Removal of manganese by adsorption onto newly synthesized TiO2-based adsorbent during drinking water treatment

ABSTRACT Manganese is naturally present in water, but its increased concentration in potable water is undesirable for multiple reasons. This study investigates an alternative method of demanganization by a newly synthesized TiO2-based adsorbent prepared through the transformation of titanyl sulphate monohydrate to amorphous sodium titanate. Its adsorption capacity for Mn2+ was determined, while a range of influential factors, such as the effect of contact time, adsorbent dosage, pH value, and added ions was evaluated. The adsorbent appeared highly effective for Mn2+ removal owing to its unique characteristics. Besides adsorption via electrostatic interactions, ion-exchange was also involved in the Mn2+ removal. Although the Mn2+ removal occurred within the whole investigated pH range of 4–8, the maximum was achieved at pH 7, with qe  = 73.83 mg g-1. Equilibrium data revealed a good correlation with Langmuir isotherm in the absence of any ions or in the presence of monovalent co-existing ions, while the results in the presence of divalent co-existing ions showed a better fit to Freundlich isotherm. Additionally, the presence of monovalent cations (Na+, K+) only slightly decreased the Mn2+ removal efficiency as compared to divalent cations (Ca2+, Mg2+) that caused a greater decrease; however, the effect of anions (Cl-, SO4 2-) was insignificant. To provide insight into the adsorbent safety, the toxicity assessment was performed and showed no harmful effect on cell activity. Furthermore, the residual concentration of titanium after adsorption was always below the detection limit. The results imply that the synthesized TiO2-based adsorbent is a safe promising alternative method for demanganization. Highlights The synthesis of amorphous TiO2-based adsorbent was presented. The TiO2-based adsorbent was found to be efficient for Mn2+ removal. The Mn2+ removal mechanisms were adsorption and ion-exchange. Increasing pH enhanced the efficiency of Mn2+ removal. Divalent cations decreased the Mn2+ removal efficiency more than monovalent cations. GRAPHICAL ABSTRACT


Introduction
Manganese is a natural element commonly occurring in many water resources. The concentration of Mn in the surface water is on the order of hundredths or tenths of mg L −1 , while in groundwater, it can be by orders of magnitude higher and reach units of mg L −1 . However, its presence in drinking water is undesirable and causes several operational and aesthetic problems and health risks. Concentrations of Mn 2+ in water > 0.1 mg L −1 negatively affect its organoleptic properties, especially taste, odour, and colour [1]. Furthermore, Mn 2+ ions are considered neurotoxic agents, and long-term environmental exposure is a supposed accelerator of the onset of Parkinson's disease [2]. Recent epidemiological studies concerning the consumption of water with Mn 2+ concentrations on the order of 50-100 µg L −1 reported significant cognitive deficits among children [3]. From the operational point of view, Mn 2+ causes incrustations in pipelines at concentrations > 0.02 mg L −1 [4] and is an undesirable nutrient substrate for the development of bacteria in water distribution networks [5]. From these findings, it is clear that maintaining acceptable concentrations of this element in drinking water is of great importance. The World Health Organization determined the maximum acceptable limit of Mn concentration in drinking water of 0.1 mg L −1 [6]; however, many countries follow the stricter limit according to the Environmental Protection Agency that recommends the maximum Mn concentration in drinking water of 0.05 mg L −1 [7].
Conventional methods used for demanganization are based on the conversion of soluble Mn 2+ to insoluble removable precipitates. In theory, the oxidation of Mn 2+ can be achieved by atmospheric oxygen via mechanical aeration of water, but the oxidation rate is very slow, and pH-dependent and alkalization is necessary to enhance the process [8]. When using oxidizing agents such as potassium permanganate or ozone, Mn 2+ oxidation can be achieved even at pH of approximately neutral; however, these are associated with high operational costs. They may also be consumed for the oxidation of other compounds present in water, which increases oxidant demand [9]. Another possibility is so-called contact filtration, where Mn 2+ is catalytically oxidized and trapped on the surface of a prepared filter cartridge coated with manganese dioxide [10]. There are disadvantages of this method, such as the higher amount of sludge and clogging of sand filters. For the above-described reasons, it would be advantageous to develop an alternative method of removing manganese that would be less dependent on pH value, more economical, and technologically easier [11].
