Ionic liquid silver salt complexes for propene/propane separation

Article abstract of DOI:10.1039/c0cp01104e

Properties of the room-temperature liquid complex salt [Ag(propene) _x][Tf₂N] have been studied to probe its suitability for acting as active separation layer in immobilised liquid membrane (ILM) concepts for propane/propene separation. The pressure/temperature range of complex formation has been determined and the thermal properties of Ag[Tf₂N] and [Ag(propene)_x][Tf₂N] have been studied by DSC (differential scanning calorimetry) and TGA (thermogravimetric analysis) measurements. Pressure dependent measurements of solubility and diffusivity showed that the observed membrane selectivity is dominated by the solubility selectivity. The self-diffusion coefficient of propene is always smaller compared to propane as propene is temporarily bound to the silver ion in the [Ag(propene)_x][Tf₂N] ionic liquid.

Full text of DOI:10.1039/c0cp01104e

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/pccp | Physical Chemistry Chemical Physics
Ionic liquid silver salt complexes for propene/propane separationw
Friederike Agel,*a Fee Pitsch,b Florian Felix Krull,b Peter Schulz,a
Matthias Wessling,b Thomas Melinb and Peter Wasserscheida
Received 7th July 2010, Accepted 4th October 2010
DOI: 10.1039/c0cp01104e
Properties of the room-temperature liquid complex salt [Ag(propene)x][Tf2N] have been studied to
probe its suitability for acting as active separation layer in immobilised liquid membrane (ILM)
concepts for propane/propene separation. The pressure/temperature range of complex formation
has been determined and the thermal properties of Ag[Tf2N] and [Ag(propene)x][Tf2N] have been
studied by DSC (differential scanning calorimetry) and TGA (thermogravimetric analysis)
measurements. Pressure dependent measurements of solubility and diffusivity showed that the
observed membrane selectivity is dominated by the solubility selectivity. The self-diffusion
coefficient of propene is always smaller compared to propane as propene is temporarily bound to
the silver ion in the [Ag(propene)x][Tf2N] ionic liquid.
In liquid membranes the membrane selectivity is primarily
1. Introduction
dominated by the selectivity of solubility Ssol
. Liquid
The separation of olefin–paraffin gas mixtures by low-
temperature distillation is one of the most important but
also most energy-consuming separation processes in the
petrochemical industry.1 Therefore, the development of
alternative separation methods such as absorption/desorption
or membrane separation technologies is highly desired.
In general, membrane separation technologies have several
advantages over adsorption/desorption processes, such as
the possibility to work in a continuous operation mode at
constant temperature and pressure. This results in significantly
less complex process designs with lower investment in moving
parts, pumps and process inventory.
membranes can be high-permeable and high-selective as
diffusion in liquids is much faster than that in solids
(e.g. polymers) and as it is possible to functionalise liquid
membranes with selectivity-enhancing carriers. The selectivity
improvement is caused by an increased solubility of one of the
components to be separated. Possible carriers for olefin/
paraffin separations (via liquid membrane as well as via
absorption/desorption) are copper(I) and silver(I) salts as these
metal ions are able to form p-complexes with olefins reversibly,
that means the olefin can be liberated again through temperature
or pressure changes.4 One problem with copper(I) salts is their
instability towards disproportionation or oxidation if humidity
or oxidisers are present. The concept to use copper– and
silver–olefin complexes for ene/ane separation technologies is
not new, but many examples using these salts dissolved in
molecular solvents like water, glycerol or polyalcohols suffer
from the problem of selectivity loss over time due to solvent
evaporation.5
Apart from longterm stability of the membrane itself the
economic efficiency of membrane processes is determined by
the membrane selectivity and the membrane productivity (i.e. the
flux that can be realized through the membrane). For the
separation of olefin/paraffins, selectivity is the most important
factor as low capacity can be compensated by a bigger membrane
area but low selectivity requires a multi level process.2
Functionalised ionic liquids (ILs) provide the opportunity
to develop liquid membranes avoiding solvent loss problems
due to the very low vapour pressure of ionic liquids.6 First
examples making use of this concept have appeared in the last
years. The well-known fact that Ag[BF4] forms solid
coordination compounds with olefins7 has been applied by
Ortiz et al. to investigate solutions of this compound in
Transport in liquid membranes follows the solution/diffusion
mechanism, where the permeability P is given by the product of
solubility coefficient K and diffusion coefficient D.3
P = KꢀD
(1)
The membrane selectivity Smem is given by the ratio of the
permeabilities of the components to be separated.
