Performance of oxalic acid-chitosan/alumina ceramic biocomposite for the adsorption of a reactive anionic azo dye

A biocomposite system was developed and tested for the removal of the azo dye Reactive Red (RR195) from wastewater. The biocomposite was synthesized using ceramic particles containing 75% alumina which were coated using chitosan cross-linked with oxalic acid. The biocomposite showed high performance at low pH (maximum adsorption capacity = 345.3mg.g−1 at pH=2.0). The physicochemical and structure characteristics of the matrix were evaluated by Z-potential, FTIR-ATR, SEM-EDS, XRD, and porosity. Langmuir sorption isotherm and pseudosecond-order model gave the best fit. The electrostatic interaction between RR195 (due to the sulfonate groups) and the free amino groups of chitosan, enabled successive desorption/regeneration cycles. The maximum removal percentage (>80%) occurred at pH=2.0 due to the cross-linking effect. Experiments at different temperatures allowed the calculation of thermodynamic parameters (ΔG, ΔS, ΔH); adsorption was spontaneous, exothermic, and enthalpy controlled. The presence of inorganic ions (SO42−>NO3−>Cl−\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {\mathrm{SO}}_4^{2-}>\mathrm{N}{\mathrm{O}}_3^{-}>{\mathrm{Cl}}^{-} $$\end{document} ) was analyzed during the adsorption process. This novel biocomposite can be applied as a cost-effective and environmentally friendly adsorbent for anionic azo dye removal from wastewater. The application of chitosan cross-linked with oxalic acid as a coating of the ceramic support enhanced the adsorption capacity and enabled its use under acidic conditions without solubilization.