Among other methods of Mn 2+ removal are adsorption or cation-exchange. Many materials have been proposed for this purpose, such as activated carbon [12,13], oxidized multiwalled carbon nanotubes [14], biomatrixes prepared from various lignocellulosic wastes from agro-industry [15], zeolites [16] or modified multivalent metal hydroxides [17,18]. These adsorbents are more or less efficient for Mn 2+ removal from aqueous solution, but the required dose is mostly very high, which makes these methods expensive. In the case of adsorbents made of waste products, the costs could be reduced. On the other hand, their application in drinking water treatment can raise concerns, particularly if the adsorbent contains undesirable elements (e.g. NH 4 + ) [19], which might be released into the treated water.
Recently, more attention has been focused on adsorbents based on TiO 2 . The outstanding properties of TiO 2 -based adsorbents are, for instance, large surface area, high cation-exchange capacity, and high density of functional hydroxyl groups on the adsorbent surface [20,21]. Several variants of TiO 2 -based adsorbents have already been investigated for the removal of natural organic matter [22,23], humic acids [24,25], or synthetic organic matter [26]. Particular attention has then been paid to the use of TiO 2 -based adsorbent for the high-efficiency removal of metals [19,27] or radionuclides [28]. Nevertheless, to the best of our knowledge, the efficiency and usability of TiO 2 -based adsorbents for Mn 2+ removal during drinking water treatment have not yet been studied. Various sorbents prepared by hydrolysis of aqueous solutions of titanium salts [29] or titanium tetrachloride [19] lead to the formation of colloids of hydrated TiO 2 nanoparticles that are difficult to separate from the aqueous phase [30]. In addition, the corresponding synthetic procedures are difficult to control and result in products with variable properties, which complicates their practical use. In this study, the process for the synthesis of an amorphous alkaline titanate sorbent by reacting a solid titanyl sulphate monohydrate in an aqueous medium by reaction with an alkaline solution was developed. Under the reaction conditions, SO 4 2anions were extracted from the solid phase, leaving the Ti-O framework intact to give solid amorphous titanate with excellent sorption capacity, e.g. for heavy metals or selected radionuclides [27,28,31]; therefore, it is expected to be effective for Mn 2+ removal as well.
The main aims of this research were to (i) characterize the novel TiO 2 -based adsorbent; (ii) assess its efficiency for Mn 2+ removal and describe the removal mechanisms; and (iii) investigate the effects of solution conditions such as pH and the presence of other cations and anions on Mn 2+ removal. To the best of our knowledge, this work is the first to investigate the usability of a TiO 2based adsorbent for Mn 2+ removal as an alternative method for demanganization during drinking water treatment.

Synthesis of TiO 2 -based adsorbent
Based on alkaline-controlled hydrolysis, sodium titanium oxide was prepared according to the following procedure [31]: 100 mL of cooled distilled water was mixed with ice and 0.14 mol NaOH in the form of a solution. After adding the solid titanyl sulphate (monohydrate, provided by local supplier PRECHEZA a. s.), the suspension had a temperature of 0°C and pH of 10. While the mixture was magnetically stirred for 2 h, its temperature rose to room temperature. Then, the solid residue was decanted twice and filtered off. The solid product was dried at room temperature.

Characterization of TiO 2 -based adsorbent
The total content of alkali metal in the prepared material was determined by dissolving 0.1 g of the sample in 50 mL of concentrated HNO 3 under heating. The solution was then analysed via atomic absorption spectroscopy (Varian AA240FS, LabX, Canada) in absorption mode at a wavelength of 589.0 nm and a slit width of 0.5 nm using a hollow cathode lamp in air/acetylene gas. The total content of the alkali metal cation was 133.8 mg L −1 .
Structural and morphology characterizations of the samples were observed by scanning electron microscopy (SEM). A JEOL JSM-6510 microscope (Wcathode, 20 nm resolution at 1 kV) was used for initial observation. Chemical analysis and mapping were performed using the attached energy-dispersive X-ray spectroscope (EDS) analyser from Oxford Instruments. Measurements were performed on native samples without coating. Semiquantitative EDS analyses were measured on compressed tablets.
