Corncob as an effective, eco-friendly, and economic biosorbent for removing the azo dye Direct Yellow 27 from aqueous solutions

The corncob is an agricultural waste generated in huge quantities during corn processing. In this paper, we tested the capacity of corncob particles for water purification by removing the azo dye Direct Yellow 27 (DY27) via biosorption. The biosorption process was investigated in terms of the kinetics, equilibria, and thermodynamics. Batch biosorption studies showed that the biosorption performance has strong inverse correlations to the solution pH and the corncob particle size, and it increases quickly with increasing contact time and initial dye concentration. The pseudo-second-order kinetic model provides the best fit to the experimental data, whereas the Redlich-Peterson isotherm model is most suitable for describing the observed equilibrium biosorption. The biosorption process is exothermic, spontaneous, and physisorption in character. Fourier transform infrared (FTIR) spectroscopy and confocal scanning laser microscopy (CSLM) studies suggest that lignocellulose and proteins play key roles in the biosorption of DY27 from aqueous solutions by corncob. Furthermore, after biosorption onto the corncob, the dye can be effectively desorbed using 0.1 M NaOH solution. Therefore, the corncob can be used as a promising biosorbent to remediate DY27-contaminated water and wastewater.


Introduction
Synthetic organic dyes are extensively used in many technologies, such as textiles, printing, paper making, leather tanning, rubber, food processing, plastics, cosmetics, hair coloring, pharmaceuticals, photography, agricultural research, light-harvesting arrays, and photoelectrochemical cells [1][2][3]. Currently, there are estimated to be more than 100,000 commercially available dyes with a total production rate of more than 0.7-1 × 10 6 tons per year [1].
Approximately 12% of the synthetic dyes produced annually in the world are believed to be lost during manufacturing and processing operations, and about 20% of the lost dyes enter a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Ethics statement
No specific permits were required for the corncob sample collection for this study. We used corncob material that came from the private lands from one of the authors (NKBV). No material was taken from protected land or National Parks. The studies conducted in this work did not involve any endangered or protected species.

Chemicals
DY27 dye (molecular formula: C 25 H 20 N 4 Na 2 O 9 S 3 , molecular weight: 662.62 g mol -1 , Fig 1) [27] was purchased from Sigma-Aldrich Chemicals. All other reagents were of analytical grade. A DY27 stock solution of 2 g L -1 was prepared by dissolving an accurately weighed amount of DY27 in distilled deionized water. Test solutions of DY27 were prepared by diluting the stock solution with distilled deionized water. In this work, the initial DY27 concentration (C o ) varied from 7 to 293 mg L -1 , and the pH of each solution was adjusted to the desired value using 0.1 M HCl or NaOH solutions.

Biosorbent preparation
Corncobs (Zea mays L.) were collected in the municipality of Villa del Carbón (Latitude: 194 3' 41"N; Longitude: 99˚28' 28"W), State of Mexico, Mexico. The collected corncobs were first washed thoroughly with distilled deionized water, and then oven-dried at 60˚C until the weight was constant. The dried samples were subsequently ground in a hammer mill (Glen Creston Ltd.), and the resulting particles were screened using ASTM standard sieves to obtain fractions with different particle sizes. The sieved fractions were stored in airtight plastic containers until use.

Proximate chemical composition of corncob
The biosorbent's proximate chemical composition was analyzed in triplicate, following methods described by the Association of Official Analytical Chemists [28]. The total protein (TP) was analyzed by the Kjeldahl method, using 6.25 as the conversion factor from total nitrogen to total protein. Ash (A) was determined by incineration of samples in a muffle furnace at 600˚C until the weight was constant. Ether extract (EE) was determined by the Soxhlet method. Crude fiber (CF) was measured as the loss on ignition of dried lipid-free residues after digestion with H 2 SO 4 and NaOH standard solutions. Nitrogen-free extract (NFE) was calculated as the difference between dry matter and the sum of total protein, ether extract, ash, and crude fiber (Eq 1):

