The Melt of Sodium Nitrate as a Medium for the Synthesis of Fluorides

The preparation of NaLnF4 complexes, LnF3 (Ln = La, Ce, Y) rare earth binary fluorides, CaF2 and SrF2 alkali earth fluorides, as well as mixtures of these compounds from their nitrates dissolved in molten NaNO3 has been studied in order to select the ideal solvent for fluoride synthesis by spontaneous crystallization from flux. Sodium fluoride (NaF) was used as a fluorinating agent. The results of our experiments have confirmed that NaNO3 melt is one of the most promising media for precipitating said inorganic fluoride materials within a broad temperature range (300–500 ◦C). Also, in contrast with precipitation/co-precipitation from aqueous solutions, our syntheses have resulted in obtaining equilibrium phases only.


Introduction
Inorganic fluorides can be prepared using a broad variety of techniques [1][2][3][4][5][6].However, using these methods in practice is always accompanied by some experimental obstacles.The most important problem for the fluoride syntheses is the reaction of said fluorides with water, i.e., hydrolysis, which leads to an accumulation of hydroxyl and/or oxygen impurities in the obtained materials [7][8][9].Atmospheric water vapor, water from aqueous solutions and/or water adsorbed on the solid particle surfaces can participate in hydrolyzing fluorides.The rate of hydrolytic chemical transformation increases significantly when temperature goes up and hydrolysis at an elevated temperature has been called pyrohydrolysis.In order to suppress hydrolysis, an excess of fluorinating agent has been widely implemented to create a proper fluorinating atmosphere [10,11].Direct oxidation by molecular oxygen in the course of fluorination is usually unfavorable from a thermodynamics point of view and, thus, as a result, it usually does not create additional problems for fluoride syntheses.
Undesirable batch reactions with crucible materials are important factors in solid phase fluorination and syntheses in molten media [2,12,13].Typically, the use of ceramic crucibles is not possible due to their corrosion accompanied by the formation of highly volatile fluoride by-products, such as SiF 4 and/or AlF 3 .On the other hand, metal (Mo, Ni, Cu, steel, etc.) or carbon crucibles can efficiently reduce fluorides, lowering the oxidation state of the fluoride-forming metal (europium, ytterbium) [14,15] or even completely reducing the latter to the free element (bismuth) [16].Gold [17] and copper [18] seem to be convenient materials for fluoride preparation, but, nevertheless, one has to consider that hermetically sealed reactors are preferred in comparison to unsealed crucibles [17,19], for even platinum can become a reducing agent in the opened systems [20,21].Teflon and/or other organofluorine polymers also can be used for the synthesis of fluorides with lower melting points and higher reaction activities [22,23].
The sintering time that is necessary for reaching the solid-phase equilibrium increases exponentially with a temperature decrease, and this factor becomes an overwhelming obstacle for the low-temperature phase synthesis [13,24].
Non-equilibrium specimens with unique functional properties (e.g., enhanced ionic conductivity) can be obtained by mechanochemical techniques.The use of such methods leads to product contamination with milling materials, such as ZrO 2 or WC.However, a detailed study of possible oxygen contamination of the mechanochemical synthesis products has yet to be carried out [6,25].
Various functional fluoride materials with highly developed surfaces have been synthesized by thermal decomposition of the corresponding trifluoroacetates [6,26,27].This is a very attractive synthetic method, but it leads to the product's contamination with elemental carbon.
In contrast with the solid-phase synthesis, the use of solvents leads to a very significant rate acceleration.Water is the simplest solvent, and most fluorides, having low solubility in it, can be easily precipitated from aqueous solutions [3,4,28].Such precipitation is very inexpensive.It requires very simple experimental arrangements and, very frequently, leads to the formation of nanofluorides [3,4,28].Precipitation at room temperature minimizes hydrolysis by-products, but a well-developed surface of fluoride nanoparticle makes said hydrolysis easier.For example, hydrolytic oxygen contamination has been observed for bismuth fluoride precipitation [29].
