Effective adsorption of antimony (V) from contaminated water by a novel composite manganese oxide/oxyhydroxide as an adsorbent

Antimony (Sb) is a metalloid that has both metallic and nonmetallic properties. Release of Sb into the environment is related to both natural processes (rock weathering, sedimentation, etc.) (Wedepohl 1995) and human activities (mining, smelting, combustion, etc.) (Filella et al. 2002). It is reported that dissolved Sb concentrations in pollution-free systems are usually less than 1 μg/L (Filella et al. 2002). High concentrations of Sb in air, soil and water bodies have been reported near mining areas, smelter plants and other Sb-containing product sites (Scheinost et al. 2006; Foster et al. 2019). For instance, water samples from a mine area (Xikuangshan antimony mine of Hunan Province, China) were found to be contaminated with Sb concentration of up to 6,384 μg/L, which is three orders of magnitude above the natural background (Wang et al. 2011). More attention has been focused on antimony pollution because of the growing awareness of its risk to human health, such as respiratory irritation, pneumoconiosis, antimony spots on the skin and gastrointestinal symptoms (Sundar & Chakravarty 2010). Both the Environmental Protection Agency of the United States (USEPA) (Callahan et al. 1979) and the European Union (Council of the European Communities) (EEC 1976) have included Sb as a priority contaminant. The former even recommended a level of 6 μg/L as the maximum contamination concentration of Sb in drinking water (USEPA 2009).

In natural environments (waters, soil and sediments), the inorganic Sb usually exists in two oxidation states, namely trivalent antimony (Sb(III)) and pentavalent antimony (Sb(V)). It heavily depends on the ambient environmental conditions (Filella & May 2003). Sb(III) is generally present under reducing conditions, where it abundantly exists in the form of the antimonous acid Sb(OH)₃ with pKa = 11.9 (Mitsunobu et al. 2010). Sb(V) exists under oxic environments, where it predominates as the deprotonated form of Sb(OH)₆⁻ (the pH value ranges from 2 to 11) (Filella et al. 2002). The toxicity of Sb can effectively be reduced by the oxidation of Sb(III) to Sb(V) through biogeochemical or chemical oxidation (Chen et al. 2003). However, the oxidation of Sb(III) leads to the release of Sb(V) as the dominant sorbed species in sediment. It proved to be more difficult to remove Sb(V) than Sb(III) by either adsorption or coagulation (Guo et al. 2009). Thus, it becomes an urgent requirement to develop feasible adsorbents with high adsorption capacity, chemical stability and low cost for Sb(V) removal from water.

There have been many studies devoted to the development of effective methods for removing heavy metal or toxic substances from contaminated water, including electrodeposition (Mosivand et al. 2018; Nepel et al. 2019), reverse osmosis (Yang et al. 2017), precipitation/coagulation (Park et al. 2017; Liu et al. 2019), adsorption (Elgarahy et al. 2020; Elwakeel et al. 2021; Shalaby et al. 2021) and membrane filtration (Garba et al. 2019). Among those methods, adsorption is hailed as one of the most potential treatment methods due to its advantageous properties in terms of flexibility, ease of operation, cost-effectiveness and eco-friendliness (Long et al. 2020). The adsorption behaviour of Sb on various adsorbents, such as metal oxides (Thanabalasingam & Pickering 1990; Wang et al. 2019a), inorganic minerals (Qiu et al. 2018), carbon-based materials (Luo et al. 2015a; Jiang et al. 2020) etc., has been elaborated. Manganese (Mn) oxides (or oxyhydroxides), such as manganese dioxide (MnO₂) and manganese oxyhydroxide (MnOOH), can provide a large accessible surface area and potentially serve as an important adsorbent for environmental pollutants. Batch equilibrium experiments combined with various spectroscopic techniques have suggested specific Sb(V) adsorption at the edge sites of δ-MnO₂ by forming a monodentate mononuclear complex (Sun et al. 2018). However, MnO₂ suffers from an inherent limitation of the low removal efficiency due to the low affinity of MnO₂ to Sb. Although Mn-related binary oxides were proved to have high adsorption capacity for the Sb removal in wastewater treatment, Mn oxides mainly accounted for the Sb(III) oxidation rather than the Sb(V) removal (Xu et al. 2011). Therefore, there are still some challenges in improving the adsorption capacity and accelerating the adsorption process of Mn oxides during the Sb(V) removal.

