New applications of supercritical fluids and supercritical fluids processes in separation

Transcription

1 Separation and Purification Technology 32 (2003) 57/63 New applications of supercritical fluids and supercritical fluids processes in separation S. Sarrade a,1, *, C. Guizard b,2, G.M. Rios b,3 a Commissariat a l Energie Atomique Supercritical Fluids and Membranes Laboratory-CEA VALRHO, DTE/SLP, Pierrelatte BP , France b European Institute of Membranes, UMR CNRS 5635-UMII-CCO 47-2, place Eugène Bataillon, Montpellier, France Abstract This presentation is a review of new ideas recently proposed in the field of supercritical fluid (SCF)-and membrane coupled processes or membranes preparation. First, few studies are presented in order to illustrate the preparation of ceramic membranes using the supercritical route. New materials obtained in supercritical conditions lead us to develop original ceramic membrane for filtration or application. The other applications concern coupled supercritical CO 2 (SC CO 2 ) extraction with nanofiltration separation to purify low molecular weight compounds. This process has been proposed as an environment safeguarding solution for coupling extraction and separation. Cross-flow filtration applied to viscous liquids fluidified with SC CO 2 is also investigated. The basic idea is to reduce liquid viscosity at ordinary temperature by injecting a gas (CO 2 ) in supercritical conditions before filtration. Finally, these works, as well as other examples involving new membrane contactor or reactor types and supercritical fluids as solvents, bear testimony of the main interest to use such hybrid techniques regarding performance and environmental constraints. # 2003 Elsevier Science B.V. All rights reserved. Keywords: Supercritical carbon dioxide; Inorganic membrane preparation; Cross-flow filtration; Viscous fluid fluidification; Membrane contactors/reactors 1. Introduction * Corresponding author. Tel.: / ; fax: / address: stephane.sarrade@cea.fr (S. Sarrade). 1 Joint Research Team Supercritical Fluids and Porous Media (CEA/CNRS/UMMII/ENSCM). 2 Joint Research Team Supercritical Fluids and Porous Media (CEA/CNRS/UMMII/ENSCM). 3 Joint Research Team Supercritical Fluids and Porous Media (CEA/CNRS/UMMII/ENSCM). Supercritical fluids (SCF) have been widely used in extraction and recovery of high-value compounds. Experience accumulated in recent years on the use of SCF and SCF processes have reached the step that it is possible to explore and envision their uses beyond the common practice of extraction. The purpose of the present article is to describe some of such possible applications, based on studies carried out in our laboratory and elsewhere. Some of the more significant potential application includes: /03/$ - see front matter # 2003 Elsevier Science B.V. All rights reserved. doi: /s (03)

2 58 S. Sarrade et al. / Separation and Purification Technology 32 (2003) 57/63 Ceramic filtration membrane prepared using the supercritical CO 2 (SC CO 2 ) route. Solvent recovery or separation of CO 2 from complex mixtures. CO 2 extraction coupled with cross-flow filtration. CO 2 fluidification of viscous fluids during cross-flow filtration. Membrane contactor/reactor operated under SC CO 2 conditions. This article will present these topics in order to have a better view of all the applications possible to reach when the coupled concept is use, and several examples will allow us to focus on nearindustrial scale processes. 2. Membranes preparation The use of supercritical fluids as a chemical reaction medium offers new possibilities. SC CO 2 can replace usefully number of typical organic solvents. Although SC CO 2 is a non-polar solvent, which limits solubility of polar reactants, its combination of liquid-like density and gas-like viscosity and diffusivity leads to high reaction rate and easy recovery of products. Commonly used in organic chemistry, carbon dioxide is now proposed to be used in inorganic synthesis. A number of very interesting applications: chromatographic stationary phase, adsorbents, catalytic supports or mineral membrane can be achieved using ultrafine metal-oxide particles. SCFs have been investigated as suitable and novel reaction media for the synthesis of such particles starting from titanium alkoxide precursors. Brasseur-Tilmant [1] presented in 1999 a work dealing with modification of a macroporous alumina media by TiO 2 particles deposition using supercritical isopropanol. The aim was to prepare inorganic membranes for cross-flow filtration. Anatase particles were deposited on plane alumina support after thermal decomposition of titanium alkoxide precursors. A slight infiltrated zone is observed and a pore size reduction is achieved from 110 to 5 nm, leading to obtain fine ultrafiltration membranes. The main problem is to control the reaction at the membrane interface and not in the porosity, and moreover this process is suitable for tubular membrane preparation. Papet et al. [2] described the preparation of titanium hydroxide particles in a high-pressure stirred vessel using SC CO 2 followed by particle recovery. Because of the high reactivity of the titanium alkoxide for hydrolysis, highly volatile amorphous fine powders of titanium hydroxide was formed at the pressure vessel surfaces. SEM observations of this powder (Fig. 1) revealed a relatively homogeneous size distribution of microsized spherical particles depending on the reaction parameters. Later, Papet [3] presented an alternative process for preparing tubular ceramic cross-flow filtration membranes. Papet s method consists of the casting of tubular mineral microfiltration membranes with titanium dioxide suspensions. The deposited particles on the porous support were then compressed and finally the layer was consolidated by firing. One issue to control the final pore size of the consolidated layer is to control the size of the deposited particles. Actually, using the submicronic titanium dioxide particles previously described, the preparation process allows the authors to control the particle size (i.e. the layer pore diameter) by tuning the operating parameters during the process (temperature, pressure, contact time). The influence of the synthesis parameters on morphology and texture of titanium dioxide particles and the preliminary results concerning membrane preparation using the layer compression process, show that the obtained membranes are still in the ultrafiltration range, due to shrinkage during sintering related to the low primary particles size and high specific surface area. Further work is in progress in our group concerning other precursors for the production of ceramic oxide materials (Ce 1x Gd x O 2x /2 or doped ZrO 2 and LaGaO 3 ) which are oxides of interest as oxygen ion conductors. 3. Solvent recovery In 1983, Schell et al. [4] proposed the use of gas diffusion membrane for intensive hydrocarbon