Introduction
Dyes are synthetic organic compounds used to color different materials (i.e., silk, wool, cellulose, cotton, and jute) (Srivastava et al. 2020). The global production of dyes reaches nearly 800,000 tons per year (Manzoor and Sharma 2020), and the market size is expected to reach USD 49.1 billion by 2027 (Grand View Research 2020). According to estimates, in the textile industry, around 100 tons of dyes are discharged due to the coloring of fibers, which generates a great environmental problem (Yagub et al. 2014). The wastewater containing these pollutants directly affects the photosynthetic capacity of aquatic plant species. Additionally, the chemical structure of the dyes contributes to a high level of organic matter content negatively affecting the water quality (Crini 2006).
Dyes can be classified according to the chemical structure (azo, anthraquinone, indigoids, etc.); the interaction with the substrate or fiber to generate the color (acid, basic, reactive, direct, and disperse); or the ion charge in solution being cationic (basics), anionic (direct, acid, and reactive), and nonionic (disperse) (Tan et al. 2015). More than 10,000 dyes are used in the textile industry; from these, 70% are azo dyes (Hassaan and Nemr 2017). The presence of azo-reactive dyes cause human health problems being responsible for mutagenesis and leading to several pathologies such as carcinogenesis, respiratory deficiency, and jaundice (Alver and Metin 2012;Vakili et al. 2014). Structurally, azo dyes are ionic molecules consisting of a chromophore (azo group, −N = N−) and sulfonate groups (−SO − 3 ) (Hunger 2003;Hassaan and Nemr 2017).
Reactive Red 195 (RR195) is an azo dye (molecular structure is shown Fig. A1 Supplementary Information) that is classified as reactive; these azo dyes dominated the market in 2019 with a market share of 55.7% (Grand View Research 2020). The azo dyes are composed of highly colored organic substances and are mainly used in tinting textiles; it is estimated that by 2025, the market of reactive dyes will generate $438 million (Wise Guy 2019). The high use of reactive dyes is due to the presence of covalent bonds between the dye molecules and the functional groups of the fiber (wool, cotton, and nylon) (Hunger 2003). About 40% of the reactive dyes are persistent in the wastewater because they are recalcitrant compounds in the water bodies, exhibiting serious difficulties for their removal and producing diseases such as cancer as well as respiratory and skin problems (Alver and Metin 2012;Zaharia and Suteu 2012;Vakili et al. 2014).
Adsorption is applied for dye removal of wastewater (Crini and Badot 2008); it is highly used due to the facility of fullscale operation; in addition, different inexpensive materials with low toxicity levels can be used as adsorbents (Crini et al. 2019). The introduction of efficient sorbent materials is an alternative to conventional adsorbents that are expensive or have a high carbon footprint such as activated carbon. Biocomposite materials are attractive alternatives for water purification technologies, since these materials are heterogeneous systems composed of two or more constituents that produce a synergistic effect by improving their properties such as their adsorption capacity (Ranđelovic et al. 2012). Biocomposite materials also display excellent adsorption properties, mechanical stability, and low cost (Srinivasan and Viraraghavan 2010;Li et al. 2019). In order to formulate biocomposite sorbents, two types of materials are necessary: an inorganic support and a biodegradable active component.
Wastewater contaminated with dyes is often acidic enabling the solubilization of chitosan due to the protonation of the amine groups; therefore, biopolymer modification using cross-linking agents can usually circumvent the problem of instability at low pH (Crini et al. 2019). Covalent crosslinking agents have been used, in particular, glutaraldehyde and epichlorohydrin (Crini et al. 2019); although these compounds generate high performance adsorbent materials, they are highly toxic (Leung 2001); therefore, their application in the treatment of wastewater is impractical. Another type of mechanism is the ionic cross-linking, consisting in the formation of ionic bonds between the amino groups of Ch and the anions of the cross-linking agent (Jóźwiak et al. 2017). The main advantage of chitosan ionic cross-linking is that it generates materials with good chemical stability in acidic media (Crini et al. 2019) and the possibility of functionalizing the biopolymer material in the synthesis of biocomposite sorbents. In this work, the salt forming ion of oxalic acid namely the oxalate ion −C 2 O 2− 4 À Á will be evaluated as a cross-linking agent and as a binding molecule with the inorganic support.
Dicarboxylic acids are used as ionic cross-linkers; in this case, the OA has been used as a Ch cross-linking agent with low toxicity compared to other types of cross-linkers (Fadzallah et al. 2014;Jóźwiak et al. 2017) due to that OA is a natural compound in some plants (i.e., black tea ) (Ghosh and Ali 2012). OA has a pK a =1.2, otherwise it dissociates in an aqueous medium into oxalate ions −C 2 O 2− 4 À Á ; these conditions cause the Ch to dissolve and reticule due to electrostatic interactions between the amino and oxalates groups (Fadzallah et al. 2014).
Cross-linked chitosan/oxalic acid hydrogels (ChOxb) have been used for the removal copper (II) (Mi et al. 2015) and Reactive Black 5 (Jóźwiak et al. 2017). Pérez-Calderón et al. (2020) reported the use of ChOxb for RR195 removal, showing that it improved the overall performance of this novel adsorbent enabling the application of this material in a higher acidic media without solubilization at pH=2.5 and increasing the adsorption capacity by 34.8% with respect to coacervated chitosan hydrogels (CP) (Pérez-Calderón et al. 2018).
Good bonding between constitutes of the biocomposite provides a high level of improvement in terms of mechanical properties and chemical stability, which are important aspects of the adsorption process. The bonding between constitutes can be established by (a) the presence of intermolecular forces or covalent bonds; (b) mechanical action in the material processing (e.g., pelletization); (c) the presence of an adhesive agent (Ranđelovic et al. 2012).
In the present work, the properties of chitosan cross-linked with oxalic acid are used to form active films by coating alumina ceramic particles. Issues have to be considered in order to formulate these new biocomposites such as the ability of the biopolymer to interact with the inorganic support and form a robust coating without suffering detachment when the adsorption process of azo dyes is produced. Additionally, the biopolymer should maintain its structure minimizing solubilization under acidic conditions that is usually an impediment if unmodified or coacervated Ch is applied.
Biocomposites formulated with pure alumina and chitosan cross-linked with oxalic acid have been developed to increase the mechanical stability of the adsorbent material. These biocomposites have been used for the adsorption of chromium (Boddu et al. 2003;Darjito et al. 2014) and fluorine (Li et al. 2013). There is scarce information concerning the combined use of inorganic materials such as alumina ceramic as support for the active biopolymer (chitosan) where the adsorption of azo dyes occurs at the films surface.
The objectives of the present study were (a) to synthetize cross-linked oxalic acid-chitosan/alumina ceramic biocomposite (BChA) for the adsorption of a reactive azo dye (RR195); (b) to characterize the morphology and structure of BChA by using Fourier transform infrared spectroscopy (FTIR-ATR), scanning electron microscopy with energy dispersive X-ray analysis (SEM-EDS), X-ray diffraction, thermogravimetric analysis, mercury intrusion porosimetry, and Z-potential; (c) to analyze the interactions present between the alumina ceramic support and the ionic cross-linked chitosan film; (d) to determine the removal percentage and adsorption capacity in batch adsorption experiments; (e) to model the adsorption kinetics; (f) to obtain the adsorption isotherms and to select the appropriate equations; (g) to calculate thermodynamic parameters of the sorption process; (h) to evaluate the effect of different competitive ions present in the solution on the adsorption of the dye; (i) to study the desorption process and regeneration of the biocomposite by applying adsorption/desorption cycles.
CerPa were formulated with 75 wt% alumina, 20 wt% bauxite, 2 wt% kaolin, and 2 wt% talc as support material. CerPa were synthesized using an Eirich High Intensity Mixer Machine (Model R20E, Erich Industrial Ltd., Brazil) where the two process steps, mixing and granulating, are performed in the same equipment.
The particles were obtained through batch experiments where the material and water were combined to form a homogeneous mixture, and the spherical geometry was achieved by adjusting the speed of the mixing tool. The particles were sieved to obtain diameters between 14 and 16 ASTM meshes (diameters between 1.18 and 1.40 mm). The CerPa were dried in a conventional oven and calcined at 1500°C for 2 h in a high temperature oven MHI (USA). The lack of RR195 adsorption capacity of CerPa was tested by mixing CerPa and RR195; this experiment corroborated that this individual material did not show adsorption properties.
BChA was prepared by coating the CerPa with the crosslinked chitosan solution. The coating capacity of CerPa was directly related to the presence of alumina in the inorganic material; this assumption was corroborated by formulating CerPa without alumina and observing in this case that filmforming capacity was weak and the biopolymer was detached from the CerPa after being in contact in a solution. The stability of the active chitosan coating was carried out using the method described by Boddu et al. (2003) with some modifications; these include a first stage of activation of the ceramic material and two consecutive stages where a coating is formed using cross-linked chitosan with OA; briefly, (i) dried CerPa was activated by stirring the particles in a OA 10% w.v −1 solution for 4 h at 25°C; after this step, the material was filtered and washed with distilled water; finally, the activated CerPa was dried in a vacuum oven LiTekvo model DZF-6030A (LiTekvo Instruments, Minhang District, Shanghai, China) at 70°C and reduced pressure of 100 mmHg to a constant dry weight. (ii) The activated CerPa was mixed with 1% w.v −1 cross-linked chitosan solution under constant shaker stirring at 25°C for 17h. To finish the first coating, the material was dried at 55°C in a Heratherm OMS 60 forced convection oven (Thermo Scientific, Germany) for 24 h. (iii) The second coating was performed by mixing the product of the previous step with 2% w.v −1 cross-linked chitosan solution under constant stirring at 25°C for 17h. The excess of filmogenic solution was removed; in addition, it was neutralized with 1 N sodium hydroxide, standing for 3 h for subsequent washing with distilled water until neutral, obtaining the BChA. (iiv) BChA was dried at 55°C for 24h in a forced convection oven.
The amount of biopolymer in the BChA was determined by weight loss after calcination of the biocomposite at 750°C for 10h using a muffle. The test was performed in triplicate, and the results were expressed as a percentage of cross-linked chitosan coating (% CCh) calculated according to Eq. 1: where W m is the mass of the dry sample and Wc of the ashes. This information is important to express the amount of adsorbent in terms of the active material during the adsorption process.

