Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Batch sorption–desorption of As(III) from waste water by magnetic palm kernel shell activated carbon using optimized Box–Behnken design

Batch sorption–desorption of As(III) from waste water by magnetic palm kernel shell activated... Appl Water Sci (2017) 7:4573–4591 https://doi.org/10.1007/s13201-017-0610-9 O R I G IN AL ARTI CL E Batch sorption–desorption of As(III) from waste water by magnetic palm kernel shell activated carbon using optimized Box–Behnken design 2 1 1 • • • Chinedum Anyika Nur Asilayana Mohd Asri Zaiton Abdul Majid 1 2 Jafariah Jaafar Adibah Yahya Received: 25 March 2017 / Accepted: 29 August 2017 / Published online: 8 September 2017 The Author(s) 2017. This article is an open access publication Abstract In this study, we converted activated carbon the MPKSF functional groups and As(III). The findings (AC) into magnetic activated carbon (MAC), which was suggested that the MPKSF exhibited a strong capacity to established to have removed arsenic (III) from wastewater. efficiently remove As(III) from wastewater, while the Arsenic (III) is a toxic heavy metal which is readily soluble desorption studies showed that the As(III) was rigidly in water and can be detrimental to human health. The MAC bound to the MPKSF thereby eliminating the possibility of was prepared by incorporating Fe O into the AC by using secondary pollution. 3 4 Fe O extracted from a ferrous sulfate solution, designated: 3 4 magnetic palm kernel shell from iron suspension Keywords As(III)  Sorption  Desorption  Waste water (MPKSF). Batch experiments were conducted using two treatment  Magnetic palm kernel shell activated carbon methods: (1) one-factor-at-a-time and (2) Box–Behnken statistical analysis. Results showed that the optimum con- ditions resulted in 95% of As(III) removal in the wastew- Introduction ater sample. The adsorption data were best fitted to the Langmuir isotherm. The adsorption of As(III) onto the Removal of contaminants such as arsenic (As) from MPKSF was confirmed by energy dispersive X-ray spec- wastewater by adsorption mechanisms remains the most trometry analysis which detected the presence of As(III) of effective method (Elizalde-Gonzalez et al. 2001). Most 0.52% on the surface of the MPKSF. The Fourier transform studies in the field of adsorption for the removal of heavy infrared spectroscopy analysis of the MPKSF–As presented metals from water have mainly focused on the use of AC, -1 a peak at 573 cm , which was assigned to M–O (metal– activated alumina, sand impregnated with iron, polymer oxygen) bending, indicating the coordination of As(III) resins, hydrous ferric oxide and natural ores (Addo Ntim with oxygen through the formation of inner-sphere com- and Mitra 2011). Although AC has been found to be more plexation, thereby indicating a covalent bonding between effective relative to the other adsorbents mentioned above, especially for the removal of heavy metals from aqueous solutions, with percentage removal ranging from 82 to 96% (Ribeiro et al. 2006), this technique may not be adequate when it comes to a heavy metal like As(III) which is known to be highly soluble in water. Arsenic exists in two dif- ferent oxidation states (1) arsenite, As(III) and (2) arsenate, & Zaiton Abdul Majid As(V). As(III) is different from As(V) in a number of ways. zaiton@kimia.fs.utm.my; zaitonmajid@utm.my Firstly, difficulties arise when it comes to the removal of As(III) from wastewater compared to arsenate As(V) due to Department of Chemistry, Faculty of Science, Universiti its high solubility in the aqueous environment, hence Teknologi, Malaysia, 81310 Johor Bahru, Malaysia As(III) which is the most toxic is also the most difficult to Environmental Biotechnology Laboratory, Faculty of remove from water (Pattanayak et al. 2000). Secondly, Biosciences and Medical Engineering, Universiti Teknologi, whereas As(V) is mostly removed by outer-sphere Malaysia, 81310 Johor Bahru, Malaysia 123 4574 Appl Water Sci (2017) 7:4573–4591 complexation, i.e., by electrostatic attraction (Cheng et al. sorption–desorption of As(III) onto the prepared magnetic 2016), As(III) can be removed by inner-sphere complexa- activated carbon. tion, which is a covalent bonding as demonstrated in this study. On the other hand, the long-term effects of drinking Experimental water contaminated with arsenic include cancer of the skin, lung, bladder, and kidney, skin thickening, neurological Chemicals and reagents disorders, muscular weakness and nausea (Jain and Ali 2000;WHO 1981). This has led to numerous studies Phosphoric acid (H PO , 85 wt%, Merck, Germany) was 3 4 regarding the improvement of AC by magnetic modifica- used to pretreat and impregnate the raw materials. The tion to increase its capacity to remove heavy metal con- chemical reagents for the preparation of magnetic activated taminants. The MAC adsorbents exhibited magnetic carbon were of three kinds: (1) iron (III) chloride ([96%): properties with greater efficiency for the adsorption of Sigma-Aldrich; (2) iron (II) sulfate (99.5%), Qrec; (3) contaminants from aqueous solution (Xu et al. 2010). sodium hydroxide, Merck Germany. A 1000 ppm stock Whereas palm kernel shells were used in the production of solution of arsenite, As(III), was prepared in double dis- AC due to its high carbon content and low organic content tilled water from 0.05 M sodium arsenite (NaAsO ) pre- (Daud and Ali 2004), while at the same time high quality pared by Fluka. Hydrochloric acid (HCl) and sodium of AC can be synthesized from PKS waste (Adhoum and hydroxide (NaOH) were used to adjust the solution pH of Monser 2002; Budinova et al. 2006; Hussein et al. 1996). As(III). One major advantage of the MAC is that it exhibits magnetic characteristics, in addition to having demon- Experimental methods strated to be effective for adsorption in dilute solutions, while its high specific surface area due to the presence of Material development and characterization microporous structure results in greater capacity for the adsorption of heavy metals (Nakahira et al. 2006). So far, The raw PKSs (100 g) used in this study were obtained magnetization has been identified as being potentially from a palm oil estate located at Jalan Sawah, Pekan important in improving the sorption characteristics of Nenas, Johor Bahru, Malaysia. Sample pretreatment was organic adsorbents with well-developed structures, i.e., of carried out by weighing out a 50-g portion of the raw PKS, woody feedstock such as PKS (Trakal et al. 2016). How- which was ground and sieved to particle sizes in the range ever, previous studies on the removal of As(III) from of 75–250 lm, then soaked and impregnated with 10 mL wastewater employed the use of biochar made from empty of 30 wt% of phosphoric acid, H PO , at room temperature 3 4 fruit bunch and rice husks (Samsuri et al. 2013), which are with PKS-to-acid ratio of 1:1. The sample was left fibrous in nature, unlike the PKS feedstock used in this impregnated for 24 h, and afterwards it was washed with study, which is a woody feedstock. Moreover, they used distilled water and dried at room temperature. biochar which is a different material. Another problem with A 10-g portion of the pre-treated PKSs was transferred their study was that they did not carry out a desorption test into five conical flasks containing different concentrations to ascertain the stability of the Fe-coated organic adsor- of H PO , i.e., 10, 20, 30, 40 and 50% wt/wt, respec- 3 4 bents. Further, their study was on competitive adsorption tively. Similarly, the pre-treated PKS was treated again between As(III) and As(V). Similarly, in the paper by using dilute acid (10, 20 and 30% wt/wt H PO ). This 3 4 Payne and Abdel-Fattah (2005) using Fe-coated AC for the was done to determine the appropriate surface area at removal of As(III) and As(V) from water, they relied on which if the surface area obtained by using dilute acid commercially procured AC, which had a poorly developed was high, then the use of concentrated acid can be structure and hence was only able to remove 60% As(III) reduced and the experimental costs will be less. This ratio from waste water unlike in this study where 95% removal (1:1) implied that 10 g of the raw PKSs was soaked in was achieved by using MPKS which has a well-developed 10 g of H PO for 24 h. The excess acid was then filtered 3 4 structure. In addition, they did not look at the magnetic and the soaked PKSs were placed in a muffle furnace and properties of the Fe-coated AC. heated at 200 C for 30 min to initiate the carbonization Therefore, the influence of the MAC on the removal of process. As(III) from waste water has not been completely eluci- Subsequently, the temperature of the furnace was dated. The objectives of this study are: (1) to prepare increased to the range of 400–550 C and held for 2 h magnetic activated carbon for the removal of arsenic from followed by cooling to room temperature. Afterwards, the water; (2) to optimize the parameters for As(III) sorption samples were thoroughly washed and rinsed using vacuum using Box–Behnken design; (3) to study the filtration with hot distilled water to remove all the excess 123 Appl Water Sci (2017) 7:4573–4591 4575 acid until the pH of the filtrate was approximately 7. The under pH 6 and 7, (5) effect of temperature, in this case, a samples were then dried in the oven at a temperature of temperature of 30 C was considered preferable for the 110 C for 24 h. The AC samples were then stored in adsorption of the As(III) onto the MPKSF. desiccators for further characterization and adsorption The residual arsenic solution was analyzed using gra- studies. The preparation of the MPKSF was achieved by phite furnace atomic absorption spectrometry (GFAAS). utilizing a suspension of ferric chloride/ferrous sulfate. The And the same procedures were repeated using a real water characterization experiments were conducted on the waste sample from Skudai River, Johor Bahru Malaysia, MPKSF by Fourier transform infrared spectroscopy spiked with arsenite to analyze the percentage of arsenite (FTIR), X-ray diffraction (XRD), particle size analysis, removal using MPKSF. nitrogen adsorption analysis, scanning electron microscopy (SEM), field emission scanning electron microscopy Determination of pHpzc (point of zero charge) of the (FESEM), energy dispersive X-ray spectrometry (EDX) samples The pHpzc of both PKSAC and MPKSF was and the point of zero charge (pHpzc) was also determined determined. A 50-mL solution of 0.01 M NaCl was placed as part of the characterization. The magnetic saturation of in a closed Erlenmeyer flask. The pH of the solution was the MAC sample was characterized using vibrating sample adjusted to achieve a suspension pH of between 2 and 12 magnetometer (VSM). by adding 0.1 M HCl or 0.1 M NaOH solutions in ten conical flasks. Approximately 0.15 g each of the PKSAC Batch experimental procedure and testing methods and MPKSF were added and the final pH was measured after 48 h. The pH of each solution recorded was plotted. Batch experiments were carried out by grinding the The intersection of pH and pH of the solution was initial final MPKSF into fine powder of 75 lm particle sizes. The stock then taken as the pHpzc. solution was prepared by diluting 0.05 M of sodium arsenite (NaAsO ) with distilled water up to a concentra- Desorption procedure of magnetic activated carbon tion of 1000 ppm to give a 1000 ppm arsenite (As(III)) (MPKSF) The optimum amount of MPKSF loaded with -1 stock solution. The pH of the solution was then adjusted arsenite (48.08 lgg ) obtained after the adsorption pro- using hydrochloric acid (HCl) until it reached a pH of 7. A cess was then added into a 50 mL of distilled water in a 0.05–0.30 g portion of the powdered MPKSF was placed centrifuge tube. The solution was shaken at 150 rpm and into a conical flask together with 200 mL arsenite solution. then agitated at specific time intervals for up to 48 h. The The solution was then shaken for 24 h at different tem- solution was then centrifuged and the supernatant was peratures ranging from 10 to 40 C. The contact time in the collected for further analysis to examine the concentration shaker was also varied ranging from 5 to 720 min. The of As(III) desorbed from the MPKSF. The desorption initial concentration of the arsenite solution was varied procedure was repeated three times and the MPKSF -1 ranging from 5 to 100 lgL . This was done to mimic the adsorbent was reserved for further analysis. drinking water standard. The preliminary experiment was conducted at an equilibrium time of 180 min. After that, Graphite furnace atomic absorption spectrometry the suspension was filtered through 0.45-lm pore size (GFAAS) The detection and the concentration of arsen- membrane filter. ite were conducted by a graphite furnace atomic absorp- The preliminary study on the adsorption of As(III) on tion spectrometry (GFAAS). The samples were analyzed MPKSF, was divided into 6 factors and are described in triplicates, to obtain the optimum result of -1 briefly namely (1) effect of contact time, which was found 10–40 lgL . A small amount of the sample which was to be 180 min and established to be the time taken to reach in the range of 20–100 lL was placed into the graphite adsorption–desorption equilibrium; (2) effect of initial tube manually. Further, the arsenite samples were acidi- -1 As(III) concentration, which was found to be 5–70 lgL fied with nitric acid to a pH of less than 2. Upon injection with percentage removal of 87.58–89.42%; (3) effect of of the samples into the graphite tube, they are vaporized. adsorbent dosage, was found to be 0.3 g, which resulted in Subsequently, the amount of light energy absorbed in the an increase in percentage removal of As(III) from 42.17 to vapor was considered to be proportional to the atomic 96.58%. This was due to a higher dose of adsorbent utilized concentration. which provides greater proportion of adsorption sites for As(III) to bind on the MPKSF surface (Yao et al. 2014); (4) Box–Behnken design effect of pH, in this case, the effect of the initial solution pH in the range of 6–8 on the adsorption of As(III) onto the In this study, the Box–Behnken design was used to opti- MPKSF, indicated that the removal of As(III) was better mize the number of experiments to be conducted with the 123 4576 Appl Water Sci (2017) 7:4573–4591 aim of determining the probable interactions between the Results and discussion parameters under study and their influence on the adsorp- tion of As(III) onto the MPKSF. Further characteristics of Characterization of adsorbents the Box–Behnken design has been described elsewhere (Kumar et al. 2008). The characterization results (detailed characterization of In Table 1, the experimental parameters and a the adsorbent (s) has been described elsewhere, Anyika 3-level 5-factorial Box–Behnken experimental design et al. 2017) revealed that the MPKSF presented better are presented. This was applied to examine and validate characteristics and was therefore selected as the sole the adsorption system parameters influencing the adsorbent for the adsorption studies. It presented a higher 2 -1 adsorption of As(III) onto the MPKSF. Contact time BET surface area of 257 m g , higher pore volume of -1 (5–720 min), pH (6–8), adsorbent dose (0.05–0.30 g), 0.1124 cc g and higher magnetic properties with a -1 -1 initial As(III) concentration (5–100 lgL ), tempera- magnetic saturation of 49.55 emu g . The FTIR spectrum -1 ture (20–40 C) represent the variable input parameters. of the MPKSF exhibited intense OH bending at 1629 cm The factor levels (3) were coded as (-1 = low, which can be attributed to the presence of oxygen in the -1 0 = medium level or central point, and ?1 = high samples, while the absorption bands at 1093 and 579 cm level). Response surface methodology (RSM) was indicated the presence of C–O stretching and metal–oxy- applied to the experimental data using the Design gen (M–O) bands due to the interaction of iron and oxygen. Expert statistical software version 7.1.6 by Stat-Ease, To illustrate that the MPKSF acquired magnetic properties, Inc., Minneapolis, USA. the XRD data were analyzed. The MPKSF exhibited the The regression equation of the designed experiment was presence of Fe O at 2h 30.75, 35.95, 57.35, 63.20 3 4 obtained by applying four models namely linear, interac- from the XRD diffractogram. Similarly, three peaks at 2h tive, quadratic and cubic models which were fitted to the 19.30, 43.45, 54.10 and a peak at 2h 24.40 which can experimental data obtained from the design system. be assigned to c-Fe O and a-Fe O , respectively, were 2 3 2 3 To select the best model, i.e., after the responses have detected. been recorded, the data were analyzed using three different The point of zero charge of the MPKSF was determined tests in order to decide the adequacy of the models stated to explain the surface charge phenomena as well as the above to represent the adsorption process of MPKSF–As. magnetic properties of the MPKSF. With respect to the These validation tests are the sequential model sum of surface charge, the point of zero charge of MPKSF squares (F test), lack-of-fit test and the model summary occurred at pH 5.94. Since As(III) is positively charged, it statistics. Further, a quadratic polynomial was used to is critical for the surface of the MPKSF to be negatively explain the relationship between the parameters and As(III) charged in order for the adsorption process to occur. As the residual concentration (%). solution pH is higher than 5.94, the surface of the MPKSF The second-order polynomial is represented by Eq. (1): exhibits greater formation of hydroxide ions. However, at pH 6 and 7, the MPKSF demonstrated greater adsorption k k k X X X ¼ b þ b x þ b x þ b x x þ e; ð1Þ compared to pH 8. This was presumed to have resulted i i j 0 i ii i ij i1 i1 1s s i j from the formation of As(III) precipitate at a higher pH hence reducing the adsorption efficiency. Additionally, at where the terms have their established meanings (Ku- higher pH, As(III) has the potential to be oxidized to namneni and Singh 2005; Shehzad et al. 2016). A design of As(V), which may significantly reduce the adsorption of 46 tests was formulated. As(III) onto the MPKSF (Vance 2002). In Table 2, the Table 1 Independent parameters and their levels used for Box–Behnken design Parameters, unit Factors Levels -10 1 Contact time (min) A 5 362.5 720 pH B 67 8 Adsorbent dosage (g) C 0.05 0.17 0.30 -1 Initial As concentration (lgL ) D 5 52.5 100 Temperature (C) E 20 30 40 123 Appl Water Sci (2017) 7:4573–4591 4577 Table 2 Initial and final pH value of PKSAC and MPKSF reaction MPKSF, i.e., pHpzc = 5.94, its surface becomes negatively solutions charged, hence the protonated As(III) will have a greater affinity towards the MPKSF surface. Based on the Box– Initial pH Final pH Behnken optimization, the optimum adsorption occurred at PKSAC MPKSF pH 6.55, which conformed to the aquatic environmental 4.00 3.94 4.26 range of pH of 5–9 (Zou et al. 2016). 