Insight into mechanisms of fluoride removal from contaminated groundwater using lanthanum-modified bone waste

The current worldwide issue of fluoride contamination in groundwater has resulted in an increased demand for efficient adsorbents. Meanwhile discard and landfill of bone waste has led to environmental pollution. In order to achieve bone waste recycling and enhance the efficiency of fluoride removal, a lanthanummodified bone waste (LBW) composite was synthesized and tested to remove fluoride from contaminated groundwater. The adsorbent characterization was conducted by SEM, BET, XRD, FTIR and XPS. The fluoride adsorption performance was evaluated by batch experiments. SEM and BET revealed that the introduction of lanthanum could modify the porous structure of the adsorbent and enhance its specific surface area. The LBW composite had a high pHzpc of 11.4 and the fluoride adsorption was barely affected by the solution pH over a wide pH range of 2.5–10.0. The influence of common coexisting oxygen anions in the range of 0–100 mg L 1 was not significant. The fluoride adsorption was a typical chemisorption process and followed PSO and F-L PSO equations. The FVER model provided a more accurate prediction of a larger surface coverage degree with respect to equilibrium at the initial stage during adsorption of fluoride. Isotherm studies revealed that the reaction obeyed the Langmuir model, indicating that this process was monolayer adsorption. Possible defluoridation and regeneration were proposed. The fluoride adsorption was mainly controlled by the processes of electrostatic attraction on the LBW surface with a positive charge and ion exchange between fluoride and hydroxide ions. This research provides an alternative method for fluoride removal from contaminated groundwater in practical applications.


Introduction
Fluoride contamination in groundwater has been recognized as one of the increasingly serious worldwide environmental problems and has received considerable attention for many years. 1 The breakdown and leaching of uoride bearing rocks and soils, the runoff from agricultural elds and the uncontrolled discharge from industrial establishments are primarily responsible for uoride contamination. 2 The uoride content is a function of many factors, such as the availability and solubility of uoride minerals, the velocity of groundwater, pH, temperature, and the concentration of calcium and bicarbonate ions. 3As an essential micronutrient of the human body, whether uoride in drinking water is benecial to human health depends on its concentration and the duration of continuous uptake. 4Low concentrations of uoride in drinking water can prevent dental caries and facilitate the mineralization of hard tissues, 5 while high concentrations of uoride can contribute to various physical disorders, including dental and skeletal uorosis, infertility, brain damage and thyroid disorder. 6It is estimated that more than 200 million people worldwide rely on drinking water with a uoride concentration that exceeds 1.5 mg L À1 (ref.7) and high uoride concentrations in groundwater can be found in many parts of the world, particularly in parts of India, China, Central Africa and South America. 8Therefore, the fast and effective removal of uoride is of prime importance for providing safe drinking water and maintaining sustainable water resources.
Various techniques have been developed to remove uoride from aqueous solution, including precipitation, 9 adsorption, 10 electrocoagulation, 11 ion exchange, 12 reverse osmosis, 13 and electrodialysis. 14Among these methods, adsorption has developed very mature for deuoridation of water due to its low maintenance cost, exible operation and simple design. 15The viability of such a technique is greatly dependent on the development of suitable adsorptive materials.The adsorbents available for treatment of uoride include activated alumina, 16 bone char, 17 hydroxyapatite (HAP), 18 chitosan, 19 activated carbon 20 and zeolite-based adsorbents. 21However, impertinent discard and landll of some raw materials, such as bones waste, led to environmental pollution, and some other raw materials such as HAP, chitosan are expensive and with complicated regeneration.Therefore, materials waste made into adsorbents for uoride removal from contaminated groundwater has become a signicant direction to explore nowadays, especially in regions with limited resources.
With the acceleration of urbanization and an increasing population, the annual consumption of large amounts of meat in China results in a large amount of bone waste.Bone waste is calcined to obtain bone char, whose main inorganic component is HAP, Ca 10 (PO 4 ) 6 (OH) 2 . 22HAP, as an important inorganic material, has attracted signicant attention in the last two decades due to its structure, ion-exchange capacity and adsorption affinity. 23HAP has been used as an adsorbent for the efficient and selective removal of uoride from aqueous solution through adsorption and ion exchange. 24It was reported that bone waste had been the subject topic of a number of studies using it as uoride adsorbent.For example, Medellin-Castillo reported the uoride removal performance of bone char made from cattle bones waste, which the adsorption capacity was 2.8 and 36 times greater than those of a commercial activated alumina and a commercial activated carbon.This study concluded that the uoride adsorption on bone char was due to its HAP content. 25On the other hand, Rojas-Mayorga reported the optimization of a pyrolysis process at the temperatures of 700 C for the synthesis of bone char.This study concluded that pyrolysis temperatures higher than 700 C cause the dehydroxylation of the hydroxyapatite of bone char reducing its uoride adsorption capacity. 26In other study, Medellin-Castillo determined the effects of solution pH on the adsorption of uoride onto bone char, which the adsorption capacity drastically increased while decreasing the pH from 7.0 to 5.0. 23It is obvious that the operating conditions for bone waste play an important role to determine the nal adsorption capacities of the adsorbent.In general, the uoride adsorption properties of commercial bones waste may be limited, and range from about 0.5 to 3.0 mg g À1 . 10,25Therefore, bone obtained from modied process may show better adsorption capacities.The uoride ion is classied as a hard base, which has a strong affinity towards multivalent metal ions due to its high electro-negativity and small ionic size. 27Meanwhile, earth metal, such as lanthanum, classied as a hard acid, show a high chemical attraction to uoride ions. 28In other words, it is expected that using lanthanum-modied bone waste (LBW) as an adsorbent can not only achieve the recycling of waste, but also can achieve excellent uoride removal due to its synergistic effect.
In order to achieve bone waste recycling and enhance the efficiency of uoride removal, a LBW composite was prepared and tested to remove uoride from aqueous solution.The surface properties and morphology of the LBW composite were obtained by characterization analysis.The adsorption performance of uoride ions on the LBW composite was evaluated by batch experiments.The adsorption behaviors of uoride were analyzed by adsorption kinetic and isotherm models.The peculiarity and applicability of the LBW composite for deuoridation in actual contaminated groundwater were examined.Finally, by combining the aforementioned results, the possible mechanisms of uoride adsorption are also proposed.

