Journal of Environmental Management 92 (2011) 472e479 Contents lists available at ScienceDirect Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman Magnetic composite prepared from palm shell-based carbon and application for recovery of residual oil from POME Worawan Ngarmkam a, Chitnarong Sirisathitkul c, Chantaraporn Phalakornkule a, b, * a Department of Chemical Engineering, Faculty of Engineering, King Mongkut’s University of Technology North Bangkok, Bangkok 10800, Thailand The Research and Technology Center for Renewable Products and Energy, King Mongkut’s University of Technology North Bangkok, Bangkok 10800, Thailand c The Institute of Science, Walailak University, Nakhon Si Thammarat 80161, Thailand b a r t i c l e i n f o a b s t r a c t Article history: Received 18 June 2010 Received in revised form 28 August 2010 Accepted 30 August 2010 Available online 6 October 2010 Magnetic separation combined with adsorption by activated carbon has been found to be a useful method for removing pollutants. In this paper, the use of palm shell as a source of activated carbon for the removal and recovery of oil from palm oil mill effluent (POME) is studied. In the first part of the study, the properties of samples of activated carbon prepared from palm shell under a variety of different conditions were characterized for their hydrophobicity, surface areas and pore size distribution. The most effective of the activated carbon samples was prepared by impregnation with ZnCl2 followed by combined physical/chemical activation under carbon dioxide flow at 800  C. Four grams of these samples adsorbed 90% of the oil from 50 mL POME. In the second part, the palm shell-based carbon samples were given magnetic properties by the technique of iron oxide deposition. Ninety-four percent of the activated carbon/iron oxide composite containing the adsorbed oil could be extracted from the POME by a magnetic bar of 0.15 T. Four grams of the composite can remove 85% of oil from 50 mL POME and a total of 67% of the initial oil can then be recovered by hexane extraction. Powder X-ray diffractometry showed the presence of magnetite and maghemite in the activated carbon/iron oxide composite. Ó 2010 Elsevier Ltd. All rights reserved. Keywords: Activated carbon Adsorption Magnetic Iron oxide POME Oil 1. Introduction Activated carbon can be combined with a magnetic material to provide an effective magnetic extraction method for first adsorbing contaminants from aqueous effluents and then separation by a simple magnetic process (Booker et al., 1991). Application of magnetic particle technology for remediating environmental problems has received attention in recent years. For example, Oliveira et al. (2002) prepared composites of activated carbon and iron oxide and employed them to remove chloroform, phenol, chlorobenzene and drimaren red from aqueous solutions. The magnetic particles were subsequently removed from the medium by a simple magnetic procedure. Castro et al. (2009) prepared composites of activated carbon and iron oxide and employed them to remove atrazine from aqueous solution. Activated carbon is one of the substances that have been proved to be fruitful for tackling environmental problems. In recent years, interest has been growing in the use of low-cost and abundantly * Corresponding author. The Research and Technology Center for Renewable Products and Energy, King Mongkut’s University of Technology, North Bangkok, Bangkok 10800, Thailand. E-mail addresses: cphalak21@yahoo.com, cpk@kmutnb.ac.th (C. Phalakornkule). 0301-4797/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2010.08.031 available lignocellulosic materials as precursors for the preparation of activated carbon. In Thailand, one of these low-cost lignocellulosic materials is palm shell, which is abundantly available from the palm oil processing mills. It has been used as the starting material for preparing activated carbon for various applications. In addition to palm shell, a palm oil milling process generates significant amounts of oily wastewater. Palm oil mill effluent (POME) is a highly polluting wastewater with typical properties: biochemical oxygen demand of 25,000 mg L 1, chemical oxygen demand of 50,000 mg L 1, suspended solids of 20,000 mg L 1, total solids of 40,000 mg L 1 and oil and grease of 8000 mg L 1 (Ma, 1995). The oil has to be removed from the wastewater to avoid problems in the subsequent water treatment units and biological treatment stages. Removal of oil from POME before anaerobic biological treatment is desirable as the oil reduces the efficiency of the anaerobic treatment. In addition, the recovered oil will be available for sale as low-grade oil. Previous studies have shown the potential of oil removal by activated carbon (Ahmad et al., 2005), of preparation of activated carbon from palm shell (Daud et al., 2000; Guo and Lua, 2002a, 2002b, 2002c, 2003; Tan and Ani, 2004; Daud and Ali, 2004; Guo et al., 2005, 2007; Adinata et al., 2007; Sumathi et al., 2009), and of preparation of activated carbon/iron oxide composites (Oliveira W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 et al., 2002; Castro et al., 2009). In this work, we integrated the three techniques for the preparation of palm shell-based magnetic composites for oil recovery. Various conditions of carbonization and activation were then applied to prepare the palm shell-based carbons, and the samples were characterized for their hydrophobicity, surface areas and pore size distribution. The application for oil removal from POME was then shown as an example. The palm shell-based carbons with acceptable oil-removing capacity were further modified to be magnetic composites. The carbon/iron oxide composites were employed to remove oils from POME, and the oils were further recovered by hexane extraction. 