Studies on fluorination of Fe₃O₄ (magnetite) by NH₄HF₂

Synopsis Fluorination of magnetite (Fe₃O₄) by NH₄HF₂ was investigated using simultaneous thermogravimetry and differential thermal analysis (TG-DTA), and observing the morphology and phase changes using scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS) and X-ray diffractometry (XRD). The results indicate that fluorination with the involvement of oxygen begins at room temperature, peaks at 178.4 0 C, and is completed at 200 0 C with the formation of only (NH₄)₃FeF₆. On heating, (NH₄)₃FeF₆ gradually releases NH₄F by the formation of NH₄FeF₄ at 259 0 C, then (NH₄) 0.18 FeF₃ at 327 0 C, and finally FeF₃ with minor FeF₂ at 400 0 C due to the partial reduction of Fe (III) to Fe (II). At 550 0 C, FeF₃ is oxidized to FeOF/Fe₂O₃.


Introduction
Transitional metal fluorides such as FeF₃ have gained growing attentions due to their potential for use as electrode materials in lithium ion batteries owing to their low cost and high specific capacities (Ignatiev et al., 2020;Shimoda et al., 2020;Zhou et al., 2017;Zhou et al., 2018). Fe is the fourth most abundant element in the Earth's crust and the cheapest metal in the market. In particular, the theoretical capacity of FeF₃ is up to 712 mA h g −1 because of its unique reaction mechanism during the charge and discharge processes. However, FeF₃ prepared by hydrometallurgical processes always contains crystal water such as FeF₃·3H₂O. During the dehydraion process, iron oxides form (Sophronov et al., 2016) because Fe fluorides are unstable in the presence of water vapour. The formation of iron oxides significantly decreased the capacity. FeF₃ can also be prepared via thermal process using anhydrous HF or F₂ gas at high temperature in special corrosion-resistant equipment (Johnson., 1981). NH₄F and NH₄HF₂ are recognized as cheaper and versatile fluorinating agent used at low temperatures (<240°C) (Andreev, 2008;Claux et al., 2016;Gordienko et al., 2017;Juneja et al., 1995;Laptash and Maslennikova, 2012;Laptash and Polyshchuk, 1995;Mukherjee et al., 2011;Pourroy and Poix, 1989;Sophronov et al., 2016). However, excess NH₄F should be added in order to produce oxygen-free fluorides due to the highly hygroscopic nature of NH₄F (Mukherjee et al., 2011;Pourroy and Poix, 1989;Sophronov et al., 2016). Fluorination of different oxides by NH₄HF₂ therefore appears to be the most convenient method for obtaining oxygen-free fluorides.

Experimental
Commercial analytical grade magnetite (Fe₃O₄, 99.8 wt.%) and ammonium bifluoride (99.5 wt.%) were supplied by Sinopharm Group (China). To ensure complete fluorination, the theoretical mass ratio of NH₄HF₂:Fe₃O₄ is 1.8450 according to Equation [3]. In order to investigate the reaction progress, two mass ratios of 2.5 and 3.5 (higher than the theoretical value) were chosen for investigation.
TG-DTA runs with pure NH₄HF₂ and Fe₂O₃/NH₄HF₂ mixtures were carried out in a Shimadzu DTG-60 unit at a rate of 5°C/min from 25°C to 600°C under 20 mL/min N₂ gas. Derivative thermogravimetry (DTG) curves were obtained as the first derivative of the TG curves. Based on TG-DTG-DTA results, the critical reaction temperatures of DTA curves were determined. In order to analyses the composition and determine the morphologies and phases of products before and after each reaction stage, Fe₃O₄ was first mixed with NH₄HF₂ at different mass ratios (NH₄HF₂: Fe₃O₄ = 2.5 or 3.5), put into a pure nickel crucible, then placed in a furnace for the assays. Heating was carried out at a rate of 5°C/min. Once the selected temperature was reached, the samples remained isothermal for 1 hour, and then allowed to cool to room temperature for further characterization. In order to increase repeatability, each test was repeated three times using 100 g Fe₃O₄.
