KMnO4 Modified Carbon Prepared from Waste of Pineapple Leaf Fiber Production Processing for Removal of Ferric Ion from Aqueous Solution

Corresponding Author: Sumrit Mopoung Department of Chemistry, Faculty of Science, Naresuan University, Phitsanulok, Thailand Email: sumritm@nu.ac.th Abstract: KMnO4 modified carbon materials from waste of pineapple leaf fiber production processing were prepared and characterized. The effects of % weight of KMnO4 (0-5 %wt) and carbonization temperature (500-700°C) were studied by SEM-EDS, XRD, FTIR and BET. The KMnO4 modified waste carbon was used for Fe removal from aqueous solutions. The effects of KMnO4 modification, contact time, pH and loading were evaluated. The Langmuir isotherm and Freundlich isotherm were also used for evaluation of Fe adsorption by KMnO4 modified waste carbon. The results show that MnO2 deposited on the surface of KMnO4 modified waste carbon with some heterogeneity. The OH, CO and MnO groups are major functional groups on the surface of KMnO4 modified carbon. The BET surface area and total pore volume of modified waste carbon is increased with increasing weight % of KMnO4 and carbonization temperature, while the average pore size follows an inverse trend. The Fe adsorption on modified waste carbon reaches equilibrium within 60 min. The amount of Fe adsorbed on the modified waste carbon increases with increasing pH, reaches maximum at pH of about 4.7 and then sharply decreases at pH above 7. Both Langmuir and Freundlich isotherms were fitted for data a Fe adsorption on modified waste carbon with 25.25 mg g as theoretical maximum adsorption capacity.


Introduction
Pineapple is a food plant, which is grown in many countries. After harvesting, large quantities of pineapple leaves are left behind as waste causing various problems for farmers. This material has been used for pineapple leaf fiber production by mechanical milling process (Kengkhetkit and Amornsakchai, 2012). Milled materials form mechanical milling processing are pineapple leaf fiber and non-fibrous material. The nonfibrous material (or pineapple fiber waste) constitutes waste from this process and has been previously modified with H 3 PO 4 . The carbon content of pineapple fiber waste is high amounts sufficient (49.03%) to produce charcoal and activated carbon. After carbonization at 500-700°C, the charcoal products of pineapple fiber waste have BET surface area of 48.4335-413.3908 m 2 /g .
Potassium permanganate has the formula KMnO 4 and is highly reactive. The reactivity of potassium permanganate depends on pH. It is a strong oxidant at pH lower than 7, but only a mild oxidant in alkaline solutions (Zhang et al., 2013b). The surface carbons modified with MnO 4 − are amphoteric and therefore can adsorb both negatively and positively charged contaminants. At pH <pH pzc , manganese oxides are positively charged, which leads to attraction of anions such as fluorides. However, in basic solution, manganese oxides become negatively charged and attract cations (Daifullah et al., 2007). A composite material with negative charge has also been produced by oxidation with KMnO 4 (Arulraj and Rajamathi, 2013). It has been used for as catalysts for oxidation of phenolic compounds at 100°C (Abecassis-Wolfovich et al., 2005), cyclization of carbon fiber (Mathur et al., 1994), oxidation of hydroxylamine for H 2 O 2 production (Song et al., 2010) and oxidation of trichloroethene (Liang et al., 2014). The oxidation of carbon based materials by KMnO 4 with the aim of removal of natural organic matter has been studied by Zhang et al. (2013b). It has been used for water treatment from domestic and industrial sources (Zeid et al., 1995). It is also used for drinking water production (Tian et al., 2013). It could also improve the surface of active carbons for adsorption of contaminants such as NH 3 (Shan et al., 2008), Cd 2+ (Luo et al., 2013), Pb 2+ (Wang et al., 2012), Hg, NO and SO 2 (Ping et al., 2012), Cu 2+ (Baccar et al., 2009), Bi 3+ (Zhang et al., 2009) and the removal of Microcystis Aeruginosa cell (Waller et al., 2014).
In this research we studied the modification of Pineapple Fiber Waste Carbon (PFWC) with KMnO 4 and the properties of the resultant material for Fe 3+ removal from aqueous solution. The effect of KMnO 4 concentration was evaluated. The isotherms of Fe adsorption were also evaluated.

