Recovery and valorization of tannins from a forest waste as an adsorbent for antimony uptake

Accepted Manuscript Recovery and valorization of tannins from a forest waste as an adsorbent for antimony uptake Hugo Bacelo, Bárbara R.C. Vieira, Sílvia C.R. Santos, Rui A.R. Boaventura, Cidália M.S. Botelho PII: S0959-6526(18)32059-6 DOI: 10.1016/j.jclepro.2018.07.086 Reference: JCLP 13540 To appear in: Journal of Cleaner Production Received Date: 09 March 2018 Accepted Date: 09 July 2018 Please cite this article as: Hugo Bacelo, Bárbara R.C. Vieira, Sílvia C.R. Santos, Rui A.R. Boaventura, Cidália M.S. Botelho, Recovery and valorization of tannins from a forest waste as an adsorbent for antimony uptake, Journal of Cleaner Production (2018), doi: 10.1016/j.jclepro. 2018.07.086 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT 10900 WORDS Recovery and valorization of tannins from a forest waste as an adsorbent for antimony uptake Hugo Bacelo, Bárbara R.C. Vieira, Sílvia C.R. Santos *, Rui A.R. Boaventura, Cidália M.S. Botelho Laboratory of Separation and Reaction Engineering - Laboratory of Catalysis and Materials (LSRE-LCM), Chemical Engineering Department, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal *Corresponding author. E-mail address: scrs@fe.up.pt Abstract In the current context, it is imperative to seek for a sustainable management and an efficient use of natural resources. Pine pinaster bark is a forest and industrial waste whose chemical richness is commonly ignored. In this work, tannins were extracted from P. pinaster bark and converted into adsorbents through polymerization. Aqueous (alkaline) extraction yielded more formaldehyde-condensable phenols than an organic extraction using ethanol (Soxhlet) (53±8 vs. 13±4 mg of gallic acid equivalents per g of bark). The polymerization reaction was optimized and higher amounts of adsorbent were produced using 6.0 mL of 0.25 mol L–1 sodium hydroxide solution and 0.40 mL of formaldehyde (36 % wt) per g of extract. The performance of the produced adsorbent was assessed on the sequestration of Sb(III) and Sb(V) species from water. The adsorbent was effective for both species, in diluted and heavily-contaminated waters, providing maximum adsorption capacities (Langmuir model) of 24±3 mg g–1 (pH 6) and 27±7 mg g–1 (pH 2), respectively for Sb(III) and Sb(V). No significant effect was observed due to the presence of arsenic, chloride, nitrate, sulfate or phosphate and little influence was obtained when a tailings water from a mine site was used as aqueous matrix. Electrostatic attraction and Sb(III) and Sb(V) complexation with polyol groups of tannin-adsorbents were the suggested adsorption mechanisms. Moreover, tannin-adsorbents were stable at different pH (no color leaching; total dissolved carbon ≤16 mg L‒1) and their production does not require high energy or expensive chemicals. Keywords: Maritime pine; water; antimonite; antimonate; adsorption ACCEPTED MANUSCRIPT 1. Introduction In 2015, countries adopted a set of goals as part of a new sustainable development agenda of the United Nations (UN, 2015). Among others, targets include the achievement of affordable potable water for all, the improvement of water quality by minimizing the release of hazardous chemicals, the increase in water recycling, safer and cheaper reuse. Adsorption is a simple, cost-effective technique, applicable in water treatment or at the polishing stages of wastewater remediation that can contribute significantly to meet these targets. Biosorption, in particular, has been widely studied (Nishikawa et al., 2018; Rao and Khatoon, 2017; Ungureanu et al., 2017; Yargıç et al., 2015) but, despite the great efforts to achieve efficient and economic biosorbents, scarce exploitation in industrial scale was found (Fomina and Gadd, 2014). One of the most important limitations is probably related to the adsorbent stability. Many of the reported “low-cost adsorbents” are not stable enough, causing secondary pollution (e.g. leaching of organic matter, metals), which is a barrier to their use in more “clean” applications (e.g. water treatment for human consumption, wastewater treatment for further reuse). In addition, some of these materials have unsuitable particle size and mechanical properties that can cause clogging problems in packed beds. In fact, there is a need to balance the cost of acquisition/production of adsorbents, their efficiency and these properties. The present work emerges as an answer to these gaps and proposes a novel biosorbent, using tannins, which are natural and ubiquitous polyphenols, as precursors. Due to their phenolic nature, tannins are highly reactive towards aldehydes and a stepgrowth polymerization reaction occurs between tannins and formaldehyde (cross-linking agent) leading to the formation of an insoluble non-linear polymer where properties of interest are available in a stable material that can be used as biosorbent. Tannin-adsorbents have been studied for the removal and separation of heavy metals, critical metals and organics (Alvares Rodrigues et al., 2015; Beltran-Heredia et al., 2012; Luzardo et al., 2017; Nakano et al., 2001; Oo et al., 2009; Sanchez-Martin et al., 2011; Slabbert, 1992; Tondi et al., 2009), a topic that was recently reviewed by Bacelo et al. (2016). Tannins obtained from Mimosa, quebracho and persimmon have been preferred, particularly for the separation of precious and rare elements (Qiu et al., 2017; Shen et al., 2016). 2 ACCEPTED MANUSCRIPT In this work, bark of Pinus pinaster (maritime pine) was used as source of tannins. Pinus pinaster represents roughly a third of all forestry area in Portugal (Fradinho et al., 2002). Its bark is particularly rich in condensed tannins (Pepino et al., 2001), mostly procyanidins (Navarrete et al., 2010, 2013), and is a common forest waste, appearing also as a residue from timber industry. The production of tannin-adsorbents, although involving some processing, which is optimized in the present work, requires much less energy than the conventional adsorbents currently used in commercial systems (activated carbon and oxidic adsorbents). Additionally, this kind of material is stable and can be produced in different particle sizes. The successful production of this new adsorbent can also contribute to meet other targets of the “2030 Agenda for Sustainable Development”, namely the decrease of waste generation through recycling and the efficient use of natural resources (UN, 2015). The tannin-adsorbents produced were evaluated on the uptake of antimony. To the best of the authors knowledge, the removal of this metalloid from water has never been assessed by this kind of material. Antimony is a natural occurring trace metalloid present in water mostly under trivalent and pentavalent oxidation states (Filella et al., 2002). Both forms are subjected to strong hydrolysis forming neutral or negatively-charged species; the formation of positively-charged antimony species is rare, only occurring in strongly acidic conditions (Tella and Pokrovski, 2009). The strong Sb affinity to OH– groups in solution limits its complexing ability with other inorganic and organic ligands, making difficult its uptake from water, in comparison to metals that simply exist as cations (Tella and Pokrovski, 2009), well researched before. Antimony trioxide is classified as possibly carcinogenic to humans (IARC, 1989) and various health adverse effects are related to antimony exposure by oral route (WHO, 2003). It is also known that Sb(III) is generally more toxic than Sb(V) (Stemmer, 1976). Anthropogenic sources of antimony such as mining, industry, coal burning, smelting, military training, pharmaceuticals and pesticides use (Herath et al., 2017; Ungureanu et al., 2015) are of particular concern. Mining industry may be the main anthropogenic source generating wastewaters with Sb levels that can reach 25-30 mg L–1 (Wang et al., 2013; Zhu et al., 2011). The development of techniques to handle antimony-contaminated waters prior to their discharge is then critically important. According to the World Health Organization (WHO, 2003), the 3 ACCEPTED MANUSCRIPT guideline value for antimony concentration in drinking water is 20 µg L–1. In many sites, especially the ones related to localized anthropogenic sources, antimony levels in surface and ground waters can reach or even exceed 1 mg L–1 (Ungureanu et al., 2015). Techniques to remove antimony from these sources, considering drinking water production or irrigation, are also necessary. Coagulation with ferric salts is commonly applied (Guo et al., 2009), but may be environmentally disadvantageous due to the considerable doses of chemicals used and the formation of toxic sludge. 