Recovery of Calcium Fluoride from Highly Contaminated Fluoric/Hexa Uorosilicic Acid Wastewater

Recovery of Calcium Fluoride from Highly Contaminated Fluoric/Hexa uorosilicic Acid Wastewater

Masao Morita1, Giuseppe Granata2 and Chiharu Tokoro2,*

1School of Creative Science and Engineering, Waseda University, Tokyo 169–8555, Japan 2Faculty of Science and Engineering, Waseda University, Tokyo 169–8555, Japan

We investigated the recovery of calcium uoride (CaF2) from highly concentrated hexa uorosilicic acid wastewater by neutralization– purication. Neutralization was achieved by dosing calcium hydroxide, whereas purication was carried out by alkaline leaching with sodium hydroxide. X-ray diffraction, X-ray absorption ne structure, mineral liberation analyzer, Fourier transform infrared spectroscopy and Inductively-coupled plasma atomic emission spectroscopy were used to quantify the neutralization and leaching performance and to elucidate their mechanisms. The precipitation behavior was strongly dependent on the calcium (Ca)/silicon (Si) molar ratio. For a Ca/Si ratio of 1.12, approximately 25% of the total uorine was precipitated selectively as CaF2. By increasing the Ca/Si ratio to 3.91, the recovery yield in- 2− creased to 100% because of the precipitation of CaSiF6 and the hydrolytic decomposition of hexa uorosilicate ion (SiF6 ) to CaF2. The hy- 2− drolytic decomposition of SiF6 resulted in the precipitation of silicon dioxide on the surface of the previously formed CaF2. Alkaline leaching by sodium hydroxide at 70°C resulted in an efcient removal of the silicon phase from the neutralized sludge with the formation of a CaF2 product with a grade above 90%. Leaching parameters, such as the kinetic constant and the activation energy, were determined by assuming 2− rst-order kinetics. The residual silicon phase in the nal product could not be dissolved because of the formation of non-leachable SiO3 . [doi:10.2320/matertrans.M-M2017850]

(Received June 3, 2017; Accepted October 30, 2017; Published December 8, 2017)

