CHAPTER SIX: DISCUSSION
6.1. Fluorinated Ionic Liquids investigated in this work
6.1.1. Gas-Liquid Chromatography
6.1.1.1. Phosphonium-based Ionic Liquids
Infinite dilution activity coefficients of various organic
solutes in the ionic liquids [3C6C14P] [Tf2N], [3C6C14P]
[BF4], [3C6C14P] [PF6] are listed in tables 5-2 through 5-4,
5-6 through 5-8 and 5-10 through 5-12. For two different solvent loadings IDAC
values obtained with the same solutes generally remained close to each other.
It is an indication that no adsorption took place onto the column packing, even
for polar solutes such as ketones and alcohols. For all the three
phosphonium-based ionic liquids, infinite dilution activity coefficients
decrease in the order: nalkanes alk-1-enes alcohols > cycloalkanes >
alk-1-ynes > alkylbenzenes > ketones. IDAC values for ketones were the
smallest for all the investigated solutes, an indication of strong
solute-solvent interactions. This can be attributed to the interaction between
the two pairs of electrons on the oxygen atom of the ketone with the cation of
the IL, as well as between the positive pole of the ketone and the ionic
liquid`s anion. Banerjee and Khanna (2006), Letcher et al. (2008) and Revelli
et al. (2009) reported infinite dilution activity coefficients of various
organic solutes in [3C6C14P] [Tf2N] and/or [3C6C14P]
[BF4]. In this work, a larger number of solutes have been considered.
Additionally, IDAC data of organics in the ionic liquid [3C6C14P][Tf2N]
obtained from the inert gas stripping technique are presented. Tables 6-1 and
6-2 allow a quick comparison of experimental infinite dilution activity
coefficient data obtained in this work to literature data. Results from this
study are in good agreement with those published by Letcher et al. (2008) and
Revelli et al. (2009). However, there are large discrepancies between data
reported in this work and those obtained by Banerjee and Khanna (2006). This
may be due to different purities as far as ionic liquids samples are concerned.
Moreover, the methodological approach used by Banerjee and coworkers is
different from the one used in this study.
During experiments, it was observed that alcohols led to very
long retention times. Resulting peaks were so broad that a very accurate
determination of the retention time was not possible. This is the reason why
the linear regression of experimental data shown in Figures 5-5 and 5-13 was
not convincingly successful.
Table 6-1: Comparison of experimental IDACs
in the ionic liquid [3C6C14P] [Tf2N] at 313.15 K
from this work
to literature data;*Interpolated data.
Solute
|
*Banerjee and Khanna
(2006)
|
Letcher et al.
(2008)
|
*Revelli et al.
(2009)
|
This
work
|
n-Hexane
|
0.759
|
1.130
|
1.064
|
1.096
|
Hex-1-ene
|
0.498
|
0.920
|
0.889
|
0.907
|
Cyclohexane
|
0.611
|
0.830
|
0.802
|
0.795
|
Methanol
|
0.951
|
1.230
|
1.128
|
1.083
|
Benzene
|
0.405
|
0.400
|
0.392
|
0.391
|
Butan-2-one
|
-
|
-
|
0.270
|
0.321
|
|
Table 6-2: Comparison of experimental IDACs
in the ionic liquid [3C6C14P] [BF4] at 313.15 K
from this work to
literature data; * Interpolated data.
Solute
|
*Banerjee and
Khanna (2006) This work
|
n-Hexane
|
0.956
|
1.400
|
Hex-1-ene
|
0.565
|
1.109
|
Cyclohexane
|
0.725
|
0.975
|
Methanol
|
6.524
|
0.537
|
Benzene
|
0.369
|
0.410
|
|
Partial molar excess enthalpies at infinite dilution were
calculated from the plots of
as a function of provided in Figures 5-1 through 5-7, 5-9
through 5-15 and 5-17
through 5-23, using the Gibbs-Helmholtz equation:
(2-11)
Tables 5-5, 5-9 and 5-13 show that partial molar excess
enthalpies at infinite dilution are
generally small. Thus, limiting activity
coefficients values vary little with temperature. (The
greatest variation was 0.023 K-1). Positive values of
partial molar excess enthalpies imply
that infinite dilution activity coefficients decrease with
increasing temperature. In addition, positive infinite dilution partial molar
excess enthalpies indicate that dissociation effects outweigh association
effects in very dilute mixtures involving the ionic liquid and the solute under
consideration. Negative infinite dilution partial molar excess enthalpies
indicate that association effects outweigh dissociation effects. Figures 5-8,
5-16 and 5-24 show that
extending the solute alkyl chain weakens its interaction with
the ionic liquid as values
increase when the number of carbon atoms is increased.