Characterization of biocomposite
Characterization of BChA and the materials used in this work were performed using Fourier transform infrared spectroscopy with an attenuated total reflection accessory (FTIR-ATR), scanning electron microscopy equipped with energy dispersive X-ray spectrometry (SEM-EDS), X-ray diffraction (XRD), thermogravimetric analysis (TGA), and Z-potential (ZP).
FTIR-ART was carried out using a Thermo Nicolet iS10 spectrometer (Thermo Scientific, MA, USA). FTIR Spectra were obtained in %Transmittance mode with 4 cm −1 spectral resolutions and 34 scans. ATR accessory consisted of diamond crystal (nominal angle of incidence=42°). XRD was performed on sample powders; diffractograms were obtained using an X-ray diffractometer Philips PW-3710 using Cu-Kα radiation and Ni at constant voltage (40 kV, 35 mA). TGA was performed with a thermogravimetric analyzer Rigaku (Thermo Plus EVO2, Osaka, Japan) using N 2 with inert gas. Surface morphology was studied using SEM-EDS; therefore, a NeoScope Benchtop JEOL JCM-6000 microscope (JEOL, USA) equipped with energy dispersion spectrometry X-ray (EDS) unit JEOL WX-36210DPP was employed. ZP was measured to characterize the surface charges of the material; a nanoparticle analyzer SZ-100-Z (Horiba Instruments Inc., Kyoto, Japan) provided with a laser diode model JUNO 10G-HO (Showa Optronics Co., Ltd., Yokohama, Japan) operating at 532 nm was used. ZP of the material was determined by using a solution-containing grinded BChA (40 mg) and 40-mL KCl 1 mM using an electrode cell (carbon, 6 mm); ZP was reported as the average of five determinations per sample. The isoelectric point (pH IEP ) of the material was established by means of the ZP determined at different pH values adjusted with hydrochloric acid and sodium hydroxide 0.1M.
Microstructural properties such as pore size distribution and specific surface were determined by mercury intrusion porosimetry using a Pascal 440 Porosimeter (Thermo Fischer Scientific, Belgica). The mean diameter of the biocomposite and the layer of the cross-linked chitosan were determined by analyzing at least fifty micrographs of BChA using a MZ-10F stereomicroscope with DFC490 camera (Leica Microscopy Ltd, Germany) that were processed using the Image-J software.

Batch adsorption studies
Batch assays were carried out for the adsorption of RR195 onto the cross-linked oxalic acid-chitosan/alumina ceramic biocomposite (BChA) varying dosage of the biocomposite (0.43-4.25 g.L −1 ), pH (2.0-12.0), and contact time (0-24 h). Initial concentration (C 0 ) and dye solution volume were 150 mg L −1 and 0.02 L, respectively. The initial pH of the dye solution was adjusted with hydrochloric acid and sodium hydroxide 0.1M. Ionic strength was controlled with a KCl 1mM solution. Electrical conductivity was determined using a conductivity meter SevenMulti S47 (Mettler Toledo, Switzerland) equipped with an electrode InlLab 731.
An orbital shaker with temperature control was used to maintain constant conditions during the experiment (125rev min −1 , 298K). The final dye concentration (C f ) was determined by UV-visible spectrophotometry using Hach DR-2800 spectrophotometer (Loveland, CO, USA) at a wavelength of 538 nm.
The results were evaluated by determining the percentage of dye removal (%RE) and the adsorption capacity (Q) at a given time t (Q t ) using Eqs. 2 and 3.
where C t is the concentration at a given time t, V is the volume of the solution (L), and W is the dose of BChA (g) in terms of the mass of coating using the cross-linked chitosan.
Under equilibrium conditions, the percentage removal (%RE e ) was calculated using Eq. 2 by replacing C f with C e (equilibrium concentration), and the equilibrium adsorption (Q e ) was calculated using Eq. 3 by replacing C t with C e .

Kinetics, adsorption isotherms, and thermodynamic studies
The assays were carried out using the doses that achieved the best adsorption performances. Moreover, different pH values were tested to evaluate the cross-linking effect of chitosan on the chemical stability of the biocomposite. The equations used for kinetics, isotherm, and thermodynamic studies are shown in Table 1 (Eqs. 4 to 25).
To determine the thermodynamic parameters, the sorption isotherms were determined at different temperatures (298, 308, and 318 K). Using Van't Hoff equation (Eq. 21), the enthalpy (ΔH) and entropy (ΔS) were calculated. In this case, equilibrium partition constant (K P ) was obtained in order to determine the thermodynamic parameters (Zhang et al. 2014;Tran et al. 2017;Pérez-Calderón et al. 2018, 2020. K P was estimated as described in Pérez-Calderón et al. (2020, 2018; K P (Eq. 22 in Table 1) considers the ratio between the activity of the dye adsorbed by the solid (a s ) and the activity of the dye in the solution at equilibrium (a e ). Another possibility is to express K P in terms of the ratio between activity coefficients (γ between activit, K P is the ratio between the activity coefficient of the adsorbed dye (γ s ) and the activity coefficient of the dye in the solution at equilibrium (γ e ). If γ is used, the adsorbed dye concentration (C S Eq. 24) and the dye concentration in equilibrium (C e ) must be considered.
The activity coefficient (γ) approaches unity when the (C S )➔0 and (C e )➔0, therefore Eq. 22 can be written as Eq. 23. By plotting Ln (C S /C e ) vs C s and extrapolating to C e ➔0, K P was calculated at each experimental temperature (Tran et al. 2017). The Gibbs free energy (ΔG) was determined using Eq. 25.

Effect of competitive ions
The effect of competitive ions and the changes of ionic strength were tested by varying the different inorganic ions present in effluents from the textile industry (Zhu et al. 2010(Zhu et al. , 2016Radovic et al. 2014). The presence of inorganic ions was analyzed in the adsorption process by evaluation of %RE; the tested ions were Cl − , NO − 3 , and SO 2− 4 NO − 3 ; at a concentration of 5mmol L −1 . In all cases, the solutions were prepared with Milli-Q water. The C 0 of dye was 200 mg L −1 , and the salts used were KCl, KNO 3 , and K 2 SO 4 .

Regeneration of the biocomposite
Consecutive cycles of desorption/adsorption were conducted; for the regeneration stage, RR19 was desorbed from the BChA used in a previous adsorption experiment with an initial dye concentration of C 0 =200mg L −1 . BChA was kept in contact with Milli-Q water (0.020 L) at highly alkaline conditions (pH=12), and the test was performed at a constant agitation and temperature. The regenerated biocomposite material was dried in a forced convention oven at 55°C until constant weight and submitted to adsorption cycles with a RR195 initial concentration of C 0 =200 mg L −1 (V=0.020 L); in this way, three adsorption cycles were completed with two intermediate desorption cycles.