2. Materials and methods 2.1. Preparation of tannin-adsorbent 2.1.1. Tannins extraction Pinus pinaster bark was collected, milled in a regular coffee mill and used in this work. Tannins were extracted following the alkaline batch procedure reported by Pepino et al. (2001) and by Sanchez-Martin et al. (2011). The aqueous extraction was carried out using 50 g of dried bark and the following conditions: NaOH 5% (w/w, in respect to the bark) and solid:liquid ratio of 1:6. Such conditions were shown to be optimal by Vázquez et al. (2001). The mixture was kept at 80 °C for 90 minutes, in a heating plaque/magnetic stirrer (Heidolph MR 3001). Subsequently, the solids were separated from the liquid by filtration (Whatman qualitative paper filter). The liquid was neutralized using 1 mol L–1 HCl (prepared from 37% w/w analytical grade commercial solution, Valente e Ribeiro, Lda) and evaporated using a heating plaque and a glass crystallizer. Finally, the humid precipitate was dried in an oven at 65 °C and the resultant material considered the tannin extract. Additionally, an alcoholic extraction in a Soxhlet apparatus was also carried out for comparison. Milled bark was subjected to 20 cycles with ethanol 70 % (v/v) and a solid:liquid ratio of 1:12. The extraction efficiency and the extract characteristics provided by both methods were then compared. Extraction yield (ηE) denotes the ratio between the amounts (in mass) of extract produced and dry bark initially used. Extract characteristics were assessed by the determination 4 ACCEPTED MANUSCRIPT of total polyphenolic content (TPC), Stiasny number (SN) and formaldehyde-condensable phenolic content (FCPC). TPC was determined using the Folin-Ciocalteu method (Singleton and Rossi, 1965) and adapted from Lazar et al. (2016). For each analysis, 1.0 mL of extract solution (25 mg of tannin extract dissolved in 50.0 mL of distilled water) was mixed with 0.50 mL of Folin-Ciocalteu reagent (Panreac), 2.0 mL of 100 g L–1 Na2CO3 (analytical grade, Merck) and 5.0 mL of distilled water. In order to avoid precipitate formation, sodium carbonate solution was added last (Cicco and Lattanzio, 2011). The mixture was kept in the dark at room temperature for 90 minutes. The absorbance of each solution was measured by an UV-vis spectrophotometer (VWR UV-6300PC Double Beam Spectrophotometer) at a wavelength of 765 nm. TPC values were calculated considering a predetermined calibration curve obtained using gallic acid standard solutions (15-100 mg L–1) and expressed as mg of gallic acid equivalents (GAE) per gram of tannin extract. The procedure carried out to estimate the amount of formaldehyde-condensable phenols that are reactive towards was adapted from the one described by Hoong et al. (2009). 50.0 mL of extract solution (250 mg of extract dissolved in 50.0 mL of distilled water) were added to 5.0 mL of formaldehyde (36 %, analytical grade, Labsolve) and 5.0 mL of 10 mol L–1 HCl and the mixture was kept at 80 °C for 30 minutes under reflux in a heating digester (Velp Scientifica DK6). The reaction mixture was filtrated under vacuum using membrane filters with 0.45 µm porosity. The precipitate was then dried in an oven up to constant weight. The quantification of the phenolic material that is condensable with formaldehyde was then performed in two ways (Pepino et al., 2001): (i) Stiasny number (SN), defined as the ratio between mass of the precipitate obtained and mass of the total dissolved extract, expressed here as g-precipitate per g-extract; (ii) FCPC which was calculated by the difference between the phenolic content of the solution, determined by the Folin-Ciocalteu method, before and after reaction with formaldehyde and expressed as mg of gallic acid equivalents per gram of tannin extract. Comparing alkaline and organic extraction methods described above, the former was selected to produce tannin extract to be used for the biosorbent preparation. 5 ACCEPTED MANUSCRIPT 2.1.2. Gelification Tannin gelification was achieved by reaction with formaldehyde in basic medium (Nakano et al., 2001; Sanchez-Martin et al., 2011). The tannin extract was dissolved in 0.25 mol L–1 NaOH solution at room temperature, which was followed by addition of formaldehyde (36 wt %, analytical grade, Labsolve) to function as a crosslinking reagent. After gelification at 80 °C for 8 hours, the precipitate was dried, milled and washed successively with 0.05 mol L–1 HNO3 solution and distilled water to remove unreacted substances. Finally, the obtained adsorbent was once more dried at 65 °C. The resultant product was considered the tannin-adsorbent. To optimize the gelification reaction, the effect of two experimental conditions was studied at different levels: (1) the volume of NaOH solution (4, 6, 8 and 12 mL per gram of tannin extract); and (2) formaldehyde amount (0.05, 0.10, 0.20, 0.40 and 0.80 mL per gram of tannin extract). The gelification efficiency (denoted as ηG) defined as the ratio between the mass of produced adsorbent (gelified product) and the mass of dissolved extract was considered the response to be maximized. The optimized conditions were used to produce the adsorbent to be tested on the uptake of antimony from aqueous solution. 2.1.3. Characterization of tannin-adsorbents Infrared (IR) spectra of tannin extract and tannin-adsorbent before and after saturation with antimony were obtained by FTIR (Fourier Transform Infrared Spectroscopy) in a Shimadzu FTIR (model IRAffinity) spectrometer, in a wavenumber range from 400 to 4000 cm-1, through 50 scans and with a resolution of 8.0 cm-1. Scanning electron microscopy (SEM) and Energy Dispersion Spectroscopy (EDS) were used to obtain images and chemical composition of tannin-materials surface. The SEM/EDS exam was performed at CEMUP-LMEV (Materials Centre of the University of Porto - Laboratory for Scanning Electron Microscopy and X-Ray Microanalysis) using a high resolution (Schottky) Environmental Scanning Electron Microscope with X-Ray Microanalysis and Electron Backscattered Diffraction analysis: Quanta 400 FEG ESEM/EDAX Genesis X4M. Samples were coated with a gold/palladium thin film, by sputtering, using the SPI Module Sputter Coater equipment. 6 ACCEPTED MANUSCRIPT The point of zero charge (PZC) is the pH value at which concentrations of negatively and positively charged groups at the surface are equal and then surface charge is null. It is an important parameter since it could allow to predict and understand the influence of pH on adsorption. The pH-drift method (Rivera-Utrilla et al., 2001) was used in this work to estimate the pHPZC value of the tannin-adsorbent at two different background electrolyte concentrations (0.01 mol L–1 or 0.1 mol L–1 NaCl solutions): 30.0 mL of solution at different pH (in the range 3-9, adjusted by small additions of NaOH or HCl solutions) were stirred with 15 mg of adsorbent for 24 h in closed Erlenmeyer flasks. Experiments were made in duplicate and blanks were also performed to discount the effect of CO2 from air. The final pH of each solution was measured and plotted against the initial pH (corrected by blank experiments). The pHPZC for each electrolyte concentration was estimated as the initial pH value for which no further pH change occurred. To evaluate the stability of the tannin-adsorbent and the possible organic matter leaching from the liquid phase, 50.0 mL of distilled water at different pH values were stirred with 25 mg of adsorbent for 24 h. Liquid phase was separated by filtration and dissolved organic carbon (DOC) was measured by catalytic oxidation at 680 ºC, in a Schimadzu TOC-L CSH analyzer. 2.2. Adsorption studies 2.2.1. Analytical methods Antimony concentrations in the liquid phase were measured by atomic absorption spectrometry (AAS) in air-acetylene flame (GBC 932 Plus) or by electrothermal atomization (equipment: GBC GF 3000, SenAA Dual spectrometer), at a wavelength of 217.6 nm, using background correction, 0.2 nm slit width and lamp currents of 10 mA or 8 mA, respectively. Flame was used to measure concentrations in the range 2-30 mg L–1 (detection limit: 0.8 mg L–1) and graphite furnace used to assess concentrations lower than 2.0 mg L–1 (detection limit: 3 µg L–1), after proper dilution. For both techniques, calibration curves were obtained and accepted for a determination coefficients (R2) higher than 0.995. Antimony speciation in solution was achieved by measuring Sb(III) and obtaining Sb(V) by difference from the total dissolved antimony. Sb(III) concentration was measured by liquid/liquid 7 ACCEPTED MANUSCRIPT extraction, using ammonium pyrrolidinedithiocarbamate (APDC) as complexing reagent and MIBK (methyl isobutyl ketone) as solvent (Smichowski et al., 1998). The procedure is based on the selective chelate formation of Sb(III) with APDC, followed by its quantitative extraction in MIBK and direct measure of antimony in the organic phase by AAS. 2.2.2. Equilibrium studies Firstly, the effect of adsorbent dosage on the uptake of Sb(III) by the tannin-adsorbent was assessed. Adsorption experiments were carried out by adding 50.0 mL of solution (initial adsorbate concentration: 20 mg L–1) at pH 6.0 to different adsorbent amounts (12.5 – 250 mg). Suspensions were stirred at 180 rpm in an orbital shaker for 24 h. Samples were then taken, filtrated using mixed cellulose ester membrane filters (0.45 µm porosity) and the liquid phase analyzed for total dissolved antimony. The amount of Sb adsorbed per gram of tannin-adsorbent (q) was calculated by Eq. 1, where Cin and C are initial and final Sb concentrations, respectively, and