[C4C1Im][BF4] for propene/propane separation using
a
membrane contactor device.8 Very recently, the group of
Meindersma studied several silver salts dissolved in ionic
liquids for ethene/ethane separation.9
P1 K1 ꢀ D1
¼
D1
D2
Smem
¼
¼ Ssol
ð2Þ
P2 K2 ꢀ D2
a Chemical Reaction Engineering, Friedrich-Alexander-University
Erlangen-Nuremberg, Egerlandstrasse 3, D-91058 Erlangen,
Germany. E-mail: friederike.agel@crt.cbi.uni-erlangen.de;
Fax: +49 (0)9131 8527421; Tel: +49 (0)9131 8527424
b Aachener Verfahrenstechnik, Chemical Process Engineering,
RWTH Aachen University, Turmstrasse 46, D-52056 Aachen,
Germany
Both, liquid membrane separation and absorption/
desorption processes, need a high silver content in the liquid
for good selectivity and capacity. However, the maximum
silver content that can be incorporated into a classical ionic
liquid is limited by the solubility of the silver salt in the liquid
salt. Unfortunately, high loadings of silver salts close to the
saturation level result in a high viscosity of the solution leading
to slow mass transport. Another possibility to absorb olefins in
w Electronic supplementary information (ESI) available: NMR- and
IR-spectra of Ag[Tf2N] and [Ag(propene)x][Tf2N]. DSC, TGA and
ESI-MS of Ag[Tf2N]. See DOI: 10.1039/c0cp01104e
c
This journal is the Owner Societies 2011
Phys. Chem. Chem. Phys., 2011, 13, 725–731 725

View Article Online
liquid salts is to form systems that contain complex
[Ag(olefin)x]+ ions as the only cations. For example, Ag[NO3]
is known to form a liquid complex with propene but it is only
stable under pressures near the vapour pressure of propene
and temperatures below 36 1C.10
Room temperature liquid salts of the general type
[Ag(olefin)x][Tf2N] have been described in the literature by
Dai and coworkers for higher olefins (C5–C6).11 These
complex salts show surprisingly low viscosities even at room
temperature.
In the present paper we describe that the solid salt Ag[Tf2N]
forms a liquid complex salt with propene at room temperature
and ambient pressure. As we consider this ionic liquid system
to be a very promising candidate for the use in propene/
propane separation concepts, we report important thermo-
dynamic (solubility) and kinetic (diffusion) properties relevant
for potential future immobilized liquid membrane concepts.
Fig. 1 DSC of [Ag(propene)x][Tf2N], heating rate 5 K minꢁ1
.
2. Results and discussion
2.1 Synthesis and physical properties
In the earlier literature11 the preparation of [Ag(olefin)x][Tf2N]
(olefin = 1-pentene or 1-hexene) was carried out by mixing
aqueous solutions of Ag[NO3] and Li[Tf2N] in the presence of
the liquid 1-olefin (molar ratio 1 : 1 : 1). The hydrophobic
complex [Ag(olefin)][Tf2N] ionic liquid formed as a second
heavier phase and was washed with water to remove Li[NO3]
residues.11
Fig. 2 TGA of [Ag(propene)x][Tf2N], heating rates 5 K minꢁ1
.
However, to form the silver–propene complexes within our
work, this synthesis route was found to be unfavourable due to
the volatility of propene and generally moderate product
yields with respect to silver.
The complex ionic liquid [Ag(propene)x][Tf2N] is synthe-
sised by contacting the solid Ag[Tf2N] with gaseous propene.
The reaction is reversible. By purging the resulting liquid with
nitrogen, Ag[Tf2N] recrystallises.
Therefore, we utilized an alternative route to prepare the
[Ag(olefin)x][Tf2N], with nearly no loss of the high cost
components, silver-cation and Tf2N-anion, which is rather
similar to earlier literature procedures.12
The complex salt [Ag(propene)x][Tf2N] prepared under
normal pressure propene atmosphere shows a diffuse melting
range between ꢁ6 and +7 1C. Such a melting behaviour has
also been observed for [Ag(propene)x][NO3] by Francis who
ascribed the behaviour to the presence of different complexes
(x = 1 or 2) in the mixture.10 Crystallisation on cooling occurs
at ꢁ2.7 1C as shown in Fig. 1.
For this preparation route, freshly precipitated Ag2O is
dissolved in an aqueous solution of H[Tf2N] (Scheme 1). The
latter is obtained by sublimation from a mixture of Li[Tf2N]
and concentrated sulfuric acid. After removal of the water,
Ag[Tf2N] is isolated as a white solid. The yield is nearly
quantitative and the effective cleaning procedures of the
reactants (Ag2O by washing and H[Tf2N] by sublimation)
prevent contamination of the product with salt impurities.