Keywords: uorine, calcium uoride, recovery, neutralization, mechanism

1. Introduction to the immobilization operations that are used most fre- quently, such as coagulation, ion-exchange and adsorp- Fluorine (F) is used widely in many industrial elds, such tion15). Fluorine recovery from high concentrations of uo- as in the manufacture of hydro uoric acid (HF), uorocar- rine wastewater mitigates the depletion of natural raw bons and glass. A natural source of uorine is mineral ore material. For these reasons, the selective recovery of uorine that contains uorite (CaF2), which is processed mostly in from CMP wastewater is highly anticipated. China and Mexico1). Because of restrictions in the export of Previous researchers have focused on the recovery of uo- 2) rare-earth metals from China between 2005 and 2010 , it is rine as CaF2 from wastewater that is generated in the silicon believed that a short supply of uorine could result because industry16) or in fertilizer plants17) by neutralization/precipi- of the international geopolitical situation3). Therefore, F re- tation. However, in their research, the only source of uorine 7,16) cycling in addition to mineral-ore availability is desirable to was HF or, when hydrogen hexa uorosilicate (H2SiF6) ensure a stable F supply4). was also present, the total uorine concentration was lower 18) Fluoride can be recovered from wastewater that is gener- than ~1 g/L . This implies a direct recovery of CaF2 with- ated in the manufacture of glass5), semiconductors6), hydro- out any purication from the silicon impurities. 7) 8) uoric acid , organic uorine chemical products and fertil- The goal of this research was to recover uorine as CaF2 9) izers . An ideal source of recoverable uorine is wastewater from a highly concentrated H2SiF6 CMP wastewater by pre- that is produced in the chemical mechanical polishing cipitation/neutralization with calcium hydroxide (Ca(OH)2). 10) (CMP) of silicon wafers ; this source is even more promis- Ca(OH)2 results from the lower solubility of calcium uo- −11 3 −3 ing considering the constant expansion of the electronics ride (KsCaF2 (25°C) = 3.46 × 10 mol × L ) compared with industry11). other uorides (e.g., the solubility of sodium uoride ex- 19) A successful selective recovery of uorine as CaF2 from ceeds 40 g/L at 20°C) . Unlike in previous research, we fo- CMP wastewater would be benecial from an economic and cused on wastewater that exceeded 100 g/L H2SiF6, which environmental perspective. CaF2 that is used in the produc- can contaminate CaF2 with silicon impurities because of the tion of HF requires a grade (acid grade) higher than possible precipitation of SiO2 and CaSiF6. Therefore, to re- 12) 97 mass% , whereas that to be used in steel manufacture cover CaF2, we investigated the precipitation behavior and must have a CaF2 content of 60–72.5% (metallurgical the formation of silicon compounds in the neutralization grade)13). Because the achievement of these grades requires process. the removal of impurities, such as silicon dioxide (SiO2), X-ray diffraction (XRD), X-ray absorption ne structure calcium carbonate and phosphorus14), the recovery of uo- (XAFS), mineral liberation analysis (MLA), ion-chromatog- rine would reduce the costs related to the purication opera- raphy (IC) and inductively-coupled plasma spectrometry tions signicantly. Because an excessive absorption of uo- (ICP-AES) were used to investigate the in uence of the Ca/ rine is hazardous for human health, the recovery of uorine Si molar ratio on the crystallization behavior and the recov- from wastewater would lower the remediation costs related ery performances. We also investigated the removal of silicon impurities * Corresponding author, E-mail: [email protected] from the precipitation sludge by alkaline leaching with so- Recovery of Calcium Fluoride from Highly Contaminated Fluoric/Hexa uorosilicic Acid Wastewater 291 dium hydroxide (NaOH). NaOH was preferred over other 2.2 Alkali leaching experiment possible strong bases, such as potassium hydroxide (KOH), The neutralized sludge was dried at 45°C and crushed by magnesium hydroxide (Mg(OH)2) and Ca(OH)2 because of using a mortar to obtain 850–1700-µm samples for leaching its lower cost and because of its ability to dissolve silicon by NaOH. The 1 M NaOH solution that was to be used as a without forming insoluble silicate salts (e.g., the solubility leaching agent was heated on a hot stirring plate to the de- of calcium silicate is 0.095 g/L20)). The in uence of tem- sired temperature. The leaching pulp density was xed at perature on leaching products and performance was eluci- 50 g sample/L NaOH. After