6.00 4.22 5.94 As seen in Fig. 1b, the pHpzc value of MPKSF decreased its acidity to attain a pH of 5.94. This may be 8.00 4.42 5.31 10.00 4.29 6.54 due to the iron oxide extracted from the ferric chlo- ride/ferrous sulfate solution (FeOF) which was used in the 12.00 11.41 11.59 production of MPKSF. To further illustrate that the Point of zero charge MPKSF acquired magnetic properties, the hydration of Fe O in aqueous solution resulted in the formation of a- 3 4 initial pH and final pH of the reaction solutions for both Fe O in an acidic condition as depicted by the reaction in 2 3 adsorbent samples, PKSAC and MPKSF are presented. Eq. (2). To illustrate that a-Fe O was formed, previous 2 3 In Fig. 1a, b, a plot of pH versus pH for adsorbent final initial studies have reported that the pHpzc for the untreated PKSAC and MPKSF is presented. From the graph, the point Fe O was 6.5 while for c-Fe O , it was at pH 5.9 3 4 2 3 of zero charge (pHpzc) of the sample represents the point (Milonjic et al. 1983) which indicate that the value of pH where the plot of final pH versus initial pH intersects with for both was nearly acidic even in the untreated condition. the line at which the final pH equals to the initial pH. The Further, Schwertmann and Murad (1983) had reported that blue line in both graphs indicates the line of pH = - final a-Fe O is predominantly formed at pH 7–8. In this study, 2 3 pH , while the red and green curves indicate the plots of initial it was demonstrated that upon impregnation of the PKSAC pH against pH for PKSAC and MPKSF, respec- final initial with FeOF, the acidity of the modified PKSAC (MPKSF) tively. Figure 1a shows that the pHpzc of PKSAC adsorbent was reduced from 3.94 to 5.94 due to the formation of a- was 3.94, which indicated that the sample was acidic, due to Fe O from the reaction of Fe O with water as represented 2 3 3 4 the impregnation of PKSAC using phosphoric acid by the Eq. (2) below: (H PO ). A suitable acid activation results in the production 3 4 2Fe O þ H O  3Fe O þ 2H þ 2e : ð2Þ 3 4 2 2 3 of high quality and high surface area AC (Yakout and Sharaf El-Deen 2016). In the adsorption process of As(III) Box–Behnken statistical analysis onto MPKSF, at a pH above the point of zero charge of the In Table 3, the most important parameters influencing the efficiency of adsorption of As(III) onto the MPKSF are represented by the letters: A, B, C, D and E, which repre- sents the coded symbols for the respective factors: contact time, pH, adsorbent dosage, initial As(III) concentration and the temperature parameters. The combined effects of these factors were evaluated by performing experiments on the different combinations of these parameters. The applied Box–Behnken model can be expressed as Eq. (3): Y ¼ X þ X A þ X B þ X C þ X D þ X E þ X AB 0 1 2 3 4 5 6 þ X AC þ X AD þ X AE þ X BC þ X BD 7 8 9 10 11 þ X BE þ X CD þ X CE þ X DE þ X A 12 13 14 15 16 2 2 2 2 þ X B þ X C þ X D þ X E ; ð3Þ 17 18 19 20 where Y is the response, X and X depicted the global mean 0 i and other regression coefficients, respectively, while A, B, C, D and E are the coded symbols for the respective fac- tors: contact time, pH, adsorbent dosage, initial As(III) concentration and the temperature parameters. Fig. 1 a Graph of pH versus pH for the adsorbent PKSAC final initial In Table 4, the statistical significance of the ratio of suspension. b Graph of pH versus pH graph for the adsorbent final initial mean square variation due to regression and mean square MPKSF suspension 123 4578 Appl Water Sci (2017) 7:4573–4591 Table 3 Experimental, actual and predicted values of Y for As(III) onto MPKSF Standard run order AB C D E Actual value Predicted value 1 5 6 0.17 52.5 30 86.93 86.91 2 720 6 0.17 52.5 30 99.81 99.80 3 5 8 0.17 52.5 30 81.67 81.65 4 720 8 0.17 52.5 30 93.96 93.94 5 362.50 7 0.05 5 30 57.41 57.43 6 362.50 7 0.3 5 30 51.48 51.50 7 362.50 7 0.05 100 30 18.24 18.22 8 362.50 7 0.3 100 30 92.46 92.44 9 362.50 6 0.17 52.5 20 91.63 91.61 10 362.50 8 0.17 52.5 20 88.46 88.47 11 362.50 6 0.17 52.5 40 96.80 96.78 12 362.50 8 0.17 52.5 40 88.80 88.81 13 5 7 0.05 52.5 30 28.03 28.05 14 720 7 0.05 52.5 30 70.81 70.81 15 5 7 0.3 52.5 30 92.36 92.37 16 720 7 0.3 52.5 30 74.79 74.78 17 362.50 7 0.17 5 20 79.93 79.91 18 362.50 7 0.17 100 20 83.05 83.05 19 362.50 7 0.17 5 40 84.96 84.94 20 362.50 7 0.17 100 40 83.52 83.53 21 362.50 6 0.05 52.5 30 63.35 63.36 22 362.50 8 0.05 52.5 30 29.41 29.41 23 362.50 6 0.3 52.5 30 69.10 69.11 24 362.50 8 0.3 52.5 30 91.95 91.95 25 5 7 0.17 5 30 78.30 78.28 26 720 7 0.17 5 30 84.89 84.89 27 5 7 0.17 100 30 73.18 73.17 28 720 7 0.17 100 30 91.72 91.73 29 362.50 7 0.05 52.5 20 35.69 35.68 30 362.50 7 0.3 52.5 20 96.26 96.26 31 362.50 7 0.05 52.5 40 64.88 64.86 32 362.50 7 0.3 52.5 40 72.58 72.57 33 5 7 0.17 52.5 20 87.86 87.88 34 720 7 0.17 52.5 20 98.26 98.28 35 5 7 0.17 52.5 40 88.42 88.45 36 720 7 0.17 52.5 40 103.20 103.22 37 362.50 6 0.17 5 30 81.57 81.59 38 362.50 8 0.17 5 30 75.49 75.50 39 362.50 6 0.17 100 30 81.90 81.92 40 362.50 8 0.17 100 30 76.89 76.90 41 362.50 7 0.17 52.5 30 85.53 85.53 42 362.50 7 0.17 52.5 30 85.52 85.53 43 362.50 7 0.17 52.5 30 85.51 85.53 44 362.50 7 0.17 52.5 30 85.54 85.53 45 362.50 7 0.17 52.5 30 85.52 85.53 46 362.50 7 0.17 52.5 30 85.53 85.53 123 Appl Water Sci (2017) 7:4573–4591 4579 Table 4 ANOVA for response surface quadratic model (Y) Source Sum of squares Df Mean square F value p value (p [ F) Model 6726.26 20 335.36 99.63 \0.0001 A: contact time 218.30 1 218.30 64.85 \0.0001 B: pH 32.86 1 32.86 9.76 0.0045 C: adsorbent dosage 2084.38 1 2084.38 619.20 \0.0001 D: initial conc. As 2242.50 1 2242.50 666.17 \0.0001 E: temperature 5.87 1 5.87 1.74 0.1987 AB 0.03 1 0.03 9.098E-003 0.9248 AC 248.38 1 248.38 73.78 \0.0001 AD 40.51 1 40.51 12.04 0.0019 AE 2.89 1 2.89 0.86 0.3630 BC 221.71 1 221.71 65.86 \0.0001 BD 4.49 1 4.49 1.34 0.2588 BE 1.59 1 1.59 0.47 0.4986 CD 732.51 1 732.51 217.61 \0.0001 CE 190.85 1 190.85 56.70 \0.0001 DE 0.26 1 0.26 0.08 0.7833 A 0.34 1 0.34 0.10 0.7516 B 1.85 1 1.85 0.55 0.4654 C 493.91 1 493.91 146.73 \0.0001 D 120.20 1 120.20 35.71 \0.0001 E 6.70 1 6.70 1.99 0.1706 Residual 84.16 25 3.37 Lack of fit 84.11 20 48.21 4.79 0.0687 Pure error 0.04 5 Corr. total 7154.40 45 residual error was tested using ANOVA. The results of the (3) model summary statistics. The results showed that the ANOVA indicated that the F values obtained for all the p value for majority of the regression were \0.05. This regressions were higher, which indicated that that majority implied that one of the terms in the regression equation was of the variation in the response can be explained by the significantly correlated to the response variable. Further, regression equation (Kumar et al. 2008). To determine the quadratic model was found to yield the best fit of R , 2 2 whether the F is large enough to result in a statistical Adjusted R and predicted R values of 0.9876, 0.9777 and significance, the p value is examined. In this case, the 0.9505, respectively (Table 5). This also implied that the model is considered to be statistically significant if the model does not explain 1% of the experimental results. values under the column p [ F value is \0.05 (Table 4) Again, the high R values in Table 5 and the p value of (Segurola et al. 1999). The ANOVA result for the MPKSF– \0.0001 in Table 4, indicate that the quadratic polynomial As design system shows that the F value of 99.63 and its was highly significant in explaining the relationship p value of \0.05 imply that the model was significant between the parameters and As(III) residual concentration towards the response. Hence in this analysis, A, B, C, D, (%). 2 2 AC, AD, BC, CD, CE, C and D were the significant terms In Table 5, since the cubic model was established to be (Table 4). Besides, the ANOVA results for the MPKSF–As aliased, the quadratic model was therefore, chosen to be adsorption system showed that the F value is 99.63 used for further analysis. Further, under the lack of fit (Table 4), indicating that the terms on the model are having (Table 5) the F value is not significant, with an F value of a significant effect on the response. 4.79 and the p value of 0.0687. This shows that the lack of In Table 5, the adequacy of the model for the adsorption fit was not significant relative to the pure error. However, of As(III) onto the MPKSF was determined by three tests: the lack-of-fit value indicates that there is 6.87% possibility (1) sequential model sum of squares; (2) lack-of-fit tests; that the error resulted from noise. Thus, the non-significant 123 4580 Appl Water Sci (2017) 7:4573–4591 Table 5 Adequacy of the model tested for MPKSF–As design system Source Sum of squares Df Mean square Fp value (p [ F) Remark Sequential model sum of squares Mean vs total 5863.8 1 5863.8 – – – Linear vs mean 4583.91 5 916.78 16.61 \0.0001 – 2FI vs linear 1443.23 10 144.32 5.67 \0.0001 Quadratic vs 2FI 680.13 5 136.03 40.41 \0.0001 Suggested Cubic vs quadratic 59.05 15 3.94 1.57 0.2386 Aliased Residual 25.10 10 2.51 – – – Total 12,655.22 46 275.11 – – – Lack-of-fit tests Linear 2207.47 35 63.07 7345.18 \0.0001 – 2FI 764.24 25 30.57 3560.12 \0.0001 – Quadratic 84.11 20 48.21 4.79 0.0687 Suggested Cubic 25.06 5 5.01 583.72 \0.0001 Aliased Pure error 0.043 5 8.587E-003 – – 2 2 2 Source Std. dev. R Adjusted R Predicted R PRESS Remark Model summary statistics Linear 7.43 0.6750 0.6343 0.5575 3005.16 – 2FI 5.05 0.8875 0.8312 0.7113 1960.64 – Quadratic 1.83 0.9876 0.9777 0.9505 336.51 Suggested Cubic 1.58 0.9963 0.9834 0.7638 1603.97 Aliased value for the lack of fit showed that the model was valid for shows the antagonistic effect on the adsorption process further analysis. The final mathematical equation given by (Tan et al. 2008). Further, the positive value of the model the Box–Behnken design in terms of actual values deter- term indicates the effect that favors the optimization mined by Design Expert software is presented in Eq. (4). process while the negative model term value represents the Removal of As(III) (%) is given by: inverse interaction between the parameters with the response. Y ¼ 85:53 þ 6:29  A  2:78  B þ 17:07  C þ 0:43  D þ 1:38  E  0:15  A  B Adsorption studies 15:09  A  C þ 2:99  A  D þ 1:09  A  E þ 14:20  B  C þ 0:27  B  D  1:21  B  E In this study, the Box–Behnken design was used for the optimization of the selected parameters. Thus, the effects þ 20:04  C  D  13:22  C  E  1:14  D  E of these parameters on the MPKSF–As adsorption was 2 2 2 2 þ 4:05  A þ 1:01  B  23:07  C  7:56  D shown by the response surface plots using two parameters þ 4:89  E : ð4Þ simultaneously while the remaining parameters were set at their center points. However, it should be noted that since The equation implied that the constant with a value of two parameters were used simultaneously, hence, only one 85.53 (see Eq. (4)) which is independent of any factor or set of the figures were presented per variable, while the rest interaction between factors suggests that the average were referred to accordingly under the affected variables, removal of As(III) by MPKSF was 85.53%. Although i.e., from ‘‘Effect of temperature’’ to ‘‘Effect of contact this average removal is independent of the factors in the time’’. experimental setup (Kumar et al. 2008). In addition, Eq. (4) shows that pH and initial As(III) concentration Effect of selected variables and response surface plots had a positive effect while contact time, temperature and adsorbent dosage had a negative effect on the adsorption Effect of pH As seen in Fig. 2a–d, the combined effects percentage of As(III) by MPKSF. In general, the positive of pH with contact time, initial As(III) concentration, sign represents the synergistic effect and the negative sign adsorbent dosage and temperature, respectively, were 123 Appl Water Sci (2017) 7:4573–4591 4581 86.25 95.25 82.5 90.5 78.75 85.75 6.00 6.00 6.50 6.50 7.00 720.00 7.00 100.00 B: pH 541.25 76.25 B: pH 7.50 362.50 7.50 52.50 183.75 28.75 5.00 8.00 8.00 5.00 A: contact time D: initial conc As 93.5 78.5 86.5 45.5 83 6.00 8.00 6.50 7.50 0.30 7.00 0.24 7.00 B: pH 40.00 35.00 0.17 7.50 6.50 B: pH 30.00 0.11 25.00 0.05 6.00 8.00 20.00 C: adsorbent dosage E: temperature Fig. 2 3D response surface plots for As(III) removal vs a contact c adsorbent dose (0.05–0.3 g) and the pH of the adsorbent MPKSF time (5–720 min) and the pH of the adsorbent MPKSF suspension suspension (6–8); d temperature (20–40 C) and the pH of the -1 (6–8); b initial As concentration (5–100 lgL ) and pH (6–8); adsorbent MPKSF suspension (6–8) presented. From the responses, the lower value of As(III) As reported previously, the pHpzc value for MPKSF residual indicates a higher percentage of removal as given was 5.94. At a pH above the pHpzc value, the As(III) by Eq. (5): cations are attracted to the negatively charged surface of MPKSF, which resulted from the deprotonation of MPKSF C  C initial final % Removal ¼  100%: ð5Þ hydroxyl groups. Meanwhile, at pH lower than pHpzc, the initial adsorption of As(III) cations occurs only slightly because Figure 2a shows the interaction between the effect of pH the surface of MPKSF is positively charged due to the and the contact time of As(III) solution with the MPKSF protonation of MPKSF hydroxyl group. Thus, a solution adsorbent. The optimum removal of As occurred at pH 6.55 pH of 6.55 provides the optimum condition for adsorption and at a contact time of 44.73 min with a removal percentage of MPKSF–As. of 94.76% as the other parameters were set to their center points. Figure 2b illustrates the interaction of initial As(III) Effect of temperature concentration with pH with the optimum removal of As(III) achieved at pH 6.55 at an initial As(III) concentration of The experiment was conducted within a temperature range -1 5 lgL with a removal percentage of 99.10%. of 20–40 C. As seen in Fig. 2d (refer to Fig. 2din‘‘Effect Further, Fig. 2c, d shows the relationships between pH of pH’’ above for the effect of temperature and pH) and 3a– with adsorbent dosage and temperature, respectively. The c, the relationships between temperature and pH, contact excluded parameters for the respective response surface time, adsorbent dose and initial concentration, respectively, plots remain at their center points. Figure 2c shows that the are presented. All the four response surface plots show that removal of As(III) attained a maximum of 100% at pH 8 the removal of As(III) increases with temperature. Thus, with an adsorbent dose of 0.3 g, while the other parameters the adsorption process could be described as an endother- were at their center points. In Fig. 2d, the relationship mic process, as the removal of As(III) is optimum at higher between pH and temperature gives an optimum removal of temperatures with efficiency of 91–100% removal. 