Fluid Phase Equilibria 386 (2015) 65–74

Measurement and correlation for the thermophysical properties ofnovel pyrrolidonium ionic liquids: Effect of temperature and alkylchain length on anion

Somenath Panda, Ramesh L. Gardas *,1

Department of Chemistry, Indian Institute of Technology Madras, Chennai 600 036, India

A R T I C L E I N F O

Article history:Received 15 May 2014Received in revised form 18 November 2014Accepted 26 November 2014Available online 28 November 2014

Keywords:Ionic liquidThermophysical propertyVogel–Tammann–Fulcher equationStructure–property correlationGardas and Coutinho model

A B S T R A C T

In this work, a new series of low viscous protic ionic liquids based on N-methylpyrrolidone cation andcarboxylate anions with different chain length has been synthesized and characterized. Density, speed ofsound and viscosity have been measured for these ionic liquids in the temperature range from 293.15 to343.15 K. The density data have been fitted to linear equation as well as correlated with groupcontribution method developed by Gardas and Coutinho. From the density correlation, molecular volumedata for cation and anion constituents of studied ionic liquids have been proposed. By using experimentaldensity and speed of sound data, the isentropic compressibility, coefficient of thermal expansion,standard entropy, intermolecular free length and lattice energy have been calculated for studied ionicliquids. Viscosity data were correlated with the Vogel–Tammann–Fulcher equation. To understand thenature of the ionic liquids and structure–property relationship, the effect of anion alkyl chain length hasbeen studied and correlated. Ionic liquids studied in this work are comparatively cheaper in price and canbe promising in bio-macromolecule solubilization and other commercial fields.

ã 2014 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Fluid Phase Equilibria

journal homepage: www.elsevier .com/ locate /fluid

1. Introduction

Since the time of its introduction to the chemical community,ionic liquid (IL) has made a distinct place by its unique propertiesand promising applicability. The incomparable physical character-istics as low volatility, large liquidus range, and good electricalconductivity have opened unlimited opportunities to the scientistsfrom all branches for further exploration. The expedition hasshown ILs to have amazing applicability in the fields as solventmedium [1,2], catalysis [3], biomass processing [4], petroleumrefining [5], electrochemistry [6], as magnetic fluid [7] andenergetic materials [8].

ILs are generally composed of a bulky organic cationand relatively small anion, so it is theoretically possible toget innumerous number of combinations. As it is practicallyimpossible to study all of them, an accurate and systematicinvestigation of physicochemical properties of selected systemsis necessary, which can help to build structure propertycorrelation and enhance predictive modeling [9]. These studiesare useful for both fundamental and applied research as theprimary screening to find a suitable material for an application

* Corresponding author. Tel.: +91 44 2257 4248; fax: +91 44 2257 4202.E-mail address: gardas@iitm.ac.in (R.L. Gardas).

1 http://www.iitm.ac.in/info/fac/gardas.

0378-3812/$ – see front matter ã 2014 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.fluid.2014.11.024

is based on its thermophysical properties. Furthermore, tounderstand and relate the macroscopic and microscopic behaviorof ILs, a detailed knowledge of the thermophysical properties is ofutmost importance.

Over the past few decades, to fulfill the work specific demand oracademic quest, many new ILs have been discovered and studied.Though the number of publications on ILs is enormous, most of theresearches are centered on imidazolium, pyridinium, pyrrolidi-nium, piperidinium and phosphonium [10–13] with some recentworks on ammonium [14] and triazolium [15] based ILs. Theprimary factor for an extensive research and application is the highcost of the ILs; therefore search for cheaper species is alwayseconomically preferable. The high viscosity of the ILs is anotherburden which narrows down the field of applicability in theindustrial processes. Water solubility and toxicity are among othercriteria confining the applicability to bio-related fields. In otherwords, there is always a scope to study new ILs, as the presentlyavailable data or correlations are not exhaustive anyways.

In this retrospect, ionic liquids based on N-methylpyrrolidone(NMP) can present a blend of new properties as well as lowtoxicity, less viscosity and cheapness. Carboxylates anions, on theother hand, can fulfill both the research demand of studyingnature of ILs as well as can be applicable in cellulose and lignintreatment [16,17]. To the best of our knowledge, there are onlyfew open literature available on NMP based ILs with inorganicanions [18–20], but no literature on ILs with organic acid anions.

List of symbols

N Avogadro numbera Coefficient of thermal expansionr DensityLf Inter molecular free lengthbs Isentropic compressibilityK Jacobson’s constantUPOT Lattice potential energiesM Molecular weightVm Molecular volumeVa Molecular volume of the anionVc Molecular volume of the cationP Pressureu Speed of soundS0 Standard EntropyTd Thermal decomposition temperatureh Viscosity

66 S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74

Furthermore, none of the literature reports the thermophysicalproperties of NMP based ILs. Therefore, in the present work, wehave synthesized a series of NMP carboxylate ILs as listed inScheme 1 in which only the anion carbon chain length varies.Densities, ultrasonic sound velocities and viscosities of this serieshave been experimentally measured as a function of temperature.Some important thermodynamic properties such as molecularvolume, isobaric expansion coefficient, isoentropic compressibili-ty, standard molar entropy, lattice energy have been calculated forthese ILs. The present work attempts to understand the molecularinteraction as well as the effect of alkyl chain length in the anion onthe physicochemical properties of the ILs. These data have apotential to add in developing structure-property correlation andmolecular modeling. Furthermore, these carboxylate ILs can beapplicable in various fields such as ammonia based refrigerationsystems and others.

