Kinetic and Isotherm Modelling of the Adsorption of Phenolic Compounds from Olive Mill Wastewater onto Activated Carbon

The olive oil industries generate huge amounts of wastes with great environmental impact (1). Olive oil wastewater, as one of the main olive oil extraction residues is an acidic liquid (pH=5–5.5) that contains traces of olive pulp, pectins and oil. Moreover, the presence of organic substances, particularly polyphenols, has been reported to be associated with the negative eff ects of this waste on soil, aquatic and air ecosystems (2). Large amount of phenolic compounds, free fatt y acids and inorganic salts are known as the major causes of the high phytotoxicity and poor biodegradability of olive mill wastewater (3). As a consequence, increasing att ention has been given to fi nding suitable methods for the control of this pollutant. Several works have concentrated on this problem taking approaches such as treatment with calcium hydroxide or fermentation with microorganisms, or even co-composting with olive stone wooden residue (4). All of these methods lead to degradation of phenolic compounds in this effl uent in an irreversible and nonrecoverable manner.


Introduction
The olive oil industries generate huge amounts of wastes with great environmental impact (1).Olive oil wastewater, as one of the main olive oil extraction residues is an acidic liquid (pH=5-5.5)that contains traces of olive pulp, pectins and oil.Moreover, the presence of organic substances, particularly polyphenols, has been reported to be associated with the negative eff ects of this waste on soil, aquatic and air ecosystems (2).Large amount of phenolic compounds, free fatt y acids and inorganic salts are known as the major causes of the high phytotoxicity and poor biodegradability of olive mill wastewater (3).As a consequence, increasing att ention has been given to fi nding suitable methods for the control of this pollutant.Several works have concentrated on this problem taking approaches such as treatment with calcium hydroxide or fermentation with microorganisms, or even co-composting with olive stone wooden residue (4).All of these methods lead to degradation of phenolic compounds in this effl uent in an irreversible and nonrecoverable manner.
On the other hand, the importance of polyphenols as bioactive compounds with many interesting activities has been widely discussed in the past years (5).Phenolic compounds act as antioxidants (6)(7)(8)(9), neurosedative, anti-infl ammatory, antiviral and anticancer agents (10).It has been indicated that verbascoside and its analogues fr om olive fruit can consistently contribute to the intake of antioxidants in the diet (11).Therefore, an appropriate application of olive oil residues could not only improve the economic status of olive oil producers but could also reduce such an environmental problem.Thus, the possibili-ty of recovering an extract enriched with phenolic compounds, obtained from a low-cost and widely available by-product, is of great interest especially in the Mediterranean area.
Recently, a large amount of literature has been increasingly devoted to the study of adsorption for the removal of aqueous organic species, such as substituted and unsubstituted phenols using activated carbon (AC) (12)(13)(14).AC has perfect adsorption ability for relatively low molecular mass organic compounds, such as phenols.Adsorption onto AC is the physical and/or chemical process in which a substance is accumulated at an interface between two phases.Removal by AC is obtained as a result of att ractive interaction of the molecules with the walls of micro-or mesopores with dimensions comparable to those of the molecules (15).AC has been utilised as an effi cient sorbent for odour removal, solvent recovery, decolourisation, dechlorination, ozone annihilation, H 2 S/ CS 2 removal, gold recovery, fi ltration, fuel gas cleaning, industrial wastewater treatment, drinking water conditioning, etc. (16).Although several papers about the adsorption of phenolic compounds from aqueous media on AC have been reported, there are few publications that discuss the characteristics of adsorption from multicomponent solutions.The importance of polyphenols as bioactive compounds encouraged us to recover polyphenols from wastewater by the adsorption onto activated carbon, because the process of regenerating the adsorbent by desorption of the organic compounds poses an important problem for the current studies since they have high affi nity towards the sorbent surface.Several techniques such as chemical (17), thermal (18) degradation and ultrasound (19) have been used for the desorption of phenols from AC. NaOH solution proved to be effi cient in recovering the adsorbed phenolics from AC by formation of sodium salt of phenols, which may facilitate desorption.Özkaya (14) and, more recently, Ena et al. (20) reported that 0.15 M NaOH is required for quantitative desorption of phenols from AC.The desorbed phenolic fraction was more than 60 % when using 0.15 M NaOH eluent.Diff erent factors aff ect adsorption capacity such as the physical nature of the pore structure of the adsorbent, functional groups, the nature of the adsorbate, molecular mass, size and solution conditions (pH and ionic strength).Changes in surface charges with pH can aff ect the reversible and irreversible uptakes of phenolics.This may be due to the increase in activity of phenolic compounds with decreasing pH and this infl uence on reversible uptake should be minimal at compound pH values well below the solute pK a .It can be concluded from the dependence of the irreversible adsorption that oxidative coupling of phenolic compounds is more facile in alkaline than in acidic media (21).
The aim of this work is to study the adsorption capacity of a cost-eff ective and environmentally compatible adsorbent for the removal of phenolic compounds present in olive mill wastewater.In this study commercial activated carbon was used.The eff ect of some operating conditions, such as adsorbent quantity (1 to 8 g per 100 mL) and temperature (10, 25 and 40 °C), were investigated.The kinetic data were fi tt ed with pseudo-fi rst and -second order kinetic models.The Langmuir and Freundlich models were used to describe the equilibrium isotherms.The results obtained from the present study will be used in further research of recovery of adsorbed phenolic compounds.Aft er purifi cation processes, these valuable compounds can be used in food, cosmetic and pharmaceutical industries.