The phases, morphologies, and composition of Fe₃O₄ powder, NH₄HF₂ agent, and fluorides produced were determined by XRD (D/Max-2500 pc type X-ray diffractometer) and scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS) (Oxford Instruments, INCA) Figure 1 shows a SEM image and the corresponding EDS trace of Fe₃O₄ particles. Fe₃O₄ particles exhibit a spherical morphology with particle size less than 500 nm. EDS results in Figure 1b indicated that only Fe and O were detected. The atomic ratio of Fe to O is close to 3:4, which coincides well with the chemical formula of Fe₃O₄. The XRD results indicate that this phase is Fe₃O₄ (#19-0629) (magnetite). Figure 2 shows the XRD spectrum of NH₄HF₂ agent (#12-0302). Clearly, NH₄HF₂ agent has a crystalline nature. Figure 3 shows the TG-DTG-DTA curve of NH4HF2 between 25 and 600°C. A weak endothermic peak is observed at 126.8°C due to the melting of NH 4 HF 2 (Carling and Westrum, 1976;House and Rippon, 1981;Resentera et al., 2020;White and Pistorius, 1972). The second well-defined endothermic peak overlaps the previous peak, having a maximum at 160.2°CThis peak corresponds to the decomposition and total removal of NH₄HF₂ (Carling and Westrum, 1976;House and Rippon, 1981;Resentera et al., 2020;White and Pistorius, 1972), as observed on the TGA-DTG curve.

Thermal analysis of NH4HF2
Furthermore, the mass loss of NH₄HF₂ reagent begins from room temperature, as found in previous investigations (Carling and Westrum, 1976;House and Rippon, 1981;Resentera et al., 2020;White and Pistorius, 1972). Figure 4 shows the TG-DTG-DTA curves of Fe 3 O 4 /NH 4 HF 2 mixtures between 25 and 600°C at different mass ratios. Clearly, two endothermic peaks are observed for a mass ratio of 2.5: 144.9°C and 259.9°C as seen in Figure 4a. Moreover, a weak endothermic peak appears at 327°C. For a mass ratio of 3.5, two new endothermic peaks appear at 126.8°C and 178.4°C, as seen in Figure 4b. However, the peak at 144.9°Cdisappears or is overlapped by the peaks at 126.8°C and 178.4°C; while the peak at 327°C increases significantly. From Figure 4a, it can also be seen that the mass loss of about 2-3% begins at room temperature for a mass ratio of 2.5, the same as for pure NH₄HF₂ ( Figure 3); while a minor mass gain (less than 1%) is observed for a mass ratio of 3.5 before 100°C (Figure 4b). With increasing temperature, a mass loss of about 10% is observed between 100 and 150°C for a mass ratio of 2.5; after which a levelling off occurs between 150 and 200°C. However, significant mass loss (approx. 28.9%) is observed between 100 and 200°C for a mass ratio of 3.5 without the curve flattening. By comparison, there are at least three endothermic peaks, which coincide well with the peaks in the DTG curves with large mass loss at 178.4°C, 259.2−259.9°C,and 327−327.6°C. The masses of residues for different temperatures are listed in Table  II. In this temperature range, Fe₃O₄ is stable even at ambient condition (Ouertani et al., 2020). These results indicate that the formation of the above three peaks may be due to chemical reactions.

Characterization of the fluorination products of Fe₃O₄
In order to identify and analyse the products involved in TG-DTG-DTA curves of Figure 4, samples were prepared by direct thermal treatment at different temperatures for 1 hour and then analysed using XRD. The results are shown in Figure 5. Clearly, the products between 135 and 180°C at both mass ratios consist chiefly of (NH₄)₃FeF₆ (#22-1040) with minor NH₄HF₂ (#12-0302) and Fe₃O₄ (#19-0629); the peak intensity of NH₄HF₂ decreases with increasing temperature and disappears at 270°C; the peak of Fe₃O₄ appeared between 135 and 180°C disappears at 270°C; only NH₄FeF₄ (#20-0503) is detected at 270°C at both mass ratios.
