Removal of Cadmium(II) by hydrated manganese dioxide: behaviour and mechanism at different pH

ABSTRACT Homogeneous precipitation was proposed to prepare hydrated manganese dioxide (HMO) with KMnO4 as oxidant, NaCl as reductant and HNO3 as reaction auxiliary. HMO was applied to remove Cd(II) and the effect of contact time, initial concentration, adsorbent dose and pH value on adsorption efficiency were investigated. The removal mechanisms at various pH values were analysed in detail. Adsorption thermodynamics parameters were calculated as ΔG < 0, ΔH > 0 and ΔS > 0, which meant that the adsorption process was endothermic. The result of adsorption kinetics indicated the adsorption process conformed to pseudo-second-order kinetics. When adsorbing Cd(II) with initial concentration equaling 100 mg·L−1, the activation energy (Ea) was 62.740 kJ·mol−1. The Langmuir model could describe adsorption behaviour on HMO better than the Freundlich model, indicating that the adsorption sites of HMO were homogeneous and that single-layer adsorption was a dominant way in this process. The maximum adsorption capacity of Cd(II) on MnO2 calculated by the Langmuir model was 267 mg·g−1. The adsorbent HMO could be recycled and reused for several times with a high efficiency above 70% by adding HCl. SEM, EDS, FTIR and XPS were used to analyse the mechanisms of removal of Cd(II) at pH = 3,7 and 10. The mechanisms included electrostatic attraction, ion exchange and chemical precipitation. With pH increasing, the zeta potential decreased and the surface negative charge increased, promoting Cd(II) removal through enhanced electrostatic attraction. Meanwhile, ion exchange mechanisms including inner-sphere complexation and outer-sphere complexation occurred during adsorption process at different pH. GRAPHICAL ABSTRACT


Introduction
Water pollution caused by heavy metal ions increasingly threatens the survival of human beings and aquatic organisms. Various industrial and agricultural activities and atmospheric pollution subsidence lead to the inflow of large number of heavy metal ions into water and soil. Due to the high toxicity and non-biodegradation, heavy metal ions will migrate and accumulate in water and soil, causing extremely serious damage to the ecological environment [1][2][3]. Cadmium is one of the most toxic heavy metals and it can be absorbed by humans and other organisms through the food chain [4,5]. Once Cadmium is absorbed by human body, it will remain in the organisms and accumulate in body, causing damage to the immune system, central nervous and reproductive system and posing a serious threat to human life and health [6][7][8]. The World Health Organization recommends a guideline concentration of Cd(II) in the drinking water below 0.005 mg·L −1 [9]. Therefore, it is imperative to develop an effective and simple method to remove heavy metal ions before industrial and agricultural wastewater is discharged into the ecosystem, which is conducive to environmental sustainability and human life and health.
Various conventional wastewater treatment techniques such as chemical precipitation [10], chemical coagulation [11], membrane separation [12], reverse osmosis [13], ion exchange [14] and adsorption [15] have already been reported. In the past few years, the adsorption method has received increasing attention in the removal of heavy metal ions and organic compounds from wastewater due to its simple operation, wide range of sources of adsorbents, low cost, excellent removal effect and environmental friendliness [8,16]. Various sorbents such as modified activated carbon [17], modified biochar [18], clay minerals [19] and magnetic composite materials [20,21] have been widely applied to research on the removal of heavy metal ions. However, the practical application of above sorbents is limited because of the complicated preparation methods, material cost and limited adsorption performance far inferior to the adsorption efficiency of some metal oxides, especially manganese oxides. Manganese oxides are excellent adsorbents because of their lowcost, eco-friendly, biocompatibility, their ability to form complexes with heavy metal ions (e.g. Cd(II), Cu(II), Zn (II) and Pb(II)) and good chemical stability under basic and acidic condition [22,23]. Meanwhile, manganese oxides have a large surface area, microporous structures and high affinity for metal ions, which makes manganese dioxide (MnO 2 ) considered as the most important scavenger of aqueous trace metals in soil, sediments and rocks because of its dominant adsorptive behaviour [24,25]. Qin et al. [25] generated manganese dioxide in situ through the reaction of manganese sulphate and potassium permanganate to remove Pb(II) and Cd(II), and the result showed that the maximum adsorption capacity of Pb(II) and Cd(II) reached 917 and 176 mg·g −1 , respectively. Qi et al. [26] prepared manganese dioxide by the reaction of manganese sulphate and potassium permanganate, and found it could removal heavy metal ions Pb(II), Ni(II) and Cd(II); Meanwhile they found potassium permanganate could improve the efficiency of removal of heavy metal ions by manganese dioxide. Liu et al. [27] synthesized hydrous manganese dioxide with different zeta potentials and was applied to adsorb As(III). And it is reported that HMO could remove trace amounts of As(III) and HMO with smaller zeta potential achieving higher removal rate of As(III). Besides, manganese dioxide and its composites have been applied to water treatment for the removal of Tl(I) [28,29] and organic contaminants [30,31]. These reports about MnO 2 and its composites exhibit potential application in the removal of heavy metals.
However, many studies have focused on the preparation methods of MnO 2 and adsorption performance of different heavy metal ions on MnO 2 , and the study of surface properties has not received sufficient attention [32]. There are studies attaching importance to the mechanism of adsorption of heavy metal ions by MnO 2 and its composites [33,34], but there are few studies on the removal mechanisms at different pH values. The pH value is one of the most important physical and chemical parameters to adsorption, which influences the species of heavy metal ions in solution and the surface characteristics of adsorbent especially the surface charge of adsorbent [27,35]. Therefore, it is meaningful to investigate the removal mechanisms at different pH values.
Herein, a simple method called homogeneous precipitation was proposed to prepare two kinds of MnO 2 . One was synthesized under acidic conditions using sodium chloride and potassium permanganate as precursors. The other was synthesized by potassium permanganate as oxidant and manganese chloride as reductant. Through the comparison, MnO 2 synthesized by the first method performs better adsorption effect. The influence of various adsorption conditions like contact time, adsorbent dose, initial concentration of Cd(II) and pH on the adsorption process were explored. Meanwhile, the adsorption thermodynamics and kinetics were examined to evaluate the adsorption behaviour further. The lattice structure, internal structure and surface morphology of HMO were characterized by Xray diffraction (XRD), scanning electron microscope (SEM) and Brunauer-Emmett-Teller (BET), while the adsorption mechanisms in different pH were discussed in detail by Fourier transform infrared spectroscopy (FTIR), SEM-EDS analysis and X-ray photoelectron spectroscopy (XPS).

