journal of environmental sciences 121 (2022) 1–12 Available online at www.sciencedirect.com www.elsevier.com/locate/jes Decoupling the adsorption mechanisms of arsenate at molecular level on modified cube-shaped sponge loaded superparamagnetic iron oxide nanoparticles Xiang-Yang Lou 1, Roberto Boada 1,∗, Verónica Verdugo 1, Laura Simonelli 2, Gustavo Pérez 1,3, Manuel Valiente 1 1 GTS-UAB Research Group, Department of Chemistry, Facultat de Ciències, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain 2 ALBA Synchrotron, Carrer de la llum 2-26, Cerdanyola del Vallès, 08290 Barcelona, Spain 3 Centro de Excelencia en Nanotecnologia, Leitat Chile, Santiago 7500724, Chile a r t i c l e i n f o a b s t r a c t Article history: In this study, a commercial cube-shaped open-celled cellulose sponge adsorbent was mod- Received 6 July 2021 ified by in-situ co-precipitation of superparamagnetic iron oxide nanoparticles (SPION) and Revised 31 August 2021 used to remove As(V) from aqueous solutions. Fe K-edge X-ray absorption spectroscopy Accepted 1 September 2021 (XAS) and TEM identified maghemite as the main iron phase of the SPION nanoparticles with Available online 31 January 2022 an average size 13 nm. Batch adsorption experiments at 800 mg/L showed a 63% increase of adsorption capacity when loading 2.6 wt.% mass fraction of SPION in the cube-sponge. Keywords: Experimental determination of the adsorption thermodynamic parameters indicated that Adsorption the As(V) adsorption on the composite material is a spontaneous and exothermic process. Cube composite adsorbent As K-edge XAS results confirmed that the adsorption enhancement on the composite can Arsenate removal be attributed to the nanoparticles loaded. In addition, adsorbed As(V) did not get reduced to Regeneration more toxic As(III) and formed a binuclear corner-sharing complex with SPION. The advan- X-ray absorption spectroscopy tageous cube-shape of the sponge-loaded SPION composite together with its high affinity and good adsorption capacity for As(V), good regeneration capability and the enhanced- diffusion attributed to its open-celled structure make this adsorbent a good candidate for industrial applications. © 2022 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Introduction Arsenic (As) is a major concern regarding environmental pol- lution and human health due to its high toxicity and car- cinogenicity. Although natural processes such as soil weath- ∗ ering and volcanic eruptions release As into the environment, Corresponding author. E-mail: roberto.boada@uab.cat (R. Boada). https://doi.org/10.1016/j.jes.2021.09.001 1001-0742/© 2022 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 2 journal of environmental sciences 121 (2022) 1–12 anthropogenic activities are the primary source of arsenic is necessary to develop new methodologies for synthesizing pollution (Alam et al., 2019). The world health organization composite materials in sizes that could be used in column (WHO) establish the maximum discharge level of As at 10 μg/L applications. (Shakya and Ghosh, 2018). Unfortunately, As concentrations In this study, we propose to use SPION supported on com- higher than 50 μg/L can still be found in drinking water of sev- mercially available cube-shaped open-celled cellulose sponge eral countries. As a result, more than 200 million people in the (hereafter referred to as “sponge-loaded SPION”) as a compos- world are exposed to unacceptable levels of As (Castriota et al., ite material for the removal of As(V) from aqueous solution. 2020). The synthesis of the SPION was performed in-situ in the pres- Although As exists in both organic and inorganic forms in ence of the cube-sponge to overcome the drawbacks of the the environment, inorganic forms are more toxic and the more SPION loading obtained by nebulization for adsorbents that common ones (Liu and Xie, 2021). In water, As is mostly found are not in powdered form. Here we report the comprehensive as oxyanions of trivalent arsenite, As(III), or pentavalent arse- characterization of this cube composite adsorbent material, nate, As(V) (He et al., 2020). The latter, As(V), is the thermo- including a comparison of its adsorption performance with dynamically stable form that generally predominates in oxic recently reported Fe-based adsorbent materials. In addition, surface waters. exploiting the sensitivity of X-ray absorption spectroscopy to Among the available methodologies for As removal, ad- the local coordination environment of the proben As atoms, sorption is considered to be the best one for low concen- we have been able to decouple the adsorption occurring at trations, and the most promising one for industrial appli- the different components of the composite sorbent. These re- cations thanks to its fast and easy operation, and the ad- sults have allowed us to get a better understanding of the vantage of being environmentally-friendly and cost-effective As(V) adsorption mechanisms. Besides, the regeneration ca- (Ploychompoo et al., 2020). An ideal adsorbent for water re- pability of the material was also evaluated by performing sev- mediation should be non-toxic, possess excellent chemical eral adsorption-desorption cycles and monitoring the varia- resistance, have very low solubility in water and, ultimately, tion of the adsorption capacity. offer the possibility to be used in practical applications (e.g. in columns or membrane systems). Many types of adsorbents have been reported, such as biomass, metal oxides (especially iron oxide), fly ash, graphene and graphene oxide (Long et al., 1. Materials and methods 2020). Among all these materials, superparamagnetic iron ox- ide nanoparticles (SPION) are considered to be one of the most 1.1. Materials promising adsorbents due to their remarkable features such as small size (< 30 nm), large specific surface area, non-toxic All reagents used were of analytical grade and purchased and facile magnetic separation (Bilgic and Cimen, 2021). from Sigma-Aldrich (Missouri, USA), Panreac (Barcelona, Many efforts are being devoted to optimize the design, syn- Spain), Scharlab (Barcelona, Spain) or Honeywell, (Char- thesis, and usage of SPION to improve their structural stabil- lotte, USA). The stock aqueous solution of 1000 mg/L As(V) ity, adsorption capacity, and adsorption selectivity (Bilgic and was prepared from sodium arsenate dibasic heptahydrate Cimen, 2021). However, one of the main remaining drawbacks (Na2 HAsO4 •7H2 O). The pH was adjusted to 3.6 prior to the of nanoparticles is their easy aggregation when they are sus- adsorption experiments