2. Materials and methodology 2.1. Materials Palm shells were provided by Suksomboon Co., Ltd (Thailand). The shells were cleaned and dried for several days under the sun. Then they were crushed and sieved and were categorized according to their size and form: (1) Granular palm shell (GP) with the size between 1.0 and 2.0 mm; and (2) Powder palm shell (PP) with the size less than 355 mm. Finally, the GP and PP products were dried at 110  C for about 24 h to remove residual moisture. The results of a proximate analysis of the dried palm shell are shown in Table 1. The chemicals used in this study were of analytical grade: ZnCl2 (Ajax Chemicals, Australia), FeCl3 (Ajax Chemicals, Australia), FeSO4 (Ajax Chemicals, Australia), n-hexane (Ajax Chemicals, Australia), nitrogen gas with 99.9% purity, and carbon dioxide gas with 99.9% (Thonburi wattana Ltd., Thailand). Commercial activated carbon was also used in this study for comparison purposes. This was obtained from C. Gigantic Carbon Co., Ltd. The activated carbon (commercial name CGC 200 PÒ) was produced from palm shell, was in powder form and had the following properties: moisture content 4.2%, ash content 6.1%, pH value 10.1, bulk density 0.63 g cm 3 and iodine adsorption 862 mg g 1. POME was collected from Suksomboon Co., Ltd. The discharged temperature was between 70 and 80  C. POME was filtered through two-layer cheesecloth to remove dirt, plant cell debris, fibers and other solid particles with size in the order of millimeters. Portions of this filtered POME were withdrawn and were analyzed. The initial oil content was found to be 8100 mg L 1, and total solid (TS) content was 48,600 mg L 1, chemical oxygen demand (COD) 69,250 mg L 1 and pH 4.5. The POME samples were stored at 4  C prior to use. 2.2. Preparation of the palm shell-based carbons 473 For carbonization, the GP was heated from room temperature to a desired temperature of either 400  C, 600  C or 800  C under N2 flow of 300 mL min 1 at the heating rate of 10  C min 1. The hold time was 2 h. After being cooled to room temperature, the GP was crushed and sieved to size less than 355 mm. For physical activation, the GP was first carbonized at 400  C, 600  C or 800  C under N2 flow of 300 mL min 1 at the heating rate of 10  C min 1. The carbonization was then followed by switching the gas flow to CO2 at 300 mL min 1 to activate the carbon. The activation was continued for 2 h at the carbonization temperature. After being cooled to room temperature, the samples were crushed and sieved to size less than 355 mm. For the combined physical and chemical activation, the GP was first impregnated with ZnCl2 solution. The weight ratio of ZnCl2 to GP was 2:1. The mixture was dried overnight at 110  C. The sample was then carbonized under nitrogen atmosphere at either 400  C, 600  C, 800  C or 900  C for 1 h. The carbonization was followed by switching the gas flow to CO2 at 300 mL min 1 to activate the carbon. The activation was continued for a period of 2e4 h. The activated products were then cooled to room temperature and washed with deionized water, 0.1 M HCl solution and hot deionized water to remove the remaining chemicals. The samples were dried overnight at 110  C. Then they were crushed and sieved to size less than 355 mm. The preparation methods of the samples and their abbreviated names are summarized in Table 2. The yield (wt%) was calculated according to equation (1): Yield ¼ Wt  100% Wo (1) where Wo and Wt is the initial mass of the sample (g) and the mass of the sample (g) after the thermal treatment, respectively. 2.3. Preparation of the carbon/iron oxide composites The method of preparation followed that of Oliveira et al. (2002) and Castro et al. (2009). The composites were prepared from a suspension of activated carbon in a 400 mL solution of FeCl3 (19.5 g L 1) and FeSO4 (9.75 g L 1) at 70  C for 24 h. NaOH solution (100 mL, 5 M) was added dropwise to precipitate the iron oxides. The amount of activated carbon was adjusted in order to obtain activated carbon/iron oxide weight ratios of 1:1 (MC 1:1 for the abbreviated name), 3:1 (MC 3:1) and 6:1 (MC 6:1). The obtained materials were dried in an oven at 100  C for 3 h. Then they were crushed and sieved to size less than 355 mm. 2.4. Oil removal and recovery experiments The palm shell-based carbons were prepared from the GP either by carbonization alone or by carbonization combined with physical activation or physical/chemical activation. Table 