Preparation of adsorbent
Manganese dioxide was prepared by homogeneous precipitation method, which regarded KMnO 4 as oxidant, NaCl as reducing agent and HNO 3 as reaction auxiliary to provide the acidity required for the reaction. The oxidation-reduction reaction is 116 mL 0.16 mol·L −1 of KMnO 4 solution and 696 mL 0.08 mol·L −1 of NaCl solution were mixed evenly, then concentrated HNO 3 of 68% was added in batches, and they were stirred under the magnetic agitator (Re>120 rpm) until the solution became colourless. Then the MnO 2 precipitate was washed with distilled water until the conductivity touched 0.02 μs·cm −1 . The obtained MnO 2 was kept in aqueous solution to prevent surface dehydration, marked as HMO.
We also prepared α-MnO 2 by co-precipitation with potassium permanganate as oxidant and manganese chloride as reductant [32]. The oxidation-reduction reaction is 200 mL 0.1 mol·L −1 of KMnO 4 solution and 300 mL 0.1 mol·L −1 of MnCl 2 solution were stirred under a magnetic stirrer. After the reaction was completed, the mixture was continued to stir for 1 h, and then washed until the conductivity reached 0.02 μs·cm −1 . Then the obtained MnO 2 was kept in an aqueous solution and marked as HMO-2.

Analysis method
The concentration of Cd(II) and Mn(II) was measured by atomic absorption spectrometry (AAS). To calculate the concentration of MnO 2 in HMO suspension, we dissolved MnO 2 with hydrogen peroxide and dilute nitric acid into Mn(II), and then the concentration of Mn(II) was measured by AAS. The crystal structure and surface morphology of MnO 2 was analysed by X-ray diffraction (XRD) and scanning electron microscope (SEM); the change of atomic proportion and surface functional groups of MnO 2 before and after adsorption were analysed by EDS and Fourier transform infrared spectrometer (FTIR). The surface elements composition and element oxidation state of MnO 2 were detected by X-ray photoelectron spectroscopy (XPS).

