Effect of Cation Symmetry on the Morphology and Physicochemical

ARTICLE pubs.acs.org/JPCB

Effect of Cation Symmetry on the Morphology and Physicochemical Properties of Imidazolium Ionic Liquids Wei Zheng,† Ali Mohammed,‡ Larry G. Hines, Jr.,§ Dong Xiao,§ Omar J. Martinez,§ Richard A. Bartsch,§ Sindee L. Simon,† Olga Russina,|| Alessandro Triolo,*,^ and Edward L. Quitevis*,‡,§ Department of Chemical Engineering, ‡Department of Physics, and §Department of Chemistry & Biochemistry, Texas Tech University, Lubbock, Texas 79409, United States Department of Chemistry, University of Rome “Sapienza”, Rome, Italy ^ Istituto Struttura della Materia, National Research Council of Italy, Rome, Italy

)

†

bS Supporting Information ABSTRACT: In this paper, the morphology and bulk physical properties of 1,3-dialkylimidazolium bis{(trifluoromethane) sulfonyl}amide ([(CN/2)2im][NTf2]) are compared to that of 1-alkyl-3-methylimidazolium bis{(trifluoromethane)sulfonyl} amide ([CN
1C1im][NTf2]) for N = 4, 6, 8, and 10. For a given pair of ionic liquids (ILs) with the same N, the ILs differ only in the symmetry of the alkyl substitution on the imidazolium ring of the cation. Small-wide-angle X-ray scattering measurements indicate that, for a given symmetric/asymmetric IL pair, the structural heterogeneities are larger in the asymmetric IL than in the symmetric IL. The correlation length of structural heterogeneities for the symmetric and asymmetric salts, however, is described by the same linear equation when plotted versus the single alkyl chain length. Symmetric ILs with N = 4 and 6 easily crystallize, whereas longer alkyl chains and asymmetry hinder crystallization. Interestingly, the glass transition temperature is found to vary inversely with the correlation length of structural heterogeneities and with the length of the longest alkyl chain. Whereas the densities for a symmetric/asymmetric IL pair with a given N are nearly the same, the viscosity of the asymmetric IL is greater than that of the symmetric IL. Also, an even
odd effect previously observed in molecular dynamics simulations is confirmed by viscosity measurements. We discuss in this paper how the structural heterogeneities and physical properties of these ILs are consistent with alkyl tail segregation.

1. INTRODUCTION Room-temperature ionic liquids (ILs) are defined as salts with melting points at or below 373 K. A typical IL is composed of a bulky organic cation and an inorganic or organic anion. The ability to tune the physicochemical properties of ILs by changing the structure of the ions has led to their being called “designer” solvents.1,2 Because of their unique properties and potential use in a variety of areas such as synthesis, chemical separations, biocatalysis, electrochemistry, solar cells, tribology, and ion
gel technology, ILs have recently been the subject of a large amount of research.3
13 Given that the number of ILs is estimated to be a million,14 a molecular-level understanding of ILs is clearly necessary for the rational design of ILs for current and future applications. There have been numerous attempts to gain a fundamental understanding of the physicochemical properties of ILs.15
23 In particular, Watanabe and co-workers15
17 used the method of pulsed-field gradient spin
echo (PGSE) NMR to obtain cationic and anionic self-diffusion coefficients of imidazolium-based ILs. By correlating microscopic self-diffusion coefficients and macroscopic transport properties, such as viscosity and ionic r 2011 American Chemical Society

conductivity, of a series of ILs with varying anion and alkyl chain length, Watanabe and co-workers were able to provide insights into the nature of the interionic forces that underlie the physicochemical properties of ILs. Another unique property of ILs is the nanoscale heterogeneity in their liquid state. Small-wide-angle X-ray scattering (SWAXS) measurements have provided the most direct evidence for this structural heterogeneity. SWAXS measurements24
26 on ILs based on the 1-alkyl-3-methylimidazolium cation [CnC1im]þ with the anions [Cl]
, [BF4]
, and [PF6]
for even n and bis{(trifluoromethane)sulfonyl}amide ([NTf2]
) for both even and odd n have shown that a low Q (momentum transfer) peak can be observed in the diffraction pattern from ILs bearing a long enough alkyl chain, in agreement with structural findings from molecular dynamics (MD) simulations.27
32 These MD simulations indicate that the polar head groups and the alkyl chains tend to segregate, thus leading to structural domains whose average Received: December 5, 2010 Revised: March 31, 2011 Published: May 02, 2011 6572