2. Experimental

2.1. Synthesis of ILs

The source and purity of all the chemicals used in thisexperiment are summarized in Table 1. The detailed syntheticprocedure has been described elsewhere [21]. The generalprocedure involves the exothermic neutralization of the base withequimolar amount of different acids as shown in Scheme 1. Thesynthesis of N-methylpyrrolidonium acetate ionic liquid is givenhere. Similar procedure has been followed for the synthesis ofother ILs. 9.9 g (0.1 mol) of N-methylpyrrolidone (NMP) was takenin a 250 mL double necked round bottomed flask fitted with areflux condenser, a thermometer and a pressure equalizing

Scheme 1. General procedure for the synthesis

dropping funnel and kept in an ice bath. 6.0 g (0.1 mol) aceticacid was added drop wise from the dropping funnel with vigorousstirring by a magnetic stirrer. During addition, the temperature wasmaintained at �278 K to dissipate the heat generated in thereaction. After complete addition, the flask was taken out from theice bath and stirring was continued in room temperature for 12 h.The flask was then connected to high vacuum for 24 h with heatingat 313.15 K to remove any unreacted starting materials as well asmoisture. The structures of the synthesized ILs along with theirabbreviations are summarized in Table 2.

2.2. Characterization

The 1H and 13C NMR was recorded in a Bruker Avance 500 MHzspectrometer with CDCl3 as solvent and TMS as internal standard.The IR data were recorded in a JASCO FT/IR-4100 spectrometerusing NaCl disk. The 13C and IR data has been given in supportinginformation S1. 1H-NMR (CDCl3, d ppm): [NMP][For] d = 2.32(m,4H) 2.93(s,3H) 3.34(t,2H) 6–8(broad NH+ and HCOO�); [NMP][Ace] d = 2.04(q,2H) 2.44(t, 2H) 2.49(s,3H) 2.86(s,3H) 3.4(t,2H) 6.4(broad NH+); [NMP][Pro] d = 1.10(t,3H) 2.01(m,2H) 2.29(m,2H) 2.37(t,2H) 2.83(s,3H) 3.42(t,2H) 6.1(broad NH+); [NMP][But] d = 0.94(t,3H) 1.64(q,2H) 2.00(m,2H) 2.28(m,2H) 2.38(m,2H) 2.83(s,3H)3.38(t,2H) no broadening; [NMP][Pen] d = 0.88 (t,3H) 1.32(m,2H)1.56(m,2H) 1.99(m,2H) 2.29(m,2H) 2.38(q,2H) 2.82(s,3H) 3.36(t,2H) (no broad peak for NH+); [NMP][Hex] d = 0.89(t,3H) 1.31(m,4H) 1.63(q,2H) 2.01(m,2H) 2.35(m,4H) 2.83(s,3H) 3.38(t,2H)(no broad peak for NH+)

2.3. Measurement technique

The density and speed of sound of the ionic liquids weremeasured with an Anton Paar (DSA 5000M) in the temperaturerange of 293.15–343.15 K and at atmospheric pressure. For densitymeasurement, the instrument uses the vibrating U-shaped cellkept inside a cavity of a metallic block with peltier devices allowingprecise temperature control and stability. The sound velocitymeasurement cell is connected to one end of the U tube so that thesame liquid can pass to it without sample loss and the samemetallic block can control the temperature. The viscosity measure-ments were done by Anton Paar Lovis 2000ME instrument. It usesthe rolling ball in a capillary method and temperature is keptconstant through a built-in peltier device with an accuracy of0.02 K. Approximately 4 mL of sample was inserted to the inlet portcarefully (in bubble free condition) of density meter which also fillsthe sound velocity cell and viscometer capillary respectively. Priorto measurements, the internal calibration was verified by themeasurements with double distilled water and atmospheric air.Similar to our earlier work [21], the instrument was also calibratedwith reference ionic liquid namely, 1-hexyl-3-methylimidazoliumbis(trifluoromethylsulfonyl) imide, [C6Mim][Tf2N] for density,speed of sound and viscosity measurements. The data reportedhere is the average value of three consecutive measurements. The

of N-methylpyrrolidone based ionic liquids.

Table 1Source and purity of chemicals used in this work.

Chemicals name CAS number Source Mass fraction purity (%) Purification method

N-methylpyrrolidone (NMP) 872-50-4 Sigma–Aldrich >99 Molecular sievesFormic acid 64-18-6 Sigma–Aldrich 99.9 NoneAcetic acid 64-19-7 Merck 99.6 NonePropionic acid 79-09-4 Sigma–Aldrich 99 NoneButanoic acid 107-92-6 Sigma–Aldrich 99 NonePentanoic acid 109-52-4 Sigma–Aldrich >99 NoneHexanoic acid 142-62-1 Sigma–Aldrich 99 None

S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74 67

standard uncertainties in density, speed of sound, viscositytemperature and pressure are u(r) = 0.005 kg m�3; u(u) = 0.5 m s�1;u(h) = 0.005 mPa s; u(T) = 0.01 K and u(P) = 0.01 kPa, respectively.

The water contents of the ILs were measured with Analab(Micro Aqua Cal 100) Karl Fischer Titrator and presented in Table 3.It works on the principle of conductometric titration, and candetect water from less than 10 ppm to 100% with its dual platinumelectrode. The instrument was calibrated with milipore wateraccording to the supplier’s manual.

Thermogravimetric analysis (TGA) was done on the TGAinstrument (TA instruments Hi-Res. TGA Q500). It measures thechange in sample weight kept on an Aluminium pan in nitrogen

Table 2Name and abbreviation of the synthesized ionic liquids.

Cation Anion

atmosphere with a precession of �0.01% and suitable heating rate.The heating rates are controlled precisely by two thermocoupleskept adjacent to the sample. Prior to the measurement, instrumentwas calibrated with Nickel metal and in this experiment heatingrate was set as 10 K min�1.