Detailed phase analysis, including imaging and electron diffraction, was carried out on a JEOL JEM 3010 transmission electron microscope (TEM) operated at 300 kV (LaB6, cathode, point resolution 1.7 Å). Images were recorded on a Gatan CCD camera with a resolution of 1024 × 1024 pixels using the Digital Micrograph software package. The powder samples were dispersed in ethanol, and the suspension was ultrasonicated for 2 min. A drop of the very dilute suspension was placed on a holey carbon-coated Cu grid and allowed to dry by evaporation at ambient temperature.
X-ray diffraction patterns were collected with a PANalytical X'Pert PRO diffractometer equipped with a conventional X-ray tube (CuK radiation, 40 kV, 30 mA, point focus) and a multichannel detector X'Celerator with an anti-scatter shield. The sample was measured in transmission mode.
The surface area was determined by the BET method or V-t method using a Quantachrome Nova 4200e instrument. Nitrogen adsorption was carried out at -196°C. Before analysis, the non-annealed samples were pretreated at RT under vacuum for 35 h.
The dependence of zeta potential on pH was obtained using a Zetasizer Nano ZS (Malvern). Then, 0.1 g of the sample and 100 mL of water were prepared, and an adequate portion was dispersed by ultrasonication and injected into the electrophoretic cell. Zeta potential values were recorded in the pH range from 2 to 10. The pH adjustments were made with solutions of 0.25 and 0.025 M HCl and 0.25 M NaOH. The zeta potential was measured three times for each pH, and the average values are presented.

Preparation of Mn 2+ aqueous solution
An aqueous solution of Mn 2+ ions was prepared by dissolving the desired amount of Mn(NO 3 ) 2 ·4 H 2 O in ultrapure water (GW 65, Goldman Water, CZ) with the total alkalinity adjusted to 1.5 mmol L −1 (a value common for natural raw water) by 0.125 M NaHCO 3 . For purposes of investigating the pH effect on Mn 2+ adsorption, the pH of model water was adjusted to pH 4-9 by using 0.1 M or 1 M NaOH and by 0.1 M or 1 M HCl, respectively. To investigate the effect of cations (Na + , K + , Ca 2+ , Mg 2+ ) and anions (Cl -, SO 4 2-) typically present in natural water, NaCl, KCl, K 2 SO 4 , CaCl 2 ·2 H 2 O, MgCl 2 , or MgSO 4 ·7 H 2 O were added to Mn 2+ solutions to attain amounts equal to 1 mmol L −1 . All chemicals were purchased from Sigma-Aldrich, USA.

Adsorption experiments and data evaluation
Equilibrium batch adsorption experiments were performed to evaluate the effectiveness of TiO 2 -based adsorbent for Mn 2+ removal and determine the effect of the solution pH value on this process. For this purpose, Mn 2+ concentration range of 0.1-5 mg L −1 was selected because these concentrations typically occur in natural waters. Solutions of 250 mL with known initial concentrations of Mn 2+ were then adjusted to the desired pH of 4-9, and 40 mg L −1 TiO 2 -based material was added. The samples were placed in 250 mL borosilicate glass flasks and shaken on a magnetic stirrer (130 rpm) at room temperature (22 ± 0.5°C ) for 24 h. The applied time interval of 24 h sufficient to reach adsorption equilibrium was predetermined by the kinetic adsorption experiments performed for all initial concentrations of Mn 2+ with an adsorbent dosage of 40 mg L −1 and with predetermined optimal adsorbent dosage. After this time, the solutions were filtered through a 0.22 µm membrane filter (Millipore, USA) to remove TiO 2 -based adsorbent particles and analysed for residual Mn 2+ concentration. The residual concentration of titanium was also measured. For each initial concentration of Mn 2+ and pH value, control samples without the TiO 2 -based adsorbent were also processed to reveal potential adsorbate loss during the experiments. Similarly, experiments investigating the effect of coexisting ions on the adsorption of Mn 2+ were conducted at pH 7 with samples prepared as described above. Each experiment was repeated three times.