Determination of point of zero charge
The biosorbent's point of zero charge (pH pzc ) is defined as the pH at which the amounts of positive and negative charges on its biosorbent surface are equal, leading to a neutral net charge [29]. In the present work, pH pzc was determined using a batch equilibrium method, following the procedures described by Lim et al. [30]. Briefly, corncob samples (1 g L -1 ) were mixed with 0.1 M KNO 3 solutions which were previously adjusted to initial pH (pH i ) values of 2, 4, 6, 8, 10, and 12 by the addition of 0.1 M HNO 3 and/or 0.1 M NaOH. The suspensions were agitated continuously at 150 rpm at 18 ± 1˚C for 48 h to ensure that an equilibrium pH was reached in each case. Subsequently, the suspensions were filtered through filter paper (Whatman, grade 42), and the final pH (pH f ) of each filtrate was recorded. A graph of the pH change in the KNO 3 solution (ΔpH = pH i −pH f ) against pH i was then constructed to determine the pH pzc of corncob.

Batch kinetic experiments for DY27 biosorption
Batch kinetic biosorption studies were conducted to evaluate the effect of relevant operational parameters, such as solution pH, corncob particle size, shaking contact time, and C o on the biosorption of DY27 by corncob. All experiments were performed in 500-mL Erlenmeyer flasks, using 100 mL of DY27 solution of known pH and dye concentration and 1 g (dry weight) L -1 of biosorbent. During each experiment, the pH of the solution was maintained at a constant value (± 0.1 pH unit) by periodic checks and adjustments with 0.1 M HCl or NaOH solution. The flasks were placed in a shaker bath (Cole Parmer1) at a constant shaking speed of 150 rpm for 72 h. The influence of solution pH was examined at C o = 50 mg L -1 at different pH values (1.5, 2.0, 2.5, 3.0, 4.0, 5.0, and 6.0) at 18 ± 1˚C. The effect of particle size was explored in DY27 solution at C o = 50 mg L -1 , pH 1.5, and with different corncob particle sizes (0.297-0.50, 0.50-0.80, 0.8-1.0, 1.0-1.18, 1.18-1.41, 1.41-1.70, and 1.70-2.0 mm) at 18 ± 1˚C. The influence of shaking contact time was investigated for 0, 0.1, 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 4.0, 5.0, 6.0, 7.0, 24, 48, and 72 h in solutions with different initial dye concentrations (7,13,28,50,74,85,100,145, and 293 mg L -1 ) at pH 1.5 and 18 ± 1˚C. These C o values were also used to examine the kinetic biosorption profiles at pH 1.5 and 18 ± 1˚C.
Corncob-free controls were used to evaluate the loss of DY27 dye arising from precipitation and/or adsorption onto glass wall.
For the kinetic studies, after certain contact times, samples were taken from the mixture and centrifuged at 5000 rpm for 10 min. The DY27 concentration in the supernatants was analyzed by spectrophotometry (Thermo Scientific™ Evolution 201) at 398 nm. The time-dependent DY27 biosorption capacity (q t , mg g -1 ) was estimated using the following mass balance relationship: Here, C o and C t are the initial and residual DY27 concentrations (mg L -1 ) at time t = 0 and t (h), respectively, and M is the corncob concentration (g L -1 ).

Isotherm studies of DY27 biosorption
For DY27 equilibrium biosorption, corncob biomass was added at 1 g (dry weight) L -1 to DY27 solutions at pH = 1.5 at C o = 7-293 mg L -1 and different temperatures (18, 35, 50, and 60˚C), with a constant agitation of 150 rpm for 72 h to ensure that biosorption equilibrium was reached. Afterwards, samples were taken, centrifuged at 5000 rpm for 10 min, and then the dye concentrations in the supernatants were measured. The DY27 equilibrium biosorption capacity (q e exp ) was calculated using Eq (2), by replacing C t with C e (the equilibrium concentration of DY27 in the aqueous solution in mg L -1 ).

Kinetic modeling of DY27 biosorption
Modeling the kinetics of target pollutant removal is crucial for understanding relevant aspects of the biosorption process, such as the biosorption rate and its influencing factors, designing effective and efficient biosorption systems, selecting operation conditions for full-scale batch process, and establishing the time dependence of biosorption systems under different process conditions [31,32]. In the present work, the dynamics of DY27 biosorption onto corncob biomass was examined at different solution pH levels, corncob particle sizes, and C o . The data were analyzed using five different kinetic models: pseudo-first-order, pseudo-second-order, Elovich, intraparticle diffusion, and fractional power ( Table 1).