leaching, solid–liquid ltration dated by Fourier-transform infrared spectroscopy (FT-IR) was performed with 0.10-µm membrane lters and a kinetic analysis of the reaction was performed by cou- (H010A090C, ADVANTEC). The Ca and Si concentrations pling experimental data with thermodynamic calculations by were measured by ICP-AES, whereas the F concentration using PHREEQC software (User’s Guide to PHREEQC). was determined by IC. The leaching residues were washed and dried at 45°C for 24 h. After drying, Fourier-transform 2. Materials and Methods infrared spectroscopy (FT-IR4200, JASCO, Japan) was used to investigate the silica transformation after leaching. FT-IR 2.1 Neutralization test analyses were carried out by using the transmission method, Neutralization tests were performed at room temperature by diluting the sample with potassium bromide (KBr) to on 1-L samples of synthetic wastewater solutions that con- 0.3 mass%. The sample and KBr were mixed in an agate tained 1.28 M HF and 0.90 M H2SiF6 that were prepared by mortar and pressed into a 10-mm-diameter pellet. The mea- mixing 46.5 mL of 48% HF, 232.5 mL of 40% H2SiF6 and surement was conducted with a scanning speed and a resolu- −1 −1 pure water (Aquarius RFD 240NA，ADVANTEC, Japan). tion of 2 mm·s and 4 cm , respectively. Amorphous SiO2 Synthetic wastewater was prepared based on the concentra- and Na2SiO3 were measured as references. tions of a real wastewater that was produced in Japan from the cleaning of glass substrates. In the neutralizations tests, 3. Results and Discussion 15 mass% lime hydrate (Ca(OH)2) slurry was dripped at 100 mL/h (0.20 mol-Ca/30 mins) into the wastewater solu- 3.1 Neutralization tion with stirring. After 210 min, the suspension was centri- In the neutralization tests, for a Ca/Si molar ratio below fuged (Himac CR-21, Hitachi, Japan) for 10 min at 1.12, only a small amount of Si (2.42%) precipitated 5000 rpm. Suction ltration was conducted with 0.10-µm (Fig. 1). By increasing the Ca/Si molar ratio, the amount of membrane lters (A010A090C, ADVANTEC). The concen- precipitated Si also increased. However, for a Ca/Si molar trations of Ca and Si in the ltrate were measured by ICP- ratio of 3.91, wastewater neutralization was achieved (pH AES (SPS-4000, Seiko Instruments, Chiba, Japan), whereas 7.43) along with a quantitative removal of all components as the F concentration was determined by IC (IC850, Metrohm, precipitates. Up to a Ca/Si ratio of 1.12, whereas ~25% of Tokyo, Japan). the uorine was precipitated, almost all silicon remained All chemicals used were analytical-grade reagents from dissolved in the wastewater. This result suggests that the re- Wako-Chemical, Japan. moved uorine was uoride ion from HF, which precipitated Sludge that was generated in the neutralization process as CaF2. In our wastewater solutions, uoride com- was dried at 25°C and analyzed by XRD (RINT Ultima III, prised ~25% of the total uorine, whereas the remaining 2− Rigaku, Japan) to determine the phase composition. To iden- 75% was hexa uorosilicate ion (SiF6 ), which is well- tify the chemical form of silicon in the sludge, XAFS analy- known for its difcult decomposition10) and higher solubility 2 −2 22) sis (BL6N1 in Aichi Synchrotron Radiation Center) was (KsCaSiF6 (25°C) = 0.303 mol × L ) . conducted by using a uorescence method. Specically pre- The XRD spectra of the neutralized sludge conrmed that 21) pared CaSiF6 and amorphous SiO2 (Kanto Chemical, the precipitated solid was crystalline CaF2 (ICSD:00-035- highest quality of JIS, Japan) were measured as reference 0816). By increasing the Ca/Si ratio to 3.36, the XRD peak materials. of CaF2 sharpened as a result of the increased crystallinity. MLA was conducted to elucidate the evolution of mineral species in the sludge during the precipitation of CaF2 and silicate. The MLA equipment was composed of a scanning electron microscope (SEM) (Quanta 250G, FEI) and an energy dispersive X-ray spectrometer (EDS) (XFlash Detector 5030, Bruker). To perform the MLA analysis, solid samples were molded into resin blocks whose surface had been pol- ished by using sandpaper. Diamond paste and DP-Lubricant Blue (Struers, Japan) were used for the nal polishing. After polishing, the sample was coated (SC-701, SANYU ELECTRON, Japan) to prevent charging. MLA mapping, was performed in the grain X-ray mapping measurement mode (GXMAP) mode because of the close back scattered electron (BSE) value of the CaF2 and the silicate. Fig. 1 Distribution ratio of each elements to solid against total amount of input by neutralization at different Ca/Si molar ratios. 292 M. Morita, G. Granata and C. Tokoro