S/L, the solid:liquid ratio. q ( C in  C ) S/L (1) The effect of pH on the adsorbed amount of Sb(III) and Sb(V) was studied in the range 2-8, following the same procedure and using 0.50 g L–1 as adsorbent dosage. Initial pH of solutions was adjusted to different values, using diluted HCl or NaOH solutions. During the contact time pH was measured and readjusted if necessary to be approximately constant (maximum variations of 0.5). Competitive effect of arsenic, chloride, nitrate, sulfate and phosphate anions on the uptake of antimony by the tannin-adsorbent was studied individually, using initial adsorbate concentrations of 20 mg L-1, pH 6 (Sb(III)) or pH 2 (Sb(V)), S/L ratio of 0.50 g L–1 and the following levels of contaminants: arsenic 1 mg L-1 (As(III) for Sb(III) and As(V) for Sb(V)), chloride 50 mg L-1, nitrate 50 mg L-1, sulfate 200 mg-S L-1, and phosphate 20 mg-P L-1. Adsorption equilibrium isotherms were determined at 25 ºC, for different pH conditions (2, 4 and 6 for Sb(III), 2 and 4 for Sb(V)), using an adsorbent dosage of 0.50 g L–1 and different initial Sb concentrations (1-30 mg L–1). To check whether conversion between the two oxidation states 8 ACCEPTED MANUSCRIPT occurs, solutions from 20 mg L–1 initial concentration experiments were analyzed to assess Sb(III)/Sb(V) distribution. Adsorption isotherms were also obtained using a tailings water from a Portuguese mining site. Antimony levels in mine waters can vary widely. In this case, as the effluent presented a low level of Sb (<10 µg L–1), it was decided to spike it with antimony. Other properties measured in the effluent were: pH 4.2, phosphate <3 mg-P L–1, 378 mg-Ca L–1, 430 mg-Mg L–1, <0.5 mg-Fe L–1, 25 mg-Zn L–1, 0.6 mg-Cu L–1. The experiments were conducted in the same conditions as those adopted for the synthetic solutions, namely 25ºC, adsorbent dosage 0.50 g L–1, pH adjusted to 6 (antimonite) or 2 (antimonate) and Sb concentrations in the range 130 mg L-1. 2.2.3. Kinetic studies The effect of contact time on antimony adsorption by the tannin-adsorbent was studied in batch mode at constant temperature (25 ºC), adsorbent dosage (0.50 g L–1) and pH (6.0±0.3 for Sb(III) and 2.0±0.2 for Sb(V)). Experiments were conducted using different initial adsorbate concentrations (1, 5 and 20 mg L–1). A volume of 0.50 L of antimony solution was continuously stirred with tannin-adsorbent at 400 rpm. The pH of solutions was initially adjusted to the desired values, frequently checked and readjusted when necessary. Samples (5 mL) were regularly withdrawn, filtrated and analyzed for Sb concentration. 2.3. Desorption studies Desorption of antimony from the spent tannin-adsorbent was evaluated using four different eluents, HNO3 0.1 mol L–1, NaOH 0.1 and 0.5 mol L–1, NaCl 0.5 mol L–1, EDTA 0.1 mol L–1, a solid:liquid ratio of 2.5 g L-1 and 12 h stirring time. NaOH 0.5 mol L–1 was selected to be used in regeneration experiments, which were carried out through 2 adsorption/desorption cycles, using 20 mg L-1 Sb(V) solution at pH 2 to load the adsorbent. 9 ACCEPTED MANUSCRIPT 3. Results and Discussion 3.1. Optimization of tannin-adsorbent preparation 3.1.1. Tannin extraction Tannins were extracted from maritime pine bark by a batch alkaline method and by organic extraction (ethanol) in a Soxhlet apparatus. The extraction yields were calculated and the obtained extracts were analyzed and compared (Table 1). [Table 1] The yield (15±1 %) for the alkaline extraction was considerably higher than for the alcoholic one (3.2±0.4 %), as also reported by Pepino et al. (2001). The yield found in this work for the organic extraction is slightly lower than the value obtained from Pinus radiata bark (4.67±0.14 %) using ethanol/water (3:1, w/w) in a bench-scale reactor (120 ºC; S/L: 1:20; 120 min) (Bocalandro et al., 2012). Regarding aqueous extractions, the yield obtained depends on the alkaline solution concentration, temperature, time, solid:liquid ratio, degree of agitation and particle size of the bark. More aggressive conditions, such as higher NaOH concentration and temperature, seem to favor the extraction yield (Aspe and Fernandez, 2011). Literature reports extraction yields from Pinus pinaster varying roughly between 15 and 34 %, obtained using alkali concentrations between 1 and 5% w/w and temperatures in the range 70-100ºC (Chupin et al., 2013; Fradinho et al., 2002; Vázquez et al., 2001). The value here found is slightly below that range suggesting that further optimization of the bark sample extraction might be possible. Beyond extraction conditions, the age of the tree and its exposure to sunlight can also influence the amount of phenols contained in its bark (Alonso-Amelot et al., 2007) and then the extracted amount. Looking at the extract characteristics (Table 1), TPC in the alcoholic extract is higher than in the alkaline one. This indicates that ethanol has a higher specificity to phenolics than the alkaline solution and this is also in line with results reported by Pepino et al. (2001). Regarding SN and FCPC values, both methods generated extracts with similar properties (no statistical difference between the SN and FCPC values was observed). Thus, even though the alcoholic extract was richer in phenolic content, it was not in the formaldehyde-condensable material, which is a more 10 ACCEPTED MANUSCRIPT important property than the alkaline extract. Moreover, in the alkaline extract 93 % of total phenols were found to be condensable with formaldehyde, a higher percentage than that of the organic extract (84 %). Using extract properties and extraction yields obtained, the following quantities were calculated: TEP, denoting the total phenols (mg-GAE) extracted per gram of bark; ηSN defined here as the mass of gelified product (precipitate) obtained per gram of bark; and EFCP, representing the amount of formaldehyde condensable phenols (measured in mg-GAE) that were extracted per gram of bark. Alkaline extraction led to higher amounts of total extracted phenols (57±8 mg-GAE g–1 against the 16±3 mg-GAE g–1 obtained by the alcoholic method). In addition, alkaline extraction also generated higher ηSN and EFCP values (about 3 and 2.5 times, respectively), in comparison to the organic extraction. These two parameters, ηSN and EFCP, must be considered, as they reflect directly the amount of adsorbent that is possible to obtain per gram of Pinus pinaster bark. Table 2 presents a summary of the amount of total phenols, formaldehyde-condensable phenols, and gelified product that can be obtained per gram of bark of different Pinus species, calculated with values reported in literature. Clearly, the amount of total phenols extracted per gram of bark, in the present work (57 mg g–1) is within the range reported in literature. The ratio ηSN here recorded (100 mg of gelified product per gram of bark) is within the lowest ones of those gathered in Table 2. Variability in phenolic content between bark samples may explain this observation, as well as the possibility that the procedure here implemented needs further optimization. [Table 2] In conclusion, the alkaline method was found to be overall better suited for this work and the extraction procedure was carried out in a larger scale to have sufficient extract to be used in the subsequent gelification process. 3.1.2. Gelification The gelification procedure was optimized in terms of the amount of NaOH solution (solvent and catalyst), and formaldehyde. Fig. 1 presents the results of the influence of both variables (expressed per gram of tannin extract) on the gelification yield. 11 ACCEPTED MANUSCRIPT [Fig. 1] As it can be seen (Fig. 1a), an increase in the amount of formaldehyde added to the reaction generally increases the gelification efficiency. Such behavior was somewhat predictable since formaldehyde is the crosslinking reagent of the reaction. Using 4.0 mL of NaOH solution per g of extract, efficiency consistently increased until reaching a maximum for 0.80 mL of formaldehyde added per g of extract. On the other hand, using 6.0 mL of NaOH per g of extract the efficiency increased until a maximum of 0.40 mL g–1, evening out above that value. This indicates that the amount of formaldehyde (0.40 mL per gram of extract) is enough to react with total dissolved tannins. To achieve the best efficiency and, simultaneously, use the minimum amount of formaldehyde, it was decided that best conditions, within those assessed in this assay, are 6.0 mL of NaOH and 0.40 mL of formaldehyde per g of extract. Additionally, a higher amount of NaOH, given a fixed amount of formaldehyde, was tested in order to promote an increase in efficiency. It can be observed in Fig. 1b that, although efficiency may initially increase with the amount of NaOH, it decreases rapidly if the amount of NaOH is further increased, independently of the amount of formaldehyde used. This can be explained by the higher amount of solvent that keeps the tannins dissolved, decreasing the driving force to gelification. Then, it was considered that 6.0 mL of NaOH solution per g of extract was the optimal amount for a given amount of formaldehyde, upholding the conclusion of the formaldehyde assay. Tannin-adsorbents were then obtained by gelification in such optimal conditions and were used in the adsorption assays. 