3. Results and discussion

3.1. Density

The experimental density data measured at atmosphericpressure and over the temperature range of 293.15–343.15 K for

Name Abbreviation

N-methyl-2-pyrrolidonium formate [NMP][For]

N-methyl-2-pyrrolidonium acetate [NMP][Ace]

N-methyl-2-pyrrolidonium propionate [NMP][Pro]

N-methyl-2-pyrrolidonium butanoate [NMP][But]

N-methyl-2-pyrrolidonium pentanoate [NMP][Pen]

N-methyl-2-pyrrolidonium hexanoate [NMP][Hex]

Fig. 1. Temperature dependence of density for the studied ILs. & [NMP][For]; ~

[NMP][Ace]; * [NMP][Pro]; ^ [NMP][But]; � [NMP][Pen]; * [NMP][Hex].

Table 3Water content, decomposition temperature (Td), standard entropy (S0) and lattice energy (UPOT) of the investigated ILs.

IL Water content/ppm Molar mass Td/K aS0/J K�1mol�1 aUPOT/kJ mol�1

[NMP][For] 2743 145.16 359 309.8 489.57[NMP][Ace] 1974 159.18 371 344.4 474.89[NMP][Pro] 2873 173.21 390 378.3 462.49[NMP][But] 2049 187.24 429 414.7 450.81[NMP][Pen] 2626 201.26 433 450.2 440.74[NMP][Hex] 2137 215.29 421 484.7 432.01

a At T = 298.15 K.

68 S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74

the series are presented in Table 4. Density data as a function oftemperature are displayed in Fig. 1. As expected, the densitydecreases linearly with increase in temperature. It is also clearlyobserved that the density decreases as the anion chain increasesdepicting the role of anions on the molecular geometry. Onepossible reason may be that with the increase in chain length onthe anion part, the nonpolar regions increases which takes up morespace and results in a lower overall density [22]. This trend is quitesimilar to earlier report for the ILs with different anionic alkylchain length. [23,24]

The experimental density data were fitted with the linearequation in the given temperature range and 0.1 MPa pressure.

r ¼ A þ BT (1)

where r is the density, T is the temperature, A and B are arbitraryparameters. These fitting parameters along with their averageabsolute relative deviation (ARD) obtained by least square analysisare listed in Table 5. The ARD were calculated by the followingequation.

ARD ¼ 1n

X jrexp � rcaljrexp

!100 (2)

From the fitting, it is clear that the data are in good correlation withthe aforesaid equation. The ARD value ranges from (0.189 to0.310) � 10�3 which shows that the studied ILs density changesalmost linearly with temperature.

These experimental density data were further used to calculatethe molecular volume (Vm) of each ionic liquids by the followingequation.

Vm ¼ MNr

(3)

where M is the molar mass, N is the Avogadro’s number, andr is the density of the ILs. The calculated Vm values of the ILsare listed in Table 6. It can be observed that, for the given cation

Table 4Densities of studied ILs at T = (293.15 to 343.15) K and at atmospheric pressure.

Density/kg m�3

T/K [NMP][For] [NMP][Ace] [NMP][Pro]

293.15 1076.38 1050.71 1032.46

298.15 1071.66 1046.06 1027.86

303.15 1066.93 1041.42 1023.32

308.15 1062.20 1036.77 1018.74

313.15 1057.45 1032.12 1014.16

318.15 1052.71 1027.46 1009.56

323.15 1047.96 1022.79 1004.97

328.15 1043.20 1018.12 1000.36

333.15 1038.43 1013.44 995.75

338.15 1033.66 1008.75 991.12

343.15 1028.88 1004.05 986.49

Standard uncertainty u is u(r) = 0.005 kg m�3,u(T) = 0.01 K and u(P) = 0.01 kPa.

N-methyl-2-pyrrolidonium, [NMPH]+, the molecular volumesof the ILs increase with the effective anion size in the order:[HCOO]�< [CH3COO]�< [CH3CH2COO]�< [CH3(CH2)2COO]�<

[CH3(CH2)3COO]�< [CH3(CH2)4COO]�. Furthermore, at any giventemperature, a linear relationship was observed betweenmolecular volume of the ILs and number of Carbon atoms inalkyl chain of the carboxylate anions. From the figure, it is seenthat such a linear plot at 298.15 K have a slope of 0.028 nm3,indicating that a regular addition per —CH2— group in the anionalkyl chain of the ILs contributes to an increase of 0.028 nm3

in their molecular volumes. This value is in good agreement withthe earlier reports of —CH2— contribution of 0.028 nm3 from

[NMP][But] [NMP][Pen] [NMP][Hex]

1010.56 994.31 983.011006.11 990.01 978.811001.65 985.70 974.60997.18 981.38 970.39992.71 977.06 966.18988.24 972.73 961.96983.74 968.39 957.74979.25 964.05 953.51974.76 959.71 949.28970.26 955.36 945.04965.75 951.01 940.79

Table 7The absolute relative deviation (ARD) and maximum relativedeviaton (MD) in estimated density using Gardas and Coutinhomodel (Eq. (5)).

IL ARD MD

[NMP][For] 0.35 0.69[NMP][Ace] 0.36 0.71[NMP][Pro] 0.44 0.95[NMP][But] 0.35 0.78[NMP][Pen] 0.38 0.90[NMP][Hex] 0.33 0.78

Table 5Coefficients of Eq. (1) and absolute relative deviation (ARD).

IL A � 10�3/kg m�3 B � 10/kg m�3 K�1 ARD � 103

[NMP][For] 1.3549 �9.50 0.230[NMP][Ace] 1.3242 �9.33 0.241[NMP][Pro] 1.3019 �9.19 0.310[NMP][But] 1.2733 �8.96 0.218[NMP][Pen] 1.2483 �8.66 0.214[NMP][Hex] 1.2306 �8.44 0.189

S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74 69

Caprolactam ILs, 0.028 nm3 from n-alcohols, 0.027 nm3 fromn-amines, and 0.027 nm3 from n-paraffins [24,25].