We have performed systematic studies of phase formation in the fluoride systems using co-precipitation from aqueous nitrate and/or citrate solutions by various fluorinating agents (e.g., hydrofluoric acid, sodium fluoride, ammonium fluoride, etc.) [3,4,[29][30][31][32][33][34][35][36][37][38][39].Our results are valuable from both practical (luminophores, laser ceramics precursors, solid electrolytes) and theoretical (studying non-classic crystallization mechanism via oriented attachment crystal growth [40]; making novel non-equilibrium phases of variable compositions [30]) points of view.Our results have revealed a broad variety of the synthesized material properties compared to the existing data for the known phase diagrams obtained under equilibrium conditions [34]: low solubilities of the precipitated fluorides hindered achieving the chemical equilibrium.
It is worth noting that nanofluorides, obtained by co-precipitation, are hydrated.Highly active nanopowder surfaces causes multilayered water adsorption [41].This hydration leads to the formation of quite unusual chemical compounds, such as (H 3 O)Ln 3 F 10 (Ln = rare-earth elements) [30,42], and causes luminescence quenching by OH − groups and/or water molecules in freshly prepared nanopowders.It is not easy to remove water that has been adsorbed on nanoparticle surfaces but their pyrohydrolysis under thermal treatment can be prevented by forming compounds with fluorinating agents, such as BaF 2 •HF [37] or ammonium fluoride solid solutions in strontium fluoride, Sr 1−x−y Ln x (NH 4 ) y F 2+x−y [34,38,39].
By definition, hydrothermal syntheses occur at elevated temperatures.This leads to a solubility increase and a surface energy decrease (particle faceting) for the synthesized nanofluorides, and equilibrium conditions can be reached.We successfully used the aforementioned hydrothermal techniques to prepare potassium-rare earth fluoride compound series [43][44][45] and, thus, refined phase equilibria data (including the ones for the lower temperatures) for the KF-RF 3 systems [46].However, temperature increase is accompanied with hydrolysis acceleration, and the content of the contaminating oxygen admixtures also increases in the K 2 RF 5 < KRF 4 < KR 2 F 7 < KR 3 F 10 series due to the isomorphous substitution of fluorine ions by OH − fragments with preservation of the phase homogeneity.
Molten salt synthesis (MSS) represents an additional group of synthetic techniques.These techniques include phase formation and crystallization from molten solutions (flux) and it requires the used solvents to have lower melting temperatures, be chemically inert (i.e., solvents shall not form compounds and/or solid solutions with the target compounds or phases), have sufficient solubility for the starting materials, be able to separate solvent from the products, and be non-volatile and display low toxicity [50].Of course, there is no such thing as an ideal solvent, but one still can try to find the best fitting materials for the aforementioned MSS technology.Sodium fluoride can be considered as one of the most promising MSS materials due to its sufficient water solubility and relatively low melting point (994 • C).However, it easily forms numerous phases with the other fluorides, like in the NaF-BaF 2 -GdF 3 system [51].Other fluoride flux materials for MSS methods include the following examples: Garton and Wanklyn grew K 2 NaGaF 6 and Rb 2 KGaF 6 single crystals using PbF 2 solvent [52], Hoppe utilized toxic thallium fluoride as a flux [53], and other authors [54] implemented the highly reactive NH 4 HF 2 as a flux and fluorinating agent for KMgF 3 preparation [54].
Oxide glasses can be considered as some kind of fluxes, too, as nanofluorides can be obtained from oxyfluoride glass ceramics by a dissolution of the silicate matrix in hydrofluoric acid [58].However, the aforementioned oxyfluoride glass ceramics are not the only ones representative of exotic type fluxes.Thus, an orthorhombic single CaF 2 crystal polymorph has been grown under high pressure from molten Ca(OH) 2 flux [59].
All of the above-listed reasons illustrate the need to search for novel solvents and, as a result, we have explored those opportunities and suggested nitrate melts as such alternative fluxes.Namely, we present below our results for the use of NaNO 3 melt for fluoride syntheses.