KMnO 4 Modified PFWC Preparation
Pineapple Fiber Waste (PFW), which is collected from mechanical milling process for the pineapple leaf fiber production, was used as the precursor material in this study. It consists of 1.15% ash, 49.03% C, 45.54% O, 1.16% Si, 1.15% P, 0.93% K and 0.86% Ca . It was cleaned with tap water and oven (SL 1375 SHEL LAB 1350 FX) dried at 105°C for 6 h. The dried PFW was carbonized at 500°C with an increment rate of 10°C min −1 and then kept at constant temperature for 1 h. The obtained Pineapple Fiber Waste Carbon (PFWC), which consists of 3.55% ash, 68.27% C, 22.61% O, 1.62% Si, 1.66% P, 3.87% K and 1.96% Ca, was modified with KMnO 4 (Merck, Germany) using 1, 3, or 5 %wt ratios. KMnO 4 modified Pineapple Fiber Waste Carbon (KPFWC) was oven (SL 1375 SHEL LAB 1350 FX) dried at 105°C for 3 h and then used for Fe 3+ adsorption experiments.

Analysis and Characterization
The modified products (KPFWC) were characterized by a Fourier transform infrared spectrometer (Spectrum GX, Perkin Elmer), X-ray diffractometer (PW 3040/60, X' Pert Pro MPD), scanning electron microscope-energy dispersive X-ray spectrometer (PHILIPS LEO 1455 VP) and Brunauer Emmett Teller surface area analyzer (Micromeritics TriStar II). The ash content of all samples was also analyzed by standard methods (ASTM, 1996).

SEM-EDS Analysis
Scanning electron microscopy was used to visualize the surface morphology of samples. The samples were coated with gold by a gold sputtering device for a clear vision of the surface morphology. Elemental composition of the samples was also observed using energy dispersive spectrometer by scanning through the surface of the samples.

XRD Analysis
X-ray patterns of modified products were recorded by X-ray powder diffractometer with a Cu tube anode.

FTIR Analysis
The distribution of surface functional groups of the samples was studied using a Fourier transform infrared spectrometer in the range 400-4000 cm −1 . The samples were prepared as pellets in IR grade KBr (Yang et al., 2011).

BET Measurement
Textural characteristics of samples were determined by N 2 adsorption at -196°C on a Brunauer Emmett Teller surface area analyzer. The samples were degassed at 250°C for 12 h under vacuum before the measurements. The multipoint Brunauere Emmette Teller equation was used to estimate the specific surface areas.

Fe Adsorption Experiments
The 1000 mg dm −3 Fe 3+ stock solution was prepared from 0.4827 g of FeCl 3 ·6H 2 O (Merck, Germany) was dissolved in deionized water, acidified with 1 cm 3 of 1 M HCl added and then diluted to 100 cm 3 with deionized water. Further solutions (5 mg dm −3 ) were prepared from the stock solution by dilution.
Fe 3+ adsorption experiments were followed using a method of Üçer et al. (2005). A KPFWC with 1 wt% was added to 50 cm 3 of Fe 3+ solution (5 mg dm −3 ) in a conical flasks and shaken continuously at 120 rpm at a temperature of 32±2°C. Following the adsorption, the aqueous phase was separated by centrifugation at 4000 rpm for 10 min and the final Fe 3+ ion concentration in the solution was determined by FAAS (Varian Spectr AA 220, Australia) with air-acetylene and cathode on Fe-hollow cathode lamp at 248.3 nm.
The amount of Fe 3+ removal was calculated by the difference in initial and final concentrations. The optimum condition for Fe 3+ removal was determined at different effects (e.g., pH 2-9, contact time (20-180 min) and KPFWC loading (0.05-2.0 g)).
Final concentration (C f ) of Fe 3+ was measured and calculated of Fe 3+ removal percentage as shown in the following equation (Shrestha et al., 2013): Where: The adsorption capacity (q t , mg g −1 ) at any time was calculated as shown below (Shrestha et al., 2013): Where: V = The volume of the solution (dm 3 ) W = The mass of dry adsorbent used (g)