to optimize the adsorption in accor- pended in water. To overcome this issue, nanoparticles have dance with previous works (Morillo et al., 2015). The pH was been often loaded on support materials such as sand, fly ash, adjusted by adding the appropriate amount of hydrogen chlo- and biomass (Hosseini Asl et al., 2017). Our previous stud- ride (HCl) or sodium hydroxide (NaOH). ies have shown that commercially available open-celled cel- Cube-shaped MetalZorb® adsorbent was kindly supplied lulose porous MetalZorb® sponge can be used as a support to by CleanWay Environmental Partners, Inc. (Portland, USA). effectively decrease the aggregation of SPION (Morillo et al., The material is an open-celled cellulose sponge (hereafter 2015). In that work, the SPION was synthesized prior to load- referred simply as “sponge”) which incorporates a water- ing them by spray nebulization over powdered sponge. This insoluble polyamide chelating polymer. The sponge contains method generates a homogeneous dispersion of SPION over free available ethyleneamine and iminodiacetate groups that a very fine powdered sample, but it does not assure a uni- interact with heavy metals ions by ion exchange mechanism form permeation of the SPION into the pores for coarse pow- and chelation, respectively (Muñoz et al., 2002). The sponge der. In addition, this sponge material shows higher As(V) ad- has low impedance of diffusion and it is light in weight. When sorption capacity than other support materials. However, one used in a column, flow rates of three bed volumes per minute of the main limitations for industrial application of previ- can be obtained at hydrostatic pressures only 60 cm above the ously reported works is the use of such composite materials bed and without any additional pressurization. These prop- in very fine powdered form. Hence, when used in adsorption erties make this porous material a good candidate for sup- columns, they would limit the flow rate of the solution and porting nanoparticles in industrial water remediation appli- even clog the column blocking the flow. Besides, the separa- cations. Besides, another very important fact to use this spe- tion of powdered adsorbents from wastewater is a tedious job cific sponge stands on their industrial production and com- and requires sophisticated instrumentations that are hardly mercialization for the removal of metal pollutants in contami- available in developing countries. This reduces the feasibil- nated waters. The averaged dimensions of the cube are (length ity of its implementation at those places with serious con- x width x height): 12.8 ± 1.8 × 10.1 ± 1.1 × 7.3 ± 1.3 mm. Each tamination issues (Kaljurand and Koel, 2011). Therefore, it cube has an average weight of 0.20 ± 0.02 g. journal of environmental sciences 121 (2022) 1–12 3 1.2. Preparation of cube sponge-loaded SPION 1.4. Adsorption experiments Following the manufacturer’s advice, the sponge was pre- The experiments were performed in batch mode using 50 mL treated with 1.0 mol/L HCl to remove all possible pre-absorbed plastic tubes containing 25 mL As(V) solution and one cube cations. This protonates the amino functional groups, which of the adsorbent material (sponge or sponge-loaded SPION, improves the adsorption capacity of the adsorbent. In addi- 0.20 g). The tubes were agitated mechanically at 300 rpm/min tion, this activation step facilitates the immobilization of the and at 293, 318 or 343 K by using a shaker inside a tempera- nanoparticles on the sponge surface. The activation of the as- ture controlled incubator. The solutions were separated from received sponge was performed in batches of 6 cubes in a the adsorbents by filtration and As(V) concentration in the so- 50 mL tube as follow: immersion of the cubes in Milli-Q wa- lution was determined by UV–visible spectrometric technique ter for 24 hr, immersion in 1.0 mol/L HCl solution for 4 hr to based on the molybdenum blue method (see Appendix A Sec- fully protonate the iminodiacetate groups, rinsing with Milli- tion S1). The adsorption capacity and removal percentage at Q water until a pH value of 4.0, immersion in HCl solution at the equilibrium was calculated by using Eq. (1) and Eq. (2), re- pH 2.5 for 2 hr to equilibrate the pH value, and washing 5 times spectively: with Milli-Q water to remove the excess of acid solution. Af- V terwards, the water was decanted, and the cubes were dried qe = (Co − Ce ) (1) m in an oven at 50◦ C for 24 hr. SPION was synthesized by co-precipitation using an alka- line mixture of iron chloride hexahydrate (FeCl3 •6H2 O) and (Co − Ce ) Removal percentage = × 100% (2) ferrous chloride tetrahydrate (FeCl2 •4H2 O) solution as de- Co scribed elsewhere (Uheida et al., 2006). The synthesis was car- where qe (mg/g) is the adsorption capacity, V (L) is the volume ried out while bubbling nitrogen gas to avoid SPION oxidiza- of the arsenic solution, m (g) is the weight of the adsorbent, tion and while stirring at 300 rpm/min to ensure a proper mix- Co (mg/L) and Ce (mg/L) are initial and equilibrium concentra- ing of all the reagents during the synthesis. The loading of tions of As(V) solution, respectively. SPION in the cube sponge was carried out in-situ as follows: first, 100 mL Milli-Q water was added to 500 mL Erlenmeyer 1.5. Adsorption-desorption cycle experiments flask, deoxygenated for 30 mins and heated at 40 °C. Then, the Fe(II) and Fe(III) precursors (0.02 and 0.04 mol/L, respectively), For regenerating the sponge-loaded SPION cube adsorbent and six sponge cubes were added to the solution and leave it material after adsorption (initial concentration 200 mg/L), the stirring for 30 min. Afterwards, 120 mL of 0.5 mol/L NaOH was cube was separated by filtration, washed three times with added to the solution dropwise and the mixture was stirred Milli-Q water, and immersed in 25 mL 0.5 mol/L NaOH solution for an hour to leave enough time for completing the reaction. to desorb the arsenic. After shaking at 300 rpm for 60 min, the The resulting composite material was separated by decanting, adsorbent was separated by filtration and rinsed with 25 mL washed three times with 50 mL deoxygenated Milli-Q water Milli-Q water. Finally, the cube sponge-loaded SPION was dried and dried in an oven at 50◦ C for 24 hr. at 50 °C for 24 hr before carrying out the next adsorption run. To characterize the SPION particles at different depths of This adsorption/desorption cycle was repeated five times to the cube material, several cube samples were cut and di- determine the regeneration capacity of the composite adsor- vided in three