These results indicate that the mass ratio of NH₄HF₂ to Fe₃O₄ has no influence on the fluoride phases between 135 and 270°C. In this case, only the fluorides with the mass ratio of 2.5 after direct thermal treatment were chosen for analysis. Macroscopic morphology investigation showed that the fluorides formed between 135 and 180°C exhibit a similar grey color with increasing temperature: light grey, grey, and dark grey. Figure 6 shows the SEM images and the corresponding EDS data for fluorides produced between 135 and 270°C. Clearly, the fluorides formed between 135 and 180°C exhibit a similar large, faceted-grain morphology, as seen in Figures 6a, 6c, and6e. Furthermore, the particle size increases with increasing temperature, although, the average grain size at 180°C is still less than 1 μm. The corresponding EDS results in Figures 6b, 6d, and.6f indicate that the fluorides consist of Fe, F, and N, with minor O. Hydrogen was below the detection limit. In order to obtain more precise results for Fe and F, O is omitted during the quantitative analysis. From Figure 6, it can be seen that the content of Fe and N increases while the content of F decreases with increasing temperature. The atomic ratios of F:Fe of the fluorides at 135°C, 150°C, and 180°C are 6.4, 6.2, and 6.0, respectively. These results suggest that NH₄HF₂ is lost between 135 and 180°C, which is corroborated by the decreasing peak of NH₄HF₂ in Figure 5.
The fluorides at 270°C become white ( Figure 6g). Clearly, the faceted particles formed between 135 and 180°C disappear while finer spherical particles with average size less than 100 nm appear. The EDS results in Figure 6h indicate that the fluorides consist of Fe, F, and N without O, and the atomic ratio of F:Fe further decreases to 4.0.
The fluorides become green at 330°C. The size of the spherical particles slightly increases to larger than 150 nm and significant agglomeration occurs, as seen in Figure 7a. The EDS results in Figure 7b indicate that the fluorides at 330°C also consist of Fe and F with minor N. However, the atomic ratio of F:Fe further decreases to 3.0, which is close to the formula of FeF 3 . However, the XRD results in Figure 8a indicate that this is not FeF₃ but a new phase comprising (NH₄) 0 .₁₈FeF₃ (#47-0646). The results suggest that an intermediate phase, (NH₄) 0 .₁₈FeF₃ (#47-0646), forms between NH₄FeF₄ and FeF₃. In other words, a new chemical reaction occurs between 270 and 330°C.
The fluorides formed at 400°C (Figure 7c) are also green. Clearly, the particle size is the same as that formed at 330°C. However, the EDS results in Figure 7d indicate that the fluorides at 400°C consist of Fe and F without N. The atomic ratio of F:Fe is close to 3, the formula ratio of FeF₃. However, the XRD results in Figure 8b indicate that this is FeF₃ (#33-0647) with minor FeF₂ (#45-1062).  The fluorides become red at 550°C (Figure 9a). Clearly, the average particle size increases up to 500 nm. The EDS results in Figure 9b indicate that the fluorides consist of Fe and O with minor F. The atomic ratio of O:Fe is close to 3:2 of the formula ratio of Fe₂O₃. The XRD results in Figure 9c indicate that the fluorides consist mostly of Fe₂O₃ (#33-0664) with minor FeF₃(#33-0647) and FeF₂ (#45-1062).