As suggested by Gardas et al. [26] a simple model proposedby Esperança et al. [27] can be used to predict the molecularvolume of an IL in which total volume is considered as the sum ofthe effective molecular volume occupied by individual cation andanion.

Vm ¼ V�c þ V�

a (4)

According to this approach, if the molecular volume of an anionof a particular IL is known, it is possible to determine the molecularvolume of the cation or vice-versa. The ionic volumes of cationsand anions were estimated from the density data fitting to theGardas and Coutinho model [28] by taking the available ionicvolume for acetate ion (85.5 Å3) as Ref. [27] and the contribution byeach —CH2— group as 28 Å3, reported by Ye and Shreeve [29]. Theestimated data is in well agreement with the reported data forformate anion (56 Å3) [30], though the other ionic volume datacould not be compared due to the lack of such literature reports.The effective molecular volume for the [NMPH]+ cation wasestimated as 169 Å3. The estimated Vm along with their percentrelative deviation from the experimental value is given in Table 6. Itis observed that our estimated Vm is in good accordance with theexperimentally determined value at 298.15 K.

Gardas and Coutinho proposed a model [28] to relate thedensity of an ionic liquid with its molecular volume in a wide rangeof temperatures (273.15–393.15 K) and pressures (0.10–100 MPa).The model was used to calculate density by minimizingthe objective function jrexp� rcalj, through optimization of themolecular volume (Vm). The model equation, giving both theestimated value of Vm and density was represented as:

rcal ¼M

NVmða þ bT þ cPÞ (5)

where rcal is the density in kg m�3, M is the molar mass inkg mol�1, N is the Avogadro’s number, Vm is the estimatedmolecular volume in nm3, T is the temperature in K and P isthe pressure in MPa. The coefficient data for a, b and c were takenfrom literature, obtained by optimizing about 800 experimental

Table 6Effective molecular volume of anion ðV�

aÞ and cation ðV�cÞ and estimated

molecular volumes (Vm) along with relative deviation (RD) of the investigatedILs at 298.15 K

IL V�c/Å

3 V�a/Å

3 Vm(estd)/Å3 bVm(expt)/Å3 % Vm RD

[NMP][For] 169a 57.5a 226.5 225 0.71[NMP][Ace] 85.5 [29] 254.5 253 0.73[NMP][Pro] 113.5a 282.5 280 0.97[NMP][But] 141.5a 310.5 309 0.49[NMP][Pen] 169.5a 338.5 338 0.29[NMP][Hex] 197.5a 366.5 365 0.36

a Estimated in this work.b Calculated from Eq. (3).

density data points [9,26,28]. The values of a, b and c weregiven as 0.8005 � 0.00023; (6.6520 � 0.0069) � 10�4 K�1 and(�5.919 � 0.024) � 10�4MPa�1 respectively, at 95% confidencelevel. The maximum average deviation and averagerelative deviation from the predicted density value are listed inTable 7. From the correlation plot in Fig. 2 a correlation asrcal = (0.92258 � 0.012) rexp is obtained with R2 = 0.98871 showinga good accordance. These results shows that the equationproposed by Gardas and Coutinho may be applied for newfamilies of ionic liquids as well, with an acceptable confidencelevel.

According to Glasser, as ionic liquids consist of large organiccation, it can be thought of having the average properties of ionicsolid and organic liquids. Therefore, the linear equation relating thestandard entropy and molecular volume (as in ionic solids andorganic liquids) can be applicable to ILs as well. By taking the averagevalue of constants, the equation has been expressed as [31]:

S0ðJ K�1mol�1Þ � 1246:5Vmðnm3Þ þ 29:5 (6)

where, S0 is the standard entropy for the ILs in liquid state atambient temperature (298 K) and Vm is the molecular volume.The standard entropies calculated according to this equation arelisted in Table 3. From Fig. 4, it is noticed that the standard entropyof the ILs increases linearly with increasing number of Carbonatoms in alkyl chain of the carboxylate anions. It suggests that asalkyl chain length in carboxylate anions increases, the molecules inthe ILs become less organized. This linear relationship betweenmolecular volume of the ILs and number of Carbon atoms in alkylchain of their anions (shown in Fig. 4) are in accordance withEq. (6). From the plot, the slope of this straight line is found to be

Fig. 2. Correlation between the experimental and calculated density. & [NMP][For]; ~ [NMP][Ace]; * [NMP][Pro]; [NMP][But]; � [NMP][Pen]; * [NMP][Hex].

Table 8Speed of sound data for the investigated ILs at T = (293.15 to 343.15) K and at atmospheric pressure.

Speed of sound/m s�1

T/K [NMP][For] [NMP][Ace] [NMP][Pro] [NMP][But] [NMP][Pen] [NMP][Hex]

293.15 1495.99 1491.50 1460.35 1435.64 1427.71 1431.40298.15 1478.27 1473.45 1442.56 1418.02 1410.50 1414.15303.15 1460.38 1455.14 1424.49 1400.10 1392.96 1396.69308.15 1442.53 1436.86 1406.43 1382.18 1375.43 1379.28313.15 1424.82 1418.66 1388.44 1364.38 1357.99 1361.97318.15 1407.20 1400.59 1370.59 1346.67 1340.66 1344.76323.15 1389.66 1382.61 1352.80 1328.55 1322.95 1327.63328.15 1372.23 1364.67 1335.10 1310.98 1305.74 1310.59333.15 1354.84 1346.90 1317.49 1293.55 1288.68 1293.65338.15 1337.64 1329.15 1300.00 1276.27 1271.75 1276.84343.15 1320.51 1311.54 1282.64 1259.13 1254.95 1260.16

Standard uncertainty u is u(u) = 0.5 m s�1, u(T) = 0.01 K and u(P) = 0.01 kPa.