1 Property of palm shell. Property Value Volatile matter (wt%) Fixed carbon (wt%) Ash (wt%) Moisture (wt%) Element (wt%) C H N Others Density (g cm 3) 67.87 24.54 1.92 5.67 50.01 6.85 1.90 41.24 1.46 The amount of each adsorbent was varied between 0.5 and 4 g in treating 50 mL POME. A series of batch experiments was performed at a mixing rate of 150 rpm at room temperature for 24 h. Two methods were used for the solid/liquid separation. In the first method, a filtering method with cellulose filter paper (Whatman No. 40, United Kingdom) was used to separate non-magnetic adsorbents from liquid phase. In the second method, two magnetic bars of 0.15 T were used to remove magnetized adsorbents. Suspensions were allowed to flow from top to bottom in a burette with the magnetic bars attached to its sides. The non-magnetic suspension was collected at the bottom. The magnetized adsorbents were collected after removing the magnetic bars. The residual oil content in the filtrate was conducted according to ASTM D4281-95. The oil content in the collected solids was measured using the oil and grease method recommended by a SoxtecÔ 2043 (Foss, Denmark) with n-hexane as the oil-extraction 474 W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 Table 2 Designated names and corresponding preparation conditions of palm shell-based carbons. Samplea PP PC 400-2 PC 600-2 PC 800-2 Phy 400-2 Phy 600-2 Phy 800-2 Chem 400-2 Chem 600-2 Chem 800-2 Chem 800-3 Chem 800-4 Chem 900-2 Chem 900-3 Chem 900-4 CGC 200 PÒ a a Impregnation Carbonization Activation Temp. ( C) Time (h) ZnCl2 (w/w) Temp. ( C) Time (h) Gas atmosphere Temp. ( C) Time (h) Gas atmosphere e e e e e e e 400 600 800 800 800 900 900 900 e e e e e e e e 24 24 24 24 24 24 24 24 e e e e e e e e 2:1 2:1 2:1 2:1 2:1 2:1 2:1 2:1 e e 400 600 800 400 600 800 400 600 800 800 800 800 800 800 850 e 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 e N2 N2 N2 N2 N2 N2 N2 N2 N2 N2 N2 N2 N2 N2 N2 e e e e 400 600 800 400 600 800 800 800 900 900 900 850 e e e e 2 2 2 2 2 2 3 4 2 3 4 2 e e e e CO2 CO2 CO2 CO2 CO2 CO2 CO2 CO2 CO2 CO2 CO2 H2O b, Process temperatures ( C) hold time (h). solvent. First, the samples to be analyzed were dried at 103  C for 2 h and were transferred into thimbles. The sample-containing thimbles were placed in an extraction unit connected to a chamber containing 50 mL of hexane. The three-step extraction procedure consisted of boiling at 130  C for 25 min, rinsing for 45 min and solvent recovery for 10 min. The oil content was calculated from the ratio of the amount of the extracted oil to the sample weight. The percent oil removal or ‘the oil removal capacity’ was determined from the equation (2): Oil removal capacity ð%Þ ¼ 100    Co C Co (2) where Co and C are the initial oil concentration in POME and the oil concentration in the filtrate, respectively. The percent oil recovery was determined from the equation (3): Oil recovery ð%Þ ¼ 100  moil moil;o ! (3) where moil,o and moil are the amount of oil (g) in the initial POME and the amount of oil (g) extracted from the collected solids by hexane extraction, respectively. The percent recovery and residues of the activated carbon/iron oxide composites from suspension were determined from the equations (4) and (5), respectively: Magnetic composite recovery ð%Þ ¼ 100  M Mo   Magnetic composite residues ð%Þ ¼ 100  1  M Mo (4)  (5) where Mo and M are the initial mass of the magnetic composites (g) in suspension and the mass of magnetic solids (g) collected by the magnetic process, respectively. 2.5. Characterization and chemical analysis Analyses of POME characteristics including TS and COD were conducted according to the Standard Methods (American Public Health Association, 1998). Oil content was conducted according to ASTM D4281-95 with the SoxtecÔ 2043. The size distribution of colloids in POME was measured by a nanosizer (Malvern Instruments, NanoSX, USA). The components of palm shell were obtained from proximate analyses (JIS M 8812-1984) and elemental composition (CHNS/ O Analyzer PE2400 Series II, PerkineElmer, USA). Bulk density and ash content were determined according to ASTM D2854-09 and ASTM D2866-94 (2004), respectively. Surface area, pore size and pore volume were measured by a surface area analyzer (BELSORP-mini II, Japan). The water contact angles on lawns of carbon powders were measured using Contact Angle System OCA. (OCA 15plus, DataPhysics Instruments GmbH, Germany). Phases of iron oxide were examined by powder thetaetheta rotating anode X-Ray diffractometer (Rigaku, TTRAXIII, Japan). Images of the adsorbent surfaces were investigated by Scanning Electron Microscope (SEM, LEO 1455 VP, United Kingdom). Magnetic measurement was performed based on magnetic induction followed by magnetic flux measurement. Activated carbon/iron oxide composites were filled in standard 500 mg-drug capsules. The capsules were then subjected to 60 G 50 Hz magnetic field supplied by a Helmholz coil (Lakeshore FH-2.5, USA). Magnetic flux density induced in each sample was detected by a pick-up coil wound around the capsule and connected to a flux meter (Lakeshore 480, USA). 