91.89% at pH 6 and a temperature of 40 C. Removal % Removal % Removal % Removal % 4582 Appl Water Sci (2017) 7:4573–4591 98.75 93.5 88.25 0.30 0.24 20.00 25.00 0.17 720.00 30.00 541.25 C: adsorbent dosage 0.11 40.00 362.50 35.00 35.00 30.00 A: contact time 183.75 25.00 0.05 40.00 E: temperature 20.00 5.00 E: temperature 88.25 84.5 80.75 5.00 28.75 40.00 35.00 52.50 30.00 D: initial conc As 76.25 25.00 E: temperature 100.00 20.00 Fig. 3 3D response surface plots for As(III) removal vs a temperature (20–40 C) and contact time (5–720 min); b temperature (20–40 C) and -1 adsorbent dosage (0.05–0.3 g); c temperature (20–40 C) and initial As(III) concentration (5–100 lgL ) Effect of adsorbent dosage the percentage removal data showed increasing pattern while other percentages removal data became slightly Refer to Fig. 2c under ‘‘Effect of pH’’ and Fig. 3b under decreased. In this case, other parameters need to be con- ‘‘Effect of temperature’’ above and Fig. 4a, b under ‘‘Effect sidered especially the adsorbent dosage which provides the of contact time’’ below, for the combined effect of adsor- information on the available vacant adsorption sites for bent (MPKSF) dose with pH, temperature, contact time and adsorption of As(III) at higher concentrations. initial As(III) concentration, respectively. The results showed that the percentage removal increases as the Effect of contact time adsorbent dose increases. The increase in the amount of adsorbent used can be attributed to the formation of a Refer to Fig. 2a (under ‘‘Effect of pH’’, and Fig. 3a under greater surface area and provides more available vacant ‘‘Effect of temperature’’) above, and Fig. 4a, c below. The sites for the adsorption of As(III) on MPKSF surface. results illustrate the combined response surface plots of contact time with pH of the adsorbent MPKSF suspension, Effect of initial As(III) concentration temperature, adsorbent dosage and initial As concentration, respectively. It shows that the increase in contact time Adsorption of MPKSF–As was carried out with different duration causes the removal percentages of As(III) to -1 initial concentrations ranging from 5 to 100 lgL . Refer increase. As the contact time increased, the adsorbate had to Fig. 2b under ‘‘Effect of pH’’ and 3c under ‘‘Effect of enough time to disperse and be adsorbed onto the surface temperature’’ above, and Fig. 4b, c, under ‘‘Effect of and into the pores of MPKSF until the adsorption reached contact time’’ below. The results illustrates the combined equilibrium at an optimum contact time. effect of initial As(III) concentration with pH of the adsorbent MPKSF suspension, temperature, adsorbent dose Optimization using desirability function and contact time, respectively, while two parameters are kept constant. As shown by the response surface plots, it A desirable value for each input parameter and the revealed that as the initial concentration increases, some of response can be selected. The multiple response methods Removal % Removal % Removal % Appl Water Sci (2017) 7:4573–4591 4583 96 75 79 56 720.00 541.25 100.00 76.25 0.30 0.30 362.50 0.24 0.24 52.50 A: contact time 0.17 0.17 183.75 D: initial conc As 28.75 0.11 0.11 0.05 5.00 5.00 0.05 C: adsorbent dosage C: adsorbent dosage 5.00 720.00 28.75 541.25 52.50 362.50 76.25 183.75 D: initial conc As A: contact time 5.00 100.00 Fig. 4 3D response surface plots for As(III) removal vs a contact time (5–720 min) and adsorbent dose (0.05–0.3 g); b initial As(III) -1 -1 concentration (5–100 lgL ) and adsorbent dose (0.05–0.3 g); c initial As(III) concentration (5–100 lgL ) and contact time (5–720 min) was applied to obtain the optimum condition for the five Table 6 Optimal condition and model validation for As(III) removal parameters used including contact time, pH, adsorbent by MPKSF dose, initial As(III) concentration and temperature. The Parameters Optimal conditions numerical optimization reveals the points that maximize Contact time (min) 45 the desirability function. In this study, the maximum level pH 6.55 for As(III) removal was set for desirability at a minimum Adsorbent dosage (g) 0.29 level of contact time (5 min), minimum adsorbent dosage -1 Initial As(III) concentration (lgL ) 100.00 (0.05 g), the maximum level of initial As(III) concentration -1 (100 lgL ), the level of solution pH in the range of 6–8 Temperature (C) 26 Removal (%) and the level of temperature within a range of 20–40 C (data not shown for the desirability ramp). Predicted 96 Using Eq. (5) reported previously to determine the Experimental 95 removal percentage, 96% removal of As(III) was achieved Error (%) 1 using the optimal conditions at contact time of 45 min, -1 initial As(III) concentration maximized to 100 lgL , adsorbent dosage of 0.29 g, initial solution pH of 6.55 and Adsorption equilibrium studies of As(III) onto MPKSF adsorbent the temperature at 26 C as shown in Table 6. Experi- mentally, these optimum values are applied in a verifica- Adsorption isotherm models tion experiment and resulted in 95% removal of As(III) by MPKSF adsorbent with a percentage error of 1% as com- In this study, several isotherm models were applied to pared to the predicted removal of As(III). Thus, this indi- determine the type of adsorption that occurred on the cated that the Box–Behnken design model is reliable for the optimization of various parameters used in an adsorp- magnetic activated carbon surface, by fitting the data to the Langmuir, Freundlich, and Temkin isotherm models. The tion process. correlation coefficient, R obtained from the isothermal Removal % Removal % Removal % 4584 Appl Water Sci (2017) 7:4573–4591 plots was used to identify the isotherm model that best Further, the inherent feature of this isotherm namely a described the adsorption of As(III) onto MPKSF. dimensionless constant separation factor, R can be expressed as an equilibrium parameter. The parameter is The Langmuir adsorption isotherm model calculated using Eq. (9). The effect of the separation factor on the adsorption nature of the Langmuir isotherm is The Langmuir is represented by the following equations; summarized in Table 7. (1) Eq. (6) which was formulated based on the kinetic R ¼ ; ð9Þ theory; (2) Eq. (7) depicts the equation for the value of 1 þ K C L o adsorbate adsorbed on the adsorbent; (3) the linear form of -3 where C = initial concentration of surfactant (mg dm ), the Langmuir equation is given by Eq. (8): K = Langmuir isotherm constant. x q K C max L e ¼ q ¼ ; ð6Þ To illustrate, that the homogeneous adsorption process m 1 þ K C L e predicted by the Langmuir isotherm assumes a monolayer ðÞ C  C V o e adsorption of molecules onto the surface of the adsorbent. q ¼ ; ð7Þ m The Langmuir linear plot of the specific adsorption (C /q ) e e against the equilibrium concentration (C ) for adsorption of C C 1 e e e ¼ þ ; ð8Þ As(III) onto MPKSF is depicted in Fig. 6. The C was q q q K e max max L obtained after 180 min. The correlation coefficient (R ) where x = mass of adsorbate adsorbed (lg), m = mass of obtained from the Langmuir isotherm is 0.9973, which adsorbent (g), V = volume of solution used for adsorption indicates that the adsorption data of the As(III) onto the -1 process (L), C = equilibrium concentration (lgL ), MPKSF surface was well fitted to the Langmuir isotherm -1 C = initial concentration (lgL ), q = amount of o e model. The monolayer adsorption capacity (q )of max -1 adsorbate adsorbed at equilibrium (lgg ), q = max- -1 max As(III) onto MPKSF was found to be 48.08 lgg . imum adsorption at monolayer coverage, K = Langmuir A dimensionless equilibrium parameter, R was used to -1 isotherm constant (L lg ). express the nature of adsorption of the Langmuir isotherm. To assess the Langmuir isotherm model, a graph of C / The R value calculated from the adsorption data is 0.0765 q against C is plotted, a straight line is obtained and the e e indicating that the adsorption of As(III) on MPKSF was a values of K and q are computed from the values of the L max favorable process as the R value lies between 0 to 1 slope and intercept of the graph. The fitted isotherm in this study was illustrated by a plot of equilibrium concentration of adsorbate C on the adsorbent, q , i.e., C /q against the e e e e Table 7 Effect of separation factor on Langmuir isotherm equilibrium concentration of adsorbate, C . Figure 5 R Adsorption nature of isotherm depicts the fitted As(III) adsorption isotherm which indi- cated the presence of a linear part of plotted curve at low R [ 1 Unfavorable C followed by a slight curvature around C = 2.56 - e e R = 1 Linear -1 lgL towards the completion of coverage of the mono- 0 \ R \ 1 Favorable layer with the R value of 0.9599. Hence, this indicated that R = 0 Irreversible adsorption of arsenite onto MPKSF is L-type (having no strict plateau). A similar observation was reported by El- Said et al. (2009) whose study demonstrated the adsorption of As(III) and As(V) using Nigella sativa L. Fig. 6 Langmuir isotherm representing the variation of specific Fig. 5 Langmuir equilibrium isotherm for the adsorption of As(III) adsorption (C /q ) against the equilibrium concentration (C ) for e e e onto MPKSF adsorption of As(III) onto MPKSF 123 Appl Water Sci (2017) 7:4573–4591 4585 (0 \ R \ 1) at an equilibrium temperature of 30 C (Table 7). The Freundlich isotherm model This isotherm was applied as an empirical model which considers the data often fit to the empirical equation stated in Eq. (10): ¼ q ¼ K C ; ð10Þ e F e where x = weight of solute adsorbed, m = weight of Fig. 7 Freundlich isotherm indicating the variation of log q with adsorbent, C = equilibrium concentration of adsorbate -1 respect to log C for adsorption of As(III) onto MPKSF (mg L ), q = amount of solute adsorbed at equilibrium -1 (mg g ), K and n = Freundlich isotherm constant -1 (mg g ) Table 8 Effect of n values on Freundlich isotherm From the isotherm, the n value reveals the nature of the n value Adsorption nature of isotherm adsorption, i.e., how favorable the adsorption process was. n = 1 Linear The value of n was used as the linearity parameter implying n \ 1 Chemical process that if the n value lies between one and ten, this indicates a n [ 1 Physical process favorable sorption process of the adsorbent (Freundlich 1906). The effect of the n value on the nature of adsorption as represented by the Freundlich isotherm is shown in of As(III) and Fe O in MPKSF is depicted in Fig. 8 and 3 4 Table 4. The linear form of the equation is given by supported by the FTIR analysis of the MPKSF–As. Eq. (11). A graph of log (x/m) against log C results in a It is well known that sodium arsenite can be represented straight line at which the value slope and intercept are the as sodium ortho-arsenite (Na AsO ) and sodium meta- 3 3 value of 1/n and log K , respectively. arsenite (NaAsO ), the latter was used in this study for x 1 adsorption. Based on Fig. 9, MPKSF reacted with arsenous log ¼ log q ¼ log K þ log C : ð11Þ e F e m n acid (H AsO ) which is produced according to the chem- 3 3 A multilayer adsorption for the heterogeneous surface is ical reaction (Eq. 11). Firstly, sodium meta-arsenite reacts with water producing arsenic trioxide with sodium ions and indicated by the Freundlich isotherm model. The heterogeneous system of adsorption assumes that there is hydroxide ions followed by the slow hydrolysis of arsenic trioxide. The reaction proceeds in a basic condition, finally no formation of monolayer adsorption of As(III) on MPKSF. Figure 7 illustrates the plot of log q against log producing arsenous acid: C . The correlation coefficient, R obtained from the graph þ 2NaAsO þ H O ! As O þ 2Na 2 2 2 3 illustrates that the Freundlich isotherm model is not well slow hydrolysis þ 2OH !2H AsO : ð12Þ 3 3 fitted to the adsorption data with an R value of 0.9744. The Freundlich constants, K and the n value are shown in Liu et al. (2015) reported that magnetite (Fe O )is 3 4 Table 8. known as a mixture of two iron oxides which are composed The slope of the Freundlich isotherm (Fig. 7) shows that of 67% of Fe(III) and 33% of Fe(II). In the adsorption the 1/n value is less than 1 and this has shown that the study, the major Fe(III) species was reported to interact process is a favorable physical adsorption process. The with As(III) to form either inner-sphere monodentate or smaller 1/n value indicates that a strong bond is present bidentate-binuclear complex. This can be attributed to the between the adsorbent and adsorbate molecules (Okeola oxidation reaction which takes place in the presence of and Odebunmi 2010). oxygen under atmospheric experimental conditions hence In the paper by Mayo et al. (2007), it was demonstrated suggesting the oxidation of Fe(II) to Fe(III). that the adsorption of As(III) and As(V) onto Fe O 3 4 nanoparticles exhibits the surface complexation reaction by Temkin isotherm forming either inner-sphere monodentate or bidentate-bin- uclear complex with iron oxide (Fe O ). A similar reaction 3 4 The Temkin isotherm model was applied to consider the of magnesium with iron oxide, i.e., Fe O and c-Fe O was 3 4 2 3 effect of the interactions between the adsorbent and the reported by Jolstera ˚ et al. (2012). The proposed adsorption adsorbate on an adsorption isotherm. The model assumes 123 4586 Appl Water Sci (2017) 7:4573–4591 Fig. 8 Proposed adsorption of As(III) onto MPKSF. (adapted from Ciuro Juncosa 2008; O’Reilly et al. 2001) where b = Temkin isotherm constant, A = equilibrium binding constant correspond to maximum binding energy, -1 -1 R = gas constant (8.314 J mol K ), T = absolute temperature, K. The values of constant A and B are obtained from a plot of q against ln C . e e Figure 9 shows the plot of q against ln C whereby the e e slope and intercept values obtained from the graph plot are used to calculate Temkin constant A, and the heat of sorption constant B. The R value obtained from the linear plot of Temkin isotherm model is 0.9816 indicating that the adsorption data are applicable to this model. Similar observations have been reported by Itodo and Itodo (2010) Fig. 9 Temkin isotherm showing the variation of q against ln C for e e on the adsorption of atrazine onto sheanut shell. Similarly adsorption of As(III) onto MPKSF by Hamdaoui and Naffrechoux (2007) whose study demonstrated the adsorption of phenol and chlorophenol that the heat of adsorption of the molecules present in the onto granular activated carbon. Maurya and Mittal (2006) adsorbed layer is reducing linearly with the coverage of the had also established the linear Temkin plot for the molecules instead of in logarithmic pattern due to this adsorption of methylene blue and Rhodamine B onto interaction (Temkin 1941). This means that as the coverage activated carbon. of adsorbed layer increased, the heat of adsorption Comparing the three correlation coefficients; R for the decreased. three isotherms, the Langmuir isotherm gave the best fit of Temkin isotherm is given by Eq. (13): adsorption isotherm with highest correlation coefficient, R RT value of 0.9973 followed by Temkin (0.9816) and Fre- q ¼ lnðAC Þ: ð13Þ e e undlich (0.9744) isotherm models (Table 9). Langmuir isotherm implies that adsorption of As(III) onto MPKSF The linear form of Eq. 13 is given by: adsorbent occurs in a monolayer adsorption at which when RT RT q ¼ ln A þ ln C : ð14Þ the available sorption sites of MPKSF are fully occupied, e e b b no further adsorption process can take place at those sites. Substituting RT/b with B and hence, It is corroborated by the formation of inner-sphere com- plexes between iron oxide and As(III) molecules on q ¼ B ln A þ B ln C ; ð15Þ e e MPKSF surface. 123 Appl Water Sci (2017) 7:4573–4591 4587 Fig. 10 Desorption of As(III) from MPKSF. Data obtained: n = 3 for each desorption experiment The R value for Langmuir and Freundlich are 0.9973 desorption of As(III) from MPKSF. During the first stage and 0.9744, respectively (Table 9). The Temkin isotherm of desorption using distilled water, the percentage of -1 shows that the heat of adsorption is low (9.72 J mol ) As(III) desorbed in the solution was 0.72% followed by indicating physical adsorption. Furthermore, the calculated 0.78 and 0.97% at second and third stages of desorption, -1 Freundlich (q ) (54.48 lgg ) was higher than the respectively. The small amount of As(III) detected after the max -1 adsorption capacity (q = 48.08 lgg ) determined first desorption (Des1) was presumably due to the com- max from the Langmuir isotherm. Based on the q value plexation reaction between As(III) ions with the iron in max obtained, the adsorption of As(III) is more of a physical MPKSF adsorbent (Ciuro ´ Juncosa 2008) which prevents the adsorption as described by the Freundlich isotherm. Studies dissociation of As(III) from iron oxide on the surface of by Liu et al. (2015) reported that the presence of chemical MPKSF. This is also consistent with the results of the interaction between As(III) and iron oxide forming inner- Langmuir adsorption isotherm, thus explaining the forma- sphere surface complex can be best explained by Langmuir tion of chemical bonds between As(III) and the surface of isotherm, suggesting a monolayer As(III) adsorption onto MPKSF which prevent the As(III) from desorbing easily the Fe O surface. However, they also reported that the from MPKSF. 