parts: surface (1.5 mm from the surface), quar- bent material. ter (1.5–3.0 mm from the surface) and center (remaining part of the cube) as shown in Appendix A Fig. S1. These parts 1.6. Characterization of sponge-loaded SPION account for the 68%, 28% and 4% of the total cube volume, respectively. The XAS experiments were carried out at BL22 CLÆSS beam- line of ALBA CELLS synchrotron (Simonelli et al., 2016). Fe K-edge XAS measurements were collected in transmission 1.3. Characterization of sponge-loaded SPION mode at room temperature to study the iron oxide phase in both SPION and sponge-loaded SPION. A Si(111) double crystal Fourier transform infrared spectroscopy (FT-IR, Tensor 27, monochromator. For the sponge-loaded SPION samples, 22 mg Bruker, USA) measurements were performed to identify of the material was pressed into 5 mm pellets. For SPION sam- the functional groups of SPION, sponge, and sponge-loaded ples and Fe reference compounds, 5 mg of SPION was thor- SPION. Scanning electron microscope (SEM, Merlin, Carl Zeiss oughly mixed with 100 mg of cellulose and pressed into 13 mm AG, Germany) images were collected to get information about pellets. the morphology of the materials. The morphology and size of As K-edge XAS measurements were performed to unravel the SPION were characterized by Transmission Electron Mi- the chemical state and coordination environment of the ad- croscope (TEM, JEM-1400, JEOL, Japan) in a section of the cube sorbed arsenic after exposing the adsorbents to an initial con- composite material. X-ray powder diffraction (XRD, X-Pert, centration of 200 mg/L As(V). In order to get information at Philips, Netherlands) was carried out to identify the iron ox- different depths of both materials, cube sponge and sponge- ide phase and the crystallinity of the nanoparticles. X-ray ab- loaded SPION, the cubes were split in three regions as men- sorption spectroscopy (XAS) measurements at the Fe K-edge tioned before: surface, quarter and center. A Si(311) double were performed at BL22 CLÆSS beamline of ALBA CELLS syn- crystal monochromator was used. Measurements of the As- chrotron (Simonelli et al., 2016). sponge, As-sponge-loaded SPION adsorbents and As reference 4 journal of environmental sciences 121 (2022) 1–12 Fig. 1 – SEM images for (a) SPION, (b) sponge, and (c) sponge-loaded SPION. TEM images and particle size histogram for (d, e, f) surface and center (g, h, i) parts of sponge-loaded SPION. samples were performed in transmission mode using ioniza- see Fig. 1a. The open-celled structure of sponge composed tion chambers. These samples were milled and pressed into by connected pores can be seen in Fig. 1b. The loading of pellets without using any binder. For the As-SPION sample, SPION into sponge is clearly indicated by the darkening of the spectra was collected in fluorescence mode using a multi- the final material with respect to the original sponge (see Ap- element silicon drift detector with Xspress3 electronics. In this pendix A Fig. S2). After loading the nanoparticles, the sur- case, to reduce the self-absorption effects, the As-SPION sam- face of sponge becomes more heterogeneous, see Fig. 1c. It ple was homogeneously mixed with cellulose using a 1:10 ra- is clear that having the sponge as support helps in reducing tio. The measurements were carried out at liquid nitrogen the nanoparticle aggregation to a large extent. SEM images of temperature to minimize any radiation damage. sponge and sponge-loaded SPION with smaller magnification XAS data analysis was performed according to standard are also shown in Appendix A Fig. S3. procedures using ATHENA and ARTEMIS software of the The TEM images of sponge-loaded SPION and the corre- Demeter package (Ravel and Newville, 2005). sponding particle size distribution histograms calculated us- ing Image-Pro-Plus software (Media Cybernetics, Silver Spring, USA) are shown in Fig. 1 (middle and bottom panels). The 2. Results and discussion nanoparticles are coating the surface of the sponge and they are not blocking the pores, which preserves the advantageous 2.1. Characterization of adsorbent materials convection properties of the original sponge material. The nanoparticles are mostly spherical, and the mean nanoparti- The SEM images (top panels in Fig. 1) revealed the microstruc- cle size found for the surface and the center parts of the cube ture and morphology of the adsorbent materials. A severe ag- sponge-loaded SPION were 12.7 and 13.8 nm, respectively. The gregation of the nanoparticles was found for SPION sample, histogram shows that the particle size dispersion, considering journal of environmental sciences 121 (2022) 1–12 5 gates. On the other side, the spectrum of the SPION nanopar- ticles loaded in the sponge is slightly different from the one of the SPION sample. It also resembles the spectral profile of the maghemite reference, but the spectral features A, B and C are smeared out respect to SPION. Similarly, we found that, when comparing the extracted extended X-ray absorp- tion fine structure (EXAFS) signal for both samples (see in- set in Fig. 2b), the amplitude of the oscillation is smaller for sponge-loaded SPION. This is related with the decrease of the coordination number of Fe atoms close to the surface of the iron oxide nanoparticles and the increase of disorder respect to the bulk structure. These observations point out that the main iron phase of the nanoparticles in the sponge-loaded SPION is maghemite, i.e. the Fe oxidation state is dominated by Fe(III). This is beneficial in terms of the adsorption capac- ity since maghemite provides a higher adsorption capacity and a more stable Fe-As complex after As(V) adsorption than magnetite (Mejia-Santillan et al., 2018). Regarding the differ- Fig. 2 – XRD spectra of SPION, cube sponge-loaded SPION, ent parts of the cube, no significant change in the chemical simulated γ-Fe2 O3 and simulated Fe3 O4 . state of the iron oxide was found at different parts of the cube as shown in Appendix A Fig. S4. A rough estimation of the atomic Fe concentration obtained from the absorption jump of the XAS spectrum revealed that there is an average of ∼15% a normal distribution, has similar width for both, surface and less Fe in the surface part than in the center part. This ob- center parts of the cube, see Fig. 1f and Fig. 1i. However, almost servation agrees with the smaller particle size and the higher a bimodal distribution is found at the surface part which high- surface/volume ratio of the SPION loaded on the surface part lights the important contribution of the smaller