Thus in this paper, a new composite manganese oxide (CMO) was synthesized by a simple hydrothermal synthesis method in our laboratory. It can be expected as a potentially promising adsorbent to remove Sb(V) from contaminated water. The objectives of the study are: (I) to identify the construction of the CMO adsorbent; (II) to determine the adsorption characteristics of the CMO adsorbent by the adsorption isotherms and thermodynamics; (III) to study the influence of different conditions on the Sb(V) adsorption, including pH, coexisting anions, etc.; and (IV) to investigate the Sb(V) adsorption mechanism by the CMO adsorbent.

Manganese powder (Mn, 2N grade), potassium antimonite [KSb(OH)₆], potassium permanganate, manganese sulfate monohydrate, sodium carbonate, sodium bicarbonate, sodium sulfate, sodium nitrate, sodium hydroxide, and ethanol were brought from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China) and all chemical reagents were of analytical grade. All solutions were prepared using deionized water with a resistivity of 18 MΩ cm (Millipore Corp., Billerica, USA). Sb(V) stock solution was obtained by dissolving KSb(OH)₆ and 2 M HCl in an appropriate amount of deionized water.

Potassium permanganate (KMnO₄) and hydrate manganese sulfate (MnSO₄·H₂O) were separately dissolved into deionized water at room temperature in a ratio of 2:3 (substance amount). The potassium permanganate solution was dropwise added to the hydrate manganese sulfate under magnetic stirring. After stirring for 3 h, the resulting homogeneous solution was transferred into an autoclave and subjected to hydrothermal treatment at a constant temperature of 160 °C for 6 h. After the reaction was completed, the suspension was taken out and cooled down to room temperature naturally. The product was centrifuged and repeatedly washed with deionized water to get rid of the remaining ions, and finally dried at 110 °C to obtain nano-manganese dioxide.

The CMO adsorbent was prepared using the following hydrothermal synthesis procedures. The manganese powder was prepared by finely grinding and sieving in advance and then mixed with the synthesized nano-MnO₂ in deionized water. After stirring at room temperature for 1 h, the solution was heated to 65 °C and kept for another 6 h. Subsequently, the resulting mixed precipitate was ultrasonically dispersed for 30 minutes and further filtered, followed by repeatedly washing with deionized water. After drying treatment at 60 °C, the resulting powder was stored under nitrogen protection for further experiments.

Morphologies of the CMO adsorbent were collected by scanning electron microscopy (SEM, Nano SEM-4800, HITACHI, Japan). The elemental analyses were performed using an Energy-dispersive System (EDS). The phase composition in the CMO adsorbent was analysed using an x-ray diffractometer (XRD, Smart Lab 3KW, Japan). The specific surface area and the pore size distribution were determined by the Brunauer–Emmett–Teller (BET, Quadrasorb EVO, USA) N₂ adsorption/desorption method and the Brunauer–Joyner–Hallenda (BJH) method. Both the Fourier transform infrared spectrometer (FTIR, Nicolet iS50, USA) and the X-ray photoelectron spectroscope (XPS, Thermo Escalab 250Xi, USA) were applied to analyse the structures of the samples before and after Sb(V) adsorption.

The adsorption isotherms of Sb(V) were carried out at varying temperatures (i.e., 25, 35 and 45 °C). 0.05 g of CMO adsorbent was added into a conical flask containing 100 mL of Sb(V) solutions with different initial concentrations (ranging from 2 to 100 mg/L). In each test, the solution pH was controlled to 7.0 ± 0.1 by acid-base buffer solution, and the ionic strength was maintained with 100 mg/L Na₂CO₃. The flasks were shaken on a shaker (BS-2E, Changzhou, China) at a certain temperature for 12 h. Then, all the samples were filtered by 0.45 μm membrane and analysed for residual antimony by the AFS-8230 atomic fluorescence spectrophotometer (Beijing Yoshida, China).