3 4 adsorption of As(III) onto AC showed the best fit with Freundlich isotherm. Based on the experimental data, it is Identification of the proposed mechanism suggested that both chemisorption, involving the formation of the adsorption of As(III) onto the MPKSF of inner-sphere complex and physisorption on activated carbon occurred in the adsorption of As(III) onto MPKSF. Fourier transform infrared spectroscopy (FTIR) analysis Desorption of As(III) from MPKSF As stated previously a mechanism was proposed for the adsorption of As(III) onto the MPKSF, in this section, this Desorption experiment was conducted to examine the mechanism was identified using the results of the spec- reusability of the MPKSF adsorbent and the reversibility of troscopy studies and supported by the results of the char- the adsorption process. Figure 10 illustrates the percentage acterization of the MPKSF after the adsorption. Figure 11 shows the FTIR spectrum of MPKSF after adsorption of As(III) (MPKSF–As). Five peaks could be observed in the Table 9 Langmuir, Freundlich and Temkin isotherm model param- spectrum identified at wavenumber of 3430, 1625, 1387, -1 eters and correlation coefficient for adsorption of As(III) on MPKSF 1078 and 573 cm . These are assigned to the functional groups OH, C=O, C–C, C–O, and M–O, respectively. Isotherm Parameters Both spectra of MPKSF before and after adsorption -1 Langmuir q (lgg ) 48.08 max show the presence of peaks at a wavenumber of 3430, 1625 -1 K (L lg ) 0.1208 -1 and 573 cm . Additionally, in MPKSF–As, two new R 0.9973 -1 -1 peaks were detected at 1387 cm and around 800 cm Freundlich K 10.86 which are assigned to C–C bending (Mayo et al. 2007) and 1/n 0.36 As–O interaction (Ito et al. 1995), respectively, on MPKSF -1 q (lgg ) (calculated) 54.48 max surface after adsorption. However, the spectrum showed a -1 R 0.9744 less significant band at 800 cm . Thus the EDX analysis -1 Temkin A (L g ) 1.436 data are used to validate the presence of As(III). According -1 B (J mol ) 9.72 to Sayle (2000), at a pH lower than the pKa value, the R 0.9816 molecules will be mostly protonated. Thus, the negatively 123 4588 Appl Water Sci (2017) 7:4573–4591 As-O 1078.32cm-1 26 1078.32 3914.78cm-1 1387.99cm-1 24 1387.99 3698.94cm-1 635.83cm-1 57 573. 3.35 3cm- 5 1 22 MPKSF-As C-C 1625.71cm-1 1625.71 3780.52cm-1 1093.92cm-1 445.06cm-1 MPKS 3430.14cm-1 3430.14 12 1629.52cm-1 636.19cm-1 579.58cm-1 3435.28cm-1 C=O C-O M-O OH 4000 3500 3000 2500 2000 1500 1000 500 400 cm-1 Fig. 11 FTIR spectrum of MPKSF and MPKSF–As Fig. 12 XRD diffractogram of MPKSF–As charged MPKSF provides greater affinity towards the of Mandal et al. (2013) on the adsorption of As(III) by protonated As(III) ions at pH lower than 9.2 which result in zirconium polyacrylamide hybrid material. However, no the formation of inner-sphere complexes between As(III) crystalline As(III) was detected using this analysis, prob- and Fe O in MPKSF. Additionally, Bundschuh et al. ably due to the small amount of As(III) adsorbed on 3 4 (2005) reported that as the pH increases, the dominant MPKSF. negative charges are present on adsorbent surface hence interference with the adsorption of As(III) and As(V) are Scanning electron microscopy (SEM) analysis significantly governed by the surface charge of adsorbent. The morphology of MPKSF–As under magnification of X-ray diffraction (XRD) analysis 15509 is depicted in Fig. 13a illustrating that MPKSF–As surface experienced a distinct change as compared to Figure 12 illustrates the diffractogram pattern of MPKSF MPKSF. In addition, the presence of a layer of coating on after adsorption (MPKSF–As). The pattern obtained is the surface is presumably assigned to the adsorbed layer of quite similar to the diffractogram of MPKSF before the As(III) complexes on the MPKSF. The entire coverage of adsorption process. There are four intense peaks assigned As(III) in the pores as depicted in Fig. 13b under higher to Fe O at 2h 30.45, 35.75, 57.85 and 63.00, mean- magnification (62009) shows that the adsorption of As(III) 3 4 while one peak at 2h 43.35 is assigned to c-Fe O , occurs evenly over the surface and in the pores of MPKSF. 2 3 respectively. The increment in crystallite size of MPKSF– Further, it can be seen that As(III) particles do not block As is probably due to crystal defect after the adsorption of the external pores as it is still observed clearly even after As(III) onto the MPKSF. This is consistent with the work adsorbing As(III). With respect to both figures, it shows %T Appl Water Sci (2017) 7:4573–4591 4589 Fig. 13 Micrograph of MPKSF–As under magnification a 91550 and b 96200 Fig. 14 EDX spectrum for MPKSF–As sample that the pore size of MPKSF was affected after the Table 10 Elemental analysis for sample MPKSF and MPKSF–As adsorption process. The size of the pores reduced compared Sample MPKSF MPKSF–As to the pore size of MPKSF before adsorption. This is Weight (%) Weight (%) possibly due to the formation of the inner-sphere com- Carbon (C) 17.14 18.16 plexes of As(III) ion with Fe O in MPKSF on the walls of 3 4 Oxygen (O) 31.63 30.03 the external pores. Phosphorus (P) 1.78 3.24 Iron (Fe) 49.45 48.05 Energy dispersive X-ray analysis (EDX) analysis Arsenic (As) – 0.52 The energy dispersive X-ray analysis is used to detect the presence of arsenic after adsorption by MPKSF. Figure 14 Conclusions illustrates EDX spectrum of MPKSF–As. The spectrum showed the peaks similar to the untreated MPKSF for This study demonstrated that the MPKSF effectively carbon, oxygen, iron and phosphorus. However, after the removed As(III) from waste water by the formation of adsorption process, the presence of As(III) was detected inner-sphere complexes between the Fe and the As(III). with a composition of 0.52% as illustrated in Table 10 Further, desorption of the sorbed As(III) was found to be in below. The amount of arsenic detected is small due to the very minute concentrations, thereby suggesting that the low concentration of As used in the adsorption process sorbed As(III) was rigidly bound by inner-sphere com- -1 which is in the range of 5–100 lgL . The composition of plexation mechanism. An efficient adsorption process was other elements present in the sample was unaffected by the revealed to take place at pH 6 and 7 and at a longer contact adsorption process as the composition of these elements are time. The initial As(III) concentration and adsorbent dose quite similar to the composition of the untreated MPKSF. 123 4590 Appl Water Sci (2017) 7:4573–4591 Arsenic in Groundwater’’, 32nd international geological con- were concluded to be dependent on each other due to the gress, Florence, Italy, 18–19 August 2004 availability of adsorption binding sites for As(III) presented Cheng W et al (2016) Competitive sorption of As(V) and Cr(VI) on by greater dose of the MPKSF. Further, the higher reaction carbonaceous nanofibers. Chem Eng J 293:311–318 temperature was shown to generate more residual As(III) in Ciuro ´ Juncosa E (2008) Adsorption properties of synthetic iron oxides: as(V) adsorption on goethite (alpha-FeOOH). Dissertation. https:// the solution hence reducing the removal efficiency of scholar.google.com/scholar?q=Ciur%C3%B3Juncosa?E?%2820 As(III) by MPKSF. 08%29?Adsorption?properties?of?synthetic?iron?oxides%3A The predicted results from the designed experiment on ?As%28V%29?adsorption?on?goethite?%28alpha-FeOOH% the adsorption of As(III) onto MPKSF using Box–Behnken 29&btnG=&hl=en&as_sdt=0%2C5. Accessed 04 Sept 2017 Daud WMAW, Ali WSW (2004) Comparison on pore development of statistical data revealed that 96% of As(III) removal was activated carbon produced from palm shell and coconut shell. achieved utilizing the optimal conditions at contact time of Biores Technol 93:63–69 44.73 min, initial As(III) concentration maximized to Elizalde-Gonza ´lez M, Mattusch J, Einicke W-D, Wennrich R (2001) -1 100 lgL , adsorbent dosage of 0.29 g, initial solution pH Sorption on natural solids for arsenic removal. Chem Eng J 81:187–195 of 6.55 and the temperature at 26.38 C. A verification El-Said S, Alamri M, El-Barak A-BS, Alsogair O (2009) Adsorptive experiment conducted using a real waste water sample removal of arsenite As(III) and arsenate As(V) heavy metals resulted in a 95% As(III) removal, which indicated that from waste water using Nigella sativa L Asian. J Sci Res 0.96% error had occurred. 2:96–104 Freundlich HMF (1906) Over the adsorption in solution. J Phys Chem The adsorption studies suggest that the adsorption of 57:385–471 As(III) onto the MPKSF was highly dependent on pH and Hamdaoui O, Naffrechoux E (2007) Modeling of adsorption contact time relative to the initial As(III) concentration, isotherms of phenol and chlorophenols onto granular activated adsorbent dose and temperature. The adsorption data of the carbon: part I. Two-parameter models and equations allowing determination of thermodynamic parameters. J Hazard Mater As(III) onto the MPKSF was well described by the Lang- 147:381–394 muir isotherm followed by Temkin and Freundlich Hussein M, Tarmizi R, Zainal Z, Ibrahim R, Badri M (1996) isotherms. Preparation and characterization of active carbons from oil palm shells. Carbon 34:1447–1449 Acknowledgements This work was funded by the Universiti Ito F, Nakanaga T, Takeo H, Essig K, Jones H (1995) The FTIR Teknologi Malaysia research Grant no. 12H29. absorption spectrum of the fundamental band of the AsO radical. J Mol Spectrosc 174:417–424 Open Access This article is distributed under the terms of the Itodo A, Itodo H (2010) Sorption energies estimation using Dubinin– Creative Commons Attribution 4.0 International License (http:// Radushkevich and Temkin adsorption isotherms. Life Sci J Acta creativecommons.org/licenses/by/4.0/), which permits unrestricted Zhengzhou Univ Overseas Ed 7:31–39 use, distribution, and reproduction in any medium, provided you give Jain C, Ali I (2000) Arsenic: occurrence, toxicity and speciation appropriate credit to the original author(s) and the source, provide a techniques. Water Res 34:4304–4312 link to the Creative Commons license, and indicate if changes were Jolstera ˚ R, Gunneriusson L, Holmgren A (2012) Surface complex- made. ation modeling of Fe O –H? and Mg(II) sorption onto 3 4 maghemite and magnetite. J Colloid Interface Sci 386:260–267 Kumar A, Prasad B, Mishra IM (2008) Optimization of process Publisher’s Note Springer Nature remains neutral with regard to parameters for acrylonitrile removal by a low-cost adsorbent jurisdictional claims in published maps and institutional affiliations. using Box–Behnken design. J Hazard Mater 150:174–182. doi:10.1016/j.jhazmat.2007.09.043 Kunamneni A, Singh S (2005) Response surface optimization of enzymatic hydrolysis of maize starch for higher glucose References production. Biochem Eng J 27:179–190 Liu C-H, Chuang Y-H, Chen T-Y, Tian Y, Li H, Wang M-K, Zhang Addo Ntim S, Mitra S (2011) Removal of trace arsenic to meet W (2015) Mechanism of arsenic adsorption on magnetite drinking water standards using iron oxide coated multiwall nanoparticles from water: thermodynamic and spectroscopic carbon nanotubes. J Chem Eng Data 56:2077–2083 studies. Environ Sci Technol 49:7726–7734 Adhoum N, Monser L (2002) Removal of cyanide from aqueous Mandal S, Sahu MK, Patel RK (2013) Adsorption studies of arsenic solution using impregnated activated carbon. Chem Eng Process (III) removal from water by zirconium polyacrylamide hybrid 41:17–21 material (ZrPACM-43). Water Resour Ind 4:51–67 Anyika C, Asri NAM, Majid ZA, Yahya A, Jaafar J (2017) Synthesis Maurya NS, Mittal AK (2006) Applicability of equilibrium isotherm and characterization of magnetic activatedcarbon developed models for the biosorptive uptakes in comparison to activated from palm kernel shells Nanotechnology for Environmental carbon-based adsorption. J Environ Eng 132:1589–1599 Engineering 2:16. doi:10.1007/s41204-017-0027-6 Mayo J et al (2007) The effect of nanocrystalline magnetite size on Budinova T, Petrov N, Razvigorova M, Parra J, Galiatsatou P (2006) arsenic removal. Sci Technol Adv Mater 8:71–75 Removal of arsenic (III) from aqueous solution by activated Milonjic ´ S, Kopec ˇni M, Ilic ´ Z (1983) The point of zero charge and carbons prepared from solvent extracted olive pulp and olive adsorption properties of natural magnetite. J Radioanal Nucl stones. Ind Eng Chem Res 45:1896–1901 Chem 78:15–24 Bundschuh J, Bhattacharya P, Chandrashekharam D (2005) Natural Nakahira A, Nishida S, Fukunishi K (2006) Adsorption of contam- arsenic in groundwater: occurrences, remediation and manage- inants by MACs and magnetic separation. J Ceram Soc ment. In: Proceedings of the pre-congress workshop ‘‘Natural 114:135–137 123 Appl Water Sci (2017) 7:4573–4591 4591 Okeola F, Odebunmi E (2010) Freundlich and Langmuir isotherms Shehzad A, Bashir MJ, Sethupathi S, Lim J-W (2016) An insight into parameters for adsorption of methylene blue by activated carbon the remediation of highly contaminated landfill leachate using derived from agrowastes. Adv Nat Appl Sci 4:281–288 sea mango based activated bio-char: optimization, isothermal O’Reilly SE, Strawn DG, Sparks DL (2001) Residence time effects on and kinetic studies. Desalin Water Treat 57:22244–22257 arsenate adsorption/desorption mechanisms on goethite. Soil Sci Tan I, Ahmad A, Hameed B (2008) Preparation of activated carbon Soc Am J 65:67–77. doi:10.2136/sssaj2001.65167x from coconut husk: optimization study on removal of 2,4,6- Pattanayak J, Mondal K, Mathew S, Lalvani S (2000) A parametric trichlorophenol using response surface methodology. J Hazard evaluation of the removal of As(V) and As(III) by carbon-based Mater 153:709–717 adsorbents. Carbon 38:589–596 Temkin MI (1941) Adsorption equilibrium and the kinetics of Payne KB, Abdel-Fattah TM (2005) Adsorption of arsenate and processes on nonhomogeneous surfaces and in the interaction arsenite by iron-treated activated carbon and zeolites: effects of between adsorbed molecules. Zh Fiz Chim 15:296–332 ˇ ˇ pH, temperature, and ionic strength. J Environ Sci Health Trakal L, Veselska ´ V, Safar ˇ´ ıkI,V´ıtkova ´ M, C´ ıhalova ´ S, Koma ´rek M 40:723–749 (2016) Lead and cadmium sorption mechanisms on magnetically Ribeiro MA, Goncalves AR, Rocha GJM, Quintana G (2006) modified biochars. Biores Technol 203:318–324 Cadmium and nickel removal from water by using activated Vance DB (2002) Arsenic chemical behavior and treatment. Internet carbon obtained from sugarcane lignin. Reunion Iberica de WHO (1981) Arsenic, environmental health criteria 18. World Health Adsorcion 116:14 Organization, Geneva Samsuri AW, Sadegh-Zadeh F, Seh-Bardan BJ (2013) Adsorption of Xu J, Gao N, Tang Y, Deng Y, Sui M (2010) Perchlorate removal As(III) and As(V) by Fe coated biochars and biochars produced using granular activated carbon supported iron compounds: from empty fruit bunch and rice husk. J Environ Chem Eng synthesis, characterization and reactivity. J Environ Sci 1:981–988 22:1807–1813 Sayle R (2000) Physiological ionization and pKa prediction. Euro Yakout SM, Sharaf El-Deen G (2016) Characterization of activated MUG. https://scholar.google.com/scholar?hl=en&q=Sayle%2C? carbon prepared by phosphoric acid activation of olive stones. R?Physiological?ionization?and?pKa?prediction?Metaphorics Arab J Chem 9:S1155–S1162. doi:10.1016/j.arabjc.2011.12.002 ?LLC%3A?2000?http%3A%2F%2Fwwwdaylightcom%2F Yao S, Liu Z, Shi Z (2014) Arsenic removal from aqueous solutions meetings%2Femug00%2FSayle%2Fpkapredicthtml&btnG=&as_ by adsorption onto iron oxide/activated carbon magnetic com- 20. Accessed 04 Sept 2017 posite. J Environ Health Sci Eng 12:1 Schwertmann U, Murad E (1983) Effect of pH on the formation of Zou Y et al (2016) Environmental remediation and application of goethite and hematite from ferrihydrite. Clays Clay Miner nanoscale zero-valent iron and its composites for the removal of 31:277–284 heavy metal ions: a review. Environ Sci Technol 50:7290–7304 Segurola J, Allen NS, Edge M, Mc Mahon A (1999) Design of eutectic photoinitiator blends for UV/visible curable acrylated printing inks and coatings. Prog Org Coat 37:23–37 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Water Science Springer Journals

Batch sorption–desorption of As(III) from waste water by magnetic palm kernel shell activated carbon using optimized Box–Behnken design

Download PDF
19 pages

Loading next page...
