- Email: [email protected]
Solvent impregnated resins for MTBE removal from aqueous environments
Reactive & Functional Polymers 70 (2010) 41–47
Contents lists available at ScienceDirect
Reactive & Functional Polymers journal homepage: www.elsevier.com/locate/react
Solvent impregnated resins for MTBE removal from aqueous environments Bernhard Burghoff a,*, Joao Sousa Marques b, Bart M. van Lankvelt a, Andre B. de Haan a a b
Eindhoven University of Technology, Den Dolech 2, 5600 MB Eindhoven, Netherlands Universidade do Porto, Faculdade de Engenharia, 4000 Porto, Portugal
a r t i c l e
i n f o
Article history: Received 6 July 2009 Received in revised form 22 September 2009 Accepted 11 October 2009 Available online 5 November 2009 Keywords: Extraction MTBE Solvent impregnated resins
a b s t r a c t MTBE removal from groundwater is difficult. State-of-the-art processes such as air stripping or adsorption can have significant drawbacks, like high energy consumption, low capacity and low selectivity. This is why the alternative technology of solvent impregnated resins is investigated. The extractant used is 3iodophenol diluted with propylbenzene. From solids screening and impregnation experiments macroporous polypropylene (MPP) particles appear as a suitable solid support. The MTBE capacity of impregnated MPP is lower than that of a carbonaceous resin, but the selectivity of the SIR between MTBE and humic acid is significantly higher. Solvent impregnated resin (SIR) regeneration can be easily achieved with hot gas. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Methyl tert-butyl ether (MTBE) is mainly used as a fuel oxygenate added to petrol as an octane enhancer which helps to prevent engine ‘knocking’, and to increase oxygen content [1]. Due to its high water solubility compared with other fuel constituents (water solubility of MTBE at 25 °C: 51 g L1 [2], water solubility of benzene at 25 °C: 1.79 g L1 [3], its low soil adsorption coefficient and its low Henry coefficient leakages from storage facilities cause immediate ground water contamination [1]. In the United States, of the 1000 groundwater sites tested, more than 80% were shown to be contaminated with MTBE and concentrations as high as 23 mg L1 have been measured [4]. Since MTBE has a very low taste and odor threshold in water of 40 and 15 lg L1 [5], respectively, it needs to be removed during the production of high-quality potable water from ground water. Conventional processes for MTBE removal are air stripping and granular activated carbon (GAC) adsorption [5]. Although air stripping is an established technology, large air streams are required for the removal of low concentration MTBE, which results in high process costs [4]. GAC adsorption is a commercial technology and regularly used in water treatment. It appears to be efficient for MTBE concentrations below 200 lg L1 [5]. However, when MTBE is present in concentrations above 2000 lg L1 [6], this process can only be used as a polishing step followed by biological or chemical treatment or air stripping [5]. This is because of the low affinity * Corresponding author. Present address: TU Dortmund, Emil-Figge-Str. 70, 44227 Dortmund, Germany. Tel.: +49 2317557148; fax: +49 2317552431. E-mail address: [email protected] (B. Burghoff). 1381-5148/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.reactfunctpolym.2009.10.002
of MTBE for adsorption on solids. Another drawback of GAC is the reduced life-time of typically 10–20 cycles [7]. Furthermore a competitive adsorption of other organic compounds in ground water, such as either fuel components or humic acid, results in a GAC affinity decrease for MTBE [8]. MTBE removal with synthetic resins was first investigated in the 1970s as an alternative to GAC [6]. Synthetic resins have a higher number of possible adsorption sites than GAC. Synthetic resins can be divided into two categories, which are polymeric resins and carbonaceous resins. The polymeric resins are usually based on cross-linked polymers. These polymers can have polystyrene, phenolformaldehyde, or acrylate matrices [6]. Carbonaceous resins are produced as a product of the partial pyrolysis of macroporous polymer particles. Due to the pyrolysis micropores will form, which are critical for the application of carbonaceous resins in water treatment. An alternative process for removal of low concentration solutes from water is extraction. In a previous study, 3-iodophenol is suggested as an effective MTBE extractant [9]. The immobilization of the extractant in commercially available macroporous resin particles, so-called solvent impregnated resins (SIRs), avoids emulsification and simplifies the phase separation after extraction [10,11]. Another advantage of SIR technology compared to conventional liquid–liquid extraction is that due to the impregnation the loss of extractant into the aqueous phase during extraction decreases [10,12]. Possible drawbacks of SIR technology, such as leaching of the extractant or clogging of a fixed bed by attrition of the particles, can be remedied by choosing the proper system. This implies a suitable extractant with low water solubility, which is sufficiently retained inside the pores, and the selection of sturdy
42
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
Nomenclature c KD KF n q Vpore
concentration (mol L1) overall distribution coefficient (mol L1) (mol L1)1 Freundlich isotherm parameter (mol g1) (mol L1)n Freundlich isotherm parameter () solute loading of particle (mol g1) particle pore volume (mL)
Greek letters selectivity H constant separation factor
Subscripts 0 parameter at time t = 0 E extractant HA humic acid max maximum MTBE methyl tert-butyl ether S solute SIR solvent impregnated resin
a
Superscripts aq aqueous phase org organic phase particles as a solid support for the extractant. Applications of SIRs for heavy metal recovery are already well known [13–15]. Recently, also other applications have been developed, e.g. the recovery of hydrocarbons, such as aliphatics, aromatics and halogenated hydrocarbons, on a large-scale on offshore oil platforms [16]. In such an application, where the SIR particles are contained in a packed bed, flow rates higher than 0.5 m3 h1 without maximum flow restrictions can be treated cost competitive to air stripping/ activated carbon, steam stripping and bio treatment systems [16]. The removal of polar organics like phenol [11], aldehydes [17], organic acids [18], amino acids [19] and flavonoids from aqueous media [20] is only carried out on a bench-scale or pilotscale. Large-scale applications for the separation of polar solutes, like e.g. MTBE, do not yet exist. The design of a SIR containing a suitable MTBE extractant can close this gap. The objective of this study is to show that MTBE can be removed from aqueous solutions with a SIR. For this purpose, a preliminary SIR is designed to demonstrate the proof-of-principle. As yet, for the SIR design only the preliminary extractant 3-iodophenol is available based on a previous MTBE extractant screening [9]. The extractant 3-iodophenol is not optimal in terms of water solubility and environmental impact, which should both be negligible for an environmental application. In addition to this, it is solid at standard conditions and requires a diluent for application in an extraction process. Adding a branched alkyl chain in para position at the aromatic ring of 3-iodophenol will decrease water solubility and thus the possible release into the environment appreciably. In order to decrease the melting point below standard temperature and thus avoid the need for a diluent, an extractant blend consisting of components with differing alkyl chain lengths can be created. However, the synthesis of such an optimized extractant including the necessary evaluation of physical properties, environmental impact and MTBE capacity implies great experimental effort and is time consuming. This extractant optimization is not within the scope of this study. The preliminary SIR is thus simply based on 3-iodophenol dissolved in a diluent and the current study focuses mainly on the proof-of-principle to remove MTBE from water with SIRs. In order to show that SIRs can in principle be used to remove MTBE from aqueous solutions a suitable SIR is developed. In addition to an extractant, the solid matrix for the solvent impregnated resin needs to be selected. Preferably, the particles should have an affinity for MTBE, in order to have a supportive effect by adsorption on the particle surface in addition to the extraction by the immobilized extractant. This is why adsorbents with a high MTBE capacity are preferred. A literature review gives information about suitable adsorbents. Four different types of adsorbents appear to be suitable. These adsorbents are carbonaceous resins (e.g. Ambersorb XE-348F), activated carbon (e.g. AquaSorb 202) [6,21] and
synthetic resins, such as non-ionic aliphatic acrylic particles (e.g. Amberlite XAD 7) [5]. Macroporous polypropylene (e.g. MPP) particles are tested, as they have been shown in other SIR applications to give a good performance [11,16] due to a high extractant capacity inside the pores and a high mechanical stability. Adsorbents with a high MTBE capacity, a high extractant capacity inside the pores and a high mechanical stability are then used as solid supports for the extractant. A further issue of investigation is the MTBE selectivity in the presence of humic acid. This is an important topic in environmental applications, when MTBE is supposed to be removed from groundwater. Humic acid is one of the main constituents of humic substances in groundwater. The stability of humic substances is very high compared to other organic fractions in the soil [22]. High concentrations of natural organic matter, such as humic acid, compete with MTBE adsorption on activated carbon and synthetic resins, and can decrease the adsorbent usage efficiency and consequently increase the process costs [5]. A kinetic study of the prepared SIR particles is carried out. In the scope of this study, the MTBE extraction profiles of Ambersorb XE-348F, the activated carbon AquaSorb 202 and prepared SIRs are compared. The influence of the particle size on the determined concentration profiles is investigated. A zero-length-column setup, which will be explained below, is used to minimize axial dispersion and external mass transfer [23,24]. Finally, the regenerability of the selected SIR is investigated. In an exploratory regeneration study, the feasibility of using a hot gas stream to evaporate MTBE from the SIR particles after MTBE extraction is the focus of the regenerability considerations.
