Prévia do material em texto
ORIGINAL PAPER Characterization of the Solvent Properties of Glycerol Using Inverse Gas Chromatography and Solubility Parameters Joel D. Vincent • Keerthi Srinivas • Jerry W. King Received: 13 October 2011 / Revised: 11 February 2012 / Accepted: 5 April 2012 / Published online: 27 April 2012 � AOCS 2012 Abstract The production of glycerol from the synthesis of biodiesel has led to a market surplus of this polyhydric alcohol and additional research to find new applications for this versatile chemical. This study involves the use of inverse gas chromatography (IGC) to characterize the solute–solvent interactions between glycerol and a homologous series of aliphatic alcohols, in which the latter components are at infinite dilution in the glycerol, which is the stationary phase contained in a packed GC column. The IGC experiments were conducted between 51.5 and 111 �C for the n-alcohols ranging from methanol to n-butanol. All of the n-alcohol homologs exhibited positive deviations from Raoult’s law as based on mole fraction activity coefficients values ranging from 1.86 to 14.4. The mea- sured mole fraction activity coefficients of the alcoholic solutes in glycerol showed good agreement with literature values, and in some cases with those predicted using existing theoretical models. The mole fraction activity coefficients increased going from methanol to n-butanol, reflecting the change in the alcohol’s cohesive energy densities relative to that for glycerol. The total solubility parameter of glycerol calculated from IGC data was found to be 34.8 MPa1/2 which is in good agreement with that obtained using Hansen solubility parameter approach (31.6 MPa1/2). This data can be used to characterize the solvent properties of glycerol as well as to provide ther- modynamic data for the removal of the alcoholic solutes from glycerol. Keywords Activity coefficient � Alcohols � Glycerol � Inverse gas chromatography � Solubility parameter Introduction Glycerol (1,2,3-propane triol or glycerine) is a polyhydric alcohol that is produced as a by-product of the transeste- rification of fatty acids present in fats and oils to yield fatty acid methyl esters, i.e. ‘biodiesel fuel’ [1]. Glycerol can also be produced from fats and oils as a by-product of their saponification process to form soaps [2] and subsequent hydrolysis to fatty acids [3]. Studies have indicated that since 1970, the amount of glycerol produced from fats and oils has increased from 40 million tons to around 144 million tons in 2005 [4]. It has been estimated that 1 ton of any biomass material containing 2.5 % (by weight) of fatty acids on transesterification with methanol yields about 946 kg of fatty acid methyl esters, 89 kg of glycerol and 23 kg of unreacted fatty acids [5]. This indicates that there is a yield of 1 kg of crude glycerol for every 9 kg of bio- diesel produced glycerol thus is a potential economic credit in the manufacturing of biodiesel. However with the increasing market for biodiesel as a fuel worldwide, there is an excess of glycerol available in world market which has lowered its economic value substantially from approximately 1,700$/ton in 2001 to about 550$/ton in 2009 [6]. This abundance of glycerol in the market and its rela- tively low cost has opened a number of venues for the application of glycerol in food industries, coatings, lubri- cants, as a humectant in personal care products and in separation processes [6]. Glycerol is also used as raw material in chemical syntheses [7], in medicinal applica- tions [8] and as a substitute for sweeteners [9]. Though, J. D. Vincent � K. Srinivas � J. W. King (&) Department of Chemical Engineering, 3202, Bell Engineering Center, University of Arkansas, Fayetteville, AR 72701, USA e-mail: jwking1@uark.edu 123 J Am Oil Chem Soc (2012) 89:1585–1597 DOI 10.1007/s11746-012-2070-6 glycerol cannot be directly used as a fuel due to the pro- duction of toxic and corrosive polymers of acrolein on burning, their acetylated derivatives have been used as fuel additives [10], and glycerol has been used as a boiler fuel. Despite these potential applications, there is a need to explore other possibilities for glycerol utilization. One pos- sibility is to extend the use of glycerol as a bio-renewable solvent medium. Due to its low toxicity, relatively higher boiling point and water miscibility, glycerol is a potential ‘‘green’’ solvent [11] with the potential of replacing more hazardous chemical solvents. Some studies have been reported using glycerol as a solvent to extract phenol from coal tar hydrocarbons [12] and for accelerating the rate of a number of organic reactions. However, there is limited data on the solubility and miscibility of glycerol, as a function of temperature, with other organic solvents. This study seeks to reconcile this lack of data and explore the relationship between glycerol and a homologous series of alcohols (methanol, ethanol, 1-propanol, and 1-butanol) at infinite dilution as a function of temperature. Such thermo- dynamic data for multi-component systems at infinite dilu- tion allow on a more fundamental level, the study of molecular interactions between the alcoholic solutes and glycerol. A number of techniques have been used to study the solvent characteristics of glycerol including inverse gas chromatography [13], differential ebulliometry [14], dew point techniques [15], steady state and unsteady state gas– liquid chromatographic and liquid–liquid chromatographic methods [16]. The inverse gas chromatographic technique (IGC) has been widely used as a relatively simple and reproducible technique to measure thermodynamic data of solutes in vegetable oils [17], polymers [18], biodiesel [19] and organic solutes [20] as a function of temperature. In this study, the mole fraction activity coefficients, Henry’s law constants, heats of solution and vaporization for the alcoholic-solute interactions with glycerol are determined at infinite dilution as a function of temperature between 51.5 and 111 �C using the IGC technique. The solubility parameter of glycerol