particles. In- of the cube obtained from TEM. Hence, this corroborates that deed, a detailed analysis of the particle size data revealed that, the particles at the surface part of the cube are smaller respect in average, due to the smaller particle size, the surface/volume to the ones at the center. ratio is 10% larger at the surface part. XRD was carried out to identify the iron oxide phase of 2.2. Adsorption performance the nanoparticles. Fig. 2 shows the comparison of the experi- mental diffraction patterns with the simulated ones for Fe2 O3 To assess the As(V) adsorption performance of cube sponge (American Mineralogist Crystal Structure Database, code: and cube sponge-loaded SPION, batch adsorption experi- 0007898) (Barrón et al., 2003) and Fe3 O4 (American Mineralo- ments with various initial concentrations, contact times and gist Crystal Structure Database, code: 0002400) (Haavik et al., different temperatures were performed. As seen in Fig. 4a, the 2000). The diffraction pattern of both samples displayed the adsorption capacity of sponge and sponge-loaded SPION in- characteristic reflections of the spinel structure; however, it crease when increasing the As(V) initial concentration. Also, was not possible to determine the iron phase since the diffrac- both adsorbents show better adsorption at 293 K than at tion patterns of maghemite, γ -Fe2 O3 , and magnetite, Fe3 O4 , 343 K, while sponge-loaded SPION has a higher adsorption are very similar and the width of the diffraction peaks yield capacity than sponge alone. These adsorption experiments by the nanoparticles were quite wide. In order to further iden- did not reach saturation within the concentration range stud- tify the iron oxide phase, XAS analysis at the Fe K-edge was ied. However, the maximum initial concentration used in this performed. Fig. 3 shows the comparison of the X-ray absorp- study is 800 mg/L, which is much higher than the concentra- tion near edge structure (XANES) region of the Fe K-edge XAS tion found in polluted groundwater (Ali et al., 2019). However, spectra collected on SPION, sponge-loaded SPION (surface) it is worth mentioning that for the lowest As(V) concentration and the reference bulk iron oxide materials, magnetite (Fe3 O4 ) considered for this study (25 mg/L), which is much more rel- and maghemite (γ -Fe2 O3 ). Magnetite and maghemite refer- evant for environmental remediation applications, a 98% re- ences show a clearly differentiated XANES spectra. The posi- moval was achieved. tion of the absorption rising edge, which is highly influenced Adsorption results with different contact times, Fig. 4b, re- by the oxidation state of the Fe ions, appears at higher en- vealed that there is a transient initial stage in which the ad- ergy for maghemite which contains only Fe(III) ions. Indeed, sorption capacity sharply increases. This can be attributed to maghemite can be considered as a Fe(II)-deficient magnetite. the large number of adsorption sites initially available on the In addition, the energy shift of the edge is also influenced by surface of the adsorbent. The adsorption capacity becomes al- the Fe−O bond length (the shorter the interatomic distance, most constant after ∼60 min. When comparing the two dif- the higher the edge energy) and this can account for bond ferent adsorbents studied, sponge-loaded SPION showed a length expansion in non-stochiometric iron oxide nanoparti- higher equilibrium adsorption capacity and a slightly faster cles due to the reduced coordination. In our case, the XANES adsorption process. spectrum of SPION is almost identical to the maghemite ref- Our results show that the adsorption capacity of the erence reflecting the bulk character of the nanoparticle aggre- sponge increases notably after loading SPION. Considering an 6 journal of environmental sciences 121 (2022) 1–12 Fig. 3 – Fe K-edge XANES spectra (a) and EXAFS spectra (b) of SPION, sponge-loaded SPION (surface) and the two iron oxide reference materials (Fe3 O4 , γ-Fe2 O3 ). Fig. 4 – As(V) adsorption capacity using cube sponge and sponge-loaded SPION with different (a) initial concentration and (b) contact time. initial concentration of 200 mg/L and at 293 K, the adsorption centration of 800 mg/L an adsorption capacity of 69.68 mg/g is capacity increases from 18.7 to 24.0 mg/g after loading SPION. obtained for sponge-loaded SPION which outperforms those This means that the adsorption capacity of sponge increases other Fe-based adsorbents. 28.3% after loading only 2.6 wt.% mass fraction of SPION (63.2% The independent adsorption performance of SPION sup- at 800 mg/L initial concentration). This can be attributed to ported in the sponge can be assessed when the adsorption the high affinity of SPION to As(V) as it has been reported capacities of the two materials forming the composite ad- previously (Hao et al., 2018). In fact, this adsorption capacity sorbent are disentangled. When comparing with the adsorp- is considerably higher than the ones reported for other iron- tion capacity of the SPION sample, 27.7 mg/g, we found that, based composite materials, see Table 1. For example, Bakshi in proportion, the supported SPION adsorbs ∼8 times more and coworkers reported a maximum adsorption capacity of As(V) than the unsupported nanoparticles. This reflects the 15.66 mg/g when loading 59.64 wt% zero-valent iron on red reduction of the surface area due to the aggregation of the un- oak biochar (particle size < 0.04 mm) (Bakshi et al., 2018). A supported nanoparticles in the SPION samples as it has been widely used iron-hydroxide based granulate commercial ad- shown above in the SEM images. Indeed, the different behav- sorbent GEH-102 (Fe(OH)3 and β-FeO(OH), size 300–2000 μm) ior of the SPION when loaded in the sponge is also reflected shows a maximum As(V) adsorption capacity of 25.24 mg/g in the temperature variation of the adsorption capacity. For (Moraga et al., 2019). In our case, the mass fraction of the iron SPION sample, there is a 21.7% adsorption increase when go- nanoparticulate adsorbent loaded is much lower than in those ing from 293 to 343 K, whereas the SPION, when loaded on works, and, in addition, we are not quoting the calculated the sponge, displays a larger increase, 38.6%, see Appendix A maximum adsorption capacity as it is usually reported by sev- Table S1. This explains why the temperature variation of the eral authors. In this study, for a maximum initial As(V) con- whole composite adsorbent decreases less, −9.6%, than the journal of environmental sciences 121 (2022) 1–12 7 Table 1 – Comparison of adsorption capacities with similar system. Iron Maximum loading adsorption Support material Size (mm) (wt.