2. Materials and methods 2.1. Substances MTBE (99.5%, Merck KGaA, Germany) were used without further treatment. As extractant 3-iodophenol (98%, Sigma–Aldrich, Germany) was used. Propylbenzene (99%, Fluka Chemie AG, Switzerland) was used as a diluent for 3-iodophenol. Above listed substances were all utilized without further treatment. The polymeric resins used were MPP (VWS MPP Systems, Netherlands), Amberlite XAD7 (Rohm and Haas Company, France) and Ambersorb XE-348F (Sigma–Aldrich Chemie GmbH, USA). The activated carbon used was AquaSorb 202 (Jacobi Carbons, Germany). AquaSorb 202 and Ambersorb XE-348F were used as received. The MPP particles were sieved, then washed with acetone and left in a rotary evaporator (BÜCHI Rotavapor R-200, BÜCHI Labortechnik AG, Switzerland) equipped with a heating bath (BÜCHI Heating
43
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47 Table 1 Particle characteristics.
a b c d
Property Material
Ambersorb XE-348F Carbonaceous adsorbent
AquaSorb 202 Activated carbon of bituminous coal
Amberlite XAD7 Macroreticular aliphatic acrylic cross-linked polymer
MPP Semi-crystalline polypropylene
Porosity Density (g mL1) Surface area (m2 g1) Pore size (nm) Particle size (mm)
0.45b 1.32b 750a N.A. 0.25–0.5a
0.51a 1.22b 1100a N.A. 0.425–1.7a
P0.5a 1.07a P380a N.A. 0.56–0.71a
0.68a 1.15b 11.5c <1000d 0.8–1.18 (98%)b
Manufacturer’s data. Own measurement. Babic´ et al. [17]. Based on analysis of SEM images.
Bath B-490, BÜCHI Labortechnik AG, Switzerland) for 12 h at vacuum and 70 °C to remove solvent residues. Before initial use Amberlite XAD7 was rinsed with demineralized water, decanted, rinsed with acetone, decanted again and finally rinsed with hexane and subsequently decanted. The Amberlite XAD7 particles were left in a rotary evaporator for 12 h at vacuum and 70 °C to evaporate solvent residues. 2.2. Screening of adsorbents/solid matrices The characteristics of the different particles used are given in Table 1. Isotherm experiments were conducted in closed 50 mL glass flasks. In each bottle, 0.3 g of adsorbent and 10 mL of solutions of different concentrations of MTBE were mixed by magnetic stirring at 20 °C. For all experiments the aqueous concentration of MTBE was in the range of 0.25–5 g L1. This concentration range was selected due to the available analytical equipment. In case of Amberlite XAD7 and MPP particles the contacting time was 24 h in order to ensure that equilibrium was reached. Previously performed tentative experiments suggested that this period of 24 h is sufficient to reach equilibrium. In case of Ambersorb XE-348F and AquaSorb 202 the contacting time was 5 days [21]. The stirring speed during these experiments was adjusted to 500 rpm. The MTBE content of the aqueous phase was determined with head space gas chromatography (HSGC) using a Varian CP-3800 gas chromatograph (Varian Inc., Netherlands) equipped with a CombiPal Atas Focus (Atas GL International B.V., Netherlands) including an 11/70 injector. The column used was a factor four capillary column CP 8945, VF5 ms, with 30 m length, 0.25 mm internal diameter and 0.50 lm film thickness (Varian Inc., Netherlands) connected to an FID. Samples for analysis were prepared in 20 mL vials. The transfer of MTBE to the gas phase was enhanced by thermostatic control of the samples at 70 °C [25]. To ensure that the vapor–liquid phase equilibrium was reached each vial was agitated for 10 min before the injection. The concentration of MTBE in organic and solid phases was calculated by a simple mass balance [12]. The systematic error of this analytical method was below approximately 9%. 2.3. Solvent impregnated resin preparation and characterization The carbonaceous resin Ambersorb XE-348F is impregnated with 3-iodophenol dissolved in propylbenzene. The concentration of 3-iodophenol in propylbenzene is 717 g L1 (3.26 mol L1), which was the approximate maximum solubility of 3-iodophenol in propylbenzene. Ambersorb XE-348F particles were dispersed in the extractant solution in a closed flask for 48 h. After that, the flask was put in an ultrasonic bath to facilitate full impregnation. This was necessary, as Ambersorb XE-348F particles were partially microporous. The particles were assumed to be fully impregnated when no air bubbles were noticed for at least 30 min. After the
impregnation, the particles were filtered, washed with water and dried between two sheets of dry tissue paper, which absorbed extractant solution adhering to the particle surface. This drying procedure was utilized to avoid the possible solvent evaporation, when a drying oven with elevated temperatures is used. The impregnated particles were considered dry when they did not form clusters anymore. For impregnation of Amberlite XAD7 and MPP, the particles were dispersed in solutions of 3-iodophenol in propylbenzene 1 or 3.26 mol L1) for 24 h. Due to the lack of (corg E;0 = 717 g L micropores in these particles, no ultrasound was needed to facilitate the impregnation. After the impregnation, the particles were filtered, washed with water and dried as explained above. The impregnated particles were considered dry when they did not form clusters anymore. The solute removal experiments with solvent impregnated resins (SIRs) were performed with fully impregnated Ambersorb XE348F, fully impregnated Amberlite XAD7 and fully impregnated MPP, which were prepared as described above. The experiments were done in closed glass flasks (50 mL). In each flask, 0.3 g of SIR and 10 mL of solutions of different concentrations of MTBE were mixed by magnetic stirring. The volume of the gas phase above the liquid level was kept minimal to avoid MTBE evaporation. For all experiments the aqueous concentration of MTBE was in the range of 0.25–5 g L1. In order to avoid erroneous KD determinations due to extractant loss during the experiments, the aqueous phase in each experiment was pre-saturated with the extractant/ diluent solution. The pre-saturation was done by contacting demineralized water with a solution of 3-iodophenol in propylbenzene 1 or 3.26 mol L1) for 24 h in a closed and magneti(corg E;0 = 717 g L cally stirred glass flask. After 24 h, the phases were separated gravimetrically and the saturated aqueous phase was decanted from the organic phase. After this, the aqueous phase was spiked with MTBE and the respective SIR was dispersed in this aqueous phase. To ensure that equilibrium was reached, the phases were magnetically stirred for 5 days in case of impregnated Ambersorb XE-348F and 24 h in case of impregnated Amberlite XAD7 and impregnated MPP. The different contact times were based on the adsorption experiments described above. The stirrer speed used was 750 rpm. The experiments were performed at room temperature (20 °C) and atmospheric pressure. The MTBE analysis was performed with the same HSGC method as described above. 2.4. Selectivity The resins used were unimpregnated Ambersorb XE-348F, unimpregnated AquaSorb 202 and MPP impregnated with propyl1 or 3.26 mol L1). The benzene and 3-iodophenol (corg E;0 = 717 g L selectivity of the resins was studied between MTBE and humic acid (technical grade, Sigma–Aldrich, Germany), a common groundwater component. The selectivity experiments were carried out in
44
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
10 mL amber bottles. The selectivity experiments were performed in the same way as the isotherm experiments. The concentration of MTBE was 1 mg L1 and the concentration of humic acid was 0.5 mg L1 according to literature [6]. The dispersions of unimpregnated Ambersorb XE-348F and unimpregnated AquaSorb 202 were magnetically stirred for 5 days at 20 °C, the dispersion of impregnated resin particles was magnetically stirred for 24 h at 20 °C. The selectivity was calculated according to Eq. (1), in which the selectivity aMTBE/HA between MTBE and humic acid (HA) depends on the ratio between the distribution coefficients KD,MTBE of MTBE and KD,HA of humic acid for the studied extractant system.
aMTBE=HA ¼
K D;MTBE K D;HA
ð1Þ
The analysis of MTBE in the aqueous phase was performed with the same HSGC method as described above. The humic acid content of the aqueous phase was analyzed photometrically (Cary 300 Conc UV/vis spectrophotometer, Varian, Netherlands) at 310 nm. The systematic error of this photometric method was approximately 5%. The MTBE content of the organic phase in equilibrium could be determined via a simple mass balance. 2.5. Determination of kinetic MTBE concentration profiles The determination of the MTBE concentration profiles over time was done using a zero-length-column setup [17,23,24], see Fig. A, supplementary information. In such a setup, the bed length is very low compared to a completely filled column, thus ‘‘zero” length. This way, phenomena like axial dispersion or wall effects can be neglected. The setup consisted of a pump (WellChrom HPLC-pump K-1001, Knauer GmbH, Germany), a differential refractometer (Model 2142, LKB Bromma, Sweden) and an Omnifit Column (Bio-Chem Valve/Omnifit, England) fixed in a column oven (ProStar 510 Spark, Varian Inc., Netherlands). The dead volume of the setup was determined to be 5.06 ± 0.095 mL. The borosilicate heavy wall glass column had an inner diameter of 15 mm and a length of 150 mm. The bed length did not exceed 2 mm. The applied flow rate was 29.0 mL min1. The temperature at which the measurements were performed is 20 °C. All kinetic experiments were performed according to the same procedure. The column was loaded with 0.2 g particles. When Ambersorb XE-348F and AquaSorb 202, respectively, were used the setup was rinsed with demineralized water prior to the determination of the MTBE concentration profiles. When impregnated MPP parti1 or 3.08 mol L1) were used, the setup was cles (corg E;0 = 678 g L rinsed with demineralized water pre-saturated with the extractant/diluent solution. After this, the setup was run in a recycle mode until equilibrium was reached. During the recycle mode the inlet and outlet tube of the setup were placed in the same glass vessel. The glass vessel contained 30 mL of demineralized water pre-saturated with the extractant/diluent solution and spiked with MTBE, and was heated with a water bath (F25-HP, Julabo Labortechnik GmbH, Germany) to 20 °C. The initial MTBE concentration was 0.04 mol L1. During the runs the signal of the refractometer was monitored online with the software Penlab v6.21. 2.6. Regeneration For the regeneration experiments a quantity of 3.5 g of impregnated MPP particles was used. The pores of the impregnated MPP particles were completely filled with a solution of 3-iodophenol 1 or 3.03 mol L1). The MTBE loadin propylbenzene (corg E;0 = 667 g L ing of the SIR particles was 1 mol L1 MTBE in the extractant phase inside the pores. The SIR particles were transferred into a semi-pre-
Fig. 1. Equilibrium isotherms of different particles and MTBE (point: experiments, lines: calculated isotherms).