Reagents
Methanol, acetonitrile, sulphuric acid, acetic acid (HPLC grade), Folin-Ciocalteu reagent and caff eic acid standard were purchased from Sigma-Aldrich Chemical Co (St. Louis, MO, USA).Glucose and fructose standards were obtained from Merck KGaA (Darmstadt, Germany).Standard stock solutions were prepared, wrapped in aluminium foil and stored at -20 °C.
The commercial activated carbon (AC; Sigma-Aldrich Chemical Co) was used in the batch experiments without any pretreatment.The cost of commercial AC can vary from $800 to 2000 per metric tonne depending on the quality of the product.The AC used in this study has the pore size of 2.98 nm, specifi c surface area (Brunnauer, Emmett and Teller (BET) theory) 920.3 m 2 /g, maximum moisture 2.0 %, iodine number 800, bulk density 0.336 kg/L, pH=9 in water extract, ash 6 %, methylen blue adsorption of 12.5 g per 100 g per min and molasses decolourising number 440.Powder X-ray diff raction (XRD; Philips PW 1830 generator, Eindhoven, the Netherlands) and Fourier transform infrared spectroscopy (FTIR) were used, and the sample patt erns are illustrated in Fig. 1.For FTIR analysis the matrices were dispersed on a pressed disk of KBr and analysed by a FTIR spectrometer (Nicolet 380, Thermo Scientifi c, Waltham, MA, USA), the spec- trum was analysed using the Omnic™ Lite soft ware (Thermo Electron Corporation, Madison, WA, USA).X--ray patt ern (Fig. 1a) shows the amorphous form of the used sample.Fig. 1b indicates the functional groups and surface properties of the adsorbent determined by FTIR spectra.The spectra of the adsorbent were measured in the wavenumber range of 4000-400 cm -1 .The most intense peak at 3425 cm -1 can be assigned to alcohol or phenol -OH stretching.Peaks at 1968 and 2850 cm -1 are attributed to symmetric and asymmetric CH 2 stretching, respectively.The peak around 1081 cm -1 shows the appearance of C-O stretching on the carbon surface.

Olive oil wastewater and characteristics
Olive mill wastewater from the treatment of Taggiasca cultivar obtained from three-phase oil extraction decanter was supplied by an olive oil production plant in Imperia, Liguria region (Italy).Adequate quantities of samples were centrifuged by a PK 131 centrifuge (ALC, Alberta, Canada) at 6000×g for 10 min in order to separate suspended solids.
The value of chemical oxygen demand (COD) was estimated using the method described by Knechtel (22).Simple carbohydrates such as glucose and fructose were determined directly by high-performance liquid chromatography (1100 Series, Hewlett Packard, Palo Alto, CA, USA) as reported by Garrote et al. (23), using a refractive index detector and a Supelcogel H59304-U column (Sigma-Aldrich Corp., Bellefonte, PA, USA).Samples were fi ltered through 0.45-μm membranes (Millipore, Billerica, MA, USA).Separation was achieved using 0.005 M H 2 SO 4 as mobile phase at a fl ow rate of 0.5 mL/min at 50 °C.
Total polyphenol (TP) concentration was measured using the Folin-Ciocalteu assay (24,25).Briefl y, 4.8 mL of pure water, 0.2 mL of sample, and 0.5 mL of Folin-Ciocalteu reagent were mixed and 1 mL of a 20 % sodium carbonate solution was added.Pure water was added to reach the fi nal volume of 10 mL.Solutions were mixed and allowed to stand at room temperature in the dark for 1 h.Sample aliquots were used for the determination of total phenolic concentration using an UV-VIS spectrophotometer (Perkin Elmer, Wellesley, MA, USA) at a wavelength of 725 nm.TP was standardised against caff eic acid and expressed as mg of caff eic acid equivalents (CAE) per mL of wastewater.The method response was described with the following linear equation within the range of 0.1-1.0mg/mL with R 2 =0.9962.
The concentration of o-diphenols (OD) in the methanolic extract, also expressed as CAE in mg/mL, was determined by the molybdate method (26): 0.2 mL of extract was diluted with water to reach 1.0 mL, and then 1.0 mL of 0.1 M phosphate buff er (pH=6.5)and 2.0 mL of 5 % Na 2 MoO 4 •2H 2 O were added.The content was mixed and the absorbance was measured aft er 15 min at 350 nm against a blank reagent using the same spectrophotometer as above.The calibration curve was made with standard solutions of caff eic acid in the range of 0.01-0.25 mg/ mL, giving the following equation:

Adsorption experiments
In order to simplify the adsorption process, batch experiments were carried out in 100-mL Erlenmeyer fl asks containing (25±0.2) mL of centrifuged wastewater and different quantities of adsorbent (0.25, 0.5, 1, 1.5 and 2 g).The fl asks were placed at 10, 25 and 40 °C, using a thermo-regulated water bath, model SWB25 (Enco, Spinea, Venice, Italy) in order to study the infl uence of temperature on sorption.When the desired temperature was reached, a known amount of adsorbents was added to each fl ask and the solutions were agitated on a rotary shaker at 200 rpm.At predefi ned time intervals (10,20,30,40,50, 60, 80, 100 and 120 min), a constant quantity of sample (1.0 mL) was taken.The adsorbent was separated from the samples using membrane fi lters of 0.2 μm (Millipore).Residual concentration of total polyphenols in each sample was determined using Folin-Ciocalteu method as described above.All experiments were conducted in duplicate, and the error was always less than 6 %, calculated by dividing standard deviation by average of diff erent quantities of polyphenols and expressed as percentage.Each adsorption experiment was conducted in duplicate while total polyphenol concentration analyses were determined in triplicates.

Sorption kinetics and isotherms
Adsorption kinetics was evaluated by applying the Lagergren model ( 27) and the pseudo-second-order equation of Ho and McKay (28).The linearized form of the model of Lagergren gave the following equation: /3/ where q is the amount of TP (expressed as CAE in mg) adsorbed per g of adsorbent at a given time (t) and was used in plots of log (q e −q) vs. time to estimate the fi rst-order-rate constant of sorption, k 1 (min -1 ).
The capability of activated carbon as sorbent was evaluated in terms of sorption capacity at equilibrium (q e ), expressed in g of CAE per g of activated carbon: /4/ where γ 0 and γ e are the initial and equilibrium liquid--phase phenol concentrations (expressed as mg of CAE per L), respectively, V is the volume of the wastewater (mL) and m is the mass of activated carbon used (g).The effi ciency of the removal of phenolics (Y e ) was expressed in percentage: /5/ The pseudo-second-order model (29), based on the sorption capacity of the solid phase and consistent with the chemisorptions mechanism, is described by the following equation: /6/ Plots of t/q vs. time at diff erent activated carbon quantities were used to estimate the second-order-rate constant of sorption, k 2 (g of AC per mg of CAE per min).
e e 1 t t q k q q Several models have been published in the literature that describe experimental data of adsorption isotherm (14-16).The Langmuir (29) and Freundlich models (30) are simple and most frequently applied.The Langmuir isotherm is based on the assumption of monolayer adsorption onto a surface containing fi nite number of adsorption sites of uniform energies of adsorption with no transmigration of adsorbate in the pores of the adsorbent surface.The linearized form of the Langmuir isotherm model (Eq.7) was used to determine equilibrium data: /7/ where q m is the maximum sorption capacity (mg of CAE per g of AC) and K L is the Langmuir equilibrium constant (mL per mg of CAE).
The Freundlich model (Eq.8) assumes heterogeneous surface energies, where adsorption energy varies as a function of the surface coverage due to variation in the adsorption heat (16)./8/ in which K F (mg of CAE per g of AC)(mg of CAE per L) n and n (dimensionless) are constants for a given sorbate and sorbent, respectively, at a particular temperature, which are related to the sorption capacity and intensity, respectively.
For bett er understanding of the mechanism of adsorption of phenolics on activated carbon, in this work the experimental data of the adsorption of phenolics were fi tt ed and compared with the well-known Freundlich and Langmuir models and the goodness of fi t was evaluated.The Freundlich model is widely applied (31)(32)(33) in heterogeneous systems especially of organic compounds and highly interactive species on activated carbon and molecular sieves.The parameters and constants appearing in both equations were estimated by linear regression, by plott ing γ e /q e vs. γ e in the former case and ln q e vs. ln γ e in the latt er.