3. Results and discussion 3.1. Oil removal and oil recovery efficiencies using native palm shell and palm shell-based carbons It was found from the nanosizer that most colloidal particles in the POME samples including oil droplets had sizes in the range 1e2 mm or 4e6 mm. The yields, surface characteristics and oil removal and recovery efficiencies obtained using the native palm shell in powder form (PP), the various palm shell-based carbons derived from the granular form and the commercial form (CGC 200 PÒ) are shown in Table 3. PP had a limiting oil sorption capacity. Four grams of PP can remove 32% of oil from 50 mL POME. This is equivalent to 33 mg oil per g adsorbent. The PP had little porosity and surface area. However, its surface was hydrophobic, and it was this surface hydrophobicity that was responsible for its oilremoving capacity. Carbonization was found to enhance oil sorption and the oilremoving capacity was found to increase with increasing carbonization temperature. When the native palm shell was carbonized at 400  C for 2 h, the surface hydrophobicity was slightly lower than that of the PP. However, PC 400-2 had higher oil sorption capacity 475 W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 Table 3 Yields, surface characteristics and oil removing and recovery capacities of the palm shell-based carbons. Adsorbent Yield (%) PP PC 400-2 PC 600-2 PC 800-2 Phy 400-2 Phy 600-2 Phy 800-2 Chem 400-2 Chem 600-2 Chem 800-2 Chem 800-3 Chem 800-4 Chem 900-2 Chem 900-3 Chem 900-4 CGC 200 PÒ e 42 35 29 41 33 29 37 34 33 30 26 29 26 24 NA a b c d Surface characteristicsa /contact angle SBET (m2 g Hydrophobic/130 Hydrophobic/113 Hydrophilic/<90 Hydrophilic/<90 Hydrophobic/95 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 Hydrophilic/<90 1.5d 2.7d 127d 365 41d 213d 511 368 414 1041 1334 1066 1017 1053 1080 906 1 ) % Mesopore volume % Oil removalb % Oil recoveryb NAc NAc 12 18 NAc 5 8 14 18 22 35 22 19 17 16 26 32 54 57 62 57 66 74 63 67 86 90 88 84 86 88 84 30 51 55 62 54 66 69 60 63 70 74 74 70 71 71 69                 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1                 0.5 1 1 1 1 1 0.5 0.5 1 1 1 1 1 1 1 0.5 Hydrophobic: water contact angle >90 , Hydrophilic: water contact angle <90 . The value was based on the use of 4 g of adsorbent in 50 mL of POME. NA: Not available. Incomplete adsorption isotherm. than PP because a degree of macropore development due to carbonization added to the capacity associated with the surface hydrophobicity. However, as the carbonization temperatures were increased further to 600  C and 800  C, the surface hydrophobicity was greatly reduced and the surfaces became hydrophilic. The increase in the oil-removing capacity of PC 600-2 and PC 800-2 was then mainly due to significant pore development indicated by increasing surface areas. Physical activation increased the oil-removing capacity of the palm shell at all three temperatures used, 400  C, 600  C and 800  C, with the increase in capacity rising from 3% at 400  C to 12% at 800  C. As in the case of the increased oil-removing capacity of carbonized samples compared with PP samples, pore development was used to explain the 12% increase in oil-removing capacity of the activated Phy 800-2 compared with the carbonized only PC 800-2. The surface area of Phy 800-2 was 511 m2 g 1 compared with 365 m2 g 1 of PC 800-2. However, even though the activation appreciably increased the surface area, the mesopore volume was not changed appreciably. This suggests that the increased surface area was due to micropore development. ZnCl2 activation has been known to provide high mesoporosity and to increase pore width due to large external localized decomposition (Caturla et al., 1991; Hu and Srinivasan, 2001). The combined physical and chemical activation helped to increase the oil-removing capacity. The capacity increased to greater than 85% for physical and chemical activation temperature of 800  C and 900  C. The surface area of Chem 800-2 increased to 1041 m2 g 1 and the oil-removing capacity increased 24% relative to PC 800-2. As in the case of carbonization and physical activation, pore development was used to explain the increased oil-removing capacity due to chemical activation. The chemically activated Chem 800-2 had a surface area of 1041 m2 g 1 and a removal efficiency of 86% compared with 511 m2 g 1 and 74% for the physically activated Phy 800-2. The development of both micropore and mesopore by chemical activation was significant (around 20% mesopore volume). The chemical activation temperature of 900  C increased neither the surface area nor the oil-removing capacity, and the chemical activation of 800  C was concluded to be the optimum temperature for the combined physical and chemical activation. It has been reported in the literature that the primary devolatilization of the low-molecular-weight volatiles originally present in native palm shell occurred between 400 and 700  C, and the secondary devolatilization of the high-molecular-weight volatiles occurred between 700 and 900  C. In the low temperature regime, increasing the activation temperature from 400  C to 