Chemical modi ﬁ cation, characterization, and application of chicken feathers as novel biosorbents

The current work exclusively pertains to preparing arsenic removing biosorbents using chicken feathers (CF) as a raw material. CF consist of keratinous proteins with many functional groups, and this paper contributes to the debate on how functional groups, especially – COOH, – NH 2 and – S – S – , interact with arsenic species before and after their modi ﬁ cation through sorption phenomena. Chemically modi ﬁ ed CF biosorbents were investigated regarding their ability to remove As( III ) from water. The modi ﬁ cation results suggest that reactions occur mainly on the surface with noticeable changes on reactive sites in the interior of the modi ﬁ ed chicken feathers (MCF). To prepare the modi ﬁ cations labelled as MCF-I and MCF-II, the feathers were treated with aqueous NaOH and sodium sul ﬁ te, respectively, to change their structure and morphology. Then, maleimide terminated poly( N -isopropylacrylamide) (PNIPAM) was added to the reaction mixture to enhance the mass and reactivity of the biosorbents. However, MCF-I and MCF-II displayed a relatively low sorption capacity ( $ 25 – 50%) for the removal of arsenic, but a higher capacity than raw CF. The methyl alcohol supported modi ﬁ cation MCF-III, on the other hand, exhibited a signi ﬁ cant performance in segregating negatively charged arsenic species from water. Therefore, esteri ﬁ ed carboxylic groups ( – COOH) in keratin were identi ﬁ ed as particularly e ﬀ ective promoters of the arsenic uptake. Around 80 – 90% sorption capacity was observed within the ﬁ rst hour of contact between the MCF-III biosorbent and arsenic polluted water. Characterizations such as FTIR, XRD, DSC, TGA and SEM supported the modi ﬁ cation of chicken feathers and the subsequent e ﬀ ect of this modi ﬁ cation on the sorption, particularly when MCF-III was applied as a biosorbent. The role of pH was highly signi ﬁ cant for changing the surface behaviour, and high uptake was observed at low pH for MCF-I, MCF-II and MCF-III. The kinetics of the biosorption capacities of CF, MCF-I, MCF-II and MCF-III were also evaluated and compared at di ﬀ erent pH. The experimental kinetic data for MCF-III at variable pH followed pseudo second order model ﬁ ts and are typical for chemisorption. Similarly, the Freundlich isotherm model supports our data with a high correlation value ( R 2 ) and demonstrates both monolayer and multilayer biosorption.