At a Ca/Si ratio of 3.91, the CaF2 peak broadened again, sisted of CaSiF6, whereas at a Ca/Si ratio of 3.91, ~95% of which suggests that the CaF2 precipitate could be contami- silicate was present as amorphous SiO2. These results sup- nated by other compounds. This evidence implies that, up to port the abovementioned discussion on the enhancement of 2− a Ca/Si ratio of 1.12, neutralization can ensure a partial se- hydration and decomposition of SiF6 by CaF2 precipitation 22) lectivity for uorine removal. for large additions of Ca(OH)2 . The pH decreased slightly up to a Ca/Si ratio of 2.24 and From XRD and XANES analyses, we see that neutraliza- then increased again (Fig. 1). This evidence suggests that tion by Ca(OH)2 at a lower Ca/Si molar ratio was controlled 2− SiF6 decomposed to HF and Si(OH)4 because of the hy- by the direct precipitation of CaF2 and CaSiF6 as in (2) and 2− 18) dration of SiF6 as shown in eq. (1) : (3). 2 + SiF − + 4H O 4H + 6F− + Si(OH) (1) 2HF + Ca(OH) CaF + 2H O (2) 6 2 → 4 2 → 2 2 where the hydration and decomposition of SiF 2− could be 6 H SiF + Ca(OH) CaSiF + 2H O (3) enhanced by the simultaneous removal of uoride from 2 6 2 → 6 2 solution as CaF2 precipitate. The pH increase that was ob- By adding more Ca(OH)2, the neutralization was con- served from a Ca/Si ratio of 2.24, along with the quantita- trolled by the precipitation of CaF2 and amorphous SiO2 as 2− tive precipitation of all elements at a Ca/Si ratio of 3.36, result of the hydrolytic decomposition of SiF6 , as in (4). 2− suggests that the hydrolytic decomposition of SiF6 was H2SiF6 + 3Ca(OH)2 3CaF2 + SiO2 + 4H2O (4) complete and free uoride and Si(OH)4 could be precipi- → tated easily. For a proper choice of purication operation, it is conve- Because silicon compounds were not detected by XRD nient to know where the impurities are concentrated. For this 2− (Fig. 2), despite the hydrolytic decomposition of SiF6 , we reason, to elucidate whether the silicon products that form speculate that they precipitated as amorphous materials. To for a high Ca/Si molar ratio coat the surface of the precipi- elucidate this aspect, the edge region of the XAFS spectra tated CaF2, we conducted MLA analysis and performed im- (XANES) of the neutralized precipitate was studied at the Si age analysis of the SEM-EDS mapping (Fig. 4) to determine K-edge by measuring CaSiF6 and amorphous SiO2 as refer- the free-surface ratio of each precipitated particle. ence materials. The free-surface ratio of the precipitated CaF2 particles For the precipitate that was obtained at a Ca/Si ratio of 1.12, a double peak in the XANES region at ~1853 eV and 1855 eV was observed (Fig. 3). In comparison with the reference CaSiF6, we see that the double peak corresponded to CaSiF6. By increasing the Ca/Si molar ratio to 2.24, the CaSiF6 peak weakened, whereas another peak at ~1849 eV, which corresponded to amorphous SiO2 appeared. With a further increase of Ca/Si ratio to 3.91, the only peak that was observed in the XANES spectrum resulted from amorphous SiO2. These results conrm the previous hypothesis on the par- 2− tial precipitation of SiF6 as CaSiF6 in addition to CaF2 for low amounts of Ca(OH)2, and the selective precipitation of 2− CaF2 and amorphous SiO2 upon decomposition of SiF6 for larger amounts of Ca(OH)2. To quantify the CaSiF6 and amorphous SiO2 in the precipitate, a tting of the XANES spectra was conducted against reference CaSiF6 and amorphous SiO2. As shown in Table 1, at a Ca/Si ratio of 1.12, most silicon in the precipitate con-