800  C increased the release of low-molecular-weight volatiles from the matrix structure, resulting in increasing rudimentary pores. However, when the activation temperature reached 900  C, softening and sintering of the high-molecular-weight volatiles occurred, resulting in the depolymerization of the melt and shrinkage of the total volume of the sample (Lua et al., 2006). In order to investigate the effect of activation time, the activation time was varied between 2, 3 and 4 h. The activation time was found to be optimal at 3 h for the activation temperature of 800  C. The surface area at this condition was the highest, reaching 1334 m2 g 1, and the percent mesopore volume was 35%. The oil-removing capacity was also highest at 90%, equivalent to 90 mg oil per g adsorbent. It was observed that the increase in the oil-removing capacity was not in the same proportion as the increase in the surface area. The data showed that the oil-removing capacity could not be further increased by increasing the surface area. A possible explanation was that the development of inner micropores was responsible for the increased surface area but the oil could not reach the innermost micropores. Increasing the activation time to 4 h decreased both the surface area and the oil-removing capacity. The adsorbed oil on the carbons can be recovered by n-hexane extraction. When the surface area of an adsorbent was lower than 500 m2 g 1, the percent oil removal and the percent oil recovery were close in value with less than 5% difference. The similar values indicated that most of the adsorbed oil can be extracted out by n-hexane. However, when the surface area exceeded 1000 m2 g 1, the percent oil removal was approximately 14e17% greater than the percent of oil recovery. An explanation was that it was difficult for the oil adsorbed in the inner micropore to transfer out through the complex textural structure of the activated carbons. Another possible explanation was that it was difficult for hexane to penetrate the micropores and to dissolve the oil adsorbed in them. Among the palm shell-based carbons, Chem 800-3 was the best in terms of oil removal and oil recovery. Its performance in recovering oil was 5% higher than that of the commercial activated carbon, CGC 200 PÒ. Therefore, Chem 800-3 was selected for the study of the preparation of palm shell-based carbon/iron oxide composite. It should be mentioned that the oil sorption capacity of native palm shell and their activated carbons was relatively very low (30e90 mg g 1) compared to other natural adsorbents such as wood, kenaf, cotton, kapok and milkweed. Wood and kenaf have 476 W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 a sorption capacity of 4000e8000 mg g 1, whereas cotton, kapok and milkweed have a sorption capacity of 30,000e40,000 mg g 1 (Deschamps et al., 2003). However, palm shell-based carbon has the unique advantage of being a local resource for oil recovery that is available at no cost as it is a by-product of the palm oil production process. therefore rather insensitive to excitation by alternating current magnetic field. Fig. 1 presents the X-ray data (XRD) of the precursor, iron oxide and the composites MC 1:1, MC 3:1 and MC 6:1. No phases of iron 3.2. Characterization of the palm shell-based carbon/iron oxide composites The data shown in Table 4 indicates that the densities of the composites were significantly affected by the presence of iron oxide. The apparent density of the activated carbon Chem 800-3 was about half that of the native palm shell (r ¼ 1.46). The decrease in the apparent density was due to the burn-off of the carbon, accompanied by porosity generation. As the density of the activated carbon was less than that of water, the adsorbent became suspended in the water and was not effective for solid/liquid separation. The Chem 800-3/iron oxide composites had higher densities than the precursor activated carbon because of the presence of iron oxide (r ¼ 4.21). As expected, the density decreased with increasing activated carbon/iron oxide weight ratios used in the preparation of the composites. The densities of MC 1:1, 3:1 and 6:1 were greater than 1, making them settle down in water and therefore easy to separate from the water by gravitational methods. Besides density, the BET surface areas of the composites were significantly affected by the presence of iron oxide. As expected, the surface area increased with increasing activated carbon/iron oxide weight ratios used in the preparation of the composites. However, the Chem 800-3/iron oxide composites had significantly lower BET surface areas than the precursor activated carbon because of the presence of iron oxide (SBET ¼ 116 m2 g 1). It should be noted that the remarkable drop of surface area could not be explained merely by the presence of iron oxide on the carbon surface. The calculation of the composite surface area by the weight factor of the carbon and iron oxide yielded a surface area significantly greater than the measured surface area. Therefore, the result implied that iron oxide was also present in the pores of the activated carbon and might block