Introduction
Arsenic and its compounds are extremely fatal and carcinogenic to all living organisms.Arsenic in exceeding amounts, once ingested, can cause severe nausea and gastrointestinal symptoms, and thus, the toxicity of arsenic is an established and scientic fact. 1,23][4] As(III), on the other hand, has a tendency to bind to the sulydryl (-SH) groups of dehydrogenase enzymes, i.e. pyruvate, dihydrolipoate and a-ketoglutarate, causing severe metabolic complications, and at times, cellular mutagenesis. 5Arsenic is naturally found on the Earth's crust, mostly in the common mineral arsenopyrite (FeAsS). 3The arsenic concentration in soil may vary from 1.0 mg kg À1 (apatite, calcite and uorite samples) to 77 000-126 000 mg kg À1 (pyrite or arsenopyrite samples). 6Chemical industries, especially those related to electronic devices, 7,8 use small amounts of arsenic and its compounds to manufacture components for cosmetics, wood preservatives, laser equipment, pesticides and glasses.Mineral and metallurgical industries contribute to arsenic containing wastes.The contaminated solid and liquid waste is a potential source of surface and groundwater pollution.Arsenic contaminated groundwater has been reported worldwide, and especially Canada, the USA, Chile, Argentina, India and Bangladesh are places where the natural leaching of As-enriched soil and rock has been the main cause of arsenic contamination. 3,6,7][11] Statistical data show that more than 25 million people have been exposed to water with elevated arsenic concentrations, i.e. $50 mg L À1 in Bangladesh and the West Bengal province of India. 12The variable oxidation state of arsenic species inuences its stability and immobilization in natural environments or under rectifying measures.It mainly exists in two water soluble oxidation states of As(III) and As(V) derived from arsenous acid (H 3 AsO 3 ) and arsenic acid (H 3 AsO 4 ), respectively.Of these two states, As(III) not only shows considerably higher toxicity and solubility in water but is also important with regard to its increased mobility in soils.Arsenical mobility in natural waters is dependent on the pH, Eh conditions and presence of other chemical species. 6It has been determined experimentally that the oxidation of As(III) to As(V) is kinetically very slow, which results in the co-existence of both species even under oxic conditions. 13Bacterial activity causes the reduction of As(V) species, which causes toxicity in biological environments. 1,14part from inorganic forms, arsenic also exists as organic species such as dimethyl arsenic acid at very low concentrations in natural waters. 157][18][19][20][21][22][23] These processes have, however, numerous drawbacks, which include selective or partial metal removal, and high capital and operational cost with increased disposal of residual metal sludge, making them unsuitable and unsustainable for small scale industries.Adsorption techniques have long been used in water and waste water industries, but the real challenge in such techniques is the employment of inexpensive, excessively available and effective adsorbents.5][26][27] However, the high price for the preparation and regeneration of these adsorbents encouraged the application of biosorbents for arsenic removal from water.An unspe-cic ion exchange reaction mechanism is usually involved in the biosorption uptake.For instance, negatively charged arsenate or arsenite species are potentially attracted by reactive sites of amino groups to form adsorptive complexes. 28The continuous and sustainable growth of the poultry industry and an ever increasing demand for poultry consumption is leading to an oversupply of byproducts.The efficient utilization of byproducts is a challenge, but one which is extremely important to overcome with this crucial worldwide industrial waste.0][31] Poultry feathers contain about 90% protein (keratin) and are a cheap and renewable source of protein bers.There is very limited use of feathers in industrial applications.At present, in addition to animal feed 32 and a few applications in composites and other products, 33,34 the majority of the poultry feathers are disposed of in landlls.[37][38][39][40] Adsorption results show that aer chemical modication, CF have a higher dye/ion uptake as compared to their unmodied form. 38We studied structural changes during modication and exclusively extended the use of modied CF by designing arsenic removal lters.Chemically, several modications of CF were generated by treatment with different doping agents.The extent of modication of the lter containing material was evaluated by characterization techniques such as SEM, FTIR, DSC, TGA and XRD, and the adsorption efficiency was evaluated using kinetic and isothermal studies of the biosorption.

Feather processing
The CF supplied by the Poultry Research Center of the University of Alberta were washed several times with soapy hot water.The washed feathers were dried by spreading them in a closed fume hood for one week to evaporate the water and thereaer, they were kept in a ventilated oven for 24 h at 50 C to completely remove the remaining moisture.The hollow sha, calamus, was trimmed from the vane of the CF with a pair of scissors.The processed CF were ground using a Fritsch cutting mill (Pulverisette 15, Laval Lab. Inc., Laval Canada) with a sieve insert size of 0.25 mm.The batches of ground CF (30 g each) were further treated in a Soxhlet (extraction tube with 50 mm internal diameter) for 5 h with 250 mL of petroleum ether.Aer evaporating the petroleum ether, the dried CF were stored in a desiccator at room temperature until they were used for the experimental work.

Modication of chicken feathers
The chemical modication of the CF was carried out by keeping in mind three important functional groups in keratinous protein such as sulydryl (SH) groups from reduced disulde (-S-S-) linkages, amino groups (-NH 2 ) and carboxylic groups (-COOH), and accordingly, we divided our experimental work into three types of modications i.e.MCF-I, MCF-II, MCF-III.