3.1.3. Characterization of tannin-adsorbents Fig. 2 display the IR spectra obtained for the tannin extract and the tannin-adsorbent. The broad bands observed at 3394 cm-1 (extract) and 3537 cm-1 (adsorbent) are characteristic of the -OH stretching vibration. These bands are distributed over a wide region, which is explained by the distribution of different degrees of polymerization in natural tannins and by the different positions of OH substituents (Ricci et al., 2015). The small peak observed at 2905 cm-1 for the tannin extract is due to aromatic C-H stretching. In the IR spectrum of tannin-adsorbent this band and probably other bands from the methylene bridges are overlapped by the -OH absorption. No distinctive signals related to carbonyl were observed for the tannin extract. In fact, in condensed tannins 12 ACCEPTED MANUSCRIPT absorption signals due to carbonyl groups are not expected, although weak absorptions may occur at 1750-1740 cm-1, possibly related to oxidation of OH groups on the flavanol molecules as a result of the extraction procedures (Ricci et al., 2015). Murugananthan et al. (2005) reported sharp and intense bands at 1620 and 1610 cm-1 for catechol and resorcinol, respectively. In the present case, as the tannins involve catechol (B-ring) and resorcinol (A-ring) moieties, it is possible that both absorptions are combined, resulting in the single band observed at 1618 cm-1 region, in extract IR. For the tannin-adsorbent, the broad absorption visible in the region 1800-1600 cm-1 may be a combination of the previous referred absorptions as well as weak signals related to the presence of carbonyl group, resultant from a residual oxidation of tannins. Tannins also generate absorption bands in the region 1400-1620 cm-1, attributed to the presence of aromatic rings (Ricci et al., 2015), although in the IR spectra here obtained and possibly due to the complex nature of the extract, which was not further purified, these peaks are not pronounced and cannot be individualized. [Fig. 2] Fig. 3 presents SEM images obtained for the tannin extract and adsorbent. The extract particles have in general a rough surface. On the other hand, the adsorbent particles present a smooth surface with deformities and do not appear to be porous. Sanchez-Martin et al. (2011) obtained SEM images for tannin gels and observed as well smooth-surface of non-porous particles. [Fig. 3] Adsorbents undergo surface charging depending on the solution pH. The solid surface is expected to be positively charged for pH values lower than pHPZC and negatively charged for pH values higher than it. Two points of zero charge were estimated by pH-drift method for the tanninadsorbent: 6.3±0.1 (using 0.01 mol L–1 NaCl solution) and 6.9±0.1 (0.1 mol L–1 NaCl). Different values were obtained in different electrolyte concentrations and a common intersection point (CIP) was not observed (data not shown). This means that the surface charge of the tanninadsorbent is not exclusively due to protonation and deprotonation of surface sites, but is also a function of the ionic strength and expectably of the solid-to-liquid ratio (Kosmulski, 2012). The electrolyte concentrations used here were believed to be representative of the ionic strength of 13 ACCEPTED MANUSCRIPT waters and wastewaters, and then the values found are useful to discuss antimony adsorption results. To assess the adsorbent stability in aqueous phase, DOC measurements were performed in distilled water after contact with the tannin-adsorbent. DOC values obtained were 3±2, 15±1, 16±1 mg-C L–1 for pH 2, 4 and 6 respectively. These very low DOC values (< 20 mg-C L–1) show that no significant secondary pollution was generated by the adsorbent. Additionally, color was not visually detected in solution. 3.2. Adsorption studies 3.2.1. Effect of adsorbent dosage Fig. 4 shows the effect of the solid:liquid ratio on the amount of Sb(III) adsorbed per gram of tannin-adsorbent. As expected, the removal percentage of antimonite increased with S/L, reaching 97 % at 5.0 g L–1. Adsorption capacity decreased with increasing adsorbent dosage; the highest adsorption capacities, 19±2 mg g–1 and 20.4±0.9 mg g–1, were respectively found for 0.25 and 0.50 g L–1, which are comparable values. The best use of the adsorptive capacity of the adsorbent was then obtained in these conditions. Subsequent studies were conducted using 0.50 g L–1, instead of 0.25 g L–1, to decrease the error related to the use of very low adsorbent dosages and to provide higher removal percentages. [Fig. 4] 3.2.2. Effect of pH and coexisting compounds The influence of pH on the adsorbed amount of antimonite and antimonate by the tanninadsorbent is depicted in Fig. 5(a). As it can be seen, pH significantly affects the uptake of both adsorbates. [Fig. 5] Regarding the uptake of antimonite, in the pH range 2-8, it gradually increased with the pH and optimum removals were found in the range pH 6-8 (20-22 mg g–1). The same behavior was observed on the uptake of Sb(III) by dead seaweeds, where biosorbed amounts increased with increasing pH (Ungureanu et al., 2017; Vijayaraghavan and Balasubramanian, 2011). According 14 ACCEPTED MANUSCRIPT to the antimony speciation diagram (Takeno, 2005; Tella and Pokrovski, 2012), in the pH range 1.3-12 Sb(III) is expected to predominate as a neutral complex (Sb(OH)3). For very low pH, Sb(OH)2+ co-occurs and then adsorption is harmed due to the electrostatic repulsion between this cationic compound and the positively charged groups on the adsorbent. At pH 6-7, close to the PZC, the repulsion is minimum and then maximum adsorbed amounts were obtained. In addition to the electrostatic interaction, adsorption mechanism probably involves complexation reactions. Tella and Pokrovski (2009) studied the stability of Sb(III) complexes formed with organic ligands in aqueous solution. They reported the establishment of Sb(III) complexes via Sb-O-C bonds with ligands having two or more adjacent carboxyl and/or hydroxyl functional groups (Fig. 6(b)). In the case of di-hydroxy-phenol (catechol), which serves for comparison purposes with the present work, two types of complexes are expected to be formed with Sb(OH)3 respectively at acidic conditions (where catechol is protonated) and at basic conditions (monodeprotonated catechol). This observation suggests an adsorption mechanism based on Sb(III) complexation with the dihydroxy-phenol surface groups, present in B-rings of tannin-materials (Fig. 6) and explains why Sb(III) is reasonably adsorbed in the whole pH range under study. The infrared spectrum of Sb(III)-loaded adsorbent (Fig.6(c)) seems to be in line with the proposed mechanism as a decrease in the intensity of the –OH stretching vibration bands and a change in the peak frequency were observed. [Fig. 6] Concerning the antimonate (Fig. 5a), optimum removal was found at pH 2 (16.7±0.9 mg g–1), although considerable values were also recorded at pH 3 and 4. Noticeably, the increase in pH beyond 4 suppressed Sb(V) uptake, with negligible removals found in the pH range 5-8. According to Eh-pH diagrams (Takeno, 2005; Tella and Pokrovski, 2012), Sb(V) only occurs as the neutral complex Sb(OH)5 in very strong acidic conditions (pH below 2.6); in mild acidic, neutral and alkaline solution, Sb(V) is expected to occur as the oxyanion Sb(OH)6–. The adsorption mechanism probably involves electrostatic attraction between positive adsorbent surface groups (pH<pHZPC) and this oxyanion, and also involves the complexation of polyol surface groups on the tannin-adsorbent and Sb(OH)6– (Fig. 6(c)). The formation of these type of 15 ACCEPTED MANUSCRIPT complexes in aqueous solution and in the pH range 2-4 was also previously demonstrated (Tella and Pokrovski, 2012) and this is perfectly in line with the pH effect observed here, including the uptake inhibition for pH higher than 4. The main differences between the infrared spectrum of the tannin-adsorbent before and after Sb(V) uptake (Fig. 2) are also observed in the –OH stretching region. Other authors also reported acidic conditions as the most suitable for Sb(V) uptake: for commercial activated alumina, the optimum pH range was identified as 2.8-4.3 and a dramatically decrease was also reported for higher pH values (Xu et al., 2001); optimum pH of 2-3 were observed for untreated and modified aerobic granules (Wang et al., 2014) and for freshwater cyanobacteria Microcystis biomass (Sun et al., 2011). The competitive effect of As(III) or As(V), Cl-, NO3-, SO42- and PO43- on the uptake of Sb(III) and Sb(V) by the tannin-adsorbent was studied and results depicted in Fig. 5(b). As it can be seen, As(III) and the studied anions did not exert significant influence on the uptake of antimonite by the tannin-adsorbent. The observed differences between the adsorbed amounts in the control experiment (absence of possible interfering components) and in the presence of the referred competitors was also statistically insignificant for Sb(V), although a minor influence seems to exist due to sulfate and phosphate. The results here obtained indicate that under typical conditions the tannin-adsorbent presents a good selectivity towards Sb(III) and Sb(V). Literature also reports negligible or minor effects of the studied anions on the uptake of Sb(III) by seaweeds and ferric hydroxide (He et al., 2015; Ungureanu et al., 2017). A little effect (≈16% decrease) was reported for the performance of a green seaweed on the uptake of Sb(V) due to sulfate and phosphate (Ungureanu et al., 2016). 