Thermal analysis of Ag[Tf2N] indicates a melting point of
248 1C and fast thermal decomposition above 300 1C
(Tonset = 309 1C at 10 K minꢁ1 heating rate). As onset
temperatures in TGA measurements are only a very rough
indication of thermal stability we performed also a TGA
with isothermal steps. Up to 100 1C no mass loss can be
detected.
In the TGA (Fig. 2) a mass loss of B15% below 100 1C
implies that in [Ag(propene)x][Tf2N] x is about 1.7 at
atmospheric propene pressure.
The viscosity of the complex salt is independent of shear
rate and low compared to other ionic liquids involving the
[Tf2N]ꢁ-anion and organic cations. This is important, as
diffusion rates are inversely dependent on viscosity.
As expected, the viscosity is not only temperature dependent
(Fig. 3) but also propene pressure dependent. At ambient
propene pressure the viscosity is 40 mPas, at 10 bar propene
pressure viscosity drops to 15 mPas.
The surface tension of [Ag(propene)x][Tf2N] in propene
atmosphere under normal pressure was determined to be
29.6 mN mꢁ1
.
2.2 Complex stability
Crystallisation of Ag[Tf2N] leads to a significant volume
reduction from 256.7 cm3 molꢁ1 for the liquid complex
Scheme 1 Synthesis of Ag[Tf2N].
c
726 Phys. Chem. Chem. Phys., 2011, 13, 725–731
This journal is the Owner Societies 2011

View Article Online
Fig. 5 Van’t Hoff plot of the complex stability, T = 25–90 1C.
Fig. 3 Shear rate dependent viscosity at different temperatures.
Scheme 2 Complex dissociation reaction.
be calculated as DH1 = 29 ꢂ 2 kJ molꢁ1 and DS1 = 73 ꢂ
6 J molꢁ1 Kꢁ1
.
2.3 Gas solubilities
at 1 bar propene to 135.2 cm3 molꢁ1 for the solid silver salt. As
this volume contraction will lead unavoidably to membrane
leakage in the case of the ILM-application of this material,
crystallisation in the membrane is very undesirable. Therefore
it is of particular importance to know the thermodynamic
stability of the liquid complex. Scheme 2 shows the dissociation
reaction that leads to crystallisation.
Propene. At low propene pressure the propene uptake curve
in Fig. 6 shows a very sharp increase up to 2.0 bar that can be
ascribed to complex formation. Above 2.0 bar propene
pressure, the constant slope indicates a physical sorption of
propene in the liquid complex rather than a higher coordination
number of the silver ion. The concentration of propene can be
fitted with the following mathematical functions.
The reaction equilibrium depends on temperature and
propene pressure pen and can be described by the following
equations:13
for p(en) o 2 bar:
cen = 0.31ꢀln[170(pen ꢁ 0.05) + 1]
(6)
where 0.05 bar is the required minimum propene pressure for
complex formation at room temperature.
for p(en) > 2 bar:
Kp = pen/p0
(3)
(4)
(5)
DG1 = DH1 ꢁ TDS1 = ꢁRTꢀln Kp
ln (pen/p0) = ꢁDH1/(RT) + DS1/R
cen = 0.1646ꢀpen + 1.4710
(7)
Propane. In the system described here, the propane solubility
can only be measured in the presence of propene as the liquid
complex [Ag(propene)x][Tf2N] has to be generated first. As
can be seen from Fig. 7 the solubility of propane is much
smaller than that of propene and depends on propene pressure.
In the measured pressure range the propane concentration has
the following relationship to the partial pressures:
The propene partial pressures at which crystallisation takes
place were experimentally determined for temperatures
between 25 and 90 1C (Fig. 4).
The logarithmic plot of (pen/p0) against the reciprocal
temperature (in K) gives a linear relationship indicating that
the temperature dependencies of enthalpy and entropy can be
neglected (Fig. 5). From slope and intercept the values for the
standard reaction enthalpy DH1 and standard entropy DS1 can
can = 0.0041ꢀp2an + (0.0098ꢀpen + 0.0254)ꢀpan
(8)
Fig. 4 Propene pressure and temperature dependent crystallisation
Fig. 6 Pressure dependent propene uptake in mol propene molꢁ1
silver T = 22 1C (three series of measurements).
points.
c
This journal is the Owner Societies 2011
Phys. Chem. Chem. Phys., 2011, 13, 725–731 727

View Article Online
Fig. 7 Propane pressure dependent propane uptake at four different
propene pressures. T = 22 1C.