1
 
/lp/springer_journal/batch-sorption-desorption-of-as-iii-from-waste-water-by-magnetic-palm-QZFIauT0eR

References (54)

Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Earth Sciences; Hydrogeology; Water Industry/Water Technologies; Industrial and Production Engineering; Waste Water Technology / Water Pollution Control / Water Management / Aquatic Pollution; Nanotechnology; Private International Law, International & Foreign Law, Comparative Law
ISSN
2190-5487
eISSN
2190-5495
DOI
10.1007/s13201-017-0610-9
Publisher site
See Article on Publisher Site

Abstract

Appl Water Sci (2017) 7:4573–4591 https://doi.org/10.1007/s13201-017-0610-9 O R I G IN AL ARTI CL E Batch sorption–desorption of As(III) from waste water by magnetic palm kernel shell activated carbon using optimized Box–Behnken design 2 1 1 • • • Chinedum Anyika Nur Asilayana Mohd Asri Zaiton Abdul Majid 1 2 Jafariah Jaafar Adibah Yahya Received: 25 March 2017 / Accepted: 29 August 2017 / Published online: 8 September 2017 The Author(s) 2017. This article is an open access publication Abstract In this study, we converted activated carbon the MPKSF functional groups and As(III). The findings (AC) into magnetic activated carbon (MAC), which was suggested that the MPKSF exhibited a strong capacity to established to have removed arsenic (III) from wastewater. efficiently remove As(III) from wastewater, while the Arsenic (III) is a toxic heavy metal which is readily soluble desorption studies showed that the As(III) was rigidly in water and can be detrimental to human health. The MAC bound to the MPKSF thereby eliminating the possibility of was prepared by incorporating Fe O into the AC by using secondary pollution. 3 4 Fe O extracted from a ferrous sulfate solution, designated: 3 4 magnetic palm kernel shell from iron suspension Keywords As(III)  Sorption  Desorption  Waste water (MPKSF). Batch experiments were conducted using two treatment  Magnetic palm kernel shell activated carbon methods: (1) one-factor-at-a-time and (2) Box–Behnken statistical analysis. Results showed that the optimum con- ditions resulted in 95% of As(III) removal in the wastew- Introduction ater sample. The adsorption data were best fitted to the Langmuir isotherm. The adsorption of As(III) onto the Removal of contaminants such as arsenic (As) from MPKSF was confirmed by energy dispersive X-ray spec- wastewater by adsorption mechanisms remains the most trometry analysis which detected the presence of As(III) of effective method (Elizalde-Gonzalez et al. 2001). Most 0.52% on the surface of the MPKSF. The Fourier transform studies in the field of adsorption for the removal of heavy infrared spectroscopy analysis of the MPKSF–As presented metals from water have mainly focused on the use of AC, -1 a peak at 573 cm , which was assigned to M–O (metal– activated alumina, sand impregnated with iron, polymer oxygen) bending, indicating the coordination of As(III) resins, hydrous ferric oxide and natural ores (Addo Ntim with oxygen through the formation of inner-sphere com- and Mitra 2011). Although AC has been found to be more plexation, thereby indicating a covalent bonding between effective relative to the other adsorbents mentioned above, especially for the removal of heavy metals from aqueous solutions, with percentage removal ranging from 82 to 96% (Ribeiro et al. 2006), this technique may not be adequate when it comes to a heavy metal like As(III) which is known to be highly soluble in water. Arsenic exists in two dif- ferent oxidation states (1) arsenite, As(III) and (2) arsenate, & Zaiton Abdul Majid As(V). As(III) is different from As(V) in a number of ways. zaiton@kimia.fs.utm.my; zaitonmajid@utm.my Firstly, difficulties arise when it comes to the removal of As(III) from wastewater compared to arsenate As(V) due to Department of Chemistry, Faculty of Science, Universiti its high solubility in the aqueous environment, hence Teknologi, Malaysia, 81310 Johor Bahru, Malaysia As(III) which is the most toxic is also the most difficult to Environmental Biotechnology Laboratory, Faculty of remove from water (Pattanayak et al. 2000). Secondly, Biosciences and Medical Engineering, Universiti Teknologi, whereas As(V) is mostly removed by outer-sphere Malaysia, 81310 Johor Bahru, Malaysia 123 4574 Appl Water Sci (2017) 7:4573–4591 complexation, i.e., by electrostatic attraction (Cheng et al. sorption–desorption of As(III) onto the prepared magnetic 2016), As(III) can be removed by inner-sphere complexa- activated carbon. tion, which is a covalent bonding as demonstrated in this study. On the other hand, the long-term effects of drinking Experimental water contaminated with arsenic include cancer of the skin, lung, bladder, and kidney, skin thickening, neurological Chemicals and reagents disorders, muscular weakness and nausea (Jain and Ali 2000;WHO 1981). This has led to numerous studies Phosphoric acid (H PO , 85 wt%, Merck, Germany) was 3 4 regarding the improvement of AC by magnetic modifica- used to pretreat and impregnate the raw materials. The tion to increase its capacity to remove heavy metal con- chemical reagents for the preparation of magnetic activated taminants. The MAC adsorbents exhibited magnetic carbon were of three kinds: (1) iron (III) chloride ([96%): properties with greater efficiency for the adsorption of Sigma-Aldrich; (2) iron (II) sulfate (99.5%), Qrec; (3) contaminants from aqueous solution (Xu et al. 2010). sodium hydroxide, Merck Germany. A 1000 ppm stock Whereas palm kernel shells were used in the production of solution of arsenite, As(III), was prepared in double dis- AC due to its high carbon content and low organic content tilled water from 0.05 M sodium arsenite (NaAsO ) pre- (Daud and Ali 2004), while at the same time high quality pared by Fluka. Hydrochloric acid (HCl) and sodium of AC can be synthesized from PKS waste (Adhoum and hydroxide (NaOH) were used to adjust the solution pH of Monser 2002; Budinova et al. 2006; Hussein et al. 1996). As(III). One major advantage of the MAC is that it exhibits magnetic characteristics, in addition to having demon- Experimental methods strated to be effective for adsorption in dilute solutions, while its high specific surface area due to the presence of Material development and characterization microporous structure results in greater capacity for the adsorption of heavy metals (Nakahira et al. 2006). So far, The raw PKSs (100 g) used in this study were obtained magnetization has been identified as being potentially from a palm oil estate located at Jalan Sawah, Pekan important in improving the sorption characteristics of Nenas, Johor Bahru, Malaysia. Sample pretreatment was organic adsorbents with well-developed structures, i.e., of carried out by weighing out a 50-g portion of the raw PKS, woody feedstock such as PKS (Trakal et al. 2016). How- which was ground and sieved to particle sizes in the range ever, previous studies on the removal of As(III) from of 75–250 lm, then soaked and impregnated with 10 mL wastewater employed the use of biochar made from empty of 30 wt% of phosphoric acid, H PO , at room temperature 3 4 fruit bunch and rice husks (Samsuri et al. 2013), which are with PKS-to-acid ratio of 1:1. The sample was left fibrous in nature, unlike the PKS feedstock used in this impregnated for 24 h, and afterwards it was washed with study, which is a woody feedstock. Moreover, they used distilled water and dried at room temperature. biochar which is a different material. Another problem with A 10-g portion of the pre-treated PKSs was transferred their study was that they did not carry out a desorption test into five conical flasks containing different concentrations to ascertain the stability of the Fe-coated organic adsor- of H PO , i.e., 10, 20, 30, 40 and 50% wt/wt, respec- 3 4 bents. Further, their study was on competitive adsorption tively. Similarly, the pre-treated PKS was treated again between As(III) and As(V). Similarly, in the paper by using dilute acid (10, 20 and 30% wt/wt H PO ). This 3 4 Payne and Abdel-Fattah (2005) using Fe-coated AC for the was done to determine the appropriate surface area at removal of As(III) and As(V) from water, they relied on which if the surface area obtained by using dilute acid commercially procured AC, which had a poorly developed was high, then the use of concentrated acid can be structure and hence was only able to remove 60% As(III) reduced and the experimental costs will be less. This ratio from waste water unlike in this study where 95% removal (1:1) implied that 10 g of the raw PKSs was soaked in was achieved by using MPKS which has a well-developed 10 g of H PO for 24 h. The excess acid was then filtered 3 4 structure. In addition, they did not look at the magnetic and the soaked PKSs were placed in a muffle furnace and properties of the Fe-coated AC. heated at 200 C for 30 min to initiate the carbonization Therefore, the influence of the MAC on the removal of process. As(III) from waste water has not been completely eluci- Subsequently, the temperature of the furnace was dated. The objectives of this study are: (1) to prepare increased to the range of 400–550 C and held for 2 h magnetic activated carbon for the removal of arsenic from followed by cooling to room temperature. Afterwards, the water; (2) to optimize the parameters for As(III) sorption samples were thoroughly washed and rinsed using vacuum using Box–Behnken design; (3) to study the filtration with hot distilled water to remove all the excess 123 Appl Water Sci (2017) 7:4573–4591 4575 acid until the pH of the filtrate was approximately 7. The under pH 6 and 7, (5) effect of temperature, in this case, a samples were then dried in the oven at a temperature of temperature of 30 C was considered preferable for the 110 C for 24 h. The AC samples were then stored in adsorption of the As(III) onto the MPKSF. desiccators for further characterization and adsorption The residual arsenic solution was analyzed using gra- studies. The preparation of the MPKSF was achieved by phite furnace atomic absorption spectrometry (GFAAS). utilizing a suspension of ferric chloride/ferrous sulfate. The And the same procedures were repeated using a real water characterization experiments were conducted on the waste sample from Skudai River, Johor Bahru Malaysia, MPKSF by Fourier transform infrared spectroscopy spiked with arsenite to analyze the percentage of arsenite (FTIR), X-ray diffraction (XRD), particle size analysis, removal using MPKSF. nitrogen adsorption analysis, scanning electron microscopy (SEM), field emission scanning electron microscopy Determination of pHpzc (point of zero charge) of the (FESEM), energy dispersive X-ray spectrometry (EDX) samples The pHpzc of both PKSAC and MPKSF was and the point of zero charge (pHpzc) was also determined determined. A 50-mL solution of 0.01 M NaCl was placed as part of the characterization. The magnetic saturation of in a closed Erlenmeyer flask. The pH of the solution was the MAC sample was characterized using vibrating sample adjusted to achieve a suspension pH of between 2 and 12 magnetometer (VSM). by adding 0.1 M HCl or 0.1 M NaOH solutions in ten conical flasks. Approximately 0.15 g each of the PKSAC Batch experimental procedure and testing methods and MPKSF were added and the final pH was measured after 48 h. The pH of each solution recorded was plotted. Batch experiments were carried out by grinding the The intersection of pH and pH of the solution was initial final MPKSF into fine powder of 75 lm particle sizes. The stock then taken as the pHpzc. solution was prepared by diluting 0.05 M of sodium arsenite (NaAsO ) with distilled water up to a concentra- Desorption procedure of magnetic activated carbon tion of 1000 ppm to give a 1000 ppm arsenite (As(III)) (MPKSF) The optimum amount of MPKSF loaded with -1 stock solution. The pH of the solution was then adjusted arsenite (48.08 lgg ) obtained after the adsorption pro- using hydrochloric acid (HCl) until it reached a pH of 7. A cess was then added into a 50 mL of distilled water in a 0.05–0.30 g portion of the powdered MPKSF was placed centrifuge tube. The solution was shaken at 150 rpm and into a conical flask together with 200 mL arsenite solution. then agitated at specific time intervals for up to 48 h. The The solution was then shaken for 24 h at different tem- solution was then centrifuged and the supernatant was peratures ranging from 10 to 40 C. The contact time in the collected for further analysis to examine the concentration shaker was also varied ranging from 5 to 720 min. The of As(III) desorbed from the MPKSF. The desorption initial concentration of the arsenite solution was varied procedure was repeated three times and the MPKSF -1 ranging from 5 to 100 lgL . This was done to mimic the adsorbent was reserved for further analysis. drinking water standard. The preliminary experiment was conducted at an equilibrium time of 180 min. After that, Graphite furnace atomic absorption spectrometry the suspension was filtered through 0.45-lm pore size (GFAAS) The detection and the concentration of arsen- membrane filter. ite were conducted by a graphite furnace atomic absorp- The preliminary study on the adsorption of As(III) on tion spectrometry (GFAAS). The samples were analyzed MPKSF, was divided into 6 factors and are described in triplicates, to obtain the optimum result of -1 briefly namely (1) effect of contact time, which was found 10–40 lgL . A small amount of the sample which was to be 180 min and established to be the time taken to reach in the range of 20–100 lL was placed into the graphite adsorption–desorption equilibrium; (2) effect of initial tube manually. Further, the arsenite samples were acidi- -1 As(III) concentration, which was found to be 5–70 lgL fied with nitric acid to a pH of less than 2. Upon injection with percentage removal of 87.58–89.42%; (3) effect of of the samples into the graphite tube, they are vaporized. adsorbent dosage, was found to be 0.3 g, which resulted in Subsequently, the amount of light energy absorbed in the an increase in percentage removal of As(III) from 42.17 to vapor was considered to be proportional to the atomic 96.58%. This was due to a higher dose of adsorbent utilized concentration. which provides greater proportion of adsorption sites for As(III) to bind on the MPKSF surface (Yao et al. 2014); (4) Box–Behnken design effect of pH, in this case, the effect of the initial solution pH in the range of 6–8 on the adsorption of As(III) onto the In this study, the Box–Behnken design was used to opti- MPKSF, indicated that the removal of As(III) was better mize the number of experiments to be conducted with the 123 4576 Appl Water Sci (2017) 7:4573–4591 aim of determining the probable interactions between the Results and discussion parameters under study and their influence on the adsorp- tion of As(III) onto the MPKSF. Further characteristics of Characterization of adsorbents the Box–Behnken design has been described elsewhere (Kumar et al. 2008). The characterization results (detailed characterization of In Table 1, the experimental parameters and a the adsorbent (s) has been described elsewhere, Anyika 3-level 5-factorial Box–Behnken experimental design et al. 2017) revealed that the MPKSF presented better are presented. This was applied to examine and validate characteristics and was therefore selected as the sole the adsorption system parameters influencing the adsorbent for the adsorption studies. It presented a higher 2 -1 adsorption of As(III) onto the MPKSF. Contact time BET surface area of 257 m g , higher pore volume of -1 (5–720 min), pH (6–8), adsorbent dose (0.05–0.30 g), 0.1124 cc g and higher magnetic properties with a -1 -1 initial As(III) concentration (5–100 lgL ), tempera- magnetic saturation of 49.55 emu g . The FTIR spectrum -1 ture (20–40 C) represent the variable input parameters. of the MPKSF exhibited intense OH bending at 1629 cm The factor levels (3) were coded as (-1 = low, which can be attributed to the presence of oxygen in the -1 0 = medium level or central point, and ?1 = high samples, while the absorption bands at 1093 and 579 cm level). Response surface methodology (RSM) was indicated the presence of C–O stretching and metal–oxy- applied to the experimental data using the Design gen (M–O) bands due to the interaction of iron and oxygen. Expert statistical software version 7.1.6 by Stat-Ease, To illustrate that the MPKSF acquired magnetic properties, Inc., Minneapolis, USA. the XRD data were analyzed. The MPKSF exhibited the The regression equation of the designed experiment was presence of Fe O at 2h 30.75, 35.95, 57.35, 63.20 3 4 obtained by applying four models namely linear, interac- from the XRD diffractogram. Similarly, three peaks at 2h tive, quadratic and cubic models which were fitted to the 19.30, 43.45, 54.10 and a peak at 2h 24.40 which can experimental data obtained from the design system. be assigned to c-Fe O and a-Fe O , respectively, were 2 3 2 3 To select the best model, i.e., after the responses have detected. been recorded, the data were analyzed using three different The point of zero charge of the MPKSF was determined tests in order to decide the adequacy of the models stated to explain the surface charge phenomena as well as the above to represent the adsorption process of MPKSF–As. magnetic properties of the MPKSF. With respect to the These validation tests are the sequential model sum of surface charge, the point of zero charge of MPKSF squares (F test), lack-of-fit test and the model summary occurred at pH 5.94. Since As(III) is positively charged, it statistics. Further, a quadratic polynomial was used to is critical for the surface of the MPKSF to be negatively explain the relationship between the parameters and As(III) charged in order for the adsorption process to occur. As the residual concentration (%). solution pH is higher than 5.94, the surface of the MPKSF The second-order polynomial is represented by Eq. (1): exhibits greater formation of hydroxide ions. However, at pH 6 and 7, the MPKSF demonstrated greater adsorption k k k X X X ¼ b þ b x þ b x þ b x x þ e; ð1Þ compared to pH 8. This was presumed to have resulted i i j 0 i ii i ij i1 i1 1s s i j from the formation of As(III) precipitate at a higher pH hence reducing the adsorption efficiency. Additionally, at where the terms have their established meanings (Ku- higher pH, As(III) has the potential to be oxidized to namneni and Singh 2005; Shehzad et al. 2016). A design of As(V), which may significantly reduce the adsorption of 46 tests was formulated. As(III) onto the MPKSF (Vance 2002). In Table 2, the Table 1 Independent parameters and their levels used for Box–Behnken design Parameters, unit Factors Levels -10 1 Contact time (min) A 5 362.5 720 pH B 67 8 Adsorbent dosage (g) C 0.05 0.17 0.30 -1 Initial As concentration (lgL ) D 5 52.5 100 Temperature (C) E 20 30 40 123 Appl Water Sci (2017) 7:4573–4591 4577 Table 2 Initial and final pH value of PKSAC and MPKSF reaction MPKSF, i.e., pHpzc = 5.94, its surface becomes negatively solutions charged, hence the protonated As(III) will have a greater affinity towards the MPKSF surface. Based on the Box– Initial pH Final pH Behnken optimization, the optimum adsorption occurred at PKSAC MPKSF pH 6.55, which conformed to the aquatic environmental 4.00 3.94 4.26 range of pH of 5–9 (Zou et al. 2016). 