Batch adsorption experiments
The batch adsorption experiments were carried out at 303.15 ± 0.1 K in the thermostatic water bath with magnetic stirrers at 200 rpm. 100 mL of Cd(II) solution was added into a 200 mL beaker. The pH value was adjusted with 0.1 mol·L −1 HCl and 0.1 mol·L −1 NaOH, and then 5 mL HMO was added in solution system above. The concentration of Cd(II) in the water samples was determined by AAS after the water samples were filtered through 0.22 μm membrane filter after being stirred for 3 h.
Equilibrium adsorption capacity of Cd(II) on HMO Q e (mg·g −1 ) was calculated by Equation (3): where C 0 (mg·L −1 ) and V 0 (L) are the initial concentration and volume of Cd(II) solution, respectively. C e (mg·L −1 ) and V e (L) represent equilibrium concentration and volume of Cd(II) solution. W (g) represents the mass of adsorbent. Adsorption capacity at specific time of Cd(II) on HMO, Q t (mg·g −1 ) was calculated by Equation (4): where C t (mg·L −1 ) and V t (L) are concentration and volume of Cd(II) solution at a specific time respectively. Removal rate of Cd(II) (R%) was calculated by Equation (5): 3. Results and discussion  [36]. Figure 1(b) shows that HMO and HMO-2 had the same functional groups, and there were obvious peaks around the wavenumber of 517, 1620 and 3371cm −1 , corresponding to the stretching vibration of Mn-O, -OH bending vibration of water molecule and -OH stretching vibration of associated state, respectively [37]. However, -OH stretching vibration peak of HMO was obviously weaker than that of HMO-2, which might be related to the preparation of HMO under strong acid conditions. The crystal structure of manganese dioxide prepared by the two methods were similar through XRD and FTIR analysis, but the adsorption efficiency of HMO was higher than HMO-2.
To compare the adsorption performance of two kinds of MnO 2 , 5 mL HMO and HMO-2 were used to adsorb 100 mL solution with Cd(II) initial concentration varying from 30 to 100 mg·L −1 with pH = 7 at T = 303.15 K, and the relationship between adsorption capacity and equilibrium concentration was compared. Figure 1(c) shows that the adsorption capacity of HMO was slightly smaller than that of HMO-2 at low equilibrium concentration, but with the concentration increased, the adsorption capacity of HMO was larger than that of HMO-2. The equilibrium adsorption capacity maximum of Cd(II) on HMO achieved 267 mg·g −1 , while that of Cd(II) on HMO-2 reached 192 mg·g −1 . Therefore, we explored the adsorption behaviours and mechanism of HMO in detail.
XRD and FTIR analysis showed that the crystal structure of manganese dioxide prepared by the two methods were similar, but their adsorption efficiency was different. We examined the surface morphology by SEM. Figure 2(a-d) shows that morphology of HMO and HMO-2 was totally different. HMO was regular  spherical particles,while the HMO-2 was irregular, amorphous and uneven surface layer structure.
BET method was applied to measure the specific surface area and pore size of HMO and HMO-2. The N 2 adsorption and desorption isotherms and pore size distribution of the two kinds of manganese dioxide were shown in Figure 3. The BET surface area and average pore size were shown in Table 1. The results indicated that both HMO and HMO-2 were mesoporous materials and HMO had larger specific surface area and smaller average pore size than HMO-2. From Figure 3(c,d), the pore size distribution of HMO was more homogeneous than HMO-2, which is consistent with the results obtained by SEM. Therefore, HMO could provide more adsorption sites so that HMO could remove Cd(II) more efficiently than HMO-2.

Effect of contact time on Cd(II) removal
To remove heavy metal ions effectively and reduce energy consumption, it is necessary to investigate the time for adsorption to reach equilibrium. In this experiment, Cd(II) with C 0 = 50 and 100 mg·L −1 were studied and analysed for adsorption time from 1 to 360 min at pH = 7, T = 303.15 K. Figure 4(a,b) indicated concentration decreased with time, and the removal rate increased with time. Furthermore, adsorption equilibrium could reach within 25-30 min for wastewater containing Cd(II) with C 0 = 50 mg·L −1 , while equilibrium could reach within 2 h for wastewater containing 100 mg·L −1 of Cd(II). The removal rate for C 0 = 100 mg·L −1 could achieve 86%, while for 50 mg·L −1 , the removal rate could achieve 99.92%.

Effect of HMO dose on Cd(II) removal
The adsorbent dose plays a significant role in the economic benefit evaluation of cleaning wastewater containing large number of heavy metal ions. Herein, HMO with volume increasing from 1 to 5 mL was applied to adsorb 100 mL wastewater containing Cd(II) of 50 mg·L −1 for 2 h. Then we explored the variation of the adsorption capacity and adsorption removal rate with HMO volume increasing. As shown in Figure 5, the adsorption capacity decreased with the increasing adsorbent dose, while the removal rate increased significantly. This phenomenon might own to the increasing available adsorption sites with growing HMO volume. With adding low volume of HMO, the surface of HMO was saturated with Cd(II) and the residual Cd(II) in the  solution was higher. With adding more HMO, the HMO could provide sufficient adsorption sites for Cd(II), which led to residual concentration of Cd(II) decreased and removal rate increased. When we added 4 mL of HMO, the equilibrium concentration could reach an extremely low value of 0.05 mg·L −1 Cd(II), and the removal rate could achieve 99.97%.

Effect of initial concentration on Cd(II) removal
We took 5 mL HMO to adsorb wastewater with C 0 increased from 60 to 110 mg·L −1 at pH = 6 and pH = 7, respectively. The adsorption was carried out on the thermostatic oscillator for 3 h to complete adsorption at the fixed temperature of 303.15 K. Figure 6 shows that the equilibrium concentration (C e ) of adsorption occurred at pH = 6 and pH = 7 increased with increasing initial concentration, while the removal rate (R%) decreased with initial concentration increasing. Adsorption sites on HMO surface are sufficient for Cd(II) with low concentration, which made the adsorption independent of the initial concentration. However, with Cd(II) increasing, the competition for adsorption sites got fierce [38]. Therefore, at low initial concentration, the removal rate reached 99.77% and 99.94% and the equilibrium concentration decreased as low as 0.1374 and 0.0354 mg·L −1 at pH = 6 and 7 respectively. When initial concentration increased to 70 mg·L −1 and higher, with the certain adsorption sites being occupied, the equilibrium concentration of adsorption occurred at pH = 6 and pH = 7 increased and the removal rate decreased quickly with initial concentration increasing. There was little difference between pH = 6 and pH = 7 with C 0 ≤ 70 mg·L −1 in equilibrium concentration and removal rate. However, the equilibrium concentration of pH = 7 was significantly lower than that of pH = 6, and the removal rate of pH = 7 was greatly higher than that of pH = 6, indicating that pH was a vital parameter for removing Cd(II).