Results
In the present paper, we have studied the preparation of well-known LnF 3 (Ln = La, Ce, Y) rare earth and CaF 2 , SrF 2 alkali earth fluorides as well as their mixtures and NaLnF 4 complexes from solutions of corresponding metal nitrates in molten NaNO 3 by spontaneous crystallization (precipitation) in melt solutions.Despite the fact that actual chemical composition of the aforementioned NaLnF 4 phases is Na 3x Ln 2−x F 6 (gagarinite-type structure, derived from UCl 3 -type; x ~0.5) [2,32,46], we will name such solid solution compounds formed in the NaF-LnF 3 system as NaLnF 4 in agreement with modern literature naming conventions.
We have used sodium fluoride (NaF) as the fluorinating agent in the studied systems; the corresponding chemical transformations can be described by the following equations: Results of our work are presented in Table 1 and Figures 1-10.

The NaF-YF 3 System
X-ray diffraction patterns of NaYF 4 powders, obtained by precipitation from the NaNO 3 melt, before (a) and after (b) washing with water are presented in Figure 1.The most intensive peaks in sample (a) X-ray diffraction pattern correspond to NaNO 3 phase (R−3c space symmetry group (SSG); PCPDFWIN # 85-1466).After washing with water, sample (b) contained only one phase, NaYF 4 or β-Na(Y 1.5 Na 0.5 )F 6 (P63/m SSG; PCPDFWIN # 16-0334).* See Appendix A, Table A1.As our example with 3 and 4 samples of Y(NO 3 ) 3 :NaF:NaNO 3 = 7:52:41 starting material ratio has demonstrated, raising the temperature of the specimen synthesis from 320 to 435 • C did not affect its X-ray diffraction pattern: in both cases, the synthesized products contained NaYF 4 phase only with nearly the same crystal lattice parameters (Table 1), but with different particle morphology (Figure 4).NaYF 4 particles, formed at 320 • C (Figure 4b,d), contained relatively large hexagonal-shaped micron-size particles.Such particles, made up by ca.50 nm thick platelets, were hollow inside.In turn, a similar composition sample, formed at 320 • C (Figure 4b,d), also contained elongated micron-size particles with habitus, typical for the hexagonal system, but the temperature increase has led to the formation of dense, completely filled (bulk) particles.

The NaF-LaF 3 and NaF-CeF 3 Systems
Our attempts to prepare NaLaF 4 phase in the NaF-LaF 3 system were unsuccessful: for any ratios of the starting materials, only LaF 3 microcrystals were formed (Figure 5a).

The CaF 2 -SrF 2 System
The results of our syntheses in the calcium and strontium fluoride systems are presented in Figures 7-11.The values of yield were 68-91% (see Appendix A, Table A2).Intrinsic fluorides form easily washable micron-sized particles with fluorite-type crystal structures (Figure 7) with lattice parameters coinciding with the literature data (CaF 2 , a = 5.463 Å, PCPDFWIN # 35-0816; SrF 2 , a = 5.800 Å, PCPDFWIN # 06-0262).However, the morphology of the aforementioned powder particles is different: strontium fluoride formed faceted particles, whereas calcium fluoride did not (Figure 8).The use of equimolar Ca:Sr = 1:1 starting composition produced a mixture of two different fluorite-type phases with crystal lattice parameters different from both unit cell parameters for intrinsic CaF 2 and SrF 2 (Figure 9).SEM data unequivocally indicate that fluoride particles, containing more strontium fluoride, had a much larger size than the particles, containing more calcium fluoride (Figures 10 and 11).