6.00 4.22 5.94 As seen in Fig. 1b, the pHpzc value of MPKSF decreased its acidity to attain a pH of 5.94. This may be 8.00 4.42 5.31 10.00 4.29 6.54 due to the iron oxide extracted from the ferric chlo- ride/ferrous sulfate solution (FeOF) which was used in the 12.00 11.41 11.59 production of MPKSF. To further illustrate that the Point of zero charge MPKSF acquired magnetic properties, the hydration of Fe O in aqueous solution resulted in the formation of a- 3 4 initial pH and final pH of the reaction solutions for both Fe O in an acidic condition as depicted by the reaction in 2 3 adsorbent samples, PKSAC and MPKSF are presented. Eq. (2). To illustrate that a-Fe O was formed, previous 2 3 In Fig. 1a, b, a plot of pH versus pH for adsorbent final initial studies have reported that the pHpzc for the untreated PKSAC and MPKSF is presented. From the graph, the point Fe O was 6.5 while for c-Fe O , it was at pH 5.9 3 4 2 3 of zero charge (pHpzc) of the sample represents the point (Milonjic et al. 1983) which indicate that the value of pH where the plot of final pH versus initial pH intersects with for both was nearly acidic even in the untreated condition. the line at which the final pH equals to the initial pH. The Further, Schwertmann and Murad (1983) had reported that blue line in both graphs indicates the line of pH = - final a-Fe O is predominantly formed at pH 7–8. In this study, 2 3 pH , while the red and green curves indicate the plots of initial it was demonstrated that upon impregnation of the PKSAC pH against pH for PKSAC and MPKSF, respec- final initial with FeOF, the acidity of the modified PKSAC (MPKSF) tively. Figure 1a shows that the pHpzc of PKSAC adsorbent was reduced from 3.94 to 5.94 due to the formation of a- was 3.94, which indicated that the sample was acidic, due to Fe O from the reaction of Fe O with water as represented 2 3 3 4 the impregnation of PKSAC using phosphoric acid by the Eq. (2) below: (H PO ). A suitable acid activation results in the production 3 4 2Fe O þ H O  3Fe O þ 2H þ 2e : ð2Þ 3 4 2 2 3 of high quality and high surface area AC (Yakout and Sharaf El-Deen 2016). In the adsorption process of As(III) Box–Behnken statistical analysis onto MPKSF, at a pH above the point of zero charge of the In Table 3, the most important parameters influencing the efficiency of adsorption of As(III) onto the MPKSF are represented by the letters: A, B, C, D and E, which repre- sents the coded symbols for the respective factors: contact time, pH, adsorbent dosage, initial As(III) concentration and the temperature parameters. The combined effects of these factors were evaluated by performing experiments on the different combinations of these parameters. The applied Box–Behnken model can be expressed as Eq. (3): Y ¼ X þ X A þ X B þ X C þ X D þ X E þ X AB 0 1 2 3 4 5 6 þ X AC þ X AD þ X AE þ X BC þ X BD 7 8 9 10 11 þ X BE þ X CD þ X CE þ X DE þ X A 12 13 14 15 16 2 2 2 2 þ X B þ X C þ X D þ X E ; ð3Þ 17 18 19 20 where Y is the response, X and X depicted the global mean 0 i and other regression coefficients, respectively, while A, B, C, D and E are the coded symbols for the respective fac- tors: contact time, pH, adsorbent dosage, initial As(III) concentration and the temperature parameters. Fig. 1 a Graph of pH versus pH for the adsorbent PKSAC final initial In Table 4, the statistical significance of the ratio of suspension. b Graph of pH versus pH graph for the adsorbent final initial mean square variation due to regression and mean square MPKSF suspension 123 4578 Appl Water Sci (2017) 7:4573–4591 Table 3 Experimental, actual and predicted values of Y for As(III) onto MPKSF Standard run order AB C D E Actual value Predicted value 1 5 6 0.17 52.5 30 86.93 86.91 2 720 6 0.17 52.5 30 99.81 99.80 3 5 8 0.17 52.5 30 81.67 81.65 4 720 8 0.17 52.5 30 93.96 93.94 5 362.50 7 0.05 5 30 57.41 57.43 6 362.50 7 0.3 5 30 51.48 51.50 7 362.50 7 0.05 100 30 18.24 18.22 8 362.50 7 0.3 100 30 92.46 92.44 9 362.50 6 0.17 52.5 20 91.63 91.61 10 362.50 8 0.17 52.5 20 88.46 88.47 11 362.50 6 0.17 52.5 40 96.80 96.78 12 362.50 8 0.17 52.5 40 88.80 88.81 13 5 7 0.05 52.5 30 28.03 28.05 14 720 7 0.05 52.5 30 70.81 70.81 15 5 7 0.3 52.5 30 92.36 92.37 16 720 7 0.3 52.5 30 74.79 74.78 17 362.50 7 0.17 5 20 79.93 79.91 18 362.50 7 0.17 100 20 83.05 83.05 19 362.50 7 0.17 5 40 84.96 84.94 20 362.50 7 0.17 100 40 83.52 83.53 21 362.50 6 0.05 52.5 30 63.35 63.36 22 362.50 8 0.05 52.5 30 29.41 29.41 23 362.50 6 0.3 52.5 30 69.10 69.11 24 362.50 8 0.3 52.5 30 91.95 91.95 25 5 7 0.17 5 30 78.30 78.28 26 720 7 0.17 5 30 84.89 84.89 27 5 7 0.17 100 30 73.18 73.17 28 720 7 0.17 100 30 91.72 91.73 29 362.50 7 0.05 52.5 20 35.69 35.68 30 362.50 7 0.3 52.5 20 96.26 96.26 31 362.50 7 0.05 52.5 40 64.88 64.86 32 362.50 7 0.3 52.5 40 72.58 72.57 33 5 7 0.17 52.5 20 87.86 87.88 34 720 7 0.17 52.5 20 98.26 98.28 35 5 7 0.17 52.5 40 88.42 88.45 36 720 7 0.17 52.5 40 103.20 103.22 37 362.50 6 0.17 5 30 81.57 81.59 38 362.50 8 0.17 5 30 75.49 75.50 39 362.50 6 0.17 100 30 81.90 81.92 40 362.50 8 0.17 100 30 76.89 76.90 41 362.50 7 0.17 52.5 30 85.53 85.53 42 362.50 7 0.17 52.5 30 85.52 85.53 43 362.50 7 0.17 52.5 30 85.51 85.53 44 362.50 7 0.17 52.5 30 85.54 85.53 45 362.50 7 0.17 52.5 30 85.52 85.53 46 362.50 7 0.17 52.5 30 85.53 85.53 123 Appl Water Sci (2017) 7:4573–4591 4579 Table 4 ANOVA for response surface quadratic model (Y) Source Sum of squares Df Mean square F value p value (p [ F) Model 6726.26 20 335.36 99.63 \0.0001 A: contact time 218.30 1 218.30 64.85 \0.0001 B: pH 32.86 1 32.86 9.76 0.0045 C: adsorbent dosage 2084.38 1 2084.38 619.20 \0.0001 D: initial conc. As 2242.50 1 2242.50 666.17 \0.0001 E: temperature 5.87 1 5.87 1.74 0.1987 AB 0.03 1 0.03 9.098E-003 0.9248 AC 248.38 1 248.38 73.78 \0.0001 AD 40.51 1 40.51 12.04 0.0019 AE 2.89 1 2.89 0.86 0.3630 BC 221.71 1 221.71 65.86 \0.0001 BD 4.49 1 4.49 1.34 0.2588 BE 1.59 1 1.59 0.47 0.4986 CD 732.51 1 732.51 217.61 \0.0001 CE 190.85 1 190.85 56.70 \0.0001 DE 0.26 1 0.26 0.08 0.7833 A 0.34 1 0.34 0.10 0.7516 B 1.85 1 1.85 0.55 0.4654 C 493.91 1 493.91 146.73 \0.0001 D 120.20 1 120.20 35.71 \0.0001 E 6.70 1 6.70 1.99 0.1706 Residual 84.16 25 3.37 Lack of fit 84.11 20 48.21 4.79 0.0687 Pure error 0.04 5 Corr. total 7154.40 45 residual error was tested using ANOVA. The results of the (3) model summary statistics. The results showed that the ANOVA indicated that the F values obtained for all the p value for majority of the regression were \0.05. This regressions were higher, which indicated that that majority implied that one of the terms in the regression equation was of the variation in the response can be explained by the significantly correlated to the response variable. Further, regression equation (Kumar et al. 2008). To determine the quadratic model was found to yield the best fit of R , 2 2 whether the F is large enough to result in a statistical Adjusted R and predicted R values of 0.9876, 0.9777 and significance, the p value is examined. In this case, the 0.9505, respectively (Table 5). This also implied that the model is considered to be statistically significant if the model does not explain 1% of the experimental results. values under the column p [ F value is \0.05 (Table 4) Again, the high R values in Table 5 and the p value of (Segurola et al. 1999). The ANOVA result for the MPKSF– \0.0001 in Table 4, indicate that the quadratic polynomial As design system shows that the F value of 99.63 and its was highly significant in explaining the relationship p value of \0.05 imply that the model was significant between the parameters and As(III) residual concentration towards the response. Hence in this analysis, A, B, C, D, (%). 2 2 AC, AD, BC, CD, CE, C and D were the significant terms In Table 5, since the cubic model was established to be (Table 4). Besides, the ANOVA results for the MPKSF–As aliased, the quadratic model was therefore, chosen to be adsorption system showed that the F value is 99.63 used for further analysis. Further, under the lack of fit (Table 4), indicating that the terms on the model are having (Table 5) the F value is not significant, with an F value of a significant effect on the response. 4.79 and the p value of 0.0687. This shows that the lack of In Table 5, the adequacy of the model for the adsorption fit was not significant relative to the pure error. However, of As(III) onto the MPKSF was determined by three tests: the lack-of-fit value indicates that there is 6.87% possibility (1) sequential model sum of squares; (2) lack-of-fit tests; that the error resulted from noise. Thus, the non-significant 123 4580 Appl Water Sci (2017) 7:4573–4591 Table 5 Adequacy of the model tested for MPKSF–As design system Source Sum of squares Df Mean square Fp value (p [ F) Remark Sequential model sum of squares Mean vs total 5863.8 1 5863.8 – – – Linear vs mean 4583.91 5 916.78 16.61 \0.0001 – 2FI vs linear 1443.23 10 144.32 5.67 \0.0001 Quadratic vs 2FI 680.13 5 136.03 40.41 \0.0001 Suggested Cubic vs quadratic 59.05 15 3.94 1.57 0.2386 Aliased Residual 25.10 10 2.51 – – – Total 12,655.22 46 275.11 – – – Lack-of-fit tests Linear 2207.47 35 63.07 7345.18 \0.0001 – 2FI 764.24 25 30.57 3560.12 \0.0001 – Quadratic 84.11 20 48.21 4.79 0.0687 Suggested Cubic 25.06 5 5.01 583.72 \0.0001 Aliased Pure error 0.043 5 8.587E-003 – – 2 2 2 Source Std. dev. R Adjusted R Predicted R PRESS Remark Model summary statistics Linear 7.43 0.6750 0.6343 0.5575 3005.16 – 2FI 5.05 0.8875 0.8312 0.7113 1960.64 – Quadratic 1.83 0.9876 0.9777 0.9505 336.51 Suggested Cubic 1.58 0.9963 0.9834 0.7638 1603.97 Aliased value for the lack of fit showed that the model was valid for shows the antagonistic effect on the adsorption process further analysis. The final mathematical equation given by (Tan et al. 2008). Further, the positive value of the model the Box–Behnken design in terms of actual values deter- term indicates the effect that favors the optimization mined by Design Expert software is presented in Eq. (4). process while the negative model term value represents the Removal of As(III) (%) is given by: inverse interaction between the parameters with the response. Y ¼ 85:53 þ 6:29  A  2:78  B þ 17:07  C þ 0:43  D þ 1:38  E  0:15  A  B Adsorption studies 15:09  A  C þ 2:99  A  D þ 1:09  A  E þ 14:20  B  C þ 0:27  B  D  1:21  B  E In this study, the Box–Behnken design was used for the optimization of the selected parameters. Thus, the effects þ 20:04  C  D  13:22  C  E  1:14  D  E of these parameters on the MPKSF–As adsorption was 2 2 2 2 þ 4:05  A þ 1:01  B  23:07  C  7:56  D shown by the response surface plots using two parameters þ 4:89  E : ð4Þ simultaneously while the remaining parameters were set at their center points. However, it should be noted that since The equation implied that the constant with a value of two parameters were used simultaneously, hence, only one 85.53 (see Eq. (4)) which is independent of any factor or set of the figures were presented per variable, while the rest interaction between factors suggests that the average were referred to accordingly under the affected variables, removal of As(III) by MPKSF was 85.53%. Although i.e., from ‘‘Effect of temperature’’ to ‘‘Effect of contact this average removal is independent of the factors in the time’’. experimental setup (Kumar et al. 2008). In addition, Eq. (4) shows that pH and initial As(III) concentration Effect of selected variables and response surface plots had a positive effect while contact time, temperature and adsorbent dosage had a negative effect on the adsorption Effect of pH As seen in Fig. 2a–d, the combined effects percentage of As(III) by MPKSF. In general, the positive of pH with contact time, initial As(III) concentration, sign represents the synergistic effect and the negative sign adsorbent dosage and temperature, respectively, were 123 Appl Water Sci (2017) 7:4573–4591 4581 86.25 95.25 82.5 90.5 78.75 85.75 6.00 6.00 6.50 6.50 7.00 720.00 7.00 100.00 B: pH 541.25 76.25 B: pH 7.50 362.50 7.50 52.50 183.75 28.75 5.00 8.00 8.00 5.00 A: contact time D: initial conc As 93.5 78.5 86.5 45.5 83 6.00 8.00 6.50 7.50 0.30 7.00 0.24 7.00 B: pH 40.00 35.00 0.17 7.50 6.50 B: pH 30.00 0.11 25.00 0.05 6.00 8.00 20.00 C: adsorbent dosage E: temperature Fig. 2 3D response surface plots for As(III) removal vs a contact c adsorbent dose (0.05–0.3 g) and the pH of the adsorbent MPKSF time (5–720 min) and the pH of the adsorbent MPKSF suspension suspension (6–8); d temperature (20–40 C) and the pH of the -1 (6–8); b initial As concentration (5–100 lgL ) and pH (6–8); adsorbent MPKSF suspension (6–8) presented. From the responses, the lower value of As(III) As reported previously, the pHpzc value for MPKSF residual indicates a higher percentage of removal as given was 5.94. At a pH above the pHpzc value, the As(III) by Eq. (5): cations are attracted to the negatively charged surface of MPKSF, which resulted from the deprotonation of MPKSF C  C initial final % Removal ¼  100%: ð5Þ hydroxyl groups. Meanwhile, at pH lower than pHpzc, the initial adsorption of As(III) cations occurs only slightly because Figure 2a shows the interaction between the effect of pH the surface of MPKSF is positively charged due to the and the contact time of As(III) solution with the MPKSF protonation of MPKSF hydroxyl group. Thus, a solution adsorbent. The optimum removal of As occurred at pH 6.55 pH of 6.55 provides the optimum condition for adsorption and at a contact time of 44.73 min with a removal percentage of MPKSF–As. of 94.76% as the other parameters were set to their center points. Figure 2b illustrates the interaction of initial As(III) Effect of temperature concentration with pH with the optimum removal of As(III) achieved at pH 6.55 at an initial As(III) concentration of The experiment was conducted within a temperature range -1 5 lgL with a removal percentage of 99.10%. of 20–40 C. As seen in Fig. 2d (refer to Fig. 2din‘‘Effect Further, Fig. 2c, d shows the relationships between pH of pH’’ above for the effect of temperature and pH) and 3a– with adsorbent dosage and temperature, respectively. The c, the relationships between temperature and pH, contact excluded parameters for the respective response surface time, adsorbent dose and initial concentration, respectively, plots remain at their center points. Figure 2c shows that the are presented. All the four response surface plots show that removal of As(III) attained a maximum of 100% at pH 8 the removal of As(III) increases with temperature. Thus, with an adsorbent dose of 0.3 g, while the other parameters the adsorption process could be described as an endother- were at their center points. In Fig. 2d, the relationship mic process, as the removal of As(III) is optimum at higher between pH and temperature gives an optimum removal of temperatures with efficiency of 91–100% removal. 