Modeling of DY27 biosorption equilibrium isotherms
Equilibrium biosorption isotherm models are useful for describing the interactions between the adsorbate and biosorbent, and therefore they are essential for the design of biosorption facilities on a large scale [32]. Various two-parameter (Langmuir, Freundlich, Temkin, Halsey, and Dubinin-Radushkevich) and three-parameter (Sips, Toth, Radke-Prausnitz, and Redlich-Peterson) isotherm models (Table 1) were applied in the present work to fit the experimental equilibrium data of DY27 biosorption.

Desorption studies
Corncob samples were mixed with a 293 mg L -1 DY27 solution (100 mL) at a ratio of 1 g L -1 , pH 1.5, and 18 ± 1˚C with constant agitation at 150 rpm for 72 h to ensure biosorption equilibrium. Afterwards, the DY27-loaded sample was separated from the dye solution by centrifugation at 5000 rpm for 10 min, and the concentration of residual DY27 in the supernatant was measured to calculate the amount of dye biosorption. The DY27-loaded precipitate was gently washed with distilled deionized water to remove any unbiosorbed dye. After another centrifugal separation, the dye-loaded corncob was oven-dried at 60˚C until a constant dry weight was obtained, and stored in an airtight plastic container until use.
All batch desorption experiments were conducted in 500-mL Erlenmeyer flasks containing 100 mL of given desorption solution and 1 g (dry weight) L -1 of DY27-loaded corncob. The flasks were agitated in a shaker bath (Cole-Parmer1) at 150 rpm and 18 ± 1˚C.
To select the best desorption solutions, different acids (0.1 M HCl, H 2 SO 4 , HNO 3 , NH 4 Cl, and (NH 4 ) 2 SO 4 ), alkali (0.1 M NaOH, KOH, and Na 2 CO 3 ), neutral salts (0.1 M NaCl and NaNO 3 ), and organic compounds (0.1 M acetone and chloroform) were mixed with DY27-loaded corncob over a 48 h period. The best agent for DY27 desorption was used for batch desorption kinetic studies. The amount of desorbed DY27 was determined spectrophotometrically at a wavelength of 398 nm. The desorption efficiency (E Dt , %) was estimated according to the following equation: Here, C Dt is the residual DY27 concentration (mg L -1 ) in the desorption solution after time t D , and q e is the initial (t D = 0 h) dye concentration in the corncob (mg g -1 ), i.e., the equilibrium biosorption capacity.

Statistical data analysis
All biosorption and desorption experiments were repeated three times, and the values reported herein represent the average values. The biosorption and desorption data were statistically analyzed by analysis of variance (ANOVA) with Tukey's test (overall confidence level = 0.05) using GraphPad Prism software (version 7.0; GraphPad Software, Inc.). The same software was also used to estimate all the biosorption model parameters by nonlinear regression analysis. The different biosorption models used for fitting were assessed using the coefficient of determination (r 2 ), standard deviation of residuals (Sy.x), absolute sum of squares (ASS), and Aikake information criterion (AIC). A high value of r 2 (close to 1.0) and small Sy.x, ASS, and AIC values correspond to better representation of experimental data by a particular model.

SEM analysis
To analyze the morphological features and surface characteristics of native (i.e. not loaded with DY27) and DY27-loaded corncob samples, micrographs were obtained using a high-resolution scanning electron microscope (JSM-7800F, JEOL) at an accelerating voltage of 5 kV after gold coating.

CSLM analysis
DY27-loaded corncob samples were stained with 1% fluorescein or calcofluor white M2R (Sigma-Aldrich) in order to detect proteins and cellulose, respectively. Excess dye was removed with deionized water, and the samples were dried at 60˚C for 24 h. Micrographs were obtained using a multiphoton laser scanning confocal microscope (LSM710 NLO, Carl Zeiss) equipped with an argon/krypton laser with four excitation lines (405, 488, 561, and 633 nm) and powers of 100, 100, 2, and 2%. All images were acquired at 10× magnification without zooming, and saved in TIFF format at 512 × 512 pixels.