Fig. 3 Si K-edge XANES spectrum of the precipitate from neutralization; (dashed lines are the result of tting with CaSiF6 and amorphous SiO2 reference materials).

Table 1 Molar ratio of CaSiF6 and amorphous SiO2 in precipitate; results obtained from tting of the XANES spectrum.

Ca/Si Amorphous SiO R value CaSiF (%) 2 Molar ratio 6 (%) 1.12 100 0 0.033 2.24 27.4 72.6 0.023 3.35 17.5 82.5 0.022

Fig. 2 XRD patterns of the neutralized sledges. 3.91 4.93 95.1 0.016 Recovery of Calcium Fluoride from Highly Contaminated Fluoric/Hexa uorosilicic Acid Wastewater 293 and their relationship with the Ca/Si molar ratio is shown in tatively, but the recovered CaF2 exhibits a lower grade. Fig. 5. Therefore, a process that involves the quantitative precipita- At a Ca/Si ratio of 1.12, CaF2 exhibited a mostly sili- tion of uorine at a high Ca/Si molar ratio followed by puri- con-free surface as a result of selective precipitation. At a cation seems to be the most desirable process option. Ca/Si ratio of 2.24, the free surface of the CaF2 phase de- Based on the abovementioned results that show the pres- creased partially. By increasing the Ca/Si ratio, a large part ence of mixed phases of CaF2–CaSiF6 coated by amorphous of the CaF2 phase was coated by silicate, which conrmed SiO2, we chose to purify CaF2 by alkaline leaching. that the precipitation of SiO2 followed the precipitation of CaF2 and CaSiF6. 3.2 Neutralization mass balance Based on results of the XANES tting and stoichiometric The XRD and XANES analyses indicate that the neutral- considerations ( uorine that was present as CaF2 and ization by Ca(OH)2 at a lower Ca/Si molar ratio was con- CaSiF6; silicon that was present as amorphous SiO2), we trolled by the direct precipitation of CaF2 and CaSiF6 as calculated the chemical composition of the precipitates that shown in eqs. (1) and (2). In particular, at a Ca/Si ratio of were obtained for different Ca/Si ratios. The results are 1.12, the silicon removal was ~2.5%, which corresponded to shown in Table 2. 0.59 mol/L of residual silicon in solution. In this case, the At a low Ca/Si molar ratio, a high grade CaF2 can be re- precipitates consisted mostly of CaF2 as shown in Table 2, covered, but the removal efciency towards the uorine is whereas the dissolved concentrations of uorine and cal- probably very low for practical applications. In contrast, for cium were 3.5 mol/L and 0.28 mol/L, respectively. Because a higher Ca/Si molar ratio, uorine can be removed quanti- the added Ca(OH)2 was consumed by the formation of insoluble CaF2 and neutral CaSiF6, whereas the hydrolysis of 2− + SiF6 released H in solution, the pH decreased slightly. For a Ca/Si ratio of 2.24, the silicon removal increased to ~30%, which corresponds to 0.31 mol/L of silicon in solution. According to the XANES tting, 84.11% of CaF2 and several percent of CaSiF6 and SiO2 were produced and 1.5 mol/L of uorine and 0.20 mol/L of calcium were obtained in solution. Therefore, in this case as well, the forma- 2− tion of CaSiF6 along with the increased hydrolysis of SiF6 prevented the pH from increasing. By adding more Ca(OH)2 (Ca/Si ratio of 3.36), the experimental silicon removal increased to ~90%, which corresponds to 0.03 mol/L of silicon in solution. In this case, Fig. 4 SEM-EDS mapping of the precipitate products. most uorine and calcium was precipitated (Fig. 1 and Table 2), which leaves 0.13 mol/L of uorine and 0.01 mol/L of calcium in an acidic solution. The further addition of Ca(OH)2 to a Ca/Si ratio of 3.96 yielded the complete removal of silicon and uorine with a consequent increase in pH because of the excess of Ca(OH)2.

3.3 Purication by alkali leaching: leaching performances To purify the uorine product that was obtained at a Ca/ Si molar ratio of 3.91, we performed an alkali leaching of the precipitated CaF2 by NaOH and investigated the effect of three different temperatures on the leaching performances. The results are shown in Fig. 6. According to the obtained results, regardless of the tem-