Statistical analysis
For the batch adsorption assays, the equilibrium isotherms, kinetic equations, and thermodynamic parameters were evaluated from linear and nonlinear regressions using Origin-Pro 8 software (Origin Lab Corporation, Northampton, MA, USA). The goodness of fit was evaluated taking into account the determination coefficient (R 2 ), mean average perceptual error (MAPE), and chi-square distribution error (χ 2 ); MAPE and χ 2 were determined with the Eqs. 26 and 27 respectively, where y e is the experimental value and y p is the value predicted by the model.
The analysis of variance (ANOVA) was conducted to assess significant differences between the samples; means were compared by Fisher LSD test using a 95% confidence level (significant difference, P≤0.05).

Results and discussion
Characterization of the biocomposite

Surface properties and morphology
The mean diameter of cross-linked oxalic acid-chitosan/alumina ceramic biocomposite (BChA) and alumina ceramic particles (CerPa) was 1.35 mm (SD=0.98) and 1.18 mm (SD=0.35), respectively; these values were obtained from micrographs ( Fig. 1a-b) of the materials, which were analyzed using Image-J software. Micrographs of the surface morphology of CerPa and BChA obtained by SEM before the adsorption process are shown in Fig. 1a-b. SEM micrograph of BChA (Fig. 1b) showed the presence of chitosan (Ch) evidenced by a smoother and softer surface of the biocomposite  a=k C 0 (w eq −1) w eq : Relative equilibrium uptake 12 b=2 k C 0 τ 0.5 (w eq −1) Adsorption isotherm models 13 Langmuir  compared with the surface of CerPa (Fig. 1a). The surface morphology of the BChA was attributed to the cross-linked Ch film that coated the surface of the ceramic particle. EDS microanalysis spectrum (Fig. 1b) shows that the surface of BChA contained oxygen (O), carbon (C), nitrogen (N), and aluminum (Al). The weight percentage of C in BChA was 28.57%, being this value higher than in CerPa (8.26 %, Fig.  1a). In addition, the EDS spectrum of BChA showed the presence of N, due to the cross-linked chitosan coating (CCh) on the ceramic material. Surface microstructural properties were determined by mercury intrusion porosimetry; the pore size diameter determined was 1165.2 nm for CerPa and 394.5 nm for BChA, and the specific surface area of these materials were 0.496 m 2 g −1 and 1.42 m 2 g −1 for CerPa and BChA, respectively. The increase in the specific surface of the biocomposite is attributed to the cross-linked chitosan coating over the surface (micropores and crevices) of CerPa; similar results were reported (Zhang et al. 2012).
The amount of biopolymer that is coating the CerPa was determined by calcination of BChA, and the results were expressed as percentage cross-linked chitosan coating (%CCh). According to Eq. 1, %CCh was 1.7% (SD=0.4); this result is important since it is the amount of active sorbent material that generates interactions with dye molecules.  (Jagtap et al. 2011); in addition, bands between 500 and 1000 cm −1 characteristic for aluminum oxides (Jagtap et al. 2011) were observed.
OA was used to bind the alumina of the CerPa and Ch; BChA was synthesized by (i) activating CerPa with OA and (ii) coating it with an ionic modified Ch solution. Different chemical reactions are generated in these stages. In the first stage, the use of OA generates the esterification reaction between the alumina of the CerPa and the oxalate ion, producing the alumina oxalate (III) complex (Boddu et al.,2003;Dobson and McQuillan 1999); this reaction is represented in Fig. 2a. Figure 1d shows in the spectrum of BChA, the presence of the band at 1723 cm −1 characteristic of the stretching of −C = O group of the ester group confirming the presence of this type of bond.
In the second stage, an electrostatic interaction between the alumina oxalate (III) complex and the protonated free amino groups of Ch (Boddu et al.,2003;Darjito et al. 2014) occurs, being this reaction represented in Fig. 2b. Results show the chemical interaction between the different constituents (inorganic support and biopolymer) of the new synthesized biocomposite generates an active and stable matrix.
The coating of the CerPa was carried out using ionic crosslinked chitosan where the ionic interaction occurs between the carboxylate anions of OA and the protonated amino group of the Ch (Pérez-Calderón et al. 2020). Figure 1d shows the existence of oxalic/chitosan ion cross-linking was evidenced (a) at 1685 cm −1 band shift of the symmetric stretching of the carboxylic group originally present at 1698 cm −1 of the OA spectrum (Mi et al.2015); (b) at 1517 cm −1 the shift and deformation of the band of the amino group originally overlapped at 1550 cm −1 , due to the ionic interaction of the protonated amino group of the chitosan molecule with the carboxyl anions of the OA (Fadzallah et al. 2014;Mi et al. 2015;Pérez-Calderón et al. 2020); the reaction is represented in Fig.  2c. Figure 1d confirms the presence of Ch in BChA; the observed bands were (a) at 2951 cm −1 and 2885 cm −1 , the symmetrical stretching of the -CH aliphatic and vibration of the C − H tensions, respectively; b) at 1065 cm −1 and 1027 cm −1 , the bands correspond to the vibrations of the stretching of the -C − OH group characteristic of polysaccharides. In addition, in the spectrum of the BChA, the band at 1397 cm −1 , which is characteristic of the asymmetric stretching of -C = O provided by the OA, and at 659 cm −1 , the -Al − O − Al bending, were observed. Figure 1e shows the diffractograms for CerPa and BChA. For CerPa, it was confirmed that the predominant phase was α-alumina due to the presence of several diffraction peaks at 2θ=25. 59°, 35.16°, 27.58°, 43.36°, 52.44°, 57.48°, 66.48°, and 68.16°(Feret et al. 2000). In addition, small peaks were observed at 2θ=36.84°, 59.56°, and 64.94°that are of spinel (MgAl 2 O 4 ) and at 22.24°due to cristobalite (SiO 2 ); this was corroborated with results obtained by EDS microanalysis presented in Fig. 1a showing the existence of magnesium (Mg) and silicon (Si). For BChA, the characteristic diffraction peaks of the CerPa were observed. In addition, three new peaks were presented in the positions 2θ=14.42°, 24.69°, and 30.30°, attributed to the OA/Ch crosslinking (Mi et al. 2015).