70 S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74

35.09 J K�1mol�1, which is in reasonable agreement with theearlier reported value of 34.6 J K�1mol�1 from [Cnmim][Ala],33.9 J K�1mol�1 from [Cnmim][BF4], 35.1 J K�1mol�1 from[Cnmim][NTf2], and 32.2 J K�1mol�1 from an extended group oforganic compounds [24], 34.6 J K�1mol�1 for butyrolactam ILsand 34.5 J K�1mol�1 for caprolactam ILs [21]. It shows that theaddition of each —CH2— group in the alkyl chain of the anionscontributes to an increase of 35.09 J K�1mol�1 in the standardentropy of the carboxylate ILs, which is quite close to predictedvalue from the data of [Cnmim]+ based ILs and some organiccompounds.

Another important method for the estimation of latticepotential energy (UPOT) of simple MX type (1:1) ionic solids,developed by Glasser [31] is given by the equation:

UPOT ¼ grM

� �1=3þ d (7)

where g and d are fitted coefficients of the salts, r is thedensity, and M is the molecular weight. Glasser suggested that,as the density of the condensed melt and solid phase aresimilar, the calculated lattice potential can be applied toboth phases (the constants given as g = 1981.2 kJ mol�1 andd = 103.8 kJ mol�1). The lattice energies of the ILs in liquid phaseand at 298.15 K as calculated from the above equation have beenlisted in Table 3. The calculated potential energies are found to bemuch lower than fused salts as CsI (613 kJ mol�1) and for

Fig. 3. Speed of sound recorded as a function of temperature (lines are guide to eye).& [NMP][For]; ~ [NMP][Ace]; * [NMP][Pro]; ^ [NMP][But]; � [NMP][Pen]; *[NMP][Hex].

imidazolium based ILs with same anion series [24]. This lowlattice energy justifies the liquid states of the formed compound.Furthermore, the lattice energy decreases steadily with theincrease in the carbon chain length which is consistent withthe similar studies done earlier for [Cnmim] Carboxyllic acidsand [Cnmim][Ala] with different anion chain length [24,25]. Thissteady decrease could be attributed to the lower packing efficiencydue to the increased chain length [21].

The density data at different temperatures has been used tocalculate the isobaric expansion coefficient according to thefollowing equation:

a ¼ �1r

drdT

� �P

(8)

For studied six ILs, in the temperature range of 293.15–343.15 K,isobaric expansion coefficient value fall in small ranges from(8.59 to 9.23) � 10�4 K�1 which are similar to the reportedvalues for other ILs such as butyrolactam or caprolactam ILs(7.70 to 8.39) � 10�4 K�1 [21]. This shows that the effect ofcarbon chain length in anion and temperature on the isobaricexpansion coefficient is less significant, though it shows smalldecrease with temperature [9,11]. Also, the a values of studied ILsare much close to that of water but smaller than the conventionalorganic solvents, however those are higher than imidazoliumbased ILs consisting similar anions [24].

Fig. 4. Linear relationship between the molecular volume or the standard entropyof the ILs and the C atom number (nc) in alkyl chain of the carboxylate anions at298.15 K. * Molecular volume � standard entropy.

Table 9Isobaric expansivity (a) and isoentropic compressibility (bs) data for the studied ILs at T = (293.15 to 343.15) K and at atmospheric pressure.

T/K [NMP][For] [NMP][Ace] [NMP][Pro] [NMP][But] [NMP][Pen] [NMP][Hex]

a � 104/K�1 bs/TPa�1 a � 104/K�1 bs/TPa�1 a � 104 /K�1 bs/TPa�1 a � 104/K�1 bs/TPa�1 a � 104/K�1 bs/TPa�1 a � 104 /K�1 bs/TPa�1

293.15 8.83 415.13 8.88 427.83 8.90 454.17 8.87 480.12 8.71 493.40 8.59 496.50298.15 8.86 427.01 8.92 440.32 8.94 467.52 8.91 494.30 8.75 507.71 8.62 510.87303.15 8.90 439.47 8.96 453.49 8.98 481.58 8.95 509.29 8.79 522.85 8.66 525.98308.15 8.94 452.42 9.00 467.18 9.02 496.25 8.99 524.92 8.82 538.62 8.70 541.69313.15 8.98 465.82 9.04 481.41 9.06 511.49 9.03 541.14 8.86 554.99 8.74 557.97318.15 9.02 479.71 9.08 496.15 9.10 527.29 9.07 557.99 8.90 571.96 8.77 574.85323.15 9.07 494.13 9.12 511.46 9.14 543.73 9.11 575.92 8.94 590.01 8.81 592.38328.15 9.11 509.07 9.16 527.41 9.19 560.81 9.15 594.17 8.98 608.40 8.85 610.58333.15 9.15 524.62 9.21 543.92 9.23 578.57 9.19 613.11 9.02 627.44 8.89 629.47338.15 9.19 540.68 9.25 561.14 9.27 597.02 9.23 632.74 9.06 647.18 8.93 649.05343.15 9.23 557.38 9.29 579.00 9.32 616.16 9.28 653.12 9.11 667.67 8.97 669.35

Standard combined uncertainties uc are uc(a) = 0.002 kK�1, uc(bs) = 0.05 TPa�1, u(T) = 0.01 K and u(P) = 0.01 kPa.

S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74 71

3.2. Speed of sound

The speeds of sound data as a function of temperature over therange of 293.15–343.15 K are compiled in Table 8. This variation isfurther plotted in Fig. 3. As a normal trend the speed of sounddecreases with the increase in temperature [32]. Furthermore,there is a common trend of decrement in speed of sound withthe addition of carbon chain on the anionic part except for theN-methyl-2-pyrrolidonium hexanoate [NMP][Hex] which showsanomalous effect. Similar trend of decrease in speed of sound withthe increase in carbon chain is reported in literature, though thereis no universal rule relating the speed of sound to carbon chainlength [32,33].One important thermodynamic parameter, theisentropic compressibility (bs) has been derived with the followingwell-known Newton–Laplace equation and presented in Table 9.

bs ¼1ru2 (9)

where r is the density of the IL and u is the speed of sound.bs gives a measure of the free space available in the molecular

structure. In the present case, it is observed that the bs increaseswith the increase of temperature as well as with the increase incarbon chain length in the anion part. It can be inferred from theprevious measurements that as the chain length increases, thearrangements becomes less efficiently packed having more voidspace hence increasing the compressibility.