Treatment with alkaline aqueous solution (MCF-I)
The CF (10 g) were added to 100 mL of a 0.1 M L À1 aqueous NaOH solution and the mixture was stirred at 200 rpm at 70 C for four hours.Aer cooling the mixture to room temp, the pH was maintained at 7.5 and then the desired amount of NIPAM was added to the reaction mixture in an inert atmosphere under a stream of N 2 gas.The reaction mixture was le overnight.The modied material was ltered, washed with distilled water, dialysed for 48 hours using a dialysis membrane with a molecular weight cut off (MWCO) at 8000 to remove any unreactive contents, and then freeze dried to remove the water.
Sodium sulte supported modication (MCF-II) 0.5 g of Na 2 SO 3 were added to a mixture of 100 mL water containing 10 g CF, 10% by volume of each EDTA and tris base in the presence of urea in a double necked ask equipped with a magnetic stirring bar.The reaction mixture was stirred for about 4 hours at 40 C. The mixture was cooled to room temperature and the pH was maintained at 9.00.NIPAM was added slowly to the ask aer deoxygenating it with nitrogen gas.The reaction mixture was le for 24 hours at room temperature and constant stirring at 100 rpm.The CF solution was washed with water and submitted to dialysis.The dialysis was performed at room temperature over 48 hours using a dialysis membrane (MWCO 8000) and freeze dried.

Treatment with methyl alcohol (MCF-III)
In this modication, we used 10 g CF, 6% (v/v) CH 3 OH and 2% (v/v) HCl in a 250 mL closed double necked ask and placed it on a heating plate at 80 C with a 150 rpm stirring rate for ve hours.The reaction mixture was ltered, washed with distilled water and subjected to dialysis followed by lyophilisation for three days.

Characterizations
FTIR spectra of both the modied and unmodied samples of solid feathers in KBr pellets were recorded with a Thermo Nicolet 750, Madison, WI, USA.All spectra were collected over a frequency range of 4000-650 cm À1 , 32 scans and 4 cm À1 resolution by averaging of two replicate measurements for each sample.The Thermo Scientic OMNIC soware package (version 7.1), and second derivative spectra were obtained to support the initial identication of band positions.The band positions obtained from the above steps were then used as the initial guess for curve-tting of the original spectra with Gaussian bands.DSC analysis of all samples was carried out under a continuous nitrogen purge on a Perkin-Elmer (Pyris 1, Norwalk, CT, USA) calorimetric apparatus.A sample of pure indium was used for the heat ow and temperature calibration of the instruments.A temperature range of 25-250 C at 10 C was set for the DSC analysis of all samples.TGA was performed on a Perkin-Elmer (Pyris 1, Waltham, MA, USA) thermogravimetric analyzer in a temperature range of 25-600 C, at heating rate of 10 C min À1 under a nitrogen atmosphere.X-ray powder diffraction patterns were recorded using a Rigaku Ultima IV, Geigerex Powder Diffractometer with Cu-Ka radiation (l ¼ 0.154 nm).Scanning electron microscopy (SEM) images were scanned with a Philips-FEI model Quanta 20.For easier comparison, intensities were normalized in all spectra at 10 , and the spectra are offset.

Batch sorption experiments
In a batch process, each modied biosorbent (MCF-I, MCF-II and MCF-III) was employed besides the untreated chicken feathers (CF) to assess the "As(III)" removal efficiency of the prepared material.Aqueous arsenic solutions with concentrations in the range of 100-800 mg L À1 were prepared by diluting 0.05 M NaAsO 2 .The tests were conducted by using 1.0 g of each modied biosorbent in a ask containing 100 mL of an aqueous solution of arsenic with a predetermined concentration.Each sorption experiment was run for ten minutes and the suspensions were agitated in a shaker at 20 C and nally ltered.Untreated chicken feathers were tested in the batch experiments as well.The residual arsenic concentrations in the ltrates were determined by a Perkin Elmer Elan 6000 quadrupole ICP-MS.The instrumental conditions were as follows: RF power of 1200 W, dual detector mode, blank subtraction applied subsequent to internal standard correction, measurement units cps (counts per second), auto lens on, four points calibration curve (0, 0.005, 0.010, and 0.020 ppm for As), typical count rate for a 10 ppb arsenic (As) solution: 150 000-200 000 cps.The sample uptake rate was approximately 1 mL min À1 with 35 sweeps per reading, 1 reading per replicate and 3 replicates.The dwell times were 100 ms for As.The relative standard deviation for As is between 5 and 10% of the reading.
The sorption capacity (q) (mg g À1 ) of the biosorbents was calculated from the initial concentration (C i ) (mg L À1 ), and the nal concentration (C f ) (mg L À1 ) of the arsenic metalloid in solution was calculated according to the following equation: where "V" is the volume of the solution (L) and "m" is the dry mass of the biosorbent (g).