3.2.3. Equilibrium isotherms Equilibrium isotherms for the adsorption of Sb(III) and Sb(V) on the tannin-adsorbent are presented in Fig. 7. The adsorbed amounts (qe) in equilibrium with the solution concentration (Ce) were calculated by Eq. 1. [Fig. 7] As previously explained, pH 6-8 is the optimum pH range to bind Sb(III) to the tannin-adsorbent. The isotherm was measured at pH 6 considering that most of the Sb-polluted waters present an 16 ACCEPTED MANUSCRIPT acidic or slightly acidic pH. Fig. 7(a) corroborates previous observation and shows that a significant effect of pH is observed in almost the entire concentration range. In the selected operating conditions (pH 6, adsorbent dosage 0.50 g L–1), antimonite removal efficiencies varied between 38±2 % (Cin=30 mg L–1) and 90±1 % (Cin=1 mg L–1). These values clearly illustrate the considerable ability to uptake this toxic metalloid from water in typical soluble concentrations. Regarding Sb(V) (Fig. 7b), considerable adsorbed amounts were also recorded, reaching 21.0±0.9 mg g–1 and 16.0±0.9 mg g–1 at pH 2 and 4, respectively. A slightly better performance of the adsorbent was observed at pH 2 with efficiencies between 33±2 % (Cin=30 mg L–1) and 69±4 % (Cin=1 mg L–1). Final solutions from adsorption experiments with 20 mg L–1 initial Sb(III) and Sb(V) concentrations were subjected to speciation analysis. Final trivalent and pentavalent concentrations were found to be the same as total dissolved Sb, which means that no significant oxidation or reduction reactions involving antimony occurred in solution. Ungureanu et al. (2017) also did not observe Sb(III)/Sb(V) conversions in solution during antimony adsorption by seaweeds, but Wu et al. (2012) reported up to 6.9% of Sb(V) in final solutions resultant from Sb(III) adsorption experiments using Microcystis biomass. Fig. 7 also presents equilibrium adsorption isotherms measured using a mine tailings water (ME) as aqueous matrix. In the case of antimonite (Fig. 7a), there seems to be a small matrix effect for low adsorbate concentrations, which ceases for Ceq higher than ≈7 mg L-1, as the adsorbed amounts are similar to the ones obtained in the distilled water matrix. These results are in line with the previous observations (section 3.2.2), which show insignificant effects of As(III) and various anions on antimonite uptake from 20 mg L-1 solutions. Regarding antimonate (Fig. 7b), a moderate matrix effect was observed in the mining effluent, with a decrease of up to 38% in the ability of the tannin-adsorbent to uptake Sb(V). Langmuir (1918) and Freundlich (1906) equilibrium models were fitted to the equilibrium experimental data by non-linear regression. In the Langmuir model (Eq. 2), Qm symbolizes the maximum adsorption capacity of the adsorbent and KL the Langmuir constant. In the Freundlich 17 ACCEPTED MANUSCRIPT model equation (Eq. 3), KF is a constant related to the adsorption capacity and n a constant related to the intensity of adsorption. 𝑞𝑒𝑞 = 𝑄𝑚 ∙ 𝐾𝐿 ∙ 𝐶𝑒𝑞 1 + 𝐾𝐿𝐶𝑒𝑞 1/𝑛 𝑞𝑒𝑞 = 𝐾𝐹 ∙ 𝐶𝑒𝑞 (2) (3) Langmuir and Freundlich parameters are presented in Table 3 and model curves plotted in Fig. 7. As it can be seen, both models described quite well the experimental data, with most of the correlation coefficients (R) close to 1. Except for the Sb(V) isotherm measured at pH 2, Langmuir fitting provided lower regression standard errors (SE). Limited adsorption capacities were obtained for Sb(III) at pH 2 and 4 but at pH 6 a very considerable value was achieved (24±3 mg g–1). Similar values were also obtained for Sb(V) at pH 2 and 4 (27±7 and 25±10 mg g–1). Observed results show the ability of tannin-adsorbent to uptake both forms of antimony at selected pH values, according to the antimony predominant oxidation state in solution. From the thermodynamic point of view, the pentavalent form is the most stable in oxygenated waters and the trivalent one in reducing/middle reducing conditions. However, it has been reported the occurrence of both oxidation forms in thermodynamically unforeseen situations, which has been attributed to slow kinetics of conversion or biological activity (Filella et al., 2002), and reinforces the importance to study the uptake of both forms. [Table 3] Antimony needs to be removed from heavy-contaminated solutions, e.g. mine drainage and mine flotation wastewaters and/or from much more diluted solutions, such as surface or groundwaters, where levels are typically lower than 1 mg L–1. In the latter case, the affinity of the adsorbent to the target adsorbates is also an important parameter. It is a measure of the adsorbent ability to uptake contaminants from very dilute solutions and can be calculated from the slope of the isotherm when equilibrium concentration tends to zero, i.e. the product of KL and Qm from the Langmuir model. The following conclusions were obtained from calculated values for tanninadsorbent: (i) a much higher affinity for Sb(III) than for Sb(V); (ii) similar affinities for Sb(III) at 18 ACCEPTED MANUSCRIPT pH 4 and 6; and (iii) similar affinities for Sb(V) at pH 2 and 4. Last observations indicate that pH does not significantly affect the adsorption from low-Sb levels waters. Tannin-adsorbents obtained by gelification of tannins usually present a good ability to uptake heavy metal ions, such as Cr(III), Zn(II), Pb(II), Cu(II), with maximum adsorption capacities reported in the range 0.42-1.3 mmol/g (Bacelo et al., 2016; Huang et al., 2010; Sengil and Ozacar, 2009; Yurtsever and Sengil, 2009). In this work, the performance of the tannin-adsorbent for antimony is understandably lower (Qm values corresponding to ≈ 0.2 mmol/g), considering the completely different chemical behavior of Sb(III) and Sb(V), in comparison to heavy metals that simply occur as cations. The maximum adsorption capacities determined in this work were compared to values reported in literature for conventional and non-conventional adsorbents (Table 4). Higher or comparable Qm values were obtained in this work, comparing with those reported in literature for ferric hydroxide, biochars and biosorbents. Only bimetal oxides presented a much better performance, although with two possible disadvantages: possible metal leaching to solution and high cost. There are not many studies on the use of activated carbon as adsorbent for antimony, but it is known that an additional treatment (usually modification by FeCl3) is required to make it effective (Ungureanu et al., 2015). Even so, the performance of the tannin-adsorbent seems to be slightly higher, considering that for a modified iron-activated carbon adsorbed amounts of ≈ 3 mg g–1 were obtained under Sb(III) equilibrium concentration of approximately 1 mg L–1 (pH 7), while tannin-adsorbent uptake was ≈ 8 mg g–1 (pH 6). [Table 4] 3.2.4. Adsorption kinetics Contact time influence on Sb(III) and Sb(V) uptake by tannin-adsorbent was studied for different initial adsorbate concentrations, at constant pH, temperature and solid:liquid ratio. Results are presented in Fig. 8. [Fig. 8] Pseudo-first (Lagergren, 1898) (Eq. 4) and pseudo-second order (Blanchard et al., 1984; Ho, 1995) (Eq. 5) models, commonly used to describe kinetic adsorption data, were fitted to the 19 ACCEPTED MANUSCRIPT experimental data by non-linear regression. In Eq. 4 and 5, q and qeq denote adsorbed amounts per unit mass of adsorbent, at time t and at equilibrium, respectively; k1 and k2 are kinetic constants. q  qeq 1  exp k1t  q  qeq k 2 qeq t (4) (5) 1  k 2 qeq t The kinetic parameters obtained are presented in Table 5 and the modelled curves depicted in Fig. 8. Both models describe adsorption dynamics considerably well, although pseudo-second order regressions provided slightly lower standard errors and the pseudo-first order predicted qeq values closer to the experimental ones. It is evident that the adsorption of Sb(III) is faster than Sb(V), suggesting that different kinds of complexes are formed for each oxidation state. Complexation of Sb with catechol is established with Sb/ligand stoichiometries of 1:1 and 1:2 reactions for Sb(III) and 1:3 for Sb(V) (Tella and Pokrovski, 2009, 2012). The conjugation with more hydroxyl groups can explain the lower Sb(V) uptake kinetics comparing with Sb(III). Initial adsorption rates (dq/dt for t=0) were calculated using pseudo-first order parameters. For initial Sb concentrations of 1, 5 and 20 mg L–1, the values respectively obtained were 0.060±0.005, 0.17±0.02 and 0.16±0.02 mg g–1 min–1 for Sb(III) and 0.011±0.005, 0.03±0.01 and 0.05±0.02 mg g–1 min–1 for Sb(V). For both adsorbates, initial adsorption rates increased when the initial Sb