%) Iron phase capacity (mg/g) Reference ∗ Cellulose sponge cube 1.80 γ -Fe2 O3 69.68 This work (13×10×7) Sponge half-cylinder ∼12 iron oxide 4.6 (Nguyen et al., 2010) (30 Ø x 15) NONE 0.3-2 52-62 Fe(OH)3 ; 25.24 (Moraga et al., 2019) β-FeO(OH) Corn straw biochar 0.5-1 6.05 Fe 6.80 (He et al., 2018) Activated carbon 0.85 3.5 iron 4.96 (Nieto-Delgado et al., 2019) oxyhydroxides Red oak biochar <0.04 59.64 Zero-valent iron 15.66 (Bakshi et al., 2018) Switchgrass biochar <0.04 55.72 Zero-valent iron 6.48 (Bakshi et al., 2018) Clay-activated carbon 2–3 – iron oxide 5.0 (Pawar et al., 2018) Fungal biomass <0.4 – iron oxide 10.3 (Cárdenas-González et al., 2017) Darco 20×50 0.3-0.85 4.22 β-FeOOH 1.95 (Chang et al., 2010) ∗ at 800 mg/L initial concentration. value that can be estimated when combining the data from The adsorption isotherm data of sponge and sponge- the individual components, −22.2%. loaded SPION at 293 and 343 K were fitted to these models, see Fig. 5. The results from the fits are displayed in Table 2. A 22.3. Adsorption isotherm modeling relatively good fit was found for the three models (R2 >0.90), however, the best model is Freundlich which highlights the Adsorption isotherms were performed to understand the ad- importance of the heterogeneous surface of the adsorbents. sorption mechanism of cube sponge-loaded SPION. Langmuir, In Freundlich isotherm model, the values of 1/n are less than Freundlich, and Temkin, the most used isotherm models, were 1 for both adsorbents which indicates that the As(V) adsorp- applied to interpret the experimental data (Tran et al., 2020). tion is favorable in both materials. Besides, significant higher Langmuir model assumes that the adsorption occurs on a ho- KF values for the sponge-loaded SPION system at both temper- mogeneous surface and that all binding sites have equal affin- atures means that adsorption capacity increased after loading ity for the adsorbate (Eq. (3)). Freundlich isotherm assumes the nanoparticles. multilayer adsorption, with non-uniform distribution of ad- sorption heat and affinities over the heterogeneous surface 2.4. Adsorption kinetic studies (Eq. (4)). Temkin isotherm contains a factor that explicitly con- siders the adsorbent–adsorbate interactions and it is mostly Kinetic parameters are important to determine the efficiency used for electrostatic interaction-based chemical adsorption and mechanism of the adsorption process that is a key knowl- (Eq. (5)). edge to scale up the As(V) removal. To study the rate and Langmuir isotherm model: mechanism of As(V) adsorption onto sponge-loaded SPION,   the experimental kinetic data were fitted to the pseudo-first 1 1 1 1 = + (3) order kinetic model (Eq. (6)) and pseudo-second order kinetic qe KL qm Ce qm model (Eq. (7)) (Tran et al., 2017): Freundlich isotherm model: Pseudo-first order kinetic model: 1 K1 log qe = log KF + log Ce (4) log (qe − qt ) = log qe − t (6) n 2.303 Temkin isotherm model: Pseudo-second order kinetics model: RT RT t 1 t qe = ln AT + ln Ce (5) = + (7) bT bT qt K2 q2e qe where qe (mg/g) and qm (mg/g) are the equilibrium and max- Where, qe (mg/g) and qt (mg/g) are the adsorption capacity imum adsorption capacity, respectively. Ce (mg/g) is the con- at equilibrium and at any time t (min), respectively; K1 and K2 centration remaining in the solution. KL (L/mg) is the Lang- are the rate constants of each respective model. muir constant, which is related to the pollutant affinity to Sponge and sponge-loaded SPION adsorption data at 293 binding sites. KF ((mg/g)(L/mg)1/n ) and n are Freundlich ad- and 343 K were fitted to these kinetic models, see Fig. 6. Re- sorption constants, and 1/n gives an indication of the prefer- gression coefficients and rate constants are shown in Table 3. ential adsorption process. R is the ideal gas constant 8.3145 In this case, the regression coefficients of the pseudo-second [J/(mol•K)] and T (K) is the absolute temperature. AT (L/g) and order model are higher than 0.99 for both sponge and sponge- bT (kJ/mol) are Temkin constants, which are linked to the max- loaded SPION, while regression coefficients of pseudo-first or- imum binding energy and heat of adsorption, respectively. der were below 0.80. Hence, the adsorption kinetics of both 8 journal of environmental sciences 121 (2022) 1–12 Fig. 5 – Isotherm modeling of As(V) adsorption on sponge and sponge-loaded SPION experimental data. (a) at 293 K and (b) at 343 K. Table 2 – The parameters of different isotherms models for sponge and sponge-loaded SPION, expressed in the following units: T (K), qm (mg/g), KL (L/mg), KF ((mg/g) (L/mg)1/n ), AT (L/g), bT (kJ/mol). Langmuir isotherm Freundlich isotherm Temkin isotherm Adsorbents T qm KL R2 KF 1/n R2 AT bT R2 sponge 293 45.00 0.012 0.95 3.49 0.39 0.96 0.31 0.32 0.95 343 39.36 0.008 0.99 2.16 0.43 0.94 0.13 0.39 0.97 sponge-loaded 293 72.24 0.036 0.94 10.12 0.34 0.98 2.21 0.24 0.96 SPION 343 58.97 0.019 0.98 6.69 0.35 0.98 1.35 0.36 0.93 Fig. 6 – Kinetic modeling of As(V) adsorption on sponge and sponge-loaded SPION at 293 K and 343 K. (a) Pseudo-first order model and (b) Pseudo-second order model. systems are better described by the pseudo-second order functional groups and As(V) during adsorption (Hu et al., 2019; model. Therefore, chemisorption is controlling the speed of Singh et al., 2012). the adsorption process. FT-IR results (Appendix A Fig. S5) The activation energy Ea for adsorption of As(V) onto confirm the existence of free available hydroxyl group (-OH), sponge-loaded SPION can be calculated from the pseudo- ethyleneamine (-NH2 –CH2 –CH2 -) and iminodiacetate (–(C = second order rate constant, K2 , using the following Arrhenius O)-NH-(C = O)-) groups in the adsorbents. This implies the equation (Eq. (8)) (Lima et al., 2019): influence of ligand exchange and ion exchange mechanisms Ea through the sharing/exchange of electrons between adsorbent ln K2 = ln A − (8) RT journal of environmental sciences 121 (2022) 1–12 9 Table 3 – Kinetic parameters for As(V) adsorption onto sponge and sponge-loaded SPION. T Pseudo-first order Pseudo-second order Adsorbent (K) −1 2 K1 (min ) R K2 (g/(mg•min)) R2 sponge 293 0.0268 0.5671 0.0782 0.9999 343 0.0167 0.2468 0.0929 0.9964 sponge-loaded 293 0.0316 0.4977 0.0519 0.9996 SPION 343 0.0645 0.7299 0.0297 0.9997 Table 4 – Thermodynamic parameters for As(V) adsorp- tion onto sponge and sponge-loaded SPION. Adsorbent T(K) Go (kJ/mol) Ho (kJ/mol) So (J/mol) sponge 293 −14.22 −12.30 6.56 318 −14.39 Fig. 7 – As K-edge XANES spectra: (a) compare of the spectra 343 −14.55 collected on the adsorbent materials with the As reference sponge- 293 −19.28 −22.80 −12.02 compounds and (b) compare of the spectra collected on the loaded 318 −18.98 different parts of cube sponge and sponge-loaded SPION. SPION 343 −18.68 Values of H° are −12.30 and −22.80 kJ/mol for sponge and where K2 is the adsorption rate constant, A is the Arrhenius sponge-loaded SPION respectively, which indicates the As(V) constant, R is the ideal gas constant 8.3145 (J/(mol•K)) and T (K) adsorption on these two adsorbents is an exothermic pro- is the absolute temperature. The Ea values obtained for sponge cess. The loading of 2.6 wt.