Table 5-32 shows selectivities and capacities at infinite
dilution for selected ionic liquids and some industrial solvents in relation
with n-hexane/benzene, n-hexane/hex-1-ene and methanol/acetone separation
problems. It appears that the studied phosphonium-based fluorinated ionic
liquids are poor solvents for separating aromatics from aliphatics. However,
they are fairly more selective when it comes to the methanol/benzene and
methanol/acetone separation problems. For example, the limiting selectivity
value of the ionic liquid [3C6C14P] [Tf2N] for the methanol/ acetone mixture
was calculated as 3.85 at 313.15 K, with the limiting capacity found to be 2.3.
These values lead to one of the best selectivity-capacity combinations at
infinite dilution for ionic liquids reported in the open literature.
6.1.1.2. Ammonium-based Ionic Liquid.
Experimental limiting activity coefficients of the
investigated solutes in the ionic liquid [C13C8N] [Tf2N] are listed in Tables
5-14 through 5-16. They decrease in the following order: n-alkanes >
alk-1-enes > alcohols > cycloalkanes > alk-1-ynes > alkylbenzenes
> ketones. Alk1-ynes, alkylbenzenes and ketones have low IDAC values in the
ionic liquid. This is due to the interaction between the electrons of these
unsaturated inorganics and the cation of the ionic liquid [C13C8N] [Tf2N].
From Figure 5-32, it can be seen that for all classes of
solutes, except ketones, IDAC values increase with the increasing number of
carbon atoms in the structure of the solutes. The magnitude of solute-solvent
interactions which is related to the infinite dilution activity coefficient is
determined by the solute polarity. The more polar the solute, the stronger the
interaction and the smaller the limiting activity of the solute in the ionic
liquid.
Temperature-dependence of IDAC values is depicted by Table
5-17 obtained from Figures 5-25 through 5-31. IDACs of unsaturated organic
solutes, except alk-1-enes increase with increasing temperature. Thus, they
exhibit negative partial molar excess enthalpies. Limiting selectivities and
capacities displayed in Table 5-32 reveal that similar to the three phosphonium
ionic liquids discussed in the previous section, [C13C8N] [Tf2N] exhibits poor
separation performance for the n-hexane/benzene system. It is likely to
represent a fair separation agent for ketones/alcohols and aromatic
compounds/alcohols mixtures.
6.1.1.3. Imidazolium-based Ionic Liquids.
Infinite dilution activity coefficients and partial molar
excess enthalpies of different organic
solutes are provided in tables 5-18
through 5-21 for [BMIM] [SbF6], 5-22 through 5-25 for
[EMIM] [TfO], 5-26
through 5-29 for [MOIM] [PF6]. Polar solutes have strong interaction with
the investigated imidazolium-based ionic liquids. The following
hierarchies in the variation of IDAC values have been observed:
· [BMIM] [SbF6]: n-alkanes > cycloalkanes alk-1-enes
> alcohols > alk-1-ynes >
alkylbenzenes > ketones;
· [EMIM] [TfO]: n-alkanes > cycloalkanes alk-1-enes >
alk-1-ynes > alkylbenzenes > alcohols;
· [MOIM] [PF6]: n-alkanes > alk-1-enes cycloalkanes >
alcohols alk-1-ynes >
alkylbenzenes.
For all three imidazolium-based ionic liquids, infinite dilution
activity coefficients increase with increasing number of carbon atoms as shown
in Figures 5-40, 5-47 and 5-54.
From the plots presented in Figures 5-33 through 5-39 for
[BMIM] [SbF6], 5-41 through 5-46 for [EMIM] [TfO] and 5-48 through 5-53 for
[MOIM] [PF6], the partial molar excess enthalpy at infinite dilution of all
investigated unsaturated solutes, except alk-1-enes was found to be
negative.
These three ionic liquids are potentially effective
separation agents for aromatic/aliphatic compounds mixtures (Table 5-32). Their
limiting selectivity and capacity data are high compared to NMP and
sulfolane.