FTIR analysis
FTIR spectroscopic analysis of native and DY27-loaded corncob samples was used to identify the main functional groups on the corncob surface that may be involved in the biosorption of DY27 from aqueous solutions. For this purpose, finely ground native or DY27-loaded corncob were thoroughly mixed with dried spectroscopic KBr in a 1:3 ratio, and immediately analyzed using a Thermo Scientific Nicolet iS10 FTIR spectrometer equipped with an EasiDiff diffuse reflectance hemisphere accessory (Pike Technologies). FTIR spectra were obtained using 64 scans over the range of 4000-400 cm -1 with a resolution of 1 cm -1 . The Kubelka-Munk function was calculated from the diffuse reflectance data using the OMNIC software (Thermo Fischer Scientific). The FTIR spectra in the present work are presented in Kubelka-Munk function vs. wavenumber.

Chemical composition of corncobs
The proximate chemical composition of corncobs (dry matter basis) was found to be: total protein, 4.28 ± 0.63%; total ash, 3.44 ± 0.08%; ether extract, 2.03 ± 0.22%; crude fiber, 26.29 ± 0.52%; and nitrogen-free extract, 63.96 ± 1.5%. Therefore, the corncob has low fat, protein, and mineral contents, and rich in carbohydrate and lignin contents. This is consistent with the knowledge that the main components of corncobs are cellulose, hemicellulose, and lignin, which together comprise approximately 90% of the dry matter [20,21]. The polymeric components in corncob possess many functional groups that can be active sites for the biosorption of dyes [20].

pH pzc of corncob
The solution pH strongly affects the surface charge of a biosorbent and its biosorption capacity. The surface charge can be qualitatively assessed by the point of zero charge (pH pzc ) or isoelectric point [2,23]. The pH pzc of corncob was determined to be 6.83 (Fig 2). Therefore, for solution pH below this value, the corncob particles carry positive charges, and vice versa. The value of 6.83 is close to that reported by Leyva-Ramos et al. [38] (pH pzc = 6.2), who attributed the weak acidity of natural corncob to the slightly higher concentration of acid sites than that of the basic ones.

SEM analysis
SEM has been widely used to directly observe the surface structure and morphology of different biosorbents [18]. Fig 3 shows the SEM micrographs of DY27-unloaded (native) and DY27-loaded corncob. SEM micrographs revealed that native corncob has a rough and porous surface ( Fig 3A), with pores of different shapes and with sizes exceeding 50 nm, thus indicating that corncob has a macroporous structure. Likewise, corncob has an irregular structure, which makes biosorption of DY27 dye on different parts of the biosorbent possible. From the SEM images of the native ( Fig 3A) and DY27-loaded corncob (Fig 3B), there is no evidence of changes in the surface morphology of the biosorbents. Contrastingly, some studies have mentioned that the surface morphology of biosorbents may change once dyes have been loaded [18,39].

Effect of solution pH on DY27 biosorption kinetics
No measurable changes in the DY27 dye concentration were detected in any of the assayed corncob-free controls, thus indicating that any evident DY27 removal was solely due to the biosorbent. As mentioned earlier, the pH can affect the biosorption of dyes from aqueous solutions via changing the surface charge of biosorbents. The solution pH may also affect the kinetics of biosorption through the ionization state and solubility of dyes [2]. Fig 4 depicts the kinetic profiles of DY27 biosorption by corncob at different solution pH (1.5-6.0). Solutions with pH < 1.5 could not be assayed because of dye precipitation. Corncob was found to take up DY27 from aqueous solution at all other pH levels, although the rate and extent of biosorption varied greatly. The capacity of DY27 biosorption increased as the solution pH decreased, reaching its highest level at pH 1.5. This trend can be explained with the aid of pH pzc . All the DY27 solutions assayed had pH < pH pzc . Under these conditions, the surface charge of corncob is positive because the active biosorption binding sites are protonated. Such protonation favors the biosorption of anionic DY27 molecules as a result of electrostatic attraction. The lower the solution pH, the stronger the corncob surface is protonated and the greater the biosorption of DY27.
The present results clearly show that the optimal solution pH for the biosorption of DY27 onto corncob is 1.5, and this pH was therefore used in all further studies. To the best of our knowledge, no previous works have addressed the biosorption of DY27 from aqueous solution. Only a single study reported the removal of DY27 from aqueous solution using an inorganic adsorbent (crystalline hydroxyapatite), for which the optimal pH for adsorption was found to be 3.5 (the lowest pH value assayed by those authors) [26].