XRD and TGA analysis
The TGA curves of Ch, OA, and BChA are shown in Fig.  1f. For Ch, a weight loss was observed at 100°C, typical of water release. Weight losses occurred also at 277°C and 642°C that are characteristic of the total melting and decomposition of the biopolymer (Rekik et al. 2017). OA TGA curve showed at 143°C a distinct weight loss attributed to the complete dehydration of the sample (Mi et al. 2015); at 173°C, unstable intermediates such as H 2 O, CO, and CO 2 are Fig. 2 Schematic diagram of the a complexation of alumina from the ceramic particle by oxalic acid reaction. b Electrostatic interaction between the free amino group of chitosan and the alumina (III) oxalate complex. c oxalic acid cross-linking the chitosan molecule by means of ionic interactions released, and finally at 214°C, the total decomposition of the OA is produced.
For BChA, an initial weight loss attributed to the release of water is shown; subsequently at 244°C, the thermal event is due to the rupture of the electrostatic interaction in the ionic cross-linking of Ch with OA (Mi et al. 2015). The weight loss at 458°C can be attributed to the breakdown of the ester bond of the alumina (III) oxalate complex, and finally at 638°C, the total weight loss characteristic of biopolymer fusion occurred. The percentage of total weight loss for the BChA corresponded to 2.04%; this is comparable to the %CCh determined by weight loss the biocomposite from calcination (%CCh=1.7).

Experimental adsorption conditions: effect of time, doses, and pH
The removal of RR195 using BChA was analyzed, and the effect of contact time on Q, testing different sorbent doses, is shown in Fig. 3a. Q increased with time achieving asymptotic values at equilibrium (Q e ); at this condition, the amount of dye that is adsorbed is in a dynamic equilibrium state with the amount of dye that is desorbed from the sorbent (Li et al. 2013). Figure 3a shows that for concentrations ranging between 0.85 and 4.25 g L −1 , the equilibrium was achieved after 15h; however using the lowest dose (0.43 g L −1 ), this equilibrium condition was not reached even after 25 h. Figure 3b shows %RE e and Q e as functions of sorbent doses; nonsignificant differences in %RE e were found for sorbent doses ranging between 0.85 and 4.25 g L −1 , therefore the lowest dose (0.85 g.L -1 ) was chosen to be used in the subsequent studies, because it led to the highest Q e .
The pH of the solution is an important environmental factor that affects the adsorption process (Azimi et al. 2017). Figure 3c shows the variation of the %RE e dye with pH; the pH of the solution influences the degree of protonation of the sorbent, thus determining the specific charge of the binding sites and the absorption performance of the sorbent (Crini et al. 2019). Most of the functional properties of sorbents made from chitosan depend on the presence of free amino groups (NH þ 3 ) since they contribute to the cationic nature of the biosorbent. It should be noted that some hydroxyl groups (especially at the C-3 position) may contribute to adsorption but to a lesser extent than the amino groups (Crini and Badot 2008). Pérez-Calderón et al. (2020, 2018 reported that RR195 adsorption was favored by using chitosan-based sorbent materials, under acidic conditions (pH<4); this is due to the electrostatic attraction between the positive charged surface of the material by protonation of NH þ 3 (Eq. 28) and the negative charge of the ionic dye. The pH affects the structural stability of dye molecules (in particular, the dissociation of their ionizable sites, Eq. 29); in acid conditions, the sulfonated groups of the RR195 are negative (−SO − 3 ) since pK a <1 (Crini and Badot 2008;Yaman and Gündüz 2015); under this conditions, the interaction between RR195 dye and biocomposite sorbent was present, reaching the adsorption of the dye (Eq. 30). Figure 3c shows that the highest %RE e was at pH=2.0 (%RE e =91.33 SD=0.15). The ability to adsorb at this very low pH condition can be attributed to the stable structure of the biopolymer gained by the ionic crosslinking of Ch with OA which avoids the solubilization of the biopolymer (Gonçalves et al. 2019). Pérez-Calderón et al. (2020, 2018 reported that the maximum adsorption of RR195 onto ChOxb and CP materials was at pH=4.0. In comparison, cross-linked chitosan bond to the ceramic material (BChA) showed maximum adsorption at pH=2.0 resulting in a material with better adsorption performance and mechanical stability. The reasons for this improvement are the larger exposed active surface of the material and the increase of the number of free amino groups allowing them to interact more effectively with the sulfonic groups of the dye (Li et al. 2013).
Since the pK a of chitosan is 6.5, at pH> 6.5, the amino groups are not likely to be protonated, and the adsorption process of the anionic azo dye will not be favored which in turn will decrease the %RE e . At pH>7.0, the existence of electrostatic repulsion forces between the dye anions and the unprotoned surface of the adsorbent causes the further decrease of the %RE e .
The surface charge of the biocomposite is a key factor when there is an electrostatic interaction during the adsorption process. Z-potential (ZP) measurements allowed determining the surface charge of adsorbents. The isoelectric point (pH IEP ) occurs when the surface charge of the adsorbent is close to zero (ZP near zero), indicating that the surface charge becomes more positive at pH values near and at below the pH IEP . In order to know the pH IEP of BChA, the change of ZP at different pH values before and after removal was studied (Fig. 3c).
According to the results obtained, the pH IEP was between pH 6 and 7 before and after adsorption; in this range, the biocomposite has a surface charge near to neutrality. In BChA, the decrease of ZP, after the adsorption, shows that the amino groups of the material interact electrostatically with the dye molecules (Pérez-Calderón et al. 2020).