3 S. Sarrade et al. / Separation and Purification Technology 32 (2003) 57/63 59 Fig. 1. Spherical nanophased titania particles obtained in SC CO 2 from hydrolysis and condensation of tetraisopropyl-orthotitanate at 323 K and 30 MPa. recovery. Because of the improved membrane (in terms of chemical and mechanical resistances) obtained in the last decade, Semenova and Ohya [5] studied the fractionation of SC CO 2 /ethanol and SC CO 2 /iso-octane mixtures using an asymmetric Kapton membrane. The investigators concluded that CO 2 transfer across was predominantly by convection rather than diffusion. At approximately the same time, Hsu and Tan [6] proposed to use reverse osmosis membranes to fractionate water/ethanol mixtures in the presence of SC CO 2. Under these conditions ethanol rejection is improved from 20 to 70%. The authors attributed the improved rejection to the formation of CO 2 and ethanol clusters. When SC CO 2 is used as a solvent for extraction, it is interesting to recover CO 2 without performing a pressure drop. Birtigh [7], Sartorelli and Brunner [8] studied the supercritical fluid cycle, and they demonstrated that inorganic or organic membranes separation process can be propose instead of typical regeneration cycle in case of supercritical extraction to reduce drastically the energy losses. In fact, a stream of low volatile compounds (LVC) extracted by SC CO 2 can be discharged of 80/90% of LVC using a nanofiltration membrane (mean pore diameter of about 1 nm) without a significant drop of pressure. The main point to manage is the chemical resistance to CO 2 of organic membrane to be used in such conditions. 4. Nanofiltration plus SCF extraction process [9 / 11] Considering that nanofiltration and supercritical fluid extraction processes are intended for the separation/recovery of same chemical species (i.e. low molecular weight compounds up to 1500 Da), and using new membrane materials able to resist to SC CO 2,wedeveloped a new hybrid process which couples the two functions, but with a single pumping device and, therefore, makes it possible to develop synergistic effects leading to improved performance. Moreover, with CO 2 as solvent, the process is intended environmentally safe and yields products with high quality. Fig. 2 provides a general outline of the process. The exact location Fig. 2. Schematic diagram of combined nanofiltration and supercritical fluid extraction process.