In adsorption kinetics experiments, 0.5 g of CMO adsorbent was added into a 1,000 mL solution with an initial Sb(V) concentration of 10 mg/L. 100 mg/L Na₂CO₃ was used to control background ionic strength at a constant level, and the solution pH was controlled to 7.0 ± 0.1 by acid-base buffer solution. The kinetics experiments were conducted at 25 °C with shaking at 160 rpm in a shaker. To determine the remaining antimony, a small aliquot of aqueous samples was taken and filtered through a 0.45 μm membrane at a specific time interval from 5 min to 10 h.

In order to explore the effect of solution pH on Sb adsorption, adsorption experiments were performed at various pH values. 0.05 g of CMO adsorbent was added into 100 mL solution with an initial Sb(V) concentration of 10 mg/L and a constant level of ionic strength. Because the dominant Sb(V) species is Sb(OH)₆⁻ as the pH value ranges from 2.7 to 10.4, the pH in those adsorption experiments was manually controlled in the range of 3.0–10.0 using 0.1 M acid-base buffer solution. The solutions were shaken at 160 rpm for 6 h when the temperature was set at 25 °C.

To investigate the influence of commonly co-existing anions (SO₄²⁻, NO₃⁻, PO₄³⁻ and HCO₃⁻) and humic acid on the adsorption process, 0.05 g of CMO adsorbent was added into 100 mL Sb(V)-containing solution with relevant anion concentrations (ranging from 0 to 500 mg/L). The initial Sb(V) concentration was 10 mg/L, and the pH was controlled to 7.0 ± 0.1 by 0.1 M acid-base buffer solution. The experiments were performed with constantly shaking at 160 rpm for 6 h at a constant temperature (25 °C). Moreover, the blank sample was prepared in a similar way without adding coexisting anion.

The regeneration of adsorbents is an important aspect since it determines whether an adsorbent can be brought into practical application. A total of four consecutive adsorption/desorption of the Sb(V) processes were performed in triplicate to evaluate the reusability of the CMO adsorbent. The saturated adsorbent was respectively soaked in different solutions, including deionized water, 0.1 mol/L HNO₃, 0.1 mol/L citric acid, 0.1 mol/L NaCl and 0.019 g/L EDTA, and then shaken at 160 rpm for 6 h at a specific temperature (three levels, i.e., 25, 35 and 45 °C). After high-speed centrifugation for 10 minutes, the supernatant was taken to determine the content of Sb in solution for calculating the desorption rate. The resulting adsorbents, after being washed with deionized water at least three times, were dried in an oven at 60 °C for use in the next cycle.

As displayed by the XRD in Figure 1(a), there are two dominant manganese oxide(s)/oxyhydroxide(s), i.e., Mn₃O₄ and MnOOH, existing in the as-prepared CMO adsorbent. The strong peaks at 18.01°, 28.89°, 31.01°, 32.31°, 36.08°, 44.45°, 50.71°, 58.51° and 59.84 are attributed to (101), (112), (200), (103), (211), (220), (105), (321) and (224) planes of the structure of Mn₃O₄ (PDF#24-07343), respectively. Other observed peaks at 25.66°, 33.54°, 37.11°, and 64.72° are indicative of MnOOH (PDF#12-0733). The EDS spectrum for determining the specific elemental contents is depicted in Figure 1(d), which indicates the presence of the expected elements in the prepared CMO adsorbent, i.e., manganese and oxygen (hydrogen is not included in EDS). The elemental analysis of the CMO adsorbent identified the weight percentage of manganese and oxygen as 75.16% and 24.84%, respectively. Further structural characterization of the CMO adsorbent was analysed by the SEM-mapping as shown in Figure 1(e) and 1(f). The mapping shapes of the CMO adsorbent regarding Mn and O elements show that the prepared adsorbent possesses a characteristic of uniform element composition and dispersion.