Influence of corncob particle size on DY27 biosorption
Fig 5 displays the effects of corncob particle size on DY27 biosorption at pH 1.5. It is apparent that the rate and capacity of biosorption increased for finer corncob particles. This trend is attributed to the enlarged specific surface area, as well as a lower intraparticle diffusion resistance that improves accessibility of internal sites on the corncob [1,40,41]. Based on this result, the smallest particle size (0.297-0.50 mm) was used in further studies.

Effect of contact time and initial DY27 concentration on biosorption
From the results shown in Fig 6, it is evident that the contact time between biosorbent and DY27 solution has a significant effect on the biosorption. At all assayed C o values, the biosorption capacity gradually increased with the experiment time, eventually reaching a maximum constant value that corresponded to the equilibrium biosorption capacity (q e exp ). After this point, the DY27 removal ceased. Furthermore, irrespective of C o , the initial biosorption rate was quite fast because (1) there were a large number of positively charged binding sites available for biosorption and (2) the driving force to transfer DY27 molecules from the aqueous solution to the surface of corncob was high [2,30]. Afterwards, both the number of vacant binding sites and the mass transfer driving force decreased gradually, therefore the biosorption rate decreased and eventually reached dynamic equilibrium [1,42]. Fig 6 also shows that the DY27 equilibrium biosorption capacity increased from 6.5 to 73.00 mg g -1 as C o was increased from 7 to 145 mg L -1 . The higher availability of DY27 molecules in the solution increased their interactions with the corncob and the equilibrium biosorption capacity. A higher C o also increases the DY27 concentration gradient near the particle surface, which is the thermodynamic driving force for the mass transfer of DY27 from the solution to the corncob surface. Consequently, the probability of interaction between the dye molecules and the active biosorption sites increases, and so does the DY27 biosorption capacity [16]. However, the equilibrium biosorption capacities were very similar at C o = 145 and 293 mg L -1 , probably due to saturation of the biosorption sites.
The above results clearly indicate that there is a direct relation between the dye concentration and the available biosorption sites on the corncob surface, the latter is the limiting factor for DY27 biosorption, and that the saturation of biosorbent depends strongly on C o [1,16,18].

Modeling of biosorption kinetics
The kinetics of DY27 biosorption onto corncob was analyzed using five different models at pH = 1.5-6.0, corncob particle size from 0.297-0.5 to 1.7-2.0 mm, and C o = 7-293 mg L -1 . Tables 2-4 present the experimental equilibrium biosorption capacity (q e exp ), the kinetic parameter values (pseudo-first-order model: q e1 and k 1 , pseudo-second-order model: q e2 and k 2 , Elovich model: A E and B E , intraparticle diffusion model: k id and c, and fractional power model: k fp and v) under the different solution pH levels, corncob particle sizes, and C o values assayed, together with the corresponding r 2 , ASS, Sy.x, and AIC values.
The r 2 values were higher with the pseudo-second-order kinetic model; and the corresponding ASS, Sy.x, and AIC values were lower than those obtained with the other four kinetic models. Therefore, the pseudo-second-order model is the most suitable for describing the experimental data obtained in this study, as shown by the solid lines in Figs 4-6. In addition, the equilibrium biosorption capacity calculated from this model (q e2 ) matched the experimental value (q e exp ) closely.
The pseudo-second-order model has also been successfully used to describe the biosorption kinetics of other dye-biosorbent systems. Examples include red 27 onto the leaves of Eichhornia crassipes [16]; direct red-31 and direct orange-26 onto rice husks [18]; methylene blue onto tamarind fruit shell [39]; crystal violet onto Artocarpus altilis (breadfruit) skin [30]; acid orange 7, basic red 46, and basic blue 3 onto Spirogyra sp. [43]; and reactive black 5 onto Laminaria sp. [44]. Tables 2-4 show the variation in the pseudo-second-order model rate constant (k 2 ) according to the solution pH, corncob particle size, and C o . A lower k 2 value means a longer time needed to reach the equilibrium biosorption capacity under the given experimental condition, because k 2 plays the key role of a time-scaling factor in the corresponding kinetic equation [45]. The same trend was observed in other dye biosorption systems [16,18,41,44].  Biosorption of Direct Yellow 27 dye from aqueous solutions by corncob the x-axis (concentration axis). The concave shape and initial slope indicate that as more biosorption binding sites are occupied, it becomes more difficult for the remaining dye molecules to find available active sites on the corncob surface, which suggests a progressive saturation of corncob with DY27 [16,46].