In an effort to relate the speed of sound with the intermolecularinteraction in a liquid, Jacobson formulated an empirical relationfor calculating intermolecular free length (Lf) as [34].

Lf ¼ K

ffiffiffiffiffiffiffiffiffi1ru2

s(10)

where K is a temperature dependent variable, called as Jacobsonsconstant and r is the density in g cm�3 and u is the speed of soundin m s�1. The values of Lf at different temperature are given inTable 10. As seen from Table 10, at atmospheric pressure, the

Table 10Intermolecular free length (Lf/Å) data of studied ILs at T = (293.15 to 343.15) K and at at

T/K [NMP][For] [NMP][Ace] [NMP][Pro]

293.15 0.398 0.404 0.416

298.15 0.408 0.415 0.427

303.15 0.418 0.425 0.438

313.15 0.438 0.445 0.459

323.15 0.458 0.466 0.481

Standard combined uncertainties uc are u(T) = 0.01 K and u (P) = 0.01 kPa.

Lf increases with the increase in temperature. As suggested byEyring’s liquid state theory, there is a time lag between the acousticwaves to pass through the intermolecular distance from onemolecule to the next. Hence, with the increase in Lf, the acousticwaves will take longer time to pass through [33], which is supportedby the speed of sound data in Table 8. Furthermore, the rise intemperature and increase in carbon chain length, both leads to theincrease in Lfor otherwise decreasing the ultrasonic speed of sound,validating our assumption. A Similar trend was reported byChhotaray and co-workers, for lactam based ILs [21] and forammonium based ILs [33].

3.3. Viscosity

The measured dynamic viscosity data for the studied ILs atdifferent temperature are reported in Table 11 and the temperaturedependence is shown in Fig. 5. The temperature dependence ofviscosity is observed to be in reverse order of density. It is seen thatthe viscosity increases with the increase in alkyl chain lengthwhich is contrary to the common anticipation. Generally, it can beexpected that, with the increase in side-chain length, the overallcontribution of the strong, associating terms to the interactionsdiminishes, and the contribution of weaker dispersion forcesincreases, leading to decrease in viscosity [35]. The long alkyl chainincreases the van dar Waals interaction between the ions as well asincreases the bulkiness, hence increasing the viscosity [36].

Viscosity of the studied ILs is quite less compared to that ofcommonly studied imidazolium based ILs [36] and much closer tothe common organic solvents. Also, a dramatic decrease inviscosity with temperature is noticed. This change is lessprominent at higher temperature and it becomes nearly equalas the temperature increases further. A similar trend has also beenreported earlier for lactam based ILs and others [21,24,35].

As suggested by Seddon et al., [37] the Vogel–Tammann–Fulcher (VTF) equation can be applied to these class ofnon-Arrhenius type ILs to analyse the variation of transportproperties with respect to temperature.

mospheric pressure.

[NMP][But] [NMP][Pen] [NMP][Hex]

0.428 0.434 0.4350.439 0.445 0.4470.450 0.456 0.4580.472 0.478 0.4800.495 0.501 0.502

Table 11Viscosity data of studied ILs at T = (293.15 to 343.15) K and at atmospheric pressure.

Viscosity/mPa s

T/K [NMP][For] [NMP][Ace] [NMP][Pro] [NMP][But] [NMP][Pen] [NMP][Hex]

293.15 2.451 2.933 3.136 3.996 4.448 5.211298.15 2.217 2.622 2.801 3.526 3.910 4.549303.15 2.020 2.364 2.518 3.136 3.465 3.998308.15 1.850 2.144 2.280 2.809 3.093 3.543313.15 1.705 1.959 2.075 2.531 2.779 3.163318.15 1.580 1.799 1.900 2.295 2.513 2.843323.15 1.472 1.660 1.750 2.093 2.284 2.570328.15 1.377 1.540 1.619 1.918 2.087 2.336333.15 1.295 1.435 1.506 1.771 1.917 2.135338.15 1.223 1.344 1.407 1.639 1.769 1.960343.15 1.159 1.264 1.320 1.527 1.640 1.808

Standard uncertainty u is u(h) = 0.005 mPa s, u(T) = 0.01 K and u(P) = 0.01 kPa.

Table 12Fitting parameters obtained from the viscosity data fitted with the VTF Equation(Eq. (11)) along with standard deviation.

IL h0/mPa s B/K T0/K ARD

[NMP][For] 0.035 986 60.2 0.844[NMP][Ace] 0.031 997 73.5 0.747[NMP][Pro] 0.031 989 78.7 0.689[NMP][But] 0.032 955 93.9 0.590[NMP][Pen] 0.037 907 103.5 0.420[NMP][Hex] 0.042 850 116 0.342

72 S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74

h ¼ h0expB

T � T0

� �(11)

where B, T0 and h0 are adjustable parameters obtained from thefitting of the curve. These parameters along with their standarddeviation are reported in Table 12.

Though the experimental data are in good agreement with theVTF equation, the logarithmic form of Arrhenius equation(generally used to describe common liquid properties) can alsobe used to test the validity of the data for the cations of lowersymmetry [35].

lnh ¼ lnh1 þ EhRT

(12)

where h is the dynamic viscosity, T is the temperature in K, R is theuniversal gas constant, h1 is the viscosity at infinite temperatureand Eh is the activation energy. The h1 and Eh are calculated fromthe intercept and slope of the Arrhenius plot in Fig. 6. Table 13 givesthe calculated values and standard deviations.