Chemicals
All chemicals used in this study, including NaF, LaCl 3 $nH 2 O, NaOH, HCl, NaHCO 3 , Na 2 CO 3 , NaNO 3 , Na 2 SO 4 , Na 3 PO 4 , AgNO 3 and Na 3 C 6 H 5 O 7 $2H 2 O, were of analytical grade without the need for further purication and were purchased from Sinopharm Chemical Reagent Co., Ltd., (Shanghai, China).The uoride stock solution (100 mg L À1 ) were prepared by adding NaF (0.2210 g), which was dried at 105 C for 2 h prior to use, to deionized water (1000 mL).The test solutions were prepared by suitable dilution of the standard stock solution.

Preparation of sodium form of bone waste
The creatural bone wastes collected from local restaurants were dried in an oven at 100 C for 3 h aer cleanly removing surcial fat and meat pieces.The dried bones were crushed and further calcined at 500 C for 4 h.Bone particles with sizes between 75 and 165 mm was selected aer milling and sieving.A fullysaturated sodium form of bone waste (Na-BW) was used in the whole study due to its high ion-exchange capacity, promoting the composition between metal cations and bone waste particles. 29,30According to the mass ratio of NaCl to bones (1 : 5), the Na-BW was prepared by adding a 10 wt% NaCl solution into a 10 wt% bone waste aqueous suspension.Aer 1 h of vigorous stirring at 70 C, the resulting suspension was kept for 24 h at room temperature (25 AE 2 C).Then the solid and liquid parts were separated by centrifugation and washed with deionized water until no chloride was tested in the supernatant (detected with a 0.01 mol L À1 AgNO 3 solution).Finally, the lter cake was dried under 100 C for 3 h to obtain the Na-BW.

Preparation of LBW composite
The LBW composite was prepared on the basis of the previous research 31 with some modications according to the following steps: (1) 5 g of LaCl 3 $nH 2 O was dissolved in 100 mL of deionized water by ultrasonic dispersion for 5 min.Aer 12 h of aging under 60 C, the solution was stabilized at room temperature (25 AE 2 C) for 12 h to obtain solution A; (2) a certain amount of Na-BW (mass ratio of Na-BW to LaCl 3 $nH 2 O 2 : 1) was added into 100 mL of deionized water under vigorous stirring for 10 min to obtain solution B; (3) solution A was dropped into solution B under strong stirring at a slow speed of 0.5 mL s À1 , followed by stirring at room temperature (25 AE 2 C) for 2 h.Then, 0.2 N NaOH solution was immersed into the system for 2 h to keep the pH of the mixture within a range of 5.0-6.0;(4) the precipitate was centrifuged and further dried completely.
Finally, the prepared LBW composite was powdered and used for uoride adsorption experiments.