DY27 biosorption isotherms and their mathematical modeling
The DY27 biosorption performance decreased with increasing temperature from 18 to 60˚C, revealing the exothermic nature of this process. This lower biosorption at higher temperature could be attributed to many possible factors: increased solubility of DY27 in water, stronger dye-solvent interaction than that between DY27 molecules and corncob, increased Brownian movement of DY27 molecules in solution, the breaking of existing intermolecular hydrogen bonding between DY27 and corncob, and/or damaged active biosorption sites that decrease the surface activity [1,47]. A decrease in the dye biosorption performance at higher temperatures has also been reported for the biosorption of FD&C red No. 40 and acid blue 9 by Spirulina platensis [48]; acid blue 9, food yellow 3, and FD&C yellow No. 5 by chitosan [49]; methylene blue, eriochrome black T, and alizarin S by prickly pear cactus (Opuntia ficus indica) cladodes [50]; and astrazon red by the leaves of Posidonia oceanica (Linnaeus) Delile [51], among others. The DY27 biosorption isotherms were modeled with several two-parameter (Langmuir, Freundlich, Temkin, Halsey, and Dubinin-Radushkevich) and three-parameter (Sips, Radke-Prausnitz, Redlich-Peterson, and Toth) models. The parameters obtained from nonlinear regression analysis for different temperatures, along with their corresponding r 2 , ASS, Sy.x, and AIC values are shown in Tables 5-8. The equilibrium biosorption of DY27 onto corncob is best described by the Redlich-Peterson model, as it yielded a higher r 2 value and lower ASS, Sy.x and AIC values at each assayed temperature compared to the other isotherm models. The good agreement between this model and experimental results are shown in Fig 7 (lines and  points, respectively). The Redlich-Peterson isotherm model is a hybrid model of Langmuir and Freundlich models, and it incorporates three parameters into a single empirical equation. This versatile model has been successfully used to represent biosorption equilibria over a wide concentration range, and can be applied to both homogeneous and heterogeneous systems [52]. Examples include  [54], reactive blue 19 onto cross-linked chitosan/oil palm ash composite beads [55], Gemazol turquoise blue-G reactive dye onto dried sugar beet pulp [56], and methylene blue onto activated carbons [57]. The maximum experimental biosorption capacity (q m exp ) was found to be 73.7 mg g -1 at 18˚C ( Table 5). As mentioned above, to the best of our knowledge, this is the first report of using a biosorbent to remove DY27 from aqueous solutions. The only previous work on DY27 removal used an inorganic adsorbent, namely poorly crystalline hydroxyapatite prepared via a precipitation method [26]. In that study, the maximum adsorption occurred at pH 3.5 to be 6.8 mg g -1 , while the corresponding value predicted by the Langmuir model was 89.26 mg g -1 [26]. The above results indicate that corncob may be used as a low-cost, abundant, and green biosorbent with high capacity to remove DY27 from contaminated water and wastewater.

Thermodynamics of DY27 biosorption
The DY27 biosorption isotherm data at different temperatures were used to calculate important thermodynamic parameters, namely changes in Gibbs free energy (ΔG˚), enthalpy (ΔH˚), and entropy (ΔS˚), in order to provide in-depth information about the inherent energetic changes during the biosorption process [37,58]. According to Table 9, the ΔG˚values were negative at all the temperatures (18-60˚C), indicating that the biosorption of DY27 onto corncob was spontaneous and thermodynamically favorable (i.e., there is an affinity of corncob towards the DY27 dye) [37,44]. These ΔG˚values are within the general range of -20 to 0 kJ mol -1 for physisorption processes [37,55].
The ΔG˚value was more negative at 18˚C, indicating that the biosorption was more energetically favorable at lower temperatures [1,37]. This is due to the positive ΔS˚value that indicates increased randomness at the solid-liquid interface after biosorption [37]. The positive ΔS˚value also suggests the possibility of some structural changes or readjustments in the DY27-corncob complex [1].
The negative ΔH˚values indicate that the biosorption is an exothermic process, in agreement with the observed decrease in biosorption as the temperature increases. The magnitude of ΔH˚value confirms that it is a physical sorption process (physisorption), for which the values of |ΔH˚| are usually lower than 84 kJ mol -1 [37]. It is well known that the physisorption is exothermic and enhanced at lower temperature, because the interaction between adsorbates and active sites is weakened at higher temperatures [59].