Table 2 Freundlich parameters of the determined isotherms. Particles
KF ((mol g1) (mol L1)n)
n
R2
Ambersorb XE-348F AquaSorb 202 Amberlite XAD7 Impregnated MPP Impregnated Amberlite XAD7
0.026 0.012 0.0093 0.0107 0.0115
0.40 0.47 0.66 0.65 0.67
0.999 0.996 0.999 0.945 0.987
parative NovoGROM stainless steel column, 125 mm 8 mm (Alltech Grom GmbH, Germany). The column was placed in a HP 5890A column oven (Hewlett Packard, USA) and connected to a flame ionization detector (FID), see Fig. B, supplementary information. A gas flow of 17 mL min1 nitrogen, heated with a Julabo F25/ MW Thermostat (Julabo GmbH, Germany) to 120 °C, was applied to the SIR bed inside the column. The organics content of the gas stream after leaving the column was detected by the FID and the signal was recorded with the software Penlab v6.21. 3. Results and discussion 3.1. Selection of an adsorbent as solid matrix for a SIR The main objective of performing adsorption experiments is to determine which adsorbent has the highest affinity for MTBE. For this purpose four different adsorbents were tested, namely macroporous polypropylene MPP, non-ionic aliphatic acrylic Amberlite XAD 7, the carbonaceous resin Ambersorb XE-348F and the activated carbon AquaSorb 202. Some of these adsorbents were later on investigated as a solid support for the extractant. In Fig. 1, the experimentally determined isotherms of the MTBE adsorption experiments are given. The isotherms of Ambersorb XE-348F, AquaSorb 202 and Amberlite XAD7 could be described very well by the Freundlich isotherm Eq. (2), see Fig. 1. The Freundlich parameters for these isotherms are given in Table 2.
qS;F ¼ K F caq S
n
ð2Þ
The MPP isotherm could be accurately described with the constant separation factor isotherm, see Eq. (3), which is based on the mass action law and is a simple mathematical sub-set of the Langmuir equation [26].
q¼
caq S;max
caq S qmax H þ ð1 HÞ caq S
ð3Þ
The separation factor H for the determined MPP isotherm is H = 1.285. The coefficient of determination R2 of this isotherm is 0.985.
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
As can be seen in Fig. 1, the Ambersorb XE-348F particles had the highest MTBE capacity. The reason why Ambersorb XE-348F was the best of the tested MTBE adsorbents is the presence of micropores in the carbonaceous resin. These micropores, formed by the pyrolysis of the ion-exchange resin, are the reason for its high efficiency during water treatment [6], because the micropores are only accessible to comparatively small molecules and thus reduce fouling [27]. However, the adsorption inside the micropores is also the reason for the long time needed to achieve equilibrium [21]. The MTBE capacity of the activated carbon AquaSorb 202 was approximately two thirds of the capacity of Ambersorb XE348F. Within the concentration range of the performed experiments, Amberlite XAD7 reached a maximum MTBE capacity of about half the highest determined MTBE capacity of AquaSorb 202. MPP particles had the lowest MTBE capacity of the tested particles. Due to its high MTBE capacity, Ambersorb XE-348F was selected as solid support for the extractant/diluent system. In addition to the relatively small Ambersorb XE-348F particles, see Table 1, the comparably large Amberlite XAD7 and MPP particles were impregnated. Due to the larger particle size compared to Ambersorb XE-348F and the comparatively high porosity and the large pores, see Table 1, the extractant capacities of Amberlite XAD7 and MPP particles were higher compared to Ambersorb XE-348F particles. This becomes clear, when the particle pore volumes of Ambersorb XE-348F (Vpore = 1.08 105 mL), Amberlite XAD7 (Vpore = 5.8 105 mL) and MPP particles (Vpore = 29.89 105 mL) are compared. Although the MTBE capacity of unimpregnated Amberlite XAD7 and unimpregnated MPP particles was rather low compared to unimpregnated Ambersorb XE-348F, it has been shown before that the impregnation with a suitable extractant could enhance the solute capacity of resin particles with an otherwise low solute capacity drastically [11]. The activated carbon AquaSorb 202 was not considered for impregnation experiments due to a lack of mechanical strength during the performed experiments. 3.2. Solvent impregnated resin for MTBE extraction 1 or Fully impregnated Ambersorb XE-348F (corg E;0 = 717 g L 1 3.26 mol L ) showed a surprising decrease of MTBE capacity compared to unimpregnated Ambersorb XE-348F, see Fig. 2. The factor by which the MTBE capacity decreased in the investigated concen1 . The tration range was as high as 24 at caq eq (MTBE) = 0.04 mol L micropores of Ambersorb XE-348F are crucial for its high solute capacity during water treatment [6]. An explanation for the pronounced decrease of MTBE capacity could be a blocking of the micropores by the extractant and diluent molecules. Micropores can have diameters below 2 nm, so that extractant–extractant, extractant–diluent or diluent–diluent aggregations could block
Fig. 2. Isotherm comparison of unimpregnated and impregnated Ambersorb XE348F.
45
these pores. Furthermore, the penetration of the micropores could be hindered by a disadvantageous viscosity and lack of wetting of the pore walls. Additionally, the adsorption of extractant and/or diluent could also influence the extraction of MTBE. Besides, the particles are relatively small and have only a comparatively low extractant capacity inside the particle pores, see above. The isotherm of impregnated Ambersorb XE-348F could be described with the constant separation factor isotherm (H = 6.43, R2 = 0.998). In case of MPP the unfavourable adsorption isotherm of unimpregnated MPP changed to a more favourable isotherm after impregnation of the MPP particles, see Fig. 3a. Unimpregnated MPP particles do not provide functional groups on the particle surface, as they consist of polypropylene. Thus, MTBE molecules can hardly adsorb on the MPP particle surface via molecular interactions such as e.g. hydrogen bonds with the MPP particles surface. The MTBE capacity of the impregnated MPP particles compared to unimpregnated MPP particles was more than 3 times higher in the investigated MTBE concentration range. Based on the observation that unimpregnated MPP particles have only a negligible MTBE capacity, this increase of MTBE capacity of the impregnated MPP particles could be attributed to the MTBE extraction with the immobilized extractant/diluent system. The effectiveness of the selected extractant diluent system could already be shown by Burghoff et al. [9]. The impregnation with the extractant solution introduced a high number of functional groups into the particles. These functional groups could be accessed by the MTBE molecules to form complexes via hydrogen bonding [9]. Such an effect of improving the solute capacity of an adsorbent with otherwise low adsorptive capacity by introduction of functional groups via impregnation could already be shown for the removal of phenol from aqueous solution with SIRs [11]. When comparing the isotherms of fully impregnated Amberlite XAD7 and unimpregnated Amberlite XAD7, see Fig. 3b, it occurs that also in this case the MTBE capacity of fully impregnated Amberlite XAD7 was higher than the MTBE capacity of unimpregnated Amberlite XAD7, albeit only by a factor of maximum 1.2. This was lower than the increase of MTBE capacity observed for impregnated MPP particles. In comparison to the more effective improvement of the MTBE capacity of MPP particles by impregnation, this rather low increase of MTBE capacity after impregnation of Amberlite XAD7 could be attributed to the adsorption of extractant on the surface of the resin particles. It was shown before, that Amberlite XAD7 has a moderate adsorptive capacity for phenols [11]. Thus, an adsorption of 3-iodophenol on the particle surface seems likely. Such an adsorption of the extractant could decrease the number of available functional groups for hydrogen bonding with MTBE. This way, less extractant moieties are accessible by the MTBE molecules. Nonetheless, this extractant adsorption still allows an improvement of the MTBE capacity of Amberlite XAD7 particles, as shown above. The experimentally determined isotherms of impregnated MPP and impregnated Amberlite XAD7 were well described by the Freundlich isotherm, see Fig. 3. The isotherm parameters of the evaluated isotherms are given in Table 2. The comparison of the data of fully impregnated MPP, Fig. 3a, and fully impregnated Amberlite XAD7, see Fig. 3b, shows that the MTBE capacity of fully impregnated MPP was comparable to the MTBE capacity of fully impregnated XAD7. Both fully impregnated MPP and fully impregnated Amberlite XAD7 had a better affinity for MTBE than impregnated Ambersorb XE-348F. These results show that the impregnation had a significant influence on the capacity of an adsorbent. The MTBE capacity of each MPP and Amberlite XAD7 was increased after impregnation. However, the MTBE capacities of impregnated MPP and impregnated Amberlite XAD7 were still lower than the MTBE capacity of the carbonaceous resin Ambersorb XE-348F.
46
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
Fig. 3. Comparison of SIR isotherms and adsorption isotherms, (a) MPP and (b) Amberlite XAD7.