Fig. 8 Propene pressure dependent solubility selectivity at two
different propane pressures. T = 22 1C.
The more propene is coordinated around the silver ion the
more unpolar is the solvent character of the ionic liquid phase
and the solubility of propane rises. Thus, the selectivity of
solubility cannot be correctly determined in this case from
solubility measurement of the single gases.
Table 1 Selectivity and capacity of different ionic liquid/Ag systems
c(Ag)/
c(en)/
Systema
mol Lꢁ1 T/K Ssol mol Lꢁ1 Ref.
b
Ag[BF4]/[C4C1Im][BF4]/C3 b
Ag[OTf]/[C2C1Im][OTf]/C2
0.25
1.20
298 16 0.4
303 45
303 12 0.6
303 70 2.6
298 40c 6.1
8
9
9
9
—
—
2.4 Selectivity of solubility
b
Ag[Tf2N]/[C2C1Im][Tf2N]/C2b 0.45
Ag[Tf2N]/[C2C1Im][Tf2N]/C2 1.80
The selectivity Ssol is the ratio of the solubility coefficients K
for propene and propane. K is concentration divided by the
partial pressures.
b
d
Ag[Tf2N]/C3
3.90
a
b
All partial pressures = 1 bar. C2: ethene/ethane, C3: propene/
c
propane. Calculated from mixed gas measurements. This work.
d
Ken cen ꢀ pan
Ssol
¼
¼
ð9Þ
pressure until the minimum propene pressure for complex
formation is reached at 0.05 bar.
Kan pen ꢀ can
Table 1 compares the performance of different ionic liquid/
silver systems for olefin/paraffin separations. Remarkably, the
pure Ag[Tf2N] investigated in detail here shows very good
results in comparison with silver salt solutions in ionic liquids.
Solubility selectivity is not the highest, but in all the cited
references single gas measurements were performed in contrast
to our mixed gas experiments. Furthermore, our system has by
far the highest capacity which becomes understandable from
the diluting ionic liquids in the other literature examples.
Solubility selectivities are often calculated from single gas
measurements and therefore called ‘‘ideal’’ as mixed gas
selectivities can be lower in some cases. Scovazzo et al.
compared single- and mixed-gas permeability measurement
of CO2/CH4 and CO2/N2 with several ionic liquid membranes
without carrier for facilitated transport and found them to be
in good agreement for CO2 partial pressure up to 2 bar.14 But
according to Brennecke et al. the mole fraction of CO2 in
imidazolium–Tf2N ionic liquids is only about 0.06 (2 bar CO2
and 25 1C)15 and this relatively low value can explain the good
agreement in the mixed gas experiment. In our case, a
selectivity determination based on single gas solubilities is
not possible as a minimum pressure of 0.05 bar is necessary
to form the silver ionic liquid. Moreover, the use of the
[Ag(olefin)x][Tf2N] system leads to a much higher propene
uptake and involves a chemical reaction which both leads to a
dramatic change in ionic liquid properties. In our case the
mole fraction of propene at 2 bar and 25 1C is 0.64 which is
about one magnitude higher than that for the CO2 solubilities
reported by Brennecke. Furthermore, propene coordination
affects a change in the charge distribution at the silver ion so
that interaction with the [Tf2N]-anion is also influenced. The
change in the liquid’s polarity is clearly revealed by the fact
that olefin coordination to Ag[Tf2N] in aqueous solution leads
to phase separation of water and [Ag(olefin)x][Tf2N].11
2.5 Diffusion
For molecules of similar size, diffusion in liquid membranes
plays only a minor role for membrane selectivity. In carrier
mediated transport, however, diffusion selectivity may be
important as propene is bound to the silver complex during
transport while propane is not. Therefore, we determined the
self-diffusion coefficients of propene and propane in
[Ag(propene)x][Tf2N] at varying propene pressures applying
PFG-NMR (pulsed field gradient) measurements.16 Even
though self-diffusion coefficients do not fit to macroscopic
diffusion along a concentration gradient, still the ratio of the
self-diffusion coefficients should be an appropriate representation
of concentration driven diffusion of propene and propane in
[Ag(propene)x][Tf2N]. As the exchange between coordinated
and uncoordinated propene is fast compared to the NMR-time
scale, only one signal and therefore an average self-diffusion
coefficient is detected for propene.