FTIR analysis
FTIR spectra were collected for native and DY27-loaded corncob samples, in order to examine the interactions between dye molecules with the functional groups on the corncob. Fig 8 displays the spectra for the corncob samples and that of DY27 alone, and Table 10 lists the peak positions of the major absorption bands. Cellulose, hemicellulose, and lignin are the main constituents of corncob. They contain functional groups that exhibit common absorption bands, including the OH stretching at 3428 cm -1 [60] and the symmetric stretching of aliphatic chains (-CH) at 2916 cm -1 [61]. Absorption bands related mainly to cellulose were observed at 1374, 1160, and 900 cm -1 , which are associated with the symmetric C-H deformations, C-O-C stretching of glycosidic bonds, and antisymmetric out-of-plane ring stretching, respectively [66,67,70]. Two characteristic absorption bands of hemicellulose were identified, the first one at 1732 cm -1 is associated with the stretching vibration of C = O group [62], and the second at 1250 cm -1 is attributed to the stretching of C-O bond [68,71]. Representative absorption bands of lignin were found at 1515, 1603, 1426, and 1070 cm -1 , which correspond to the aromatic skeletal vibration, the aromatic skeletal vibration breathing with C = O stretching [64,66,72], the C-H in-plane deformation with aromatic ring stretching, and the C-O deformations of secondary alcohols and aliphatic ethers [64], respectively. The characteristic absorption band of amide I C = O stretching at 1653 cm -1 [63] was also detected in the spectrum of native biosorbent, and this band most probably corresponds to the amide groups in the proteins.
After loading DY27 onto the native biosorbent, the FTIR spectrum showed a few changes. The relative intensity decreased for the bands at 1653, 1603, 1374, and 1070 cm -1 . Based on the assignments described above, these changes indicate the involvement of the proteins, cellulose, and lignin of corncob in the biosorption of DY27 dye molecules. Additionally, the absorption band at 1514 cm -1 became more intense, which can be explained by the combination between the band corresponding to the aromatic skeletal vibration of lignin at 1514 cm -1 and a characteristic absorption band of DY27 (the stretching of the azo functional group) at 1507 cm -1 [65]. A new absorption band appeared at 1458 cm -1 , which could be related again to the presence of the DY27 dye, because this wavenumber is close to the C = C stretching frequency of its aromatic ring at 1464 cm -1 [65]. The presence of these functional groups belonging to DY27 in the dye-loaded corncob may come from the dye molecules that either do not participate directly in the biosorption process or were biosorbed through weak forces. Dyes with linear molecular structures (including DY27) are used mainly for dyeing paper. The paper dyeing process basically involves the penetration of dye molecules into the capillary spaces of cellulose, followed by dye adsorption on the fiber surface. The adsorption is driven by the effects of charge, precipitation, and intermolecular forces. It is known that lignin has a considerable influence on the dyeability, but not the hemicellulose [73]. This finding in paper dyeing is consistent with the analysis of the FTIR spectra of DY27 biosorbed on corncob.