The amount of Mn adsorbed per unit mass of adsorbent at equilibrium (q e , mg g −1 ) was calculated as follows (1): where C 0 and C e are the initial and equilibrium solution concentrations of Mn (mg L −1 ), respectively, V is the solution volume (L), and m is the mass of the adsorbent (g). The data obtained from the adsorption isotherm experiments were fitted to the Langmuir (2) and Freundlich (3) models given by the adapted equations as follows (2,3): and where q e (mg g −1 ) and C e (mg L −1 ) represent adsorbate uptake and solution concentration at equilibrium, respectively. Parameters a m (mg g −1 ) and K f ((mg g −1 ) (L mg −1 ) 1/n ) are reflective of adsorption capacity; constants b (L mg −1 ) and 1/n represent the surface affinity and the heterogeneity of surface site energy distribution, respectively. The amount of Mn adsorbed at each time interval per unit mass of adsorbent, (qt, mg g −1 ), was calculated as follows (4): where C 0 is the initial solution concentration of Mn (mg L −1 ), C t is the solution concentration of Mn at a specific time t (h), V is the volume of the solution (L) and m is the mass of the adsorbent (g).

Analytical methods
The concentrations of Mn 2+ and coexisting cations (Na + , K + , Ca 2+ , Mg 2+ ) were analysed before and after adsorption and were measured by inductively coupled plasma optical emission spectrometry (ICP OES, Optima 2000 DV, Perkin-Elmer, USA). The concentration of coexisting anions (Cl -, SO 4 2-) was measured by the argentometric Mohr method with a standard solution of silver nitrate and potassium chromate as an indicator (analysis of Cl -) and gravimetric analysis with barium chloride (analysis of SO 4 2-). Measurements of all samples were carried out in triplicate, and errors were less than 3%. Due to unclear effects and suspicion of titanium toxicity, its residual concentration in the solution was measured in all samples after adsorption by ICP OES as well.

TiO 2 -based adsorbent characterization
The starting material titanyl sulphate monohydrate is formed by aggregates of isometric crystals with a broad size distribution, and according to EDS analysis, it contains Ti, S, and O in a ratio corresponding to the value expected from the chemical formula. The crystals of titanyl sulphate remained unchanged throughout the preparation process, and after immersion in an aqueous solution of sodium hydroxide, they provided material composed of aggregates of irregular planar crystals of size 1-2 µm (Figure 1(a-b)). The elemental composition of the final product involves (average of 3 measurements) 64.94 wt. % of oxygen, 6.77 wt. % of sodium, and 28.29 wt. % of titanium (Figure 1(c)). No impurities were found in the concentration detectable by the EDS method (approximately ≥ 0.01 wt. %), which is beneficial because the adsorbent application in drinking water treatment could raise concerns if it contained any undesirable elements (e.g. NH 4 + ) [19].
Moreover, the toxicity assessment of the prepared adsorbent was conducted and confirmed its harmlessnesssee details in section S1, Supplementary Data (SD); the results are depicted in Fig. S1, SD. For a detailed investigation of the prepared sample, TEM was applied; individual crystals were observed (Figure 2(a)). In the selected area electron diffraction patterns, only wide indefinite rings were found (inset Figure 2(a)), implying that the prepared sample is amorphous. The powder XRD pattern shows the amorphous character of the adsorbent (Figure 2(b)). After subtracting the sample holder (Mylar foil) from the measured data, only very broad maxima can be seen centered at 30, 45, and 62°2 Θ CuKα.
The surface area and porosity measurements showed the purely mesoporous character of the prepared adsorbent with a BET surface area (S BET ) of 2.8 m 2 g −1 , and the pore volume was 0.008 cm 3 g −1 , which is in good agreement with data observed for similar adsorbents [31]. The average pore diameter in mesoporous parts was 3.5 nm.
The resulting product of the synthesis has the original morphological characteristics (of the starting titanyl sulphate) well preserved while maintaining the high surface area and porosity typical for nanoparticles. The TiO 2 -based adsorbent is mechanically highly resistant, it has no tendency to form colloids, and does not disintegrate in water even after a long time therefore it is easily separable from the aqueous environment.