some pores. The remarkable drop of surface area was a drawback of the iron oxide entering the pores, but a benefit was that the iron oxide did not separate from the activated carbon in the presence of the magnetic field. The magnetic flux density induced in the Chem 800-3/iron oxide composites by 60 G 50 Hz magnetic field is shown in Table 4. The value is predictably increased with increasing fraction of iron oxide, but the magnetization was not linearly dependent on the activated carbon/iron oxide weight ratios. In addition, it is noted that the flux density in a capsule with iron oxide is lower than those with composites. This can be explained by the nature of bulk iron oxide whose coercivity is of order of hundreds oersted. It is Table 4 Characteristics of activated carbon/iron oxide magnetic composites. Sample Fe2O3 MC 1:1 MC 3:1 MC 6:1 Chem 800-3 Amount of Amount of Apparent SBET product per (m2 g product per density total reactants amount of (g cm 3) Chem 800-3 0.26 0.48 0.63 0.71 e e 2.33 1.44 1.17 e 4.21 3.85 3.31 2.76 0.72 116 378 427 544 1334 Magnetization ) (G) 1 0.9a 10.3 6.4 5.1 0.1 a The nature of bulk iron oxide whose coercivity is of order of hundreds oersted is relatively insensitive to the excitation by alternating current magnetic field. Fig. 1. XRD of Chem 800-3, iron oxide, MC 1:1, MC 3:1 and MC 6:1. W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 oxide were observed for the precursor Chem 800-3, while all phases of iron oxide, i.e., maghemite, magnetite and goethite, can be observed for the iron oxide. Compared with the XRD of iron oxide, the XRD of MC 1:1 had a lower goethite phase. Similarly, when compared with the XRD of MC 1:1, the XRD of MC 3:1 and MC 6:1 had a lower fraction of goethite. Therefore, it was concluded that the phase of iron oxide depended on the activated carbon/iron oxide weight ratios used in the preparation of the composites. The fraction of goethite seemed to decrease with increasing activated carbon content. Maghemite and magnetite have been known to have stronger magnetization than goethite (Oliveira et al., 2002; Castro et al., 2009). Therefore, this was another benefit of increasing activated carbon content in the preparation of the composite. The morphologies investigated by SEM of the native palm shell, the precursor Chem 800-3 and the Chem 800-3/iron oxide composite are shown in Fig. 2a, b and c, respectively. The native palm shell had macroporous structure with an average pore size of 1.2 mm. The precursor Chem 800-3 had a cracked surface morphology due to the heat treatment and the average pore size on 477 the surface decreased to 0.8 mm. In Fig. 2c, iron oxide can be distinguished from the activated carbon because the former appeared brighter than the supporting surface, which was the activated carbon. The surface of the Chem 800-3/iron oxide composite was smoother than that of its precursor due to the covering of the surface by iron oxide. In addition, small aggregates of iron oxides were observed on the surface and some can be seen in the pores. 3.3. Utilization of the palm shell-based carbon/iron oxide composite for oil removal and recovery from POME The efficiency for the composite recovery of this magnetic process was examined. As shown in Fig. 3, MC 1:1 and MC 3:1 can be recovered from both water and POME with more or less the same degree as pure Fe2O3. However, when the ratio of activated carbon to iron oxide increased to 6:1, the efficiency for the composite recovery dropped approximately 8%. The drop in the solid recovery efficiency corresponded to the drop in the degree of magnetization. The drop in the solid recovery was expected to be Fig. 2. SEM micrographs ×5000 (left) and ×1000 (right) of (a) palm shell, (b) the activated carbon Chem 800-3 and (c) the activated carbon/iron oxide composite MC 6:1. 478 W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 % Recovery or Residual 100 80 60 40 20 0 Pure Fe2O3 MC 1:1 MC 3:1 MC 6:1 Type of Adsorbent %Adsorbent recovery in water % Adsorbent recovery in POME % Residual adsorbent in water % Residual adsorbent in POME Fig. 3. Percent recovery and residues of Chem 800-3/iron oxide composites separated by the magnetic process with two magnetic bars of 0.15 Tesla from water and from POME. solved by increasing the power of the magnetic bars used for the magnetic force induction. A recovery of MC 6:1 with two magnetic bars of 0.5 T rose to 95%. Fig. 4 compares the oil-removing capacity and percent oil recovery by n-hexane of the precursor Chem 800-3, the composites MC 1:1, MC 3:1 and MC 6:1, and pure Fe2O3. The oil-removing capacity increased with the increasing portion of activated carbon. This was expected because the activated portion was responsible for oil sorption. The oil removal capacity of MC 6:1 was only 5% lower than Chem 800-3. In addition, the difference between the percent oil removal and the percent oil recovery of MC 6:1 was more or less the same as that of Chem 800-3. Besides removing oil, the composites also led to the reduction of COD and TS in POME. It is noted that the COD removal increased in proportion with increasing activated carbon ratio. This was as expected because the activated carbon portion was responsible for adsorption. The magnetic composites could be separated from the liquid portion by the introduction of a 0.15 T magnetic bar. Due to their higher density than water, some portions of the composites settled down to the bottom. The increased density was another benefit of the composite. Otherwise, the precursor activated carbon would remain suspended in water due to its low density compared to water. 