Fig. 8 shows the mixed gas selectivity Ssol calculated with
eqn (6)–(8) and plotted against propene pressure for two
different propane pressures. Ssol is mainly dependant on the
propene partial pressure and increases with decreasing
Fig. 9 shows that the ratio of the self-diffusion coefficients of
propene and propane is always less than unity. Propene moves
c
728 Phys. Chem. Chem. Phys., 2011, 13, 725–731
This journal is the Owner Societies 2011

View Article Online
This means that the selectivity of solubility is diminished by
the ratio of diffusion coefficients to afford the selectivity of the
membrane. As a result selectivity Smem is best at low partial
pressures where the difference in solubilities between propene
and propane is the largest due to complexation of propene to
the silver carrier (Fig. 10). With increasing propane partial
pressure, the contribution of selective complexation based
propene transport diminishes, whereas the physical absorption
based reverse selective diffusion of propane over propene
counteracts and reduces the selectivity. However, for
the investigated pressure range, we predict the permeability
selectivity to be in favor of the propene.
Fig. 9 Propene pressure dependent self-diffusion coefficients of
3. Experimental
propene and propane.
3.1 Materials
much slower than propane although the molecules have nearly
the same size and shape. This is due to the coordination of
propene to the silver cation and its interaction with the large
[Tf2N]-anion whereas the propane can move freely in the liquid.
The shape of the ratio-curve can be explained by different
binding conditions of propene. Below 1 bar propene pressure,
the coordination number of propene around the silver ion drops
significantly and thus the shielding of the cation’s charge is
reduced. Interaction between cation and anion increases and
the ratio Den/Dan decreases. Above 2 bar propene pressure the
portion of uncoordinated propene increases as more and more
uncoordinated propene dissolves in the complex ionic liquid. As
the latter carries no charge there is no interaction with the anion
anymore and this part of the propene can move as freely as
propane and therefore much faster. The ratio Den/Dan can be
fitted by the following equation:
Silver nitrate was purchased from AppliChem (reinst Ph. Eur.,
USP), Sodium hydroxide (analysis grade) and lithium
bis(tri-fluoromethylsulfonyl)imide (99%) from Merck KGaA
and sulfuric acid (GPR Rectapur 98%) from BDH Prolabo
VWR. All chemicals were used without further purification.
Propene and propane were purchased from Linde AG in
2.5 quality.
3.2 Synthesis
Bis(trifluoromethylsulfonyl)imide H[Tf2N]. 106.2 g (0.37 mol)
lithium bis(trifluoromethylsulfonyl)imide Li[Tf2N] dissolved
in 140 ml conc. sulfuric acid were heated to 90 1C under
argon atmosphere. The formed bis(trifluoromethylsulfonyl)
imide H[Tf2N] was distilled under reduced pressure over a
curved glass tube into a (liquid nitrogen) cooled flask. The yield
was 102.7 g (99%).
Den
¼ 0:0449 ꢀ p3en ꢁ 0:1850 ꢀ pe2n þ 0:2608 ꢀ pen þ 0:2420
Silver bis(trifluoromethylsulfonyl)imide Ag[Tf2N]. 63.0 g
(0.37 mol) silver nitrate (AgNO3) were dissolved in 100 ml of
demineralised water and sodium hydroxide solution (20 w%)
was added until the pH-value reached 11.3. The dark brown
precipitate was filtered, washed several times with demineralised
water and was then put in a solution of 103.2 g bis(trifluoro-
methylsulfonyl)imide H[Tf2N] (0.99 equiv. to silver) dissolved
in 100 ml of demineralised water. After stirring for 2 h at
60 1C, the excessive silver hydroxide was filtered off at RT and
the water was removed at the rotary evaporator. The product
was dried at 50 1C under high vacuum. 141.2 g (99%) of silver
bis(trifluoromethylsulfonyl)imide Ag[Tf2N] was obtained as a
white solid (nmax(compressed powder)/cmꢁ1 734, 770w, 793,
1024, 1105, 1197, 1307, 1331; dF(376.14 MHz, d4-MeOD,
Cl3FC) ꢁ78.5; dC(100.52 MHz, d4-MeOD) 120.5 (q,
JCF = 321 Hz, CF3); dAg(18.60 MHz, d4-MeOD, capillary
with AgNO3 sat. in D2O) 51.8; m/z (ESI-MS negative mode)
582.8 (9%), 279.9 (100), 211.0 (6), 147.0 (42), 78.0 (11); m/z
(ESI-MS positive mode) 107.0 (100%), 108.9 (99), 124.9 (5),
126.9 (4), 142.9 (3), 145.0 (2).