concentration was changed from 1 to 5 mg L–1, due to the increase of the driving force for mass transfer but remained almost constant between initial concentrations of 5 and 20 mg L–1. For Sb(III), the time required to reach equilibrium varied between 5 h (Cin=1 mg L–1) and 10 h (Cin=20 mg L–1), although after 5 h of contact time almost 90% of the maximum adsorbed amount was already attained in the experiment conducted with Cin=20 mg L–1. The times required to reach equilibrium in Sb(V) uptake were estimated to be longer: 11 h for Cin=1 mg L–1, and 20 h for Cin=20 mg L–1. Adsorbent samples saturated with Sb(III) and Sb(V), obtained at the end of kinetic assays using initial concentrations of 5 and 20 mg L–1, were analyzed by SEM. The antimony did not cause any visible alteration of the adsorbent (data not illustrated), although EDS analysis 20 ACCEPTED MANUSCRIPT confirmed antimony presence in the solids, suggesting a homogeneous coverage by the adsorbates. [Table 5] 3.3. Desorption In order to study the possible antimony recovery, the regeneration of the adsorbent and an additional understanding of the mechanism involved in Sb uptake by tannin-adsorbents, some desorption experiments were done. The desorption of antimony from Sb(III) and Sb(V)-saturated adsorbents was studied using different eluents and the obtained efficiencies are presented in Table 6. Low desorbed amounts were observed in saline (<8 %) and acid solutions (<13 %), suggesting the involvement of strong and stable chemical bonds between antimony and the tannin -adsorbent. The regeneration of the exhausted material was only found to be feasible using alkaline solutions, with best results observed for NaOH 0.5 mol L-1 (desorption efficiencies of 69-75 %). Limited Sb desorption efficiencies were also reported in literature for different biosorbents. Ungureanu et al. (2016) indicated desorption efficiencies in the range 12-23 % from Sb(IIII)saturated seaweeds and using different saline (0.5 mol L-1), alkaline and acid solutions (0.5 mol L-1). Wu et al. (2012) obtained values (≈63 %) closer to the ones here found, from exhausted Microcystis biomass, using HCl 4 mol L-1 as eluent. In order to assess the regeneration capacity of the tannin-adsorbent, two adsorption/desorption cycles were performed, using NaOH 0.5 mol L-1 as eluent. The second adsorption step took place with no loss on the adsorption capacity (16±2 mg g-1 and 18±2 mg g-1 were the Sb(V) adsorbed amounts recorded in the first and second adsorption steps, respectively), which show the ability of the alkaline solution to regenerate the adsorbent. However, a strong decrease on the desorption efficiency was observed in the second desorption step, reaching only 28%. In addition, some color leaching was observed in these conditions. 21 ACCEPTED MANUSCRIPT 4. Conclusions Tannins were extracted from Pinus pinaster bark by two different methods: an alkaline extraction, in batch mode, and a Soxhlet extraction using ethanol. The extraction by sodium hydroxide solution, despite providing extracts poorer in phenolic content, led to formaldehyde-condensable phenolic content and Stiasny numbers comparable to the ones obtained by organic extraction. On the other hand, thanks to its greater extraction yield (15±1 vs 3.2±0.5 %), a higher amount of biosorbent can be obtained per gram of bark used (100±10 mg g-1) through the alkaline procedure. Thus, extraction in aqueous alkaline solution was found to be better suited to produce tanninadsorbents from pine bark. Extracted tannins were converted into biosorbents (insoluble matrices) by gelification. The amounts of solvent/catalyst (NaOH) and reactant (formaldehyde) were optimized to maximize gelification yield while minimizing chemicals use. Optimum conditions (gelification efficiency: 71±4 %) were found when extracts were dissolved in 6.0 mL of NaOH 0.25 mol L–1 per g of extract and when 0.40 mL of formaldehyde per g of extract was added. The biosorbents produced in optimized conditions were studied for the uptake of antimony from solution. The results showed that Sb(III) uptake occurs extensively in the entire pH range studied, with optimum removals found at pH close to the neutrality (pH 6-8). The uptake of Sb(V) was only efficient from strong acidic waters (pH 2-4). The adsorption of Sb(III) and Sb(V) is not significantly affected by the presence of As, chloride, nitrate, sulfate or phosphate, at typical levels. Adsorption kinetics was well described by both pseudo-first and pseudo-second order models. The time required to reach equilibrium depends on the adsorbate (higher time for Sb(V)) and its initial concentration and varied between 5 and 20 h. Equilibrium results showed that tannin-adsorbents present a strong affinity to antimony species, especially to the trivalent species, which is advantageous since this is the most toxic Sb form. Maximum adsorption capacities for Sb(III) and Sb(V) were obtained by Langmuir model as 24±3 mg g–1 (pH 6) and 27±7 mg g–1 (pH 2), respectively, in synthetic solutions. Similar values (30±5 and 17±4 mg g–1, respectively) were obtained when a tailings water from a mining site was used as aqueous matrix. Tannin-adsorbents here produced are stable in solution, which does not usually happen with many of the low-cost materials reported in literature; can be produced in varied particle sizes, which is important for 22 ACCEPTED MANUSCRIPT application at full-scale; and are believed to be economically competitive with conventional adsorbents, due to the observed efficiency and the relatively simple production. Bark from Pinus pinaster was then found to be an interesting precursor for this effective adsorbent and this can be used as an alternative way to manage and give value to such forest waste whose chemical richness is commonly ignored. Acknowledgements This work is a result of project “AIProcMat@N2020 - Advanced Industrial Processes and Materials for a Sustainable Northern Region of Portugal 2020”, with the reference NORTE-010145-FEDER-000006, supported by Norte Portugal Regional Operational Programme (NORTE 2020), under the Portugal 2020 Partnership Agreement, through the European Regional Development Fund (ERDF) and of Project POCI-01-0145-FEDER-006984 – Associate Laboratory LSRE-LCM funded by ERDF through COMPETE2020 - Programa Operacional Competitividade e Internacionalização (POCI) – and by national funds through FCT - Fundação para a Ciência e a Tecnologia. H. Bacelo acknowledges his PhD scholarship funded by FCT (PD/BD/135062/2017). 23 ACCEPTED MANUSCRIPT References Alonso-Amelot, M.E., Oliveros-Bastidas, A., Calcagno-Pisarelli, M.P., 2007. Phenolics and condensed tannins of high altitude Pteridium arachnoideum in relation to sunlight exposure, elevation, and rain regime. Biochemical Systematics and Ecology 35, 1-10. Alvares Rodrigues, L., Koibuchi Sakane, K., Alves Nunes Simonetti, E., Patrocínio Thim, G., 2015. Cr total removal in aqueous solution by PHENOTAN AP based tannin gel (TFC). Journal of Environmental Chemical Engineering 3, 725-733. Aspe, E., Fernandez, K., 2011. The effect of different extraction techniques on extraction yield, total phenolic, and anti-radical capacity of extracts from Pinus radiata Bark. Industrial Crops and Products 34, 838-844. Bacelo, H.A.M., Santos, S.C.R., Botelho, C.M.S., 2016. Tannin-based biosorbents for environmental applications - A review. Chemical Engineering Journal 303, 575-587. Beltran-Heredia, J., Palo, P., Sanchez-Martin, J., Dominguez, J.R., Gonzalez, T., 2012. Natural Adsorbents Derived from Tannin Extracts for Pharmaceutical Removal in Water. Industrial & Engineering Chemistry Research 51, 50-57. Blanchard, G., Maunaye, M., Martin, G., 1984. Removal of Heavy-Metals from Waters by Means of Natural Zeolites. Water Research 18, 1501-1507. Bocalandro, C., Sanhueza, V., Gomez-Caravaca, A.M., Gonzalez-Alvarez, J., Fernandez, K., Roeckel, M., Rodriguez-Estrada, M.T., 2012. Comparison of the composition of Pinus radiata bark extracts obtained at bench- and pilot-scales. Industrial Crops and Products 38, 21-26. Chupin, L., Motillon, C., Charrier-El Bouhtoury, F., Pizzi, A., Charrier, B., 2013. Characterisation of maritime pine (Pinus pinaster) bark tannins extracted under different conditions by spectroscopic methods, FTIR and HPLC. Industrial Crops and Products 49, 897-903. Cicco, N., Lattanzio, V., 2011. The Influence of Initial Carbonate Concentration on the FolinCiocalteu Micro-Method for the Determination of Phenolics with Low Concentration in the Presence of Methanol: A Comparative Study of Real-Time Monitored Reactions. American Journal of Analytical Chemistry 2, 840-848. Cui, X.Q., Ni, Q.J., Lin, Q., Khan, K.Y., Li, T.Q., Khan, M.B., He, Z.L., Yang, X.E., 2017. Simultaneous sorption and catalytic oxidation of trivalent antimony by Canna indica derived biochars. Environmental Pollution 229, 394-402. Filella, M., Belzile, N., Chen, Y.W., 2002. Antimony in the environment: a review focused on natural waters I. Occurrence. Earth-Science Reviews 57, 125-176. Fomina, M., Gadd, G.M., 2014. Biosorption: current perspectives on concept, definition and application. Bioresource Technology 160, 3-14. Fradinho, D.M., Neto, C.P., Evtuguin, D., Jorge, F.C., Irle, M.A., Gil, M.H., de Jesus, J.P., 2002. Chemical characterisation of bark and of alkaline bark extracts from maritime pine grown in Portugal. Industrial Crops and Products 16, 23-32. Freundlich, H.M.F., 1906. Over the adsorption in solution. Journal of Physical Chemistry 57, 385471. Garcia, D.E., Glasser, W.G., Pizzi, A., Lacoste, C., Laborie, M.P., 2014. Polyphenolic resins prepared with maritime pine bark tannin and bulky-aldehydes. Industrial Crops and Products 62, 84-93. Guo, X.J., Wu, Z.J., He, M.C., 2009. Removal of antimony(V) and antimony(III) from drinking water by coagulation-flocculation-sedimentation (CFS). Water Research 43, 4327-4335. He, Z., Liu, R.P., Liu, H.J., Qu, J.H., 2015. Adsorption of Sb(III) and Sb(V) on Freshly Prepared Ferric Hydroxide (FeOxHy). Environmental Engineering Science 32, 95-102. 