Table 3 Selectivity values of MTBE and humic acid.
aMTBE/HA
Resin Impregnated MPP
(corg E;0
= 3.26 mol L
1
AquaSorb 202 Ambersorb XE-348F
)
190 20 5
3.3. Selectivity The selectivity of unimpregnated Ambersorb XE-348F, unimpregnated AquaSorb 202 and impregnated MPP for MTBE over humic acid was studied, as humic acid is a common component in groundwater. As can be seen in Table 3, the impregnated MPP particles had the highest selectivity of the investigated resins for MTBE when humic acid was present. The selectivity values of AquaSorb 202 and Ambersorb XE-348F were approximately 10 and 40 times lower, respectively, than the selectivity of impregnated MPP particles. Thus, it can be expected that the MTBE capacity of Ambersorb XE-348F and AquaSorb 202 will decrease significantly when considerable amounts of humic acid are present due to the competitive adsorption of humic acid. This means that the SIR utilization efficiency is higher than compared to the evaluated adsorbents. This is a significant advantage of the prepared SIR over conventional adsorbents. 3.4. Kinetic concentration profiles The determination of the initial kinetic MTBE concentration profiles over time using a zero-length-column setup [17,23,24] showed that the initial MTBE uptake of Ambersorb XE-348F was very fast and took less than 5 min, see Fig. 4. After this initial uptake, the concentration of MTBE decreased further, albeit at a significantly lower rate. In the performed experiment, even after more than 2 days the equilibrium MTBE concentration had not been reached. This is in agreement with previous observations [21]. The initial extraction rate of Ambersorb XE-348F was fast compared to activated carbon and impregnated MPP particles. However, the particles of Ambersorb XE-348F (0.25–0.5 mm) were significantly smaller than the particles of activated carbon (0.425– 1.7 mm) and impregnated MPP (0.8–1.18 mm). The particle size influences the extraction rate. A decrease of the particle diameter can increase extraction rate [17,28]. The effect of the particle size on the overall extraction rate of impregnated MPP particles was investigated with particle fractions of 0.5–0.63 and 1.0–1.6 mm. In Fig. 4 it can be seen that a decrease of the MPP particle size increased the MTBE extraction rate, which was also concluded for the sorption of pentanal on impregnated XAD16 by Babic et al. [17]. As can be seen, a reduction of the SIR particle size from 1.0–1.6 mm to
Fig. 4. Concentration profiles of MTBE extraction kinetics of AquaSorb 202 (particle 1 or 0.04 mol L1, diameter 0.425–1.7 mm), impregnated MPP (caq MTBE;0 = 3.5 g L 1 1 = 676 g L or 3.08 mol L ) (particle diameter 0.8–1.18 mm) and Ambersorb corg E;0 XE-348F (particle diameter 0.25–0.5 mm) determined with a zero-length-column setup.
0.5–0.63 mm decreased the time until the equilibrium MTBE concentration was reached from more than 60 min down to 30 min. This can be attributed to the faster diffusion in smaller particles, as the diffusion time constant is inversely proportional to the square of the particle radius. However, if the MTBE removal with the prepared SIR would entirely depend on diffusion, the extraction rate should increase by a factor of 4 instead of a factor of 2, as observed in this case. Possibly, the MTBE removal depends on a combination of the diffusion of the solute, the extractant and the formed complex inside the pores, and on the other hand the chemical complexation between MTBE and the extractant [17,29]. Nonetheless, diffusion inside the SIR particle can be considered as one of the limiting factors of the MTBE extraction in the investigated system and the extraction rate of MTBE by the impregnated MPP particles can be enhanced by decreasing the particle size. Although the equilibrium MTBE capacity of AquaSorb 202 was higher than the equilibrium MTBE capacity of impregnated MPP particles, the MTBE extraction rate of AquaSorb 202 was significantly lower compared to impregnated MPP particles with a particle size of 0.5–0.63 mm. 3.5. Regeneration The regeneration of impregnated MPP particles was achieved with a hot nitrogen gas stream (120 °C). In an integrated process, off-gas or steam might also be used. Fig. 5 depicts the concentration profile of only MTBE during thermal regeneration. The initial time lag of the concentration profile during regeneration can be ex-
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
47
Acknowledgements This work is financially supported by the Technology Foundation STW. Additional support by Veolia MPP Systems, Akzo Nobel Chemicals, Diosynth, INEOS Phenol, Vitens and our colleagues ir. R. Cuypers, Prof. Dr. H. Zuilhof at the University of Wageningen and Prof. Dr. E.J.R. Sudhölter at Delft University is also acknowledged. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.reactfunctpolym.2009.10.002. Fig. 5. Profile of thermal regeneration of impregnated MPP particles (flow rate 1 or 1 mol L1). N2 = 17 mL min1, T = 120 °C, caq MTBE;0 = 88.15 g L
plained by the time that is needed to completely heat up the whole particle bed. The initial shoulder of the profile was probably caused by the fast evaporation of MTBE still left on the surface of the wet particles after MTBE extraction. After this, the MTBE contained by the extractant inside the pores was released. The time until the complete amount of MTBE contained in the SIR particles had evaporated took approximately 20 min. In order to prevent diluent loss during regeneration, the diluent propylbenzene needs to be optimized further in terms of volatility. This can be achieved by the addition of alkyl chains or by modification of the existing propyl chain. Increasing the alkyl chain length and increasing the degree of chain branching can decrease the vapor pressure of the diluent. This is, however, not incorporated into this study and will be investigated further.
4. Conclusions In this study the use of a solvent impregnated resin (SIR) to remove MTBE from water is investigated. Of the investigated adsorbents, which serve as solid matrix for a SIR, the carbonaceous resin Ambersorb XE-348F has the highest MTBE capacity. However, the MTBE capacity of impregnated Ambersorb XE-348F significantly decreases compared to the unimpregnated particles. The impregnation of MPP particles increases the MTBE capacity by a factor of more than 3 compared to the unimpregnated MPP. This relative increase of MTBE capacity is more pronounced than in the case of impregnated Amberlite XAD7. Although Ambersorb XE-348F has a higher MTBE capacity than the prepared impregnated MPP particles, the selectivity of the impregnated MPP particles for MTBE in the presence of humic acid is 10–40 times higher compared to Ambersorb XE-348F and the activated carbon AquaSorb 202. Exploratory kinetics measurements in a zero-length-column setup show that the initial extraction rate of Ambersorb XE-348F is faster than of impregnated MPP. However, the particles of Ambersorb XE-348F are smaller than the MPP particles. It can be shown that a decrease of the MPP particle size increases the extraction rate of the used SIR. The prepared SIR particles can be successfully regenerated with a hot nitrogen gas stream (120 °C).
References [1] A.A.M. Langenhoff, Netherlands Centre for Soil Quality Management and Knowledge Transfer (SKB SV-018), 2000. [2] R. Schuhmacher, M. Führer, W. Kandler, C. Stadlmann, R. Krska, Anal. Bioanal. Chem. 377 (2003) 1140–1147. [3] W.E. May, S.P. Wasik, M.M. Miller, Y.B. Tewari, J.M. Brown-Thomas, R.N. Goldberg, J. Chem. Eng. Data 28 (1983) 197–200. [4] A.A. Keller, Health and Environment Assessment of MTBE, Report to Legislature of the State of California, vol. III, 1998. [5] U. Peters, F. Nierlich, E. Schulte-Körne, M. Sakuth, R. Deeb, M. Laugier, M. Suominen, M. Kavanaugh, Methyl tert-butyl ether, in: Ullmann’s Encyclopedia of Industrial Chemistry, John Wiley and Sons, Inc., Weinheim, 2003. [6] G. Melin, California MTBE Research Partnership: Treatment Technologies for Removal of Methyl tert-Butyl Ether from Drinking Water, National Water Research Institute, 2000. [7] B.A. Ray, A. Selvakumar, N.A. Tafuri, Treatment of methyl tert-butyl ether (MTBE) contaminated waters with Fenton’s reagent, Report to the US Environmental Protection Agency, USA, 2003, pp. 1–18. [8] T.C. Shih, M. Wangpaichitr, M. Suffet, Water Res. 37 (2003) 375–385. [9] B. Burghoff, J. Schiferli, J. Sousa Marques, A.B. de Haan, Chem. Eng. Sci. 64 (2008) 2887–2892. [10] M. Traving, H.-J. Bart, Chem. Eng. Technol. 25 (2002) 997–1003. [11] B. Burghoff, E.L.V. Goetheer, A.B. de Haan, React. Funct. Polym. 68 (2008) 1314–1324. [12] K. Babic´, A.G.J. van der Ham, A.B. de Haan, React. Funct. Polym. 66 (2006) 1494–1505. [13] J. Serarols, J. Poch, I. Villaescusa, React. Funct. Polym. 48 (2001) 37–51. [14] A. Warshawsky, J.L. Cortina, M. Aguilar, K. Jerabek, in: International Solvent Extraction Conference, Barcelona, Spain, 1999, pp. 1267–1272. [15] Y. Wang, C. Wang, A. Warshawsky, B. Berkowitz, Sep. Sci. Technol. 38 (2003) 149–163. [16] MPPSystems, Macro Porous Polymer Extraction System – Water Purification, Akzo Nobel, 2005. [17] K. Babic´, A.G.J. van der Ham, A.B. de Haan, Adsorption 14 (2008) 357–366. [18] R.-S. Juang, H.-L. Chang, Ind. Eng. Chem. Res. 34 (1995) 1294–1301. [19] A. Kostova, H.-J. Bart, Chem. Ing. Tech. 76 (2004) 1743–1748. [20] H. Kitazaki, M. Ishimaru, K. Inoue, K. Yoshida, S. Nakamura, in: International Solvent Extraction Conference, Melbourne, Australia, 1996, pp. 1667–1672. [21] H.-W. Hung, T.-F. Lin, J. Hazard. Mater. B135 (2006) 210–217. [22] F.J. Stevenson, Humus Chemistry: Genesis, Composition, Reactions, John Wiley and Sons, New York, 1994. [23] M. Eic, D.M. Ruthven, Zeolites 8 (1988) 40–45. [24] T. Djekic, V. Tripkovic, A.G.J. van der Ham, A.B. de Haan, Adsorption 13 (2007) 231–237. [25] T.C. Schmidt, H.-A. Duong, M. Berg, B.S. Haderlein, Analyst 126 (2001) 405– 413. [26] E. Arevalo, M. Rendueles, A. Fernandez, M. Diaz, Sep. Purif. Technol. 18 (2000) 217–225. [27] R.A. Rodriguez, J.M. Davidson, D.N. Creek, A.E. Flores, A.J. Stocking, M.C. Kavanaugh, H. Eylers, M. Wooden, T. Herson, Center for Groundwater Restoration and Protection, National Water Research Institute, Fountain Valley, 1999. [28] J. Serarols, J. Poch, I. Villaescusa, React. Funct. Polym. 48 (2001) 53–63. [29] B. Burghoff, E. Zondervan, A.B. de Haan, React. Funct. Polym. 69 (2008) 264– 271.