Characterization
The lanthanum crystal structure present in the LBW was analyzed using X-ray diffraction (XRD) (DX-2700, Dangdong Fangyuan, China) equipped with Cu Ka radiation (l ¼ 0.154056 nm), and the accelerating voltage and the applied current were held at 40 kV and 30 mA, respectively.The surface morphology of bone waste and LBW composite were observed using a scanning electron Microscope (SEM) (S4800, Hitachi, Japan).The specic surface area, pore volume and pore size distribution were determined by the Brunauer-Emmett-Teller (BET) theory, using N 2 as the adsorption/desorption reagent at 77 K (À196 C) (ASAP2020, MICROMERITICS, USA).The zeta potentials of LBW before and aer uoride adsorption were measured using a Zetasizer analyzer (ZEN 3600, Malvern Instruments Ltd., U.K.).Fourier transform infrared (FTIR) spectra were recorded on a Fourier transform infrared spectrometer (TENSOR27, Bruker, Germany) with a resolution of 0.4 cm À1 .The surface chemistry of LBW before and aer uoride adsorption were analyzed using X-ray photoelectron spectroscopy (XPS) (ESCALAB 250Xi, Thermo Scientic, USA) with a monochromatic Al Ka X-ray source (1486.6 eV).

Adsorption experiments and analytical methods
To investigate the ability of uoride removal by the LBW from aqueous solutions, batch adsorption tests were carried out in detail.The solution pH was adjusted by 0.2 N HCl or NaOH.Unless otherwise stated, all the experiments were conducted by adding 1.0 g of adsorbent into a 250 mL polypropylene vessel containing 100 mL of 10 mg L À1 F À solution.The vessels were agitated at 200 rpm in a thermostatic shaker at room temperature (25 AE 2 C) for 24 h to attain the equilibrium.All adsorption experiments were carried out three times.Then, the aqueous solution was sampled at desired times and ltered through a 0.22 mm cellulose acetate lter.The concentration of uoride in solution was measured using a uoride-ion-selective electrode (PF-1 Leici China).TISABI (total ionic strength adjustment buffer, pH 5-6) was added to the solution to protect the uoride measurements from the interference of other ions.
The amounts of uoride adsorbed per unit weight of LBW composite at time t (q t , mg g À1 ) and at equilibrium (q e , mg g À1 ) were calculated using the following equations, respectively: 32 where C 0 (mg L À1 ), C t (mg L À1 ) and C e (mg L À1 ) are the concentration of uoride at the initial time, a given time t and the equilibrium, respectively; V (L) is the volume of the aqueous solution; and m (g) is the dry mass of the adsorbent.The effect of pH on the removal of uoride was investigated by adjusting the initial pH of the uoride solutions to 2.5-11.5.
The kinetic studies were executed in 200 mL of uoride solution (10, 30 and 50 mg L À1 ) with 2.0 g of LBW and a required amount of sample was collected at preset time intervals to determine the residual uoride concentration.The isotherm studies were performed at different initial concentrations (5-60 mg L À1 ) at 25, 35 and 45 C.