The pH pzc of HMO and effect of pH value on Cd(II) removal
To explore the influence of pH on HMO adsorption efficiency, this research measured the Zeta potential of   HMO and adsorption behaviour at different pH. We diluted 1 mL HMO to 200 mL and took 6 mL suspension to adjust pH = 1.5-11 precisely, then took 1 mL solution with different pH to measure zeta potential by Malvern Zetasizer nano ZSE. Figure 7(a) shows that the negative charge on the surface of HMO rose with the increase of pH value at pH = 1.5-10. In addition, we found that the point of zero charge of HMO equalled to 3.31(pH pzc = 3.31). The species distribution of Cd(II) in aqueous solution changed with varying solution pH [39]. And the species distribution of Cd(II) under different pH values was calculated by Visual MINTEQ 3.1, as shown in Figure 7(b). Cd 2+ is the main existing specie of Cd(II) at pH ≤ 7, while pH > 7, the concentration of Cd(OH) + increases with pH increasing. Cd 2+ , Cd(OH) + and Cd (OH) 2 coexist in system with Cd(II) concentration equals to 100 mg·L −1 at pH > 9. When pH was higher than pH pzc , varying from 4 to 10, HMO was negatively charged, which were conducive to the removal of Cd (II) through enhanced electrostatic attraction. Meanwhile, we took 5 mL HMO to adsorb 100 mL wastewater containing 50 and 100 mg·L −1 Cd(II) for 3 h at pH varied from 3 to 11. From Figure 8, the equilibrium concentration of Cd(II) with C 0 = 50 mg·L −1 could achieve a low value at pH = 5 and the removal rate reached 99.3%, and 99.97% at pH = 7, indicating that adsorption with C 0 = 50 mg·L −1 was independent with pH when pH ≥ 5 and the surface adsorption sites were sufficient. While the adsorption sites were insufficient, the removal rate and equilibrium concentration were dependent on pH. Figure 8 shows that the equilibrium concentration of Cd(II) with C 0 = 100 mg·L −1 almost increased linearly, and the removal rate decreased linearly at pH = 3-8. As pH increased from 3 to 8, hydroxide radical on HMO augmented, and the H of -OH on HMO was replaced by Cd 2+ . The equilibrium concentration decreased slowly at pH = 9-11, and this might partly be attributed to the formation of different Cd(II) species with pH increasing [40]. The removal rate achieved higher value of 99.95% at pH = 10, which was 8.44% higher than the value of pH = 8. From Figure 7(b), as pH > 9, the level of Cd(II) in the solution dramatically decreased because Cd(OH) + and Cd(OH) 2 were generated in wastewater containing high Cd(II) concentration.
These results shown above indicated that pH value did play an important role in the adsorption behaviour of adsorbent. On the one hand, the existing species of Cd(II) was determined by pH value. From Figure 7(b), Cd 2+ is the main existing form of Cd(II) at pH ≤ 7, while at pH > 7, the concentration of Cd 2+ decreases and Cd (OH) + increases with pH growing. Cd 2+ , Cd(OH) + and Cd(OH) 2 coexist in system with Cd(II) concentration equals to 100 mg·L −1 at pH = 10. On the other hand, the surface charge of HMO is related to the pH value. While pH > 3.31, HMO surface hydroxyl gradually became negatively charged, which strengthened the electrostatic attraction between HMO and Cd(II).

Thermodynamics study of adsorption
To evaluate the adsorption behaviour deeply, we calculated the thermodynamic parameters including Gibbs free energy change (ΔG), enthalpy change (ΔH ) and entropy change (ΔS). Cd (II) with different initial concentrations (80, 90, 100 mg·L −1 ) were adsorbed on 5 mL HMO suspension at different temperatures (303.15, 308.15, 313.15 K). Thermodynamic parameters including ΔG, ΔH and ΔS were calculated by Equations (6) and (7) [41]: where R (8.314 J·mol −1 ·K −1 ) is the universal gas constant, T (K) is the temperature, the k e is the ratio of the adsorption capacity at equilibrium to the concentration of the solution at equilibrium, and the ratio is dimensionless, k e = Q e C e × 1 g/L the C e (mg·L −1 ) and Q e (mg·g −1 ) represent equilibrium concentration and equilibrium adsorption capacity.
The results were shown in Figure 9 and Table 2, from which we found that ΔG < 0, ΔH > 0 and ΔS > 0 was set up with different initial concentrations of Cd(II), which meant that the adsorption process of Cd(II) on HMO was spontaneous and endothermic. Furthermore, the absolute value of ΔG increased with temperature increasing, which meant increasing temperature facilitated the removal of Cd(II) on HMO. The absolute value of ΔG decreased with increasing concentration, indicating that it would consume more energy to adsorb high initial concentration of Cd(II).