dx.doi.org/10.1021/jp1115614 | J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B size depends on the alkyl chain length. This segregation is attributed to competition between the electrostatic interactions between charged imidazolium rings and anions and the collective short-range interactions between neutral alkyl chains. The X-ray diffraction data from Triolo and co-workers24
26,33,34 (for imidazolium, piperidinium, and phosponium salts), from Mizuhata et al.35 (for aliphatic quaternary ammonium salts), and the neutron diffraction data from Atkin and Warr36 (for alkylammonium salts) confirm this structural heterogeneity, providing a clear quantitative description of the nature of this phenomenon. Spectroscopic signatures of this structural heterogeneity have also been found in Raman (linear and nonlinear)37,38 and optical Kerr effect (OKE) measurements39
41 on ILs. Cation symmetry provides another structural “knob” with which to tweak the physicochemical properties of ILs. Dzyuba and Bartsch42 reported the synthesis of 10 1,3-dialkylimidazolium hexafluorophosphates with dialkyl moieties ranging from dipropyl to didecyl. They showed that the symmetrical substituted 1,3-dialkylimidazolium hexafluorophosphates with dibutyl, dipentyl, diheptyl, dioctyl, dinonyl, and didecyl had melting points below 100 C. This result is surprising, given that one of the factors in determining the low melting points of ILs is the unsymmetrical nature of the cation. For long enough alkyl chain lengths, symmetrical substituted 1,3-dialkylimidazolium salts exhibit liquid crystalline properties.43 Xiao et al.44 recently compared the SWAXS data and OKE spectra of symmetric [(Cn)2im][NTf2] with n = 2
5 with those of asymmetric [CnC1im][NTf2] with n = 2
5 and concluded that the local order is higher and intermolecular dynamics are higher in frequency for the symmetric series than for the asymmetric series. These conclusions are consistent with the structural picture revealed in recent MD simulations of these ILs.45 In the present paper, we describe studies of ILs that provide further insight into the relationship between structural heterogeneity and the physicochemical properties of ILs. The approach taken in this study differs from that of our previous study in that the morphology, thermal transitions, densities, and viscosities of [(CN/2)2im][NTf2] are compared to that of [CN
1C1im][NTf2] for N = 4, 6, 8, and 10. For convenience, the ILs are identified by the number of carbon atoms in the two alkyl groups at the 1 and 3 positions of the imidazolium ring, for example, [C5C1im][NTf2] as C5C1 and [(C3)2im][NTf2] as C3C3. In this numbering scheme, the cation can be thought of as being formed by the insertion of an imidazolium ring into an N-carbon alkane chain, with the degree of symmetry determined by the position of the ring insertion along the alkane chain as depicted in Figure 1. The advantage of this approach is that the morphology and physical properties of a pair of ILs having the same N (and therefore molar mass) but differing in the symmetry of the alkyl substitution can be compared. As we will show below, the SWAXS data indicate that, for such an IL pair, the structural heterogeneities occur over a larger spatial scale for the asymmetric salt than for the symmetric salt. Moreover, for an IL pair with a given N, the densities are nearly the same but thermal properties and viscosities can be quite different. Interestingly, the glass transition temperature of these ILs scales inversely with the size of the structural heterogeneities. For a given IL pair, differences in the viscosity and differences in the structural heterogeneity as reflected in the SWAXS measurements can be rationalized in terms of alkyl tail segregation. We also observe an even
odd effect in the viscosities similar to that observed in MD simulations of the ion diffusion coefficients.

ARTICLE

Figure 1. Structures of [(CN/2)2im]þ and [CN
1C1im]þ pairs with N = 4, 6, 8, and 10 and [NTf2]
ion.

2. EXPERIMENTAL SECTION 2.1. Sample Preparation. The synthesis of [CN
1C1im][NTf2] and [(CN/2)2im][NTf2] has been described previously.44,46 The water content, wH2O, of the samples was determined by Karl Fischer titration. Prior to thermal, density, and viscosity measurements, the samples were kept in tightly sealed vials and stored in a nitrogenpurged glovebox or desiccator. Samples used in the SWAXS measurements were kept in vials with septum caps. 2.2. SWAXS Measurements. The SWAXS experiment was conducted for all the samples at the high-brilliance beamline ID02, European Synchrotron Radiation Facility (ESRF), Grenoble, France,47 by use of an instrumental setup that covers the momentum range Q between 0.1 and 2 Å
1, with a wavelength λ = 0.75 Å (energy = 16.5 keV). Measurements were performed at 25 C, by use of a thermostated bath, and the samples were kept inside a temperature-controlled flow-through cell, with internal diameter of 1.9 mm. The corresponding empty cell contribution was subtracted. Calibration in absolute units (mm
1) was obtained via a neat water sample in a 2-mm capillary. 2.3. Differential Scanning Calorimetry. Calorimetric measurements were performed under nitrogen atmosphere by use of a Mettler-Toledo differential scanning calorimeter (DSC), MT DSC1, equipped with a liquid nitrogen cooling system. To remove or reduce the adventitious moisture from the materials, the DSC samples were prepared from ILs that were stored in a vacuum desiccator after their synthesis. Sample sizes varied from 8 to 13 mg. Hermetically sealed aluminum sample pans from Perkin-Elmer were used. Data were obtained on heating at 10 K/min from
120 to 40 C after cooling at 10 K/min from 40 to
120 C. When not in use, all DSC samples were stored in the desiccator. The limiting fictive temperature, Tf0 , was calculated from DSC heating scans after cooling at 10 K/min by use of the Mettler-Toledo STAR software version 9.20. The limiting fictive temperature depends only on the cooling rate and is known to be approximately equal to the glass transition temperature, Tg, measured on cooling at the same cooling rate.48
50 Tf0 was 6573

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B

ARTICLE

Table 1. Comparison of Volumetric Properties of Selected Ionic Liquids at 298.15 K IL

molar mass, g 3 mol
1

density, g 3 cm
3

molar volume, cm3 3 mol
1

molar vdW volume,a cm3 3 mol
1

free volume,b cm3 3 mol
1

C4C1

419.37

1.438c

291.6

199.6

92.0

C3C3

433.39

1.400d

309.6

210.8

98.8

C5C1

433.39

1.403d

308.9

210.7

98.2

C6C1

447.42

1.372e

326.1

221.7

104.4

Molar van der Waals (vdW) volume (cation þ anion), calculated by use of quantum chemistry program PC Spartan (Wave function, Inc.) at the semiempirical AM1 level ([NTf2]
molar vdW volume = 98.4 cm3 3 mol
1). b Free volume = molar volume
molar vdW volume. c Reference 54. d This work (water content of samples ≈ 200 μg/g). e References 52 and.53. a

obtained by integrating the heat flow curve and then extrapolating the liquid line to the glassy line, a procedure consistent with the method proposed by Moynihan et al.:51 Z T . Tg Z T . Tg ðCpl
Cpg Þ dT ¼ ðCp
Cpg Þ dT ð1Þ Tf 0