Surfaces of metal oxides usually show a positive charge at low pH and a negative charge at high pH with a point of zero charge (PZC) in between [19]. This property is in keeping with the observations where the surface charge of the synthesized adsorbent was analysed in the pH range of 2-10 ( Figure 3). The results showed that the zeta potential significantly changes with increasing pH from 2 to approximately 6 and then becomes stable at pH values in the rough pH interval 6-9. The zero zeta potential is at the value of pH PZC = 3.42. This value is significantly lower than the pH PZC of similar TiO 2 -based adsorbents in published papers [19,23].  Fig. S2, SD, the kinetic models fitting is shown in Fig. S3, SD, and the calculated kinetic parameters are shown in Table S1, SD, respectively. Kinetic adsorption experiments have shown fast initial uptake of Mn 2+ from water, especially in the first 5 min, during which more than 20% of the initial Mn 2+ concentration was removed. After 1 h more than 50% of Mn 2+ was removed. The required time to attain the adsorption equilibrium was experimentally found to be approximately 8 h. Nevertheless, the time of 24 h was selected for the following experiments so as to assure that the adsorption equilibrium will be reached under all investigated conditions. The effect of TiO 2 -based adsorbent dosage on the Mn 2+ removal efficiency was investigated for each initial concentration of Mn 2+ (0.1-5 mg L −1 ) at pH 7. At the same time, detailed results on the dose optimization are depicted in Fig. S4, SD. It was experimentally found that the optimal adsorbent dosage for the initial Mn

The influence of pH value
The adsorption isotherms of Mn 2+ adsorption onto TiO 2based adsorbent under different pH conditions (pH 4-7) are presented in Figure 4. The equilibrium data were analysed using Langmuir and Freundlich model. The Langmuir isotherm model was used to predict Mn 2+ adsorption onto TiO 2 -based adsorbent under different pH based on the R 2 values higher obtained using this model compared to the Freundlich model. The calculated Langmuir and Freundlich model parameters are summarized in Table S2 Section A, SD.
The results of blank experiments performed without the adsorbent revealed that all the concentrations of Mn 2+ precipitated at pH 9, and the concentrations of Mn 2+ above 1.5 mg L −1 also precipitated at pH 8, while these observations are in good agreement with the theoretical speciation of Mn according to hydrochemistry software Aqion 7.4.2. Due to this fact, adsorption experiments were not performed at these values. By contrast, under the experimental conditions at which the adsorption was investigated, Mn is anticipated to be present solely as Mn 2+ , and the occurrence of higher oxidation states and/or Mn precipitation can be therefore neglected.
The results of batch adsorption experiments have shown that the adsorption efficiency of Mn 2+ onto the  TiO 2 -based adsorbent increases with increasing pH, which is associated with the charge properties of the adsorbent. The maximum removal efficiency was reached at pH 7, whereas the difference between pH 7 and 8 was completely negligible. The steepest increase in adsorption efficiency was observed from pH 4 to pH 6 because of the steep decrease in zeta potential; on the other hand, the insignificant difference in adsorption efficiency at pH 7 and 8 was probably caused by the almost constant value of zeta potential under these conditions ( Figure 3). Similar to our study, Liu et al. [32] observed a steep increase in adsorption efficiency from pH 2 to pH 4 because of electrical inversion from positive to negative adsorbent surface charge.
Enhanced Mn 2+ adsorption at higher pH values was also previously observed, e.g. when using carbonbased adsorbents [33]. Generally, an adsorbent in an aqueous solution carries a surface charge that is more or less pH-dependent. Cation adsorption is typically favoured at pH values above the adsorbent´s pH PZC owing to the attractive electrostatic interactions between the negatively charged adsorbent surface and the cations [33,34]. In this study, the pH PZC of the TiO 2 -based adsorbent was 3.42; thus, a negative charge of the adsorbent prevailed within the whole investigated pH range, enabling interactions with cations. However, as mentioned above, the zeta potential further varied with increasing pH value. With regard to TiO 2 -based adsorbents, Liu et al. [32] investigated the removal of metals (Cu 2+ and Cd 2+ ) by titanate nanomaterials at pH 2-6 and confirmed higher removal efficiency at higher pH values. A higher removal efficiency with increasing pH from 2-8 was also observed in the study by Motlochova et al. [27], where different alkali metals (Pb, Cu, and Cd) were removed using different types of TiO 2 -based adsorbents was investigated. A similar trend was observed by Sounthararajah et al. [35], who focused on metal (Ni, Zn, Cd, Cu, and Pb) adsorption onto sodium titanate nanofibrous material. In addition to the increasing negative adsorbent surface charge, it was proposed that the formation of metal hydroxyl complexes that have a higher affinity for adsorption leads to an abrupt increase in metal removal [35]. The proceeding hydrolysis was also suggested to be involved in Mn 2+ removal via adsorption [35].