Adsorption kinetics
The anticipation of the kinetics is essential for designing sorption systems and determining factors responsible for the rate of reaction.The nature of the sorption process depends on the physicochemical properties of the sorbent, the experimental conditions, and the solution pH.In batch sorption phenomena, the adsorbate molecules diffuse into the interior of porous biosorbents.We studied the pseudo rst and second order models of kinetics.The pseudo rst order equation gives the adsorption in solid-liquid systems depending on the sorption capacity of solids. 42In this model, it is assumed that one arsenic species, i.e. the adsorbate, occupies one sorption site of the tested MCF: where A represents an unoccupied adsorption site on CF or the MCFs and k 1 is the rate constant in the pseudo rst order kinetic model whose linear form can be expressed as: where q e , q t (mg g À1 ) are the sorption capacities of the biosorbents at equilibrium and time t (h), respectively.Similarly, the pseudo second order rate expression which is applied to assess the chemisorption kinetics from liquid solution 43,44 and its mathematical linear form equation is as follows: where k 2 (g mg À1 h À1 ) is the rate constant for the pseudo second order adsorption and k 2 q e 2 or h (mg g À1 h À1 ) is the initial adsorption rate.In this model, it is assumed that each sorbate species covers two sorption sites at the surface of the biosorbent:

Adsorption isotherm models
An adsorption isotherm model describes the partitioning of sorbate molecules between the liquid and solid phases at equilibrium.The adsorption of an arsenic metalloid by CF and the MCFs was justied according to the Freundlich and Langmuir models in this study.The Freundlich model is applicable to both monolayer (chemisorption) and multilayer adsorption (physisorption), assuming that the surface of the biosorbents is heterogeneous. 45The linear form of this model is expressed as: where "K F " and "n" are Freundlich isotherm constants pertaining to the sorption capacity and intensity, respectively, whereas C e (mg L À1 ) is the equilibrium concentration of the sorbed As(III) species.
The Langmuir adsorption model, on the other hand, assumes monolayer adsorption on a uniform surface with a nite number of adsorption sites.According to this model, once the surface site is lled, no further sorption can take place at that specic site, leading to a surface saturation phenomenon.The linear form of this model is described as: where "K L " is the Langmuir constant related to the sorption energy and q m (mg g À1 ) is the maximum adsorption capacity. 46

Results and discussion
The chemical modication of chicken feathers was carried out with PNIPAM under different reaction conditions and with methanol as shown in Scheme 1.Reactions of maleimides with thiols 47,48 and amines 49,50 are well known.The double bonds of maleimide react specically with thiols in the pH range of 6.5-7.5, resulting in the formation of a stable thioether linkage that is not reversible.However, at more alkaline pH values (pH > 8.5), its reaction with amines becomes more evident and also increases the rate of hydrolysis of the maleimide group. 51,52ray diffraction analysis To investigate the crystallinity, XRD-patterns of untreated CF and modied chicken feathers were studied.As shown in Fig. 1, the CF have a typical pattern with a prominent 2q peak at 9.9 that corresponds to the a-helix conguration, and the more intense band at 19 is indexed to its strand secondary structure. 8he modications prepared with NaOH and Na 2 SO 3 and then jointly reacted with NIPAM give peaks with reduced intensity and a mild shi in values of 2q.The decrease in intensity at 19 is attributed to the decrease in the b sheet content as compared to neat feathers.In addition, the decrease of intensity in the peak at 9.9 and the appearance of a shoulder at 10.70 also suggests the fracture of the a-helix network.This strengthens the idea of the disruption of both the a-helix and the b-sheet content in the modied keratin material.However, there is a noticeable difference between the untreated chicken feathers and the MCFs, especially the appearance of new crystallinity peaks suggesting the formation of other crystalline patterns at a greater angle, e.g. at 2q ¼ 28.2 in the sample that was chemically treated with methyl alcohol.