Adsorption kinetics
Adsorption kinetics assays were carried out using a dose of 0.85 g L −1 of biocomposite and different initial dye concentrations (C 0 =100, 200 mg L −1 ). In order to study the effect of pH on the adsorption kinetics, experiments at pH=2.0 and 4.0 were conducted. Results of the adsorption kinetics are depicted in Fig. 4 a-b, showing that after 17h equilibrium conditions were established for both pH values and C 0 tested; the maximum adsorption performance was reached at pH=2.0 (Q e (17h)= 203.98 mg g −1 (SD=2.4) C 0 =200 mg L −1 ), with an increase in the removal capacity compared to pH=4.0 (Q t (17h)= 176.01 (SD=0.45) g L −1 C 0 =200 mg L −1 ), however differences in the process rates were not significant. Results confirm that under conditions of high acidity (pH=2.0) protonation of the free amino groups in the matrix generates a significant increase of the adsorption capacity. Experimental results were analyzed by different kinetic models (Table 1). Pseudofirst order (PS1, Eq. 4, Table 1), where k 1 (min −1 ) is the rate constant of the first-order adsorption and Q e (mg g −1 ) is the equilibrium adsorption capacity. The pseudosecond-order model (PS2) is given by Eq. 5 (Table 1) where k 2 (g mg −1 min −1 ) is the second-order rate constant (Ho and McKay 1999). Elovich's empirical model (Eq. 7, Table 1) assumes that the active sites are heterogeneous and therefore exhibit different activation. In Eq. 7, α E is the initial adsorption rate related to chemisorption, and β E is desorption rate constants at the surface coverage. The intraparticle diffusion model (Eq. 8, Table 1) describes the diffusion of the dye and has been widely used in chitosan sorbents such as hydrogels (Crini and Badot 2008). This model explains that there are two different regions in the kinetic curve corresponding to surface adsorption (rapid external diffusion) and intraparticle diffusion stage (Weber and Morris 1963).
The MSR-DCK model (Haerifar and Azizian 2013) considers that the diffusion of the dye molecules in the sorbent and its adsorption in the active sites of the biocomposite occur Equilibrium removal (RE e %) and equilibrium capacity (Q e ) as functions of BChA doses. Experimental conditions for a-b C 0 =150 mg L −1 T=298 K, shaker speed= 125 rev min −1 , pH=4, electrical conductivity=3.96 mS cm −1 . c Effect of pH on %RE e and Z potential before and after adsorption. Experimental conditions for c C 0 =150 mg L −1 T=298 shaker speed= 125 rev min −1 ,pH=4, dose BChA=0.85 g L −1 . Different letters in the %RE e and Q e indicate significant differences between the samples (P≤ 0.05) simultaneously, the analytical solution (Eq. 9, Table 1) was reported by Haerifar and Azizian (2013), where the relative equilibrium uptake (w eq ) is 0≤w eq ≤1. The effect of diffusion was introduced by the coefficient b (Eq. 12, Table 1). Table 2 shows the kinetic parameters calculated by nonlinear regressions of the models (Ps1, PS2, Elovich, intraparticle diffusion, and MSR/DCK). Goodness of fit was evaluated using the determination coefficient (R 2 ), mean average perceptual error (MAPE; Eq. 26), and chi-square distribution error (χ 2 ; Eq. 27). Analyzing the fitting of the nonlinear regressions R 2 , MAPE, and χ 2 , it can be concluded that PS2 model provided the best fit with the experimental data. Zhang et al. (2012) reported that for methyl orange adsorption using unmodified Ch/alumina biocomposite the model that best fit was PS2.
By linearization of the PS2 model equation (Eq. 6; Table 1), a linear relationship was obtained from the plot of t/Q t vs. time for each C 0 (Haerifar and Azizian 2013) (Fig. A3  (Supplementary Information). The PS2 kinetic model is based on the assumption that the rate-limiting step is the surface adsorption where the removal from a solution is due to physicochemical interactions between the two phases. Table 2 shows k 2 parameter of PS2 model for different tested conditions; k 2 decreased when C 0 of the dye increased; this is attributed to the increased competition of the dye molecules for the active sites on the surface of the sorbent material at high concentrations. In contrast, at low C 0 , the competition for the active sites of the sorbent is low, increasing the rate of the adsorption process (Zhang et al. 2014).

Adsorption isotherms
The adsorption isotherm equations that have been tested in the present work are shown in Table 1: Langmuir (Eq. 13), Freundlich (Eq. 15), Temkin (Eq. 16), Eq. 17), and Dubinin-Radushkevich (DR, Eq. 18). The isotherms were determined at pH 2.0 and 4.0, and the results are shown in Fig.4c-d. Freundlich model considers a  (Freundlich 1906); the equation that represent Freundlich's model (Eq. 15) is presented in Table 1. The parameters of this model are the adsorption capacity coefficient K F ((mg g −1 ) (mg L) −1/n ) and the value of adsorption intensity parameter n (dimensionless) which indicates the favorability of the adsorption (Mckay et al. 1982).
Temkin (Temkin and Pyzhev 1940) model is given by Eq. 16 (Table 1); A t (L mg −1 ) is the equilibrium constant representing the maximum binding energy, and B t is a coefficient related to the adsorption heat. The R-P model (Redlich and Peterson 1959) combines both Freundlich and Langmuir models, and it is an empirical equation that includes three parameters (Eq. 17, Table 1); it can be used in a range of concentrations to explain the equilibrium of adsorption in homogeneous or heterogeneous systems. Equation 17 describes this isotherm where K RP (L g −1 ) and α RP (L mg −1 ) are the constants of the R-P isotherm and β RP is an exponent with values ranging between 0 and 1; the results of this model are shown in Table 3. D-R model assumes two possible mechanisms (physical or chemical); Eq. 18 in Table 1 represents this model, where ϵ (kJ mol −1 ) is Polanyi's potential that is calculated using Eq. 19 where R represent the gas constant (8.314 ×10 −3 kJ K −1 mol −1 ) and T the temperature (K). From the value of B DR (determined by regression Eq. 18), it is possible to calculate the average free adsorption energy (E, kJ mol −1 ) using Eq. 20. According to the statistical parameters (R 2 , MAPE, and χ 2 ) shown in Table 3, Langmuir was the model that best fitted the experimental data; Fig. 4c-d shows predicted values (dashed line). Langmuir model (Eq. 13, Table 1) assumed that adsorption occurs in a single layer and can occur explicitly at a finite (fixed) number of specific localized sites, without any lateral interaction and without steric hindrance between the adsorbed molecules (Vijayaraghavan et al. 2006). The maximum adsorption capacity (Q m , mg g −1 ) and the affinity of the adsorbate for the adsorbent (K L , L mg −1 ) were calculated (Eq. 13). Q m was evaluated at different temperatures showing that higher temperatures favored the adsorption process. Table 3 shows the results obtained for this model, Q m at 298K and pH=2 was 333.9 mg g −1 and for pH=4 the value of Q m was 254.3 mg g −1 . These results show the advantages of using a cross-linked Ch with OA, since it increased the adsorption capacity at a very low pH, because the coating also stabilized the structure of the BChA avoiding biopolymer solubilization. Moreover, according to these results, Q m is the highest reported for RR195 removal in comparison with other adsorbent materials. Table 4 shows different adsorbent materials for the removal of RR195 in decreasing order of Q m . As can be observed, BChA is the material with the best adsorption capacity for the RR195 removal.
Several authors have used the separation factor (R L ) to analyze the results of Langmuir's model. R L was calculated using Eq. 14 (Table 1) which takes into account the constant (K L ) determined from the nonlinear regression of Eq. 13 and the initial concentration (C0). RL value indicates the adsorption nature to be either unfavorable (R L >1), linear (R L =1), favorable (0<R L <1), or irreversible (R L =0) (Zhang et al. 2014). According to the results shown in Table 3 for the different tests, the process is favorable. On the other hand, plotting R L vs C 0 (Fig. A4 a-b; Supplementary Information) the R L values decreased when initial dye concentration increased, indicating that the process was favored at high initial dye concentrations (Zhang et al. 2014).

