Desalination 241 (2009) 182 187
Separation properties of supported ionic liquid polydimethylsiloxane membrane in pervaporation process P. Izaka,b*, K. Friessc, V. Hynekc, W. Rutha, Z. Feid, J.P. Dysond, U. Kragla a
Institute of Chemistry, University of Rostock, Albert Einstein Str. 3a, 18059 Rostock, Germany Institute of Chemical Process Fundamentals, Rozvojova 135, 165 02 Prague 6, Czech Republic Tel. +42 029 678 0268; Fax +42 022 092 0661; email: [email protected]
c Institute of Chemical Technology, Technicka 5,160 00 Prague 6, Czech Republic d Institute of Chemical Sciences and Engineering, Swiss Federal Institute of Technology (EPFL), CH-1015 Lausanne, Switzerland
Received 16 September 2007; revised 27 November 2007; accepted 4 December 2007
Abstract Pervaporation separation properties of water butan-1-ol binary mixtures were measured with the use of three supported liquid membranes at 378C, namely PDMS, 1-ethenyl-3-ethyl-imidazolium hexafluorophosphate PDMS, and tetrapropyl-ammonium tetracyanoborate PDMS membranes. An ultrafiltration ceramic module from TiO2 with 60 nm pore size was used as a support. The diffusion coefficients of butan-1-ol in IL-PDMS were much higher than in PDMS only. However, the sorption isotherms were identical for all three measured membranes. Higher permeation flux and enrichment factors of butan-1-ol in IL-PDMS membrane were probably caused by higher diffusion coefficient. The supported ionic liquid membranes were stable during all measurements. Keywords: Ionic liquid; Diffusion coefﬁcient; Sorption isotherm; Pervaporation
1. Introduction This work tries to explain why pervaporation characteristics are different for membranes *Corresponding author.
obtained from PDMS and PDMS mixed with some ionic liquids. The use of supported liquid membranes for recovery of different solutes from aqueous solutions have been widely studied during the last 25 years . The first application of supported ionic liquid membranes for gas
Presented at the Third Membrane Science and Technology Conference of Visegrad Countries (PERMEA), Siofok, Hungary, 2–6 September 2007. 0011-9164/09/$– See front matter # 2009 Published by Elsevier B.V. doi:10.1016/j.desal.2007.12.050
P. Iz ak et al. / Desalination 241 (2009) 182 187
separations was reported by Noble et al.  and the first report about the molecular interactions between room-temperature ionic liquids (RTILs) and Nafion and PDMS membranes, proving that in contact with these polymers RTILs behaved like electrolytes rather than solvents, was published already in 2005 . The authors concluded that combining the RTILs negligible vapor pressure with the ability to produce application specific RTILs possess the potential for producing highly selective membranes with high permeabilities in comparison to classical polymer membranes . The hydrophobic ionic liquid was introduced as the third phase between the aqueous phase and the plain PDMS membrane for improving mass-transfer of acetic acid from its aqueous matrix to the PDMS membrane by Yu et al. . Their primary results indicated that the ionic liquid as an extractant prior to pervaporation was favorable for improving the permeate selectivity and the permeate flux of acetic acid compared with using only a plain PDMS membrane. Thus, pervaporation offers an alternative in downstream processing, which may have energy and capital cost advantages comparing to distillation, especially for smaller-scale systems or at lower feed concentration. A development of a membrane system with suitable flux and selectivity characteristics plays a critical role in achieving better effectiveness for pervaporation due to cost considerations. 2. Experimental The two ionic liquids were synthesized as follows. 1-Ethenyl-3-ethyl-imidazolium hexafluorophosphate (IL1): 1-ethenyl-3-ethyl-imidazolium bromide  (20.3 g, 0.1 mol) and sodium hexafluorophosphate (16.8 g, 0.1 mol) was mixed in water (75 cm3), forming a suspension. After vigorously stirring at room temperature for 2 h, the suspension was filtered. The