Adsorption Isotherms
All of the experimental adsorption data's were fitted with both Langmuir equation and Freundlich equation.
The Langmuir equation is: Where: Q e (mg g −1 ) = The Fe 3+ adsorbed amount per unit mass of adsorbent C e (mg dm −3 ) = The Fe 3+ equilibrium concentration q max (mg g −1 ) = The maximum Fe 3+ amount that forms a complete monolayer on the surface of adsorbent K L (dm 3 mg −1 ) = The Langmuir constant which associated to adsorption heat The linear form of this equation after rearrangement is: The constants q max and K L was determined from the slope and intercept of plotting C e /Q e against C e , respectively (Mahmoud, 2015).
Freundlich model is used to estimate the adsorption intensity of KPFWC towards the Fe 3+ ions and is: This equation is conveniently used in linear form as (Mahmoud, 2015): where, Q e and C e have the same definitions of the Langmuir equation. K F and n are Freundlich constants related to adsorption capacity and heterogeneity factor, respectively. The constants K F and n were determined from the slope and intercept of plotting Log C e against Log Q e , respectively.

SEM and EDS Analyses
The SEM image of PFW (non fiber part) (Fig. 1a) exhibits a notched surface, while 3% KMnO 4 modified fresh pineapple waste (3KPFW) (Fig. 1b) displays a smooth surface. This reveals that KMnO 4 covered the surface of PFW after KMnO 4 modification. Figure 1c shows the image of PFWC after carbonization at 500°C. It can be seen that the PFWC is in a furrow form with a fairly smooth surface. However, the images modified pineapple waste (3KPFWC) after carbonization at 500°C and modification with 3 %wt of KMnO 4 ( Fig. 1d-e) display a rougher surface with cracks and a large number of small particles. This is because the PFW was significantly oxidized by KMnO 4 , degraded and became brittle and opaque (Fávaro et al., 2007). The small particles are quite non-uniformly dispersed on the surface of KPFWC. These particles are rich in K and Mn content as determined by point analysis using EDS (Table 1). This analysis also confirmed that C, O, Si, K and Mn are the main composition in the all of the KPFWC materials. Furthermore, it indicated that MnO 2 deposited on the surface of KPFWC. The high amounts of K and Mn elements on the surface of KPFW and KPFWC after carbonization at 500°C indicated that KMnO 4 is fixed on the surface of modified waste with massively (Kaushik, 2000).   Figure 2a showed the XRD spectrum of PFW. It exhibits the diffraction peaks of cellulose at 2θ of about 15-17 and 22° (Hajji et al., 2016). On the other hand, the XRD spectra of PFWC after carbonization at 500°C (Fig. 2b) show amorphous phase of carbon with broad peaks at 24 and 43°. These spectra also contain a peak at 29.5° and a little peak at 34.5°, which were related to a calcium and a potassium compound, respectively . This indicates that the crystalline cellulose has decomposed into amorphous materials during the carbonization process (Hajji et al., 2016). For 1KPFWC obtained after carbonization at 500°C (Fig. 2c), the diffraction peaks at about 22, 37, 42, 57 and 66° show the presence of MnO 2 in an amorphous phase (Wang et al., 2012;Xie and Gao, 2007). The MnO 2 was generated by the reduction of KMnO 4 in acid solution (Fávaro et al., 2007). These results confirm that the interaction between KMnO 4 and PFW has occurred. In addition, a Si phase is also observed at about 28.5, 47 and 69° , which probably comes from the original waste material. Figure 3a shows the FTIR transmission spectrum of PFW. It exhibits the peaks of cellulose, which consist of a strong board band at 3200-3400 cm −1 (ν(OH) vibration of the intra molecular hydrogen bond from cellulose) (Habibi, 2014), weak peaks at 2850.8-2918.8 cm −1 (ν(CH) vibration) (Hajji et al., 2016), 1737.3 cm −1 (ν(C = O) vibration in the cellulose chain) (Hajji et al., 2016), 1635 cm −1 (δ(OH) vibration of adsorbed water in crystalline cellulose (Zhao et al., 2013) or COO − stretching (Zhao et al., 2013), 1515.6 cm −1 (ν(CO) group and aromatic skeletal stretching of lignin (Zhao et al., 2013), 1431.3 cm −1 (δ(CH 2 ) deformation vibrations) (Hajji et al., 2016), 1373 cm −1 (δ(CH) in plane deformation vibrations in lignin, (Zhao et al., 2013), 1323 cm −1 (δ(CH 2 ) vibrations associated to crystallized cellulose I composition), 1235 cm −1 (ν(C-O) vibration) (Zhao et al., 2013) (Hajji et al., 2016) or ν(C-OH) groups (Habibi, 2014), very weak peak 1106 cm −1 (asymmetric stretching of the glycosidic ring or ν(C-Si-O)) (Gong et al., 2016), 1056.3 cm −1 (ν(C-O) vibration of secondary alcohol), indicating the presence of lignin in the waste (Mokhtari and Faghihian, 2015), 1026 cm −1 (ν(C-O) vibrations of primary alcohol) and 897 cm −1 (C-O-C bonds in the glycosidic linkage of crystalline cellulose) (Hajji et al., 2016) and the band between 610-527 cm −1 (C-O deformation vibration (Zhang et al., 2013a)). These peaks disappear after KMnO 4 modification for both carbonized and noncarbonized materials (Fig. 3b-n), with the exception of peaks at 1373, 1235 and 1090 cm −1 . These functional groups were not observed in carbonized samples ( Fig.  3e-n) as they undergo thermal degradation (Hajji et al., 2016). Thermal degradation by dehydration occurs above 140°C and dehydroxylation takes place above 550°C. At temperatures in the range of 160-230°C, water molecules which as physically adsorbed and belong in interlayer of adsorbent are removed. Increasing the temperature to 350°C causes the degradation of hydroxyl, carboxyl and carbonyl groups of cellulose (Luo et al., 2013;Natkański et al., 2013). For 1, 3, 5% KPFW and KPFWC ( Fig. 3b-d), the results are attributed to the surface of materials being covered by MnO 2 (Xie and Gao, 2007). So, the functional groups on these samples are shielded. The remaining peaks at 1373, 1235 and 1090 cm −1 are in KMnO 4 modified samples. The peak at 1373 cm −1 is associated with H-O bending vibration in phenols and carboxyls after KMnO 4 oxidation (Chen et al., 2012). The weak peak at 1235 cm −1 is related to C-O in phenols. The presence of this peak provides evidence of oxidization of the waste materials by KMnO 4 to generate phenols (Chen et al., 2012). The peak of νC-O vibration at 1090 cm −1 , which is slightly shifted from 1106 cm −1 , could also be associated to δO-H vibrations which bound to Mn atoms (Xie and Gao, 2007). This peak could also be attributed to stretching vibration of C-Si-O moiety (Gong et al., 2016).