% SPION on sponge did not change and sponge-loaded SPION were 9.86 and 17.69 kJ/mol, respec- the exothermic nature of As(V) adsorption on the sponge, al- tively. This indicates that sponge-loaded SPION shows higher though the adsorption on iron oxide is an endothermic pro- dominance of chemical adsorption than original sponge. cess as previously reported (Liu et al., 2015). The spontaneous and exothermic nature of As(V) adsorption imply that adsorp- 2.5. Thermodynamics analysis tion of As(V) on the adsorbents is favorable, which agrees with the results obtained from fitting of Freundlich model. Thermodynamic parameters such as Gibbs’ free energy change (Go ), enthalpy change (Ho ) and entropy change 2.6. Chemical state and coordination environment of (So ) can predict the feasibility and nature of the adsorption adsorbed As process (Lima et al., 2019). The Gibbs free energy of adsorption can be expressed as (Eq. (9)): Although the study of the macroscopic adsorption behav- o o ior and the thermodynamic calculations provides some evi- G = −RT ln K (9) dence that the arsenic is chemically adsorbed on sponge and where, Ko is the standard distribution coefficient of the ad- sponge-loaded SPION, information about the oxidation state sorption (Eq. (10)): and coordination environment of As after being adsorbed is still missing. As K-edge XAS measurements were performed qe Ce Ko = / (10) to get a better insight of the adsorption process at the molec- qo Co ular scale. where, qo and Co and are the standard states of solution in Fig. 7a displays the As K-edge XANES spectra collected on solid and liquid, respectively. Here, qo is selected as 1 mol/kg, the adsorbent materials and the As reference compounds (ar- and Co is selected as 1 mol/L. senic trioxide, As(III); and arsenic pentoxide, As(V)). These two G° is also related to S° and H° as follow (Eq. (11)): As reference compounds display a markedly different spectral profile. The pentoxide is shifted 3.7 eV towards higher energy Go = Ho − T So (11) respect to the trioxide. Thus, the white-line feature (first res- onance after the rising absorption edge) appears at 11,875.5 By constructing a plot of lnKo versus 1/T, S° can be cal- and 11,871.7 eV for pentoxide and trioxide, respectively. The culated from the intercept and H° can be obtained from the direct comparison of the spectra collected on the adsorbent slope. The results are displayed in Table 4 and Appendix A Fig. samples with the As reference compounds reveals that the S6. arsenic adsorbed in the materials remains as As(V). Hence, no The negative Go values obtained for sponge and sponge- major redox reaction occurred upon As(V) adsorption on any loaded SPION at three temperatures indicate that the adsorp- of the materials. The spectral feature after the white line ap- tion of As(V) onto both adsorbent materials is spontaneous. pears at different energy for sponge and SPION reflecting the 10 journal of environmental sciences 121 (2022) 1–12 Fig. 8 – As K-edge EXAFS spectra: (a) EXAFS spectra for adsorbents after As(V) adsorption at 293 K and results of the linear combination fitting analysis for surface part of sponge-loaded SPION (line); (b) Pseudo-radial distribution function for adsorbents after As(V) adsorption at 293 K (symbols are experimental data and lines are fits). different coordination environment of As(V) in the two adsor- found in Fe K-edge XAS spectra after adsorption As for SPION bents. As expected, the spectrum for the composite adsorbent and sponge-loaded SPION samples. sample, sponge-loaded SPION, lies in between the two, being The Fourier-transformed spectra in R-space, see Fig. 8b, is more similar to sponge due to the low loading of nanopar- dominated by the As−O first-neighbor contribution. The mod- ticles. As shown in Fig. 7b, the XANES spectra does not re- elling of the EXAFS signals revealed that the As-O bond dis- markably change when increasing the adsorption tempera- tance in sponge and SPION are 1.69 and 1.70 Å, respectively ture from 293 to 343 K, therefore, the temperature does not (Appendix A Section S12 for further details about the EXAFS affect relevantly the oxidation state of the adsorbed As. In ad- modelling). In addition, for SPION and sponge-loaded SPION, dition, no change in the oxidation state of the adsorbed As was there are relatively weak features beyond the first coordina- found as a function of the penetration depth of the cube nei- tion shell that can be attributed to the As−Fe contributions. ther in sponge nor in sponge-loaded SPION, see Fig. 7b. Hence, The different arrangements of the As(V) tetrahedron with the the same adsorption process occurs at the different parts of surface of the iron oxide nanoparticle are related to the ar- the cube due to the good diffusion properties of the sponge senic surface coverage (van Genuchten et al., 2020). Among all, material and the rather homogeneity of the composite adsor- binuclear corner-sharing, 2 C, is found to be the dominant con- bent material. figuration, because it is strongly favored thermodynamically The different local coordination environment of As in and kinetically, and it is also the most energetically stable. SPION and sponge adsorbents is more evident when repre- Some authors have reported the presence of the mononuclear senting the EXAFS signal and the pseudo-radial distribution corner-sharing, 1 V, when arsenic has a relatively low surface function, see Fig. 8. SPION displays a more structured EXAFS coverage, but we can exclude this possibility in the present signal than sponge. As can be seen Fig. 8a, for SPION, feature case since no contribution appears in the pseudo-radial dis- A from sponge is double, and features B and C are sharper and tribution function and, in our study, the high initial concen- have a shoulder at lower k, whereas those features are much tration of arsenic (200 mg/L) and the good affinity of SPION simpler and