Contents lists available at ScienceDirect
Reactive & Functional Polymers journal homepage: www.elsevier.com/locate/react
Solvent impregnated resins for MTBE removal from aqueous environments Bernhard Burghoff a,*, Joao Sousa Marques b, Bart M. van Lankvelt a, Andre B. de Haan a a b
Eindhoven University of Technology, Den Dolech 2, 5600 MB Eindhoven, Netherlands Universidade do Porto, Faculdade de Engenharia, 4000 Porto, Portugal
a r t i c l e
i n f o
Article history: Received 6 July 2009 Received in revised form 22 September 2009 Accepted 11 October 2009 Available online 5 November 2009 Keywords: Extraction MTBE Solvent impregnated resins
a b s t r a c t MTBE removal from groundwater is difficult. State-of-the-art processes such as air stripping or adsorption can have significant drawbacks, like high energy consumption, low capacity and low selectivity. This is why the alternative technology of solvent impregnated resins is investigated. The extractant used is 3iodophenol diluted with propylbenzene. From solids screening and impregnation experiments macroporous polypropylene (MPP) particles appear as a suitable solid support. The MTBE capacity of impregnated MPP is lower than that of a carbonaceous resin, but the selectivity of the SIR between MTBE and humic acid is significantly higher. Solvent impregnated resin (SIR) regeneration can be easily achieved with hot gas. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Methyl tert-butyl ether (MTBE) is mainly used as a fuel oxygenate added to petrol as an octane enhancer which helps to prevent engine ‘knocking’, and to increase oxygen content [1]. Due to its high water solubility compared with other fuel constituents (water solubility of MTBE at 25 °C: 51 g L1 [2], water solubility of benzene at 25 °C: 1.79 g L1 [3], its low soil adsorption coefficient and its low Henry coefficient leakages from storage facilities cause immediate ground water contamination [1]. In the United States, of the 1000 groundwater sites tested, more than 80% were shown to be contaminated with MTBE and concentrations as high as 23 mg L1 have been measured [4]. Since MTBE has a very low taste and odor threshold in water of 40 and 15 lg L1 [5], respectively, it needs to be removed during the production of high-quality potable water from ground water. Conventional processes for MTBE removal are air stripping and granular activated carbon (GAC) adsorption [5]. Although air stripping is an established technology, large air streams are required for the removal of low concentration MTBE, which results in high process costs [4]. GAC adsorption is a commercial technology and regularly used in water treatment. It appears to be efficient for MTBE concentrations below 200 lg L1 [5]. However, when MTBE is present in concentrations above 2000 lg L1 [6], this process can only be used as a polishing step followed by biological or chemical treatment or air stripping [5]. This is because of the low affinity * Corresponding author. Present address: TU Dortmund, Emil-Figge-Str. 70, 44227 Dortmund, Germany. Tel.: +49 2317557148; fax: +49 2317552431. E-mail address: [email protected] (B. Burghoff). 1381-5148/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.reactfunctpolym.2009.10.002
of MTBE for adsorption on solids. Another drawback of GAC is the reduced life-time of typically 10–20 cycles [7]. Furthermore a competitive adsorption of other organic compounds in ground water, such as either fuel components or humic acid, results in a GAC affinity decrease for MTBE [8]. MTBE removal with synthetic resins was first investigated in the 1970s as an alternative to GAC [6]. Synthetic resins have a higher number of possible adsorption sites than GAC. Synthetic resins can be divided into two categories, which are polymeric resins and carbonaceous resins. The polymeric resins are usually based on cross-linked polymers. These polymers can have polystyrene, phenolformaldehyde, or acrylate matrices [6]. Carbonaceous resins are produced as a product of the partial pyrolysis of macroporous polymer particles. Due to the pyrolysis micropores will form, which are critical for the application of carbonaceous resins in water treatment. An alternative process for removal of low concentration solutes from water is extraction. In a previous study, 3-iodophenol is suggested as an effective MTBE extractant [9]. The immobilization of the extractant in commercially available macroporous resin particles, so-called solvent impregnated resins (SIRs), avoids emulsification and simplifies the phase separation after extraction [10,11]. Another advantage of SIR technology compared to conventional liquid–liquid extraction is that due to the impregnation the loss of extractant into the aqueous phase during extraction decreases [10,12]. Possible drawbacks of SIR technology, such as leaching of the extractant or clogging of a fixed bed by attrition of the particles, can be remedied by choosing the proper system. This implies a suitable extractant with low water solubility, which is sufficiently retained inside the pores, and the selection of sturdy
42
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
Nomenclature c KD KF n q Vpore
concentration (mol L1) overall distribution coefficient (mol L1) (mol L1)1 Freundlich isotherm parameter (mol g1) (mol L1)n Freundlich isotherm parameter () solute loading of particle (mol g1) particle pore volume (mL)
Greek letters selectivity H constant separation factor
Subscripts 0 parameter at time t = 0 E extractant HA humic acid max maximum MTBE methyl tert-butyl ether S solute SIR solvent impregnated resin
a
Superscripts aq aqueous phase org organic phase particles as a solid support for the extractant. Applications of SIRs for heavy metal recovery are already well known [13–15]. Recently, also other applications have been developed, e.g. the recovery of hydrocarbons, such as aliphatics, aromatics and halogenated hydrocarbons, on a large-scale on offshore oil platforms [16]. In such an application, where the SIR particles are contained in a packed bed, flow rates higher than 0.5 m3 h1 without maximum flow restrictions can be treated cost competitive to air stripping/ activated carbon, steam stripping and bio treatment systems [16]. The removal of polar organics like phenol [11], aldehydes [17], organic acids [18], amino acids [19] and flavonoids from aqueous media [20] is only carried out on a bench-scale or pilotscale. Large-scale applications for the separation of polar solutes, like e.g. MTBE, do not yet exist. The design of a SIR containing a suitable MTBE extractant can close this gap. The objective of this study is to show that MTBE can be removed from aqueous solutions with a SIR. For this purpose, a preliminary SIR is designed to demonstrate the proof-of-principle. As yet, for the SIR design only the preliminary extractant 3-iodophenol is available based on a previous MTBE extractant screening [9]. The extractant 3-iodophenol is not optimal in terms of water solubility and environmental impact, which should both be negligible for an environmental application. In addition to this, it is solid at standard conditions and requires a diluent for application in an extraction process. Adding a branched alkyl chain in para position at the aromatic ring of 3-iodophenol will decrease water solubility and thus the possible release into the environment appreciably. In order to decrease the melting point below standard temperature and thus avoid the need for a diluent, an extractant blend consisting of components with differing alkyl chain lengths can be created. However, the synthesis of such an optimized extractant including the necessary evaluation of physical properties, environmental impact and MTBE capacity implies great experimental effort and is time consuming. This extractant optimization is not within the scope of this study. The preliminary SIR is thus simply based on 3-iodophenol dissolved in a diluent and the current study focuses mainly on the proof-of-principle to remove MTBE from water with SIRs. In order to show that SIRs can in principle be used to remove MTBE from aqueous solutions a suitable SIR is developed. In addition to an extractant, the solid matrix for the solvent impregnated resin needs to be selected. Preferably, the particles should have an affinity for MTBE, in order to have a supportive effect by adsorption on the particle surface in addition to the extraction by the immobilized extractant. This is why adsorbents with a high MTBE capacity are preferred. A literature review gives information about suitable adsorbents. Four different types of adsorbents appear to be suitable. These adsorbents are carbonaceous resins (e.g. Ambersorb XE-348F), activated carbon (e.g. AquaSorb 202) [6,21] and
synthetic resins, such as non-ionic aliphatic acrylic particles (e.g. Amberlite XAD 7) [5]. Macroporous polypropylene (e.g. MPP) particles are tested, as they have been shown in other SIR applications to give a good performance [11,16] due to a high extractant capacity inside the pores and a high mechanical stability. Adsorbents with a high MTBE capacity, a high extractant capacity inside the pores and a high mechanical stability are then used as solid supports for the extractant. A further issue of investigation is the MTBE selectivity in the presence of humic acid. This is an important topic in environmental applications, when MTBE is supposed to be removed from groundwater. Humic acid is one of the main constituents of humic substances in groundwater. The stability of humic substances is very high compared to other organic fractions in the soil [22]. High concentrations of natural organic matter, such as humic acid, compete with MTBE adsorption on activated carbon and synthetic resins, and can decrease the adsorbent usage efficiency and consequently increase the process costs [5]. A kinetic study of the prepared SIR particles is carried out. In the scope of this study, the MTBE extraction profiles of Ambersorb XE-348F, the activated carbon AquaSorb 202 and prepared SIRs are compared. The influence of the particle size on the determined concentration profiles is investigated. A zero-length-column setup, which will be explained below, is used to minimize axial dispersion and external mass transfer [23,24]. Finally, the regenerability of the selected SIR is investigated. In an exploratory regeneration study, the feasibility of using a hot gas stream to evaporate MTBE from the SIR particles after MTBE extraction is the focus of the regenerability considerations.
