaller sam-pling, as they are less viscous, and the dynamics of theindividual particles in those systems is faster. Further analysisindicates that the liquid−vapor interface in the RTIL-richsystems is more structured than in the RTIL-poor systems andin pure water. The density of water vapor near the interfacedepends on the system composition and is larger for RTIL-poor mixtures.To recapitulate, the conducted MD simulations showed that

[C4C1IM][BF4] greatly alters the liquid−vapor diagram ofwater as well as its normal-pressure at the phase transitionpoint. This result is particularly interesting from thefundamental point of view and should be general to mostRTILs because they differ significantly from molecular liquidsin the phase transitions and volatility. The unusual phasebehavior stems from the ability of many RTILs to mix freelywith water over a wide temperature range. At the same time,one should not expect the presented trends to hold foramphiphilic RTILs and RTILs exhibiting limited solubility inwater. The ability to create aqueous systems that boil andproduce purely aqueous vapor at 125 instead of 100 °C isattractive for technological applications.

■ METHODOLOGYThe presented analysis is based on MD simulation of 25liquid−vapor interfaces and five different [C4C1IM][BF4]−water mixture compositions (Table 1). Each system wassimulated during 100 000 ps with an integration time-step of0.002 ps. The first 30 ns were disregarded as equilibration (seeFigure 6), while the equilibrium properties were obtained basedon the 70 ns trajectory part. The coordinates and pressuretensor components were saved every 10 ps, that is, every 5000time steps. MD trajectories were propagated using theGROMACS simulation suite.30−32 The analysis was carried

Figure 4. Mass densities of particles along the normal direction to thefilm of liquid. The densities of water, [C4C1IM] cation, and [BF4]anion are shown by the red solid, green dashed line, and blue dashed-dotted lines, respectively. The computation was done at 360 K.

Figure 5. Vapor density as a function of molar fraction of[C4C1IM][BF4] and temperature of the mixtures. Recall that onlywater molecules were found in the simulated vapor phase (Figure 1).

Figure 6. Vapor pressure and vapor density as functions of simulationtime for the equimolar [C4C1IM][BF4]−water mixture (the mostviscous mixture). The vertical dotted lines show simulation times,when the aqueous vapor becomes saturated; therefore, the liquid−vapor interface becomes equilibrated. The simulation was conducted at360 K.

The Journal of Physical Chemistry Letters Letter

dx.doi.org/10.1021/jz500563q | J. Phys. Chem. Lett. 2014, 5, 1623−16271625

out using in-home tools developed by V.V.C. and GROMACSstandard analysis tools, where possible.30−32 Vapor pressurewas directly recomputed from the virial.32

The force field for [C4C1IM][BF4], suggested previously byone of us,33−35 was used. The Coulomb and Lennard-Jonesparameters for the OPLS/AA implementation of TIP3P36 arecompatible with the parameters of our RTIL force field;33−35

therefore, additional parametrization was not required. Thecombination of TIP3P with the RTIL force field describes verywell the miscibility of the RTIL and water.All simulations were conducted in the constant temperature,

constant volume ensemble (NVT). The film composed of[C4C1IM][BF4] and water (Table 1) was located at the centerof the simulation box. We added 25 nm of vacuum to create aninterface. All systems were maintained at the requestedtemperature (see Table 1 for details) using velocity rescalingBussi−Parrinello thermostat.37 The time constant of 1.0 ps wasapplied for weak temperature coupling. All velocities werecoupled to the external heat bath simultaneously, withoutdivision into liquid and vapor particles. No restrictions wereimposed to particle exchange between liquid and saturatedvapor phases. The cutoff distance of 1.2 nm for the Lennard-Jones potential was employed in conjunction with shifted forcemodification between 1.1 and 1.2 nm. The electrostaticinteractions were computed using the direct pairwise Coulombpotential at separations smaller than 1.4 nm and using thereaction-field-zero scheme for all distances beyond the cutoff.The neighbor list was updated every 0.02 ps within 1.4 nm.Periodic boundary conditions in the three Cartesian directionswere simulated.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: v.chaban@rochester.edu; chaban@sdu.dk; vvchaban@gmail.com.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

The work was supported by grant CHE-1300118 from the USNational Science Foundation. MEMPHYS is the DanishNational Center of Excellence for Biomembrane Physics. TheCenter is supported by the Danish National ResearchFoundation.

■ REFERENCES(1) Brennecke, J. E.; Gurkan, B. E. Ionic Liquids for CO2 Captureand Emission Reduction. J. Phys. Chem. Lett. 2010, 1, 3459−3464.(2) Rocha, M. A. A.; Santos, L. M. N. B. F. First Volatility Study ofthe 1-Alkylpyridinium Based Ionic Liquids by Knudsen Effusion.Chem. Phys. Lett. 2013, 585, 59−62.