24 ACCEPTED MANUSCRIPT Herath, I., Vithanage, M., Bundschuh, J., 2017. Antimony as a global dilemma: Geochemistry, mobility, fate and transport. Environmental Pollution 223, 545-559. Ho, Y.S., 1995. Adsorption of heavy metals from waste streams by peat. Ph.D. Thesis. . The University of Birmingham, Birmingham, U.K. Hoong, Y.B., Paridah, M.T., Luqman, C.A., Koh, M.P., Loh, Y.F., 2009. Fortification of sulfited tannin from the bark of Acacia mangium with phenol-formaldehyde for use as plywood adhesive. Industrial Crops and Products 30, 416-421. Huang, X., Liao, X.P., Shi, B., 2010. Tannin-immobilized mesoporous silica bead (BT-SiO2) as an effective adsorbent of Cr(III) in aqueous solutions. Journal of Hazardous Materials 173, 3339. IARC, 1989. Antimony Trioxide and Antimony Trisulfide. IARC Monographs, Volume 47. International Agency for Research on Cancer. Iqbal, M., Saeed, A., Edyvean, R.G.J., 2013. Bioremoval of antimony(III) from contaminated water using several plant wastes: Optimization of batch and dynamic flow conditions for sorption by green bean husk (Vigna radiata). Chemical Engineering Journal 225, 192-201. Kosmulski, M., 2012. IEP as a parameter characterizing the pH-dependent surface charging of materials other than metal oxides. Advances in Colloid and Interface Science 171, 77-86. Lagergren, S.Y., 1898. Zur theorie der sogenannten adsorption gelöster stoffe, Kungliga Svenska Vetenskapsakademiens. Handlingar 24, 1-39. Langmuir, I., 1918. The adsorption of gases on plane surfaces of glass, mica and platinum. Journal of the American Chemical Society 40, 1361-1403. Lazar, L., Talmaciu, A.I., Volf, I., Popa, V.I., 2016. Kinetic modeling of the ultrasound-assisted extraction of polyphenols from Picea abies bark. Ultrasonics Sonochemistry 32, 191-197. Li, X.H., Dou, X.M., Li, J.Q., 2012. Antimony(V) removal from water by iron-zirconium bimetal oxide: Performance and mechanism. Journal of Environmental Sciences 24, 1197-1203. Luzardo, F.H.M., Velasco, F.G., Correia, I.K.S., Silva, P.M.S., Salay, L.C., 2017. Removal of lead ions from water using a resin of mimosa tannin and carbon nanotubes. Environmental Technology & Innovation 7, 219-228. Murugananthan, M., Raju, G.B., Prabhakar, S., 2005. Removal of tannins and polyhydroxy phenols by electro-chemical techniques. Journal of Chemical Technology and Biotechnology 80, 1188-1197. Nakano, Y., Takeshita, K., Tsutsumi, T., 2001. Adsorption mechanism of hexavalent chromium by redox within condensed-tannin gel. Water Research 35, 496-500. Navarrete, P., Pizzi, A., Pasch, H., Rode, K., Delmotte, L., 2010. MALDI-TOF and 13C NMR characterization of maritime pine industrial tannin extract. Industrial Crops and Products 32, 105110. Navarrete, P., Pizzi, A., Pasch, H., Rode, K., Delmotte, L., 2013. Characterization of two maritime pine tannins as wood adhesives. Journal of Adhesion Science and Technology 27, 2462-2479. Nishikawa, E., da Silva, M.G.C., Vieira, M.G.A., 2018. Cadmium biosorption by alginate extraction waste and process overview in Life Cycle Assessment context. Journal of Cleaner Production 178, 166-175. Oo, C.W., Kassim, M.J., Pizzi, A., 2009. Characterization and performance of Rhizophora apiculata mangrove polyflavonoid tannins in the adsorption of copper (II) and lead (II). Industrial Crops and Products 30, 152-161. Pepino, L., Brito, P., Jorge, F.C., da Costa, R.P., Gil, M.H., Portugal, A., 2001. Comparison of Quantification Methods for the Condensed Tannin Content of Extracts of Pinus Pinaster Bark, in: 25 ACCEPTED MANUSCRIPT Chiellini, E., Gil, H., Braunegg, G., Buchert, J., Gatenholm, P., Van Der Zee, M. (Eds.), Biorelated Polymers: Sustainable Polymer Science and Technology. Springer Boston, US. Qi, Z.L., Joshi, T.P., Liu, R.P., Liu, H.J., Qu, J.H., 2017. Synthesis of Ce(III)-doped Fe3O4 magnetic particles for efficient removal of antimony from aqueous solution. Journal of Hazardous Materials 329, 193-204. Qiu, X., Shen, Y., Yang, R., Zhang, H., Zhao, S., 2017. Adsorption of RE3+ from aqueous solutions by bayberry tannin immobilized on chitosan. Environmental technology, 1-8. https://doi.org/10.1080/09593330.09592017.01384072. Rao, R.A.K., Khatoon, A., 2017. Aluminate treated Casuarina equisetifolia leaves as potential adsorbent for sequestering Cu(II), Pb(II) and Ni(II) from aqueous solution. Journal of Cleaner Production 165, 1280-1295. Ricci, A., Olejar, K.J., Parpinello, G.P., Kilmartin, P.A., Versari, A., 2015. Application of Fourier Transform Infrared (FTIR) Spectroscopy in the Characterization of Tannins. Applied Spectroscopy Reviews 50, 407-442. Rivera-Utrilla, J., Bautista-Toledo, I., Ferro-Garcia, M.A., Moreno-Castilla, C., 2001. Activated carbon surface modifications by adsorption of bacteria and their effect on aqueous lead adsorption. Journal of Chemical Technology and Biotechnology 76, 1209-1215. Sanchez-Martin, J., Beltran-Heredia, J., Gibello-Perez, P., 2011. Adsorbent biopolymers from tannin extracts for water treatment. Chemical Engineering Journal 168, 1241-1247. Sengil, I.A., Ozacar, M., 2009. Competitive biosorption of Pb2+ Cu2+ and Zn2+ ions from aqueous solutions onto valonia tannin resin. Journal of Hazardous Materials 166, 1488-1494. Shen, Y.Y., Yang, R.L., Liao, Y., Ma, J., Mao, H., Zhao, S.L., 2016. Tannin modified aminated silica as effective absorbents for removal of light rare earth ions in aqueous solution. Desalination and Water Treatment 57, 18529-18536. Singleton, V.L., Rossi, J.A., 1965. Colorimetry of total phenolics with phosphomolybdicphosphotungstic acid reagents. American journal of Enology and Viticulture 16, 144-158. Slabbert, N., 1992. Complexation of condensed tannins with metal ions, in: Hemingway, W.R., Laks, P.E. (Eds.), Plant Polyphenols: Biogenesis, Chemical Properties, and Significance. Plenum Press, New York, pp. 421-435. Smichowski, P., Madrid, Y., Camara, C., 1998. Analytical methods for antimony speciation in waters at trace and ultratrace levels. A review. Fresenius Journal of Analytical Chemistry 360, 623-629. Stemmer, K.L., 1976. Pharmacology and toxicology of heavy-metals e antimony. Pharmacology and Therapeutics Part A 1, 157-160. Sun, F.H., Wu, F.C., Liao, H.Q., Xing, B.S., 2011. Biosorption of antimony(V) by freshwater cyanobacteria Microcystis biomass: Chemical modification and biosorption mechanisms. Chemical Engineering Journal 171, 1082-1090. Takeno, N., 2005. Atlas of Eh-pH diagrams. Geological Survey of Japan, Open File, Report No.419. , in: Environments, N.I.o.A.I.S.a.T.R.C.f.D.G. (Ed.), Tsukuba, Japan. Tella, M., Pokrovski, G.S., 2009. Antimony(III) complexing with O-bearing organic ligands in aqueous solution: An X-ray absorption fine structure spectroscopy and solubility study. Geochimica Et Cosmochimica Acta 73, 268-290. Tella, M., Pokrovski, G.S., 2012. Stability and structure of pentavalent antimony complexes with aqueous organic ligands. Chemical Geology 292, 57-68. Tondi, G., Oo, C.W., Pizzi, A., Trosa, A., Thevenon, M.F., 2009. Metal adsorption of tannin based rigid foams. Industrial Crops and Products 29, 336-340. 26 ACCEPTED MANUSCRIPT UN, 2015. Transforming Our World: The 2030 Agenda for Sustainable Development (A/RES/70/1). UN General Assembly. Ungureanu, G., Filote, C., Santos, S.C.R., Boaventura, R.A.R., Volf, I., Botelho, C.M.S., 2016. Antimony oxyanions uptake by green marine macroalgae. Journal of Environmental Chemical Engineering 4, 3441-3450. Ungureanu, G., Santos, S., Boaventura, R., Botelho, C., 2015. Arsenic and antimony in water and wastewater: Overview of removal techniques with special reference to latest advances in adsorption. Journal of Environmental Management 151, 326-342. Ungureanu, G., Santos, S.C.R., Volf, I., Boaventura, R.A.R., Botelho, C.M.S., 2017. Biosorption of antimony oxyanions by brown seaweeds: Batch and column studies. Journal of Environmental Chemical Engineering 5, 3463-3471. Vázquez, G., Gonzalez-Alvarez, J., Freire, S., Lopez-Suevos, F., Antorrena, G., 2001. Characteristics of Pinus pinaster bark extracts obtained under various extraction conditions. Holz Als Roh-Und Werkstoff 59, 451-456. Vijayaraghavan, K., Balasubramanian, R., 2011. Antimonite Removal Using Marine Algal Species. Industrial & Engineering Chemistry Research 50, 9864-9869. Wang, H.W., Chen, F.L., Mu, S.Y., Zhang, D.Y., Pan, X.L., Lee, D.J., Chang, J.S., 2013. Removal of antimony (Sb(V)) from Sb mine drainage: Biological sulfate reduction and sulfide oxidation-precipitation. Bioresource Technology 146, 799-802. Wang, L., Wan, C., Lee, D.J., Liu, X., Zhang, Y., Chen, X.F., Tay, J.H., 2014. Biosorption of antimony(V) onto Fe(III)-treated aerobic granules. Bioresource Technology 158, 351-354. WHO, 2003. Antimony in Drinking-water. Background document for development of WHO Guidelines for Drinking-water Quality (WHO/SDE/WSH/03.04/74). World Health Organization. Wu, F.C., Sun, F.H., Wu, S., Yan, Y.B., Xing, B.S., 2012. Removal of antimony(III) from aqueous solution by freshwater cyanobacteria Microcystis biomass. Chemical Engineering Journal 183, 172-179. Xi, J.H., He, M.C., Wang, K.P., Zhang, G.Z., 2013. Adsorption of antimony(III) on goethite in the presence of competitive anions. Journal of Geochemical Exploration 132, 201-208. Xu, Y.H., Ohki, A., Maeda, S., 2001. Adsorption and removal of antimony from aqueous solution by an activated Alumina. Toxicological & Environmental Chemistry 80, 133-144. Yargıç, A.