2. Materials and methods 2.1. Substances MTBE (99.5%, Merck KGaA, Germany) were used without further treatment. As extractant 3-iodophenol (98%, Sigma–Aldrich, Germany) was used. Propylbenzene (99%, Fluka Chemie AG, Switzerland) was used as a diluent for 3-iodophenol. Above listed substances were all utilized without further treatment. The polymeric resins used were MPP (VWS MPP Systems, Netherlands), Amberlite XAD7 (Rohm and Haas Company, France) and Ambersorb XE-348F (Sigma–Aldrich Chemie GmbH, USA). The activated carbon used was AquaSorb 202 (Jacobi Carbons, Germany). AquaSorb 202 and Ambersorb XE-348F were used as received. The MPP particles were sieved, then washed with acetone and left in a rotary evaporator (BÜCHI Rotavapor R-200, BÜCHI Labortechnik AG, Switzerland) equipped with a heating bath (BÜCHI Heating
43
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47 Table 1 Particle characteristics.
a b c d
Property Material
Ambersorb XE-348F Carbonaceous adsorbent
AquaSorb 202 Activated carbon of bituminous coal
Amberlite XAD7 Macroreticular aliphatic acrylic cross-linked polymer
MPP Semi-crystalline polypropylene
Porosity Density (g mL1) Surface area (m2 g1) Pore size (nm) Particle size (mm)
0.45b 1.32b 750a N.A. 0.25–0.5a
0.51a 1.22b 1100a N.A. 0.425–1.7a
P0.5a 1.07a P380a N.A. 0.56–0.71a
0.68a 1.15b 11.5c <1000d 0.8–1.18 (98%)b
Manufacturer’s data. Own measurement. Babic´ et al. [17]. Based on analysis of SEM images.
Bath B-490, BÜCHI Labortechnik AG, Switzerland) for 12 h at vacuum and 70 °C to remove solvent residues. Before initial use Amberlite XAD7 was rinsed with demineralized water, decanted, rinsed with acetone, decanted again and finally rinsed with hexane and subsequently decanted. The Amberlite XAD7 particles were left in a rotary evaporator for 12 h at vacuum and 70 °C to evaporate solvent residues. 2.2. Screening of adsorbents/solid matrices The characteristics of the different particles used are given in Table 1. Isotherm experiments were conducted in closed 50 mL glass flasks. In each bottle, 0.3 g of adsorbent and 10 mL of solutions of different concentrations of MTBE were mixed by magnetic stirring at 20 °C. For all experiments the aqueous concentration of MTBE was in the range of 0.25–5 g L1. This concentration range was selected due to the available analytical equipment. In case of Amberlite XAD7 and MPP particles the contacting time was 24 h in order to ensure that equilibrium was reached. Previously performed tentative experiments suggested that this period of 24 h is sufficient to reach equilibrium. In case of Ambersorb XE-348F and AquaSorb 202 the contacting time was 5 days [21]. The stirring speed during these experiments was adjusted to 500 rpm. The MTBE content of the aqueous phase was determined with head space gas chromatography (HSGC) using a Varian CP-3800 gas chromatograph (Varian Inc., Netherlands) equipped with a CombiPal Atas Focus (Atas GL International B.V., Netherlands) including an 11/70 injector. The column used was a factor four capillary column CP 8945, VF5 ms, with 30 m length, 0.25 mm internal diameter and 0.50 lm film thickness (Varian Inc., Netherlands) connected to an FID. Samples for analysis were prepared in 20 mL vials. The transfer of MTBE to the gas phase was enhanced by thermostatic control of the samples at 70 °C [25]. To ensure that the vapor–liquid phase equilibrium was reached each vial was agitated for 10 min before the injection. The concentration of MTBE in organic and solid phases was calculated by a simple mass balance [12]. The systematic error of this analytical method was below approximately 9%. 2.3. Solvent impregnated resin preparation and characterization The carbonaceous resin Ambersorb XE-348F is impregnated with 3-iodophenol dissolved in propylbenzene. The concentration of 3-iodophenol in propylbenzene is 717 g L1 (3.26 mol L1), which was the approximate maximum solubility of 3-iodophenol in propylbenzene. Ambersorb XE-348F particles were dispersed in the extractant solution in a closed flask for 48 h. After that, the flask was put in an ultrasonic bath to facilitate full impregnation. This was necessary, as Ambersorb XE-348F particles were partially microporous. The particles were assumed to be fully impregnated when no air bubbles were noticed for at least 30 min. After the
impregnation, the particles were filtered, washed with water and dried between two sheets of dry tissue paper, which absorbed extractant solution adhering to the particle surface. This drying procedure was utilized to avoid the possible solvent evaporation, when a drying oven with elevated temperatures is used. The impregnated particles were considered dry when they did not form clusters anymore. For impregnation of Amberlite XAD7 and MPP, the particles were dispersed in solutions of 3-iodophenol in propylbenzene 1 or 3.26 mol L1) for 24 h. Due to the lack of (corg E;0 = 717 g L micropores in these particles, no ultrasound was needed to facilitate the impregnation. After the impregnation, the particles were filtered, washed with water and dried as explained above. The impregnated particles were considered dry when they did not form clusters anymore. The solute removal experiments with solvent impregnated resins (SIRs) were performed with fully impregnated Ambersorb XE348F, fully impregnated Amberlite XAD7 and fully impregnated MPP, which were prepared as described above. The experiments were done in closed glass flasks (50 mL). In each flask, 0.3 g of SIR and 10 mL of solutions of different concentrations of MTBE were mixed by magnetic stirring. The volume of the gas phase above the liquid level was kept minimal to avoid MTBE evaporation. For all experiments the aqueous concentration of MTBE was in the range of 0.25–5 g L1. In order to avoid erroneous KD determinations due to extractant loss during the experiments, the aqueous phase in each experiment was pre-saturated with the extractant/ diluent solution. The pre-saturation was done by contacting demineralized water with a solution of 3-iodophenol in propylbenzene 1 or 3.26 mol L1) for 24 h in a closed and magneti(corg E;0 = 717 g L cally stirred glass flask. After 24 h, the phases were separated gravimetrically and the saturated aqueous phase was decanted from the organic phase. After this, the aqueous phase was spiked with MTBE and the respective SIR was dispersed in this aqueous phase. To ensure that equilibrium was reached, the phases were magnetically stirred for 5 days in case of impregnated Ambersorb XE-348F and 24 h in case of impregnated Amberlite XAD7 and impregnated MPP. The different contact times were based on the adsorption experiments described above. The stirrer speed used was 750 rpm. The experiments were performed at room temperature (20 °C) and atmospheric pressure. The MTBE analysis was performed with the same HSGC method as described above. 2.4. Selectivity The resins used were unimpregnated Ambersorb XE-348F, unimpregnated AquaSorb 202 and MPP impregnated with propyl1 or 3.26 mol L1). The benzene and 3-iodophenol (corg E;0 = 717 g L selectivity of the resins was studied between MTBE and humic acid (technical grade, Sigma–Aldrich, Germany), a common groundwater component. The selectivity experiments were carried out in
44
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
10 mL amber bottles. The selectivity experiments were performed in the same way as the isotherm experiments. The concentration of MTBE was 1 mg L1 and the concentration of humic acid was 0.5 mg L1 according to literature [6]. The dispersions of unimpregnated Ambersorb XE-348F and unimpregnated AquaSorb 202 were magnetically stirred for 5 days at 20 °C, the dispersion of impregnated resin particles was magnetically stirred for 24 h at 20 °C. The selectivity was calculated according to Eq. (1), in which the selectivity aMTBE/HA between MTBE and humic acid (HA) depends on the ratio between the distribution coefficients KD,MTBE of MTBE and KD,HA of humic acid for the studied extractant system.
aMTBE=HA ¼
K D;MTBE K D;HA
ð1Þ
The analysis of MTBE in the aqueous phase was performed with the same HSGC method as described above. The humic acid content of the aqueous phase was analyzed photometrically (Cary 300 Conc UV/vis spectrophotometer, Varian, Netherlands) at 310 nm. The systematic error of this photometric method was approximately 5%. The MTBE content of the organic phase in equilibrium could be determined via a simple mass balance. 2.5. Determination of kinetic MTBE concentration profiles The determination of the MTBE concentration profiles over time was done using a zero-length-column setup [17,23,24], see Fig. A, supplementary information. In such a setup, the bed length is very low compared to a completely filled column, thus ‘‘zero” length. This way, phenomena like axial dispersion or wall effects can be neglected. The setup consisted of a pump (WellChrom HPLC-pump K-1001, Knauer GmbH, Germany), a differential refractometer (Model 2142, LKB Bromma, Sweden) and an Omnifit Column (Bio-Chem Valve/Omnifit, England) fixed in a column oven (ProStar 510 Spark, Varian Inc., Netherlands). The dead volume of the setup was determined to be 5.06 ± 0.095 mL. The borosilicate heavy wall glass column had an inner diameter of 15 mm and a length of 150 mm. The bed length did not exceed 2 mm. The applied flow rate was 29.0 mL min1. The temperature at which the measurements were performed is 20 °C. All kinetic experiments were performed according to the same procedure. The column was loaded with 0.2 g particles. When Ambersorb XE-348F and AquaSorb 202, respectively, were used the setup was rinsed with demineralized water prior to the determination of the MTBE concentration profiles. When impregnated MPP parti1 or 3.08 mol L1) were used, the setup was cles (corg E;0 = 678 g L rinsed with demineralized water pre-saturated with the extractant/diluent solution. After this, the setup was run in a recycle mode until equilibrium was reached. During the recycle mode the inlet and outlet tube of the setup were placed in the same glass vessel. The glass vessel contained 30 mL of demineralized water pre-saturated with the extractant/diluent solution and spiked with MTBE, and was heated with a water bath (F25-HP, Julabo Labortechnik GmbH, Germany) to 20 °C. The initial MTBE concentration was 0.04 mol L1. During the runs the signal of the refractometer was monitored online with the software Penlab v6.21. 2.6. Regeneration For the regeneration experiments a quantity of 3.5 g of impregnated MPP particles was used. The pores of the impregnated MPP particles were completely filled with a solution of 3-iodophenol 1 or 3.03 mol L1). The MTBE loadin propylbenzene (corg E;0 = 667 g L ing of the SIR particles was 1 mol L1 MTBE in the extractant phase inside the pores. The SIR particles were transferred into a semi-pre-
Fig. 1. Equilibrium isotherms of different particles and MTBE (point: experiments, lines: calculated isotherms).