smoother in the case of sponge. As expected, the for As(V) results in a high surface coverage. In that sense, the EXAFS for the sponge-loaded SPION is similar to the one for mononuclear edge-sharing configuration, 2 E, is reported to be sponge, but it has some resemblance to SPION as can be no- present in some cases with the high arsenic surface coverage. ticed by the slight flattening of feature A and the change in However, our fitting results are compatible with a dominant shape of features B and C. A semi-quantitative analysis of the contribution of the 2 C configuration (see Table 5). In this case, ratio between the two components, sponge and SPION, in the two of the four As–O bonding structures are complexed to Fe sponge-loaded SPION sample can be obtained from the lin- atoms (i.e., As–O–Fe) by corner-sharing with the FeO6 octahe- ear combination fitting analysis of the EXAFS signal. The re- dron and the remaining two As atoms are present either both sults show that, at 293 K, around 21% of As(V) is adsorbed into as unprotonated As–O or one as unprotonated As–O and the the loaded nanoparticles (Fig. 8a and Appendix A Fig. S7a), other one as protonated As–O–H (Zhao and Guo, 2014). Despite whereas this amount increases to 42% at 343 K (Appendix A the differences regarding particle aggregation, both SPION and Fig. S7b). These results are in good agreement with those ob- sponge-loaded SPION samples are composed by maghemite, tained from the adsorption data reported above and confirm thus, the coordination on the composite will follow similar that the As(V) adsorbed on the loaded nanoparticles increases trend as in the case of the unsupported nanoparticles. when increasing the temperature. In addition, no change was journal of environmental sciences 121 (2022) 1–12 11 Table 5 – EXAFS fitting results of SPION, sponge (S) and loaded (S) at 293 K. Sample Atomic Pairs S0 2 CN R (Å) σ 2 (Å2 ) E0 (eV) R-factor SPION As-O 0.85 4 1.70 (0.02) 0.001 (0.0005) 5.7 0.031 As-O-O 12 1.82∗ (RAs-O ) = 3.10 σ 2 (As-O) (1.1) As-Fe 2 3.39 (0.04) 0.004 (0.003) sponge (S) As-O 1.1 4 1.69 (0.01) 0.003 (0.001) 5.4 (1.8) 0.009 sponge-loaded SPION (S) As-O 1.1 4 1.69 (0.01) 0.004 (0.002) 5.9 (2.0) 0.010 The isotherm and kinetic modelling indicate a multilayer and heterogeneous adsorption process in which the chemi- cal adsorption is the controlling step. The results from the thermodynamic analysis indicate that the As(V) adsorption on composite material is a spontaneous and exothermic pro- cess. In addition, we observed that As(V) was not reduced to more toxic As(III) after being adsorbed, which is desirable for the treatment of As(V) contaminated waters. The model- ing of the EXAFS signals revealed that the binuclear corner- sharing configuration, 2 C, dominates the As(V) adsorption on SPION. Moreover, we have shown that the cube-shaped sponge-loaded SPION has good regenerative capability and can be used over several adsorption/desorption cycles. In brief, this cube sponge-loaded SPION combines the advantages of both SPION (high affinity, good adsorption capacity) and cube Fig. 9 – Adsorption capacity and weight of adsorbent in commercial sponge (cheap, open-celled structure and favor- different regeneration cycles. able cube-shape) which makes it a promising material for in- dustrial applications. 2.7. Regeneration of adsorption materials Acknowledgments To assess the regeneration capacity of cube sponge-loaded We acknowledge CleanWay Environmental Partners, Inc. for SPION, several cycles of adsorption and desorption experi- kindly supplying the cube-shaped MetalZorb® sponge. We ments were performed. As shown in Fig. 9, the adsorption ca- acknowledge ALBA synchrotron facility for beamtime (No. pacity of the adsorbent did not show a significant decrease. 2018093125) at BL22-CLAESS beamline. This research was Hence, the cube adsorbent maintains a high adsorption capac- financially supported by the MINECO ministry of Spain ity even after five adsorption/desorption cycles. The slight in- (No. CTM2015–65414-C2–1-R) and the Innovation Found for crease observed for the fourth and fifth cycles, can be ascribed Competitiveness of the Chilean Economic Development to the loose powder which broke off from the sponge-loaded Agency (CORFO, No. es 13CEI2–21839). X-Y. Lou acknowl- SPION during shaking which contributes to the increase of the edges the China scholarship council for the fellowship (No. effective As adsorption from the solution. Besides its good re- 201708110179). R. Boada acknowledges funding support from generation capabilities, one of the main benefits of this com- the European Union’s Horizon 2020 research and innovation posite adsorbent is that, after its life-time span, it could be program under the Marie Skłodowska-Curie grant agreement easily compressed into a very small volume which hugely fa- (No. 665919). cilitates its transport and ultimately waste disposal. Appendix A Supplementary data 3. Conclusions Supplementary material associated with this article can be In this study, the composite material cube-shaped open- found, in the online version, at doi:10.1016/j.jes.2021.09.001. celled cellulose sponge loaded superparamagnetic iron ox- ide nanoparticles (SPION) was successfully synthesized by the references in-situ co-precipitation method and used to remove As(V) from aqueous solution. The adsorption capacity of sponge had a 63.2% enhancement after loading 2.6 wt.% of SPION Alam, M.Z., Hoque, M.A., Ahammed, G.J., McGee, R., (maghemite) reaching 69.68 mg/g at 800 mg/L initial As(V) con- Carpenter-Boggs, L., 2019. Arsenic accumulation in lentil (Lens centration. This value is higher than the maximum adsorption culinaris) genotypes and risk associated with the capacity reported for other Fe-based adsorbents. consumption of grains. Sci. Rep. 9, 9431. 12 journal of environmental sciences 121 (2022) 1–12 Ali, W., Rasool, A., Junaid, M., Zhang, H., 2019. A comprehensive Mejia-Santillan, M.E., Pariona, N., Bravo-C, J., Herrera-Trejo, M., review on current status, mechanism, and possible sources of Montejo-Alvaro, F., Zarate, A., et al., 2018. Physical and arsenic arsenic contamination in groundwater: a global perspective adsorption properties of maghemite and magnetite with prominence of Pakistan scenario. Environ. Geochem. sub-microparticles. J. Magn. Magn. Mater. 451, 594–601. Health 41, 737–760. Moraga, B., Toledo, L., Jelínek, L., Yañez, J., Rivas, B.L., Urbano, B.F., Bakshi, S., Banik, C., Rathke, S.J., Laird, D.A., 2018. Arsenic sorption 2019. Copolymer–hydrous zirconium oxide hybrid on zero-valent iron-biochar complexes. Water Res. 137, microspheres for arsenic sorption. Water Res 166, 115044. 