Adsorption kinetics study
To evaluate the whole process of Cd(II) removal by HMO better, this paper investigated the changing trend of the adsorption of Cd(II) on HMO with time (from 1 to 360 min) to investigate adsorption kinetics. In addition, the experimental data were fitted with pseudo-firstorder kinetic model (8), pseudo-second-order kinetic model (9) and intraparticle diffusion model (10) in Origi-nPro8.5 software [42].
where k 1 (min −1 ), k 2 (g·mg −1 ·min −1 ), k d (mg·g −1 ·min −1/2 ) represent the first, second order rate and particle diffusion velocity constant, respectively. And C (mg·g −1 ·) is a constant. Adsorption activation energy E a (J·mol −1 ) was calculated by Arrhenius formula shown as Equation (11): where k T1 and k T2 (g·mg −1 ·min −1 ) are reaction rate constants at T 1 and T 2 , respectively. R is universal gas constant (8.314 J·mol·K −1 ). The Root Mean Square Error (RMSE) was calculated by Equation (12): In Figure 10 (a,d), at the early stage of adsorption of Cd (II) with C 0 = 50 and 100 mg·L −1 , the adsorption capacity increased rapidly with time, while with time extension, the adsorption capacity increased slowly and at last, it kept almost the same. This trend could be explained by following reason. HMO has sufficient adsorption sites and high negative charge at the initial stage of adsorption process, which makes HMO adsorbing Cd(II) quickly. The sites were occupied and negative charge decreased over time and the adsorption achieve equilibrium gradually. From Figure 10 and Table 3, the R 2 Figure 9. Thermodynamics study of adsorption at different initial concentrations.  (12) and the results were listed in Table 4. The results demonstrated the inference that the Psuedo-second order could describe the adsorption process than Psuedo-second order could. This indicated that chemisorption of Cd(II) on HMO was the rate-limiting mechanism.
Here, we took k 2(T1) and k 2(T2) to replace the K T1 and K T2 to calculate the apparent activation energy (Ea) of the process of adsorbing wastewater containing 50 Figure 10. Adsorption kinetics study: (a)-(c) and (e)-(g) are the variation of adsorption capacity over time, pseudo-first-order kinetic model, pseudo-second-order kinetic model. Table 3. Parameters of adsorption kinetics fitted with first-order and second-order models.
Psuedo-first order Psuedo-second order and 100 mg·L −1 Cd(II) by Equation (11). Ea of adsorbing 50 mg·L −1 Cd(II) was 46.28 KJ·mol −1 and that of 100 mg·L −1 Cd(II) was 62.74 KJ·mol −1 , which meant that the apparent activation energy of adsorption process was related to initial concentration of Cd(II). Furthermore, the experimental data were fitted by the intraparticle diffusion model as shown in Figure 11  (a,b), and the parameters were listed in Table 5. The adsorption of Cd(II) with C 0 = 50 and 100 mg·L −1 by HMO could only be fitted by various lines rather than a consecutive line at different temperatures, which indicated adsorption process was not controlled by one single rate-determining step but multiple steps simultaneously [43]. The adsorption rate constant gradually decreased from the first to the third stage(k d1 > k d2 > k d3 ) indicating the adsorption rate decreased. From Table 5 the higher C 0 is, the greater k d1 is. The k d2 and k d3 were lower in T = 313.15 K (k d2 = 3.75 for 50 mg·L −1 , 4.12 for 100 mg·L −1 , k d3 = 0.0031for 50 mg·L −1 , 1.0004 for 100 mg·L −1 ) than in T = 303.15 K (k d2 = 3.94 for 50 mg·L −1 , 4.21 for 100 mg·L −1 , k d3 = 0.0094 for 50 mg·L −1 , 1.0191 for 100 mg·L −1 ). The possible reasons are listed as follows. At the initial stage, Cd(II) was adsorbed quickly to the surface of MnO 2 controlled by film diffusion due to electrostatic interaction and concentration difference, which could also explain why k d1 was higher at higher initial concentration. With adsorption going on, the sites on the outer surface of MnO 2 were occupied, the Cd(II) ions adsorbed by electrostatic interaction on outer surface of MnO 2 diffused to the inner surface with the concentration difference as a driving force. After Cd(II) diffused to the internal surface on MnO 2 , the Cd (II) adsorbed on adsorption sites gradually form inner-sphere complexation ≡MO (CdOH) or ≡MOCdOM≡ [32,44].