Fig. 5 Free surface ratio of precipitated CaF2 particles. perature, Si was leached from the precipitate preferentially, whereas F and Ca were dissolved only slightly. This evidence suggests that CaSiF6 and amorphous SiO2 were ex- tracted selectively from the precipitate along with smaller Table 2 Chemical composition of the precipitate obtained at different Ca/ amounts that released the uoride ion into solution. All ex- Si ratio (results based on XANES tting). tracted calcium precipitated as Ca(OH)2 (KsCa(OH)2 (25°C) = −6 3 −3 19) Ca/Si molar CaSiF6 SiO2 CaF2 F recovery 5.5 × 10 mol × L ) . ratio (%-g/g) (%-g/g) (%-g/g) rate (%) A higher temperature yielded an increased and faster sili- 1.12 6.42 0 93.58 23.45 con dissolution, as reported in previous research23). To eval- 2.24 8.48 7.41 84.11 56.72 uate important leaching parameters, such as kinetic con- 3.35 9.00 14.85 76.15 95.22 stants and activation energy, we assumed that the dissolution 3.91 2.53 16.11 81.36 99.97 of the solid products follows independent rst-order kinetics 294 M. Morita, G. Granata and C. Tokoro as given by eq. (5): that higher temperatures accelerated the leaching of impurities and in particular that of amorphous SiO . dC 2 = kC (5) To determine the activation energy for the alkali leaching dt − reactions that were involved in the purication operations, where C is the concentration of solid phase (SiO2, CaSiF6, we plotted the kinetic constant of Table 3 with the Arrhenius CaF2) and k is the kinetic constant. Based on this assump- eq. (6) in its linearized form (7): tion, the leaching parameters can be determined by tting a E typical integrated-linearized expression for rst-order reac- k = A exp a (6) −RT tion kinetics for the dissolution of amorphous SiO2, CaSiF6 and CaF . In the kinetic analysis, experimental data were 2 E coupled with chemical equilibrium calculations by using ln k = a + ln A (7) RT PHREEQC software24), with the Minteq.v425) database. The − obtained parameters are shown in Table 3, which conrm where k is the kinetic constant, Ea is the activation energy, R is the gas constant, T is the reaction temperature and A is the frequency factor. By a linear regression of the Arrhenius plot in Fig. 7, we determined the estimated activation energies for the leaching of CaSiF6 and SiO2 as 30.37 and 89.75 kJ/mol, respectively.

3.4 Mass balances and CaF2 grade of alkaline leaching Based on the initial composition of the neutralized sludge (Table 2), by considering the ICP results and the stoichiome- try of leaching reactions (8) and (9), CaSiF + 6NaOH 6NaF + Na SiO + Ca(OH) (8) 6 → 2 3 2 SiO + 2NaOH Na SiO + 2H O (9) 2 → 2 3 2 We performed a mass balance of solid species (Table 4) and calculated the grade of the nal CaF2 product as in (10):

mCaF2 CaF2(%) = 100 (10) mCaF2 + mSiO2 + mCa(OH)2 ×

where mCaF2, mSiO2 and mCa(OH)2 are the grams of solid products CaF2, SiO2 and Ca(OH)2 that were produced or were residues from leaching, respectively (calculated from Table 4 by multiplying the amounts as mmol for the molecu- lar weight of each chemical species).

3.5 Silicate dissolution mechanism: FT-IR investigation To elucidate the mechanism of silicate dissolution, FT-IR analyses of solid residues before, during and after alkaline leaching were performed. The results are shown in Fig. 8 along with the IR patterns of NaSiO3･9H2O and amorphous SiO2, which were analyzed as references. In general, the Si atom in silica has a tetrahedral coordina-

Fig. 6 Leaching rate of Si, F and Ca.

Table 3 Leaching parameters obtained by tting of the rst order kinetic equation.