100 % Removal or Recovery % Oil Removal % TS Removal %Oil Recovery % COD Removal The oil droplets in POME typically can be found in two phases. They are suspended in the supernatant as emulsions and also float as oil droplets on the upper layer of the suspension. Several methods have been used to remove residual oil from wastewater, such as adsorption (Ahmad et al., 2005), coagulation and flocculation (Andrew et al., 2000; Cañizares et al., 2008), electrocoagulation (Yang, 2007; Un et al., 2009) and floatation (Zouboulis and Avranas, 2000). The major difficulty in treating disposed oily residues is that separation of oil droplets by flocculation combined with gravity settling is difficult and time consuming. A conventional method for breaking up oil-in-water emulsions is the addition of hydrolyzing metal salts, especially aluminum sulfate, to wastewater. However, the chemical coagulation method not only decreases the POME pH but also generates a secondary pollutant, namely sulfate, which is known to be an important inhibitor of anaerobic digestion (Chen et al., 2008). The sulfate can be reduced to sulfide by sulfate reducing bacteria (Koster et al., 1986; Hilton and Oleszkiewicz, 1988) which suppress methane production by competing with methanogens for common organic and inorganic substrates (Harada et al., 1994). In addition, the sulfide produced can be toxic to bacteria (Anderson et al., 1982; Oude Elferink et al., 1994; Colleran et al., 1998). For these reasons, the present study is an efficient alternative strategy for oil recovery from POME without sulfate generation. 4. Conclusion This study was an example of local resource utilization, that is, palm shell was utilized for oil removal and recovery from POME. Native palm shell, by its hydrophobic nature, had a limiting oil sorption capacity; i.e., 33 mg oil per g palm shell. Modification of native palm shell by carbonization, physical and chemical activation further improved the oil sorption capacity. Especially, the combined physical and chemical activation using ZnCl2 for impregnation under CO2 flow for 3 h increased the oil sorption capacity to approximately triple. Even though the oil sorption capacity of native palm shell and their activated carbons was relatively very low compared to other natural adsorbents such as wood, kenaf, cotton, kapok and milkweed, palm shell-based carbon had the unique advantage of being a local resource for oil recovery available at no cost. The activated carbon/iron oxide composites were successfully prepared and, due to their magnetic properties, can be separated from POME by a simple magnetic process. Using the composites for oil sorption and a magnetic process for their recovery, 80% of oil can be removed and 65% of the initial oil presented in POME can be recovered by hexane extraction. Other modifications of native palm shell in order to increase oil sorption capacity are being pursued. 80 Acknowledgements 60 We wish to acknowledge the research grant from the Thailand Research Fund and the Commission on Higher Education (RMU5080053) and the scholarship by TRF-Master Research Grants to Worawan Ngarmkam. Special thanks to Dr. Elvin Moore for his suggestions and critical readings of this manuscript. 40 20 0 Pure Fe2O3 MC 1:1 MC 3:1 MC 6:1 Chem800-3 Type of Adsorbent Fig. 4. Effect of the carbon/iron oxide ratio on the oil removal, oil recovery, COD removal and TS removal. Appendix. Supplementary data Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jenvman.2010.08.031. W. Ngarmkam et al. / Journal of Environmental Management 92 (2011) 472e479 References Adinata, D., Daud, W.M.A.W., Aroua, M.K., 2007. Preparation and characterization of activated carbon from palm shell by chemical activation with K2CO3. Bioresour. Technol. 98, 145e149. Ahmad, A.L., Sumathi, S., Hameed, B.H., 2005. Residual oil and suspended solid removal using natural adsorbents chitosan, bentonite and activated carbon: a comparative study. Chem. Eng. J. 108, 179e185. American Public Health Association, 1998. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, American Public Association (APHA), Washington, DC. American Society for Testing and Materials, 2004. Standard Test Method for Total Ash Content of Activated Carbon ASTM Designation D2866-94(2004). American Society for Testing and Materials, 2005. Standard Test Method for Oil and Grease (Fluorocarbon Extractable Substances) by Gravimetric Determination ASTM Designation D4281-95 (Reapproved 2005). American Society for Testing and Materials, 2009. Standard Test Method for Apparent Density of Activated Carbon ASTM Designation D2854-09. Anderson, G.K., Donnelly, T., MACkeown, K.J., 1982. Identification and control of inhibition in the anaerobic treatment of industrial wastewater. Process Biochem. 