T , Tg

where the heat capacities Cpl, Cpg, and Cp are functions of temperature. These ILs also showed cold crystallization and melting peaks on the heating scans. The onset of these events and their associated enthalpy changes were also calculated with the Mettler-Toledo software. Enthalpies are reported per gram of material, not per gram of crystalline material, because of the presence of both amorphous and crystalline phases in the supercooled samples. On the basis of multiple runs on some samples, Tg values are considered to be (0.6 K, cold crystallization temperatures are (1 K, and melting temperatures are (0.1 K. Temperature calibration for the DSC was performed upon heating at 10 K/min with hexane (Tm =
94 C), octane (Tm =
57 C), and gallium (Tm = 29.78 C). The enthalpy of fusion of octane, 180.0 J/g, was used to calibrate the heat flow. The temperature is considered to be within (0.2 K and the heat flow to be accurate within 2%. The calibrations were checked at regular intervals on heating during the DSC studies by performing check runs with the above calibration materials. 2.4. Density Measurements. Densities of the ILs were measured to an accuracy of 0.1% as a function of temperature by use of a vibrating tube density meter (Anton Paar DMA 60 and 602) with temperature control from a recirculating water bath. Samples were transferred to a gastight syringe in a nitrogenpurged glovebox. With the syringe, samples were then introduced into the density meter through a lab-built, nitrogen-purged injection system. To determine the accuracy of our procedure for measuring density, we compared the density of C5C1 at 298.15 K to the densities of C4C1 and C6C1 at 298.15 K. The densities listed in Table 1 correspond to the IUPAC-recommended values for dry C6C152,53 and the values for dry C4C1 from the work of Jacquemin et al.,54 with “dry” being defined by a water content Tg and a different VTF equation for Tg < T < TB.77

4. DISCUSSION 4.1. SWAXS Data and MD Simulations. Whereas SWAXS measurements provide direct evidence for the existence of structural heterogeneities in ILs, MD simulations provide a molecular-level picture of the morphology in these systems. With the exception of MD simulations of ILs based on the 1,3-dimethylimidazolium cation ([C1C1im]þ),28,29,64 most MD simulations of imidazolium ILs to date have focused on asymmetric salts with small anions such as [F]
, [Cl]
, [Br]
, [NO3]
, and [PF6].27
29,78
80 Raju and Balasubramanian45 reported results from MD simulations of a series of ILs based on [NTf2]
. Lopes and Padua30 previously reported radial distribution functions (RDFs) between the alkyl tails of the cations in [C2C1im][NTf2] and [C4C1im][NTf2]. Recently, a comparison of energy-dispersed X-ray diffraction data and corresponding observables derived from MD simulations of a series of [CnC1im][NTf2] (n = 1, 4, 6, 8) was presented with the aim of validating interatomic potentials and accessing microscopic structural details.81 In contrast to these two latter studies, the MD simulations of Raju and Balasubramanian45 of ILs based on the [NTf2]
anion were more extensive and included both 6578

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B

ARTICLE

Table 4. Density Parameters of Ionic Liquidsa F0,

a, 10
4

IL

g 3 cm
2

g 3 cm
3 3 K
1

Vm(30),

M30,c 10
3

cm3 3 mol
1

mol 3 cm3

b

C3C1

1.773

10.0

275.1

3.635

C2C2

1.771

10.0

276.4

3.618

C5C1

1.692

9.67

310.2

3.224

C3C3

1.692

9.78

311.8

3.207

C7C1

1.617

9.13

344.4

2.904

C4C4

1.614

9.12

345.5

2.894

C9C1 C5C5

1.561 1.560

8.81 8.90

378.5 379.3

2.642 2.636

a Parameters correspond to the linear least-squares fit of the equation F = F0
aT to the density vs temperature data. b Vm(30), molar volume at 30 C. c M30, molar concentration at 30 C.

Figure 9. Expanded plot of density for [(C3)2im][NTf2] and [C5C1im][NTf2]. Solid lines are linear fits of the equation F = F0
aT to the data, with fit parameters given in Table 4.

Figure 10. Plot of molar volumes of [(CN/2)2im][NTf2] with N = 4, 6, 8, 10 and [CN
1C1im][NTf2] with N = 3
10 [odd N data from Tokuda et al.;17 even N data from this work) at 30 C.

the asymmetric cation [CnC1im]þ and the symmetric cation [(CnCnim]þ. These simulations were motivated by the SWAXS observations of Xiao et al.,44 who showed that the symmetric C5C5 system is more structured than the C5C1 system. Although the main goal of the simulations was to explain these specific experimental observations in terms of the intermolecular/ intramolecular correlations, the simulations provided further new

Figure 11. Arrhenius plot of viscosities for [CN
1C1im][NTf2] with N = 4, 6, 8, and 10.

Figure 12. Arrhenius plot of viscosities for [(CN/2)2im][NTf2] with N = 4, 6, 8, and 10.

insights into the liquid morphology of ILs based on [NTf2]
and guidance in understanding X-ray scattering from these systems in general; accordingly, we recollect here the most relevant findings from the simulations of Raju and Balasubramanian. (1) The very high first peak in the RDF between the terminal carbons of alkyl chains for both the symmetric and asymmetric salts is indicative of tail segregation. The intensity of the first peak in the RDF of the cation with propyl chains (C3C1 and C3C3) being greater than one provides additional support for the existence of structural heterogeneities even in relatively short alkyl chain members. (2) The structure factors of asymmetric and symmetric cation-based ILs calculated from the MD simulations exhibit peaks at 5.0, 7.8, and 13.2 nm
1, in agreement with the observed SWAXS data. Moreover, the intensity of the peaks in the calculated X-ray structure factor vary in the same way as in the observed SWAXS data, with the low-Q peak increasing, the intermediate-Q peak decreasing, and the high-Q peak remaining relatively constant with increasing chain length. (3) On the basis of the partial structure factors (PSFs), the peak at 13.2 nm
1 can be attributed to correlations between atoms of the anion and between atoms of the imidazolium ring, which explains the lack of dependence of this peak on alkyl chain length. In contrast, the tail
tail PSF makes the largest contribution to the peak at 5.0 nm
1, which explains the increase in peak intensity with increasing chain length. As for the peak at 7.8 nm
1, the PSF 6579

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B

ARTICLE

Figure 13. Comparison of viscosity data for [(CN/2)2im][NTf2] and [CN
1C1im][NTf2] pairs with N = 4, 6, 8, and 10. Solid lines are fits of the VTF equation (eq 2) to the data, with fit parameters given in Tables 5 and 6.