Eff ect of adsorbent quantity and temperature
The physicochemical characteristics of olive oil wastewater are represented in Table 1.
The concentration of the used AC is an important factor that controls adsorption (34).It is well known that temperature is an additional factor that greatly infl uences any sorption process.Therefore, tests were performed at three diff erent temperatures, room temperature (25 °C), above (40 °C) and bellow (10 °C) it.In order to observe the simultaneous eff ect of the temperature and adsorbent quantity on the adsorption of polyphenols, adsorbent concentration was varied from 1 to 8 g per 100 mL.Diff erent tests were performed using the same quantity of wastewater (25 mL) and constant phenolic concentration (3.0 mg/mL), but varying the quantity of activated carbon (0.25, 0.5, 1.0, 1.5 and 2.0 g).Adsorption trends at diff erent temperatures can be observed in Figs. 2 and 3.
As illustrated in Fig. 3, the initial velocities and adsorption percentage (Y) enhanced as the amount of actived carbon increased.This can be explained by the presence of multiple sites available initially.
At each measured temperature, adsorption capacity (q e ) decreased with the increase of the AC concentration.At higher concentration of activated carbon (8 g per 100 mL), the saturation of surface-active sites with the adsorbate molecules causes no more signifi cant adsorption (34).These fi ndings are in accordance with those obtained by Qadeer and Rehan (17), who observed that increasing the concentration of commercial activated carbon from 0.02 to 0.14 g per 10 mL of wastewater with initial phenolic concentration of 0.01 g/L resulted in 5 % increase in the adsorption of phenols.Moreover, Özkaya ( 14) demonstrated that increasing AC concentration from 0.5 to 9.0 g/L re-  Signifi cant changes were not observed at diff erent applied temperatures at the AC concentration of 8 g per 100 mL, for example adsorption capacity was 35.8, 35.4 and 36.1 mg/g at 10, 25 and 40 °C, respectively.Zogorski and Faust (35) reported that adsorptive capacity of phenols on granular activated carbon was increased with the decrease of temperature.
It can be observed in the literature that H 3 PO 4 (36), sewage sludge (37), fl y ash (38) and activated carbon from agricultural by-products (39) used in the petroleum coke treatment can be adsorbents of phenolic compounds with high adsorbent capacities (158, 94, 67 and 35.0 mg/g).

Adsorption equilibrium modelling
The Freund lich and Langmuir models were applied to the equilibrium data obtained for total polyphenol adsorption on AC.Table 2 describes the adsorption constants for both models obtained from the isotherms, and the linear regression coeffi cient for each evaluated condition.
According to the Langmuir model, the maximum values of adsorption capacity of AC were 126.6, 156.2 and 151.5 mg/g at 10, 25 and 40 °C.The adsorption increased with the increase in temperature from 10 to 25 °C, refl ecting the endothermic nature of the reaction, until equilibrium was obtained and then it decreased.The use of endothermic processes for the adsorption of organic compounds on activated carbon has been reported previousely (16,40,41).An increase in temperature from 25 to 40 °C caused a decrease in the amount of phenols adsorbed onto the surface of the adsorbent.At high temperature, the thickness of the boundary layer is expected to decrease due to the increased tendency of the phenolic compounds to escape from the surface of the adsorbent to the solution phase (due to the increase in kinetic energy of the ions), hence there are weak adsorption interactions between the adsorbent and the adsorbate (42).The Langmuir equilibrium constant (K L ), which is related to the affi nity between phenolics and sorbent, was lower at 25 °C (1.60 mL/mg) when compared to the other temperatures, indicating a bett er affi nity between the AC and these compounds.
From the data reported in Table 3, it can be observed that the Langmuir model fi tt ed bett er the equilibrium data at temperatures lower than 40 °C (R 2 =0.9401 at 10 °C and 0.9709 at 25 °C) when compared with Freundlich model (R 2 =0.7667 at 10 °C and 0.9511 at 25 °C), thus confi rming the poor applicability of Freundlich isotherm for this type of sorption process.On the other hand, at 40 °C, the Freundlich isotherm seems to explain bett er the experimental data (R 2 =0.9310 compared to 0.9194 in the case of Langmuir model).Qadeer and Rehan (17) as well as Özkaya ( 14) also confi rmed bett er fi tt ing of experimental data for adsoption of phenolic compounds onto activated carbon by Langmuir model than by the Freundlich model.