Fourier transform infrared spectroscopy (FTIR)
In Fig. 2, the IR spectra of NIPAM, neat feathers and the modied biosorbents are presented.The neat feathers and the biosorbents exhibit typical amide vibrations including amide A (N-H stretching), amide I (C]O stretching, with a minor contribution from N-H bending and C-N stretching, 1600-1700 cm À1 ), amide II and amide III (N-H bending and C-N stretching, at around 1540 and 1240 cm À1 , respectively).Signicant changes can be seen in the amide A region of the modied feathers.A broad absorption band of neat chicken ber keratin (CF) appearing at 3416 cm À1 is mainly due to hydrogen bonded N-H stretching vibrations, 53 as the peptide N-H groups form hydrogen bonds with amide C]O groups in the native secondary structure.A shi in this band towards lower wavenumbers in the modied materials (MCF-I, MCF-II, and MCF-III) was observed.This shi to lower wavenumbers can be attributed to the disruption of the hydrogen bonds of the amide groups by the modications.In addition, the appearance of the peak at 936 cm À1 is assigned to the C-S bond stretching vibration in MCF-I.In MCF-II, the disappearance of the peak intensity at 1133 and 1367 cm À1 is generally assigned to the asymmetric and symmetric C-N-C stretching of maleimide, 54,55 which indicates the ring opening of the maleimide group.The modications were further conrmed by the changes in the amide I region of the FTIR spectrum (Fig. 3).Among all the amide bands of the backbone peptide groups in the proteins, the most intense and the most widely used one is the amide I band.This band arises mainly from the C]O stretching vibration of the amide carbonyl group, which is weakly coupled with the in-plane N-H bending and the C-N stretching vibration and appears in the region between z1700 and 1600 cm À1 .Signicant changes can be seen in the amide I region of the neat feathers and modied materials.A shi in the peak at 1642 cm À1 (assigned to the b-sheet structure) 56 towards higher wavenumbers was observed in all modications, suggesting the disruption of the b-sheet structure.Particularly, in the case of the modication with alcohol (MCF-III), the peak becomes sharp at 1653 cm À1 , suggesting the formation of random coils 57 at the expense of the a-helices and b-sheets present in the native feathers.In addition to signicant changes in the amide I region, the appearance of a distinct peak at 1738 cm À1 in the methanol modied (MCF-III) material is due to the C]O stretching absorption of the carbonyl group, in the characteristic absorption range (1750-1735 cm À1 ) for aliphatic esters.
The amide I region is most oen used for secondary structure characterizations.However, due to overlapping peaks and interference of water vibrational bands, the use of the amide III vibration is suggested for more accurate analysis of the protein secondary structure. 58Therefore, to further elucidate the secondary structure of modied keratin, the amide III region (1350-1200 cm À1 ) was employed.For the enhancement of the resolution, techniques such as the second derivative can be used to locate the positions of individual amide III bands.This technique can be used as a sensitive diagnostic tool in locating the positions of bands in the secondary structure.We used the second derivative technique to locate the exact position of different peaks in the amide III region.The amide III peaks of neat feather keratin and modied keratin identied by the 2 nd derivative were further resolved by tting of Gaussian bands (Fig. 4A-D).The secondary structures were assigned according to the literature as a-helix (1330-1295 cm À1 ), b-turn (1295-1270 cm À1 ), random coil (1270-1250 cm À1 ) and b-sheet   (1250-1220 cm À1 ). 8,58Clear differences in the resolved amide III bands of the modied keratins MCF-I, MCF-II and MCF-III (Fig. 4B-D, respectively) and native feather keratin CF (Fig. 4A) can be seen.Both b-sheet and a-helix contents decreased, and the unordered structures (random coils) increased aer modi-cation, particularly in the MCF-III modication (Fig. 4D), in agreement with the X-ray diffraction observations.