Adsorption isotherm study
It is essential to investigate the adsorption isotherm, which is a reliable prediction of adsorption parameters that can help us to optimize the adsorption process. Therefore, we fitted the experimental data for the Langmuir and Freundlich models. The basic assumption of Langmuir model is much stricter, which assumes that  Figure 11. Adsorption kinetics study: (a) and (b) intraparticle diffusion model for Cd(II) with C 0 = 50 and 100 mg·L −1 , respectively. each adsorbent site can only be occupied by one adsorbate molecule due to the generation of a monolayer on the surface. In addition, all adsorption sites are considered energetically equivalent without the interaction between the adsorbed molecules [45,46]. Langmuir model is expressed as Equation (13). The Freundlich model shown as Equation (14) below describes the non-ideal and reversible adsorption which is applied to adsorption on the heterogenous surface. Different from Langmuir model, it describes reversible adsorption and is not restricted to the formation of a monolayer [47].
Q (mg/g) is the maximum adsorption capacity; K L and K F represent the Langmuir and Freundlich adsorption constant. And n is the Freundlich exponent.
The results and parameters of experimental data plots curving-fitting are shown in Figure 12(a-d), Tables 6 and  7. Combining Figure 12 with Tables 6 and 7, Langmuir model could describe the adsorption of Cd(II) on HMO better than Freundlich model could at any situation we investigated, indicating that the adsorbent sites were equivalent and monolayer adsorption was the dominant form on the surface, and there was no nucleating or precipitating at different situation we investigated [48]. The estimated maximum adsorption capacities of Cd(II) were 208.50, 232.70, 259.28, 273.04, 293.80 mg·g −1 , which were much higher than that reported by many literatures listed in Table 8.
According to Figure 12(a,b), we found at a certain pH value, the higher the temperature was, the greater the maximum adsorption capacity was, which was consistent with thermodynamic calculation result. When the temperature was constant, the maximum adsorption capacity increased with the pH value increasing, which was corresponding with the result of pH value effect.
From Table 6, the correlation coefficients R 2 of Langmuir model fitting were much higher than R 2 of Freundlich model fitting. Meanwhile, we found that the highest R 2 of Langmuir model fitting only reached 96.95% which was less than 99%. From Figure 12(a,b), at low equilibrium concentration, the experimental data fitted Langmuir extremely well, while at the high equilibrium concentration, the experimental data were higher than the Langmuir model. This phenomenon existed in all five conditions we explored. We speculated that the presence of high concentration in the solution after the adsorption sites reached saturation resulted in multiple layers of adsorption from the surface where Cd(II) was adsorbed, which made the experimental data higher than Langmuir model fitting. When the ≡MO(CdOH) was generated, it was possible to continue adsorbing Cd(II) on ≡MO(CdOH) to generate multilayer adsorption, which could explain why the maximum adsorption capacity was greater than the maximum adsorption capacity estimated by fitting Langmuir model in the adsorption isotherm.
To verify the effectiveness of the predictive models and inference above, the RMSE was calculated by Equation (12) and the results were shown in Table 7. Through calculation, the RMSE of Langmuir model at different conditions were all lower than Freundlich model which meant that the Langmuir model could describe the adsorption process better. The RMSE of Langmuir model was not close to 0 because the actual adsorbent prepared in this paper did not meet the basic assumptions of Langmuir model perfectly and it might form multiple layer adsorption, which is consistent with the inference above.

Characterization and adsorption mechanism
3.9.1. XRD analysis There were obvious peaks at 2θ = 12.78°, 18.57°, 28.56°, 37.02°, 39.64°, 65.06°as shown in Figure 13, which was corresponding to 110, 200, 310, 211, 330, 002 crystal planes of α-MnO 2 respectively (JCPDS 440141) [36]. The peak position of α-MnO 2 had no obvious variation before and after adsorption, but the intensity of 002 crystal plane (2θ = 56.06°) of HMO was weaker after adsorption. And the higher pH was, the weaker intensity was. Besides, the 330 crystal plane (2θ = 39.64°) disappeared after adsorption at pH = 7 and 10. The phenomena might be attributed due to the formation of Cd-O to replace the 330 crystal plane (2θ = 39.64°) of MnO 2 . The peak at 2θ = 33.01°of MnO 2 after adsorption under the three conditions was detected while MnO 2 before adsorption was not. Moreover, the peak at 2θ = 33.01°w as more obvious at high pH value. This peak might result from ≡MOCdOM≡ complex. Cd 2+ that replaced the hydrogen of hydroxyl group on the surface of MnO 2 to form ≡MO(CdOH), then undergoes intramolecular dehydration to form ≡MOCdOM≡ [49]. There were obvious peaks at 17.89°, 30.21°, 46.99°and 58.18°w hen the adsorption was carried out at pH = 10, which corresponded to the 110, 001, 041 and 221 crystal planes of JCPDS40-0760 standard card of Cd(OH) 2 , respectively. Figure 14(a-c), the images showed there was slight change between MnO 2 particles before and after adsorption in surface morphology. The surface of MnO 2 without adsorption presented many patterns forming surface channels, which formed large surface area and was conducive to the adsorption of heavy metal ions. However, the channels on the surface of MnO 2 after adsorption became smaller and the patterns became dilated, indicating that the morphology of  MnO 2 has been changed because of chemical adsorption. In Figure 14(d), the SEM of HMO adsorbing Cd(II) at pH = 10, we could see a regular cube structure and the particle with cube structure was identified as Cd (OH) 2 through EDS analysis. Therefore we believed that Cd(OH) 2 precipitation was indeed generated at pH = 10, which was consistent with results calculated by Visual MINTEQ 3.1. From EDS spectrum analysis as  shown in Figure S1 and Table 9, the percentage of oxygen approximate in HMO (adsorbing Cd(II) at pH = 3, pH = 7 and pH = 10) equalled to twice the sum of the metal atoms percentage, which suggested that Cd (II) adsorbed on HMO approximately existed in the form of inner-sphere complex ≡MOCdOM≡or ≡MO (CdOH). Above all, the results from SEM are consistent with the results of XRD.