Table 5. VTF Parameters of Asymmetric Cation Ionic Liquidsa

a

Table 6. VTF Parameters of Symmetric Cation Ionic Liquidsa IL

η0, mPa 3 s

B, K

T0, K

C2C2

0.447 ( 0.021

565 ( 13

163.6 ( 1.5

C3C3

0.240 ( 0.017

731 ( 19

162.0 ( 1.7

0.207 ( 0.17 0.194 ( 0.009

777 ( 208 824 ( 12

162.7 ( 16.5 163.8 ( 1.0

IL

η0, mPa 3 s

B, K

T0, K

C3C1

0.421 ( 0.031

617 ( 19

169.9 ( 1.9

C5C1

0.487 ( 0.067

567 ( 31

180.9 ( 3.0

C7C1

0.186 ( 0.10

799 ( 165

165.4 ( 9.3

C4C4 C5C5

C9C1

0.238 ( 0.015

788 ( 15

168.8 ( 1.3

a

See eq 2 for definition of VTF parameters.

See eq 2 for definition of VTF parameters.

between atoms of the anions makes the largest contribution, followed by the PSF between nitrogen atoms of the cation. There is also a contribution to this peak from the PSF between geometric centers of the imidazolium rings. The decrease in intensity at 7.8 nm
1 in the total structural factor is partly attributed to a decrease in intensity of the peak in this PSF in going from the propyl to pentyl alkyl group. They also showed that the first peak in the tail
tail RDF for C5C5 is narrower than that for C5C1, indicating a more structured morphology in the former, which is consistent with the experimental data. In agreement with our SWAXS data, the intensity of the total structure factor for Q < 7 nm
1 was found here to be higher for C5C5 than for C5C1. 4.2. Thermal Properties. Because of its effect on packing in the crystal, the symmetry of the ions plays an important role in

determining whether an organic salt will be an IL.82 The less symmetric the ions, the weaker the interionic interactions and the less efficient the packing in the crystal, resulting in lower melting points. For example, the melting points of N-butylpyridinium ([Nbupy]þ) salts are 100 C higher than those of their 1-ethyl3-methylimidazolium analogues because [Nbupy]þ has a mirror plane, whereas [C2C1im]þ does not.83 That CN/2CN/2 and CN
1C1 are liquids at room temperature is not surprising, given the propensity in general for [NTf2]
to inhibit crystallization due to its bulkiness and the delocalized nature of its charge distribution. Our DSC measurements suggest, however, that the ability of [NTf2]
to inhibit crystallization can be counteracted by making the cation more symmetric, as evidenced by the ease by which the symmetric salts C2C2 and C3C3 are able to crystallize relative to the asymmetric salts C3C1 and C5C1. However, for longer alkyl chains, the ability 6580

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B

ARTICLE

of [NTf2]
to inhibit crystallization wins out and only glass transitions are observed. 4.3. Densities and Viscosities. Our results indicate that the symmetry of the cation affects both density and viscosity of these ILs. At first glance, the viscosity of CN
1C1 being higher than that of CN/2CN/2 appears to be consistent with the molar concentration of CN
1C1 being higher than that of CN/2CN/2. A higher molar concentration should increase the electrostatic attraction between the anion and cation. However, the differences in viscosities cannot be solely attributed to differences in densities. This can readily be seen by using the Doolittle equation84 for the ratio of viscosities of the ILs: ln ðηS =ηA Þ ¼ b½ð1=fS Þ
ð1=fA Þ

ð3Þ

where fS and fA are the fractional free volumes of symmetric and asymmetric cation ILs, respectively, and b is a constant on the order of unity. In the case of C3C3 and C5C1, the differences in viscosities and densities are respectively, 19% and 0.21%. Using the free volumes listed in Table 1 and setting b equal to 1, we estimate that the viscosities of C3C3 and C5C1 at 298.15 K should differ by ∼2%, which is clearly much less than the experimental value of 19%. In addition, the differences in viscosities cannot be attributed to differences in the values of the glass transition: Tg values are approximately 2 K higher for the asymmetric salt for a given pair, whereas the Vogel temperatures T0 for the asymmetric salts are 3
19 K higher than those for the corresponding symmetric salt. 4.4. Dependence of Viscosity on Alkyl Chain Length. In their study of the physical properties of [CN
1C1im][NTf2] with N = 2, 3, 5, 7, 9, Tokuda et al.17 showed that the cation and anion self-diffusion constants vary linearly with T/η in accordance with the Stokes
Einstein (SE) equation: Dself ¼

kB T CπηRhyd

ð4Þ

where kB is the Boltzmann constant, T is the absolute temperature, Rhyd is the effective hydrodynamic radius, and C is a constant equal to 6 for stick boundary conditions and 4 for slip boundary conditions. Moreover, the sum of the cation and anion selfdiffusion coefficients follows the order C2C1 > C1C1 > C4C1 > C6C1 > C8C1, whereas the viscosity follows the order C2C1 < C1C1 < C4C1 < C6C1 < C8C1. The inverse correlation of the sum of the ion self-diffusion coefficient with viscosity is not unexpected and is consistent with the SE equation. However, the anomalously high (low) value of the viscosity (ion diffusivity) of C1C1 compared to the values for the other members of the series suggests that the dependence of ionic diffusion and viscosity on the number of
CH2
units is due to a balance between electrostatic and alkyl chain
ion inductive forces. The mechanism by which this balance of forces controls viscosity and ionic diffusion can be understood by examining the ratio Λimp/ΛNMR, where Λimp is the molar conductivity obtained from impedance measurements and ΛNMR is the molar conductivity calculated from ion self-diffusion coefficients. The ratio Λimp/ΛNMR is a measure of the fraction of ions that contribute to ionic conduction (i.e., the “ionicity” of the IL) and is analogous to the degree of dissociation in dilute electrolyte solutions. For the above series of ILs, the value of Λimp/ΛNMR was less than unity and decreased with the number of
CH2
units. The value of Λimp/ΛNMR being less than unity is an indication that not all of the diffusing species contribute to the