4 60 S. Sarrade et al. / Separation and Purification Technology 32 (2003) 57/63 and function of the nanofiltration membrane on the classical extraction loop have been indicated. The first expected advantage of the process is to be able to work with a high permeate flux because of the low viscosity of SC CO 2 (ten times lower than for water). Indeed the membranes used in the process are hybrid nanofiltration elements, constituted from an inorganic substrate (TiO 2 with a mean pore diameter of about 10 nm) on which a nafion layer had been deposited, the prevailing mass transfer mechanism of which is convection. This fact could be checked through experiments conducted successively with water and SC CO 2, flows obtained being in the opposite ratio of fluid viscosities within than less 10%. Another expected advantage of the technique is a good control of the whole extraction/separation process. It is worth recalling that with the classical supercritical fluid extraction process the choice of optimal extraction conditions is often compromised by a requirement with a suitable final fractionation of extracts. Indeed the thermodynamic conditions for maximum extraction of compounds, and those for optimal selectivity of extracted species, are often not the same; consequently it is necessary to choose intermediate working conditions or more expensive equipment involving several pressure release steps. Because with the new process the two functions are clearly separated, they may be optimised separately. All these advantages have been checked [11]. Firstly experiments were carried out with model solutions constituted from small polyethyleneglycols (PEG). Then extraction/separation of natural products was investigated: recovery of unsaturated fatty acids from fish oil such as eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) today used in human medicine for reduction of lipid absorption and prevention for heart disease; extraction and separation of b-carotene (Provitamin A) an antioxidant supposed to prevent some cancers and also a well-known yellow /orange natural colorant of agro-food industries. For these applications, the new technique provides an original and effective way to get good quality extracts in safeguarding environmental conditions. 5. SCF assisted ultrafiltration [12 /15] Ultrafiltration of highly viscous liquids (particularly oils) is a difficult and expensive operation mainly characterised by low permeate flux and high-energy consumption. These difficulties in principle can be overcome by lowering the liquid viscosity, which can be accomplished by either increasing the process temperature (up to 350 8C) or adding specific chemicals (such as surfactants). For either case, additional costs result (directly linked to requisite equipments security, energy; due to higher operating temperature, or further separation steps, necessary for preserving environment; due to chemical addition), as well as product degradation (as far as thermal sensitivity or direct contact pollution are concerned). That is the reason why a new cross-flow ultrafiltration process, taking advantage of the low viscosity, low surface tension, tunable solvent power of supercritical CO 2 has been recently proposed. Moreover, the process appears as particularly interesting for safeguarding of product quality and environment. The experimental set up consists mainly in a gas pressurised cross flow filtration circulation loop working up to 20 MPa and 150 8C (Fig. 3). Retentate velocity can be adjusted between 1 and 25 m s 1. Membranes chosen are inorganic in order to ensure drastic operating conditions without failure: ZrO 2 ultrafiltration membranes with molecular weight cut off around 50 and 300 kda, or even nanofiltration layers (MWCO: 1.5 kda). Fig. 3. Pilot-plant for SCF assisted ultrafiltration.

5 S. Sarrade et al. / Separation and Purification Technology 32 (2003) 57/63 61 Before performing the filtration experiments, the effect of CO 2 on viscosity reduction was investigated at laboratory scale. It was found (Fig. 4) that [13]: the higher the CO 2 pressure, the lower the viscosity, the lower the temperature, the higher the viscosity reduction effect, the higher the molecular mass, the higher the viscosity reduction effect. The effect of the transmembrane and absolute CO 2 pressures on the permeate flux for the case of used motor oil are shown in Fig. 5. It is clear that an increase of CO 2 increases the permeate flux; but the effect is less significant above 15 MPa and infact this value represents the optimum pressure for oil filtration. As a whole a flux increase higher than 300% may be reached. Flux stabilisation at a transmembrane pressure higher than 0.2 MPa may be explained considering the presence of additives and impurities in used oil (polarisation and fouling); but the presence of a partial demixion of gas due to high shear stresses that develop at pore wall constitute another assumption which cannot be Fig. 5. Effect of the transmembrane and CO 2 pressures on oil permeate flux (a: Jv/mixture flux; b: Jv/net flux in oil). Fig. 4. Viscosity vs. pressure; PEG 200 and PEG 400. rejected either at first. Finally from analysis of permeate and retentate it follows that there is an excellent rejection of metals, Fe, Zn, Cu, which increases with DP up to more than 99% depending on species. For mass concentration factors of about 27, residues with a total metal content less than 4% of the initial load were found: oil regeneration was very effective.