solid was washed with water (310 cm3) and the product was dried under vacuum at 308C for 24 h. In situ recrystallization by slow cooling of an overheated fluid from 808C to room temperature over a period of 24 h gave single crystals suitable for X-ray determination. 1-Ethenyl-3ethyl-imidazolium hexafluorophosphate ionic liquid (50 wt.%) was mixed with polydimethylsiloxane (50 wt.%). The PDMS was prepared by mixing a solution of RTV 615A and RTV 615B (General Electric) in a 10:1 ratio at 608C for 0.5 h. The second supported ionic liquid membrane was prepared from a 15 wt.% of tetrapropylammonium tetracyanoborate ionic liquid (IL2)  and it was mixed with 85 wt.% of PDMS (the maximum amount of IL to get a homogeneous polymer). As a support matrix for the nonporous membrane we used the ceramic ultrafiltration membrane made from TiO2 with pore size 60 nm. The ceramic asymmetric modules were 500 mm long with external diameter of 10 mm, 1.5 mm thick and effective area of 0.011 m2. They were made by Inopor GmbH, Germany. The membrane was then impregnated by this viscous blend of IL and PDMS inside the burette for 0.5 h at 808C. The impregnated membrane was then taken out from the burette and cooled down to the room temperature and left to cure for 24 h. IL-PDMS blend (18 cm3) was kept in the ceramic module during all the time necessary for the experiments. The stability of IL-PDMS blend inside the pores was checked by weighting and no weight change of the module was observed. The pervaporation experiments were performed with the impregnated module at 378C. The concentrations of the permeates were estimated by high pressure liquid chromatography (HPLC). Samples for analysis of the composition were extracted from the feed and the cold trap (permeate) at regular time intervals usually every 24 h. An internal standard was used for all samples. The content of butan-1-ol /
P. Iz ak et al. / Desalination 241 (2009) 182 187
was determined by using HPLC in ion exclusion mode. The operating parameters used were HPLC-column, Aminex HPX-87H 300 7.8 mm (BIO-RAD, USA); mobile phase, sulfuric acid 0.006 M; temperature, 658C; detection by refractive index. The measurements were carried out using HPLC-equipment from Knauer (Berlin, Germany). An external standard calibration was used as the quantification method. The ceramic ultrafiltration module with PDMS-IL1 and PDMS-IL2 membranes inside the pores was stable under low pressure of 20 Pa in aqueous solution of butan-1-ol for more than 3 months. We did not record any formation of hydrofluoric acid from IL1 and any change in weight of the IL-PDMS inside the module during our experimental condition (pH7, 378C). The sorption experiments at a wide interval of relative vapor pressures prel p/p0 ) were performed gravimetrically by two sorption apparatuses equipped with the calibrated quartz (McBain’s) spiral balances. The all-glass apparatus  with PTFE valves was used to prevent the failure of butyl rubber o-ring seals in metal glass-apparatus valves. A stripe of sample membrane (between 0.3 and0.4 g) was hanged on a spiral balance in an evacuated glass tube. Stretching of quartz spiral was monitored by the optical system (camera Sony XCD-X710 FireWireobjective SE2514, SONY, Japan and reading and scoring software was purchased by Neovision, Czech Republic) at selected time intervals (0.5100 s) from the beginning of the measurement till the equilibrium state. The liquid sample (510 cm3) was degassed by several consequent steps of heating/cooling before dosing from a glass vessel connected to the apparatus. Selected vapor pressure of each experiment was set using the vapor reservoir and determined by pressure gauge (Leybold Vacuum, Germany). Before each experiment the apparatus was evacuated to pressure lower than 0.01 Pa by rotary /
and turbo molecular pumps (Leybold Vacuum). The average error of mass determination of each experiment reaches approximately 9 30 mg. The integral diffusion coefficients D were obtained by fitting of experimental data to Eq. (1) which was derived by solving of the second Fick law under appropriate initial and boundary conditions [14,15] /
ð2i þ 1Þ
# 2 ð2i þ 1Þ 2 Dt L2
where Qt is the mass uptake at time t and Q is the mass uptake at time of reaching the sorption equilibrium, and L is the membrane thickness.