6.1.2. The inert gas stripping technique
Mixtures of n-hexane and cyclohexane as solutes in NMP were
used as test systems. Measurements were carried out in the temperature range
from 303.15 K and 323.15 K at three different stripping gas flow rates.
Experimental results presented in table 5-30 were within 2 % of published
literature data. Good agreement was also observed between data obtained from
the gas-liquid chromatography and the dilutor method. Table 5-31 lists infinite
dilution activity coefficients of six solutes, each one representing a
functional group, in the ionic liquid [3C6C14P] [Tf2N]. A comparison, in term
of relative deviations was made to assess the extent of the agreement between
the two experimental techniques used in this work with exactly the same
systems. It was found that experimental results agreed within approximately 3.4
%. This is significant as not only the reliability of the newly constructed gas
stripping set up was confirmed but also doubts on the validity of the GG
experimental procedure were dispelled. In effect, the open literature (Mutelet
and Jaubert 2006) warns of large experimental errors in the GC method when
polar solutes are investigated due to adsorption of the latter onto the column
packing and probably onto the inner walls of the GC column. To minimize
adsorption effects, large mass fractions of the ionic liquids were used during
the column packing stage. This
strategy was effective since even for methanol and acetone
results from the GC and the dilutor techniques are not very different, and are
within experimental errors.
Table 6-4 suggests that the uncertainty in determining the
flow rate was the largest contributor to the overall error of infinite dilution
activity coefficient data. Apart from the inability of the flow regulator to
stabilize the inert gas flow rate over a long period of time, no major
difficulties were encountered. All minor problems were sorted out as soon as
they had arisen. Using the dilutor technique, a typical run with flow rates
around 15 cm3. min.-1 could last three and eight hours
for results displayed in tables 5-30 and 5-31 respectively. An experiment was
terminated when, according to Krummen et al. (2000), around 15 % of the solute
was stripped out of the system. It took a month and half to generate these data
when, for the same systems and number of runs, the GLC equipment used in this
study would require only a single day. The inert gas stripping technique is
known as a more time consuming technique than GLC. These durations are not
particularly longer than those reported by other authors. In order to decrease
the experimental time, the first attempt consisted of increasing the inert gas
flow rate. The best compromise between thermodynamics and kinetics of the
stripping process was observed in the neighborhoods of flow rates about 23
cm3.min-1. Beyond this value, infinite dilution activity
coefficient values tended to depend on the stripping gas flow rate, an
indication that thermodynamic equilibrium was not achieved in the cell. For
similar systems involving NMP, Krummen et al. (2000) took only 90 minutes to
achieve a run. They used a dilutor cell with larger volume and probably higher
height than the one constructed for this study, and flow rates as high as 20 to
40 cm3.min-1.
6.1.3. Error estimation
Uncertainties in determining the various quantities required
to calculate activity coefficients at infinite dilution are provided in tables
6-3 and 6-4. The presented values were obtained by taking into account the
accuracy of sensors as stated by suppliers, the stability of experimental
parameters and deviations of parameters under different experimental
conditions. The procedures described by George (2008) were used in this work to
assign errors to mass, gas flow rate, as well as temperature measurements. As
suggested by Harris (2000), the relative standard deviation of experimental
variables were calculated according to the method presented by Skoog et al.
(1996). Table 6-5 gives the overall relative errors calculated using the law of
error propagation applied to equations (2-3), (2-4), (2-11), (3-9) and (3-88).
Additional details are provided by the experimental procedures given in Chapter
four.
Table 6-3: Uncertainties on experimental
parameters for the GLC method.
Number of moles for the solvent,
|
#177;0.03
|
%
|
Oven temperature,
|
#177;0.35
|
%
|
Inlet pressure
|
#177;0.50
|
%
|
Outlet pressure
|
#177;0.30
|
%
|
Carrier gas flow rate
|
#177;0.20
|
%
|
Saturation vapour pressures
|
#177;0.30
|
%
|
Net retention volume
|
#177;3.00%
|
Difference of retention times (Solute - inert gas)
|
#177;0.20 %
|
|
Table 6-4: Uncertainties on experimental
parameters for the dilutor method.