In our study, the maximum Mn 2+ adsorption was reached at pH 7 with q e = 73.83 mg g −1 . The q e value was then lower by approximately 20%, 40%, and 75% at pH 6 (q e = 60.22 mg g −1 ), pH 5 (q e = 43.66 mg g −1 ), and pH 4 (q e = 18.16 mg g −1 ), respectively. In the study by Kanna et al. [19], which compared the adsorption efficiency of hydrated amorphous TiO 2 and commercially available TiO 2 for Mn 2+ removal at pH 7, the maximum adsorption efficiency (q e = 24.92 mg g −1 ) was almost three times lower than that in our study, despite the higher value of S BET = 449 m 2 g −1 . This could be explained by other factors, e.g. different surface charge (pH PZC ), etc.
On the other hand, higher removal efficiency in comparison with our study was reached in the study by Niksirat et al. [13], where activated carbon prepared from tire residuals was used for Mn 2+ removal. The higher adsorbed amount of Mn 2+ (q e = 120 mg g −1 ) at pH 7 was probably caused by the interplay of several adsorbent parameters (high S BET = 550 m 2 g −1 , very low pH PZC = 2.7, and high total pore volume = 1.22 cm 3 g −1 ) [13].
As mentioned above, a significant disadvantage of some conventional oxidation-based demanganization methods is the need for an increased pH value [11]. By contrast, Mn 2+ adsorption onto the TiO 2 -based adsorbent utilized in this study resulted in significant Mn 2+ removal throughout the entire investigated pH range. Even at pH 4, the removal efficiency was up to 80% in the case of low initial Mn 2+ concentrations (0.1-0.5 mg L −1 ). Additionally, the efficiency under pH values typical for natural waters (approximately 6.5-7.5) reached 90-99%. For high initial Mn 2+ concentrations (3-5 mg L −1 ), the observed removal efficiency was much lower at the lowest pH values; however, neither such high initial Mn 2+ concentrations nor such low pH values are expected in drinking water treatment. With higher pH (pH > 6) and at higher initial Mn 2+ concentrations, the removal efficiency ranged between 50-80% at the applied adsorbent dose of only 40 mg L −1 . It should be emphasized that the used dose was optimized for Mn 2+ concentrations of 1.5 mg L −1 . Thus, it can be assumed that the removal efficiency would increase with a higher adsorbent dosage (which was confirmed by kinetic adsorption experiments with the optimal adsorbent dosage at pH 7, see Table S3, SD). However, it is worth noting that for water treatment, the relevant concentration of Mn in raw water rarely occurs above 1 mg L −1 [36]. For initial Mn 2+ concentrations in the range of 0.1-0.5 mg L −1 , the strict hygienic limit of Mn residual concentration in drinking water (0.05 mg L −1 ) was reached even using the adsorbent dose of only 40 mg L −1 under all investigated pH values except pH 4.

Impact of coexisting ions on Mn 2+ removal
The coexisting ions (cations: Na + , K + , Ca 2+ , Mg 2+ , and anions: Cl -, SO 4 2-) were chosen based on their presence/ importance in natural waters. Their effect on Mn 2+ removal by adsorption onto TiO 2 -based adsorbent is presented in Figure 5. Similar to the data analysis of the influence of pH value, the equilibrium data related to the influence of coexisting ions were also analysed using the Langmuir and Freundlich model.
In the presence of monovalent ions, a better correlation between the experimental and model data (according to the higher R 2 values) was achieved by the Langmuir adsorption model; however, in the presence of divalent ions, the Freundlich adsorption model displayed a better fit. The calculated Langmuir and Freundlich model parameters are summarized in Table  S2 Section B, SD.
It was found that the presence of anions had no significant effect on the Mn 2+ removal efficiency; on the other hand, the coexisting cations decreased the removal efficiency of Mn 2+ . The amount of adsorbed Mn 2+ decreased the most when divalent cations were added, while only a slight decrease in the adsorption of Mn 2+ was observed when monovalent cations were present in the solution. To illustrate, when there were no additional ions, the removal of Mn 2+ at the initial concentration of 0.5 mg L −1 was 99%. In the presence of Na + or K + , the efficiency decreased only slightly to 96%. By contrast, the addition of Ca 2+ and Mg 2+ ions resulted in Mn 2+ removal decreases to 40% and 74%, respectively. The inhibiting effect of coexisting cations on Mn 2 + removal decreased in the order Ca 2+ > Mg 2+ > K + > Na + .