Results and discussion
3.1.Adsorbent characterization 3.1.1.SEM and BET analysis.The SEM images (50 000Â) of Na-BW and the LBW composite before and aer uoride adsorption are shown in Fig. 1.Unlike Na-BW, which had poor pores (Fig. 1a), LBW presented a more irregular surface and porous structure (Fig. 1b).This means that the addition of lanthanum could increase the specic surface area of the adsorbent and thereby provide more available adsorption sites.However, as illustrated in Fig. 1c, the pores of LBW became fewer and smaller aer adsorption, indicating that LBW composite might be bonded by uoride ions.
The specic surface area and pore size distribution of the adsorbent exert a signicant effect on the intraparticle diffusion of uoride ions.BET analysis indicates that the specic surface area, total pore volume and average pore diameter of the LBW composite were 85.4 m 2 g À1 , 0.242 cm 3 g À1 and 10.99 nm, respectively.The N 2 adsorption-desorption isotherm of LBW sample is shown in Fig. 1d.According to the International Union of Pure and Applied Chemistry (IUPAC) classication, 33 the LBW adsorbent was of type IV with hysteresis hoops of type H3.The pore size distribution mainly concentrates in a range less than 21 nm, revealing that the composite was a mesoporous material with a signicantly large specic surface area.
3.1.2.XRD analysis.The XRD patterns of the bone waste and LBW composite are depicted in Fig. 2. By comparing with the Joint Committee on Powder Diffraction Standards (JCPDS) le, the characteristic peaks of Na-BW conrmed the presence of the calcium HAP at 2q ¼ 25.9 , 32 and 39.8 , which showed efficient adsorbing-capacity for uoride. 10However, these peaks were much less intensive or disappear in LBW, revealing a change of the original structure of bone waste aer LaCl 3 addition.The diffraction peaks at 2q ¼ 15.6 , 26 , 27 , 30.3 , 32 , 39.6 and 43.3 could be ascribed to La(OH) 3 and the peaks at 2q ¼ 25 , 29.5 , 37.9 , 54 and 58 were consistent with La 2 O 3 or organic lanthanum (R-La-OH), according to the lanthanum crystallographic letters, as discussed in the literatures. 34,35oreover, the peaks of the crystalline structure of lanthanum exhibited a weak intensity because of the low content of lanthanum in the LBW composite.Overall, the micro amount of La doped in LBW contributed to signicant deuorination. 28.1.3.FTIR analysis.The FTIR spectra of Na-BW, LBW before and aer uoride adsorption are compared in Fig. 3.The major characteristic bands for Na-BW are listed as follows: the peaks around 3455 cm À1 were the -OH bond stretching vibration modes of free water (surface-adsorbed water) and HAP.1,24 The peaks at 1635, 1416 and 871 cm À1 were attributed to stretching vibration of C]O, COO-and CO 3 2À , 1,26,36,37 respectively.Moreover, the peak at 1045 cm À1 corresponded to the stretching vibration of PO 4 3À , 37 and the peaks at 569 and 663 cm À1 were the bending vibration of P-O.38 Compared to the Na-BW sample, the major characteristic bands for the LBW sample before uoride adsorption exhibited no signicant change in the modied process.However, the strong increasing intensity at 3443 cm À1 in the LBW sample before uoride adsorption, which corresponded to -OH stretching vibration, suggests the formation of lanthanum hydroxide (La(OH) 3 ) during the modication process due to the ion-exchange reaction between Na + and La 3+ .39 The introduction of lanthanum was benecial for deuorination.Aer uoride adsorption, the slight decrease in intensity at 3453 cm À1 and the occurrence of blue shi from 3443 to 3453 cm À1 explained the fact that uoride might interact with the -OH groups present on LBW surface. 1 Furthermore, the peak of PO 4 3À shied from 1035 to  1045 cm À1 and the slight decrease of intensity at 569 and 603 cm À1 of P-O conrmed the hydroxyl groups on PO 4 3À position 37,38 or the ligand exchange reaction between uoride ion and hydroxyl group on the surface of the LBW.39 3.1.4. XPSnalysis. The XPS spctra of LBW composite before and aer uoride adsorption were examined to gain insights into the adsorption mechanism.As shown in Fig. 4a, the XPS survey spectrum of the LBW composite before uoride adsorption clearly conrmed the presence of C, O, Cl, P, La and Ca.The occurrence of a new peak at $685 eV in high resolution aer uoride removal was assigned to the F 1s spectrum, indicating that uoride was bound to the adsorbents.24,40 This result was further conrmed by the detailed F 1s spectrum in Fig. 4b.0,41 As shown in Fig. 4c, the detail O 1s spectrum was divided into three peaks, which were anion oxide (O 2À ), hydroxyl bonded to metal (M-OH) and adsorbed water (H 2 O), respectively.24,37 It is clear that the area ratio of hydroxyl groups (-OH) decreased from 33.7% to 20% aer uoride adsorption, which further indicated the supposition that uoride was adsorbed on LBW via the exchange of uoride ions with hydroxyl groups on the adsorbent surface.37,41 In addition, the area ratio of anion oxide (O 2À ) increased from 36.8% to 57.8% aer uoride adsorption, indicating that amount of organic uoride bonded to metal (R-M-F) formed aer adsorption.
Fig. 4d exhibits the high resolution La 3d region in the XPS spectrum of the LBW composite before and aer uoride adsorption.It shows that there were two sets of peaks in the La 3d region due to spin-orbit interaction and each of these peaks has a doublet of comparable intensity.The binding energies of the primary La 3d 5/2 and La 3d 3/2 bands were at 835.9 eV and 852.7 eV, respectively, and the corresponding satellites were located on their higher binding energy side at 839.2 eV and 856.3 eV, respectively. 42The spin-orbit splitting of the primary La 3d 5/2 and La 3d 3/2 bonds was 16.8 eV.These satellite peaks were observed due to the transfer of an electron from O 2p to the empty 4f shell of La, leading to the 3d 9 4f 1 nal state. 43These results conrm the presence of R-La-OH, La 2 O 3 or La(OH) 3 .It is found that the binding energies of all peaks of La 3d shied to the higher binding energy side aer uoride adsorption, which was ascribed to the fact that the electronegativity of uoride ions (3.98) was more than chloride ions (3.16). 44Therefore, ion exchange between F À and Cl À might be another approach for uoride removal.