Dan
ð10Þ
2.6 Membrane selectivity
According to the solution-diffusion model the selectivity of the
membrane Smem is described by the following equation:
Pen Ken ꢀ Den
¼
Den
Dan
Smem
¼
¼ Ssol
ð11Þ
Pan Kan ꢀ Dan
Silver bis(trifluoromethylsulfonyl)imide propene complex
[Ag(propene)x][Tf2N]. [Ag(propene)x][Tf2N] was prepared by
purging Ag[Tf2N] with propene gas. (nmax(liquid)/cmꢁ1 742,
765w, 792, 905, 946, 1012, 1059, 1128, 1189, 1345, 1371,
1414, 1433, 1455, 1591; dH(399.75 MHz, no solvent, capillary
Fig. 10 Propene pressure dependent membrane selectivity at two
different propane pressures. T = 22 1C.
c
This journal is the Owner Societies 2011
Phys. Chem. Chem. Phys., 2011, 13, 725–731 729

View Article Online
equation with two virial coefficients. Several pressure steps
were done consecutively until the total pressure reached
8–9 bar. For the first pressure step, where the liquid complex
is formed out of solid Ag[Tf2N] the volume dilatation is taken
into account.
Scheme 3 Assignment of protons in propene.
For the solubility measurement of propane different initial
pressures of propene were adjusted at first in order to get the
liquid complex. After that, the liquid is pressurised with
propane as mentioned above.
with C6D6) 1.692 (3H, d, JCD 6.59 Hz, HD), 4.8–5.3 (2H, m,
H
A & HB), 6.2–6.37 (1H, m, HC) (Scheme 3); dF(376.14 MHz,
no solvent, capillary with Cl3FC) ꢁ78.4; dC(100.52 MHz, no
solvent, capillary with C6D6) 19.4 (CH3), 104.8 (CH2), 119.9
(q, JCF = 321 Hz, CF3), 135.6 (CH); dAg(18.60 MHz, no
solvent, capillary with AgNO3 sat. in D2O) 558.39.
NMR. NMR experiments were carried out at a Jeol ECX
400 MHz spectrometer with a 2-channel (HF, LF)-probe.
Ag[NTf2] is measured in d4-methanol (reference for 13C) with
soluted CCl3F (reference for 19F) and a sealed capillary
of AgNO3 saturated in D2O (reference for 109Ag).
[Ag(propene)x][NTf2] is measured without solvent, once with
a sealed capillary containing d6-benzene (reference for 1H and
13C) and CCl3F (reference for 19F) and once with a capillary
containing AgNO3 saturated in D2O (reference for 109Ag). To
the AgNO3 solution a very small amount of Fe(acac)3 is added
in order to reduce the relaxation time of the silver nucleus and
get a higher signal.
3.3 Analysis
Infrared spectra. ATR-IR spectra were recorded on a Jasco
FT/IR-4100typeA spectrometer with a resolution of 2 cmꢁ1
.
ESI-MS. Electrospray ionisation mass spectra are measured
on a Q-trap 2000 system from applied biosystems.
Thermal analysis. Differential scanning calorimetry (DSC):
The differential scanning calorimetry (DSC) was carried out
in sealed aluminium pans using a Netzsch DSC 204.
Thermo gravimetric analysis (TGA):
NMR samples under pressure were prepared in a pressure/
vacuum valve NMR tube from Wilmad LabGlass (Part No.
522-PV-7).
A Netzsch TG 209 F3 was used to determine the decom-
position temperatures.
The procedure of bpp-led-DOSY-NMR (bpp-led: bipolar
pulse pairs stimulated echo longitudinal eddy current delay,
DOSY: diffusion ordered spectroscopy) for determination of
self-diffusion coefficients followed published procedures.16
The attenuation of the echo signal is related to the
experimental parameters by the Stejskal–Tanner equation:18
Viscosity. The viscosity was measured using the Anton Paar
Physica MCR 100 rheometer with a CC25/Pr150/In/A1/SS
pressure cell.
Density. The density of the liquid complex was measured
with a 5 ml pycnometer filled under propene atmosphere
(B1 bar). The density of Ag[Tf2N] was measured by helium
pycnometry (Pycnomatic from Thermo Electron Corporation)
at 22 1C.
ꢀ
ꢁ
ꢀ
ꢁ
S
d
3
ln E ¼ ln
¼ g2g2Dd D ꢁ
ð12Þ
S0
S is the stimulated echo signal intensity, S0 is the amplitude in
the absence of diffusion, g is the gyromagnetic ratio, g is the
magnitude of the field gradient, D is the diffusion coefficient, d
is the duration of the field gradient, and D is the interval
between the gradient pulses. D was kept constant at 100 ms,
d was adapted from 1 to 3 ms, and g was varied in the range
of 0.01 to 0.61 T mꢁ1. The temperature was kept constant at
25 1C.