One possible explanation for the suppressed Mn 2+ adsorption is the effect of ionic strength. When electrostatic interactions between the adsorbate and the adsorbent's surface are attractive, the adsorption efficiency decreases with increasing ionic strength. Conversely, when electrostatic interactions between adsorption participants are repulsive, the increase in ionic strength increases adsorption [37]. In this study, experiments were performed at pH 7, which enables strong, attractive electrostatic interactions between the adsorbent and the adsorbate; thus, the Mn 2+ removal efficiency decreased with increasing concentrations of any cation. This effect was even more pronounced when divalent cations were present. The second possible explanation of the decrease in Mn 2+ removal is divalent cations competing with Mn 2+ for active sites of the adsorbent [38]. The occurrence of this phenomenon is highly probable in our study because the concentrations of added Ca 2+ and Mg 2+ in the solution decreased after adsorption (data not shown).
The difference between the impact of Na + and K + on Mn 2+ removal was insignificant, as well as the difference between the impact of Cland SO 4 2-. On the other hand, in comparison with Mg 2+ , Ca 2+ exhibited a higher inhibiting effect on Mn 2+ removal efficiency. Based on the results of several studies [39,40], the effect of coexisting ions is linearly related to the ionic radii of the competing ions.
By contrast, diverse results were reported by Rachel et al. [41], who investigated the impact of CaCl 2 on Mn 2+ adsorption onto granular activated carbon and modified activated carbon. Based on the author's observations, the adsorption improved in accordance with the increasing concentration of CaCl 2 [41]. This could be ascribed to repulsive electrostatic forces between Mn 2+ ions and the surface of the utilized adsorbent, which were suppressed by increasing ionic strength. Similarly, a positive effect of the presence of Ca 2+ or Mg 2+ ions was observed in studies where TiO 2 -based adsorbent was used for natural organic matter removal [23,25,42].
In the study by Motlochova et al. [27], which investigated the adsorption of Pb 2+ , Cu 2+ , and Cd 2+ on different types of TiO 2 -based adsorbents, no significant impact of coexisting ions on the removal efficiency of the target compound was observed. This diverse trend can be explained by a very different concentration range of the adsorbed metals (hundreds of mg L −1 in the study by Motlochova et al. [27] vs tenths and units of mg L −1 in this study) and the concentration of coexisting ions.

Mn 2+ removal mechanisms
The above-described results of the adsorption pH dependency indicate that electrostatic interactions between Mn 2+ and negatively charged TiO 2 -based adsorbent surface are an important adsorption mechanism. However, the results of some studies investigating the removal of different metal ions (Pb 2+ , Cu 2+ , Cd 2+ ) by similar types of TiO 2 -based adsorbents [27,32] also suggested the involvement of an ion-exchange mechanism. It was proposed that the amount of adsorbed metal ions was comparable to the desorbed amount of alkali metal originally incorporated in the adsorbent structure [27].
To investigate the contribution of ion-exchange in this study, the concentration of Na + was measured in the samples before and after the adsorption experiment. It was found that the concentration of Na + slightly increased after adsorption; thus, ion-exchange apparently participated in Mn 2+ removal. However, no direct correlation between the amount of desorbed Na + from the adsorbent surface and adsorbed Mn 2+ was found. This was likely due to a very low initial concentration of Mn 2+ (0.1-5 mg L −1 ) in comparison to the concentration of Na + (20-30 mg L −1 ) in the solution originating from NaHCO 3 used for alkalinity adjustment. In the studies by Motlochova et al. [27] and Liu et al. [32], much higher concentrations of adsorbates (10 mmol L −1 of Pb 2+ , Cu 2+ , and Cd 2+ [27]; 50 mg L −1 Cu 2+ and 100 mg L −1 Cd 2+ [32] were used. Nevertheless, it can be concluded that in our case, when environmentally relevant adsorbate concentrations were applied, the predominant adsorption mechanism was electrostatic interactions. Furthermore, Doula [43] investigated Mn 2 + removal by adsorption onto Clinoptilolite and a modified Clinoptilolite-iron oxide system and found that the ion-exchange mechanism was predominant only if the initial Mn 2+ concentration did not exceed 200 mg L −1 . Above this value, another adsorption mechanism occurred. Thus, the factors determining the interaction mechanism apparently also include the adsorbent-adsorbate ratio.