Thermodynamic parameters
To determine the thermodynamic parameters, results from the adsorption isotherms obtained at different pH values (2.0 and 4.0) and temperatures (298, 308, 318 K) were analyzed. K P values were calculated from intercepts obtained from Ln(C s / C e ) vs C s plots for each of the temperatures tested (as explained in the "Kinetics, adsorption isotherms, and thermodynamic studies" section); the results of the linear regressions are presented in Fig. 5a-b. According to Eq. 21 (Table 1), ΔH and ΔS were determined by plotting Ln (K P ) as a function of T −1 (Fig. 5c-d), and the results are shown in Table 5 for pH 2.0 and 4.0; these were considered satisfactory because the R 2 of the linear regressions were near 1 (at pH=2.0, R 2 =0.99 and at pH=4.0, R 2 =0.96).
ΔG was calculated using Eq. 25; in all cases, ΔG values were <0 (Table 5). Negative ΔG values indicated that the adsorption of dye RR195 on the BChA is spontaneous (Wang and Chen 2007); besides, it can be observed that absolute value of ΔG decreased with increasing the temperature, therefore in the adsorption process the degree of spontaneity of the reaction decreased at higher temperatures (Zheng et al. 2015). The obtained values of ΔH at both pH values were negative (Table 5); these results indicated that the process was exothermic; the attraction force between adsorbent and adsorbate results in a loss of heat. Table 5 shows ΔS results; the negative values of ΔS indicated that the randomness decreases at the solid-solution interface during the RR195 adsorption onto BChA (Saha and Chowdhury 2011). Different researches using sorbents synthesized with chitosan for reactive dyes adsorption have reported similar results (process (ΔG<0), exothermic (ΔH<0), and ΔS<0) (Chiou and Li 2003;Zhang et al. 2012). Interaction and adsorption mechanism between the biocomposite and the RR195 dye The interaction between the sorbent material (BChA) and the dye was analyzed. Figure 6a shows the micrograph obtained by stereomicroscopy, and the characteristic color of the dye can be observed at the surface of the BChA. The EDS microanalysis spectrum at the surface of the BChA after adsorption (Fig. 6b) shows the presence of sulfur (S), corroborating the existence of RR195 interacting with BChA. Figure 6c shows a cross-section of BChA after adsorption, in which it was possible to observe the dye adsorbed on the coated CerPa from the image analysis using Image-J; the thickness of the attached film was 7.23 μm (SD=0.6).
FTIR-ART transmission spectra of the material after adsorption process (BChA+RR195) is shown in Fig. 6d; the FTIR-ATR spectra of the RR195 dye is shown in Fig. A2 (Supplementary Information), and the analysis of the main bands is described in Table A2 (Supplementary Information).
The main changes in the BChA+RR195 (after adsorption) spectra with respect to BChA were the presence and changes of the following bands: (a) at 3455 cm −1 due to high interactions of hydrophilic nature (stretching groups -NH, -OH, and the resonant aromatic structure =CH of the dye); (b) the reduced intensity of the band at 1517 cm −1 , can be attributed to electrostatic interactions and/or hydrogen bonds between the amino groups of chitosan and ionic sulfonated groups of the dye (Wang et al. 2016); (c) at 1319 cm −1 due to the S = O vibrations of SO 2 group of the dye molecule; (d) at 1612 cm −1 corresponding to the vibrational stretching of the aromatic cycle of the dye (this band was originally located at 1617 cm −1 in the spectrum of the dye (Fig. A2)). According to the results obtained, the pH IEP was between pH 6 and 7 before and after adsorption; in this range, the biocomposite has a surface charge near to neutrality (Fig.  3c). In BChA, the decrease of ZP, after the adsorption shows that the amino groups of the material interact electrostatically with the dye molecules (Pérez-Calderón et al. 2020).
As reported in the "Thermodynamic parameters" section, the absolute ΔH values obtained in this work (<85 kJ mol −1 ) were near the lower limit for heats that correspond to chemisorption. The chemisorption generally falls into a range of 80-200 kJ mol −1 . Additionally, kinetic adsorption results were adequately fitted with the pseudosecond-order model (Eq. 6; Table 1) in which the adsorption removal from a solution is due to physicochemical interactions between the two phases.
Based on the experimental results of pH IEP , changes in the FTIR-ATR spectra, and the influence of the pH values on the adsorption capacity, it is possible to conclude that the nature of the adsorption process could be a combination of physical and chemical interactions. This is in agreement with Crini and Badot (2008) which report that adsorption on solids can be due to chemical or physical adsorption, but the dividing line between them is not always clear.