as a function of tempera- ture was also calculated from the physicochemical data obtained using the IGC technique and verified using the Hansen three-dimensional solubility parameter approach [21]. The solubility parameter and the interaction param- eter values are seminal for studying the solution charac- teristics and selectivity of glycerol towards the selected alcoholic solutes. Theoretical solution thermodynamic models based on solute–solvent interactions, such as the Miller–Guggenheim method [22] and group contribution methods, such as UNIFAC [23], have been used exten- sively in predicting the mole fraction activity coefficient of various compounds. These calculated activity coefficients values are compared with those experimentally derived from the IGC measurements. Experimental Procedure Materials Glycerol (99.7 % pure) was purchased from BDH Chem- icals (West Chester, PA, USA). Chromasorb G (45/60 mesh, acid washed and silanized, P/N# 25651) was pur- chased from the Restek Corporation (Bellefonte, PA, USA). Ultra high pure helium was used as the carrier gas, and the flame ionization combustion gases, hydrogen and air, were also purchased from Airgas (Tulsa, OK, USA). HPLC-grade methanol (P/N# MX0475-1), 1-propanol (P/N# PX1824-6), and 1-butanol (P/N# BX1780-6) were obtained from the VWR Corporation (Batavia, IL, USA). Ethanol (200 proof) was obtained from PharmCo Product, Inc. (Brookfield, CT, CAT # E200). Chromatographic injections were made using syringes obtained from the Hamilton Company (Reno,NV, USA). Column Preparation The glycerol chromatographic columns used in the exper- iments were prepared using 0.5 m sections of copper tub- ing (0.64 cm o.d 9 0.425 cm i.d). Column packings were prepared by mixing a selected weight percentage of glyc- erol dissolved in methanol with the packing material fol- lowed by solvent removal using a Rotavapor R110 (Buchi Labortechnik AG, Flawil, Switzerland). The amount of glycerol precipitated as the stationary phase was measured by taking triplicate samples of the packed material in ceramic crucibles and pyrolyzing them in a bench top muffle furnace (Omegalux LMF A550, Omega Engineer- ing Inc., Stamford, CT, USA) at 700 �C overnight. The weight difference before and after pyrolysis was used to determine the actual weight of glycerol in the stationary phase. Columns containing weight percentages of 11.2, 17.6, 21.7 and 27.4 % of glycerol were prepared using the above described method. After packing the coated Chro- masorb stationary phase support in the columns, they were equilibrated in the gas chromatographic oven overnight using a small flow of helium carrier gas to aid in removing any residual solvent left in the packing material. As a check, the percentage liquid phase was compared both before and after use of the IGC columns via the above pyrolysis technique, the total amount of glycerol lost from the column varied from 0.95 to 3.28 %. Inverse Gas Chromatography The experimental procedure followed in the study is sim- ilar to that described previously [24, 25]. A column con- taining a specific weight fraction of glycerol was placed in a HP 5890 (Agilent Technologies, Santa Clara, CA, USA) 1586 J Am Oil Chem Soc (2012) 89:1585–1597 123 Series II gas chromatograph equipped with a HP 7673 auto sampler and a flame ionization detector (FID). The flow rate of the gases was measured at the exit of the FID using a soap bubble flow meter. The column flow rate measure- ments were done at ambient temperature and then corrected to column conditions. The air to hydrogen ratio was maintained at 6:1 in order to maintain sufficient flame in the FID sensitivity for detecting the injected solutes under infinite dilution conditions on the glycerol column. Approximately 5 ll of methane was injected along with the solute to measure the column dead volume. The four alcoholic solutes were injected in triplicate over a given set of column temperatures: approximately 50, 65, 80, 95 and 111 �C, respectively. The actual column temperatures were measured using four J-type thermocouples placed at dif- ferent points in the oven with the signals transmitted to a Cole–Parmer data acquisition board (Model # 18200-40, Vernon Hills, IL, USA) and recorded using TracerDAQ software on a Dell Inspiron laptop computer. The atmo- spheric pressure was read from a mercury barometer (Fisher Scientific, Pittsburgh, PA, USA) at 4-hour intervals. The chromatograms obtained from each injection were recorded on a desktop computer using Hewlett Packard Chemstation v. 2.0 software and the data converted to Excel spreadsheet for further analysis. Calculations The theory and the calculations involved in determining the activity coefficients at infinite dilution of glycerol in the selected alcoholic solutes and its corresponding solubility parameter has been described previously [17, 19, 24] and will only be briefly discussed in this manuscript. The specific retention volume of the alcoholic solutes in glyc- erol can be calculated from Eq. 1 as: V0g ¼ ðtr � taÞð273:16=T0Þð760=P0Þð1 � Pw=P0ÞðjÞðF0Þ w2 ð1Þ where, tr, ta are the retention time of the solute (r) and non- sorbed (dead volume marker) solute (a), T0 is the ambient temperature, Pw is the vapor pressure of water, P0 is the atmospheric pressure at ambient conditions, j is the James– Martin compressibility factor, F0 is the flow rate at ambient conditions, and w2 is the weight of the stationary phase. The infinite dilution mole fraction activity coefficient of the alcoholic solutes in glycerol can be calculated using Eq. 2 as follows: ln c1x � � ¼ ln 273:16R V0g MP 0 ! � P 0 B11 � V � � RT ð2Þ where, P0 is the vapor pressure of the alcoholic solute (i), B11 is the second pure virial coefficient of the solute, �Vi is the molar volume of the solute, R is the universal gas constant and T is the column temperature (K). The value of P0 was calculated using Antoine’s equation and the corresponding constants for the different alcoholic solutes were obtained from the literature [26]. The value of the second virial coefficient (B11) is usually very small and often