FTIR spectra
To explore the changes in functional groups that happened to HMO before and after Cd(II) adsorption under different pH, we measured the FTIR spectrum of HMO. The standard spectrum of the -OH vibration peak is at 3400 cm −1 wavenumber and the -OH bending vibration absorption peak of water molecule was at 1614 cm −1 [37]. In Figure 15, the -OH peak of HMO appeared at 3380 cm −1 , while that of HMO adsorbing Cd(II) appeared at 3308, 3231 and 3376 cm −1 at pH = 3,7 and 10, respectively. The shifted position and changed intensity of -OH vibration peak probably because Cd(II) replaced H on -OH on the surface of MnO 2 by ion exchange forming inner-sphere complexes between heavy metals and HMO as ≡MOCdOM≡ or ≡MO(CdOH) [50]. Compared with MnO 2 without adsorption, the -OH peak intensity of HMO after adsorbing Cd (II) at pH = 3 kept similar, and the peak position changed slightly. When pH = 7, the -OH peak strength was the weakest after Cd(II) adsorption, indicating that -OH was consumed, which meant that Cd(II) was approximately adsorbed on the surface of HMO in the form of inner-sphere complex ≡MOCdOM≡ through the Cd-O bonds [51]. The peak strength after adsorption at pH = 10 got stronger, demonstrating that more associated -OH was generated. The possible reasons were part of Cd(II) was changed into Cd(OH) 2 adhered on the surface of HMO and part of Cd(II) was adsorbed on the surface of HMO in the form of ≡MO(CdOH). The -OH bending vibration absorption peaks of water molecules before and after adsorption were in the range of 1607-1620 cm −1 . And there was tiny change after adsorption under different conditions. The position of Mn-O bond was basically unchanged. Mn-O of HMO adsorbing Cd(II) at pH = 3 and pH = 7 and HMO was the same, appeared at 517 cm −1 . However, Mn-O of HMO adsorbing Cd(II) at pH = 10 moved to a higher wavenumber of 536 cm −1 approximately caused by the formation of Cd(OH) 2 on the surface of MnO 2 [52]. Particularly, the HMO after adsorbing Cd(II) appeared weak peaks at 860 and 1399 cm −1 at pH = 10, which were corresponding to the fingerprint region of Cd(OH) 2 .

XPS characterization
The mechanisms of Cd(II) removal were further discussed by XPS full scan spectra and Mn2p, Cd3d and O1s narrow scanning spectra analysis. The results were shown in Figure S2 (XPS full scan spectra) and Figure  16(a-c). From Figure S2, the peak intensity of Cd(II) in the survey scan spectrum of HMO adsorbing Cd(II) was various under different conditions, which was related to the adsorption capacity of Cd(II). Figure 16(a,b) is the narrow scanning analysis of Mn2p and Cd3d, respectively. The binding energy of Mn2p3/2 and Mn2p1/2 decreased after adsorption, which might be attributed to the formation of ≡MO(CdOH) and ≡MOCdOM≡ due to the strong electron donation of Cd(II). Figure 16(b) shows that the binding energy of Cd3d3/2 and Cd3d5/2 on HMO after adsorption at varying pH values was also different, indicating that mechanisms of adsorption were distinct.
According to O1s spectral peak-splitting in Figure 16 (c), three types of surface oxides could be seen. The peak value between 529.89 and 529.99 eV belongs to lattice oxygen (O 2− ); The peak value between 531.30 and 531.37 eV belongs to hydroxyl oxygen (M-OH); The peak at 533.00 eV is oxygen in adsorbed water molecules (H-O-H) [23]. As shown in Figure 16(c) and Table 10, the binding energy of H-O-H had no change between before and after adsorbing Cd(II), while the  Cd(II) was larger than that of original MnO 2 (531.30 eV), but the content was basically unchanged. The tiny variation at pH = 7 might be attributed to the formation of inner-sphere complex ≡MOCdOM≡ and ≡MO(CdOH) through Cd(II) replacing the H on -OH by ion exchange and then part of ≡MO(CdOH) intramolecular dehydrated to ≡MOCdOM≡. Adsorption carried out at pH = 10, the binding energy of O 2− (529.89 eV) was lower than that of O 2− (529.99 eV) in original MnO 2 , and the content of O 2− (65.87%) in MnO 2 adsorbing Cd(II) was slightly lower than that of original MnO 2 (67.03%). The binding energy of M-OH (531.36 eV) was larger than that of original MnO 2 (531.30 eV), but the content increased (from 18.32% to 21.30%). The phenomena indicated that large number of -OH groups were produced. Combined with the results of XRD, FTIR and SEM, it was found that the precipitation of Cd(OH) 2 and ≡MO(CdOH) resulted in the increase of hydroxyl oxygen.  replaced the H on hydroxylated MnO 2 to form ≡MO (CdOH) and ≡MOCdOM≡ through inner-sphere complexation. At pH = 10, Cd 2+ , Cd(OH) + and Cd(OH) 2 coexisted in system, which brought out the result that part of Cd(II) precipitated in solution and part of Cd(II) adsorbed the surface in different complexes. The potential reactions mentioned above were expressed by Equations (15)- (19). Meanwhile through the isotherm adsorption, we speculated that the presence of high concentration in the solution after the adsorption sites reached saturation resulted in multiple layers of adsorption from the surface where Cd(II) was adsorbed, which made the experimental data higher than Langmuir model fitting. The potential reactions were listed Equations (20) and (21): ; MOH + Cd 2+ ; ; ; ; MO − + (CdOH) + + OH − ; MO · · · [Cd(OH) 2 ] (pH = 10), ; ; MOCdO(CdOH) ; MOCdOCdOCdOM