Effect of pH
The pH of solution was a vital parameter affecting the chemical species of the solute and the surface properties of the adsorbent, such as surface charges. 45The effect of pH on uoride adsorption and zeta potential of the surface of the LBW adsorbent before and aer uoride adsorption are shown in Fig. 5.In the present study, the pH of the point of zero charge (pH zpc ) for the LBW composite was 11.4.This means that there are abundant positive charges on the surface of LBW complex when the pH is less than 11.4, which is favorable for the electrostatic attraction of negative uoride ions.However, the pH zpc of the LBW composite has decreased from 11.4 to 6.5 aer uoride adsorption, suggesting that the negative uoride ions neutralized some positive charges on the adsorbent surface. 31As shown in Fig. 5, there was no signicant change in terms of uoride removal efficiencies (exceeding 91%) over a wide pH range of 2.5-10.0(less than pH zpc ), indicating that electrostatic attraction might play an important role in uoride removal. 46evertheless, the uoride removal efficiency decreased obviously with further increasing of the initial pH (pH > 10), and the uoride removal efficiency decreased to 81.9% at a pH of 11.5.The decrease in uoride adsorption is probably ascribed to the following facts: (i) the negatively charged adsorbent surface failed to adsorb uoride ions due to electrostatic repulsion; (ii) abundant hydroxide ions led to a competition of uoride ions for adsorption sites. 47Furthermore, the nal pH of solution was located between 5.8 and 7.0 aer uoride adsorption, suggesting that the LBW composite had a certain buffer capacity.This might be attributed to the presence of carbonate in the LBW composite, which could consume H + in acid condition.While the ion exchange between Cl À on the LBW surface and OH À in aqueous solution could decrease the solution pH at alkaline condition. 45In conclusion, LBW composite showed great water stability and could maintain high efficiency for uoride removal in a wide pH range.

Effect of co-existing oxygen anions
Some co-existing oxygen anions exist in groundwater may compete with uoride ions for adsorption sites, leading to a decrease in uoride removal efficiency. 48,49The effects of coexisting oxygen anions that are typically present in groundwater on uoride removal are shown in Fig. 6, and the ion concentrations and pH values of solution before and aer uoride adsorption for co-existing oxygen anions are shown in Table 1.It was found that the concentrations of nitrate, sulfate and phosphate reduced to different degrees (Table 1), indicating that these anions were also removed during the uoride adsorption, especially for phosphate removal.The competitive adsorption between PO 4 3À with F À appeared to be evident.This is because that La 3+ in the adsorbent had high affinity capacity for PO 4

3À
. However, the occurrence of nitrate (NO 3

À
) had hardly any adverse effects on uoride removal in the range of 0-100 mg L À1 (Fig. 6), which might be a result of the large amounts of available adsorption sites on the adsorbent and the LBW composite has good affinity for uoride. 31The pH values of solution before and aer uoride adsorption were less than pH zpc , and it was also proving that these anions were adsorbed due to electrostatic attraction and the adsorption ability gradually enhanced with the increase in charge (NO 3 À < SO 4 2À < PO 4 3À ). 46In addition, carbonate (CO 3

2À
) showed little effect on uoride removal at low concentrations, but had a relatively obvious impact at high concentrations.This phenomenon was explained by the following facts: (i) it was considered that La 3+ had good coordination ability for F À and OH À .Thus, the pH of solution signicantly increased due to CO 3 2À hydrolysis at high  concentration, and then abundant OH À ions easily associated with La 3+ forming strong surface complex and reducing surface potential.(ii) OH À ions produced from CO 3 2À hydrolysis led to increased hindrance to the diffusion of uoride ions. 50Typical concentration of CO 3 2À in natural groundwater were much lower than that of other co-existing anions.Therefore, the interference of CO 3 2À in natural environments could be negligible.Moreover, the nal pH values of solution aer uoride adsorption for these anions concentrated in the range of 5.8-6.7,which further indicates that the LBW composite had a good buffer capacity. 45