CSLM analysis
CSLM is a powerful optical technique for visualizing the structure of biopolymer mixtures. In CSLM, optical contrast is obtained by differences in fluorescence, either by auto-fluorescence of the material or by the addition of specific fluorescent dyes (fluorochromes) [74]. Each fluorochrome has its own particular emission wavelength [75]. In the present work, two fluorochromes were used: fluorescein and calcofluor white M2R. The auto-fluorescence of plant cells is mainly due to the presence of lignin and chlorophylls [76]. In the present work, the maximum emission wavelength (λ em max ) for lignin auto-fluorescence was observed at 501 nm (data not shown), which agrees with that reported by Rosas-Hernández (500 nm) [77]. Likewise, the peak of maximum emission for DY27 was found at 598 nm (data not shown).
The proteins were examined by staining DY27-loaded corncob samples with fluorescein, which binds covalently to the primary amine groups on the proteins. The λ em max for fluorescein was found at 519 nm, which is close to those reported by van [78], and the datasheet from the manufacturer (514 nm) [79].  Fig 9A), the fluorescence of DY27 dye on the corncob surface (red color, Fig 9B), and the fluorescein-induced fluorescence of corncob proteins (blue color, Fig 9C) can be observed. Furthermore, the interaction between DY27 with the corncob proteins was evidenced by a violet color in the overlay image (Fig 9D) produced by the combination of the red and blue colors from the DY27 dye and the fluorescein-induced fluorescence of proteins, respectively. These results clearly demonstrate that the DY27 dye binds to the proteins present in corncob.  The cellulose was examined by staining DY27-loaded corncob samples with calcofluor white M2R solution. This fluorochrome emits fluorescence when activated by ultraviolet radiation and has an affinity for the cellulose and chitin in cell walls [80,81]. The λ em max for calcofluor white was observed at 463 nm (data not shown), which is close to that reported in the datasheet from the manufacturer (433 nm) [82]. Fig 10 shows CSLM images of DY27-loaded corncob stained with calcofluor white. The auto-fluorescence of lignin is shown in green color (Fig 10A), the fluorescence of DY27 on the corncob surface in red (Fig 10B), and the calcofluor-induced fluorescence of corncob cellulose in blue (Fig 10C). The binding of DY27 onto the corncob cellulose was revealed by a violet color in the overlay image (Fig 10D), which is also due to the combination of the red and blue colors from DY27 and the calcofluor-induced fluorescence of cellulose, respectively. These results indicate that the DY27 dye also binds to the corncob cellulose.
The above findings clearly show that proteins and lignocellulose in the corncob play a crucial role in DY27 biosorption from aqueous solutions, which is consistent with the FTIR findings in this study.

DY27 desorption studies
Desorption of the dye from the DY27-loaded corncob was tested using several chemical agents, including acid, alkali, neutral salt, and organic solvent solutions. The performances of these eluent solutions are shown in Fig 11A. The highest DY27 desorption percentage was observed for strong alkali solutions, followed by weak alkaline, neutral salt, weak acid, organic solvent, and lastly strong acid solutions. Therefore, the DY27 desorption percentage depends strongly on the eluent pH: the higher the pH, the higher the desorption percentage. This trend may be explained on the basis of electrostatic repulsion between the negatively charged sites on the corncob biomass and the anionic DY27 molecules [16,83]. The strong acid solutions performed poorly in the desorption, eluting only negligible amounts (< 4%) of biosorbed DY27. Therefore, we selected the alkaline solutions (NaOH, KOH, and Na 2 CO 3 ) for further kinetic studies of the desorption process. Fig 11B displays the DY27 desorption percentage as a function of desorption time. It is evident that for all three alkaline solutions, the desorption was initially very fast and became slower towards equilibrium. The highest rate and extent of DY27 desorption were achieved with 0.1 M NaOH solution, with a desorption efficiency of approximately 98%. These results clearly indicate that 0.1 M NaOH is the best eluent for DY27 desorption from the corncob-based biosorbent.

Conclusions
The present work clearly demonstrates the feasibility of utilizing corncob as a novel, effective, efficient, eco-friendly, and inexpensive biosorbent for the removal of DY27 dye from aqueous solutions. The DY27 biosorption performance of corncob was affected significantly by solution pH, corncob particle size, shaking contact time, and initial dye concentration. The kinetic data of biosorption were accurately described by the pseudo-second-order model, and the biosorption isotherms were well predicted by the Redlich-Peterson model. Thermodynamic study indicated that the DY27 biosorption process was spontaneous and exothermic. FTIR and CSLM results confirmed that lignocellulose and proteins are involved in the biosorption process. Finally, a 0.1 M NaOH solution successfully eluted most of the DY27 dye from the DY27-loaded corncob biomass.