neglected. In this particular study, the values of B11 were found to be lower than the standard error associated with the experimental data and therefore, was considered negligible. The activity coefficient of the alcoholic solutes in glycerol in mole fraction units calculated using Eq. 2 can be further converted into the weight fraction activity coefficient using Eq. 3 as follows: c1w ¼ c1x Mi Msh i � � ð3Þ where, Mi is the molecular weight of the alcoholic solute and Ms is the molecular weight of the stationary phase, i.e., glycerol’s molecular weight. The variation in the retention volume (Vg 0) and mole fraction activity coefficient (cw) of the alcohols in glycerol with temperature can be used to calculate the enthalpies of solution (DHs ?) and mixing (DHm ?) at infinite dilution as given in Eqs. 4 and 5, respectively as follows: DH1s ¼ R o ln V0g � � o 1=T � � 0 @ 1 A ð4Þ DH1m ¼ R o ln c1x � � o 1=T � � 0 @ 1 A: ð5Þ The enthalpy of solution of the alcoholic solutes in glycerol is also related to its enthalpy of mixing and the enthalpy of vaporization as shown below in Eq. 6: DH1s ¼ DH1v � DH1m : ð6Þ The enthalpy of vaporization (DHv ?) is calculated from the difference between the heats of solution and mixing of the alcohols in glycerol. The enthalpy of vaporization of the alcohols was cal- culated from the Clausius–Clapeyron equation given in Eq. 7 as: ln P0 � � ¼ �DH 1 v RT þ A ð7Þ Thus, the DHv ? can be calculated from the slope of the variation of the natural logarithm of the vapor pressure (P0) as a function of inverse of temperature. As indicated J Am Oil Chem Soc (2012) 89:1585–1597 1587 123 previously, the vapor pressures of the alcohols were cal- culated from the Antoine’s correlation. The magnitude of the Henry’s Law constant at infinite dilution H?, can be used to calculate the solute vapor pressure above the glycerol phase and was calculated using Eq. 8 below: H1 ¼ RT V0g M : ð8Þ The alcoholic solute–glycerol interactions can also be evaluated by calculating the Flory–Huggins interaction parameters at infinite dilution (v?) as given in Eq. 9. The critical interaction parameter (vc) can be calculated using Eq. 10: ln vtð Þ ¼ ln c1x � �� ln m1 m2 � � � 1 þ Vi M2m2 ð9Þ vc ¼ 1 2 1 þ ffiffiffiffiffi m1 m2 r� �2 ð10Þ where, m1 and m2 are the specific molar volumes of solute and solvent at a given temperature, T. The solubility parameter of glycerol, which is the square root of its cohesive energy density, as a function of tem- perature, can also be calculated from the IGC retention data using an Eq. 11 [13]. d21 RT � v 1 V1 � � ¼ 2d2 RT � � d1 � d 2 2 RT þ vs V1 � � ð11Þ where, d1 and d2 are the solubility parameters of solute and solvent, respectively, at temperature T and vs is the entropic contribution to the total interaction parameter of the alco- hol–glycerol systems. In order to expand the database for an accurate predictionof the solubility parameter of glyc- erol, the activity coefficients of other alkanes, aromatics and chlorinated hydrocarbons in glycerol at the experi- mental temperatures were obtained from the literature [27]. The entropic contribution to the total interaction parameter (vs) for the solutes in glycerol can be calculated from Eq. 12. The solubility parameters of the solutes were obtained from literature [28] and their variation with temperature estimated using Eq. 13 [29]. vs ¼ v1 � vH ¼ v1 � m1 d1 � d2ð Þ2 RT ð12Þ and, d1;T2 ¼ d1;T1 1 � T2 1 � T1 � �0:34 ð13Þ where, T1 is at 298.15 K and T2 is the experimental tem- perature in Kelvin. The total solubility parameter of glycerol (dT) can also be expressed as a function of the contribution from the dispersive (dD), polar (dP) and hydrogen bonding (dH) forces as given by Eq. 14. d2T ¼ d2D þ d2P þ d2H: ð14Þ The utilization of the Hansen sphere approach to estimate the solubility parameter of glycerol has been discussed in detail previously [19]. The Hansen spheres based on the interaction of glycerol with various solutes reported in the literature [30] were plotted using a Hsp3D program that was kindly provided as a freeware by Dr. Fred Turner (Western Research Institute, Laramie, WY, USA). The use of this program to plot the Hansen spheres in order to predict the solubility parameter of a compound is based on its interaction with other solvents has been discussed in detail in the literature [31]. The Hansen spheres can also be used to characterize the solvent properties of glycerol as reported previously in the literature [19]. This can be achieved by comparing the relative energy difference (RED) values of various solutes in glycerol defined as ratio of Ra and the radius of the Hansen sphere calculated by the Hsp3D program (R0). Here, Ra was calculated using Eq. 15 with ‘1’ referring to the solute and ‘2’ referring to the glycerol. R2a ¼ 4 dD1 � dD2ð Þ2þ dP1 � dP2ð Þ2þ dH1 � dH2ð Þ2 ð15Þ The thermodynamic selectivity factors of glycerol for the four alcoholic solutes are calculated using Eq. 16: Table 1 Specific retention volumes (mL/g) for glycerol/alcohol systems as a function of temperature Solute Temperature (�C) 51.5 66.0 81.2 95.8 111 Methanol 232 (±2.30) 115 (±0.25) 65.7 (±0.10) 40.4 (±0.09) 26.1 (±0.10) Ethanol 190 (±2.34) 98.1 (±0.08) 57.3 (±0.10) 36.4 (±0.10) 21.6 (±0.07) 1-Propanol 220 (±0.98) 110 (±0.34) 59.0 (±0.09) 35.1 (±0.13) 21.3 (±0.53) 1-Butanol 277 (±0.28) 134 (±0.86) 69.8 (±0.26) 39.5 (±1.23) 23.0 (±0.07) Values in parentheses with ± denote standard deviations from the mean (n = 3) 1588 J Am Oil Chem Soc (2012) 89:1585–1597 123 S ¼ c i x c jx ð16Þ where cx i and cx j refers to the mole fraction activity coefficient at infinite dilution of solutes ‘i’ and ‘j’ in glycerol, respectively. The separation factors, a, between the various alcohols are calculated using Eq. 17 a ¼ c i xP 0 i c jxP0j ð17Þ utilizing the mole fraction activity coefficients and respective vapor pressures that were previously calculated for each alcoholic solute. Results and Discussion The specific retention times and the retention volume data for the glycerol–alcohol systems as a function of temper- ature is given in Table 1. The