ionic conductivity because of ion association. Tokuda et al.17 attributed the decrease in Λimp/ΛNMR to alkyl chain
ion inductive forces that increase with the number of
CH2
units. The dependence of the ionic character of an IL on the number of
CH2
units is therefore a balance between a decrease in electrostatic attraction, as evidenced by the decrease in molar concentration, and an increase in van der Waals interactions, as evidenced by the decrease in the value of Λimp/ΛNMR. Within the framework of the mechanism proposed by Tokuda et al.,17 the increase in viscosity in going from C2C1 to C8C1 can also be attributed to the dominance of the effect of increasing alkyl chain
ion inductive forces with increasing number of
CH2
units, which enhances the frictional forces between ions, aggregates, and clusters, over the effect of decreasing electrostatic interaction between the ions with increasing number of
CH2
units. Conversely, the anomalous decrease in viscosity and increase in diffusivity in going from C1C1 to C2C1 is attributed to the dominance of the effect of decreasing attractive electrostatic interaction with increasing chain length over the effect of increasing ion association with increasing chain length. In principle, the increasing viscosity of the symmetric CN/2CN/2 series with increasing number of
CH2
units can also be rationalized in terms of this balance between the effect of decreasing electrostatic interaction and the effect of increasing van der Waals interactions. As for the difference in the viscosities of CN/2CN/2 and CN
1C1 for a given N, the explanation may lie in the difference in the interaction between the long alkyl chains in CN
1C1 and CN/2CN/2. Space-filling models suggest that because of the size of anions in the first solvation shell around a cation,
CH2
units at the R and β positions on the alkyl chain of the cation are sterically prevented from coming in close contact with the corresponding
CH2
units of the alkyl chain of the secondshell cations. This steric hindrance, however, will have a smaller effect on the interactions between
CH2
units furthest from the imidazolium ring in cations with long alkyl tails. Van der Waals interactions should therefore be greater between CN
1 chains in the asymmetric ILs than between CN/2 chains in the symmetric ILs. Moreover, alkyl chains should pack more efficiently in the asymmetric CN
1C1 series than in the symmetric CN/2CN/2 series. This difference in packing efficiency would be consistent with the density of CN
1C1 being slightly greater than that of CN/2CN/2 for a given N. 4.5. Structural Heterogeneity, Ionic Diffusion, and Viscosity. Wang and Voth,28 by employing a multiscale coarse-grained MD method, first showed the nanosegregation of ILs into polar regions with the cation rings and anions homogeneously distributed and nonpolar regions formed by the aggregation of alkyl groups for C4 and longer. This result was subsequently more completely analyzed29 and confirmed by all-atom MD simulations.30,85 Formation of these structurally heterogeneous domains is also considered to be due to competition between the electrostatic interactions between charged head groups and anions and the collective short-range interactions between neutral tail groups. This tail segregation mechanism provides a systematic explanation for many experimental observations, such as the SWAXS data, diffusion coefficients, and viscosities. For example, diffusion coefficients for the [CN
1C1im][NO3] series with N = 2, 3, 5, 7, and 9 obtained from the linear part of the mean square displacements follow the same trend observed by Tokuda et al.17 for the [CN
1C1im][NTf2] series. Within the framework of the tail segregation mechanism, the diffusion coefficient 6581

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B

Figure 14. Plot of viscosities for [CN
1C1im][NTf2] (N = 2
10) and [(CN/2)2im][NTf2] (N = 4, 6, 8, 10) at 298 K versus N. Viscosities in this plot are from this work (N = 3
6, 8, 10) and values calculated from the VTF parameters of Tokuda et al.17 (N = 2, 3, 5, 7, 9). Water contents of C2C1 and C4C1 samples from this work were 30 and 143 μg/g, respectively. See Table 3 for water content of other samples in this work. Water content of CN
1C1 samples from study of Tokuda et al. was 10 μg/g.

decreases with increasing alkyl chain length because of the stronger collective short-range interactions associated with longer chains. Interestingly, the MD simulations of Wang and Voth28 also showed that the diffusion constants of C3C1 and C1C1 are smaller than the diffusion constants that one would predict from the trends for C2C1, C4C1, C6C1, and C8C1. If viscosity is a reflection of the ion dynamics in ILs, then a similar even
odd effect should be also observed for viscosity. Indeed, when the viscosity of data of Tokuda et al.17 at 298.15 K for C1C1, C2C1, C4C1, C6C1, and C8C1 are combined with our viscosity data at 298.15 K for C3C1, C5C1, C7C1, and C9C1, an even
odd effect is clearly discernible, with the viscosities of C1C1, C2C1, C3C1, and C4C1 oscillating about a trend line formed by the higher order members of the series (Figure 14). Measurements for N = 3 and 5 (open symbols) performed in our laboratories are consistent with the behavior shown in Figure 14. For comparison, the viscosity data at 298.15 K for C2C2, C3C3, C4C4, and C5C5 are also shown in Figure 14. Although not as pronounced, an even
odd effect is also discernible in the viscosities of the symmetric ILs- if C1C1 is included in the series.

ARTICLE

below room temperature.83 Our results are consistent with this principle: Symmetric ILs with N = 4 and 6 easily crystallize, whereas longer alkyl chains and asymmetry hinder crystallization. Remarkably, Tg for both symmetric and asymmetric salts can be reduced when plotted versus n, the length of the longest alkyl chain. Moreover, Tg scales inversely with the correlation length D (Figure 7, inset). (3) That the densities of asymmetric and symmetric ILs with the same N are nearly the same is not surprising and is consistent with the observation that the molar volume of ILs based on [CN
1C1im]þ increases by almost exactly the same amount with the addition of methylene units in the alkyl chain.20 We think that the 0.3% difference in molar volumes is due to the difference in packing efficiency of the alkyl chains in CN
1C1 compared to that in CN/2CN/2. MD simulations are needed to confirm this explanation. (4) The fact that the viscosity of CN/2CN/2 is lower than that of CN
1C1 for a given N is surprising at first glance but is physically reasonable in light of SWAXS data and MD simulations that indicate alkyl tail segregation. One of the drawbacks of using ILs in applications is their high viscosity. These results indicate that cation symmetry can be used to lower the viscosity of an IL significantly. For example, going from C3C1 to C2C2 lowers the viscosity at 298.15 K by 36% (from 51.6 to 33.1 mPa 3 s). (5) If ionic diffusion follows SE behavior in ILs, as indicated by the work of Tokuda et al.,17 and the viscosity of CN/2CN/2 is lower than that of CN
1C1 for a given N, then cations with one long chain should diffuse more slowly than cations with two chains that are only half as long. An interesting implication of such a result is that ion diffusion in these systems should not depend on the mass of the cations. Further studies on these symmetric/asymmetric IL pairs using PGSE NMR will allow us to test these ideas. (6) Finally, the even
odd effect exhibited by ion diffusion and viscosity is another IL property that needs further study. Although there was a hint of such an effect in the data of Tokuda et al.,17 the effect is readily discernible when viscosities of the entire series of ILs are plotted together (Figure 14).86 Further measurements on the [CN
1C1im][NTf2] series with odd N and on imidazolium-based ILs with other anions will be required to confirm these observations and to understand the underlying mechanism for this effect.