6 62 S. Sarrade et al. / Separation and Purification Technology 32 (2003) 57/63 6. Membrane contactor and/or reactor 6.1. Membrane contactor The Porocrit process [16] is different from the previous ones because the membrane (a macroporous element with a mean porous diameter equal to 20 nm) mainly acts as a contacting device that allows the stabilisation of one interface between two exchanging phases, the feed solution and the SC CO 2 extracting solvent Membrane reactor Recently, Knez et al. [17] reported studies on the development of membrane reactors using polysulphone ultrafiltration membranes. These membranes are known to withstand extreme operating conditions: high pressure up to 35 MPa in SC CO 2 and propane. Unlike ceramic membranes, organic membranes have to be tested experimentally in order to ascertain their chemical and mechanical integrity. Continuous procedures of enzymatic synthesis have certain advantages over batchwise processes for industrial developments, such as easy process control, low amount of enzyme, constant quality products, etc. The enzyme immobilisation strategy and the reactor design should be chosen in agreement with both the synthetic activity and operational stability of the biocatalytic system. The advantages of such reactions taking place in SCF medias are connected to physico-chemical properties of the fluid: low viscosity, high density, and fair diffusion coefficient. So, the idea of using enzymatic dynamic membranes to perform reaction in supercritical medium has been recently tested [18]. Enzymatic dynamic membranes were obtained by coating a-alumina microfiltration membranes with an inert protein, and then a- chymotrypsin was covalently attached, providing the appropriate microenvironment. The obtained enzyme derivative was used as catalyst for the continuous synthesis of L-tyrosine methyl ester by transesterification from L-tyrosine ethyl ester and methanol, and compared with an adsorbed a- chymotrypsin celite derivative. 7. Conclusions The review proposed in this article shows clearly that the use of a membrane in the presence of a supercritical fluid makes it possible the design of very attractive and powerful processes, to improve transfer or reaction, to set in contact phases, to fluidify highly viscous liquids. This is to be connected to the specific physicochemical properties of SCF and the particular environment that the membrane creates. Specific constraints must be accounted for which are due to the conditions of pressure and temperature particular to supercritical fields. Moreover, supercritical fluids seem to be very convenient for the preparation of new generation of inorganic crossflow membranes. Overall the major interest of all the processes thus created is to be safeguarding for environment and products, which is in particular essential when these ones are of biological nature. The principal effort requested from the researcher is a whole effort of imagination, and serious in further investigation with appropriate knowledge of membranes and supercritical fluid technologies. References [1] J. Brasseur-Tilmant, Ph.D. thesis, University of Paris XIII, France, [2] S. Papet, S. Sarrade, A. Julbe, C. Guizard, in: Proceedings of the Sixth Meeting on Supercritical Fluids, Chemistry and Materials, Nottingham, UK, 1999, p. 17. [3] S. Papet, Ph.D. thesis, University of Montpellier, France, [4] W.J. Schell, C.D. Houston, W.L. Hopper, Oil Gas J. 8 (1983) 52. [5] S.I. Semenova, H. Ohya, T. Higashijima, Y. Negishi, J. Membr. Sci. 74 (1992) 131. [6] J.H. Hsu, C.S. Tan, J. Membr. Sci. 81 (1993) 273. [7] A. Birtigh, Ph.D. thesis, Technical University of Hamburg, Germany, [8] L. Sartorelli, G. Brunner, in: Proceedings of the Fifth International Symposium on Supercritical Fluids, Atlanta, USA, [9] S. Sarrade, G. Rios, M. Carles, J. Membr. Sci. 97 (1994) 155.

7 S. Sarrade et al. / Separation and Purification Technology 32 (2003) 57/63 63 [10] S. Sarrade, G. Rios, M. Carles, J. Membr. Sci. 114 (1996) 81. [11] S. Sarrade, G. Rios, M. Carles, Sep. Purif. Technol. 14 (1998) 19. [12] D. Gourgouillon, Ph.D. thesis, University of Montpellier, France, [13] D. Gourgouillon, L. Schrive, S. Sarrade, G. Rios, Sep. Sci. Technol. 35 (2000) [14] D. Gourgouillon, L. Schrive, S. Sarrade, G. Rios, Environ. Sci. Technol. 34 (2000) [15] S. Sarrade, L. Schrive, D. Gourgouillon, G. Rios, Sep. Purif. Technol. 25 (2001) 315. [16] J. Robinson, Porocrit technology-porocrit LLC, 1012 Grayson St., Suite A, Berkeley, CA , USA. [17] V. Kremelj, M. Habulin, D. Vasic-Racki, Z. Knez, in: Proceedings of the Fifth International Symposium on Supercritical Fluids, Atlanta, USA, [18] P. Lozano, T. De Diego, M.P. Belleville, G.M. Rios, J.L. Iborra, Biotechnol. Lett. 22 (2000) 771.