The SEM and transmission electron microscopy (TEM) images were collected to confirm the morphology of the prepared CMO adsorbent, as displayed in Figure 1(b) and 1(c). It reveals that the CMO adsorbent is aggregated, consisting of nanoparticles and nanorods. According to the hydrothermal synthesis results reported by Hu (Hu et al. 2008), the nanoparticles and nanorods can be identified as Mn₃O₄ and MnOOH, respectively. The TEM image shows that Mn₃O₄ nanoparticles have an average size of 30 nm, most of which are nanocubes. Nanorods of MnOOH are formed with diameters of 10–50 nm and lengths of hundreds of nanometres. The large quantity of Mn₃O₄ nanoparticles and MnOOH nanorods in Figure 1(b) further demonstrates success in synthesizing composite oxide via a simple hydrothermal method. The aggregation of numerous Mn₃O₄ nanoparticles on MnOOH nanorods results in the presence of a porous structure, which is advantageous for the target species to adsorb on its surface. The specific surface area (BET) by nitrogen adsorption-desorption measurements was then used to determine the surface properties of CMO adsorbent, and the results are shown in Figure 2. According to the IUPAC classification, the CMO adsorbent exhibits the type-IV isotherm with a type-H3 hysteresis loop when the relative pressure (P/P0) ranges from 0.4 to 1.0. It implies a typical mesoporous structure in the CMO adsorbent. The surface area and pore size of the CMO adsorbent are 30.860 m²/g and 8.7 nm, respectively. It means that the high specific surface areas could provide considerable surface active sites for diffusion and adsorption of adsorbent.

When the pH value is 7.0 ± 0.1, the adsorption isotherms of Sb(V) on CMO adsorbent at 298, 308 and 318 K are depicted in Figure 3. With the increase of initial concentration, there has been a rapid increase in the adsorption amounts of Sb(V) on the CMO adsorbent and then a slow rise tendency. Besides, the adsorption capacity of Sb(V) on the CMO adsorbent exhibits a downward trend as the temperature increases, indicating that the adsorption process is exothermic.

The relevant parameters regarding the Sb(V) isotherms at different temperatures are presented in Table 1. According to the correlation coefficient (R²) values, the Langmuir model should be superior to the Freundlich model. For the Langmuir model, the adsorption site energy is generally considered to be a constant, and the adsorption can reach a maximum when all sorption sites are occupied as a single layer. The theoretical adsorption capacity of CMO obtained by the Langmuir model is 119.63 mg/g at 25 °C.

A comparison of adsorption capacities of other adsorbents for Sb(V) is listed in Table 2. Notably, it is not easy to compare the prepared adsorbents with other adsorbents due to different experimental conditions. The average concentration in actual Sb(V)-contaminated water is usually not too high, and hence, the adsorption capacity of the adsorbent is required to be better at low concentration. The prepared CMO adsorbent exhibits relatively high sorption capacity compared with other adsorbents for Sb(V) at the low concentration. The adsorption of Sb(V) on the CMO adsorbent gradually reached equilibrium in about 200 min. The equilibrium time is considerably faster than other new nanocomposites, such as Mn-coated biochar, which reach equilibrium in about 1,200 min. Compared to adsorbents consisting of expensive ingredients like Co₃O₄@rGO, the CMO adsorbent can be synthesized by cheap and readily available materials, which is very critical for technology popularization. Therefore, the comprehensive features and properties make CMO a promising adsorbent for removing Sb(V) in practical application.

The information regarding the type of adsorption process, whether chemical or physical, can be judged from the E value. The mean energy of the physical adsorption is recognized to be less than 8 kJ mol⁻¹, whereas it is higher than 8 kJ mol⁻¹ for the chemical adsorption (Helfferich 1962). With respect to the Sb(V) adsorption on the CMO adsorbent, the E value is calculated as 10.53 kJ mol⁻¹, meaning that the adsorption of Sb(V) on the CMO adsorbent is a chemical process.