determined specific retention volumes for the alcoholic solutes in glycerol over the dif- ferent column loadings noted previously were in excellent agreement with each other, indicating that a negligible influence of the gas–liquid adsorption effect [25] for the n-alcohols on the glycerol loadings used in this study. The mole fraction and the weight fraction activity coefficients calculated using Eqs. 2 and 3, respectively, are listed in Table 2. It can be seen from the data given in Table 2 that the alcohol–glycerol mole fraction activity coefficients changed appreciably with temperature. The mole fraction activity coefficient of the alcohols in glycerol were also greater than zero which is indicative of a positive deviation from Raoult’s law, however, these mole fraction activity coefficients were small when compared to those for alkanes, alkenes and alkyl benzenes in glycerol, indicative that the alcohols exhibit a lesser tendency to escape from glycerol [27]. The alcohol-glycerol activity coefficient values did exhibit an increasing trend with the increasing carbon number of the alcoholic solute as can be seen from the semi-logarithmic plot shown in Fig. 1. This highly linear correlation in Fig. 1 can be used for estimating the mole fraction activity coefficient of other higher carbon number alcohols in glycerol. This increase in the mole fraction activity coefficient of the alcohol in glycerol with the carbon number of the alcohols is indicative of the greater fugacity of the solute to escape from the glycerol phase. A similar increase in the mole fraction activity coefficient as a function of the carbon number is also exhibited by alkanes and alkyl benzenes in glycerol [32, 33]. These measured mole fraction activity coefficients were com- pared with the limited data available in the literature and with that predicted using theoretical models such as theT a b le 2 M o le fr ac ti o n an d m as s fr ac ti o n ac ti v it y co ef fi ci en ts fo r al co h o l– g ly ce ro l sy st em s as a fu n ct io n o f te m p er at u re S o lu te T em p er at u re (o C ) 5 1 .5 6 6 .0 8 1 .2 9 5 .8 1 1 1 c x? c w? c x? c w? c x? c w? c x? c w? c x? c w? M et h an o l 1 .8 6 (± 0 .0 2 ) 5 .3 4 (± 0 .0 6 ) 2 .0 2 (± 0 .0 0 ) 5 .8 2 (± 0 .0 1 ) 2 .0 1 (± 0 .0 0 ) 5 .7 8 (± 0 .0 1 ) 1 .9 9 (± 0 .0 0 ) 5 .7 1 (± 0 .0 1 ) 1 .9 5 (± 0 .0 1 ) 5 .6 0 (± 0 .0 2 ) E th an o l 3 .9 7 (± 0 .0 4 ) 7 .9 4 (± 0 .0 8 ) 4 .0 7 (± 0 .0 0 ) 8 .1 4 (± 0 .0 1 ) 3 .8 8 (± 0 .0 1 ) 7 .7 5 (± 0 .0 1 ) 3 .6 4 (± 0 .0 1 ) 7 .2 8 (± 0 .0 2 ) 3 .7 6 (± 0 .0 1 ) 7 .5 1 (± 0 .0 2 ) 1 -P ro p an o l 7 .8 0 (± 0 .0 3 ) 1 2 .0 (± 0 .0 5 ) 7 .8 2 (± 0 .1 2 ) 1 2 .0 (± 0 .1 5 ) 7 .6 6 (± 0 .0 1 ) 1 1 .7 (± 0 .0 2 ) 7 .3 3 (± 0 .0 3 ) 1 1 .2 (± 0 .0 4 ) 7 .2 0 (± 0 .0 4 ) 1 1 .0 (± 0 .2 5 ) 1 -B u ta n o l 1 4 .4 (± 0 .0 2 ) 1 7 .9 (± 0 .0 3 ) 1 4 .4 (± 0 .1 2 ) 1 7 .9 (± 0 .1 5 ) 1 3 .9 (± 0 .0 6 ) 1 7 .2 (± 0 .0 7 ) 1 3 .5 (± 0 .4 2 ) 1 6 .8 (± 0 .5 2 ) 1 3 .5 (± 0 .0 4 ) 1 6 .8 (± 0 .0 5 ) V al u es in p ar en th es es w it h ± d en o te st an d ar d d ev ia ti o n s fr o m th e m ea n (n = 3 ) J Am Oil Chem Soc (2012) 89:1585–1597 1589 123 Universal Functional Activity Coefficient (UNIFAC) approach [23] and the Martire method using the Miller– Guggenheim estimation for the ln(cs ?) contribution to the log of the total activity coefficient, ln(cT ?) [22]. Since the data available in the literature was limited to 62.4 �C [34] and at 25 �C [35], respectively, the experimentaland theoretically- predicted values for the activity coefficient at approximately 65 �C are provided in Table 3 for comparison purposes. Also, the activity coefficient values are listed in Dechema for C1–C6 compounds, namely the values listed for methanol, ethanol, and 1-propanol in glycerol from 60 to 80 �C [36]. These values for methanol (2.10–2.12), ethanol (4.06–4.18), and 1-propanol (7.85–8.50) agree well with our experimentally derived mole fraction activity coefficient values from this study in Table 3. However, there was discrepancy between the experi- mental and the theoretical calculated values of the mole fraction activity coefficients of the alcohols in glycerol using the two above described approaches to calculate the total mole fraction activity coefficient, cx ? for the n-alco- hols in glycerol. The UNIFAC calculations predicted lower values for mole fraction activity coefficients of the alcohols in glycerol when compared to experimental and literature values in Table 3. Predicted mole fraction activity coeffi- cients at infinite dilution using, UNIFAC are known to be unreliable, and it has even been suggested that experi- mental values of cT ? can be used to evaluate the UNIFAC parameters. Martire’s method tends to overestimate the mole fraction activity coefficients for the alcohols (espe- cially 1-propanol and 1-butanol) in glycerol at 66 �C. For the Martire method, the predicted activity coefficients are particularly sensitive to the values chosen for the glycerol solubility parameters, and hence have a considerable effect on the thermal contribution to the logarithm of the mole fraction activity coefficient, ln(cT ?), as well as the overall mole fraction activity coefficient value, ln(cT). Certainly, the limited data available for the activity coefficients of various alcohols in glycerol as a function of temperature supports the need for the current experimental study. The variation in the natural logarithm of retention vol- ume and mole fraction activity coefficient of alcohols in glycerol with the inverse of the absolute temperature is shown in Figs. 2 and 3, respectively, and their values are reported in Table 4. The experimental data conform to an excellent linear fit for Eq. 4 as indicated by correlation coefficients which were found to be greater than 0.999. It can be seen from Table 4 that the heat of solution for n-alcohols in glycerol varied in the following order: etha- nol \ methanol \ 1-propanol \ 1-butanol. The heat of solution is indicative of the energy required to overcome solute–solvent interactions at a constant pressure. The current trend indicates that