Discussion
For the reader's convenience, the properties of molten sodium nitrate are summarized in Table 2.The solubility of NaF in NaNO 3 at 350 • C is about 5 mol %, and it increases up to 10 mol % at 450 • C [68].Pure NaNO 3 decomposes at 557 • C, but its decomposition temperature lowers when NaF is added (the addition of 7 mol % NaF decreases the decomposition temperature to 502 • C) [69].Nevertheless, the use of NaNO 3 melts provides a sufficiently broad temperature interval for the corresponding syntheses.Moreover, the addition of other nitrates and/or salts like KNO 3 , NH 4 NO 3 , etc. can lower further the NaNO 3 melting temperature [70]: solid solution, formed in the NaNO 3 -KNO 3 system, has its minimum point in the melting curves at 222 • C for 52 ± 3 mol % KNO 3 [71].However, addition of the different cations in the molten reaction mixture unavoidably results in the contamination of the formed microcrystals and even in the parallel formation of supplementary parasitic phases [46].
Sodium nitrate is a well-known oxidizer, but it has not shown any oxidative properties in our experiments, including absence of Ce(III) to Ce(IV) oxidation (i.e., formation of Ce(IV) phases was not observed).Nevertheless, one has to monitor a possibility of such oxidation processes in the future.
The oxygen and nitrogen content in the synthesized samples are below the detection limit (1%) of the EDX method.
It is important to note that the simultaneous preparation of strontium and calcium fluorides has resulted in the formation of the mixture of slightly contaminated individual CaF 2 and SrF 2 , respectively, whereas the use of the other synthetic techniques, such as melting [72], co-precipitation from aqueous solutions [28], or mechanochemical synthesis [73], resulted in continuous Ca 1−x Sr x F 2 solid solution.However, in accordance with the third law of thermodynamics [24], all solid solutions must decompose under cooling, and the lack of such decomposition unequivocally indicates a non-equilibrium state of the prepared specimen.The highest critical decomposition temperature for Ca 1−x Sr x F 2 solid solution, estimated for the CaF 2 -SrF 2 -MnF 2 triple system data, is about 1160 K [74].The decomposition curve of the Ca 1−x Sr x F 2 solid solution, calculated for the regular solution model, is presented in Figure 11.Taking into account crystal lattice parameters a = 5.463 Å for CaF 2 and a = 5.800 Å for SrF 2 and considering a linear correlation between crystal lattice parameters and composition of the Ca 1−x Sr x F 2 solid solution (Vegard's Law), we carried out calculations for the concentrations of the CaF 2 -based and SrF 2 -based solid solutions (Table 3).We also used the data of Table 1 for the unit cell parameters of the phases formed in the CaF 2 -SrF 2 system.The obtained 5 ± 2 mol % solid solution concentrations are in a good agreement with the estimated data for T = 573 K (Figure 11).At the present time, we plan to continue our investigations of the influence of the synthesis durations on the aforementioned crystal lattice parameters.Therefore, in the aforementioned Ca 1−x Sr x F 2 system, only the synthesis in the NaNO 3 melt produced equilibrium phases.This observation demonstrates quite unique perspectives for the preparation of the lower temperature equilibrium phases in the fluoride systems: making and studying such phases has been quite limited up to the present moment due to the above experimental obstacles [46,75,76].Also, it is worth noting that the absence of NaLaF 4 in our synthesized specimens can be easily explained in view of the aforementioned observations, for NaLaF 4 , easily obtainable by sintering or melting, is not stable at lower temperatures, while being stable at the higher temperatures [32,60,61].
It is worth mentioning the unusual morphology of the particles in synthesized NaYF 4 powders (Figures 3 and 4).It appears that flat nanoplates were formed first, and then they merged together forming hollow hexagonal prisms (Figures 3c and 4a,c), perhaps, by some attachment crystal growth process [40] with unknown/obscure mechanism.Bulk hexagonal prisms with habitus corresponding to the crystal lattice type have been formed at the higher temperature (Figure 4b,d).
Also it is worth mentioning the difference between morphologies of CaF 2 and SrF 2 microcrystals (Figure 8): SrF 2 microcrystals are faceted (see such simple-shaped polyhedral like cubes and rhombododecahedrons in the SEM images), but the faceting of CaF 2 was much fuzzier for unknown reasons.When both CaF 2 -based and SrF 2 -based solid solutions formed in the same precipitate, the particle size of the former is less than the particle size of the latter (Figures 10 and 11).Dissolution of CaF 2 in SrF 2 led to the lower quality faceting compared to the intrinsic SrF 2 (Figure 10).