Based on the above equations, the ΔG parameters are determined as −24.33 kJ mol⁻¹, −22.51 kJ mol⁻¹ and −21.42 kJ mol⁻¹ for 298, 308 and 318 K, respectively. The decreasing trend in negative ΔG values with the increase of temperature implies that the degree of spontaneity declines as temperature increases. In the equation, the ΔH⁰ and ΔS⁰ parameters can be derived from the slope and intercept of the ln K versus 1/T plot. The results are listed in Table 3. The negative ΔH value (−32.21 kJ mol⁻¹) further demonstrates that the adsorption process is exothermic, which is in accordance with a decrease in the adsorption capacity with the increasing temperature. The negative ΔS value (−27.83 J mol⁻¹ K⁻¹) implies decreased degrees of freedom of the adsorbed species.

Adsorption kinetics is critical when evaluating the adsorption process. The kinetics of Sb(V) adsorbed on the CMO adsorbent was studied. The result is displayed in Figure 5. It is obvious that the adsorption kinetics is rapid initially within 50 min and then slows down until the adsorption equilibrium. There are considerable adsorption sites available for Sb(V) adsorption at the beginning of the adsorption process. When the adsorption reaches a certain level, it is difficult for the remaining adsorption sites to be occupied due to the repulsive forces between the surface molecules on the CMO adsorbent and adsorbate solution. The adsorption equilibrium is reached after 200 min with the adsorption capacity of nearly 19.80 mg/g.

All the kinetics parameters were determined and listed in Table 4. The correlation coefficient (R² = 0.9960) of the pseudo-second-order model is superior to that (R² = 0.9820) of the pseudo-first-order model. The result suggests that the pseudo-second-order model can fit the equilibrium data more perfectly. Because chemisorption associated with sharing or exchange of electrons between sorbent and sorbate is recognized as the rate-limiting step in the pseudo-second-order kinetic model (Ye et al. 2011), the chemical adsorption, therefore, was the rate-controlling step in the Sb(V) sorption by the CMO adsorbent.

The curve-fitting plot is shown in Figure 5, and the fitting parameters are listed in Table 4. It can be seen that the plot can be separated into three phases, namely a sharp rise portion, a moderate increase and an equilibrium phase. Three different phases in the adsorption process indicate that the adsorption of Sb(V) on the CMO adsorbent is influenced by various independent steps, i.e., liquid film diffusion, intra-particle diffusion and adsorption reaction stage. During the liquid film diffusion, the Sb(V) is transferred to the external surface of the CMO adsorbent. Due to the high concentration of Sb(V) in the solution, there are abundant active sites on the surface of the adsorbent, and the adsorption rate is very high. Therefore, at this stage, Sb(V) diffusion is mainly limited by the pore structure of the CMO adsorbent. In the last equilibrium stage, the diffusion is primarily controlled by the Sb(V) already attached to the adsorbent, and the equilibrium adsorption of the CMO adsorbent is almost achieved.

The effect of initial solution pH on the Sb(V) adsorption was specifically investigated by varying pH from 3 to 10, as displayed in Figure 6. It can be seen that the adsorption of Sb(V) on the CMO adsorbent is heavily dependent on pH values. The maximum adsorption is obtained when the pH is less than 7.0. When the pH is higher than 7.0, the adsorption capacity drops dramatically. It can be attributed to a change in the positive charge on the surface of the adsorbent. When pH < pH_pzc (zero point charge), the protonation of the adsorbent surface increases, leading to the increase of positive charge. Thus, the electrostatic attraction between Sb(OH)₆⁻ and the charged surfaces is enhanced. When pH > pH_pzc, a more negative charge on the adsorbent surface enhanced the electrostatic repulsion forces between Sb(OH)₆⁻ anions and charged surface, resulting in a decrease in the adsorption capacity. The results indicate that Sb(V) adsorption on the CMO adsorption is relatively efficient over a wide range of pH values (from 3.0 to 7.0), suggesting that the prepared CMO adsorbent has a great potential of application in natural water systems.

Besides, the leaching of the Mn element from the CMO adsorbent at different pH values was studied and displayed in Figure 6. The concentrations of Mn released in solutions change from 0.06 to 1.6 mg/L. According to the limits of Mn at 2.0 mg/L in the Discharge Standard of Pollutants for Municipal Wastewater Treatment Plant, the release of Mn element is negligible over a wide range of pH (from 3 to 10). It suggests that the prepared adsorbent exhibits favourable physical and chemical stability. In this regard, the as-prepared CMO adsorbent has the potential for treating Sb(V)-containing water due to its stable structure and good security.