(3) Chaban, V. V.; Prezhdo, O. V. Ionic and Molecular Liquids:Working Together for Robust Engineering. J. Phys. Chem. Lett. 2013,4, 1423−1431.(4) Palacio, M.; Bhushan, B. A Review of Ionic Liquids for GreenMolecular Lubrication in Nanotechnology. Tribol Lett. 2010, 40, 247−268.(5) Holbrey, J. D.; Rogers, R. D. Green Industrial Applications ofIonic Liquids: Technology Review. ACS Symp. Ser. 2002, 818, 446−458.(6) Hantal, G.; Cordeiro, M. N. D. S.; Jorge, M. What Does an IonicLiquid Surface Really Look Like? Unprecedented Details fromMolecular Simulations. Phys. Chem. Chem. Phys. 2011, 13, 21230−21232.(7) Ho, T. D.; Canestraro, A. J.; Anderson, J. L. Ionic Liquids inSolid-Phase Microextraction: A Review. Anal. Chim. Acta 2011, 695,18−43.(8) Lin, R. Y.; Taberna, P. L.; Fantini, S.; Presser, V.; Perez, C. R.;Malbosc, F.; Rupesinghe, N. L.; Teo, K. B. K.; Gogotsi, Y.; Simon, P.Capacitive Energy Storage from −50 to 100 °C Using an Ionic LiquidElectrolyte. J. Phys. Chem. Lett. 2011, 2, 2396−2401.(9) Maginn, E. J. What to Do with CO2. J. Phys. Chem. Lett. 2010, 1,3478−3479.(10) Chaban, V. V.; Prezhdo, O. V. How Toxic Are Ionic Liquid/Acetonitrile Mixtures? J. Phys. Chem. Lett. 2011, 2, 2499−2503.(11) Maciel, C.; Fileti, E. E. Molecular Interactions betweenFullerene C-60 and Ionic Liquids. Chem. Phys. Lett. 2013, 568, 75−79.(12) Lu, W.; Hartman, R.; Qu, L. T.; Dai, L. M. NanocompositeElectrodes for High-Performance Supercapacitors. J. Phys. Chem. Lett.2011, 2, 655−660.(13) Shkrob, I. A.; Marin, T. W.; Wishart, J. F. Ionic Liquids Based onPolynitrile Anions: Hydrophobicity, Low Proton Affinity, and HighRadiolytic Resistance Combined. J. Phys. Chem. B 2013, 117, 7084−7094.(14) Marin, T. W.; Shkrob, I. A.; Dietz, M. L. Hydrogen-BondingInteractions and Protic Equilibria in Room-Temperature Ionic LiquidsContaining Crown Ethers. J. Phys. Chem. B 2011, 115, 3912−3918.(15) Wishart, J. F. Ionic Liquids and Ionizing Radiation: Reactivity ofHighly Energetic Species. J. Phys. Chem. Lett. 2010, 1, 3225−3231.(16) Kurnia, K. A.; Freire, M. G.; Coutinho, J. A. P. Effect ofPolyvalent Ions in the Formation of Ionic-Liquid-Based AqueousBiphasic Systems. J. Phys. Chem. B 2014, 118, 297−308.(17) Schenk, J.; Panne, U.; Albrecht, M. Interaction of LevitatedIonic Liquid Droplets with Water. J. Phys. Chem. B 2012, 116, 14171−14177.(18) Ribeiro, M. C. C. High Viscosity of Imidazolium Ionic Liquidswith the Hydrogen Sulfate Anion: A Raman Spectroscopy Study. J.Phys. Chem. B 2012, 116, 7281−7290.(19) Pramanik, R.; Sarkar, S.; Ghatak, C.; Rao, V. G.; Mandal, S.;Sarkar, N. Effects of 1-Butyl-3-methyl Imidazolium TetrafluoroborateIonic Liquid on Triton X-100 Aqueous Micelles: Solvent andRotational Relaxation Studies. J. Phys. Chem. B 2011, 115, 6957−6963.(20) Dong, B.; Gao, Y. A.; Su, Y. J.; Zheng, L. Q.; Xu, J. K.; Inoue, T.Self-Aggregation Behavior of Fluorescent Carbazole-Tailed Imidazo-lium Ionic Liquids in Aqueous Solutions. J. Phys. Chem. B 2010, 114,340−348.(21) Neves, C. M. S. S.; Ventura, S. P. M.; Freire, M. G.; Marrucho, I.M.; Coutinho, J. A. P. Evaluation of Cation Influence on the

Table 1. List of Simulated Systems and Certain Simulation Parameters

no. no. [C4C1IM] cations no. [BF4] anions no. water molecules no. interaction centers temperatures (K)

1−4 1000 3000 320; 340; 360; 380;5−8 100 100 900 5700 320; 340; 360; 380;9−12 200 200 800 8400 320; 340; 360; 380;13−16 300 300 700 11 100 320; 340; 360; 380;17−20 400 400 600 13 800 320; 340; 360; 380;21−25 500 500 500 16 500 320; 340; 360; 380; 400