Kinetic studies
In order to further understand the mechanisms of uoride adsorption on the LBW composite, the experimental data are tted with the pseudo-rst-order (PFO) and pseudo-secondorder (PSO) kinetic models. 1 The PFO and PSO kinetic models are expressed as: where k 1 (min À1 ) and k 2 (g mg À1 min À1 ) are the rate constants of PFO and PSO, respectively, and t (min) is the adsorption time.
The amount of uoride uptake with time at different concentrations is depicted in Fig. 7a.The adsorption process was extremely rapid.The adsorption reached to equilibrium within 5 min at a low uoride concentration of 10 mg L À1 , and the adsorption capacity remained essentially unchanged aer 20 min at high uoride concentration of 50 mg L À1 .This means that the times required to reach equilibrium lengthened with the increasing of initial uoride concentration.The kinetic parameters obtained from the PFO and PSO are shown in Table 2.It can be clearly seen that the rate constants provided by PFO and PSO vary with initial uoride concentration, suggesting that k 1 and k 2 were the observed rate constants of the overall adsorption reaction rather than the intrinsic rate constants.The values of k 1 and k 2 decreased with increasing initial uoride concentration, further suggesting that the reaction was faster at lower initial concentrations.Compared with PFO, the predicted values provided by PSO were more close to the experimental values, and PSO had lower RSS, c 2 values and higher Adj.R 2 value, indicating that uoride adsorption on the LBW composite followed the PSO kinetic models well.These results indicate that uoride adsorption on the LBW adsorbent was typical of a chemisorption process involving valence forces through sharing or the exchange of electrons between uoride ions and LBW adsorbent. 51Furthermore, the adsorption rate was normally dependent upon the number of available adsorption sites on the adsorbent surface and eventually controlled by the attachment of uoride ions on the surface. 39oreover, it has been reported that the adsorption kinetics at the solid/solution interface with different types of surface sites and with different adsorption affinities could be also described by the fractal-like approach. 52The nonlinear forms of fractal-like pseudo-rst-order (F-L PFO) and fractal-like pseudosecond-order (F-L PSO) rate equations are given as: (5) where k 0 1 (min À1 ) and k 0 2 (g mg À1 min À1 ) are the F-L PFO and F-L PSO rate constants, respectively, and h is the fractal exponent.
The curve t of the F-L PFO and F-L PSO rate equations is described in Fig. 7b and their kinetic parameters are also listed in Table 2. Fluoride adsorption at different concentrations on LBW adsorbent was very rapid in the initial phase, followed by a short stage until adsorption equilibrium was reached.This result was consistent with that from PFO and PSO kinetic models.It is evident from Table 2 that the F-L PSO provided a satisfactory tting with a high Adj.R 2 value and low RSS and c 2 values compared with the F-L PFO.This means that uoride adsorption on the LBW composite also followed the F-L PSO kinetic models well.In addition, the rate constants predicted by the F-L PSO slightly vary at different concentrations compared with the PSO.This result supports the nding that the rate constant of the PFO and PSO were time dependent parameters.
In order to providing a further contribution in illustrating the effect of the adsorbent structural properties on the temporal variation of the intraparticle diffusion coefficient during adsorption of micropollutants, the nonlinear forms of the diffusive Vermeulen model (VER) and the fractal-like Vermeulen model (FVER) equations are given as: 53 QðtÞ ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where Q and d s (mm) are the surface coverage degree with respect to equilibrium and the Sauter particle diameter, respectively, and D p (m 2 min À1 ) and D 0 p (m 2 min À(1Àh) ) are the intraparticle diffusivity and the fractal diffusion kinetic constant, respectively.
A comparison of the kinetic adsorption proles at high uoride concentration of 50 mg L À1 obtained from VER and FVER models is described in Fig. 7c.The main kinetic parameters are listed in Table 3.It can be found that the FVER model provided a more accurate prediction of the kinetic adsorption data to comparing with the VER one because of lower values of the HYBRID error function and higher values of the Adj.R 2 (>0.99).When the VER and FVER models are observed for t > 1 min, the VER model underestimates the adsorption data for shorter times.For example, Q(t) at t ¼ 2 min equals 0.83 from experiments while it is 0.82 and 0.79 for the FVER and VER models, respectively.Therefore, the FVER model has been able to provide a well match with the experimental data in the whole time range.Especially in the short time at initial reaction stage of uoride removal, the FVER model has proved that the adsorbent has a larger surface coverage degree with respect to equilibrium at initial times during adsorption of uoride.However, the VER model has an undervaluation on the degree.

Equilibrium studies
Adsorption isotherms reect the distribution of adsorbate molecules between solid and solution phases at equilibrium under the given conditions.The mathematical analyses of the process offered the information about adsorption capacity, as well as the surface properties and affinity of the adsorbent, contributing to insights into adsorption mechanisms and the optimization of process design. 51The Langmuir, 2 Freundlich 6 and two-site Langmuir 48 isotherm models are employed to analyze equilibrium data of uoride removal by the LBW adsorbent at different temperature, which are expressed as: ) where Q max (mg g À1 ) is the maximum adsorption capacity, K L (L mg À1 ) and K F ((mg g À1 ) (L mg À1 ) 1/n ) are the Langmuir and Freundlich constants, respectively, n is an empirical parameter, Q 1 (mg g À1 ) and Q 2 (mg g À1 ) are the maximum adsorption capacity of the high and low bonding energy sites ( and K 1 (L mg À1 ) and K 2 (L mg À1 ) are the corresponding affinity coefficients, respectively.The isotherm plots of uoride adsorption on the LBW composite were described in Fig. 8, and the obtained isotherm parameters from Langmuir, Freundlich and two-site Langmuir models are displayed in Table 4.The shape of the general isotherm plot of C e against q e gave an indication of whether the  adsorption was favorable or unfavorable. 46It is described that adsorption capacity increased with the temperature raising from 25 to 45 C in Fig. 8, suggesting that this reaction process was an endothermic reaction, and elevating temperature was favorable for uoride adsorption. 39The equilibrium parameter R L , which is dened as R L ¼ 1/(1 + bC 0 ), where b is the Langmuir constant and C 0 is the initial concentration of uoride.The R L value indicated the shape of the isotherm, and in the range of 0 < R L < 1 reected the favorable adsorption process. 54The R L value for the adsorption of uoride on different adsorbents at initial concentration of 5 mg L À1 (lowest concentration studied) and 120 mg L À1 (highest concentration studied) were listed in Table 5.This indicated the fact that the adsorption process was very favorable and the adsorbent had a good potential for uoride removal. 55ompared with the Freundlich model, the results in Table 4 showed that the Langmuir isotherm model tted the experiment data well according to the smaller RSS and c 2 values and larger Adj.R 2 value.This indicated that the reaction was monolayer adsorption on the structurally homogeneous surface  of the LBW adsorbent in the uoride removal process. 26The predicted maximum adsorption capacity of uoride was 8.32 mg g À1 at the pH of 7.0 and the temperature of 45 C. The two-site Langmuir model was also suitable for describing uoride adsorption on the LBW adsorbent due to its lower RSS and higher Adj.R 2 value at different temperature, suggesting that there might be two types of active sites with different binding energies on the adsorbent surface.Moreover, Q 1 > Q 2 suggests that most of the adsorption sites could be characterized as high affinity sites and K 1 > K 2 indicates that the higher energy sites had a much higher affinity for uoride. 48This model showed higher predicted maximum adsorption capacity of uoride to 8.96 mg g À1 at the pH of 7.0 and the temperature of 45 C. The uoride adsorption capacities of other natural adsorbents were previously reported.For example, the uoride adsorption capacity was 0.122 mg g À1 and 0.226 mg g À1 for raw lamb bones and chicken bones, 10 2.71 mg g À1 for bone char, 25 7.32 mg g À1 for pyrolytic bone char at 700 C, 26 0.96 mg g À1 for activated alumina, 25 2.30 mg g À1 for activated alumina doped cellulose acetate phthalate (CAP) mixed matrix membrane, 56 3.192 mg g À1 for aluminum impregnated coconut ber ash, 32 5.16 mg g À1 for bauxite, 46 0.075 mg g À1 for activated carbon, 25 respectively.Compared with other reported materials, LBW showed better performance on uoride removal.