Adsorption kinetics modelling
The kinetics of the adsorption of phenolic compounds onto AC was investigated in order to understand the mech-  anisms of sorption.The rate constants, the experimental and calculated equilibrium sorption capacities and the linear regression coeffi cients were obtained at all tested temperatures, and only the satisfactory results of the second-order model are summarized in Table 3.The correlation coeffi cients obtained by the Lagergren model were found to be less than 0.8750, while those of the second--order model were higher than 0.9204 at all temperatures.These results, which are in agreement with those previously reported for the sorption of phenolics by activated carbon derived from agricultural waste material ( 16), demonstrate that only the latt er model is able to satisfactorily describe the kinetic behaviour of the sorption of polyphenols by AC.Table 3 shows that the values of rate constant, k 2 , increased from 0.0010 to 1.6131 at 10 °C, from 0.0025 to 0.1213 at 25 °C and 0.0010 to 0.0290 at 40 °C with the increase of AC concentration from 1 to 8 g per 100 mL.
The theoretical values of q e (q e (theoretical)) were calculated using the second-order equation and it was observed that these values are in satisfactory agreement (R 2 =0.9263) with the experimental q e (q e (experimental)) values at diff erent temperatures and AC concentrations (Fig. 4).These observations suggest that the studied sorption systems follow the second-order kinetic model.Similar observation was reported in the literature for diff erent adsorption processes (16,28,43,44).

Conclus ions
Batch sorption tests were performed using diff erent concentrations of activated carbon (1, 2, 4, 6 and 8 g per 100 mL) and temperature (10, 25 and 45 °C) for the removal of phenolic compounds from olive mill wastewater.Under optimum conditions (8 g of activated carbon per 100 mL), the maximum sorption capacity of activated carbon was 35.8 mg/g at 10 °C, 35.4 mg/g at 25 °C and 36.1 mg/g at 40 °C.The equilibrium data of total polyphenol sorption were fi tt ed satisfactorily by the Lang muir isotherm at three diff erent temperatures.The pseudo-second-order model (R 2 >0.9204) explained bett er the adsorption kinetics as compared to the pseudo-fi rst-order model (R 2 <0.8750).The results of this study suggest that activated carbon can be used effi ciently as an adsorbent for the removal of phenolic compounds from olive oil industry effl uent.Pretreated wastewater has a potential as a clean substrate for the bioethanol production with the additional benefi t of its use regarding the environmental and economical aspects.In addition, the adsorbed phenolic compounds can be recovered and aft er purifi cation used in food, cosmetic and pharmaceutical industries.(theoretical)/(mg/g) q e ( a l)/(mg/g) experiment Fig. 4. Correlation between sorption capacities at equilibrium, expressed in mg of CAE per g of AC, calculated by second--order-rate model (q e (theoretical)) and those obtained by experiments (q e (experimental)).CAE=caff eic acid equivalents, AC= acti vated carbon

Fig. 3 .
Fig. 3. Capability of activated carbon (AC) as sorbent in terms of: a) removal effi ciency (Y e ), or b) sorption capacity (q e ) at equilibrium of phenolic compounds (expressed as CAE in mg per g of AC) from olive mill wastewater from Taggiasca cultivar treatment at diff erent temperatures and adsorbent concentrations (in g per 100 mL):  1.0,   2.0,   4.0,   6.0 and  8.0.CAE=caff eic acid equivalents

Table 1 .
The physicochemical characteristics of olive mill wastewater from Taggiasca cultivar treatment Total polyphenols and o-diphenols are expressed as caff eic acid equivalents.COD=chemical oxygen demand

Table 2 .
Isotherm parameters of Langmuir and Freundlich models of the sorption of phenolic compounds (expressed as mg of CAE per g of AC) from olive mill wastewater from Taggiasca cultivar treatment onto activated carbon at diff erent temperatures

Table 3 .
Second-order kinetic parameters estimated for the sorption of phenolic compounds (expressed as mg of CAE per g of AC) from olive mill wastewater from Taggiasca cultivar treatment onto activated carbon at diff erent temperatures