3. Results and discussion
5.E–10 Dbutan-1-ol (m2 s–1)
The diffusion coefficients of butan-1-ol in IL-PDMS are displayed in Fig. 1. One can see that the diffusion coefficient is much higher (one order of magnitude) in IL-PDMS than in the PDMS membrane. However, with higher relative pressure of butan-1-ol this difference is getting smaller. On the other hand the sorption isotherms
4.E–10 3.E–10 2.E–10 1.E–10 1.E–11 0
PDMS + IL1 PDMS + IL2
Fig. 1. Dependence of butan-1-ol diffusion coefﬁcient on relative pressure. , PDMS; , PDMS/1-ethenyl-3ethyl-imidazolium hexaﬂuorophosphate ionic liquid; , PDMS/tetrapropylammonium tetracyano-borate ionic liquid.
Enrichment factor of butan-1-ol
Sorbed amount of butan-1-ol (mg/g)
P. Iz ak et al. / Desalination 241 (2009) 182 187
600 400 200
PDMS + IL1
PDMS + IL2
0 0.5 1 1.5 2 Feed concentration of butan-1-ol (%w/w) PDMS
Fig. 2. Dependence of butan-1-ol sorption isotherm on relative pressure at 378C. , PDMS; , PDMS/1ethenyl-3-ethyl-imidazolium hexaﬂuorophosphate ionic liquid; , PDMS/tetrapropylammonium tetracyanoborate ionic liquid.
Fig. 4. Dependence of butan-1-ol enrichment factor on feed concentration. , PDMS; , PDMS/1-ethenyl-3ethyl-imidazolium hexaﬂuorophosphate ionic liquid; , PDMS/tetrapropylammonium tetracyano-borate ionic liquid.
are identical for IL-PDMS and PDMS membranes (Fig. 2). Therefore, we can state that higher permeation flux (Fig. 3) is most probably caused by higher diffusion coefficient of butan1-ol in IL-PDMS membranes. Permeation flux is defined as Ji JwiP, where J is the total permeation flux through the supported ionic liquid membrane and wiP is the weight fraction of the component i in the permeate. Permeation
flux of butan-1-ol increased with increase of butan-1-ol concentration in the feed for all measured membranes. In downstream processes usually with higher flux we obtain lower selectivity. However, this is not the case when we compared IL-PDMS and PDMS only. We measured the concentration of feed and permeate always after 24 h and we obtained always higher concentration (higher enrichment factor; bi wiP/wiF, where wiF is the weight fraction of component i in the feed) of butan-1-ol in the permeate when IL was inside the polymer (Fig. 4). This is again most probably caused by higher diffusion of butan-1-ol, which is transported in IL-PDMS faster than in PDMS only. Enrichment factor of butan-1-ol for IL-PDMS membranes increased with decrease of butan-1ol concentration in the feed. For the PDMS membrane the enrichment factor was constant within the measured range of butan-1-ol concentration. We can speculate that ionic liquid inside PDMS form narrow channels and transport of buthan-1-ol through ILPDMS is by them. Therefore, we have much higher permeation flux and also diffusion coefficient in ILPDMS
Permeation flux of butan-1-ol (g m–2h–1)
20 15 10 5 0 0
0.5 1 1.5 2 Feed concentration of butan-1-ol (%w/w) PDMS
Fig. 3. Dependence of butan-1-ol permeation ﬂux on feed concentration. , PDMS; , PDMS/1-ethenyl-3ethyl-imidazolium hexaﬂuorophosphate ionic liquid; , PDMS/tetrapropylammonium tetracyano-borate ionic liquid.