91.89% at pH 6 and a temperature of 40 C. Removal % Removal % Removal % Removal % 4582 Appl Water Sci (2017) 7:4573–4591 98.75 93.5 88.25 0.30 0.24 20.00 25.00 0.17 720.00 30.00 541.25 C: adsorbent dosage 0.11 40.00 362.50 35.00 35.00 30.00 A: contact time 183.75 25.00 0.05 40.00 E: temperature 20.00 5.00 E: temperature 88.25 84.5 80.75 5.00 28.75 40.00 35.00 52.50 30.00 D: initial conc As 76.25 25.00 E: temperature 100.00 20.00 Fig. 3 3D response surface plots for As(III) removal vs a temperature (20–40 C) and contact time (5–720 min); b temperature (20–40 C) and -1 adsorbent dosage (0.05–0.3 g); c temperature (20–40 C) and initial As(III) concentration (5–100 lgL ) Effect of adsorbent dosage the percentage removal data showed increasing pattern while other percentages removal data became slightly Refer to Fig. 2c under ‘‘Effect of pH’’ and Fig. 3b under decreased. In this case, other parameters need to be con- ‘‘Effect of temperature’’ above and Fig. 4a, b under ‘‘Effect sidered especially the adsorbent dosage which provides the of contact time’’ below, for the combined effect of adsor- information on the available vacant adsorption sites for bent (MPKSF) dose with pH, temperature, contact time and adsorption of As(III) at higher concentrations. initial As(III) concentration, respectively. The results showed that the percentage removal increases as the Effect of contact time adsorbent dose increases. The increase in the amount of adsorbent used can be attributed to the formation of a Refer to Fig. 2a (under ‘‘Effect of pH’’, and Fig. 3a under greater surface area and provides more available vacant ‘‘Effect of temperature’’) above, and Fig. 4a, c below. The sites for the adsorption of As(III) on MPKSF surface. results illustrate the combined response surface plots of contact time with pH of the adsorbent MPKSF suspension, Effect of initial As(III) concentration temperature, adsorbent dosage and initial As concentration, respectively. It shows that the increase in contact time Adsorption of MPKSF–As was carried out with different duration causes the removal percentages of As(III) to -1 initial concentrations ranging from 5 to 100 lgL . Refer increase. As the contact time increased, the adsorbate had to Fig. 2b under ‘‘Effect of pH’’ and 3c under ‘‘Effect of enough time to disperse and be adsorbed onto the surface temperature’’ above, and Fig. 4b, c, under ‘‘Effect of and into the pores of MPKSF until the adsorption reached contact time’’ below. The results illustrates the combined equilibrium at an optimum contact time. effect of initial As(III) concentration with pH of the adsorbent MPKSF suspension, temperature, adsorbent dose Optimization using desirability function and contact time, respectively, while two parameters are kept constant. As shown by the response surface plots, it A desirable value for each input parameter and the revealed that as the initial concentration increases, some of response can be selected. The multiple response methods Removal % Removal % Removal % Appl Water Sci (2017) 7:4573–4591 4583 96 75 79 56 720.00 541.25 100.00 76.25 0.30 0.30 362.50 0.24 0.24 52.50 A: contact time 0.17 0.17 183.75 D: initial conc As 28.75 0.11 0.11 0.05 5.00 5.00 0.05 C: adsorbent dosage C: adsorbent dosage 5.00 720.00 28.75 541.25 52.50 362.50 76.25 183.75 D: initial conc As A: contact time 5.00 100.00 Fig. 4 3D response surface plots for As(III) removal vs a contact time (5–720 min) and adsorbent dose (0.05–0.3 g); b initial As(III) -1 -1 concentration (5–100 lgL ) and adsorbent dose (0.05–0.3 g); c initial As(III) concentration (5–100 lgL ) and contact time (5–720 min) was applied to obtain the optimum condition for the five Table 6 Optimal condition and model validation for As(III) removal parameters used including contact time, pH, adsorbent by MPKSF dose, initial As(III) concentration and temperature. The Parameters Optimal conditions numerical optimization reveals the points that maximize Contact time (min) 45 the desirability function. In this study, the maximum level pH 6.55 for As(III) removal was set for desirability at a minimum Adsorbent dosage (g) 0.29 level of contact time (5 min), minimum adsorbent dosage -1 Initial As(III) concentration (lgL ) 100.00 (0.05 g), the maximum level of initial As(III) concentration -1 (100 lgL ), the level of solution pH in the range of 6–8 Temperature (C) 26 Removal (%) and the level of temperature within a range of 20–40 C (data not shown for the desirability ramp). Predicted 96 Using Eq. (5) reported previously to determine the Experimental 95 removal percentage, 96% removal of As(III) was achieved Error (%) 1 using the optimal conditions at contact time of 45 min, -1 initial As(III) concentration maximized to 100 lgL , adsorbent dosage of 0.29 g, initial solution pH of 6.55 and Adsorption equilibrium studies of As(III) onto MPKSF adsorbent the temperature at 26 C as shown in Table 6. Experi- mentally, these optimum values are applied in a verifica- Adsorption isotherm models tion experiment and resulted in 95% removal of As(III) by MPKSF adsorbent with a percentage error of 1% as com- In this study, several isotherm models were applied to pared to the predicted removal of As(III). Thus, this indi- determine the type of adsorption that occurred on the cated that the Box–Behnken design model is reliable for the optimization of various parameters used in an adsorp- magnetic activated carbon surface, by fitting the data to the Langmuir, Freundlich, and Temkin isotherm models. The tion process. correlation coefficient, R obtained from the isothermal Removal % Removal % Removal % 4584 Appl Water Sci (2017) 7:4573–4591 plots was used to identify the isotherm model that best Further, the inherent feature of this isotherm namely a described the adsorption of As(III) onto MPKSF. dimensionless constant separation factor, R can be expressed as an equilibrium parameter. The parameter is The Langmuir adsorption isotherm model calculated using Eq. (9). The effect of the separation factor on the adsorption nature of the Langmuir isotherm is The Langmuir is represented by the following equations; summarized in Table 7. (1) Eq. (6) which was formulated based on the kinetic R ¼ ; ð9Þ theory; (2) Eq. (7) depicts the equation for the value of 1 þ K C L o adsorbate adsorbed on the adsorbent; (3) the linear form of -3 where C = initial concentration of surfactant (mg dm ), the Langmuir equation is given by Eq. (8): K = Langmuir isotherm constant. x q K C max L e ¼ q ¼ ; ð6Þ To illustrate, that the homogeneous adsorption process m 1 þ K C L e predicted by the Langmuir isotherm assumes a monolayer ðÞ C  C V o e adsorption of molecules onto the surface of the adsorbent. q ¼ ; ð7Þ m The Langmuir linear plot of the specific adsorption (C /q ) e e against the equilibrium concentration (C ) for adsorption of C C 1 e e e ¼ þ ; ð8Þ As(III) onto MPKSF is depicted in Fig. 6. The C was q q q K e max max L obtained after 180 min. The correlation coefficient (R ) where x = mass of adsorbate adsorbed (lg), m = mass of obtained from the Langmuir isotherm is 0.9973, which adsorbent (g), V = volume of solution used for adsorption indicates that the adsorption data of the As(III) onto the -1 process (L), C = equilibrium concentration (lgL ), MPKSF surface was well fitted to the Langmuir isotherm -1 C = initial concentration (lgL ), q = amount of o e model. The monolayer adsorption capacity (q )of max -1 adsorbate adsorbed at equilibrium (lgg ), q = max- -1 max As(III) onto MPKSF was found to be 48.08 lgg . imum adsorption at monolayer coverage, K = Langmuir A dimensionless equilibrium parameter, R was used to -1 isotherm constant (L lg ). express the nature of adsorption of the Langmuir isotherm. To assess the Langmuir isotherm model, a graph of C / The R value calculated from the adsorption data is 0.0765 q against C is plotted, a straight line is obtained and the e e indicating that the adsorption of As(III) on MPKSF was a values of K and q are computed from the values of the L max favorable process as the R value lies between 0 to 1 slope and intercept of the graph. The fitted isotherm in this study was illustrated by a plot of equilibrium concentration of adsorbate C on the adsorbent, q , i.e., C /q against the e e e e Table 7 Effect of separation factor on Langmuir isotherm equilibrium concentration of adsorbate, C . Figure 5 R Adsorption nature of isotherm depicts the fitted As(III) adsorption isotherm which indi- cated the presence of a linear part of plotted curve at low R [ 1 Unfavorable C followed by a slight curvature around C = 2.56 - e e R = 1 Linear -1 lgL towards the completion of coverage of the mono- 0 \ R \ 1 Favorable layer with the R value of 0.9599. Hence, this indicated that R = 0 Irreversible adsorption of arsenite onto MPKSF is L-type (having no strict plateau). A similar observation was reported by El- Said et al. (2009) whose study demonstrated the adsorption of As(III) and As(V) using Nigella sativa L. Fig. 6 Langmuir isotherm representing the variation of specific Fig. 5 Langmuir equilibrium isotherm for the adsorption of As(III) adsorption (C /q ) against the equilibrium concentration (C ) for e e e onto MPKSF adsorption of As(III) onto MPKSF 123 Appl Water Sci (2017) 7:4573–4591 4585 (0 \ R \ 1) at an equilibrium temperature of 30 C (Table 7). The Freundlich isotherm model This isotherm was applied as an empirical model which considers the data often fit to the empirical equation stated in Eq. (10): ¼ q ¼ K C ; ð10Þ e F e where x = weight of solute adsorbed, m = weight of Fig. 7 Freundlich isotherm indicating the variation of log q with adsorbent, C = equilibrium concentration of adsorbate -1 respect to log C for adsorption of As(III) onto MPKSF (mg L ), q = amount of solute adsorbed at equilibrium -1 (mg g ), K and n = Freundlich isotherm constant -1 (mg g ) Table 8 Effect of n values on Freundlich isotherm From the isotherm, the n value reveals the nature of the n value Adsorption nature of isotherm adsorption, i.e., how favorable the adsorption process was. n = 1 Linear The value of n was used as the linearity parameter implying n \ 1 Chemical process that if the n value lies between one and ten, this indicates a n [ 1 Physical process favorable sorption process of the adsorbent (Freundlich 1906). The effect of the n value on the nature of adsorption as represented by the Freundlich isotherm is shown in of As(III) and Fe O in MPKSF is depicted in Fig. 8 and 3 4 Table 4. The linear form of the equation is given by supported by the FTIR analysis of the MPKSF–As. Eq. (11). A graph of log (x/m) against log C results in a It is well known that sodium arsenite can be represented straight line at which the value slope and intercept are the as sodium ortho-arsenite (Na AsO ) and sodium meta- 3 3 value of 1/n and log K , respectively. arsenite (NaAsO ), the latter was used in this study for x 1 adsorption. Based on Fig. 9, MPKSF reacted with arsenous log ¼ log q ¼ log K þ log C : ð11Þ e F e m n acid (H AsO ) which is produced according to the chem- 3 3 A multilayer adsorption for the heterogeneous surface is ical reaction (Eq. 11). Firstly, sodium meta-arsenite reacts with water producing arsenic trioxide with sodium ions and indicated by the Freundlich isotherm model. The heterogeneous system of adsorption assumes that there is hydroxide ions followed by the slow hydrolysis of arsenic trioxide. The reaction proceeds in a basic condition, finally no formation of monolayer adsorption of As(III) on MPKSF. Figure 7 illustrates the plot of log q against log producing arsenous acid: C . The correlation coefficient, R obtained from the graph þ 2NaAsO þ H O ! As O þ 2Na 2 2 2 3 illustrates that the Freundlich isotherm model is not well slow hydrolysis þ 2OH !2H AsO : ð12Þ 3 3 fitted to the adsorption data with an R value of 0.9744. The Freundlich constants, K and the n value are shown in Liu et al. (2015) reported that magnetite (Fe O )is 3 4 Table 8. known as a mixture of two iron oxides which are composed The slope of the Freundlich isotherm (Fig. 7) shows that of 67% of Fe(III) and 33% of Fe(II). In the adsorption the 1/n value is less than 1 and this has shown that the study, the major Fe(III) species was reported to interact process is a favorable physical adsorption process. The with As(III) to form either inner-sphere monodentate or smaller 1/n value indicates that a strong bond is present bidentate-binuclear complex. This can be attributed to the between the adsorbent and adsorbate molecules (Okeola oxidation reaction which takes place in the presence of and Odebunmi 2010). oxygen under atmospheric experimental conditions hence In the paper by Mayo et al. (2007), it was demonstrated suggesting the oxidation of Fe(II) to Fe(III). that the adsorption of As(III) and As(V) onto Fe O 3 4 nanoparticles exhibits the surface complexation reaction by Temkin isotherm forming either inner-sphere monodentate or bidentate-bin- uclear complex with iron oxide (Fe O ). A similar reaction 3 4 The Temkin isotherm model was applied to consider the of magnesium with iron oxide, i.e., Fe O and c-Fe O was 3 4 2 3 effect of the interactions between the adsorbent and the reported by Jolstera ˚ et al. (2012). The proposed adsorption adsorbate on an adsorption isotherm. The model assumes 123 4586 Appl Water Sci (2017) 7:4573–4591 Fig. 8 Proposed adsorption of As(III) onto MPKSF. (adapted from Ciuro Juncosa 2008; O’Reilly et al. 2001) where b = Temkin isotherm constant, A = equilibrium binding constant correspond to maximum binding energy, -1 -1 R = gas constant (8.314 J mol K ), T = absolute temperature, K. The values of constant A and B are obtained from a plot of q against ln C . e e Figure 9 shows the plot of q against ln C whereby the e e slope and intercept values obtained from the graph plot are used to calculate Temkin constant A, and the heat of sorption constant B. The R value obtained from the linear plot of Temkin isotherm model is 0.9816 indicating that the adsorption data are applicable to this model. Similar observations have been reported by Itodo and Itodo (2010) Fig. 9 Temkin isotherm showing the variation of q against ln C for e e on the adsorption of atrazine onto sheanut shell. Similarly adsorption of As(III) onto MPKSF by Hamdaoui and Naffrechoux (2007) whose study demonstrated the adsorption of phenol and chlorophenol that the heat of adsorption of the molecules present in the onto granular activated carbon. Maurya and Mittal (2006) adsorbed layer is reducing linearly with the coverage of the had also established the linear Temkin plot for the molecules instead of in logarithmic pattern due to this adsorption of methylene blue and Rhodamine B onto interaction (Temkin 1941). This means that as the coverage activated carbon. of adsorbed layer increased, the heat of adsorption Comparing the three correlation coefficients; R for the decreased. three isotherms, the Langmuir isotherm gave the best fit of Temkin isotherm is given by Eq. (13): adsorption isotherm with highest correlation coefficient, R RT value of 0.9973 followed by Temkin (0.9816) and Fre- q ¼ lnðAC Þ: ð13Þ e e undlich (0.9744) isotherm models (Table 9). Langmuir isotherm implies that adsorption of As(III) onto MPKSF The linear form of Eq. 13 is given by: adsorbent occurs in a monolayer adsorption at which when RT RT q ¼ ln A þ ln C : ð14Þ the available sorption sites of MPKSF are fully occupied, e e b b no further adsorption process can take place at those sites. Substituting RT/b with B and hence, It is corroborated by the formation of inner-sphere com- plexes between iron oxide and As(III) molecules on q ¼ B ln A þ B ln C ; ð15Þ e e MPKSF surface. 123 Appl Water Sci (2017) 7:4573–4591 4587 Fig. 10 Desorption of As(III) from MPKSF. Data obtained: n = 3 for each desorption experiment The R value for Langmuir and Freundlich are 0.9973 desorption of As(III) from MPKSF. During the first stage and 0.9744, respectively (Table 9). The Temkin isotherm of desorption using distilled water, the percentage of -1 shows that the heat of adsorption is low (9.72 J mol ) As(III) desorbed in the solution was 0.72% followed by indicating physical adsorption. Furthermore, the calculated 0.78 and 0.97% at second and third stages of desorption, -1 Freundlich (q ) (54.48 lgg ) was higher than the respectively. The small amount of As(III) detected after the max -1 adsorption capacity (q = 48.08 lgg ) determined first desorption (Des1) was presumably due to the com- max from the Langmuir isotherm. Based on the q value plexation reaction between As(III) ions with the iron in max obtained, the adsorption of As(III) is more of a physical MPKSF adsorbent (Ciuro ´ Juncosa 2008) which prevents the adsorption as described by the Freundlich isotherm. Studies dissociation of As(III) from iron oxide on the surface of by Liu et al. (2015) reported that the presence of chemical MPKSF. This is also consistent with the results of the interaction between As(III) and iron oxide forming inner- Langmuir adsorption isotherm, thus explaining the forma- sphere surface complex can be best explained by Langmuir tion of chemical bonds between As(III) and the surface of isotherm, suggesting a monolayer As(III) adsorption onto MPKSF which prevent the As(III) from desorbing easily the Fe O surface. However, they also reported that the from MPKSF. 