As described earlier, though the best correlation of viscositywith temperature is given by the VTF equation for this set of ILs, it isalso fitted to the more general Arrhenius equation to obtain theactivation energy and other quantities. The significance is if theions want to move over the other, an energy barrier must have to beovercome which is described as Eh. This activation energy also

Fig. 5. Plot of experimental viscosity as a function of temperature (lines are guide toeye). & [NMP][For]; ~ [NMP][Ace]; * [NMP][Pro]; ^ [NMP][But]; � [NMP][Pen]; *[NMP][Hex].

signifies the correlation of viscosity with the structure of the ILs. Asthe Eh becomes more, it becomes more difficult for the ions tomove over each other, which might be attributed by other physicalconstraints or entanglements or the stronger interactions amongthe molecules of the liquid [38]. As can be seen from the data here,the activation energies of the ILs ranges from 12.5 to 16.7 kJ mol�1

which are much lower than the imidazolium based ILs and otherlactam based ILs [21]. It may be inferred that the moleculargeometry plays a major role which is further supported by theincrease in activation energy with anion chain length. At infinitetemperature, intermolecular interactions contribution to the

Fig. 6. Plot of the experimental values of viscosity (symbols) as a function oftemperature, linear approximations using the VTF equation (dashed lines) andArrhenius equation (solid lines) for the ILs. & [NMP][For]; ~ [NMP][Ace]; * [NMP][Pro]; ^ [NMP][But]; � [NMP][Pen]; * [NMP][Hex].

Table 13Activation energies (Eh), infinite temperature viscosities (h1) and ARD, obtained byfitting experimental viscosity data with Arrhenius equation (Eq. (12)).

IL En� sa/J mol�1 104h1� sa/mPa s ARD

[NMP][For] 12.5 � 0.2 14.14 � 1.09 1.077[NMP][Ace] 14.1 � 0.2 8.98 � 1.09 1.053[NMP][Pro] 14.5 � 0.2 8.12 � 1.08 1.039[NMP][But] 16.1 � 0.2 5.32 � 1.09 1.063[NMP][Pen] 16.7 � 0.2 4.68 � 1.08 1.029[NMP][Hex] 17.7 � 0.2 3.61 � 1.08 1.102

a Standard deviation.

S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74 73

viscosity becomes insignificant, and the factor determining theviscosity is only the structural geometry of the constituting ions.The beauty of h1 lies in this fact that it enumerates the structuralcontribution of the ions to the dynamic viscosity. The decrease inh1 with increase in carbon chain signifies that the structuralcontributions are less for long chain molecules.

3.4. Thermal study

The thermo gravimetric analysis (TGA) curves in Fig. 7 showsthe change in the weight of the ILs with temperature at a heatingrate of 10 K min�1. The studied ILs shows somewhat similarthermal pattern with a steep shouldering at around 363–423 K. Thedecomposition temperatures (Td) obtained from TGA is listed inTable 3. All the studied ILs shows medium stability towards heatingwith Td ranging from 359 K for N-methyl-2-pyrrolidonium formate[NMP][For] to 433 K for N-methyl-2-pyrrolidonium pentanoate[NMP][Pen] which are comparably low than other imidazoliumbased aprotic ILs [39]. Also, the decomposition temperature seemsto increase with the increase in carbon chain length, except for the[NMP][Hex] which shows lower Td. Primarily, it can be interpretedthat the stability of the protic IL depends on the effectiveness of theproton transfer. These lower Td values could be attributed to theweak N—H bonding due to less efficient proton transfer. Theseobservations are in overall agreement with lactam based ILs havingthe same structural skeleton [21].

Differential Scanning calorimetry (DSC) have not shown anymelting or glass transition temperature up to the lowest limit ofour instrument (193 K). This finding can be interesting in the fieldwhere low temperature liquids are of use, though further thermalcharacterizations will be needed for that.

Fig. 7. Plot of weight per cent against temperature to illustrate the thermaldecomposition of the studied NMP based PILs. a [NMP][For]; b [NMP][Ace]; b [NMP][Pro]; d [NMP][But]; e [NMP][Pen]; f [NMP][Hex].

4. Conclusions

In this work, we have synthesized, purified and characterized anew group of NMP based ionic liquids with carboxylic acid anions.The directly measurable properties as density, ultrasonic speed ofsound and viscosity as a function of temperature have beenmeasured. The density data were fitted to second order polynomialequation while the viscosity data were fitted to VTF equation, bothgiving excellent fit. Most interestingly, the viscosity data for thesenew series of ILs are much lower than common aprotic ionicliquids, thereby opening a number of new possibilities. Thermo-dynamically important parameters giving very useful informationhas also been derived. As per common approximation, the densityand speed of sound shows a detrimental effect while theisoentropic compressibility and coefficient of thermal expansionshows incremental effect with temperature increase. The effect ofcarbon chain of the anionic part on the thermodynamic propertieshas been analyzed and correlated with molecular arrangement atthe preliminary level. From the experimental data, various usefulthermodynamic parameters have been estimated. The thermalstability of this new series has also been analyzed which shows theILs could be useful in low temperature applications. Furthermore,these low viscous and cheap series of ILs will find many new fieldsof applicability.

Acknowledgements

Authors are thankful to Council of Scientific and IndustrialResearch (CSIR), Department of Science and Technology (DST) andIIT Madras for their financial support.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.fluid.2014.11.024.