Surface tension. The surface tension was measured by the
drop volume method in propene atmosphere at 22 1C and
ambient pressure.17
Complex stability. A small amount of Ag[NTf2] was placed
in a test tube which was heated in an oil bath. Mixtures of
propene and nitrogen with different volume ratios (regulated
by two mass flow controllers) were purged through the tube
with a total flow of B20 ml minꢁ1. Total pressure was
atmospheric pressure (0.93–0.95 bar). For each temperature
the partial pressure of propene at which crystallisation took
place was determined. The errors in temperature and propene
partial pressure are ꢂ 0.5 1C and ꢂ 0.005 bar. Standard
conditions are T = 298.15 K and p0 = 1 atm (1.013 bar).
4. Conclusions
The complex ionic liquid [Ag(propene)x][Tf2N] shows very
promising properties that qualify the material for its application
as liquid membrane in propene/propane separations.
The equilibrium propene pressures for complex formation
at ambient temperatures are low enough to form a stable
liquid membrane with a sufficiently high selectivity. As the
complex silver salt is not diluted in other solvents like water,
alcohols or ionic liquids, its silver content is very high.
Importantly, the system contains no volatile membrane
components that can get lost during the separation process
or contaminate the gas streams.
Gas solubility measurements. Gas solubility measurements
were run in a steel autoclave equipped with a valve for vacuum
and gas supply, a magnetic stirring bar and a digital pressure
sensor (GMSD 35 BAE; Greisinger electronic GmbH). All
measurements were run at room temperature (22 1C).
For all measurements a weighted amount of Ag[Tf2N] was
placed in the autoclave, evacuated for two hours, charged with
propene pressure by quickly opening and closing the valve and
then equilibrated for 30 min. From the pressure drop the
dissolved amount of gas was calculated with the virial
Despite the high silver content of the liquid, its viscosity is
quite low and therefore provides very interesting mass
transport properties. Due to the higher silver content the
c
730 Phys. Chem. Chem. Phys., 2011, 13, 725–731
This journal is the Owner Societies 2011

View Article Online
propene capacity is much higher compared to those measured
for silver salt solutions in standard ionic liquids.8,9 Despite this
remarkably high capacity, the solubility selectivity is as high as
40 at 1 bar propene and propane pressure. Moreover, our
mixed gas sorption measurements showed that a high propene
content leads to remarkably higher propane solubilities.
Therefore, single gas measurements are not appropriate to
determine the solubility selectivity for liquids with a high silver
content.
1943, 65, 227; J. Lucas, F. W. Billmeyer (Jr) and D. Pressman,
J. Am. Chem. Soc., 1943, 65, 230; F. R. Hepner, K. N. Trueblood
and H. J. Lucas, J. Am. Chem. Soc., 1952, 74, 1333;
K. N. Trueblood and H. J. Lucas, J. Am. Chem. Soc., 1952, 74,
1338; A. E. Comyns and H. J. Lucas, J. Am. Chem. Soc., 1957, 79,
4339.
5 D. J. Safarik and R. B. Eldridge, Ind. Eng. Chem. Res., 1998, 37,
2571; S. Duan, A. Ito and A. Ohkawa, J. Membr. Sci., 2003, 215,
53; A. S. Kovvali, H. Chen and K. K. Sirkar, Ind. Eng. Chem. Res.,
2002, 41, 347; UK Pat., 2 169 301, 1984; US Pat., 6 147 273, 2000.
6 M. J. Earle, J. M. S. S. Esperanc¸ a, M. A. Gilea, J. N. Canongia
Lopes, L. P. N. Rebelo, J. W. Magee, K. R. Seddon and
J. A. Widegren, Nature, 2006, 439, 831; P. Wasserscheid, Nature,
2006, 439, 797; J. P. Armstrong, C. Hurst, R. G. Jones, P. Licence,
K. R. J. Lovelock, C. J. Satterley and I. J. Villar-Garcia, Phys.
Chem. Chem. Phys., 2007, 9, 982.
7 H. W. Quinn and D. N. Glew, Can. J. Chem., 1962, 40, 1103;
H. W. Quinn, J. S. McIntyre and D. J. Peterson, Can. J. Chem.,
1965, 43, 2896.
With propene pressure dependent NMR measurements of
the self-diffusion coefficients of propene and propane, we
could demonstrate that membrane selectivity is diminished
compared to solubility selectivity. The reason is that propane
has a higher mobility than propene in the ionic liquid as the
latter is bound to the silver cation and interacts in this form
stronger with the Tf2N-anion.