Temperature CaSiF6 SiO2 CaF2 (K) (s−1) (s−1) (s−1) 298.15 2.0 × 10−3 8.5 × 10−5 − 318.15 5.0 × 10−3 1.5 × 10−3 − 343.15 1.0 × 10−2 1.0 × 10−2 3.0 × 10−6 Fig. 7 Arrhenius plot tting for CaSiF6 and SiO2. Recovery of Calcium Fluoride from Highly Contaminated Fluoric/Hexa uorosilicic Acid Wastewater 295 tion, with four oxygen atoms that surround a central Si in a recovery of uorine as CaF2 but with a lower grade. In con- shared SiO4 tetrahedra (net chemical formula SiO2) with trast, at higher temperatures, the uorine recovery was n 26) four bonds, namely (Q ), where n = 0,1,2,3,4 . As shown in lower, but the CaF2 grade was slightly higher. Fig. 8, Si-O bond (Q4) stretching around 1100 cm−1, which was identied in the initial precipitate and amorphous SiO2 Acknowledgements reference, shifted gradually to lower wavelengths (1000 cm−1) as the leaching proceeded for 30 min at 25°C. We thank the staff of the Aichi Synchrotron Radiation This shift corresponded to an increase in silicon portions that are bound with non-bonding oxygens (NBO)27,28). As a result, the Si-O stretching (Q3) peak appeared. After 2 h of leaching at 25°C, a new peak appeared at ~900 cm−1. This peak was assigned to the Si-O stretching (Q2)29). This Si-O (Q2) bonding was found also in the solid residues from high-temperature leaching and it was similar to the Si-O 1 bonding in Na2SiO3. A further shift to (Q ) (one more re- placement by OH−) would be required to transform the silicon compound to a soluble product30,31). Therefore, it is sug- gested that the nal chemical form of silicate in the leaching residue was not amorphous SiO2 but H2SiO3, NaHSiO3 or Na2SiO3. These results also highlight that leaching stopped as a result of the transformation of the Si-O bounding (Q4) to the Si-O bounding (Q2), which has a silicon site with non-bonding oxygens (Fig. 9).

4. Conclusion

The recovery of CaF2 from highly-concentrated hexa uorosilicic acid wastewater that consisted of 1.28 M HF and 0.9 M H2SiF6 by neutralization–purication was investi- Fig. 8 FT-IR spectra of the sludge before and after leaching at 25°C, 45°C gated. A rst operation of neutralization by Ca(OH)2 al- and 70°C. lowed for the removal of uorine from the liquid media. The precipitation behavior was found to be strongly dependent on the Ca/Si molar ratio. Although the best removal performance could be obtained for high Ca/Si molar ratios, the obtained sludge contained impurities, such as SiO2 and CaSiF6, and it required purication to achieve a higher CaF2 grade. The Si phase was removed from the precipitated CaF2 by alkaline leaching with NaOH for 10 min between 45 and 27) 70°C. Leaching at lower temperatures allowed for a larger Fig. 9 Stagewise breaking of silicon–oxygen bonds from SiO2 .

Table 4 Mass balance based on ICP and XANES results for the leaching at 25, 45 and 70°C.

Amount in Initial CaF Temperature Dissolution the solid Fluorine 2 Species amount grade (°C) (%) phase recovery (%) (mmol/g) (%) (mmol/g)