Recycling of HMO
Reusability is an important index to evaluate the economy of an adsorbent. A suitable regeneration method for HMO was developed in this paper. We took 5 mL HMO to adsorb 100 mL wastewater containing 50 mg·L −1 Cd(II), and desorbed in HCl solutions of different concentrations varying from 0.1 to 0.6 mol·L −1 for 2 h. The desorption rate research shown in Figure  S3 indicated that the desorption rate increased with the increase of concentration when the concentration was less than 0.5 mol·L −1 . When the concentration was 0.6 mol·L −1 , the desorption rate of Cd(II) decreased slightly. Therefore, we selected the concentration of HCl equalled to 0.5 mol·L −1 for desorption experiment.
To explore the regeneration stability, six cycles of adsorption-desorption were carried out. We took 5 mL HMO to adsorb 100 mL wastewater containing 50 and 100 mg·L −1 Cd(II) and desorbed in 0.5 mol·L −1 HCl solution for 2 h, then washed with pure distilled water until the conductivity touched 0.02 μs·cm −1 for the next recycle. The result was shown in Figure 17. With the increase of the recycle times, both adsorption capacity and removal rate decreased. For waste water containing 50 mg·L −1 Cd(II), the removal rate could maintain 80.04% after four cycles, and the adsorption capacity decreased slightly. While for waste water containing 100 mg·L −1 Cd(II), the removal rate drops linearly and it could only keep 63.19% after four cycles. It was obvious that the removal rate and adsorption capacity of Cd (II) both decreased with the increase in recycle times, especially after four times, the removal rate deeply decreased to 63.04% for 50 mg·L −1 Cd(II) wastewater and to 35.01% for 100 mg·L −1 Cd(II) wastewater after six cycles. There are two potential reasons outline follow. It was obvious that the loss of HMO at the adsorption-desorption cycle process contributed to the decrease of removal efficiency. For the other reason might own to adsorption sites being occupied by stable inner-sphere complex. Therefore, we believed that the HMO adsorbing Cd(II) could be regenerated several times with a high efficiency by adding HCl.

Conclusion
In summary, HMO prepared with KMnO 4 as oxidant, NaCl as reductant and HNO 3 as reaction auxiliary exhibited higher adsorption efficiency on Cd(II) than HMO-2. In this study, the influence of initial concentration, HMO dose and pH value on adsorption were investigated. Experimental results showed that under the optimal adsorption conditions, the removal rate could reach 99.97%, and the residual concentration of Cd(II) could reach the industrial discharge standard. The removal of Cd(II) was greatly affected by pH, and the result of effect of pH experiment demonstrated that the higher the pH was, the higher the removal rate of Cd(II) was. It is attributed to the degree of hydroxylation of α-MnO 2 surface binding site, the negative charge of α-MnO 2 surface and the species of Cd(II). Adsorption thermodynamics parameters were calculated as ΔG < 0, ΔH > 0 and ΔS > 0, which meant that the adsorption process was spontaneous and endothermic. The kinetics study and adsorption isotherm study demonstrated that the adsorption process conformed to pseudo-secondorder kinetics and Langmuir model, which indicated that single-layer chemisorption was the dominant way in the adsorption process of Cd(II) on HMO at pH = 7. Through the Langmuir model, the maximum adsorption capacity of Cd(II) at pH = 7 could achieve 267 mg·g −1 at T = 303.15 K and 298 mg·g −1 at T = 313.15 K. Through the recycling of HMO research, the adsorbent HMO adsorbing Cd(II) could be regenerated for several times with a high efficiency above 70%, indicating that HMO prepared here was environmentally friendly adsorbent.
In this study, the mechanisms of Cd(II) removal were investigated through XRD, FTIR, SEM and XPS analysis. The dominant mechanisms of Cd(II) removal were distinct at pH = 3,7 and 10. At pH = 3, Cd(II) absorbed at low pH value was mainly in the form of ; MOHCd 2+ through outer-sphere complexation. For adsorption occurred at pH = 7, large amount of hydroxyl oxygen(-OH) was formed on the surface of adsorptive Cd(II), and most of Cd(II) adsorbed on the surface of HMO in the form of ≡MO(CdOH) and ≡MOCdOM≡ through inner-sphere complexation. For the removal of Cd(II) at pH = 10, Cd(II) existed not only in the form of ≡MOCdOM≡ and ≡MO(CdOH) but also in the form of Cd(OH) 2 on the surface of HMO.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was kindly supported by Program of Guangdong Science and Technology Department of China (B2152990).

Data availability statement
The data support the findings of this study are available within the article and its supplementary materials.