Field study
In this study, the feasibility of the LBW complex for deuoridation and the change in the concentrations of common anions were also tested in actual groundwater.The uoridecontaminated groundwater was collected from different locations (villages) near a mine area of Deyang City, Sichuan Province, China.The detailed results of these samples before and aer addition of the LBW complex are presented in Table 6.The high F À concentrations in groundwater were obtained in a bore well (8.43 mg L À1 ), spring (4.49mg L À1 ) and dug well (3.43 mg L À1 ) from a peasant household.It is not difficult to nd that uoride concentrations from sampling points were above drinking water standard (1.5 mg L À1 , WHO 2011). 7Longterm uoride exposure in drinking water might cause adverse impacts on human health in this area.Batch adsorption studies were also carried out under identical experimental conditions using 1.0 g of LBW for a 100 mL of sample, and the time of constant agitation was at 20 min at normal groundwater temperature (15-18 C).It was found that the F À levels in three groundwater samples were less than drinking water standard (1.5 mg L À1 , WHO 2011) aer addition of the LBW complex.These results indicated that the LBW composite had little interference from other co-existing ions and showed better removal performance for uoride especially at low concentrations. 34Furthermore, it is evident from Table 6 that the LBW composite could effectively remove bicarbonate (HCO 3 À ) and phosphate (PO 4 3À ) from eld groundwater simultaneously.The removal efficiency of sulfate (SO 4 2À ) by the LBW composite was higher than that of NO 3 À .These results were in conformity with the effect of co-existing oxygen anions  in Fig. 6 and Table 1.Thus, the LBW composite has the potential for practical application in uoride-contaminated groundwater.

Desorption and regeneration
The above study suggested that uoride was poorly adsorbed by LBW adsorbent in basic solution.Therefore, desorption of uoride from the exhausted adsorbents can be easily obtained using slightly alkaline solution. 39In this research, desorption of uoride from LBW composite was carried out with different concentrations of Na 2 CO 3 solution.The uoride desorption rates were listed in Table 7, which showed that the released uoride increased with the increase of Na 2 CO 3 concentration.With the Na 2 CO 3 concentration of 1 mol L À1 , 92.96% of uoride was desorbed.Five consecutive adsorption-desorption cycles were performed to investigate the regeneration and recyclability of the LBW composite.The results were depicted in Fig. 9.
When the initial uoride concentration was 10 mg L À1 , aer ve cycles, the uoride removal were 93.68%, 88.41%, 81.32%, 75.77%, and 71.18%, respectively, which can be found that the value of uoride removal changed small.Therefore, the adsorbent was suitable for regeneration and reusing.