ethanol was a better solvent for glycerol compared to methanol, 1-propanol and 1-butanol. The heat of mixing of the alcohols in glycerol did not show a specific trend with the carbon number of alcohols and these were very small compared to the heat of solution values (Fig. 3). Such small heats of mixing indicate that the selected alcoholic solutes are near-ideal solvents for glyc- erol and show minimal chemical interaction with glycerol. This also suggests that the overall solution behavior of the alcohol–glycerol mixture is athermal with the free energy of solution being dominated by the entropy of mixing. The enthalpy of vaporization of the alcoholic solutes in glycerol as calculated from Eq. 6 is listed in Table 4. As y = 1.0341e0.6549x R² = 0.9974 1.0 10.0 100.0 0 1 2 3 4 5 6 7 8 xY ∞ (m ole fr ac tio n) Alcohol Carbon Number Fig. 1 Variation of the mole fraction activity coefficient of the n-alcohols in glycerol as a function of the carbon number of the alcohols Table 3 Comparison between the theoretical and experimental activity coefficients of n-alcohols in glycerol (cx ?) at different temperatures Solute cx ? Experimental (66.0 �C) Martire [34] (62.4 �C) Locke [35] (25.0 �C) UNIFAC [23] (66.0 �C) Martire [34] (66.0 �C) Methanol 2.02 1.63 1.74 0.631 1.58 Ethanol 4.07 2.98 2.90 1.83 5.83 1-Propanol 7.82 6.18 7.53 3.22 23.2 1-Butanol 14.4 11.4 17.2 5.80 125 1590 J Am Oil Chem Soc (2012) 89:1585–1597 123 indicated in the ‘‘Calculations’’ section, the enthalpy of vaporization can also be predicted from the slope of the variation of natural logarithm of vapor pressure plotted as a function of time as shown in Eq. 7. The theoretically-pre- dicted DHv ? of the alcohols are also given in Table 4, and are in excellent agreement with those derived from the dif- ference in the experimentally-measured values for the heats of mixing and solution (relative standard deviation of\1 %). It can be seen from the data provided in Table 5 that the Henry’s law constants increased exponentially with increases in temperature, and that the solute volatility decreased with an increase in the molecular weight of the alcohols. The high precision of this data in Table 5 as well as in Table 2 attest to the reproducibility of the experi- mental method using the HP 5890 gas chromatograph with automated injection. This data on the Henry’s law con- stants at infinite dilution can be used in design of desol- ventizing units and the activity coefficient data in Table 2 is critical for the design of phase separation techniques such as centrifugation and decantation [37, 38]. The values of the Flory–Huggins parameters of the alcoholic solutes in glycerol as a function of temperature and the range of critical interaction parameters are given in Table 6. The v? values for alcohol/glycerol systems increased until 66 �C (with the exception of 1-butanol where the values were constant) and then decreased as a function of temperature indicating higher miscibility of the alcohols with glycerol above 66o. The v? values for alcohol–glycerol systems also increased as a function of the carbon number of the alcohols at a specific temperature. It can also be seen from Table 6 that the v? values for methanol and ethanol were lower than their vc values while that of 1-propanol was closer to its vc and that of 1-butanol greater than its vc value as a function of temperature. This is indicative of the different levels of solubility-miscibility exhibited by the alcoholic solutes in glycerol which can vary since this non-ideal behavior is dependent on solute concentration [39, 40]. It was also found that this is an opposite trend that was observed for the same alcohols with a methyl soyate solvent system, where the lower alcohols showed v? values greater than their respective vc values and the v? values decreased as a function of the carbon number [19]. This is indicative of why the lower alcohols can be better separated from methyl soyate (biodiesel) than from glycerol after a transesterification reaction [41]. The v? and vc values calculated as a function of temperature are also given in Table 6. It was seen that the interaction 0.00 1.00 2.00 3.00 2. 50 2. 60 2. 70 2. 80 2. 90 3. 00 3. 10 3. 20 ln (γ x ∞ ) 1000/T (K-1) Methanol Ethanol 1-Propanol 1-Butanol Fig. 3 Variation of the natural logarithm of the weight fraction activity coefficient for the n-alcohols in glycerol as a function of the inverse of temperature (1,000 K-1) Table 4 Comparison between theoretical and experimental heats of vaporization (DHv ?) Solute DHs ? (Kcal/mol) DHm ? (Kcal/mol) Experimental DHv ? (Kcal/mol) Theoretical DHv ? (Kcal/mol) % Difference Methanol 9.08 (±0.4) -0.14 (±0.04) 8.94 8.88 0.64 Ethanol 8.92 (±0.5) 0.37 (±0.05) 9.30 9.38 0.94 1-Propanol 9.73 (±0.2) 0.37 (±0.02) 10.1 10.1 0.16 1-Butanol 10.4 (±0.5) 0.32 (±0.07) 10.7 10.7 0.22 Values in parentheses with ± denote standard deviations from the mean (n = 3) 2.003.00 4.00 5.00 6.00 2.50 2.60 2.70 2.80 2.90 3.00 3.10 3.20 ln (V g0 ) (m l/g ) 1000/T (K-1) Methanol Ethanol Propanol Butanol Fig. 2 Variation of the natural logarithm of the specific retention volume (mL/g) for the n-alcohols solutes as a function of the inverse of temperature (K-1) J Am Oil Chem Soc (2012) 89:1585–1597 1591 123 parameters of the alcohols in glycerol were smaller than that for alkanes, aromatics and chlorinated hydrocarbons, indicating relatively greater miscibility and compatibility of the alcohols with glycerol. As noted in the ‘‘Experimental Procedure’’ Section, Eq. 11 can be used to calculate the solubility parameter of glycerol at a specific temperature (ex. 111 �C in Fig. 4) and can be obtained by plotting (d1 2/RT-vi/V1) versus d1 (see Fig. 4). Figure 4 utilizes data from 24 solute–glycerol systems and shows an excellent correlation coefficient of 0.955 at 111 �C. Similarly, correlation coefficients of 0.957, 0.956, 0.955 and 0.955 were found at 51.5, 66.0, 81.2 and 95.8 �C, respectively. The solubility parameter of the glycerol calculated using Eq. 11 as a function of temperature is given in Fig. 5 and the solubility parameter value at 25 �C (298.15 K) was found to be 34.8 MPa1/2 by extrapolation of the linear relationship. The solubility parameter of glycerol at 25 �C has been reported in the literature and was found to vary between 33.8 MPa1/2 [42] and 36.1 MPa1/2 [43]. However, Table 5 Henry’s law constants (atm) for glycerol/alcohol systems as a function of temperature Solute Temperature (oC) 51.5 66.0 81.2 95.8 111 Methanol 3.68 (±0.007) 7.59 (±0.001) 13.9 (±0.000) 23.5 (±0.000) 38.0 (±0.000) Ethanol 3.08 (±0.007) 6.19 (±0.000) 11.2 (±0.000) 18.6 (±0.000) 32.4 (±0.000) 1-Propanol 2.03 (±0.003) 4.21 (±0.001) 8.25 (±0.000) 14.4 (±0.000) 24.7 (±0.002) 1-Butanol 1.31 (±0.001) 2.83 (±0.003) 5.64 (±0.001) 10.4 (±0.004) 18.5 (±0.000) Values in parentheses with ± denote standard deviations from the mean (n = 3) Table 6 Flory-Huggins interaction parameters (v?) and critical interaction parameters (vc) for glycerol/solute systems at different temperatures Solute Temperature (oC) vc 51.5 66.0 81.2 95.8 111 Methanol 0.75 0.83 0.82 0.80 0.78 2.63–2.71 Ethanol 1.40 1.42 1.37 1.31 1.33 2.16–2.22 1-Propanol 2.05 2.06 2.04 2.00 1.98 1.91–1.96 1-Butanol 2.70 2.70 2.67 2.65 2.65 1.74–1.78 Hexane [28] 6.21 6.08 5.96 5.86 5.77 1.39–1.44 Heptane [28] 6.47 6.26 6.07 5.90 5.74 1.41–1.44 Octane [28] 7.01 6.82 6.64 6.48 6.33 1.36–1.38 Nonane [28] 7.43 7.30 7.19 7.09 7.00 1.31–1.33 Decane [28] 7.80 7.63 7.48 7.34 7.22 1.27–1.29 Cyclohexane [28] 4.83 4.59 4.36 4.15 3.97 1.61–1.64 Styrene [28] 4.56 4.21 3.87 3.58 3.31 1.58–1.60 Benzene [28] 3.74 3.59 3.44 3.31 3.19 1.76–1.80 Toluene [28] 4.55 4.30 4.05 3.84 3.64 1.63–1.66 Ethyl benzene [28] 5.07 4.87 4.68 4.51 4.35 1.54–1.56 o-Xylene [28] 4.86 4.62 4.39 4.19 4.00 1.55–1.57 m-Xylene [28] 4.95 4.74 4.53 4.35 4.18 1.54–1.56 p-Xylene [28] 5.03 4.83 4.65 4.48 4.33 1.53–1.56 Tetrachloromethane [28] 5.81 5.47 5.14 4.85 4.58 1.69–1.73 Trichloromethane [28] 3.88 3.68 3.49 3.32 3.17 1.83–1.89 Chlorobromomethane [28] 3.74 3.68 3.62 3.56 3.51 2.02–2.07 Dichloromethane [28] 3.07 2.91 2.74 2.60 2.47 2.03–2.11 Dibromomethane [28] 4.20 4.10 4.00 3.91 3.83 1.99–2.03 1,2-Dichloroethane [28] 3.74 3.61 3.48 3.37 3.27 1.75–1.82 Water [28] -1.42 -1.39 -1.37 -1.34 -1.33 4.50–4.55 1592 J Am Oil Chem Soc (2012) 89:1585–1597 123 the reported values for the solubility parameter of glycerol were either estimated using empirical models and/or through functional group contribution methods. One such approach utilizing Fedors’ group contribution method estimated the value of solubility parameter of glycerol to be 33.6 MPa1/2 [44]. As can be seen, the solubility parameters of glycerol calculated from the IGC retention data was in good agreement with those reported in the literature and those that can be predicted using group contribution methods. As discussed in the ‘‘Calculations’’ section, the solu- bility parameter of glycerol was also predicted using the Hansen sphere approach with the aid of the Hsp3D pro- gram. The solubility parameter of glycerol, which is assumed to be the center of mass of the Hansen sphere shown in Fig. 6a, was found to be dD = 20.5 MPa 1/2, dP = 14.6 MPa 1/2, dH = 19.1 MPa 1/2. The total solubility parameter of glycerol calculated using Eq. 14 was found to be 31.6 MPa1/2. The glycerol solubility parameter esti- mated using this method, though slightly smaller, was found to be consistent in magnitude with that calculated using IGC. The RED values of common organic solutes in glycerol as obtained from the Hsp3D program are given in Table 7. The solubility parameter theory is based on the principle of ‘‘like dissolves like’’ which means that compounds of similar solubility parameter values become miscible with each other. In this regard, the solutes with RED values\1 show good interactions with glycerol while those with RED [1 showed partial or complete immiscibility in glycerol. Such poorly miscible solutes are represented by the dark triangles outside the Hansen sphere and those miscible are represented by the inverted triangles inside the sphere in Fig. 6a–c. It can be seen from Table 7 that mostly alcohols, amines and sulfonated hydrocarbons showed good compatibility with glycerol while n-alkanes and chlorinated hydrocarbons were found to be very poor solvents for glycerol. Salicylaldehyde showed the best compatibility with glycerol when compared with the alcoholic solutes. Table 7 indicates that the compatibility of alcohols with glycerol decreased with an increase in the carbon number of the alcohols. Moreover, the accuracy of the above such predictions can be improved by using a larger database. The IGC technique described in this study provides one such method that can be used to measure these thermodynamic properties and hence predict solute– glycerol interactions. It is interesting to compare the selectivity factor of glycerol (Eq. 16) with respect to the four n-alcohol homologues using the activity coefficient of methanol as a function of temperature as a baseline, as shown in Fig. 7. Here the selectivity factor of glycerol for ethanol, 1-propanol, and 1-butanol relative to methanol’s activity coefficient is in order of the magnitude of their positive deviation from Raoult’s Law, and the relative selectivity of glycerol remains fairly constant over a 60 �C tem- perature range. On the other hand, the separation factor defined by Eq. 17 which embraces both the solution non- ideality as reflected in the activity coefficients as well as their respective saturated vapor pressures, P0, shows quite a different trend with temperature as shown in Fig. 8. At 50 �C, the separation factors for ethanol, 1-propanol, and 1-butanol are in increasing order with respect to the carbon number of the alcohol, ranging from approximately 0.82–1.4. Whereas the selectivity factor of ethanol relative to methanol remains fairly constant over a wide range of temperature, both the selectivity for 1-propanol and y = 0.0204x-0.2821 R² = 0.9545 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 5 10 15 20 25 30 35 40 45 50 δ 1 2 /R T- i /V 1 (m ole /cc ) δ 1 (MPa1/2) Fig. 4 Determination of the solubility parameter of glycerol at 111 �C using