Differential scanning calorimetry (DSC)
The phase behavior of the prepared material was studied by DSC.The DSC trace for raw CF has higher heat ow values as compared to MCF.Typical heat ow curves are shown in distinctively different behavior with broader denaturation, suggesting a wide range of structural changes due to modication.The DSC of untreated CF shows an exothermic peak at <230 C, which is usually assigned to a-helix disordering and decomposition.However, in the modied chicken feather materials, particularly MCF-I and MCF-III, this peak was broadened and to comparatively lower temperatures. 59hese observations suggest the loss of a-helix structures and gain of amorphous behavior in all modied material, especially MCF-III, marked with a broadened melting curve trend.

Thermogravimetric analysis (TGA)
The thermal properties of untreated CF and MCF were investigated by TGA as shown in Fig. 6.The TGA curves as a function of temperature of the CF and MCF show decomposition in the temperature range of 250-380 C. The curves for untreated CF and MCF-I,II show virtually identical behavior of decomposition in contrast to the methanolic modication MCF-III, indicating that the esterication of the protein has taken place, which brought thermal stability.At this juncture, a total weight loss of ca.90% was observed.The absorbed water decomposed below 130 C for CF and MCF-I,II with an observable difference in the case of MCF-III.Obviously, it indicates that the alkaline and sodium sulte treated modications did not improve the thermal stability of the composed material and followed the TGA trend of untreated CF.Nevertheless, distinctively different TGA behavior of MCF-III in terms of decomposition and weight loss is useful for the sorption phenomenon.The chemical changes on the surface of the developed material MCF-III is consistent with our high arsenic uptake as discussed in this sorption study.