P. Iz ak et al. / Desalination 241 (2009) 182 187
than in PDMS only. With higher saturation of butan-1-ol the sorption isotherms exponentially increased and PDMS swells. The swelling at higher relative pressure of butan-1-ol will most probably plug the ionic liquid channels and therefore strong decrease of diffusion coefficients can appear. We cannot see decrease in permeation flux because coupling effect and more swollen PDMS can compensate lower diffusion coefficient. We also measured pervaporation characteristics in narrow concentration range where Clostridium acetobutylicum can live. However, the diffusion coefficients and sorption isotherms are in the whole concentration range. Especially noteworthy is the fact that IL-PDMS membranes have significantly better separation properties than PDMS membrane itself. Particularly, the tetrapropylammonium tetracyanoborate IL has much better separation properties and permeation flux than 1-ethenyl-3-ethyl-imidazolium hexafluorophosphate IL, even when taking into account the miscibility limitation with PDMS, which results in a much lower amount of IL immobilized within the PDMS-polymer. 4. Conclusions The removal of butan-1-ol from aqueous solution has a practical application in biotransformation processes such as fermentation of Clostridium acetobutylicum . The diffusion coefficient of butan-1-ol is much higher in IL-PDMS than in the PDMS membrane. However, with higher relative pressure of butan-1-ol the diffusion coefficients in IL-PDMS decreased exponentially. The sorption isotherms were identical for IL-PDMS and PDMS membranes. The higher permeation flux and enrichment factor in IL-PDMS membranes was most probably caused by much higher diffusion coefficient of butan-1-ol at low relative pressure. Especially noteworthy is the fact that IL-PDMS membranes
have significantly better separation properties than PDMS membrane itself. Acknowledgements This research was supported by Marie Curie Intra-European and Marie Curie Reintegration Fellowships within the 6th European Community Framework Programme. P. Izak is also grateful to Purkyne Fellowship from the Academy of Science of the Czech Republic. We are also grateful to Dr. Katrin Schwarz and Prof. Dr. Hubert Bahl from the Institute of Biological Science/Microbiology, University of Rostock, Germany for hopeful consultations. The partial financial support of the Ministry of Education, Sports and Youth MSM (Grant No. 6046137307) is gratefully acknowledged as well.
References  O. Loiacano, E. Drioli and R. Molinari, J. Membr. Sci., 28 (1986) 123.  R. Chiarizia, E.P. Horwitz, P.G. Rickert and K.M. Hodgson, Sep. Sci. Technol., 25 (1990) 1571.  J.J. Pellegrino and R.D. Noble, Trends Biotechnol., 8 (1990) 216.  J. Zigova, E. Šturd´ık, D. Vandak and S. Schlosser, Process Biochem., 34 (1999) 835.  A. Kiani, R.R. Bhave and K.K. Sirkar, J. Membr. Sci., 20 (1984) 125.  P. Scovazzo, A. Visser, J. Davis, R. Rogers, C. Koval, D. DuBois, R. Noble, in: R. Rogers, K. Seddon (Eds.), Industrial Applications of Ionic Liquids, American Chemical Society Books, Chapter 6, 2002.  T. Sch€afer, R.E.D. Paolo, R. Franco and J.G. Crespo, Chem. Commun., (2005) 2594.  P. Scovazzo, J. Kieft, D. Finan, R.D. Noble and C.A. Koval, J. Membr. Sci., 238 (2004) 57.  J. Yu, H. Li and H. Liu, Chem. Eng. Commun., 193 (11) (2006) 1422.  R. Marcilla, J.A. Blazquez, J. Rodriguez, A.A. Pomposo and D.J. Mecerreyes, Polym. Sci., 42 (2004) 208.
P. Iz ak et al. / Desalination 241 (2009) 182 187  P. Izak, M. K€ ockerling and U. Kragl, Green Chem., 8 (2006) 947.  K. Friess, M. Šípek, V. Hynek, P. Sysel, K. Bohata and P. Izak, J. Membr. Sci., 240 (2004) 179.  K. Friess, M. Šípek, V. Hynek, and C. Panayiotou, Desalination, 200 (1 3) (2006) 265.
 J. Crank, The Mathematic of Diffusion, Clarendon Press, Oxford, 1975.  J. Crank and G.S. Park, Diffusion in Polymers, Academia Press, London, 1968.  H. Bahl, W. Andersch and G. Gottschalk, Eur. J. Appl. Microbiol. Biotechnol., 15 (1983) 201.