Spectra FTIR
The board peak of ν(OH) at 3156-3227 cm −1 in KPFW and KPFWC (Fig. 3b-n), which is shifted from the value of 3402 cm −1 in PFW, is attributed to νH-O vibration in phenols and carboxyls, which were formed by KMnO 4 oxidation (Chen et al., 2012). This peak decreases in intensity with increasing carbonization temperature and disappears completely for samples carbonized at 700°C and above. This corresponds to the dehydration and desorption of cellulose and water molecules, respectively, which acted as a hydrogen bonding network of intra and intermolecular cellulosic chains (Hajji et al., 2016). It has been seen that the board peak at 3156-3227 cm −1 in KPFW disappeared after modification with 1% KMnO 4 modified (Fig. 3b). However, a boarder peak appears in the same position after modification with 3-5% KMnO 4 (Fig. 3c-d). This is because the content of carboxylic acids increases after oxidation at high KMnO 4 concentration (Chen et al., 2012). Therefore, νH-O vibration of carboxyl groups is enhanced after modification with 3-5% KMnO 4 . However, this peak disappeared in 1KPFW. This is attributed to the functional groups being obscured by MnO 2 at low KMnO 4 concentration as has been observed previously by Xie and Gao (2007). The new peaks appearing after KMnO 4 modification in KPFW and KPFWC materials are present at about 1560, 1035, 872, 755, 550 and 460 cm −1 . The peak at 1560 cm −1 is associated to conjugated νC-O vibration. This peak decreases in intensity with increasing carbonization temperature and KMnO 4 concentration. It indicates that the modified samples contain a small amount of carboxyl groups. This is caused by the presence of C = C moieties with red-shift characteristic of oxygen-rich surfaces leading to the diminution of the C = O peak (Ma et al., 2009). The peak at 1035 cm −1 (associated to C-O vibration), which is shifted from the normal value of 1056 cm −1 , also increased in intensity with increasing carbonization temperature due to oxidation by KMnO 4 . This is especially true at 700°C, where it has high intensity and a shoulder peak at 980 cm −1 . This is likely due to the overlapping of C-O stretching and Mn-OH vibration (Ma et al., 2009). The very weak peak at 872 cm −1 , which is shifted from the normal value of 897 cm −1 , corresponds to the δO-H bonds, which also result in a board peak at 3156-3227 cm −1 . The weak peak at 755 cm −1 is corresponding to the δC = O bond on the aromatic rings, which forms by substitution (Fávaro et al., 2007) after KMnO 4 modification. The peak at about 550 cm −1 in the spectra of 1-5 KPFW and 1-5KPFWC carbonized at 500°C are associated to the vibration of Mn-O bond (Zhang et al., 2015). In addition, the peak at 460 cm −1 , which is also associated to the νMn-O bond vibration, indicates that Mn has been successfully existed on surface of the waste materials (Wang et al., 2012). The weak peaks at about 600 cm −1 , appearing in samples prepared with carbonization at 500°C and above, increase in intensity with increasing carbonization temperature from 500 to 700°C. These peaks are attributed to C-O deformation vibrations, which likely arise from KMnO 4 addition (Zhang et al., 2013b). The OH and CO groups on the surface of samples are result of the KMnO 4 modification process (Ma et al., 2009), since KMnO 4 can oxidize unstable groups such as phenols, lactones, or lactols into carboxyls (Chen et al., 2012). These functional groups on KMnO 4 modified waste prepared with carbonization act as the initiation points for adsorbing external materials (Luo et al., 2013). Table 2 shows the BET surface area, pore volume and average pore size of PFW, PFWC (carbonized at 500°C), 1-3 KPFW and 1-3 KPFWC (carbonized at 500°C). It showed that the BET surface area and total pore volume of 1-3 KPFW and 3 KPFWC (carbonized at 500°C) are increased in comparison to PFW and PFWC (carbonized at 500°C). This was due to the addition of KMnO 4 . On the other hand, the average pore size decreased after KMnO 4 modification and carbonization. The results suggest that KMnO 4 treatment can improve the porosity of the waste materials by oxidation process, which can enlarge the pores of the materials (Wang et al., 2012). It can be seen that the BET surface area and total pore volume of 3KPFW is lower than of 1KPFW. But the average pore size trends in the opposite direction. This is because of the formation of oxygen-containing functional groups and MnO 2 blocking the pore entrances (Luo et al., 2013), thereby decreasing the surface area and pore volume of 3KPFW. However, 3KPFW is still oxidized to a higher extent than 1KPFW. Therefore, the average pore size of 3KPFW is higher than 1KPFW. For 3KPFWC obtained after carbonization at 500°C, the results are due to extensive thermal oxidation with KMnO 4 , which increases the surface area and pore volume. Meanwhile, the micropores and mesopores are formed to a greater extent resulting in decreased average pore size (Wang et al., 2012).