17, 28e32. Andrew, K., Graeme, G., Jeff, G., Brain, R.S., 2000. Flocculation and coalescence of oilin-water poly(dimethylsiloxane) and emulsion. Colloid Interface Sci. Ideal Library 227, 390e397. Booker, N.A., Keir, D., Priestley, A., Ritchie, C.D., Sudarmana, D.L., Woods, M.A., 1991. Sewage clarification with magnetite particles. Water Sci. Technol. 23, 1703e1712. Cañizares, P., Martínez, F., Jiménez, C., Sáez, C., Rodrigo, M.A., 2008. Coagulation and electrocoagulation of oil-in-water emulsions. J. Hazard. Mater. 151, 44e51. Castro, C.S., Guerreiro, M.C., Gonçalves, M., Oliveira, L.C.A., Anastácio, A.S., 2009. Activated carbon/iron oxide composites for the removal of atrazine from aqueous medium. Bioresour. Technol. 164, 609e614. Caturla, F., Molina-Sabio, M., Rodriguez-Reinoso, F., 1991. Preparation of activated carbon by chemical activation with ZnCl2. Carbon 29, 999e1007. Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: a review. Bioresour. Technol. 99, 4044e4064. Colleran, E., Pender, S., Phipott, U., O’Flaherty, V., Leahy, B., 1998. Full-scale and laboratory-scale anaerobic treatment of citric acid production wastewater. Biodegradation 9, 233e245. Daud, W.M.A.W., Ali, W.S.W., 2004. Comparison on pore development of activated carbon produced from palm shell and coconut shell. Bioresour. Technol. 93, 63e69. Daud, W.M.A.W., Ali, W.S.W., Sulaiman, M.Z., 2000. The effects of carbonization temperature on pore development in palm-shell-based activated carbon. Carbon 38, 1925e1932. Deschamps, G., Caruel, H., Borredon, M.-E., Bonnin, C., Vignoles, C., 2003. Oil removal from water by selective sorption on hydrophobic cotton fibers.1. Study of sorption properties and comparison with other cotton fiber-based sorbents. Environ. Sci. Technol. 37, 1013e1015. 479 Guo, J., Lua, A.C., 2002a. Microporous activated carbons prepared from palm shell by thermal activation and their application to sulfur dioxide adsorption. J. Colloid Interface Sci. 251, 242e247. Guo, J., Lua, A.C., 2002b. Textural and chemical characterizations of adsorbent prepared from palm shell by potassium hydroxide impregnation at different stages. J. Colloid Interface Sci. 254, 227e233. Guo, J., Lua, A.C., 2002c. Characterization of adsorbent prepared from oil-palm shell by CO2 activation for removal of gaseous pollutions. Mater. Lett. 55, 334e339. Guo, J., Lua, A.C., 2003. Adsorption of sulphur dioxide onto activated carbon prepared from oil-palm shells with and without pre-impregnation. Sep. Purif. Technol. 30, 265e273. Guo, J., Luo, Y., Lua, A.C., Chi, R.-A., Chen, Y.-L., Bao, X.-T., Xiang, S.-X., 2007. Adsorption of hydrogen sulphide (H2S) by activated carbons derived from oilpalm shell. Carbon 45, 330e336. Guo, J., Xu, W.S., Chen, Y.L., Lua, A.C., 2005. Adsorption of NH3 onto activated carbon prepared from palm shells impregnated with H2SO4. J. Colloid Interface Sci. 281, 285e290. Harada, H., Uemura, S., Monomoi, K., 1994. Interactions between sulphate-reducing bacteria and methane-producing bacteria in UASB reactors fed with low strength wastes containing different levels of sulphate. Water Res. 28, 355e367. Hilton, B.L., Oleszkiewicz, J.A., 1988. Sulphide-induced inhibition of anaerobic digestion. J. Environ. Eng. 114, 1377e1391. Hu, Z., Srinivasan, M.P., 2001. Mesoporous high-surface-area activated carbon. Microporous Mesoporous Mater. 43, 267e275. Japanese Industrial Standard, 1984. Standard for Granular Activated Carbon JIS M 8812-1984. Koster, I.W., Rinzema, A., De Vegt, A.L., Lettinga, G., 1986. Sulfide inhibition of the methanogenic activity of granular sludge at various pH levels. Water Res. 20, 1561e1567. Lua, A.C., Lau, F.Y., Guo, J., 2006. Influence of pyrolysis conditions on pore development of oil-palm-shell activated carbons. J. Anal. Appl. Pyrolysis 76, 96e102. Ma, A.N., 1995. A novel treatment for palm oil mill effluent. PORIM Occas. Pap. 29, 201e212. Oliveira, L.C.A., Rios, R.V.R.A., Fabris, R.J.D., Garg, V., Sapag, K., Lago, R.M., 2002. Activated carbon/iron oxide magnetic composites for the adsorption of contaminants in water. Carbon 40, 2177e2183. Oude Elferink, S.J.W.H., Visser, A., Hulshoff Pol, L.W., Stams, A.J.M., 1994. Sulphate reduction in methanogenic bioreactors. FEMS Microbiol. Rev. 15, 119e136. Sumathi, S., Bhatia, S., Lee, K.T., Mohamed, A.R., 2009. Optimization of microporous palm shell activated carbon production for flue gas desulphurization: experimental and statistical studies. Bioresour. Technol. 100, 1614e1621. Tan, J.S., Ani, F.N., 2004. Carbon molecular sieves produced from oil palm shell for air separation. Sep. Purif. Technol. 25, 47e54. Un, U.T., Koparal, A.S., Ogutverren, U.B., 2009. Electrocoagulation of vegetable oil refinery wastewater using aluminum electrodes. J. Environ. Manage. 90, 428e433. Yang, C.-L., 2007. Electrochemical coagulation for oily water demulsification. Sep. Purif. Technol. 54, 388e395. Zouboulis, A.I., Avranas, A., 2000. Treatment of oil-in-water emulsions by coagulation and dissolved-air flotation. Colloids Surf. 172, 153e161.