To conclude, under the conditions applied in this study, the dominant mechanism of Mn 2+ adsorption onto TiO 2 -based adsorbent was most likely electrostatic interactions, accompanied by an ion-exchange mechanism. The advantage of Mn 2+ removal driven by these mechanisms is that despite being pH-dependent, certain removal was achieved throughout the entire investigated pH range.

Practical aspects
The main achievement of this work was the invention of an effective adsorptive material usable for Mn 2+ removal in drinking water treatment in a wider pH range compared to conventional methods. The efficiency of the adsorbent was examined under different pH values and at various initial concentrations of Mn 2+ , and the results suggest that under typical raw water pH (pH 6.5-7.5) and common initial Mn concentrations (up to 1.5 mg L −1 ), this adsorbent with a dosage of only 40 mg L −1 is able to remove a sufficient amount of Mn 2+ to comply with the strict hygienic limit for potable water of 0.05 mg L −1 Mn.
To provide an insight into the adsorbent safety, residual concentrations of titanium in the solution after adsorption were measured. The maximum allowed concentration of titanium in drinking water is not established [6], but the study by Dong et al. [44] suggested that the maximum allowable concentration of titanium in drinking water may be set at 0.1 mg L −1 . In our study, the measured titanium concentrations after the adsorption experiments were always below the detection limit (0.01 mg L −1 ), which means that titanium is not released from the TiO 2 -based adsorbent and therefore does not pose any threat to drinking water. Moreover, it was verified that the adsorbent does not contain or release any other impurities (details in chapter 3.1). The toxicology assessment also confirmed the adsorbent safety (details in S1, SD).

Conclusion and future prospects
This study used a newly synthesized TiO2-based adsorbent prepared from titanyl sulphate monohydrate for Mn2+ removal, while the effects of different solution conditions (pH and coexisting ions) were investigated.
The results show that this adsorbent achieves high Mn 2+ removal efficiency due to its unique features (such as amorphous structure and low pH PZC ). Based on our findings, there are two mechanisms involved in Mn 2+ removal: (i) adsorption of Mn 2+ onto TiO 2 -based adsorbent surface via electrostatic interactions and (ii) ion-exchange of Mn 2+ for Na + . Mn 2+ removal occurred throughout the whole range of tested pH values, but the efficiency was greater at higher pH due to the increase in negative charge on the adsorbent surface. When using the optimal adsorbent dosage at pH 7, the strict limit for manganese concentration (0.05 mg L −1 ) was met for all investigated initial concentrations of Mn. The influence of added anions was insignificant. On the other hand, added cations suppressed Mn 2+ removal. A greater decrease in Mn 2+ removal efficiency was observed in the presence of divalent cations, owing to a higher increase in the repulsive forces and competitive behaviour between Mn 2+ and the additional cations; thus, the presence of divalent cations in raw water should be taken into account in future research focused on adsorbent utilization in practice. The equilibrium data were analysed using Langmuir and Freundlich model. The Langmuir isotherm model was used to predict Mn 2+ adsorption onto TiO 2 -based adsorbent under different pH and in the presence of monovalent coexisting ions, whereas in the presence of divalent coexisting ions better correlation was achieved using the Freundlich isotherm model. In addition to the adsorption efficiency, an advantage of the adsorbent is that it contains no impurities and has no tendency to form colloids. Toxicity assessments showed that the sorbent did not cause any statistically significant damage to the cells. The residual concentration of Ti in every sample was below the detection limit. Additionally, synthesis of this adsorbent is economically feasible. The results suggest that this adsorbent might be a promising alternative to conventional methods for Mn 2+ removal during drinking water treatment.
The results obtained in this study fulfil the basic knowledge of the functioning of Mn 2+ removal by the newly synthesized TiO 2 -based adsorbent. Future research should focus on the real-life applications of this adsorbent for the drinking water treatment process. Preparation of the adsorbent in a granular form is currently being investigated, while this form would be usable in pressure filters, similar to granular activated carbon. Investigation of the regeneration process of the granular form of prepared adsorbent and its influence on the removal efficiency is also needed.