Mechanism of uoride adsorption
It was seen from kinetic studies that the fast adsorption rate and the short adsorption equilibrium time indicated that the density of active sites on the LBW composite was relatively high. 39SEM images showed an irregular surface and porous structure on LBW, and BET analysis revealed that the LBW composite had a large specic surface area (85.4 m 2 g À1 ).These surface mesoporous and micropores aid uoride adsorption.
The XRD patterns conrmed the presence of lanthanum oxide or hydroxide in LBW composite, and the micro amount of La doped contributed to signicant deuorination.The FTIR spectra conrmed the presence of abundant -OH groups on the LBW composite and the zeta potential showed a positively charged surface over a wide pH range (pH < 11.4).Consequently, the protonation of these -OH groups and positively charged surface facilitated uoride removal due to electrostatic attraction and ion exchange.The high resolution XPS spectra of O 1s and La 3d showed that the content of hydroxyl groups (-OH) reduced and the peaks of La 3d shied to the higher binding energy side aer uoride adsorption.These results indicated that uoride ions might be removed by ion exchange between F À and -OH or Cl À on the LBW composite.
The possible mechanisms of uoride adsorption on LBW composite are depicted in Fig. 10.Sodium ions in Na-BW promoted the composition between lanthanum ions and bone waste particles when LaCl 3 $nH 2 O was added.This might form various forms of lanthanum, including R-La-OH, La 2 O 3 or La(OH) 3 , on the surface or in the mesoporous of the LBW composite through ion-exchange between Na + and La 3+ ions.Multivalent metal ions, especially lanthanum, show a strong affinity towards uoride ion, meaning the uoride ions in the solution might get exchanged for the OH À or Cl À ions present in  the LBW composite.F À and OH À are isoelectronic with comparable ionic radii, thus the ion exchange between F À and OH À tended to happen and formed strong chemical bonds with La 3+ or Ca 2+ ions. 57M-OH + F À / R-M-F + OH À M-OH + F À / M-F + OH À M-Cl + F À / M-F + Cl À (14) In addition, the metal ions were loaded on the adsorbent lattice and then a developed charged surface through amphoteric dissociation. 58Positively charged surface sites were maintained when the solution pH was less than the pH zpc of the adsorbent, which could attract the negatively charged uoride ions by electrostatic attraction, resulting in enhanced uoride removal. 59In this study, the presence of positively charged Ca 2+ or La 3+ ions in the LBW composite exhibited an attractive tendency towards the negatively charged uoride ions by means of electrostatic adsorption forming organic uoride bonded to metal (R-M-F) as follows: R-M-OH 2 + + F À / R-M-OH 2 /F (15) However, with the increase of solution pH, the positively charged sites on the LBW surface reduced or even became negatively charged (pH > pH zpc ), leading to the weakening of protonated hydroxyl, and thus ion exchange between F À and OH À or Cl À might become the predominant deuorination mechanism.

Conclusions
In this study, a LBW composite was synthesized and tested to remove uoride in aqueous solution.The introduction of lanthanum could improve the porous structure of the adsorbent and enhance the deuorination capacity of bone waste.The LBW composite had good temperature resistance and a buffer capacity for polluted water in an initial pH range of 2.5-10.Meanwhile, the LBW composite exhibited a good affinity for uoride and the co-existing oxygen anions had hardly any adverse effects on uoride removal in the range of 0-100 mg L À1 .The uoride adsorption on the LBW composite was fast and the reaction could reach equilibrium within 20 min.Kinetic studies revealed that the adsorption followed PSO and F-L PSO equations well, indicating that it was a typical chemisorption process.The FVER model provided a more accurate prediction of the kinetic adsorption data in the whole time range to comparing with the VER model.Isotherm studies showed that the uoride adsorption was monolayer adsorption occurred on homogeneous LBW surface and the maximum adsorption capacity was about 8.96 mg g À1 .The results of desorption and reuse research indicated that the LBW adsorbent could be employed as a promising adsorbent for uoride removal from groundwater.The possible mechanism might be the combination of electrostatic attraction and ion exchange.In summary, the LBW composite provides the potential of practical application for uoride removal from contaminated groundwater.

Fig. 4
Fig. 4 (a) Wide-scan XPS spectra of LBW before and after fluoride adsorption, XPS spectra of (b) F 1s after fluoride adsorption, (c) O 1s and (d) La 3d before and after fluoride adsorption.

Fig. 5
Fig. 5 Effect of pH on fluoride adsorption and zeta potential of the surface of the LBW adsorbent before and after fluoride adsorption.

Fig. 6
Fig. 6 Effect of co-existing oxygen anions on fluoride removal efficiency.

Table 1
The changes in concentration and solution pH before and after fluoride adsorption for co-existing oxygen anions a a Nil: not detected below detection limit.

Table 2
Kinetic parameters obtained from PFO, PSO, F-L PFO and F-L PSO

Table 4
Isotherm parameters obtained from Langmuir, Freundlich and two-site Langmuir models

Table 5
Equilibrium parameter R L for adsorption of fluoride at different temperature

Table 6
The detailed results of actual groundwater samples before and after addition of the LBW a a ND: not detected above detection limit.Nil: not detected below detection limit.

Table 7
Percentage desorption of fluoride from LBW composite by different concentrations Na 2 CO 3 solution