References

[1] T.L. Greaves, C.J. Drummond, Chem. Rev. 108 (2008) 206–237.[2] C.I. Melo, R.B. Lukasik, M.N. Ponte, E.B. Lukasik, Fluid Phase Equilib. 338 (2013)

209–216.[3] A. Brenno, D. Neto, J. Spencerb, J. Braz. Chem. Soc. 23 (2012) 987–1007.[4] S.R.E. Hassan, F. Mutelet, J.C. Moise, RSC Adv. 3 (2013) 20219–20226.[5] C. Shu, T. Sun, H. Zhang, J. Jia, Z. Lou, Fuel 121 (2014) 72–78.[6] M.C. Buzzeo, R.G. Evans, R.G. Compton, Chem. Phys. Chem. 5 (2004) 1106–1120.[7] R.E. Del Sesto, T.M. McCleskey, A.K. Burrell, G.A. Baker, J.D. Thompson,

B.L. Scott, J.S. Wilkes, P. Williams, Chem. Commun. 4 (2008) 447–449.[8] R.P. Singh, R.D. Verma, D.T. Meshri, J.M. Shreeve, Angew. Chem. Int. Ed. 45

(2006) 3584–3601.[9] R.L. Gardas, M.G. Freire, P.J. Carvalho, I.M. Marrucho, I.M.A. Fonseca,

A.G.M. Ferreira, J.A.P. Coutinho, J. Chem. Eng. Data 52 (2007) 1881–1888.[10] A. Andresovaa, J. Storchb, M. Traïkiac, Z. Wagnera, M. Bendovaa, P. Hussonc,

Fluid Phase Equilib. 371 (2014) 41–49.[11] R.L. Gardas, H.F. Costa, M.G. Freire, P.J. Carvalho, I.M. Marrucho, I.M.A. Fonseca,

A.G.M. Ferreira, J.A.P. Coutinho, J. Chem. Eng. Data 53 (2008) 805–811.[12] M. Soucková, J. Klomfar, J. Pátek, Fluid Phase Equilib. 333 (2012) 38–46.[13] G. Adamova, R.L. Gardas, L.P.N. Rebelo, A.J. Robertsond, K.R. Seddon, Dalton

Trans. 40 (2011) 12750–12764.[14] K. Machanova’, A. Boisset, Z. Sedla’kova’, M. Anouti, M. Bendova’, J. Jacquemin,

J. Chem. Eng. Data 57 (2012) 2227–2235.[15] M.A. Balazadeh, M. Foroutan, Fluid Phase Equilib. 356 (2013) 63–70.[16] B. Zhao, L. Greiner, W. Leitner, RSC Adv. 2 (2012) 2476–2479.[17] A. Brandt, J.P. Hallett, D.J. Leak, R.J. Murphy, T. Welton, Green Chem. 12 (2010)

672–679.[18] D. Deng, Y. Cui, D. Chen, N. Ai, J. Chem. Thermodyn. 57 (2013) 355–359.[19] H. Zhou, J. Yang, L. Ye, H. Lin, Y. Yuan, Green Chem. 12 (2010) 661–665.[20] Y. Zhao, J. Long, F. Deng, X. Liu, Z. Li, C. Xia, J. Peng, Catal. Commun. 10 (2009)

732–736.[21] P.K. Chhotaray, S. Jella, R.L. Gardas, J. Chem. Thermodyn. 74 (2014) 255–262.[22] C. Kolbeck, J. Lehmann, K.R.J. Lovelock, T. Cremer, N. Paape, P. Wasserscheid,

A.P. Froba, F. Maier, H.P. Steinruck, J. Phys. Chem. B 114 (2010) 17025–17036.[23] R.L. Gardas, R. Ge, P. Goodrich, C. Hardacre, A. Hussain, D.W. Rooney, J. Chem.

Eng. Data 55 (2010) 1505–1515.[24] A. Xu, J. Wang, Y. Zhang, Q. Chen, Ind. Eng. Chem. Res. 51 (2012) 3458–3465.

74 S. Panda, R.L. Gardas / Fluid Phase Equilibria 386 (2015) 65–74

[25] D.W. Fang, W. Guan, J. Tong, Z.W. Wang, J.Z. Yang, J. Phys. Chem. B 112 (2008)7499–7505.

[26] R.L. Gardas, M.G. Freire, P.J. Carvalho, I.M. Marrucho, I.M.A. Fonseca,A.G.M. Ferreira, J.A.P. Coutinho, J. Chem. Eng. Data 52 (2007) 80–88.

[27] J.M.S.S. Esperança, H.J.R. Guedes, M. Blesic, L.P.N. Rebelo, J. Chem. Eng. Data 51(2006) 237–242.

[28] R.L. Gardas, J.A.P. Coutinho, Fluid Phase Equilib. 263 (2008) 26–32.[29] C. Ye, J.M. Shreeve, J. Phys. Chem. A 111 (2007) 1456–1461.[30] H. Donald, B. Jenkins, H.K. Roobottom, Inorg. Chem. 38 (1999) 3609–3620.[31] L. Glasser, Thermochim. Acta 421 (2004) 87–93.

[32] K.J. Wu, Q.L. Chen, C.H. He, AIChE J. 60 (2014) 1120–1131.[33] P.K. Chhotaray, R.L. Gardas, J. Chem. Thermodyn. 72 (2014) 117–124.[34] B. Jacobson, J. Chem. Phys. 20 (1952) 927–928.[35] J. Jacquemin, P. Husson, A.A.H. Padua, V. Majer, Green Chem. 8 (2006)

172–180.[36] H.F.D. Almeida, H. Passos, J.A. Lopes-da-Silva, A.M. Fernandes, M.G. Freire,

J.A.P. Coutinho, J. Chem. Eng. Data 57 (2012) 3005–3013.[37] K.R. Seddon, A. Stark, M.J. Torres, Pure Appl. Chem. 72 (2000) 2275–2287.[38] O.O. Okoturo, T.J. VanderNoot, J. Electroanal. Chem. 568 (2004) 167–181.[39] J.S. Wilkes, J. Mol. Catal. A: Chem. 214 (2004) 11–17.