Effect of KMnO 4 Modification of Waste Materials on Fe 3+ Removal
The Fe 3+ removal by PFWC and 1KPFWC were compared (Fig. 4). The experiments were carried out using 0.1 g of PFWC or 1KPFWC loaded in 50 cm 3 of 5 mg dm -3 Fe 3+ solutions at pH 5 for 60 min contact time.
The results show that the Fe 3+ ion removal can be improved by material modification with KMnO 4 . This is due to the high content of MnO 2 and carboxylic functional groups in the modified materials. The modified material 1KPFWC has exhibited relatively high Fe 3+ removal capacity (98.2% or 2.455 mg g −1 ). The Fe 3+ ion removal capacity of 1KPFWC was approximately 4.5 times higher than that of PFWC. This indicates that KPFWC provides more binding sites for Fe 3+ than PFWC. This could be explained by Fe 3+ being attracted by electrostatic force to MnO 2 or other functional groups e.g., carboxylate groups, on KPFWC which can be protonated and deprotonated (Natkański et al., 2013). It is the result of complex formation of Fe 3+ and functional groups on modified material (Hu et al., 2015).  Effect of Contact Time on Fe 3+ Removal Figure 5 shows the effect of contact time (0-180 min) on Fe 3+ removal at pH 7 using 0.1 g of 1KPFWC in 50 cm 3 of 5 mg dm −3 Fe 3+ ion solution. The results show that the Fe 3+ adsorption rate is quite fast and reaches to equilibrium state after about 60 min. The initial rapid adsorption is due to the availability of high amount of vacant sites on surface of the adsorbent. At later stages the vacant sites which were adsorpted, becomes difficult more adsorpted. This is caused by repulsive forces between Fe 3+ ions adsorbed on the 1KPFWC surface and Fe 3+ ions in solution (Liu et al. 2015). Increasing of the adsorption efficiency with increasing the contact time has two functions. Firstly, the swelling of the adsorbent is increases, which increases its contact surface. The swelling property could result in higher hydrophilicity of the adsorbent (Natkański et al., 2013). Secondly, the contact between the swelled adsorbent and the metal ions is also increases, which increases the availability of interaction between the active functional groups on the surface of adsorbent and the metal ions in solution. As the result, the amount of metal ions adsorbed by surface functional groups increases, which increases metal ion uptake (Sadeek et al., 2015). It can be seen that the efficiency Fe 3+ adsorption is quite low. This is because this experiment was carried out at pH 7, which is not optimal condition. Based on these results, the contact time for further experiments was fixed at 60 min.