’ ASSOCIATED CONTENT

5. CONCLUSIONS (1) For a symmetric/asymmetric IL pair with the same N, structural heterogeneities are larger in the asymmetric IL than in the symmetric IL. In particular, SWAXS data for C9C1 and C5C5 show well-defined low-Q peaks that witness larger domains in C9C1 than in C5C5, that are formed from the segregation of the C9 side chains in the case of C9C1 and from the segregation of the C5 side chains in the case of C5C5, respectively, as proposed on the basis of experimental data26,44 and MD simulation.45 Interestingly, the correlation length D for symmetric and asymmetric salts is described by the same linear equation when plotted versus n, indicating that the single alkyl chain length determines segregation into domains and the corresponding size of the domains. (2) One of the key principles in understanding the properties of ILs is the notion that bulky unsymmetrical organic cations lower the melting points of these salts to temperatures at or

bS

Supporting Information. Three figures and one table showing comparison of viscosity data for C4C1 and C2C1 obtained in this study and from Jacquemin et al.54 and Tokuda et al.17 This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail [email protected] (E.L.Q.) or Alessandro.Triolo@ ism.cnr.it (A.T.).

’ ACKNOWLEDGMENT This research was supported by the National Science Foundation under Grant CHE-0718678 and by the Donors of American Chemical Society Petroleum Research Fund under Grant 47615-AC6 to 6582

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B E.L.Q. This research was partially supported by the Texas Tech University Research Enhanced Fund under a grant to E.L.Q. and by The Welch Foundation under Grant D-0775 to R.A.B. A.T. acknowledges support from the FIRB “Futuro in Ricerca” research project (RBFR086BOQ_001, “Structure and dynamics of ionic liquids and their binary mixtures”). O.R. and A.T. acknowledge the European Synchrotron Radiation Facility for provision of synchrotron radiation facilities and thank Dr. E. Di Cola and T. Narayanan for their kind and competent assistance in using beamline ID02. We thank Professor R. Khare for use of the rheometer and Professor Balasubramanian for a preprint of his paper (ref 45).

’ REFERENCES (1) Freemantle, M. Chem. Eng. News 1998, 76 (13), 32–37. (2) Plechkova, N. V.; Seddon, K. R. In Methods and Reagents for Green Chemistry: An Introduction; Tundo, P., Perosa, A., Zecchini, F., Eds.; John Wiley & Sons, Inc.: New York, 2007; pp 105
130. (3) Cull, S. G.; Holbrey, J. D.; Vargas-Mora, V.; Seddon, K. R.; Lye, G. J. Biotechnol. Bioeng. 2000, 69, 227–233. (4) Wang, P.; Zakeeruddin, S. M.; Ivan, E.; Gratzel, M. Chem. Commun. 2002, 2972–2973. (5) Gu, W.; Chen, H.; Tung, Y.-C.; Meiners, J.-C.; Takayama, S. Appl. Phys. Lett. 2007, 90, No. 033505. (6) Ionic Liquids in Synthesis; Wasserscheid, P., Welton, T., Eds.; Wiley
VCH: Weinheim, Germany, 2007. (7) Sanes, J.; Carrion, F.-J.; Bermudez, M.-D.; Martinez-Nicolas, G. Tribol. Lett. 2006, 21, 121–133. (8) Le Bideau, J.; Gaveau, P.; Bellayer, S.; Neouze, M.-A.; Vioux, A. Phys. Chem. Chem. Phys. 2007, 9, 5419–5422. (9) Seki, S.; Ohno, H.; Kobayashi, Y.; Miyashiro, H.; Usamim, A.; Mita, Y.; Tokuda, H.; Watanabe, M.; Hayamizu, K.; Tsuzuki, S.; Hattori, M.; Terada, N. J. Electrochem. Soc. 2007, 154, A173–A177. (10) MacFarlane, D. R.; Seddon, K. R. Aust. J. Chem. 2007, 60, 3–5. (11) van Rantwijk, F.; Shelton, R. A. Chem. Rev. 2007, 107 2757–2785. (12) Weingartner, H. Angew. Chem., Int. Ed. 2008, 47, 654–670. (13) Plechkova, N. V.; Seddon, K. R. Chem. Soc. Rev. 2008, 37 123–150. (14) Seddon, K. R. In The International George Papatheodorou Symposium: Proceedings; Boghosian, S., Dracopoulos, V., Kontoyannis, C. G., Voyiatzis, G. A., Eds., Institute of Chemical Engineering and High Temperature Chemical Processes: Patras, Greece, 1999; pp 131
135. (15) Noda, A.; Hayamizu, K.; Watanabe, M. J. Phys. Chem. B 2001, 105, 4603–4610. (16) Tokuda, H.; Hayamizu, K.; Ishii, K.; Susan, M. A. B.; Watanabe, M. J. Phys. Chem. B 2004, 108, 16593–16600. (17) Tokuda, H.; Hayamizu, K.; Ishii, K.; Susan, M. A. B.; Watanabe, M. J. Phys. Chem. B 2005, 109, 6103–6110. (18) Huddleston, J. G.; Visser, A. E.; Reichert, W. M.; Willauer, H. D.; Broker, G. A.; Rogers, R. D. Green Chem. 2001, 3, 156–164. (19) Seddon, K. R.; Stark, S.; Torres, M.-J. In Clean Solvents: Alternative Media for Chemical Reactions and Processing; Abraham, M. A., Moens, L., Eds.; ACS Symposium Series; American Chemical Society: Washington, DC, 2002; Vol. 819, pp 34
49. (20) Rebelo, L. P. N.; Najdanovic-Visak, V.; de Azevedo, R. G.; Esperanca, J. M. S. S.; da Ponte, M. N.; Guedes, H. J. R.; Visak, Z. P.; de Sousa, H. C.; Szydlowski, J.; Lopes, J. N. C.; Cordeiro, T. C. In Ionic Liquids IIIA: Fundamentals, Progress, Challenges, and Opportunities: Properties and Structures; Rogers, R. D., Seddon, K. R., Eds.; ACS Symposium Series; American Chemical Society: Washington, DC, 2005; Vol. 901, pp 270
291. (21) Gu, Z. Y.; Brennecke, J. F. J. Chem. Eng. Data 2002, 47 339–345. (22) Fredlake, C. P.; Crosthwaite, J. M.; Hert, D. G.; Aki, S. N. V. K.; Brennecke, J. F. J. Chem. Eng. Data 2004, 49, 954–964.