Table 2 Freundlich parameters of the determined isotherms. Particles
KF ((mol g1) (mol L1)n)
n
R2
Ambersorb XE-348F AquaSorb 202 Amberlite XAD7 Impregnated MPP Impregnated Amberlite XAD7
0.026 0.012 0.0093 0.0107 0.0115
0.40 0.47 0.66 0.65 0.67
0.999 0.996 0.999 0.945 0.987
parative NovoGROM stainless steel column, 125 mm 8 mm (Alltech Grom GmbH, Germany). The column was placed in a HP 5890A column oven (Hewlett Packard, USA) and connected to a flame ionization detector (FID), see Fig. B, supplementary information. A gas flow of 17 mL min1 nitrogen, heated with a Julabo F25/ MW Thermostat (Julabo GmbH, Germany) to 120 °C, was applied to the SIR bed inside the column. The organics content of the gas stream after leaving the column was detected by the FID and the signal was recorded with the software Penlab v6.21. 3. Results and discussion 3.1. Selection of an adsorbent as solid matrix for a SIR The main objective of performing adsorption experiments is to determine which adsorbent has the highest affinity for MTBE. For this purpose four different adsorbents were tested, namely macroporous polypropylene MPP, non-ionic aliphatic acrylic Amberlite XAD 7, the carbonaceous resin Ambersorb XE-348F and the activated carbon AquaSorb 202. Some of these adsorbents were later on investigated as a solid support for the extractant. In Fig. 1, the experimentally determined isotherms of the MTBE adsorption experiments are given. The isotherms of Ambersorb XE-348F, AquaSorb 202 and Amberlite XAD7 could be described very well by the Freundlich isotherm Eq. (2), see Fig. 1. The Freundlich parameters for these isotherms are given in Table 2.
qS;F ¼ K F caq S
n
ð2Þ
The MPP isotherm could be accurately described with the constant separation factor isotherm, see Eq. (3), which is based on the mass action law and is a simple mathematical sub-set of the Langmuir equation [26].
q¼
caq S;max
caq S qmax H þ ð1 HÞ caq S
ð3Þ
The separation factor H for the determined MPP isotherm is H = 1.285. The coefficient of determination R2 of this isotherm is 0.985.
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
As can be seen in Fig. 1, the Ambersorb XE-348F particles had the highest MTBE capacity. The reason why Ambersorb XE-348F was the best of the tested MTBE adsorbents is the presence of micropores in the carbonaceous resin. These micropores, formed by the pyrolysis of the ion-exchange resin, are the reason for its high efficiency during water treatment [6], because the micropores are only accessible to comparatively small molecules and thus reduce fouling [27]. However, the adsorption inside the micropores is also the reason for the long time needed to achieve equilibrium [21]. The MTBE capacity of the activated carbon AquaSorb 202 was approximately two thirds of the capacity of Ambersorb XE348F. Within the concentration range of the performed experiments, Amberlite XAD7 reached a maximum MTBE capacity of about half the highest determined MTBE capacity of AquaSorb 202. MPP particles had the lowest MTBE capacity of the tested particles. Due to its high MTBE capacity, Ambersorb XE-348F was selected as solid support for the extractant/diluent system. In addition to the relatively small Ambersorb XE-348F particles, see Table 1, the comparably large Amberlite XAD7 and MPP particles were impregnated. Due to the larger particle size compared to Ambersorb XE-348F and the comparatively high porosity and the large pores, see Table 1, the extractant capacities of Amberlite XAD7 and MPP particles were higher compared to Ambersorb XE-348F particles. This becomes clear, when the particle pore volumes of Ambersorb XE-348F (Vpore = 1.08 105 mL), Amberlite XAD7 (Vpore = 5.8 105 mL) and MPP particles (Vpore = 29.89 105 mL) are compared. Although the MTBE capacity of unimpregnated Amberlite XAD7 and unimpregnated MPP particles was rather low compared to unimpregnated Ambersorb XE-348F, it has been shown before that the impregnation with a suitable extractant could enhance the solute capacity of resin particles with an otherwise low solute capacity drastically [11]. The activated carbon AquaSorb 202 was not considered for impregnation experiments due to a lack of mechanical strength during the performed experiments. 3.2. Solvent impregnated resin for MTBE extraction 1 or Fully impregnated Ambersorb XE-348F (corg E;0 = 717 g L 1 3.26 mol L ) showed a surprising decrease of MTBE capacity compared to unimpregnated Ambersorb XE-348F, see Fig. 2. The factor by which the MTBE capacity decreased in the investigated concen1 . The tration range was as high as 24 at caq eq (MTBE) = 0.04 mol L micropores of Ambersorb XE-348F are crucial for its high solute capacity during water treatment [6]. An explanation for the pronounced decrease of MTBE capacity could be a blocking of the micropores by the extractant and diluent molecules. Micropores can have diameters below 2 nm, so that extractant–extractant, extractant–diluent or diluent–diluent aggregations could block
Fig. 2. Isotherm comparison of unimpregnated and impregnated Ambersorb XE348F.
45
these pores. Furthermore, the penetration of the micropores could be hindered by a disadvantageous viscosity and lack of wetting of the pore walls. Additionally, the adsorption of extractant and/or diluent could also influence the extraction of MTBE. Besides, the particles are relatively small and have only a comparatively low extractant capacity inside the particle pores, see above. The isotherm of impregnated Ambersorb XE-348F could be described with the constant separation factor isotherm (H = 6.43, R2 = 0.998). In case of MPP the unfavourable adsorption isotherm of unimpregnated MPP changed to a more favourable isotherm after impregnation of the MPP particles, see Fig. 3a. Unimpregnated MPP particles do not provide functional groups on the particle surface, as they consist of polypropylene. Thus, MTBE molecules can hardly adsorb on the MPP particle surface via molecular interactions such as e.g. hydrogen bonds with the MPP particles surface. The MTBE capacity of the impregnated MPP particles compared to unimpregnated MPP particles was more than 3 times higher in the investigated MTBE concentration range. Based on the observation that unimpregnated MPP particles have only a negligible MTBE capacity, this increase of MTBE capacity of the impregnated MPP particles could be attributed to the MTBE extraction with the immobilized extractant/diluent system. The effectiveness of the selected extractant diluent system could already be shown by Burghoff et al. [9]. The impregnation with the extractant solution introduced a high number of functional groups into the particles. These functional groups could be accessed by the MTBE molecules to form complexes via hydrogen bonding [9]. Such an effect of improving the solute capacity of an adsorbent with otherwise low adsorptive capacity by introduction of functional groups via impregnation could already be shown for the removal of phenol from aqueous solution with SIRs [11]. When comparing the isotherms of fully impregnated Amberlite XAD7 and unimpregnated Amberlite XAD7, see Fig. 3b, it occurs that also in this case the MTBE capacity of fully impregnated Amberlite XAD7 was higher than the MTBE capacity of unimpregnated Amberlite XAD7, albeit only by a factor of maximum 1.2. This was lower than the increase of MTBE capacity observed for impregnated MPP particles. In comparison to the more effective improvement of the MTBE capacity of MPP particles by impregnation, this rather low increase of MTBE capacity after impregnation of Amberlite XAD7 could be attributed to the adsorption of extractant on the surface of the resin particles. It was shown before, that Amberlite XAD7 has a moderate adsorptive capacity for phenols [11]. Thus, an adsorption of 3-iodophenol on the particle surface seems likely. Such an adsorption of the extractant could decrease the number of available functional groups for hydrogen bonding with MTBE. This way, less extractant moieties are accessible by the MTBE molecules. Nonetheless, this extractant adsorption still allows an improvement of the MTBE capacity of Amberlite XAD7 particles, as shown above. The experimentally determined isotherms of impregnated MPP and impregnated Amberlite XAD7 were well described by the Freundlich isotherm, see Fig. 3. The isotherm parameters of the evaluated isotherms are given in Table 2. The comparison of the data of fully impregnated MPP, Fig. 3a, and fully impregnated Amberlite XAD7, see Fig. 3b, shows that the MTBE capacity of fully impregnated MPP was comparable to the MTBE capacity of fully impregnated XAD7. Both fully impregnated MPP and fully impregnated Amberlite XAD7 had a better affinity for MTBE than impregnated Ambersorb XE-348F. These results show that the impregnation had a significant influence on the capacity of an adsorbent. The MTBE capacity of each MPP and Amberlite XAD7 was increased after impregnation. However, the MTBE capacities of impregnated MPP and impregnated Amberlite XAD7 were still lower than the MTBE capacity of the carbonaceous resin Ambersorb XE-348F.
46
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
Fig. 3. Comparison of SIR isotherms and adsorption isotherms, (a) MPP and (b) Amberlite XAD7.