The effects of commonly present anions (such as NO₃⁻, HCO₃⁻, SO₄²⁻ and PO₄³⁻) and humic acid on the adsorption of Sb(V) by the CMO adsorbent were investigated at different concentrations (ranging from 0 to 500 mg/L). The results are described in Figure 7. The Sb(V) removal efficiency varies slightly with the anion concentration (NO₃⁻, HCO₃⁻, and SO₄²⁻) and humic acid increasing from 0 to 500 mg/L. It indicates that these coexisting components do not greatly influence the Sb(V) adsorption. It has been reported that the adsorption involving the out-sphere surface complexes is highly sensitive to ion strength, since it was inhibited by weakly adsorbing anions through electrostatic competition (Mcbride 1997). On the contrary, the adsorption associated with inner-sphere surface complexes showed little sensitivity to ionic strength. In this respect, it suggests that the Sb(V) might be chemically adsorbed on the CMO adsorbent by forming inner-sphere surface complexes at the water/oxide interface. It is worth noting that the Sb(V) removal efficiency decreases sharply with the increasing concentration of PO₄³⁻. As the concentration of PO₄³⁻ varies from 0 to 500 mg/L, the removal efficiency reduces to 30.8%. A similar interfering effect of phosphate on arsenic removal has been shown (Gupta et al. 2009). The significant decrease of Sb(V) removal efficiency in the presence of phosphate can be interpreted by the strong competition between phosphate and Sb(V) for adsorbent binding sites.

The regeneration of the CMO adsorbent adsorbed with Sb(V) via various treatments (H₂O, HNO₃, citric acid, NaCl and EDTA) at different temperatures (25, 35 and 45 °C) has been evaluated, and the comparison results are displayed in Figure 8. It can be seen that the desorption capacity of EDTA is superior to others. The reason is that the amino and carboxyl groups in EDTA can form more stable complexes with Sb(V), and as a result, the adsorbed Sb(V) can be desorbed from the CMO adsorbent. In addition, it is worthwhile mentioning that the desorption capacities for all treatments gradually increase when the temperature varies from 25 °C to 45 °C, suggesting that the desorption process could be endothermic. Five adsorption-regeneration cycles for EDTA treatment were carried out, as depicted in the inset in Figure 8. Generally, the adsorption capacity of the adsorbent tends downwards with an increase in the regeneration cycle. However, it shows an unremarkable decrease for the CMO adsorbent, and the reduction in Sb(V) removal efficiency was only 7.0% even after the fifth regeneration. These results indicate that the prepared CMO adsorbent could be easily regenerated via EDTA treatment and used repeatedly. For comparison, reusability performances for several reported adsorbents are listed in Table 2. It can be seen that the prepared CMO adsorbent is comparable to other reported adsorbents.

For application, a self-made model device with a plexiglass column (350.0 mm × 50.0 mm) was built for Sb(V)-contaminated water treatment, as shown in Figure 9(a). The plexiglass column was filled with a small portion of gravel, a 100 mm column height of CMO adsorbent, and an appropriate amount of sponge as the bottom, middle, and upper layer, respectively. A certain amount of quartz sand was finally compacted at the top. When pumping Sb(V)-containing water into the column via a peristaltic pump, water will pass through the packing column from the bottom to the top. In the simulation experiment, Sb(V)-containing water with different concentration levels was continuously treated by this device with a flow rate of 6.5 mL/min. Figure 9(b) shows that 6.24 L of Sb(V)-containing water with a concentration of 3.6 mg/L can be treated with the Sb(V) removal efficiency of more than 90% via this device. As for the Sb(V) initial concentration of 0.5 mg/L, the removal volume of Sb(V)-containing water is up to 19.2 L. It indicates that the as-prepared CMO adsorbent has promising application potential in Sb(V)-pollutant water treatment.