Effect of pH Value on Fe 3+ Removal Efficiency
The effect of pHs (2-9) on the efficiency of Fe 3+ ion removal by 0.1 g of 1KPFWC was investigated using 50 cm 3 of 5 mg dm −3 solution of Fe 3+ ions with 60 min contact time. The results of this experiment are shown in Fig. 6. The amount of Fe 3+ adsorbed on the 1KPFWC increased with increasing pH and reached maximum at pH of about 4.7. This value is similar to the value found for carboxymethylated chitosan hydrogels (Wang et al., 2008). The adsorption efficiency then decreased with further increases in pH. The Fe 3+ adsorption efficiency decreased gradually at pH values between 5-6. This was followed by a sharp decline at pH >7.
The adsorption reaction of MnO 2 groups on the material surface leading to Fe 3+ adsorption can be elucidated by Equation 7 and 8 (Hu et al., 2015): where, MnOH and MnO − are the protonated and deprotonated MnO 2 sites, respectively: The solution pH affects to the form of Fe 3+ ions in solution and the properties of the adsorbent surface in terms of dissociation of functional groups and surface charges (Liu et al., 2015). At very low pH values, the concentration of H + is much higher than that of Fe 3+ ions, which increases the competition between H + ions and Fe 3+ ions for the surface sites of the modified waste materials. Therefore, the functional groups on 1KPFWC would preferentially combine with H + making the surface positively charged, which is disadvantageous for the interaction between 1KPFWC and Fe 3+ ions. At higher pH values, the amount of protonated functional groups on the KMnO 4 modified waste materials would decrease and Fe 3+ ions would thus have more opportunities to compete with H + for binding with surface functional groups of 1KPFWC (He et al., 2014). This greatly improves Fe 3+ adsorption. Therefore at pH below 4.7, the adsorption capacity for Fe 3+ ions on 1KPFWC is low. The highest Fe 3+ adsorption capacity of 1KPFWC was observed to be about 90% at the pH value of 4.7. However, at higher pH values (>5), the adsorption efficiency for Fe 3+ ions decreases. This is especially true at pH >7, where the Fe 3+ ions in the solution exist in from of Fe(H 2 O) 6 ] 3+ , which can react with hydroxide ions to form the insoluble Fe(OH) 3 (Lee et al., 2006). The formation of this species inhibits the adsorption process significantly. As a result the Fe 3+ adsorption amount decreases. Thus pH value of about 5 is suitable for the adsorption of Fe 3+ ions. Figure 7 shows the Fe 3+ removal with 0.05-2.0 g of 1KPFWC loaded in 50 cm 3 of 5 mg dm -3 Fe 3+ solution, at pH 5 and with 60 min contact time. The results show that the Fe 3+ removal efficiency reaches the maximum (98.2%) with only 0.1 g of 1KPFWC. With loading in the range of 0.5-2.0 g the adsorption efficiency is maintained. On the other hand, with 1KPFWC loading below 0.05 g low Fe 3+ removal efficiency is occurred, due to the adsorption surface saturation (Lee et al., 2006) of 1KPFWC. Therefore, the loading amount of 1KPFWC was fixed at 0.1 g for subsequent experiments.