ARTICLE

(23) Jin, H.; O’Hare, B.; Dong, J.; Arzhantsev, S.; Baker, G. A.; Wishart, J. F.; Benesi, A. J.; Maroncelli, M. J. Phys. Chem. B 2008, 112, 81–92. (24) Triolo, A.; Russina, O.; Bleif, H.-J.; Di Cola, E. J. Phys. Chem. B 2007, 111, 4641–4644. (25) Triolo, A.; Russina, O.; Fazio, B.; Triolo, R.; Di Cola, E. Chem. Phys. Lett. 2008, 457, 362–365. (26) Russina, O.; Triolo, A.; Xiao, D.; Hines, L. G., Jr.; Quitevis, E. L.; Gontrani, L.; Caminiti, R.; Plechkova, N. V.; Seddon, K. R. J. Phys.: Condens. Matter 2009, 21, No. 424121. (27) Urahata, S. M.; Ribeiro, M. C. C. J. Chem. Phys. 2004, 120, 1855–1863. (28) Wang, Y.; Voth, G. A. J. Am. Chem. Soc. 2005, 127 12192–12193. (29) Wang, Y.; Voth, G. A. J. Phys. Chem. B 2006, 110, 18601–18608. (30) Lopes, J. N. A. C.; Padua, A. A. H. J. Phys. Chem. B 2006, 110, 3330–3335. (31) Jiang, W.; Wang, Y.; Voth, G. A. J. Phys. Chem. B 2007, 111, 4812–4818. (32) Wang, Y.; Yan, T.; Voth, G. A. Acc. Chem. Res. 2007, 40, 1193–1199. (33) Triolo, A.; Russina, O.; Fazio, B.; Appetecchi, G. A.; Carewska, M.; Passerini, S. J. Chem. Phys. 2009, 130, No. 164521. (34) Gontrani, L.; Russina, O.; Celso, F. L.; Caminiti, R.; Annat, G.; Triolo, A. J. Phys. Chem. B 2009, 113, 9235–9240. (35) Mizuhata, M.; Maekawa, M.; Deki, S. ECS Trans. 2007, 3, 89. (36) Atkin, R.; Warr, G. G. J. Phys. Chem. B 2008, 112, 4164–4166. (37) Shigeto, S.; Hamaguchi, H.-o. Chem. Phys. Lett. 2006, 427, 329–332. (38) Iwata, K.; Okajima, H.; Saha, S.; Hamaguchi, H.-O. Acc. Chem. Res. 2007, 40, 1174–1181. (39) Xiao, D.; Rajian, J. R.; Li, S.; Bartsch, R. A.; Quitevis, E. L. J. Phys. Chem. B 2006, 110, 16174–16178. (40) Xiao, D.; Rajian, J. R.; Cady, A.; Li, S.; Bartsch, R. A.; Quitevis, E. L. J. Phys. Chem. B 2007, 111, 4669–4677. (41) Xiao, D.; Rajian, J. R.; Hines, L. G., Jr.; Li, S.; Bartsch, R. A.; Quitevis, E. L. J. Phys. Chem. B 2008, 112, 13316–13325. (42) Dzyuba, S. V.; Bartsch, R. A. Chem. Commun. 2001, 1466–1467. (43) Wang, X. J.; Heinemann, F. W.; Yang, M.; Mechler, B. U.; Fekete, M.; Mudring, A.-V.; Wasserscheid, P.; Meyer, K. Chem. Commun. 2009, 7405–7407. (44) Xiao, D.; Hines, L. G., Jr.; Li, S.; Bartsch, R. A.; Quitevis, E. L.; Russina, O.; Triolo, A. J. Phys. Chem. B 2009, 113, 6426–6433. (45) Raju, R.; Balasubramanian, S. J. Phys. Chem. B 2010, 114, 6455– 6463. (46) Dzyuba, S. V.; Bartsch, R. A. ChemPhysChem 2002, 3, 161–166. (47) Narayanan, T.; Diat, O.; Bosecke, P. Nucl. Instrum. Methods A 2001, 467, 1005–1009. (48) Plazek, D. J.; Frund, Z. N. J. Polym. Sci. B: Polym. Phys. 1990, 28, 431. (49) DeBolt, M. A.; Easteal, A. J.; Macedo, P. B.; Moynihan, C. T. J. Am. Ceram. Soc. 1976, 59, 16–21. (50) Badrinarayanan, P.; Zheng, W.; Li, Q.; Simon, S. L. J. Non-Cryst. Solids 2007, 353, 2603–2612. (51) Moynihan, C. T.; Easteal, A. J.; DeBolt, M. A. J. Am. Ceram. Soc. 1976, 59, 12–16. (52) Chirico, R. D.; Diky, V.; Magee, J. W.; Frenkel, M.; Marsh, K. N. Pure Appl. Chem. 2009, 81, 791–828. (53) Marsh, K. N.; Brennecke, J. F.; Chirico, R. D.; Frenkel, M.; Heintz, A.; Magee, J. W.; Petes, C. J.; Rebelo, L. P. N.; Seddon, K. R. Pure Appl. Chem. 2009, 81, 781–790. (54) Jacquemin, J.; Husson, P.; Padua, A. A. H.; Majer, V. Green Chem. 2006, 8, 172–180. (55) Widegren, J. A.; Laesecke, A.; Magee, J. W. Chem. Commun. 2005, 1610–1612. (56) Bhargava, B. L.; DeVane, R.; Klein, M. L.; Balasubramanian, S. Soft Matter 2007, 3, 1395–1400. (57) Hardacre, C.; Holbrey, J. D.; Mullan, C. L.; Youngs, T. G. A.; Bowron, D. T. J. Chem. Phys. 2010, 133, No. 074510. 6583