Table 3 Selectivity values of MTBE and humic acid.
aMTBE/HA
Resin Impregnated MPP
(corg E;0
= 3.26 mol L
1
AquaSorb 202 Ambersorb XE-348F
)
190 20 5
3.3. Selectivity The selectivity of unimpregnated Ambersorb XE-348F, unimpregnated AquaSorb 202 and impregnated MPP for MTBE over humic acid was studied, as humic acid is a common component in groundwater. As can be seen in Table 3, the impregnated MPP particles had the highest selectivity of the investigated resins for MTBE when humic acid was present. The selectivity values of AquaSorb 202 and Ambersorb XE-348F were approximately 10 and 40 times lower, respectively, than the selectivity of impregnated MPP particles. Thus, it can be expected that the MTBE capacity of Ambersorb XE-348F and AquaSorb 202 will decrease significantly when considerable amounts of humic acid are present due to the competitive adsorption of humic acid. This means that the SIR utilization efficiency is higher than compared to the evaluated adsorbents. This is a significant advantage of the prepared SIR over conventional adsorbents. 3.4. Kinetic concentration profiles The determination of the initial kinetic MTBE concentration profiles over time using a zero-length-column setup [17,23,24] showed that the initial MTBE uptake of Ambersorb XE-348F was very fast and took less than 5 min, see Fig. 4. After this initial uptake, the concentration of MTBE decreased further, albeit at a significantly lower rate. In the performed experiment, even after more than 2 days the equilibrium MTBE concentration had not been reached. This is in agreement with previous observations [21]. The initial extraction rate of Ambersorb XE-348F was fast compared to activated carbon and impregnated MPP particles. However, the particles of Ambersorb XE-348F (0.25–0.5 mm) were significantly smaller than the particles of activated carbon (0.425– 1.7 mm) and impregnated MPP (0.8–1.18 mm). The particle size influences the extraction rate. A decrease of the particle diameter can increase extraction rate [17,28]. The effect of the particle size on the overall extraction rate of impregnated MPP particles was investigated with particle fractions of 0.5–0.63 and 1.0–1.6 mm. In Fig. 4 it can be seen that a decrease of the MPP particle size increased the MTBE extraction rate, which was also concluded for the sorption of pentanal on impregnated XAD16 by Babic et al. [17]. As can be seen, a reduction of the SIR particle size from 1.0–1.6 mm to
Fig. 4. Concentration profiles of MTBE extraction kinetics of AquaSorb 202 (particle 1 or 0.04 mol L1, diameter 0.425–1.7 mm), impregnated MPP (caq MTBE;0 = 3.5 g L 1 1 = 676 g L or 3.08 mol L ) (particle diameter 0.8–1.18 mm) and Ambersorb corg E;0 XE-348F (particle diameter 0.25–0.5 mm) determined with a zero-length-column setup.
0.5–0.63 mm decreased the time until the equilibrium MTBE concentration was reached from more than 60 min down to 30 min. This can be attributed to the faster diffusion in smaller particles, as the diffusion time constant is inversely proportional to the square of the particle radius. However, if the MTBE removal with the prepared SIR would entirely depend on diffusion, the extraction rate should increase by a factor of 4 instead of a factor of 2, as observed in this case. Possibly, the MTBE removal depends on a combination of the diffusion of the solute, the extractant and the formed complex inside the pores, and on the other hand the chemical complexation between MTBE and the extractant [17,29]. Nonetheless, diffusion inside the SIR particle can be considered as one of the limiting factors of the MTBE extraction in the investigated system and the extraction rate of MTBE by the impregnated MPP particles can be enhanced by decreasing the particle size. Although the equilibrium MTBE capacity of AquaSorb 202 was higher than the equilibrium MTBE capacity of impregnated MPP particles, the MTBE extraction rate of AquaSorb 202 was significantly lower compared to impregnated MPP particles with a particle size of 0.5–0.63 mm. 3.5. Regeneration The regeneration of impregnated MPP particles was achieved with a hot nitrogen gas stream (120 °C). In an integrated process, off-gas or steam might also be used. Fig. 5 depicts the concentration profile of only MTBE during thermal regeneration. The initial time lag of the concentration profile during regeneration can be ex-
B. Burghoff et al. / Reactive & Functional Polymers 70 (2010) 41–47
47
Acknowledgements This work is financially supported by the Technology Foundation STW. Additional support by Veolia MPP Systems, Akzo Nobel Chemicals, Diosynth, INEOS Phenol, Vitens and our colleagues ir. R. Cuypers, Prof. Dr. H. Zuilhof at the University of Wageningen and Prof. Dr. E.J.R. Sudhölter at Delft University is also acknowledged. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.reactfunctpolym.2009.10.002. Fig. 5. Profile of thermal regeneration of impregnated MPP particles (flow rate 1 or 1 mol L1). N2 = 17 mL min1, T = 120 °C, caq MTBE;0 = 88.15 g L
plained by the time that is needed to completely heat up the whole particle bed. The initial shoulder of the profile was probably caused by the fast evaporation of MTBE still left on the surface of the wet particles after MTBE extraction. After this, the MTBE contained by the extractant inside the pores was released. The time until the complete amount of MTBE contained in the SIR particles had evaporated took approximately 20 min. In order to prevent diluent loss during regeneration, the diluent propylbenzene needs to be optimized further in terms of volatility. This can be achieved by the addition of alkyl chains or by modification of the existing propyl chain. Increasing the alkyl chain length and increasing the degree of chain branching can decrease the vapor pressure of the diluent. This is, however, not incorporated into this study and will be investigated further.
4. Conclusions In this study the use of a solvent impregnated resin (SIR) to remove MTBE from water is investigated. Of the investigated adsorbents, which serve as solid matrix for a SIR, the carbonaceous resin Ambersorb XE-348F has the highest MTBE capacity. However, the MTBE capacity of impregnated Ambersorb XE-348F significantly decreases compared to the unimpregnated particles. The impregnation of MPP particles increases the MTBE capacity by a factor of more than 3 compared to the unimpregnated MPP. This relative increase of MTBE capacity is more pronounced than in the case of impregnated Amberlite XAD7. Although Ambersorb XE-348F has a higher MTBE capacity than the prepared impregnated MPP particles, the selectivity of the impregnated MPP particles for MTBE in the presence of humic acid is 10–40 times higher compared to Ambersorb XE-348F and the activated carbon AquaSorb 202. Exploratory kinetics measurements in a zero-length-column setup show that the initial extraction rate of Ambersorb XE-348F is faster than of impregnated MPP. However, the particles of Ambersorb XE-348F are smaller than the MPP particles. It can be shown that a decrease of the MPP particle size increases the extraction rate of the used SIR. The prepared SIR particles can be successfully regenerated with a hot nitrogen gas stream (120 °C).
References [1] A.A.M. Langenhoff, Netherlands Centre for Soil Quality Management and Knowledge Transfer (SKB SV-018), 2000. [2] R. Schuhmacher, M. Führer, W. Kandler, C. Stadlmann, R. Krska, Anal. Bioanal. Chem. 377 (2003) 1140–1147. [3] W.E. May, S.P. Wasik, M.M. Miller, Y.B. Tewari, J.M. Brown-Thomas, R.N. Goldberg, J. Chem. Eng. Data 28 (1983) 197–200. [4] A.A. Keller, Health and Environment Assessment of MTBE, Report to Legislature of the State of California, vol. III, 1998. [5] U. Peters, F. Nierlich, E. Schulte-Körne, M. Sakuth, R. Deeb, M. Laugier, M. Suominen, M. Kavanaugh, Methyl tert-butyl ether, in: Ullmann’s Encyclopedia of Industrial Chemistry, John Wiley and Sons, Inc., Weinheim, 2003. [6] G. Melin, California MTBE Research Partnership: Treatment Technologies for Removal of Methyl tert-Butyl Ether from Drinking Water, National Water Research Institute, 2000. [7] B.A. Ray, A. Selvakumar, N.A. Tafuri, Treatment of methyl tert-butyl ether (MTBE) contaminated waters with Fenton’s reagent, Report to the US Environmental Protection Agency, USA, 2003, pp. 1–18. [8] T.C. Shih, M. Wangpaichitr, M. Suffet, Water Res. 37 (2003) 375–385. [9] B. Burghoff, J. Schiferli, J. Sousa Marques, A.B. de Haan, Chem. Eng. Sci. 64 (2008) 2887–2892. [10] M. Traving, H.-J. Bart, Chem. Eng. Technol. 25 (2002) 997–1003. [11] B. Burghoff, E.L.V. Goetheer, A.B. de Haan, React. Funct. Polym. 68 (2008) 1314–1324. [12] K. Babic´, A.G.J. van der Ham, A.B. de Haan, React. Funct. Polym. 66 (2006) 1494–1505. [13] J. Serarols, J. Poch, I. Villaescusa, React. Funct. Polym. 48 (2001) 37–51. [14] A. Warshawsky, J.L. Cortina, M. Aguilar, K. Jerabek, in: International Solvent Extraction Conference, Barcelona, Spain, 1999, pp. 1267–1272. [15] Y. Wang, C. Wang, A. Warshawsky, B. Berkowitz, Sep. Sci. Technol. 38 (2003) 149–163. [16] MPPSystems, Macro Porous Polymer Extraction System – Water Purification, Akzo Nobel, 2005. [17] K. Babic´, A.G.J. van der Ham, A.B. de Haan, Adsorption 14 (2008) 357–366. [18] R.-S. Juang, H.-L. Chang, Ind. Eng. Chem. Res. 34 (1995) 1294–1301. [19] A. Kostova, H.-J. Bart, Chem. Ing. Tech. 76 (2004) 1743–1748. [20] H. Kitazaki, M. Ishimaru, K. Inoue, K. Yoshida, S. Nakamura, in: International Solvent Extraction Conference, Melbourne, Australia, 1996, pp. 1667–1672. [21] H.-W. Hung, T.-F. Lin, J. Hazard. Mater. B135 (2006) 210–217. [22] F.J. Stevenson, Humus Chemistry: Genesis, Composition, Reactions, John Wiley and Sons, New York, 1994. [23] M. Eic, D.M. Ruthven, Zeolites 8 (1988) 40–45. [24] T. Djekic, V. Tripkovic, A.G.J. van der Ham, A.B. de Haan, Adsorption 13 (2007) 231–237. [25] T.C. Schmidt, H.-A. Duong, M. Berg, B.S. Haderlein, Analyst 126 (2001) 405– 413. [26] E. Arevalo, M. Rendueles, A. Fernandez, M. Diaz, Sep. Purif. Technol. 18 (2000) 217–225. [27] R.A. Rodriguez, J.M. Davidson, D.N. Creek, A.E. Flores, A.J. Stocking, M.C. Kavanaugh, H. Eylers, M. Wooden, T. Herson, Center for Groundwater Restoration and Protection, National Water Research Institute, Fountain Valley, 1999. [28] J. Serarols, J. Poch, I. Villaescusa, React. Funct. Polym. 48 (2001) 53–63. [29] B. Burghoff, E. Zondervan, A.B. de Haan, React. Funct. Polym. 69 (2008) 264– 271.