Isotherm of Fe 3+ Ion Adsorption
Fe 3+ ion adsorption isotherm studies were carried out with 0.05-2.0 g of 1KPFWC loaded in 50 cm 3 of 5 mg dm −3 Fe 3+ solution, at pH 5 and with 60 min contact time. Langmuir isotherm and Freundlich isotherm are used to correlate the Fe 3+ adsorption equilibrium data. Both the Langmuir isotherm (Fig. 8) and Freundlich isotherm (Fig. 9) were fitted using the experimental data, as was done by Yang et al. (2014). The R 2 values of these linear isotherms are nearly equivalent, with the values of 0.9919 and 0.9924 for Langmuir isotherm and Freundlich isotherm, respectively. Thus, the Freundlich adsorption model appears slightly better than the Langmuir model. It was shown that the Fe 3+ ion adsorption occurrs on the heterogeneous surface of 1KPFWC. However, at low Fe 3+ concentration it only takes place in a monolayer without interaction between Fe 3+ ions and with no transmigration on the surface of 1KPFWC (Liu et al., 2013). It can be seen from the SEM image of 1KPFWC ( Fig. 3d and e) that its surface is rough with cracks and a large number of non-uniform small particles and many functional group types (Fig. 3), which indicates a heterogeneous surface (Dissanayake et al., 2016). According to the Langmuir equation, the maximum Fe 3+ adsorption capacity (q max ) and Langmuir constant (K L ) values were calculated by linear regression. The theoretical maximum adsorption capacity value and K L for Fe 3+ ions on 1KPFWC were 25.25 mg g −1 and 0.2071 dm 3 g −1 , respectively. An estimation of the dimensionless factor R L from the values of C 0 and K L was obtained using the relationship R L = 1/(1+ K L C 0 ). The value of R L is 0.4912, which in the range 0<R L <1. Therefore, 1KPFWC appears to be suitable for Fe 3+ ions adsorption (Liu et al., 2013). In addition, Freundlich model provides two parameters K F and n from the slope and intercept of the plot of Log Qe against Log Ce in Fig. 9. The K F and n values are 125.92 dm 3(1/n) mg (1-1/n) g −1 and 0.5008, respectively. The K F value is quite high, which represents a high Fe 3+ ion adsorption capacity on the modified waste carbon material. But n value is< 1, which indicates relatively low heterogeneity of the adsorbent (Mahmoud, 2015).

Conclusion
The KMnO 4 modified carbon materials obtained from waste of pineapple fiber production processing were characterized by SEM-EDS, XRD, FTIR and BET. The results from SEM-EDS show a rougher surface with cracks and a large number of small particles with major content of C, O, Si, K and Mn elements in the all of KPFWC materials. The results of the XRD confirmed that MnO 2 was deposited on the surface of KMnO 4 modified waste carbon materials. In addition, the results from the FTIR analysis revealed that the OH, CO and MnO 2 groups are the major functional groups on the surface of KMnO 4 modified carbon materials. The BET surface area and total pore volume of modified waste carbon materials increases with increasing weight % of KMnO 4 used in the modification and increasing carbonization temperature. On the other hand, the average pore size follows an inverse trend. This is due to the formation of oxygen-containing functional groups and MnO 2 , which block the pore entrances. It was shown that the Fe 3+ adsorption on modified waste carbon materials reaches equilibrium within 60 min and that the amount of Fe 3+ adsorbed increases with increasing pH up to the optimal value of 4.7. The adsorption efficiency then decreases, with a sharp decrease observed at pH values >7. Both Langmuir isotherm model and Freundlich isotherm model can be fitted to the Fe 3+ adsorption data with high R 2 values of 0.9919 and 0.9924, respectively. The theoretical maximum adsorption capacity of Fe 3+ on the modified waste carbon materials was estimated as 25.25 mg g −1 . The materials also show some heterogeneity on their surfaces, which is favorable for Fe 3+ ion adsorption. Therefore it can be concluded that pineapple fiber leaf wastes from pineapple fiber production processing, which are cheap, easily available and can be used more economically on large scale, have high potential for conversion to Fe 3+ ion adsorbents by modify with KMnO 4 .