153–163. Morillo, D., Pérez, G., Valiente, M., 2015. Efficient arsenic(V) and Barrón, V., Torrent, J., De Grave, E., 2003. Hydromaghemite, an arsenic(III) removal from acidic solutions with Novel Forager intermediate in the hydrothermal transformation of 2-line Sponge-loaded superparamagnetic iron oxide nanoparticles. J. ferrihydrite into hematite. Am. Mineral. 88, 1679–1688. Colloid Interface Sci. 453, 132–141. Bilgic, A., Cimen, A., 2021. Synthesis, characterisation, adsorption Muñoz, J.A., Gonzalo, A., Valiente, M., 2002. Arsenic adsorption by studies and comparison of superparamagnetic iron oxide Fe(III)-loaded open-celled cellulose sponge. Thermodynamic nanoparticles (SPION) with three different amine groups and selectivity aspects. Environ. Sci. Technol. 36, 3405–3411. functionalised with BODIPY for the removal of Cr(VI) metal Nguyen, T.V., Vigneswaran, S., Ngo, H.H., Kandasamy, J., 2010. ions from aqueous solutions. Int. J. Environ. Anal. Chem. 1–26. Arsenic removal by iron oxide coated sponge: Experimental Cárdenas-González, J.F., Acosta-Rodríguez, I., Téran-Figueroa, Y., performance and mathematical models. J. Hazard. Mater. 182, Rodríguez-Pérez, A.S., 2017. Bioremoval of arsenic (V) from 723–729. aqueous solutions by chemically modified fungal biomass. 3 Nieto-Delgado, C., Gutiérrez-Martínez, J., Rangel-Méndez, J.R., Biotech 7, 226. 2019. Modified activated carbon with interconnected fibrils of Castriota, F., Zushin, P.J.H., Sanchez, S.S., Phillips, R.V., Hubbard, A., iron-oxyhydroxides using Mn2+ as morphology regulator, for a Stahl, A., et al., 2020. Chronic arsenic exposure impairs superior arsenic removal from water. J. Environ. Sci. 76, adaptive thermogenesis in male C57BL/6 J mice. Am. J. Physiol. 403–414. - Endocrinol. Metab. 318, E667–E677. Pawar, R.R., Lalhmunsiama, Kim, M., Kim, J.G., Hong, S.M., Chang, Q., Lin, W., Ying, W.C., 2010. Preparation of Sawant, S.Y., Lee, S.M., 2018. Efficient removal of hazardous iron-impregnated granular activated carbon for arsenic lead, cadmium, and arsenic from aqueous environment by removal from drinking water. J. Hazard. Mater. 184, 515–522. iron oxide modified clay-activated carbon composite beads. Haavik, C., Stølen, S., Fjellvåg, H., Hanfland, M., Häusermann, D., Appl. Clay Sci. 162, 339–350. 2000. Equation of state of magnetite and its high-pressure Ploychompoo, S., Chen, J., Luo, H., Liang, Q., 2020. Fast and modification: Thermodynamics of the Fe-O system at high efficient aqueous arsenic removal by functionalized pressure. Am. Mineral. 85, 514–523. MIL-100(Fe)/rGO/δ-MnO2 ternary composites: Adsorption Hao, L., Liu, M., Wang, N., Li, G., 2018. A critical review on arsenic performance and mechanism. J. Environ. Sci. 91, 22–34. removal from water using iron-based adsorbents. RSC Adv 8, Ravel, B., Newville, M., 2005. ATHENA, ARTEMIS, HEPHAESTUS: 39545–39560. Data analysis for X-ray absorption spectroscopy using IFEFFIT. He, R., Peng, Z., Lyu, H., Huang, H., Nan, Q., Tang, J., 2018. Synthesis J. Synchrotron Radiat. 12, 537–541. and characterization of an iron-impregnated biochar for Shakya, A.K., Ghosh, P.K., 2018. Simultaneous removal of arsenic, aqueous arsenic removal. Sci. Total Environ. 612, 1177–1186. iron and nitrate in an attached growth bioreactor to meet He, X., Li, Y., Yang, C., Lu, L., Nie, Y., Tian, X., 2020. Carbon drinking water standards: Importance of sulphate and empty dots-MnO2nanocomposites for As(iii) detection in bed contact time. J. Clean. Prod. 186, 1011–1020. groundwater with high sensitivity and selectivity. Anal. Simonelli, L., Marini, C., Olszewski, W., Ávila Pérez, M., Methods 12, 5572–5580. Ramanan, N., Guilera, G., et al., 2016. CLÆSS: The hard X-ray Hosseini Asl, S.M., Masomi, M., Hosseini, M., Javadian, H., Ruiz, M., absorption beamline of the ALBA CELLS synchrotron. Cogent Sastre, A.M., 2017. Synthesis of hydrous iron oxide/aluminum Phys 3, 1231987. hydroxide composite loaded on coal fly ash as an effective Singh, S.K., Townsend, T.G., Mazyck, D., Boyer, T.H., 2012. mesoporous and low-cost sorbent for Cr(VI) sorption: Fuzzy Equilibrium and intra-particle diffusion of stabilized landfill logic modeling. Process Saf. Environ. Prot. 107, 153–167. leachate onto micro- and meso-porous activated carbon. Hu, X., Jia, L., Cheng, J., Sun, Z., 2019. Magnetic ordered Water Res 46, 491–499. mesoporous carbon materials for adsorption of minocycline Tran, H.N., You, S.J., Hosseini-Bandegharaei, A., Chao, H.P., 2017. from aqueous solution: Preparation, characterization and Mistakes and inconsistencies regarding adsorption of adsorption mechanism. J. Hazard. Mater. 362, 1–8. contaminants from aqueous solutions: A critical review. Kaljurand, M., Koel, M., 2011. Recent advancements on greening Water Res 120, 88–116. analytical separation. Crit. Rev. Anal. Chem. 41, 2–20. Tran, T.Van, Nguyen, D.T.C., Le, H.T.N., Vo, D.-V.N., Nanda, S., Lima, E.C., Hosseini-Bandegharaei, A., Moreno-Piraján, J.C., Nguyen, T.D., 2020. Optimization, equilibrium, adsorption Anastopoulos, I., 2019. A critical review of the estimation of behavior and role of surface functional groups on graphene the thermodynamic parameters on adsorption equilibria. oxide-based nanocomposite towards diclofenac drug. J. Wrong use of equilibrium constant in the Van’t Hoof equation Environ. Sci. 93, 137–150. for calculation of thermodynamic parameters of adsorption. J. Uheida, A., Salazar-Alvarez, G., Björkman, E., Yu, Z., Mol. Liq. 273, 425–434. Muhammed, M., 2006. Fe3O4 and γ -Fe2O3 nanoparticles for Liu, C., Xie, F., 2021. Highly efficient removal of As(III) by Fe-Mn-Ca the adsorption of Co2+ from aqueous solution. J. Colloid composites with the synergistic effect of oxidation and Interface Sci. 298, 501–507. adsorption. Sci. Total Environ. 777, 145289. van Genuchten, C.M., Behrends, T., Stipp, S.L.S., Dideriksen, K., Liu, C.H., Chuang, Y.H., Chen, T.Y., Tian, Y., Li, H., Wang, M.K., et al., 2020. Achieving arsenic concentrations of <1 μg/L by Fe(0) 2015. Mechanism of Arsenic Adsorption on Magnetite electrolysis: The exceptional performance of magnetite. Nanoparticles from Water: Thermodynamic and Water Res 168, 115170. Spectroscopic Studies. Environ. Sci. Technol. 49, 7726–7734. Zhao, K., Guo, H., 2014. Behavior and mechanism of arsenate Long, X., Wang, X., Guo, X., He, M., 2020. A review of removal adsorption on activated natural siderite: evidences from FTIR technology for antimony in aqueous solution. J. Environ. Sci. and XANES analysis. Environ. Sci. Pollut. Res. 21, 1944–1953. 90, 189–204.