Effect of competitive ions
The results of the effect of competitive ions at the tested concentration (5 mmol L −1 ) and pH=2.0 showed for Cl − , NO − 3 , and SO 2− 4 that %RE were 83.3 (SD=1.3), 70.1 (SD=1.0), and 55.6% (SD=0.90), respectively; moreover, the electrical conductivities of the tested media were 2.62, 2.70, and 3.19 mS cm −1 for each case respectively. For the solutions without ions added, the %RM was 87.3 (SD=1.6), and the electrical conductivity was 2.30 mS cm −1 . According to the results of the ions tested, the increasing order of influence was Cl − <NO − 3 <SO 2− 4 ; Cl − was the ion that least affected the adsorption process. In contrast, the ion that most affected the removal was SO 2− 4 , due to the high electronegativity which generates electrostatic interactions between this moiety and the protonated amino groups in the BChA. This creates a competitive situation for active sites between the sulfonates groups (−SO − 3 ) of the dye and the ions from the salt solution. Similar results for −SO − 3 have been reported using other types of sorbent materials such as polystyrene and polyacrylate polymeric matrices for Orange G adsorption (Zhu et al. 2016).

Regeneration and desorption of the biocomposite
Different methodologies have been used for the regeneration of the sorbent materials made with chitosan; (a) by changing the medium pH to alkaline conditions (Chatterjee et al. 2005) or (b) by using solvents such as ethanol (Guibal 2004). In this work, the regeneration of the sorbent material was carried out by changing the solution to basic pH values.
In the desorption stage of BChA when alkaline conditions were established, there was a deprotonation of the amino groups (under acidic conditions the amino group is protonated −NH þ 3 , and under alkaline conditions it is −NH 2 ) causing the electrostatic interactions between the chitosan and the dye to decrease (Chen et al. 2011). As explained in the "Interaction and adsorption mechanism between the biocomposite and the RR195 dye" section, the adsorption process is governed by the electrostatic force between the protonated amino groups and the ionic azo dye. Figure 7 shows the results of three adsorption cycles (C 0 =200 mg L −1 ; pH=2) and two desorption cycles (pH=12) of RR195 using BChA. In the desorption stage, it was evident that the BChA was capable of being regenerated and used in consecutive cycles of adsorption.
Using BChA, the time required to complete the first and second desorption cycles was 3h and 1.9 h respectively; these values were shorter than in the case of ChOxb, with desorption time of 15 h and 15.2 h for the first and second cycle respectively (Pérez-Calderón et al. 2020). In the second and third adsorption cycles, Qe values of 182.9 mg g −1 (SD=0.04) and 122.5 mg g −1 (SD=0.18) respectively, were achieved. These results indicated a 5.5% and 36.6% reduction in BChA adsorption capacity in comparison with the first adsorption cycle. On the other hand, the %RE achieved after the third cycle of adsorption was 52.4% (SD=0.08), in comparison with the first and second adsorption cycle which attained a %RE=87.3(SD=1.6) and %RE=78.3(SD=0.5), respectively. Additionally, the sorbent material weight at the beginning of each cycle was measured, and no significant differences were observed between the first and second cycle; however after the third cycle, a reduction of the BChA was evidenced (1.59%). This value indicates that there might be a solubilization of the biopolymer coating after consecutive cycles that resulted in a loss of mass of sorbent, which correlates with a significant decrease in the %RE (78.3) and Q e (122.5 mg g −1 ). As a result, the use of BChA could be applied for two consecutive adsorption cycles maintaining its structural property and selective action for RR195 removal.

Conclusions
Cross-linked oxalic acid-chitosan/alumina ceramic biocomposite (BChA) was synthesized as a novel sorbent material for the removal of an anionic reactive azo dye (RR195) from wastewater. The biocomposite was prepared using ceramic particles containing 75% alumina which were coated with chitosan previously cross-linked with oxalic acid. The physicochemical and structural characteristics of BChA allowed identifying the presence of cross-linked chitosan in the biocomposite, interacting by electrostatic bonds with alumina (III) oxalate compound, product of the functionalization of the ceramic support. ? Fig. 7 Adsorption/desorption cycles of Reactive Red 195 (RR195) using crosslinked oxalic acid-chitosan/alumina ceramic biocomposite (BChA) Batch experiments showed that the maximum removal percentage (>80%) occurs at pH=2 due to the cross-linking effect of the oxalic acid. This result shows that the coating of crosslinked oxalic acid-chitosan improved adsorption performance for RR195 compared to other chitosan-synthesized materials; this improvement is attributed to the increase of exposed free amino groups on the active surface of BChA. Another reason is the stable structure of the chitosan gained by the ionic crosslinking with oxalic acid which avoids the solubilization of the biopolymer in acid medium.
Langmuir equation gave the best fit to the experimental sorption isotherm. The BChA showed a high adsorption capacity (367 mg g −1 at pH=2, 318 K) which is the highest reported for RR195 removal in comparison with other sorbents materials.
The governing adsorption mechanism was the electrostatic interaction between the anionic dye (due to the sulfonate groups) and the free amino groups of the cross-linked chitosan, enabling the successive desorption/regeneration cycles of the biocomposite. The kinetic behavior corresponded to a pseudosecond-order model. Experiments performed at different temperatures allowed the calculation of the thermodynamic parameters showing that the adsorption process was spontaneous, exothermic, and enthalpy controlled. The presence of different tested anions affected the removal according to SO 2− 4 > NO − 3 > Cl − due to the competing effect for the free amino sites.
The novel biocomposite showed a better stability at low pH and can be applied as a cost-effective and environmentally friendly sorbent for azo anionic dye removal from wastewater. The use of the oxalic acid cross-linked chitosan coating on a ceramic support improved the adsorption capacity compared to other materials and allowed the use of the sorbent in acidic.
Funding The authors gratefully acknowledge the financial support from the Universidad Nacional de La Plata (UNLP), Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Comisión de Investigaciones Científicas (CIC) de la Provincia de Buenos Aires, ANPCYT (Agencia Nacional de Promoción Científica y Tecnológica), and Ministerio de la producción, Ciencia y Tecnologia and the collaboration of the Centro de Tecnología de Recursos Minerales y Cerámica (CETMIC) in the characterization studies.
Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate Not applicable

Consent for publication Not applicable
Competing interests The authors declare no conflict of interest.