CaF2 10.42 0.6 10.35

CaSiF6 0.14 100 - 25 95.5 91.1 SiO2 2.68 57.4 1.14

Ca(OH)2 - - 0.14

CaF2 10.42 0.5 10.37

CaSiF6 0.14 100 - 45 95.7 91.4 SiO2 2.68 58.9 1.10

Ca(OH)2 - - 0.14

CaF2 10.42 3.3 10.07

CaSiF6 0.14 100 - 70 92.9 91.8 SiO2 2.68 62.7 1.00

Ca(OH)2 - - 0.14 296 M. Morita, G. Granata and C. Tokoro

Center (Aichi, Japan) for their support in the XAFS analy- 10) A. Henßge and J. Acker: Talanta 73 (2007) 220–226. ses. We thank Laura Kuhar, PhD, from Edanz Group (www. 11) M. Arif, M. Rahman and W.Y. San: Int. J. Adv. Manuf. Technol. 63 edanzediting.com/ac) for editing a draft of this manuscript. (2012) 481–504. 12) H.C. Genuino, N.N. Opembe, E.C. Njagi, S. McClain and S.L. Suib: J. Ind. Eng. Chem. 18 (2012) 1529–1539. Abbreviations 13) N. Altintaş, T. Ayok and T. Gozmen: Water Res. 21 (1987) 781–788. 14) L. Pelham: J. Fluor. Chem. 30 (1985) 1–17. CMP: Chemical mechanical polishing 15) J. He, T.-S. Siah and J.P. Chen: Water Res. 56 (2014) 88–97. EDS: Energy dispersive X-ray spectroscopy 16) C.J. Huang and J.C. Liu: Water Res. 33 (1999) 3403–3412. 17) M. Al-Harahsheh, M. Batiha, S. Kraishan and H. Al-Zoubi: Separ. FT-IR: Fourier transform infrared spectroscopy Purif. Tech. 123 (2014) 190–199. ICP-AES: Inductively-coupled plasma atomic emission 18) M. Gouider, M. Feki, S. Sayadi: J. Hazard. Mater. 170 (2009) spectroscopy 962–968. IC: Ion chromatography 19) P. Patnaik: Handbook of Inorganic Chemicals, McGraw-Hill (2002) MLA: Mineral liberation analyzer ISBN 0-07-049439-8. 20) A. Seidell: Inorganic and organic compounds a compilation of quanti- SEM: Scanning electron microscope tative solubility, (1919) Second edition. XRF: X-ray uorescence 21) S. Frisoni, S. Brenna and N. Masciocchi: Powder Diffr. 26 (2011) XRD: X-ray diffraction 308–312. XAFS: X-ray absorption ne structure 22) M. Yang, Y. Zhang, B. Shao, R. Oi and H. Myoga: J. Environ. Eng. XANES: X-ray absorption near edge structure 127 (2001) 902–907. 23) M.E. Simonsen, C. Sønderby, Z. Li and E.G. Søgaard: J. Mater. Sci. 44 (2009) 2079–2088. REFERENCES 24) User’s Guide to PHREEQC (Version 2), Water-resources Investigations Report (1999) 99-4259. 1) U.S. Geological Survey, 2015. Mineral Commodity Summaries 2015. 25) J.D. Allison, D.S. Brown and J. Novo-Gradac: MINTEQA2/ US Geological Survey 196. PRODEAFA2, a geochemical assessment model for environmental 2) E.W. Bond and J. Trachtman: World Trade Rev. 15 (2016) 189–209. systems: user manual supplement for version 4.0. US EPA, NERL 3) European Commission, 2011. Tackling the challenges in commodity (1999) Athens, Georgia. markets and on raw materials. 26) R.A. Martin, S. Yue, J.V. Hanna, P.D. Lee, R.J. Newport, M.E. Smith 4) T. Shojo, G. Inoue, K. Koike, A. Takeuchi, Y. Iwasa and S. Yasui: and J.R. Jones: Philos. T. R. Soc. A 370 (2012) 1422–1443. Kagaku Kogaku Ronbunshu 39 (2013) 452–457. 27) C. Karlsson, E. Zanghellini, J. Swenson, B. Roling, D.T. Bowron and 5) T. Itakura, R. Sasai and H. Itoh: Bull. Chem. Soc. Jpn. 79 (2006) L. Börjesson: Phys. Rev. B 72 (2005) 1–12. 1303–1307. 28) M.E. Simonsen, C. Sønderby and E.G. Søgaard: J. Sol-Gel Sci. Techn. 6) S.S. Chou, L.K. Tsao, S.Y. Huang, M.S. Li, S.H. Peng: Proceedings of 50 (2009) 372–382. the 19th Wastewater Treatment Technology Conference (1994) 29) C. Izutani, D. Fukagawa, M. Miyasita, M. Ito, N. Sugimura, R. pp. 310–319 (in Chinese). Aoyama and H. Oshio: J. Chem. Educ. 93 (2016) 1667–1670. 7) R. Aldaco, A. Garea and A. Irabien: Water Res. 41 (2007) 810–818. 30) S.L. Brantley: Kinetics of mineral dissolution, Kinetics of Water–Rock 8) M.T. Lee, C.W. Li and J.C. Liu: Separ. Purif. Tech. 156 (2015) Interaction, ed. by S.L. Brantley, J.D. Kubicki and A.F. White, 1057–1063. (Springer New York, 2008). 9) B. Grzmil and J. Wronkowski: Desalination 189 (2006) 261–268. 31) F.K. Crundwell: Hydrometallurgy 149 (2014) 265–275.