Ş., Yarbay Şahin, R.Z., Özbay, N., Önal, E., 2015. Assessment of toxic copper(II) biosorption from aqueous solution by chemically-treated tomato waste. Journal of Cleaner Production 88, 152-159. Yu, T., Wang, X., Li, C., 2013. Removal of Antimony by FeCl3-Modified Granular-Activated Carbon in Aqueous Solution. Journal of Environmental Engineering, American Society of Civil Engineers 140, A40114001. Yurtsever, M., Sengil, I.A., 2009. Biosorption of Pb(II) ions by modified quebracho tannin resin. Journal of Hazardous Materials 163, 58-64. Zhang, D.Y., Pan, X.L., Zhao, L., Mu, G.J., 2011. Biosorption of Antimony (Sb) by the Cyanobacterium Synechocystis sp. Polish Journal of Environmental Studies 20, 1353-1358. Zhao, X.Q., Dou, X.M., Mohan, D., Pittman, C.U., Ok, Y.S., Jin, X., 2014. Antimonate and antimonite adsorption by a polyvinyl alcohol-stabilized granular adsorbent containing nanoscale zero-valent iron. Chemical Engineering Journal 247, 250-257. Zhu, J., Wu, F.C., Pan, X.L., Guo, J.Y., Wen, D.S., 2011. Removal of antimony from antimony mine flotation wastewater by electrocoagulation with aluminum electrodes. Journal of Environmental Sciences 23, 1066-1071. 27 ACCEPTED MANUSCRIPT Tables Table 1 – Extraction results obtained by organic extraction (soxhlet apparatus) with ethanol and by batch alkaline mode. Values represent average from duplicates ± maximum deviation. Extraction in Ethanol Alkaline extraction 3.2±0.4 15±1 ηE (%) Extract properties (expressed per gram of extract): TPC (mg g–1) SN (g g–1) 482±37 378±24 0.66±0.02 0.67±0.02 389±34 352±26 81±2 93±1 FCPC (mg g–1) % FCPC Extracted Quantities (expressed per gram of bark): TEP (mg g–1) 16±3 57±8 ηSN (mg g–1) 21±2 100±10 EFCP (mg g–1) 13±4 53±8 Table 2 –Total extracted phenols and amount of condensable tannins that can be obtained per gram of Pine bark: calculated values from literature results. a Ethanol 75% ηSN (mg g–1) Reference 26 – (Bocalandro et al., 2012) (Vázquez et al., 2001) TEP (mg g–1) Experimental conditions S/L: 1:20 120°C 120 min b Water S/L: 1:8 80°C 30 min – 60 b NaOH 2.5% S/L: 1:6 90°C 30 min – 250 b NaOH 5% S/L: 1:6 90°C 30 min – 270 b NaOH 1% S/L: 1:9 80°C 120 min 62 110 b NaOH 5% S/L: 1:9 80°C 120 min 22 60 (Pepino et al., 2001) This work b NaOH 1% b NaOH 5% a S/L: 1:5 90°C 30 min 114 160 S/L: 1:6 80°C 90 min 57 100 (Chupin et al., 2013) b Pinus radiata; Pinus Pinaster Table 3 – Equilibrium models for Sb adsorption on the tannin-adsorbent (25 ºC): parameters (±standard error) and statistical data. Langmuir kL (L mg–1) 10±2 Qm (mg g–1) 5.4±0.1 0.99 SE (mg g–1) 0.23 KF (mg1–1/ng–1L1/n) 3.88±0.01 pH 4 <6 pH 6 0.5±0.1 3.3±0.5 0.84 0.83 24±2 0.99 1.2 pH 6 - ME 0.16±0.06 30±5 0.99 pH 2 0.13±0.04 27±3 pH 4 0.07±0.03 25±5 pH 2 - ME 0.2±0.1 17±4 pH 2 Sb(III) Sb(V) Freundlich n R 8±2 0.95 SE (mg g–1) 0.50 2.06±0.01 6±5 0.80 0.92 8.26±0.04 2.8±0.3 0.99 1.4 1.7 5±2 1.9±0.4 0.97 2.6 0.99 1.3 4.42±0.03 2.0±0.1 1.00 0.57 0.96 1.2 2.4±0.2 1.5±0.3 0.95 1.5 0.95 1.8 3.6±0.8 2.3±0.5 0.96 1.7 R 28 ACCEPTED MANUSCRIPT Table 4 – Maximum adsorption capacities reported in literature for the uptake of antimony from aqueous solutions by various adsorbents at T=298 K. Adsorbent Cin (mg L–1) pH Qm (mg g–1) Reference 0-20 7 6.99 (Zhao et al., 2014) Sb(III) PVA-Fe0 granules Granular activated carbon 1-4 7 0.54 (Yu et al., 2013) 1.5-4 7 2.64 (Yu et al., 2013) a 2-50 7 4 (Ungureanu et al., 2017) S. muticum (brown seaweed) a 2-50 2 2.1 (Ungureanu et al., 2017) 2-50 2 2.1 (Ungureanu et al., 2016) Ce-doped Fe3O4 10-100 7 224 (Qi et al., 2017) Green bean husk 2.5-100 4 20.1 (Iqbal et al., 2013) Cyanobacteria Synechocystis sp. 5-100 7 4.7 (Zhang et al., 2011) Biochars derived from Canna indica 0-30 5 16.1 (Cui et al., 2017) Tannin-adsorbent 1-30 6 24 This study Ferric hydroxide 0-25 7 18.5 (Li et al., 2012) Fe-Zr binary oxide 0-25 7 60.4 (Li et al., 2012) 10-100 7 188 (Qi et al., 2017) 0-20 7 1.7 (Zhao et al., 2014) FeCl3-modified activated carbon S. muticum (brown seaweed) C. sericea (green seaweed) b Sb(V) Ce-doped Fe3O4 PVA-Fe0 granules C. sericea (green seaweed) b Goethite 2-50 2 3.1 (Ungureanu et al., 2016) 0.05-15 7 18.3 (Xi et al., 2013) 1-30 2 27 This study Tannin-adsorbent a T=296 K; b T=295 K Table 5 – Kinetic models for Sb adsorption on the tannin-adsorbent (25 ºC, adsorbent dosage 0.50 g L–1, pH 6 for Sb(III) and pH 2 for Sb(V)): parameters (±standard error) and statistical data. Pseudo-first order model Pseudo-second order model Cin (mg Sb(III) Sb(V) L–1) k1∙102 (min–1) qeq (mg g–1) k2∙104 SE R (mg g–1) (g mg–1 min–1) qeq (mg g-1) SE R (mg g–1) 1 2.2±0.1 1.96±0.03 1.00 0.05 122±9 2.22±0.03 1.00 0.04 5 1.3±0.2 10.0±0.4 0.99 0.57 12±1 11.8±0.3 1.00 0.25 20 0.58±0.03 23.6±0.6 1.00 4.38 1.5±0.1 32±1 1.00 0.37 1 0.6±0.1 1.7±0.2 0.98 0.11 21±8 2.3±0.1 0.99 0.11 5 0.4±0.1 5.2±0.6 0.99 0.31 5±2 7±1 0.99 0.29 20 0.14±0.05 21±2 1.00 0.46 0.4±0.2 33±6 1.00 0.52 29 ACCEPTED MANUSCRIPT Table 6 – Desorption percentages of antimony from saturated tannin-adsorbent using different eluents (25ºC, saturated adsorbent dosage 2.5 g L-1, 12 h stirring time). Desorption % Eluent Sb(III)-loaded Sb(V)-loaded adsorbent adsorbent HNO3 0.1 M 5.4±0.9 13±2 NaOH 0.1 M 42±6 45±8 NaOH 0.5 M 75±8 69±10 NaCl 0.5 M 1.9±0.9 2±1 EDTA 0.1 M 8±1 2±1 30 ACCEPTED MANUSCRIPT Figures 80 (a) ƞG (%) ƞG (%) 80 60 40 (b) 60 40 20 20 0.4 mL/g 4.0 mL/g 6.0 mL/g 0 0 0.05 0.10 0.20 0.40 0.80 Volume of formaldehyde (mL g–1) 0.8 mL/g Volume of NaOH 4 (mL g–1) 6 solution 8 12 Fig. 1 – Influence on gelification yield by: (a) the amount of formaldehyde used in the reaction, at different volumes of NaOH solution; (b) the volume of sodium hydroxide solution, at different formaldehyde amounts. Absorbance 4.000 3.500 3.000 2.500 2.000 1.500 (a) (b) 1.000 4400 3400 2400 1400 400 wavenumber (cm-1) Fig. 2 – Infrared spectra of (a) tannin alkaline extract, (b) tannin-adsorbent and tannin-adsorbent saturated with (c) Sb(III) and (d) Sb(V). 31 ACCEPTED MANUSCRIPT 25.0 120 100 20.0 Removal % qeq (mg g-1) Fig. 3 – SEM images of (a) tannin extract and (b) adsorbent. 80 15.0 60 10.0 40 5.0 20 0.0 0 0.25 0.50 1.0 2.0 S/L (g L-1) 5.0 Fig. 4 – Influence of the solid:liquid ratio on equilibrium adsorbed amounts of antimonite (bars) and on the removal efficiency (points) (Cin=20 mg L‒1, pH 6). 32 ACCEPTED MANUSCRIPT (a) qeq (mg g‒1) 25.0 Sb(III) Sb(V) 2 4 20.0 15.0 10.0 5.0 0.0 3 5 pH 6 7 8 (b) 25.0 Sb(III) qeq (mg g-1) 20.0 Sb(V) 15.0 10.0 5.0 0.0 control — As As ClNO3- SO42SO42- PO43ClNO3PO43- Fig. 5 – Equilibrium adsorbed amounts of antimony (Cin=20 mg L‒1, S/L=0.50 g L‒1): (a) effect of pH; (b) influence of possible coexisting compounds in solution (pH 6, for Sb(III) and pH 2, for Sb(V)). Fig. 6 – Possible structure of the (a) tannin-adsorbent (adapted from Garcia et al. (2014)), and (b) Sb(III) and (c) Sb(V) complexes formed during adsorption. 33 ACCEPTED MANUSCRIPT qeq (mg g-1) 30.0 (a) Sb(III) 25.0 20.0 pH 2 pH 4 pH 6 pH 6 - ME 15.0 10.0 5.0 0.0 0.0 10.0 20.0 Ceq (mg L-1) 30.0 30.0 qeq (mg g-1) (b) Sb(V) 25.0 20.0 15.0 10.0 pH 2 pH 4 5.0 pH 2 ME 0.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Ceq (mg L-1) Fig. 7 – Equilibrium isotherms for the adsorption of (a) Sb(III) and (b) Sb(V) by tannin-adsorbent at different pH conditions (25 ºC, S/L=0.50 g L‒1), using Sb synthetic solution and a mine tailings water (ME): experimental data and model curves (— Langmuir; - - - Freundlich). 34 ACCEPTED MANUSCRIPT 1.20 (a) Sb(III) C/Cin 1.00 1 mg/L 0.80 5 mg/L 0.60 0.40 0.20 0.00 0.0 2.0 4.0 6.0 time (h) C/Cin 1.00 8.0 10.0 (b) Sb(V) 0.80 0.60 0.40 1 mg/L 5 mg/L 0.20 20 mg/L 0.00 0.0 2.0 4.0 6.0 time (h) 8.0 10.0 Fig. 8 – Adsorption kinetics for (a) Sb(III) and (b) Sb(V) uptake (25 ºC, S/L=0.50 g L‒1, pH=6 and pH=2, respectively): experimental data and model curves (- - - pseudo-first order; — pseudo-second order fittings). 35 ACCEPTED MANUSCRIPT Highlights  Tannins were extracted from maritime pine bark  The production of tannin-based adsorbents was optimized  The tannin-adsorbents successfully removed Sb(III,V) from water