8 A. Ortiz, A. Ruiz, D. Gorri and I. Ortiz, Sep. Purif. Technol., 2008,
63, 311.
9 L. M. Galan Sanchez, G. W. Meindersma and A. B. Haan, Ind.
´ ´
In summary, we think that [Ag(propene)x][Tf2N] is
excellently suited as liquid material for ILMs for the separation
of light ene/ane mixtures. Experiments to apply the system in
real membrane separation processes are on-going in the group
of Melin and Wessling at RWTH Aachen University.
Eng. Chem. Res., 2009, 48, 10650–10656.
10 A. W. Francis, J. Am. Chem. Soc., 1951, 73, 3709.
11 J.-F. Huang, H. Luo, C. Liang, D. Jiang and S. Dai, Ind. Eng.
Chem. Res., 2008, 47, 881.
12 A. Vij, Y. Y. Zheng, R. L. Kirchmeier and J. M. Shreeve,
Inorg. Chem., 1994, 33, 3281; E. I. Rogers, D. S. Silvester,
S. E. Ward Jones, L. Aldous, C. Hardacre, A. J. Russell,
S. G. Davies and R. G. Compton, J. Phys. Chem. C, 2007, 111,
13957.
Acknowledgements
The authors thank the Deutsche Forschungsgemeinschaft
DFG for their financial support. Friederike Agel and Peter
Wasserscheid thank for additional support by the Erlangen
Cluster of Excellence ‘‘Engineering of Advance Materials’’.
13 F. R. Hartley, Chem. Rev., 1973, 73(2), 163.
14 P. Scovazzo, D. Havard, M. McShea, S. Mixon and D. Morgan,
J. Membr. Sci., 2009, 327, 41.
15 C. Cadena, J. L. Anthony, J. K. Shah, T. I. Morrow,
J. F. Brennecke and E. J. Maginn, J. Am. Chem. Soc., 2004, 126,
5300; J. L. Anthony, J. L. Anderson, E. J. Maginn and
J. F. Brennecke, J. Phys. Chem. B, 2005, 109, 6366;
M. J. Muldoon, S. N. V. K. Aki, J. L. Anderson, J. K. Dixon
and J. F. Brennecke, J. Phys. Chem. B, 2007, 111, 9001.
16 F. C. Gozzo, L. S. Santos, R. Augusti, C. S. Consorti, J. Dupont
and M. N. Eberlin, Chem.–Eur. J., 2004, 10, 6187; R. Bini,
O. Bortolini, C. Chiappe, D. Pieraccini and T. Siciliano, J. Phys.
Chem. B, 2007, 111, 598; C. S. Johnson Jr., Prog. Nucl. Magn.
Reson. Spectrosc., 1999, 34, 203; Y. Saito, T. Umecky, J. Niwa,
T. Sakai and S. Maeda, J. Phys. Chem. B, 2007, 111, 11794;
G. Annat, D. F. MacFarlane and M. Forsyth, J. Phys. Chem. B,
2007, 111, 9018; P. Stilbs, Prog. Nucl. Magn. Reson. Spectrosc.,
1987, 19, 1.
Friederike Agel also wants to thank Andreas Bosmann for
¨
fruitful discussions and Judith Scholz for her support in
measuring the equilibrium constants.
Notes and references
1 S. Ernst and W. J. Petzny, Chemierohstoffe aus Erdol und Erdgas, in
¨
¨
Winnacker-Kuchler, Chemische Technik, ed. R. Dittmeyer,
W. Keim, G. Kreysa, A. Oberholz, 5th edn, vol. 4, Wiley-VCH,
Weinheim 2005, p. 523.
2 T. Melin and R. Rautenbach, in Membranverfahren, 3rd edn,
Springer-Verlag, Berlin, Heidelberg, 2007.
3 J. G. Wijmans and R. W. Baker, J. Membr. Sci., 1995, 107, 1;
P. Scovazzo, J. Membr. Sci., 2009, 343, 199.
4 S. Winstein and H. J. Lucas, J. Am. Chem. Soc., 1938, 60, 836;
S. H. J. Lucas, R. S. Moore and D. Pressman, J. Am. Chem. Soc.,
17 J. C. Earnshaw, E. G. Johnson, B. J. Carroll and P. J. Doyle,
J. Colloid Interface Sci., 1996, 177, 150.
18 E. O. Stejskal and J. E. Tanner, J. Chem. Phys., 1965, 42, 288.
c
This journal is the Owner Societies 2011
Phys. Chem. Chem. Phys., 2011, 13, 725–731 731