Number of moles for the solvent,
|
#177;0.03
|
%
|
System temperature,
|
#177;0.10
|
%
|
Dilutor cell pressure
|
#177;0.03
|
%
|
Stripping gas flow rate
|
#177;3.33
|
%
|
Saturation vapour pressures
|
#177;0.30
|
%
|
Saturation fugacity coefficients
|
#177;0.50
|
%
|
Flow-meter pressure
|
#177;0.03
|
%
|
|
Table 6-5: Overall uncertainties on
experimental data and derived quantities.
Infinite dilution activity coefficient (GLG method)
|
#177;3.68
|
%
|
Infinite dilution activity coefficient (Dilutor method)
|
#177;4.26
|
%
|
Molar excess enthalpy at infinite dilution
|
#177;7.66
|
%
|
Selectivity at infinite dilution
|
#177;7.36
|
%
|
Capacity at infinite dilution
|
#177;3.68
|
%
|
|
6.2. Limiting activity coefficients of fluorinated ionic
liquids
It was found, after an extensive literature survey, up to
October 2009, experimental infinite dilution activity coefficients of 37
fluorinated ionic liquids, including the ones studied in this thesis, were
available (See Table 6-6). In this section, experimental data, others than
those generated in this work, were taken from various research papers published
in the following journals:
· The Journal of Chemical Thermodynamics
· The Journal of Chemical and Engineering Data
· Fluid Phase Equilibria
· The Journal of Physical Chemistry B
· The Journal of Chromatography A
Numbers which are assigned to ionic liquids in table 6-6 are
used as x-coordinates in the plots presented in Figures 6-1 through
6-3. The fluorinated ionic liquids 1-butyronitrile-3- methylimidazolium bis
(trifluoromethylsulfonyl) imide ([CpMIM] [Tf2N]), 1-butyronitrile-2, 3-
dimethylimidazolium bis(trifluoromethylsulfonyl) imide ([CpMMIM][Tf2N]),
1-hexyloxymethyl-3-methylimidazolium bis
(trifluoromethylsulfonyl) imide ([H-O-MIM]
[Tf2N] , 2009b) and 1,3-dihexyloxymethyl-imidazolium
bis(trifluoromethylsulfonyl)-imide ([DH-O-MIM] [Tf2 , 2009b)
are not discussed in this dissertation due to insufficient
database for comparison.
Table 6-6: List of fluorinated ionic liquids
investigated in the literature and assigned numbers.
No. Ionic Liquid No. Ionic Liquid No. Ionic Liquid
1
|
[EMIM][BF4]
|
14
|
[C16MIM][BF4]
|
27
|
[BMIM][SbF6]
|
2
|
[EMIM] [Tf2N]
|
15
|
[3C6C14P][BF4]
|
28
|
[BMIM][PF6]
|
3
|
[MMIM][Tf2N]
|
16
|
[3C6C14P][Tf2N]
|
29
|
[EMIM][TFA]
|
4
|
[BMIM][BF4]
|
17
|
[3C6C14P][(C2F5)3PF3]
|
30
|
[HMIM][TfO]
|
5
|
[BMIM][Tf2N]
|
18
|
[3C1C4N][Tf2N]
|
31
|
[BMPyrr][TfO]
|
6
|
[BMIM][TfO]
|
19
|
[BMPy][BF4]
|
32
|
[HMPyrr][Tf2N]
|
7
|
[MMPIM][BF4]
|
20
|
[BMPyrr][Tf2N]
|
33
|
[OMPyrr][Tf2N]
|
8
|
[EMMIM][Tf2N]
|
21
|
[Et3S][Tf2N]
|
34
|
[CpMIM][Tf2N]
|
9
|
[HMIM][BF4]
|
22
|
[Epy][Tf2N]
|
35
|
[CpMMIM][Tf2N]
|
10
|
[HMIM][PF6]
|
23
|
[3C6C14P][PF6]
|
36
|
[H-O-MIM][Tf2N]
|
11
|
[HMIM][Tf2N]
|
25
|
[EMIM][TfO]
|
37
|
[DH-O-MIM][Tf2N]
|
13
|
[MOIM][Tf2N]
|
26
|
[MOIM][PF6]
|
|
|
12
|
[MOIM][BF4]
|
24
|
[3C8C1N][Tf2N]
|
|
|
|
Chapter 6: Discussion
| 