The Journal of Physical Chemistry Letters Letter

dx.doi.org/10.1021/jz500563q | J. Phys. Chem. Lett. 2014, 5, 1623−16271626

Formation and Extraction Capability of Ionic-Liquid-Based AqueousBiphasic Systems. J. Phys. Chem. B 2009, 113, 5194−5199.(22) Menjoge, A.; Dixon, J.; Brennecke, J. F.; Maginn, E. J.;Vasenkov, S. Influence of Water on Diffusion in Imidazolium-BasedIonic Liquids: A Pulsed Field Gradient NMR study. J. Phys. Chem. B2009, 113, 6353−6359.(23) Cromie, S. R. T.; Del Popolo, M. G.; Ballone, P. Interaction ofRoom Temperature Ionic Liquid Solutions with a Cholesterol Bilayer.J. Phys. Chem. B 2009, 113, 11642−11648.(24) Chen, T.; Chidambaram, M.; Liu, Z. P.; Smit, B.; Bell, A. T.Viscosities of the Mixtures of 1-Ethyl-3-Methylimidazolium Chloridewith Water, Acetonitrile and Glucose: A Molecular DynamicsSimulation and Experimental Study. J. Phys. Chem. B 2010, 114,5790−5794.(25) Wang, Y.; Li, H. R.; Han, S. J. a Theoretical Investigation of theInteractions between Water Molecules and Ionic Liquids. J. Phys.Chem. B 2006, 110, 24646−24651.(26) Kato, R.; Gmehling, J. Measurement and Correlation of Vapor-Liquid Equilibria of Binary Systems Containing the Ionic Liquids[EMIM] [(CF3SO2)(2)N], [BMIM] [(CF3SO2)(2)N], [MMIM][(CH3)(2)PO4] and Oxygenated Organic Compounds RespectivelyWater. Fluid Phase Equilib. 2005, 231, 38−43.(27) Passos, H.; Khan, I.; Mutelet, F.; Oliveira, M. B.; Carvalho, P. J.;Santos, L. M. N. B. F.; Held, C.; Sadowski, G.; Freire, M. G.;Coutinho, J. A. P. Vapor-Liquid Equilibria of Water plusAlkylimidazolium-Based Ionic Liquids: Measurements and Per-turbed-Chain Statistical Associating Fluid Theory Modeling. Ind.Eng. Chem. Res. 2014, 53, 3737−3748.(28) Heym, F.; Haber, J.; Korth, W.; Etzold, B. J. M.; Jess, A. VaporPressure of Water in Mixtures with Hydrophilic Ionic Liquids - AContribution to the Design of Processes for Drying of Gases byAbsorption in Ionic Liquids. Chem. Eng. Technol. 2010, 33, 1625−1634.(29) Carvalho, P. J.; Khan, I.; Morais, A.; Granjo, J. F. O.; Oliveira, N.M. C.; Santos, L. M. N. B. F.; Coutinho, J. A. P. A NewMicroebulliometer for the Measurement of the Vapor-LiquidEquilibrium of Ionic Liquid Systems. Fluid Phase Equilib. 2013, 354,156−165.(30) Berendsen, H. J. C.; van der Spoel, D.; van Drunen, R.GROMACS - A Message-Passing Parallel Molecular-DynamicsImplementation. Comput. Phys. Commun. 1995, 91, 43.(31) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS4: Algorithms for Highly Efficient, Load-Balanced, And ScalableMolecular Simulation. J. Chem. Theory Comput. 2008, 4, 435−447.(32) Lindahl, E.; Hess, B.; van der Spoel, D. GROMACS 3.0: APackage for Molecular Simulation and Trajectory Analysis. J. Mol.Model. 2001, 7, 306.(33) Chaban, V. V. Polarizability versus Mobility: Atomistic ForceField for Ionic Liquids. Phys. Chem. Chem. Phys. 2011, 13, 16055−16062.(34) Chaban, V. V.; Prezhdo, O. V. a New Force Field Model of 1-Butyl-3-methylimidazolium Tetrafluoroborate Ionic Liquid andAcetonitrile Mixtures. Phys. Chem. Chem. Phys. 2011, 13, 19345−19354.(35) Chaban, V. V.; Voroshylova, I. V.; Kalugin, O. N. a New ForceField Model for the Simulation of Transport Properties ofImidazolium-Based Ionic Liquids. Phys. Chem. Chem. Phys. 2011, 13,7910−7920.(36) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. Developmentand Testing of the OPLS All-Atom Force Field on ConformationalEnergetics and Properties of Organic Liquids. J. Am. Chem. Soc. 1996,118, 11225.(37) Bussi, G.; Donadio, D.; Parrinello, M. Canonical Samplingthrough Velocity Rescaling. J. Chem. Phys. 2007, 126, 014101.

The Journal of Physical Chemistry Letters Letter

dx.doi.org/10.1021/jz500563q | J. Phys. Chem. Lett. 2014, 5, 1623−16271627