Eq. 11 y = -0.0264x + 42.673 R² = 0.9967 32.4 32.6 32.8 33.0 33.2 33.4 33.6 33.8 34.0 34.2 34.4 320 330 340 350 360 370 380 390 So lu bi lit y pa ra m et er o f g ly ce ro l (M Pa 1/ 2 ) Temperature (K)Fig. 5 Variation of the solubility parameter of glycerol calculated from IGC retention data as a function of temperature J Am Oil Chem Soc (2012) 89:1585–1597 1593 123 1-butanol decreases significantly with temperature in going from 50 to 111 �C. For 1-propanol, selectivity decreases from approximately unity to about 0.82, while for 1-butanol over the same temperature range it decreases from 1.25 to just under 0.9. These trends argue for a decrease in selectivity of glycerol for the higher n-alcohol homologues as temperature is increased (Fig. 8). Fig. 6 a Three-dimensional solubility sphere of glycerol plotted using Hsp3D program; b two-dimensional plot of dH versus dD of the Hansen solubility sphere of glycerol; c two-dimensional plot of dH versus dP of the Hansen solubility sphere of glycerol 1594 J Am Oil Chem Soc (2012) 89:1585–1597 123 Conclusions The solution thermodynamic data of the selected alcoholic solutes in glycerol measured using inverse gas chromatography are indicative of limited molecular solute– solvent interactions at infinite dilution. The infinite dilution activity coefficients of the alcoholic solutes in glycerol as a function of temperature increased as a function of the carbon number and also showed a positive deviation from Table 7 RED values of selected solvents with glycerol obtained from Hsp3D program Solvent RED Solvent RED Salicylaldehyde 0.392 1,4-Dioxane 1.10 Phenol 0.436 Chloroform 1.15 Isopropanol amine 0.511 Trichloroethylene 1.16 Dimethyl sulfoxide 0.622 Tri-n-butyl ketone 1.18 Diethylformamide 0.628 Isoamyl acetate 1.23 Ethylene glycol 0.655 n-Heptyl acetate 1.26 Ethanol 0.690 Di-isobutyl ketone 1.29 n-Propanol 0.752 Chlorobenzene 1.36 Cresol 0.774 Ethyl ether 1.36 Triethylenetetramine 0.770 Dipropylamine 1.41 Isopropanol 0.809 Benzene 1.43 1-Butanol 0.817 Di(2-ethylhexyl) amine 1.45 Propylene glycol 0.820 Carbon tetrachloride 1.51 Cinnamaldehyde 0.826 Cyclohexane 1.56 Pyridine 0.918 Water 1.58 Tertiary amyl alcohol 0.919 Mineral spirits 1.60 Ethyl phenyl acetate 0.969 n-Dodecane 1.60 2-Phenylethyl amine 0.991 n-Decane 1.61 a-Methylbenzylamine 1.00 n-Nonane 1.61 Tetrahydrofuran 1.00 n-Octane 1.62 Acetone 1.01 n-Heptane 1.63 1-Tetradecanol 1.03 n-Hexane 1.65 Methyl isopropyl ketone 1.04 n-Pentane 1.68 Ethyl cinnamate 1.05 n-Butane 1.70 0.60 1.60 2.60 3.60 4.60 5.60 6.60 7.60 8.60 40.00 50.00 60.00 70.00 80.00 90.00 100.00 110.00 120.00 Se le ct iv ity o f g ly ce ro l f or a lc oh ol s re la tiv e to m et ha no l Temperature (°C) Ethanol 1-Propanol 1-Butanol Fig. 7 Variation in the selectivity factor of the glycerol for n-alcohols relative to methanol as a function of temperature 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 40.00 50.00 60.00 70.00 80.00 90.00 100.00 110.00 120.00 Vo la til ity o f a lc oh ol s in g ly ce ro l r el at iv e to m et ha no l Temperature (°C) Ethanol 1-Propanol 1-Butanol Fig. 8 Variation of the volatility of n-alcohols in glycerol relative to methanol as a function of temperature J Am Oil Chem Soc (2012) 89:1585–1597 1595 123 Raoult’s law, indicating a propensity to escape the glycerol medium. However, this tendency of the alcohols to volatize from the glycerol medium was found to be considerably smaller when compared to alkanes, aromatics and alkyl ben- zenes activity coefficient values reported in the literature. From the heat of solution and mixing data obtained from IGC, it was also seen that the mixing of the solute with the solvent was greatly dominated by the entropy of the solution with ethanol and methanol showing better interaction with glyc- erol. The total solubility parameter of glycerol was calculated from the IGC retention data as 34.8 MPa1/2 and compared well to the literature values and that predicted using a Hansen three-dimensional solubility parameter and Hansen solvation sphere approach (31.6 MPa1/2). The Hansen solvation sphere approach was used also used to characterize the solvent properties of glycerol. The results obtained from the solvation sphere approach in conjunction with the Flory–Huggins interaction parameter trends measured from IGC retention data indicated that the lower alcohols showed greater com- patibility with glycerol and the interaction decreased as a function of the carbon number of the alcoholic solutes. These results can play a significant role in understanding the solvent properties of glycerol and its application as a ‘‘green’’ renewable solvent [45]. Acknowledgments Joel Vincent acknowledges the University of Arkansas Honors College for a grant which made this study possible. We would also like to acknowledge Mr. Harold Watson of the Department of Chemical Engineering at the University of Arkansas for his technical assistance during various aspects of this study. References 1. Ma F, Hanna MA (1999) Biodiesel production: a review. Bior- esour Technol 70:1–15 2. Kenar JA (2007) Glycerol as a platform chemical: sweet oppor- tunities on the horizon? Lipid Technol 19:249–253 3. King JW, Holliday RL, List GR (1999) Hydrolysis of soybean oil in a subcritical water flow reactor. Green Chem 1:261–264 4. Behr A, Eilting J, Irawadi K, Leschinski J, Lindner F (2008) Improved utilisation of renewable resources: new important derivatives of glycerol. Green Chem 10:13–30 5. Petrou EC, Pappis CP (2009) Biofuels: a survey on pros and cons. Energ Fuel 23:1055–1066 6. Pagliaro M, Rossi M (2010) The future of glycerol, 2nd edn. RSC Publishing, Cambridge 7. Pagliaro M, Ciriminna R, Kimura H, Rossi M, Pina CD (2007) From glycerol to value-added products. Angew Chem Int Ed 46:4434–4440 8. Robergs RA, Griffin SE (1998) Glycerol: biochemistry, phar- macokinetics and clinical and practical applications. Sports Med 26:145–167 9. Sharma SC, Yang RK (1986) Chewing gum compositions con- taining novel sweetener delivery systems and method of prepa- ration. US Patent 4,597,970 10. Melero JA, van Grieken R, Morales G, Paniagua