3 4 adsorption of As(III) onto AC showed the best fit with Freundlich isotherm. Based on the experimental data, it is Identification of the proposed mechanism suggested that both chemisorption, involving the formation of the adsorption of As(III) onto the MPKSF of inner-sphere complex and physisorption on activated carbon occurred in the adsorption of As(III) onto MPKSF. Fourier transform infrared spectroscopy (FTIR) analysis Desorption of As(III) from MPKSF As stated previously a mechanism was proposed for the adsorption of As(III) onto the MPKSF, in this section, this Desorption experiment was conducted to examine the mechanism was identified using the results of the spec- reusability of the MPKSF adsorbent and the reversibility of troscopy studies and supported by the results of the char- the adsorption process. Figure 10 illustrates the percentage acterization of the MPKSF after the adsorption. Figure 11 shows the FTIR spectrum of MPKSF after adsorption of As(III) (MPKSF–As). Five peaks could be observed in the Table 9 Langmuir, Freundlich and Temkin isotherm model param- spectrum identified at wavenumber of 3430, 1625, 1387, -1 eters and correlation coefficient for adsorption of As(III) on MPKSF 1078 and 573 cm . These are assigned to the functional groups OH, C=O, C–C, C–O, and M–O, respectively. Isotherm Parameters Both spectra of MPKSF before and after adsorption -1 Langmuir q (lgg ) 48.08 max show the presence of peaks at a wavenumber of 3430, 1625 -1 K (L lg ) 0.1208 -1 and 573 cm . Additionally, in MPKSF–As, two new R 0.9973 -1 -1 peaks were detected at 1387 cm and around 800 cm Freundlich K 10.86 which are assigned to C–C bending (Mayo et al. 2007) and 1/n 0.36 As–O interaction (Ito et al. 1995), respectively, on MPKSF -1 q (lgg ) (calculated) 54.48 max surface after adsorption. However, the spectrum showed a -1 R 0.9744 less significant band at 800 cm . Thus the EDX analysis -1 Temkin A (L g ) 1.436 data are used to validate the presence of As(III). According -1 B (J mol ) 9.72 to Sayle (2000), at a pH lower than the pKa value, the R 0.9816 molecules will be mostly protonated. Thus, the negatively 123 4588 Appl Water Sci (2017) 7:4573–4591 As-O 1078.32cm-1 26 1078.32 3914.78cm-1 1387.99cm-1 24 1387.99 3698.94cm-1 635.83cm-1 57 573. 3.35 3cm- 5 1 22 MPKSF-As C-C 1625.71cm-1 1625.71 3780.52cm-1 1093.92cm-1 445.06cm-1 MPKS 3430.14cm-1 3430.14 12 1629.52cm-1 636.19cm-1 579.58cm-1 3435.28cm-1 C=O C-O M-O OH 4000 3500 3000 2500 2000 1500 1000 500 400 cm-1 Fig. 11 FTIR spectrum of MPKSF and MPKSF–As Fig. 12 XRD diffractogram of MPKSF–As charged MPKSF provides greater affinity towards the of Mandal et al. (2013) on the adsorption of As(III) by protonated As(III) ions at pH lower than 9.2 which result in zirconium polyacrylamide hybrid material. However, no the formation of inner-sphere complexes between As(III) crystalline As(III) was detected using this analysis, prob- and Fe O in MPKSF. Additionally, Bundschuh et al. ably due to the small amount of As(III) adsorbed on 3 4 (2005) reported that as the pH increases, the dominant MPKSF. negative charges are present on adsorbent surface hence interference with the adsorption of As(III) and As(V) are Scanning electron microscopy (SEM) analysis significantly governed by the surface charge of adsorbent. The morphology of MPKSF–As under magnification of X-ray diffraction (XRD) analysis 15509 is depicted in Fig. 13a illustrating that MPKSF–As surface experienced a distinct change as compared to Figure 12 illustrates the diffractogram pattern of MPKSF MPKSF. In addition, the presence of a layer of coating on after adsorption (MPKSF–As). The pattern obtained is the surface is presumably assigned to the adsorbed layer of quite similar to the diffractogram of MPKSF before the As(III) complexes on the MPKSF. The entire coverage of adsorption process. There are four intense peaks assigned As(III) in the pores as depicted in Fig. 13b under higher to Fe O at 2h 30.45, 35.75, 57.85 and 63.00, mean- magnification (62009) shows that the adsorption of As(III) 3 4 while one peak at 2h 43.35 is assigned to c-Fe O , occurs evenly over the surface and in the pores of MPKSF. 2 3 respectively. The increment in crystallite size of MPKSF– Further, it can be seen that As(III) particles do not block As is probably due to crystal defect after the adsorption of the external pores as it is still observed clearly even after As(III) onto the MPKSF. This is consistent with the work adsorbing As(III). With respect to both figures, it shows %T Appl Water Sci (2017) 7:4573–4591 4589 Fig. 13 Micrograph of MPKSF–As under magnification a 91550 and b 96200 Fig. 14 EDX spectrum for MPKSF–As sample that the pore size of MPKSF was affected after the Table 10 Elemental analysis for sample MPKSF and MPKSF–As adsorption process. The size of the pores reduced compared Sample MPKSF MPKSF–As to the pore size of MPKSF before adsorption. This is Weight (%) Weight (%) possibly due to the formation of the inner-sphere com- Carbon (C) 17.14 18.16 plexes of As(III) ion with Fe O in MPKSF on the walls of 3 4 Oxygen (O) 31.63 30.03 the external pores. Phosphorus (P) 1.78 3.24 Iron (Fe) 49.45 48.05 Energy dispersive X-ray analysis (EDX) analysis Arsenic (As) – 0.52 The energy dispersive X-ray analysis is used to detect the presence of arsenic after adsorption by MPKSF. Figure 14 Conclusions illustrates EDX spectrum of MPKSF–As. The spectrum showed the peaks similar to the untreated MPKSF for This study demonstrated that the MPKSF effectively carbon, oxygen, iron and phosphorus. However, after the removed As(III) from waste water by the formation of adsorption process, the presence of As(III) was detected inner-sphere complexes between the Fe and the As(III). with a composition of 0.52% as illustrated in Table 10 Further, desorption of the sorbed As(III) was found to be in below. The amount of arsenic detected is small due to the very minute concentrations, thereby suggesting that the low concentration of As used in the adsorption process sorbed As(III) was rigidly bound by inner-sphere com- -1 which is in the range of 5–100 lgL . The composition of plexation mechanism. An efficient adsorption process was other elements present in the sample was unaffected by the revealed to take place at pH 6 and 7 and at a longer contact adsorption process as the composition of these elements are time. The initial As(III) concentration and adsorbent dose quite similar to the composition of the untreated MPKSF. 123 4590 Appl Water Sci (2017) 7:4573–4591 Arsenic in Groundwater’’, 32nd international geological con- were concluded to be dependent on each other due to the gress, Florence, Italy, 18–19 August 2004 availability of adsorption binding sites for As(III) presented Cheng W et al (2016) Competitive sorption of As(V) and Cr(VI) on by greater dose of the MPKSF. Further, the higher reaction carbonaceous nanofibers. Chem Eng J 293:311–318 temperature was shown to generate more residual As(III) in Ciuro ´ Juncosa E (2008) Adsorption properties of synthetic iron oxides: as(V) adsorption on goethite (alpha-FeOOH). Dissertation. https:// the solution hence reducing the removal efficiency of scholar.google.com/scholar?q=Ciur%C3%B3Juncosa?E?%2820 As(III) by MPKSF. 08%29?Adsorption?properties?of?synthetic?iron?oxides%3A The predicted results from the designed experiment on ?As%28V%29?adsorption?on?goethite?%28alpha-FeOOH% the adsorption of As(III) onto MPKSF using Box–Behnken 29&btnG=&hl=en&as_sdt=0%2C5. Accessed 04 Sept 2017 Daud WMAW, Ali WSW (2004) Comparison on pore development of statistical data revealed that 96% of As(III) removal was activated carbon produced from palm shell and coconut shell. achieved utilizing the optimal conditions at contact time of Biores Technol 93:63–69 44.73 min, initial As(III) concentration maximized to Elizalde-Gonza ´lez M, Mattusch J, Einicke W-D, Wennrich R (2001) -1 100 lgL , adsorbent dosage of 0.29 g, initial solution pH Sorption on natural solids for arsenic removal. Chem Eng J 81:187–195 of 6.55 and the temperature at 26.38 C. A verification El-Said S, Alamri M, El-Barak A-BS, Alsogair O (2009) Adsorptive experiment conducted using a real waste water sample removal of arsenite As(III) and arsenate As(V) heavy metals resulted in a 95% As(III) removal, which indicated that from waste water using Nigella sativa L Asian. J Sci Res 0.96% error had occurred. 2:96–104 Freundlich HMF (1906) Over the adsorption in solution. J Phys Chem The adsorption studies suggest that the adsorption of 57:385–471 As(III) onto the MPKSF was highly dependent on pH and Hamdaoui O, Naffrechoux E (2007) Modeling of adsorption contact time relative to the initial As(III) concentration, isotherms of phenol and chlorophenols onto granular activated adsorbent dose and temperature. The adsorption data of the carbon: part I. Two-parameter models and equations allowing determination of thermodynamic parameters. J Hazard Mater As(III) onto the MPKSF was well described by the Lang- 147:381–394 muir isotherm followed by Temkin and Freundlich Hussein M, Tarmizi R, Zainal Z, Ibrahim R, Badri M (1996) isotherms. Preparation and characterization of active carbons from oil palm shells. Carbon 34:1447–1449 Acknowledgements This work was funded by the Universiti Ito F, Nakanaga T, Takeo H, Essig K, Jones H (1995) The FTIR Teknologi Malaysia research Grant no. 12H29. absorption spectrum of the fundamental band of the AsO radical. J Mol Spectrosc 174:417–424 Open Access This article is distributed under the terms of the Itodo A, Itodo H (2010) Sorption energies estimation using Dubinin– Creative Commons Attribution 4.0 International License (http:// Radushkevich and Temkin adsorption isotherms. Life Sci J Acta creativecommons.org/licenses/by/4.0/), which permits unrestricted Zhengzhou Univ Overseas Ed 7:31–39 use, distribution, and reproduction in any medium, provided you give Jain C, Ali I (2000) Arsenic: occurrence, toxicity and speciation appropriate credit to the original author(s) and the source, provide a techniques. Water Res 34:4304–4312 link to the Creative Commons license, and indicate if changes were Jolstera ˚ R, Gunneriusson L, Holmgren A (2012) Surface complex- made. ation modeling of Fe O –H? and Mg(II) sorption onto 3 4 maghemite and magnetite. J Colloid Interface Sci 386:260–267 Kumar A, Prasad B, Mishra IM (2008) Optimization of process Publisher’s Note Springer Nature remains neutral with regard to parameters for acrylonitrile removal by a low-cost adsorbent jurisdictional claims in published maps and institutional affiliations. using Box–Behnken design. J Hazard Mater 150:174–182. doi:10.1016/j.jhazmat.2007.09.043 Kunamneni A, Singh S (2005) Response surface optimization of enzymatic hydrolysis of maize starch for higher glucose References production. Biochem Eng J 27:179–190 Liu C-H, Chuang Y-H, Chen T-Y, Tian Y, Li H, Wang M-K, Zhang Addo Ntim S, Mitra S (2011) Removal of trace arsenic to meet W (2015) Mechanism of arsenic adsorption on magnetite drinking water standards using iron oxide coated multiwall nanoparticles from water: thermodynamic and spectroscopic carbon nanotubes. J Chem Eng Data 56:2077–2083 studies. Environ Sci Technol 49:7726–7734 Adhoum N, Monser L (2002) Removal of cyanide from aqueous Mandal S, Sahu MK, Patel RK (2013) Adsorption studies of arsenic solution using impregnated activated carbon. Chem Eng Process (III) removal from water by zirconium polyacrylamide hybrid 41:17–21 material (ZrPACM-43). Water Resour Ind 4:51–67 Anyika C, Asri NAM, Majid ZA, Yahya A, Jaafar J (2017) Synthesis Maurya NS, Mittal AK (2006) Applicability of equilibrium isotherm and characterization of magnetic activatedcarbon developed models for the biosorptive uptakes in comparison to activated from palm kernel shells Nanotechnology for Environmental carbon-based adsorption. J Environ Eng 132:1589–1599 Engineering 2:16. doi:10.1007/s41204-017-0027-6 Mayo J et al (2007) The effect of nanocrystalline magnetite size on Budinova T, Petrov N, Razvigorova M, Parra J, Galiatsatou P (2006) arsenic removal. Sci Technol Adv Mater 8:71–75 Removal of arsenic (III) from aqueous solution by activated Milonjic ´ S, Kopec ˇni M, Ilic ´ Z (1983) The point of zero charge and carbons prepared from solvent extracted olive pulp and olive adsorption properties of natural magnetite. J Radioanal Nucl stones. Ind Eng Chem Res 45:1896–1901 Chem 78:15–24 Bundschuh J, Bhattacharya P, Chandrashekharam D (2005) Natural Nakahira A, Nishida S, Fukunishi K (2006) Adsorption of contam- arsenic in groundwater: occurrences, remediation and manage- inants by MACs and magnetic separation. J Ceram Soc ment. In: Proceedings of the pre-congress workshop ‘‘Natural 114:135–137 123 Appl Water Sci (2017) 7:4573–4591 4591 Okeola F, Odebunmi E (2010) Freundlich and Langmuir isotherms Shehzad A, Bashir MJ, Sethupathi S, Lim J-W (2016) An insight into parameters for adsorption of methylene blue by activated carbon the remediation of highly contaminated landfill leachate using derived from agrowastes. Adv Nat Appl Sci 4:281–288 sea mango based activated bio-char: optimization, isothermal O’Reilly SE, Strawn DG, Sparks DL (2001) Residence time effects on and kinetic studies. Desalin Water Treat 57:22244–22257 arsenate adsorption/desorption mechanisms on goethite. Soil Sci Tan I, Ahmad A, Hameed B (2008) Preparation of activated carbon Soc Am J 65:67–77. doi:10.2136/sssaj2001.65167x from coconut husk: optimization study on removal of 2,4,6- Pattanayak J, Mondal K, Mathew S, Lalvani S (2000) A parametric trichlorophenol using response surface methodology. J Hazard evaluation of the removal of As(V) and As(III) by carbon-based Mater 153:709–717 adsorbents. Carbon 38:589–596 Temkin MI (1941) Adsorption equilibrium and the kinetics of Payne KB, Abdel-Fattah TM (2005) Adsorption of arsenate and processes on nonhomogeneous surfaces and in the interaction arsenite by iron-treated activated carbon and zeolites: effects of between adsorbed molecules. Zh Fiz Chim 15:296–332 ˇ ˇ pH, temperature, and ionic strength. J Environ Sci Health Trakal L, Veselska ´ V, Safar ˇ´ ıkI,V´ıtkova ´ M, C´ ıhalova ´ S, Koma ´rek M 40:723–749 (2016) Lead and cadmium sorption mechanisms on magnetically Ribeiro MA, Goncalves AR, Rocha GJM, Quintana G (2006) modified biochars. Biores Technol 203:318–324 Cadmium and nickel removal from water by using activated Vance DB (2002) Arsenic chemical behavior and treatment. Internet carbon obtained from sugarcane lignin. Reunion Iberica de WHO (1981) Arsenic, environmental health criteria 18. World Health Adsorcion 116:14 Organization, Geneva Samsuri AW, Sadegh-Zadeh F, Seh-Bardan BJ (2013) Adsorption of Xu J, Gao N, Tang Y, Deng Y, Sui M (2010) Perchlorate removal As(III) and As(V) by Fe coated biochars and biochars produced using granular activated carbon supported iron compounds: from empty fruit bunch and rice husk. J Environ Chem Eng synthesis, characterization and reactivity. J Environ Sci 1:981–988 22:1807–1813 Sayle R (2000) Physiological ionization and pKa prediction. Euro Yakout SM, Sharaf El-Deen G (2016) Characterization of activated MUG. https://scholar.google.com/scholar?hl=en&q=Sayle%2C? carbon prepared by phosphoric acid activation of olive stones. R?Physiological?ionization?and?pKa?prediction?Metaphorics Arab J Chem 9:S1155–S1162. doi:10.1016/j.arabjc.2011.12.002 ?LLC%3A?2000?http%3A%2F%2Fwwwdaylightcom%2F Yao S, Liu Z, Shi Z (2014) Arsenic removal from aqueous solutions meetings%2Femug00%2FSayle%2Fpkapredicthtml&btnG=&as_ by adsorption onto iron oxide/activated carbon magnetic com- 20. Accessed 04 Sept 2017 posite. J Environ Health Sci Eng 12:1 Schwertmann U, Murad E (1983) Effect of pH on the formation of Zou Y et al (2016) Environmental remediation and application of goethite and hematite from ferrihydrite. Clays Clay Miner nanoscale zero-valent iron and its composites for the removal of 31:277–284 heavy metal ions: a review. Environ Sci Technol 50:7290–7304 Segurola J, Allen NS, Edge M, Mc Mahon A (1999) Design of eutectic photoinitiator blends for UV/visible curable acrylated printing inks and coatings. Prog Org Coat 37:23–37

Journal

Applied Water Science – Springer Journals

Published: Sep 8, 2017

There are no references for this article.