The objective of this study was to investigate the effect of
structure on fluorinated ionic liquids ability to separate organic mixtures. To
assess the effectiveness of FILs as separation agents, limiting experimental
selectivities and capacities were calculated from experimental infinite
dilution activity coefficient data.
Infinite dilution activity coefficients of various organic
solutes have been measured by gas-liquid chromatography over a wide temperature
range in seven fluorinated ionic liquids:
· Trihexyltetradecylphosphonium tetrafluoroborate,
[3C6C14P] [BF4];
· Trihexyltetradecylphosphonium
bis(trifluoromethylsulfonyl) imide, [3C6C14P] [Tf2N];
· Trihexyltetradecylphosphonium hexafluorophosphate,
[3C6C14P] [PF6];
· Methyltrioctylammonium bis (trifluoromethylsulfonyl)
imide, [C13C8N] [Tf2N];
· 1-ethyl-3-methylimidazolium trifluoromethanesulfonate,
[EMIM] [TfO];
· 1-butyl-3-methylimidazolium hexafluoroantimonate, [BMIM]
[SbF6];
· 1-methyl-3-octylimidazoliumhexafluorophosphate.
[MOIM][PF6].
Two of these FILs ([3C6C14P] [BF4] and [3C6C14P]
[Tf2N]) have been investigated previously with a smaller number of solutes and
a narrower temperature range than in this study. Most of the reported
measurements are new. For each solvent, the Gibbs-Helmholtz equation was used
to derive molar excess enthalpies of solutes at infinite dilution. This was
important as the property provides information on the tendency of the solvent
to associate with a solute of interest. The large number of newly generated
infinite dilution activity coefficient values expanded the database used for
examining the effect of structure on the separation ability of FILs. If these
additional data were not produced, some of the useful evidences exploited
herein would not have emerged.
A simple empirical correlation was also proposed for limiting
activity coefficient as a function of the alkyl chain length. This is
significant since in previous work, the description of change with the alkyl
chain length has been merely qualitative. As a matter of fact, the natural
logarithm of was found to vary linearly with the carbon number of the alkyl
chain attached to the methylpyrrolidinium or methylimidazolium group. The same
applied to limiting selectivity. The model correlated well and limiting
selectivity with the cation alkyl chain length for
systems involving non-polar components as solutes. Poor
correlation coefficients were however observed for polar solutes. This outcome
can be exploited when developing a priori predictive thermodynamic models.
A detailed analysis of infinite dilution activity
coefficient, selectivity and capacity data obtained from this work and the
literature, allowed to derive the following trends, in relation to six
separation problems selected among the most interesting ones for the chemical
industry:
· n-hexane (1)/benzene (2) system:
Pyridinium and imidazolium-based FILs were potentially the best
suitable solvents for this separation problem. The best selectivity was
obtained with [EMIM] [BF4]. The ionic liquid [BMPy] [BF4] presented the best
compromise between selectivity and capacity at infinite dilution. Higher
selectivities were obtained with short-chained cations and small anion
volumes.
· Methanol (1)/benzene (2): [EMIM]
[TFA] and [3C6C14P] [(C2F5)3PF3] were found to be the best solvents for this
separation problem, depending on the stream in which methanol was to be
collected i.e. overhead or bottom stream. Phosphonium-based fluorinated ionic
liquids with large anion volume and an appropriate cation would be more
selective. FILs with short alkyl chain length for the cation and a small anion
volume were generally the least favorable ones.
· Methanol (1)/acetone (2): Limiting
capacity and selectivity varied in the same manner as for the previous system
except in case of ammonium-based FILs. [BMIM][SbF6] and [3C6C14P]
[Tf2N] lead to the highest selectivity and performance index respectively.
· n-Hexane (1)/ hex-1-ene (2):
Regardless of the class of the ionic liquid, higher selectivities were
obtained with longer alkyl chains attached to the cation and smaller anions
with sterical shielding effect around the anion charge centre. The best
selectivity was displayed by the ionic liquid [BMPyrr] [TfO]. It was seen that
the ionic liquid [3C6C14P] [(C2F5)3PF3] lead to the best compromise between
selectivity and capacity.
· Benzene (1)/ butan-2-one (2): The
best selectivity as well as the best performance index was obtained with [EMIM]
[Tf2N]. With increasing anion volume, the limiting selectivity was found to
increase. And the longer the alkyl chain of the cation, the smaller the
selectivity. Due to insufficient database, no reliable conclusion could be
reached for phosphonium-based FILs.
· Ethanol (1)/ butan-2-one (2):
[3C6C14P] [Tf2N] was found to be the best potential solvent
for this separation problem on the basis of both selectivity and performance
index. It was found that fluorinated ionic liquids consisting of long-chained
cations and big-sized anions were potentially the best separation agents for
this system.
Judging from performance index values it was observed that
the ionic liquid [3C6C14P] [Tf2N] would be a reasonable separation
solvent for all the investigated mixtures, except the n-Hexane/ Benzene system.
In some cases, no comprehensive conclusion could be made due to insufficient or
incoherent data. Nevertheless, where comparison was possible, the findings were
consistent with previously published results derived from COSMO-RS (Kumar and
Banerjee, 2009 and Lei et al. 2006, 2007). It is an indication that the
approach consisting of using experimental data was at once economic, less time
consuming and qualitatively effective.
These results emphasize the important role played by the
nature of ions comprising the ionic liquid in determining the magnitude of
activity coefficient values and consequently its effectiveness as separation
solvent. Chances to have a FIL which would act as a universal separation agent?
are very slim. For a given separation problem, there is a specific profile for
appropriate ionic liquid solvents.
The importance of this study lies in its contribution to the
understanding of how structure influences ionic liquids selectivity and
capacity in different separation problems. It expands the knowledge about FILs
which are potentially green separation agents. Additionally, the reported
infinite dilution activity coefficient data can be used to extend the
applicability range of predictive thermodynamic models. Such are UNIFAC, and
modified UNIFAC models which are already incorporated in some chemical
engineering process simulators.
Due to the limited nature of the data set used in this work, the
following was not addressed:
· The effect of the cation for phosphonium-based FILs on
limiting activity coefficient, selectivity and capacity;
· The effect of the anion for ammonium -based FILs;
· The effect of both cation and anion in other types of
ILs such as quinolinium, guadinium, sulfonium, piperidinium and
thiazolium-based FILs (though they are less popular in separation-related
studies);
· The effect of introducing functional groups other than
n-alkyl in the cation of different FILs.
These topics are worthy of interest for future work. They will
require more experimental data or the use of the quantum approach.
The calculated relative error on data obtained by the IGST was
larger than most values
reported in the literature, due to high uncertainties in
determining the flow-rate. Its stability left much to desire. Replacing the
flow-regulator is to be considered for more accurate measurements.