FTIR and XPS are generally recognized as powerful techniques in investigating surface interactions between adsorbates and adsorbents. The interactions between Sb(V) and the functional groups on the CMO adsorbent can be determined by the FTIR and XPS spectra, such as the characteristic peak shift and intensity change.

The FTIR spectra of the original CMO and after adsorbing Sb(V) were respectively collected and shown in Figure 10. The peaks at 3,435 cm⁻¹ and 1,630 cm⁻¹ in spectra correspond to the hydroxyl groups of physisorbed water molecules on the adsorbents. Three peaks at 626, 496 and 415 cm⁻¹ are associated with the vibration of Mn-O in Mn₃O₄, which is in agreement with previous FTIR studies of Mn₃O₄ (Ishii et al. 1972; Zhang et al. 2004). The peaks at 1,150, 1,120 and 1,084 cm⁻¹ are attributed to the typical OH-bending modes of MnOOH: δ-1-OH, δ-2-OH and γ-OH, respectively (Zhang et al. 2004). The peaks related to bending vibration of hydroxyl groups on the adsorbent surface in wavenumber ranging from 900 to 1,200 cm⁻¹ (Zhang et al. 2018) are not obvious because of the broad overlapping peaks in the region. But, the decreased intensity in the 900–1,200 cm⁻¹ region of the spectrum after Sb(V) adsorption can be observed, indicating that the adsorption process involves cleavage of hydroxyl groups. It suggests that the hydroxyl groups on the adsorbent surface engage in Sb(V) sorption, and the hydroxyl groups in MnOOH molecules also possibly participate in the sorption process. Besides, a new peak appears at 455 cm⁻¹ after adsorbing Sb(V), which can be attributed to the Sb(V)-O stretching vibration (Xu et al. 2011). Information regarding weakening and emerging peaks is indicative of the adsorption of Sb(V) on adsorbents by the formation of an inner sphere, as has been previously observed in studies of Sb(V) adsorption on MnO₂ (Liu et al. 2015).

To obtain further insight into the Sb(V) sorption on the CMO adsorbent, the chemical compositions of CMO before and after Sb(V) adsorption were measured by XPS. The O 1s, Sb 3d and Mn 2p spectra are illustrated in Figure 11. In the Mn 2p spectra, two peaks positioned at 530.6 eV and 539.9 eV are ascribed to Sb 3d5/2 and Sb 3d3/2, respectively (Luo et al. 2015a). The appearance of the Sb peak provides an indicator to determine that Sb(V) is adsorbed on the CMO adsorbent. The O 1 s spectrum of the CMO adsorbent in Figure 11(a) consists of three peaks positioned at the binding energy of 529.85 eV, 531.2 eV, and 531.9 eV, which can be assigned to metal oxide (M-O), hydroxyl bonded to metal (M-OH), and adsorbed H₂O in the adsorbent (H₂O), respectively (Luo et al. 2015a). It shows that the area ratio for the peak at 531.1 eV (M-OH) decreases from 37.20% to 24.31% after Sb(V) adsorption. However, the area ratio of the peak at 529.85 eV (M-O) increases from 59.42% to 68.34%. The significant decrease of M-OH suggests that the M-OH groups on the adsorbent surface play an important role in the Sb(V) adsorption process. An increase in the area ratio for the peak at 529.85 eV suggests that M-O groups are forming on the CMO adsorbent surface.