Adsorption of arsenic As(III) by MCFs
The pH effect on the arsenic uptake.We modied three types of biosorbents such as MCF-I, MCF-II and MCF-III and conducted their kinetic study at variable pH i.e. 4, 7 and 12.As shown in Fig. 7, the change in pH during the course of modi-cation has a signicant effect on the arsenic uptake.The maximum uptake was obtained at an acidic pH of 4 for all types of modications, MCF-I, MCF-II, MCF-III, and untreated CF.The keratin used in this study consists of amines and carboxyl groups which are protonated or deprotonated depending on the pH of the solution.The surface complexation theory suggests that a low pH is responsible for increasing the positive charge of the surface whose impact can be seen in terms of high sorption of negatively charged arsenic species such as H 2 AsO 4À , HAsO 4 2À , H 2 AsO 3 À , HAsO 3 2À and AsO 3 3À in our case.Thus, it is estimated that the biosorption of arsenic involves an ionexchange process in which arsenic oxyanions tend to approach positively charged active sites of the biosorbent.With greater pH, the number of positive sites decreases with an increase in the number of negatively charged arsenic species, and thus, a low uptake is observed as shown in Fig. 7.Moreover, the increase in sorption capacity for arsenic at low pH can also be explained by changes in the nature of arsenic complexation, charge density, solubility and degree of hydrolysis. 60,61The order of higher arsenic uptake regarding the pH for different modi-cations is: acidic > neutral > basic.
Effect of modication on the arsenic uptake.We chemically modied different functional groups of keratin protein extracted from CF and examined the arsenic uptake.It was determined experimentally that MCF-III, as a result of the esterication of -COOH functional groups, had a maximum arsenic uptake of 85-90% at an acidic pH 4 compared to MCF-I, MCF-II and CF.The high arsenic uptake of MCF-III suggests  that ionized carboxyl groups in keratin protein play an inhibitory role in As(III) and As(V) adsorption. 62Another biosorbent, MCF-I treated with NaOH, showed lower levels of arsenic uptake of 54% at pH 4. This indicates that the disulde in cysteine protein was not fully reduced and thus remained unreactive with dopant PNIPAM which was added to perform a reaction between the double bond of the PNIPAM ring and the reduced thiol (-SH) of the disulde (-S-S-) peptide linkage in the cysteine molecule.The observed uptake i.e. 54% is due to the surface as well as the chemical modication of the protein.
Similarly, the uptake of the sodium sulte treated modication (MCF-II) is not signicantly high on account of poor reduction, and thereby the low availability of reactive sites for added PNIPAM to make a larger molecule for efficient sorption.Thus, MCF-I performed an arsenic uptake of merely 24% under acidic conditions of the sorption process.The arsenic uptake prole of untreated CF and MCF is shown in Fig. 8.
Kinetics of arsenic adsorption.The adsorption kinetics of arsenic were studied for both the modied chicken feathers (MCF) and untreated CF at neutral pH 7 with initial concentrations of 200 mg L À1 As(III) metalloid.However, kinetic data modelling was performed using our most effective modied material, MCF-III, at three variable pH values of 4, 7 and 12.As illustrated in Fig. 8, within the rst 30-40 minutes, MCF-III followed by MCF-I, MCF-II and untreated CF attained equilibrium quickly.Thereaer, in all cases, the uptake decreases gradually, and nally, a steady trend was observed aer 8-10 hours at the adjusted pH of 4. The rapid attainment of the equilibrium and its smooth fall suggest that the exposed functional groups, particularly the carboxylic sites of MCF, are esteried, causing the depression of negative charges on the surface that favors quick uptake.Due to low pH, more hydrogen ions in solution, nonetheless, promote the protonation of negatively charged arsenic species, resulting in low uptake i.e. (H 2 AsO 3 À + H + / H 3 AsO 3 ; H 2 AsO 4 + H + / H 3 AsO 4 ). 62To evaluate the kinetics of the adsorption process, the pseudo rst order and pseudo second order models were tested for interpreting the experimental data.
Pseudo rst order kinetics.The rate constant k 1 and q e were calculated using the slope and intercept of plots of log(q e À q t ) versus the time t (Table 1).The best t lines at each pH value yielded relatively low correlation coefficients R 2 .A close inspection of the experimental observation and the model t in Fig. 9(A) suggests that eqn (3) is not appropriate because of a non-linear trend of the experimental data.In addition, a poor agreement between the experimentally observed equilibrium and that derived using eqn ( 4) is another reason that the biosorption of arsenic does not follow pseudo rst order kinetics.It further implies that, in the sorption phenomenon, one arsenic species covers more than one sorbent site of the modied material.
Pseudo second order kinetics.The pseudo second order parameters q e and k 2 were determined from the slope and intercept of the plot of t/q t versus time (t) in eqn (4).The results of the kinetic data modelling reveal that the pseudo second order model explains the rate of biosorption better than the pseudo rst order model (see Fig. 9(B)).The correlation coefficients (R 2 ) at pH 4, 7, and 12 for the pseudo second order kinetic model ts are 1.0, 0.99 and 0.98 respectively.These R 2 values are higher than for the pseudo rst order model ts.Given the good agreement between the experimentally observed biosorption capacity and the model t besides high values of R 2 , this suggests that the arsenic biosorption follows pseudo second order kinetics.This type of kinetics, as expected, favors chemisorption 46 involving the interaction between trapped arsenic species and exposed functional groups of MCF, especially -COOH, which is consistent with our other characterization results.
Adsorption isotherms.Two famous Freundlich and Langmuir models were selected to t the experimental data.The equilibrium isotherms for the sorption of As(III) species by MCF-III at pH 4, 7 and 12 were studied at an initial arsenic concentration range of 200-800 mg L À1 .The Freundlich isotherm constants K F and n are determined from the intercept and slope of a plot of log q e versus log C e (Fig. 10A).In this study, the n values are greater than unity giving idea of chemisorption. 63For the tested pH values of 4, 7 and 12, the linear values of R 2 $ 97 attained by the Freundlich model equation make it the best t for our experimental data as shown in Table 2.The slope and intercept of the plot of C e /q e versus C e at three different pH values were used to calculate q m and K L (Fig. 10B).The Langmuir isotherm parameter ts (Table 2) for the adsorption of arsenic on MCF-III yielded isotherms are in poor agreement with the observed trend (R 2 # 97) for pH 4, 7 and 12.The inclination of our isotherm data towards the Freundlich model is indicative of a heterogeneous surface of MCF-III with