Comparison of NaNO 3 and NH 4 NO 3 melt properties provides the former with two advantages: (1) NH 4 NO 3 is thermally unstable and easily undergoes solid state decomposition before reaching its melting point at 167 • C [77], producing toxic gases containing N 2 O (one has to use hood exhaust equipment while working with NH 4 NO 3 melts); and (2) as per Huang et al. [66], the use of NH 4 NO 3 melts does not necessarily produce equilibrium conditions ("BaCeF 5 " phase is absent in the BaF 2 -CeF 3 phase diagram [2], so cubic solid solution had to undergo mandatory ordering that has yet to be observed in [66]).
The use of ionic liquids for nanofluoride preparation represents a separate synthesis venue [78][79][80][81][82]. Ionic liquids are salts, containing large organic cations (usually, the unsymmetric ones), melt below 100 • C, and possess good thermal stability, high fire hazard safety, low corrosive activity, along with the low viscosity and vapor pressure.Ionic liquids appear to be good solvents for various organic and inorganic chemical compounds.Typical examples of ionic liquids include 1-butyl-3-methylimidazolium hexafluorophosphate (BmimPF 6 ), tetrafluoroborate (BmimBF 4 ), and/or chloride (BmimCl) [83].These ionic liquids have been used successfully for the LnF 3 and NaLnF 4 rare earth nanofluoride syntheses [78][79][80][81][82] as solvents (ionic transport media) and as fluorinating agents (water traces initiate ionic liquid pyrohydriolysis).Synthesized nanopowders can be easily separated from ionic liquids by rinsing with methanol.Usually, ionic liquid syntheses are carried out at the elevated temperatures under microwave or solvothermal process conditions.Nevertheless, our NaNO 3 melt-based synthetic technique can be a good replacement or supplement to the aforementioned preparation methods in ionic liquids: NaNO 3 possesses the same advantages as the aforementioned ionic liquids (lower-temperature melting point, high fire hazard safety, low corrosive activity, low toxicity), but it also is less expensive and can be washed out with water instead of toxic and flammable methanol.
We synthesized the fluoride specimens in NaNO 3 melts at 300-435 • C in ceramic crucibles with and without aluminum foil lining as well as in alundum crucibles.Separate experiments have indicated that ceramic crucibles were capable of withstanding melt corrosion, and metal lining did not provide any additional benefit to the quality and purity of the synthesized samples.
Starting materials were ground in the mortar with pestle until a homogeneous powder was formed.It was then quantitively transferred to the glazed porcelain crucible and placed in the oven for the synthesis at the elevated temperatures.Ratios of the starting materials were varied within the aforementioned ranges, including changes of the amounts of fluorinating agent excess and amount of the solvent used.All synthetic experiments were carried out at 300-435 • C (10 • C/min heating rate, 1 h exposure at maximum temperature) for all specimens unless specified otherwise.Annealed samples were cooled slowly over 10 hours.Molten products were removed from the crucibles, washed with doubly distilled water to remove remaining solvent and unreacted fluorinating agent, and dried under air at ~40 • C.

Conclusions
Results of our experiments unequivocally indicate that NaNO 3 melt is a very promising medium for the preparation of inorganic fluoride materials over a broad temperature range (300-500 • C).The latter synthetic temperature can be further lowered by using nitrate mixture, e.g., NaNO 3 -KNO 3 .Syntheses in NaNO 3 melts allowed preparation of micron-sized powders containing faceted microcrystals with the lowered surface areas and decreased adsorption capabilities.The obtained luminescent materials did not require additional thermal treatment.Synthesized specimens contained equilibrium phases, and were not contaminated by oxygen and/or carbon.Suggested synthetic protocols are inexpensive, environmentally friendly and can be used as an alternative to ionic liquid synthesis methods.

Figure 1 .
Figure 1.X-ray diffraction patterns of NaYF 4 sample 3 before (a) and after (b) washing with water.NaNO 3 phase lines are marked with triangle symbols.

Figure 2
Figure 2 data illustrate the influence of fluorinating agent stoichiometry on the phase composition of the reaction products.NaF shortage results in precipitation of the two-phase specimen (Figure 2a)

Table 1 .
Conditions and results of fluoride synthesis experiments.

Table A2 .
The values of yield.