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584

The Journal of Physical Chemistry B

ARTICLE

(58) Greaves, T. L.; Kennedy, D. F.; Mudie, S. F.; Drummond, C. J. J. Phys. Chem. B 2010, 114, 10022–10031. (59) Russina, O.; Beiner, M.; Pappas, C.; Russina, M.; Arrighi, V.; Unruh, T.; Mullan, C. L.; Hardacre, C.; Triolo, A. J. Phys. Chem. B 2009, 113, 8469–8474. (60) Fujii, K.; Soejima, Y.; Kyoshoin, Y.; Fukuda, S.; Umebayashi, Y.; Yamaguchi, M.; Ishiguro, S.; Takamuku, T. J. Phys. Chem. B 2008, 112 4329–4336. (61) Deetlefs, M.; Hardacre, C.; Nieuwenhuyzen, M.; Padua, A. A. H.; Sheppard, O.; Soper, A. K. J. Phys. Chem. B 2006, 110 12055–12061. (62) Kanzaki, R.; Mitsugi, T.; Fukuda, S.; Fujii, K.; Takeuchi, M.; Soejima, Y.; Takamuku, T.; Yamaguchi, T.; Umebayashi, Y.; Ishiguro, S. J. Mol. Liq. 2009, 147, 77–82. (63) Aoun, B.; Goldbach, A.; Kohara, S.; Wax, J.-F.; Gonzalez, M. A.; Saboungi, M.-L. J. Phys. Chem. B 2010, 114, 12623–12628. (64) Hanke, C. G.; Price, S. L.; Lynden-Bell, R. M. Mol. Phys. 2001, 99, 801–809. (65) Hardacre, C.; Holbrey, J. D.; McMath, S. E. J.; Bowron, D. T.; Soper, A. K. J. Chem. Phys. 2003, 118, 273–278. (66) Lopes, J. N. A. C.; Gomes, M. F. C.; Padua, A. A. H. J. Phys. Chem. B 2006, 110, 16816–16818. (67) Triolo, A.; Mandanici, A.; Russina, O.; Rodriguez-Mora, V.; Cutroni, M.; Hardacre, C.; Nieuwenhuyzen, M.; Bleif, H.-J.; Keller, L.; Ramos, M. A. J. Phys. Chem. B 2006, 110, 21357–21364. (68) Rivera, A.; Brodin, A.; Pugachev, A.; Rossler, E. J. Chem. Phys. 2007, 126, No. 114503. (69) Cowie, J. M. G. Eur. Polym. J. 1975, 11, 297–300. (70) Zhang, J.; Liu, G.; Jonas, J. J. Phys. Chem. 1992, 96, 3478–3480. (71) Morineau, D.; Xia, Y.; Alba-Simionesco, C. J. Chem. Phys. 2002, 117, 8966–8972. (72) Alcoutlabi, M.; McKenna, G. B. J. Phys.: Condens. Matter 2005, 17, R461–R524. (73) Xu, W.; Cooper, E. I.; Angell, C. A. J. Phys. Chem. B 2003, 107 6170–6178. (74) Vogel, H. Phys. Z. 1921, 22, 645–646. (75) Fulcher, G. S. J. Am. Ceram. Soc. 1925, 8, 339–355. (76) Tammann, G.; Hesse, G. Z. Anorg. Allg. Chem. 1926, 156, 245–257. (77) Stickel, F.; Fischer, E. W.; Richert, R. J. Chem. Phys. 1996, 104, 2043–2055. (78) Del Popolo, M. G.; Voth, G. A. J. Phys. Chem. B 2004, 108 1744–1752. (79) Margulis, C. J. Mol. Phys. 2004, 102, 829–838. (80) Annapureddy, H. V. R.; Kashyap, H. K.; De Biase, P. M.; Margulis, C. J. J. Phys. Chem. B 2010, 114, 16838–16846. (81) Bodo, E.; Gontrani, L.; Caminiti, R.; Plechkova, N. V.; Seddon, K. R.; Triolo, A. J. Phys. Chem. B 2010, 114, 16398–16407. (82) Lopez-Martin, I.; Burello, E.; Davey, P. N.; Seddon, K. R.; Rothenberg, G. ChemPhysChem 2007, 8, 690–695. (83) Seddon, K. R. J. Chem. Technol. Biotechnol. 1997, 68, 351–356. (84) Doolittle, A. K. J. Appl. Phys. 1951, 22, 1471–1475. (85) Wang, Y.; Jiang, W.; Voth, G. A. In Ionic Liquids IV: Not Just Solvents Anymore; Brennecke, J. F., Rogers, R. D., Seddon, K. R., Eds.; ACS Symposium Series; American Chemical Society: Washington, DC, 2007; Vol. 975, pp 272
307. (86) Dzyuba and Bartsch46 previously reported the viscosities for the CN
1C1 series (N = 2
11). However an even
odd effect was not observed. Also, the viscosities reported for this series in this previous study were lower than the viscosities measured in the present study, For example, for N = 3, 5, and 6 at T = 298.15 K, η = 25, 44, and 50 mPa 3 s (Dzyuba and Bartsch) versus η = 32.0, 49.5, and 61.4 mPa 3 s (this work). We attribute these discrepancies to the difference in the way the measurements were performed. In the present study, the measurements were performed by use of a rheometer with a nitrogen purge, whereas in the previous study of Dzyuba and Bartsch, measurements were performed by use of an Ostwald viscometer under ambient conditions.

6584

dx.doi.org/10.1021/jp1115614 |J. Phys. Chem. B 2011, 115, 6572–6584