M (2007) Acidic mesoporous silica for the acetylation of glycerol: synthesis of bioadditives to petrol fuel. Energ Fuel 21:1782–1791 11. Wolfson A, Dlugy C, Shotland Y (2007) Glycerol as a green solvent for high product yields and selectivities. Environ Chem Lett 2:67–71 12. Cumming APC, Morton F (1952) Solvent extraction of phenol from coal-tar hydrocarbons: the use of glycerol, triethylene glycol and their aqueous solutions as solvents. J Appl Chem 2:314–323 13. Voelkel A, Strzemiecka B, Adamska K, Milczewska K (2009) Inverse gas chromatography as a source of physicochemical data. J Chromatogr A 1216:1551–1566 14. Eckert CA, Newman BA, Nicolaides GL, Long TC (1981) Measurement and application of limiting activity coefficients. AIChE J 27:33–40 15. Trampe DB, Eckert CA (1993) Dew point technique for limiting activity coefficient in nonionic solutions. AIChE J 39:1045–1050 16. Kojima K, Zhang S, Hiaki T (1997) Measuring methods of infi- nite dilution activity coefficients and a database for systems including water. Fluid Phase Equilibr 131:145–179 17. King JW, List GR (1990) A solution thermodynamic study of soybean oil/solvent systems by inverse gas chromatography. J Am Oil Chem Soc 67:424–430 18. Zeng C, Li J, Wang D, Chen T, Zhao C, Chen C (2006) Infinite dilution activity and diffusion coefficients in polymers by inverse gas chromatography. J Chem Eng Data 51:93–98 19. Srinivas K, Potts TM, King JW (2009) Characterization of sol- vent properties of methyl soyate by inverse gas chromatography and solubilityparameters. Green Chem 11:1581–1588 20. Voelkel A, Kopczynski T (1998) Inverse gas chromatography in the examination of organic compounds: polarity and solubility parameters of isoquinoline derivatives. J Chromatogr A 795:349–357 21. Adamska K, Bellinghausen R, Voelkel A (2008) New procedure for the determination of Hansen solubility parameters by inverse gas chromatography. J Chromatogr A 1195:146–149 22. Martire DE (1963) Gas chromatography. In: Fowler L (ed) Academic Press, New York, pp 33–54 23. Fredenslund A, Jones RL, Prausnitz JM (1975) Group-contribu- tion estimation of activity coefficients in non-ideal liquid mix- tures. AIChE J 21:1086–1099 24. Bobbitt NS, King JW (2010) Physicochemical characterization of dilute n-alcohol/biodiesel mixtures by inverse gas chromatogra- phy. J Chromatogr A 1217:7898–7906 25. Nitta T, Moringa K, Katayama T (1982) Gas chromatographic study of limiting activity coefficients of organic solutes in squalene. Ind Eng Chem Fundamen 21:396–401 26. Yaws CL, Narasimhan PK, Gabbula C (2009) Yaws’ handbook of Antoine coefficients for vapor pressure. 2nd electronic edn. Knovel Corporation, available online at http://www.knovel. com 27. Ge M-N, Ma J-L, Wu C-G (2010) Activity coefficients at infinite dilution of alkanes, alkenes and alkyl benzenes in glycerol using gas–liquid chromatography. J Chem Eng Data 55:1714–1717 28. Hansen CM (2007) Hansen solubility parameters: a user’s handbook, 2nd edn. CRC Press, Boca Raton, USA 29. Jayasri A, Yaseen M (1980) Nomograms for solubility parameter. J Coatings Technol 52:41–45 30. Dow Chemical Corporation (accessed Jun. 2009) Miscibility of organic solvents with glycerine. http://www.dow.com/glycerine/ resources/table23.htm 31. Redelius P (2007) Hansen solubility parameters of asphalt, bitumen, and crude oils. In: Hansen CM (ed) Hansen solubility parameters: a user’s handbook. CRC Press, Boca Raton, pp 151–176 32. Cruickshank AJB, Gainey BW, Hicks CP, Letcher TM, Moody RW, Young CL (1969) Gas–liquid chromatographic determina- tion of cross-term second virial coefficients using glycerol. Trans Faraday Soc 65:1014–1031 1596 J Am Oil Chem Soc (2012) 89:1585–1597 123 33. Castells RC, Arancibia EL, Nardillo AM (1982) Solution and adsorption of hydrocarbons in glycerol as studied by gas–liquid chromatography. J Phys Chem 86:4456–4460 34. Martire DE (1966) Solute adsorption at the gas–liquid interface in gas–liquid chromatography. Anal Chem 38:244–255 35. Locke DC (1968) Chromatographic study of solutions of hydro- carbons in acetonitrile. J Gas Chromatogr 35:24–36 36. Tiegs D, Gmehling J, Medina A, Soares A, Bastos J, Alessi P, Kikic I, Schiller M and Menke J (1986) Activity coefficients at infinite dilution. vol. 9, Part 1. DECHEMA Chemistry Data Series, Frankfurt, Germany 37. Tizvar R, McLean DD, Kates M, Dube MA (2009) Optimal separation of glycerol and methyl oleate via liquid–liquid extraction. J Am Oil Chemists Soc 54:1541–1550 38. Chiu C-W, Goff MJ, Suppes GJ (2005) Distribution of methanol and catalysts between biodiesel and glycerin phases. AIChE J 51:1274–1278 39. King JW (1989) Fundamentals and applications of supercritical fluid extraction in chromatographic science. J Chromatogr Sci 27:355–364 40. King JW (1995) Determination of the solubility parameter of soybean oil by inverse gas chromatography. Lebensm-Wiss U-Technol 28:190–195 41. Zhou W, Boocock DGB (2006) Phase distributions of alcohol, glycerol, and catalyst in the transesterification of soybean oil. J Am Oil Chemists Soc 83:1047–1052 42. Alli A, Hazer B, Baysal BM (2006) Determination of solubility parameters of cross-linked macromonomeric initiators based on propylene glycol. Euro Polymer J 42:3024–3031 43. Roy SK, Chanda M (2006) Plastics technology handbook: plas- tics engineering series. CRC Press, Boca Raton 44. Fedors RF (1974) A method for estimating both the solubility parameters and the molar volumes of liquids. Polymer Eng Sci 14:147–154 45. Gafner S, Bergeron C, McCollom MM, Cooper LM, McPhail KL, Gerwick WH, Angerhofer CK (2004) Evaluation of the efficiency of three different solvent systems to extract triterpene saponins from roots of Pana quinquefolius using high-performance liquid chromatography. J Agric Food Chem 52:1546–1550 J Am Oil Chem Soc (2012) 89:1585–1597 1597 123 Characterization of the Solvent Properties of Glycerol Using Inverse Gas Chromatography and Solubility Parameters Abstract Introduction Experimental Procedure Materials Column Preparation Inverse Gas Chromatography Calculations Results and Discussion Conclusions Acknowledgments References