The Mn 2p spectra of the original CMO and after Sb(V) sorption are displayed in Figure 11(b). The Mn 2p_3/2 peak and the Mn 2p_1/2 peak are respectively positioned at 641.7 eV and 653.5 eV, which is in accordance with the spectra of Mn oxides (Ardizzone et al. 1998). The spectrum for Mn 2p can be divided into three pairs of doublets, i.e., 640.8 eV, 641.79 eV, and 642.99 eV. The peak at 640.8 eV can be ascribed to the Mn 2p_3/2 of MnO, the peak at 641.79 eV can be assigned to the Mn 2p_3/2 of MnOOH, and the peak at 642.99 eV can be attributed to the Mn 2p_3/2 of MnO₂. The area ratio of the peak at 640.8 eV versus the peak at 642.99 eV is about 1.6:1, i.e., a molar ratio of 2:1 for MnO versus MnO₂, which accords well with the theoretical value of Mn₃O₄ (i.e., 2MnO MnO₂) (An et al. 2008). According to the XRD spectra, it can be further demonstrated that the CMO adsorbent consists of MnOOH and Mn₃O₄, which is very well in agreement with the FTIR results. After Sb(V) sorption, the Mn 2p_3/2 spectra showed a minor shift to the lower binding energy. Previous studies of As(V) sorption found that the peak corresponding to Ti 2p_1/2 showed a shift to lower binding energy after As(V) sorption, recommending the formation of Ti-O-As (Li et al. 2010). Therefore, it can be deduced that the formation of Mn-O-Sb on the adsorbent surface leads to the Mn-O groups at low binding energies.

The FTIR and XPS analysis as described above suggested that the hydroxyl groups, including on the CMO adsorbent surface and in the MnOOH molecules, played an important role in Sb(V) sorption. The protonation of the hydroxyl groups on the adsorbent and in the MnOOH molecules led to the CMO adsorbent being filled with positive sites in acidic conditions. This promoted absorption of the negative Sb(OH)₆⁻ on the CMO adsorbent. It was consistent with the pH-dependent sorption results in Figure 6, showing a high adsorption capacity of Sb(V) in acidic or neutral conditions and a clear reduction of adsorption capacity in alkaline conditions. In Figure 7, the Sb(V) removal efficiency varied slightly with the anion concentration, implying the formation of inner-sphere complexes on the adsorbent. The FTIR analysis in Figure 10 demonstrated that the Sb(V) sorption process was heavily related to the Mn-OH groups on the adsorbent surface. The XPS analysis in Figure 11 displayed a reduction of M-OH and an increase of M-O on the adsorbent surface. It deduced that a ligand exchange existing between hydroxyl groups in Sb(OH)₆⁻ molecules and hydroxyl groups on the adsorbent surface led to the formation of Sb-O-Mn bonds and the release of water molecules. Previous studies have indicated that Sb(V) was adsorbed on metal oxides and oxyhydroxides by the formation of monodentate and bidentate complexes (Sun et al. 2018). Previous quantification calculation from the XPS analysis suggested that the stoichiometric ratios of surface hydroxyl between the original adsorbents and target saturated adsorbents were 1:2 for a monodentate surface complex and 2:1 for a bidentate one (Zhang et al. 2005). In Figure 10, the ratio value between the surface hydroxyl of CMO and Sb(V)-loaded CMO was 1.53 after Sb(V) adsorption, meaning that both monodentate and bidentate complexes may be involved on the surface of CMO adsorbent in the adsorption of Sb(V).

A new CMO adsorbent was synthesized by using a simple hydrothermal synthesis method. The synthetic CMO sorbent was mainly an aggregate of Mn₃O₄ nanoparticles and MnOOH nanorods, resulting in the presence of a porous structure to favour the adsorption of target species on its surface. The CMO adsorbent could effectively remove Sb(V). The maximal adsorption capacity of Sb(V) reached 119.63 mg/g at pH 7.0 ± 0.1. Sb(V) adsorption was heavily pH-dependent, since it showed a reduction with increasing pH. The adsorption behaviour of Sb(V) could be well described by the Langmuir isotherm and the pseudo-second-order models. The CMO adsorbent had a good anti-interference ability for coexisting ions and excellent reusability. The hydroxyl groups on the adsorbent and in the MnOOH molecules were responsible for the Sb(V) sorption. Both inner-sphere monodentate and bidentate complexes were formed on the surface of CMO adsorbent in the adsorption of Sb(V).

Not applicable.

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

The manuscript is approved by all authors for publication.

The authors declare that they have no competing interests.

This work was supported by the National Natural Science Foundation of China (No. 21806034 and No. 52070078), the Natural Science Foundation of Hunan Province (No. 2020JJ5126), and the Scientific Research Program of Hunan Education Department (No. 19C0603 and No. 20C0644).

All relevant data are included in the paper or its Supplementary Information.