SEM analysis
Both shape and of the adsorbent leave an impact on its adsorption capacity. 46The SEM images of the CF and the most effective chemically modied chicken feathers (MCF-III) were studied and are represented in Fig. 11A and C, respectively.As illustrated in Fig. 11A, the untreated feathers have long shas and barbs and a smooth surface, which is evident in the magnied image of the CF (Fig. 11B).Interestingly, the surface of MCF-III (Fig. 11C) shows shiny patches with higher levels of modication, leading to a completely amorphous structure as shown in the magnied image of MCF-III (Fig. 11E), where the shas and barbs of the feathers have disappeared and the original structure of the feather could not be identied.In addition, there are some regions where the surface of the feather remains intact with some levels of modication only on the surface.There are some regions of intermediate modication where the feathers have not completely lost their structural integrity but the surface of the feathers have become damaged and rough (Fig. 11D).These observations are in agreement with the changes as evidenced by FTIR and XRD.The structural   changes aer the modications may have contributed to the higher sorption properties of the MCF-III in this study.These observations further strengthen and support the theory that the increased arsenic sorption capacity of the biosorbents in contrast to untreated CF may be attributed to heterogeneous microstructures developed aer modication.Upon modication, the surface becomes brighter and causes roughness that is the characteristic of increased surface activity compared to CF.These structural changes aer modication may have contributed to the higher sorption properties of MCF-III in this study.

Conclusion
Modied chicken feathers can effectively be employed for removing arsenite As(III) species from contaminated water sources.The modication involving esterication showed the highest As(III) uptake on account of the overall anionic charge depression on the surface of the modied feathers.Arsenic oxyanion-biopolymer interaction leads to a rapid equilibrium within the rst hour of the biosorption reaction.At the acidic pH 4, the arsenic uptake is increased as a result of protonation making the surface positive overall which facilitates arsenic sorption.The pseudo second order kinetics model accurately described the adsorption and justied our characterization data supporting the fact that chemisorption and physisorption mechanisms are involved.The Freundlich isotherm showed a better t than the Langmuir isotherm, thus indicating the applicability of multilayer coverage of arsenic species.Results from this study recommend that MCF-III is a very suitable and cost-effective biosorbent for arsenic removal, as anticipated.

Scheme 1
Scheme 1 General routes of possible reactions for different modifications (MCF-I, MCF-II, and MCF-III) of feather keratin and the adsorption of As(III).

Fig. 2
Fig. 2 FTIR spectra of CF, MCF-I, MCF-II and MCF-III.For CF, MCF-I, MCF-II and MCF-III, the intensities are normalized at 3200 cm À1 .The spectra are offset, and curves are shifted vertically for clarity.

Fig. 4
Fig. 3 FTIR spectra of the amide I region for CF, MCF-I, MCF-II and MCF-III.For easier comparison, intensities are normalized in all spectra at 1640 cm À1 .The spectra are offset, and curves are shifted vertically for clarity.

Fig. 7
Fig. 7 As(III) uptake of CF, MCF-I, MCF-II, and MCF-III at different pH values and at 20 C temperature.

Fig. 8
Fig. 8 The As(III) sorption kinetics of 0.1 g L À1 of CF, MCF-I, MCF-II, MCF-III for concentrations of 200 mg L À1 of sodium arsenite solution at 20 C.

Fig. 10
Fig. 10 Linearized (A) Freundlich and (B) Langmuir isotherms for the arsenic adsorption on MCF-III at different pH at 20 C.

Fig. 11 SEM
Fig. 11 SEM Images of pure feather keratin CF (A) with a magnified region (B) and the modified keratin MCF-III (C) with magnifications (D) and (E).

Table 1
Adsorption kinetic model rate constants for As(III) adsorption on MCF-III at 20 C 64 Pseudo rst order Pseudo second orderk 1 (h À1 ) q e.cal (mg g À1 ) R 2 k 2 (g m À1 h À1 ) q e.cal (mg g À1 ) h (mg g À1 h À1 ) Ractivated functional groups responsible for both the monolayer (chemisorption) and multilayer (physisorption) type of adsorption in our study.Apart from our study, the Freundlich model also produced a favourable t to the data compared to the Langmuir model in the adsorption study of N 2 gas, nanosilica, nanodiamonds and protein.64

Table 2
Freundlich and Langmuir isotherm model parameters and correlation coefficients for the adsorption of As(III)