Discussion
The melting of NH₄HF₂ at 126.8°C (Carling and Westrum, 1976;House and Rippon, 1981;Resentera et al., 2020;White and Pistorius, 1972) leads to an endothermic peak in the DTA curve, as found in this work (Figure 3). This is why a clear peak is observed at 126.8°C for a mass ratio of 3.5 (Figure 4b). Even before the melting of NH₄HF₂, a minor mass loss occurs due to the decomposition of NH₄HF₂, as seen in Figures 3 and 4a. With the melting of NH₄HF₂ at 126.8°C, the mass loss increases sharply and a well-defined endothermic peak with large mass loss occurs at 178.4°C due to the fluorination of Fe₃O₄. According to Equations [1], [2], and [3], the fluorination of Fe₃O₄ should form (NH₄)₃FeF₆ and NH₄FeF₃. However, the results in Figures 5 and 6 indicate that only (NH₄)₃FeF₆ (#22-1040) with a comparable coarse faceted- grain morphology forms between 135 and 180°C. No NH₄FeF₃ was detected. The results suggest that divalent iron becomes trivalent through the oxidation of Fe₃O₄ or the involvement of oxygen during the fluorination progress (Laptash et al., 2000). In a word, oxygen is involved in the fluorination reaction according to the reaction: [4] In this case, a minor mass gain should be observed. Figure 4b indicates that a minor mass gain occurs even at room temperature, suggesting fluorination may start at room temperature. To confirm this assumption, a Fe₃O₄/NH₄HF₂ mixture with mass ratio of 2.5 was prepared and kept for one week at room temperature, then analysed using XRD ( Figure 10). Clearly, only (NH₄)₃FeF₆, and no NH₄FeF₃, was detected, the same as at 135-180°C ( Figure  5). Furthermore, the samples became warm during mixing, and a smell of ammonia was observed. The results suggest that the fluorination of Fe₃O₄ by NH₄HF₂ really begins at room temperature, the same as Fe₂O₃ (Wang et al., 2021). According to Equation [4], the fluorination of Fe₃O₄ at room temperature also causes the formation of H₂O, which is absorbed by the fluorides. The loss of absorbed H₂O between 100 and 150°C (Wang et al., 2020) plus the decomposition and removal of NH₄HF₂ between 126.8 and 160.2°C (Mukherjee et al., 2011;Resentera et al., 2020;White and Pistorius, 1972) might cause the formation of another weak peak at 144.9°C, as seen in Figure  4a. The peak at 144.9°C will be overlapped by the fluorination of Fe₃O₄ at 178.4°C at a high mass ratio of 3.5, as seen in Figure 4b.
With the melting of NH₄HF₂ at 126.8°C, the reaction rate increases sharply due to the faster liquid-solid reaction rate compared to the slower solid-solid reaction rate. Based on the above analysis, the peak at 178.4°C at mass ratio 3.5 is mainly due to the fluorination of Fe₃O₄ according to Equation [4]. However, according to Equation [4], the mass ratio of NH₄HF₂ to Fe₃O₄ for complete fluorination is 2.2140. Therefore, fluorination should be completed for both mass ratios. However, results in Figures 5a and 5b indicate that minor Fe₂O₃ and NH₄HF₂ are still detected between 135 and 180°C even after 1 hour, suggesting a slow fluorination rate of Fe₃O₄.
With increasing temperature, the further fluorination of unreacted Fe₃O₄ plus the decomposition/sublimation of NH₄HF₂ will consume all or part of the NH₄HF₂.In this case, the products at 200°Cshould consist of (NH₄)₃FeF₆, possibly with minor NH₄HF₂. Theoretical calculation indicates that the product masses at 200°C for mass ratios of 2.5 and 3.5 are 81.77-90% and 55.42-92.23%, in fair agreement with the measured values of 88% and 71.1% from TG curves, as listed in Table II. In other words, the products at 200°C contain major (NH4)3FeF6 with minor NH₄HF₂ residue. That is reasonable because the boiling point of NH₄HF₂ (238.8°C) is higher than 200°C. At 238.8°C, all NH₄HF₂ sublimes. Therefore, either no NH₄HF₂ will be present or it will be below the detection limit of XRD, as seen in Figure 5. Figure 4 indicates that a new peak appears at 259.2−259.9°C in the DTA curves, accompanied by a large mass loss. The value coincides well with the values in the literature (Juneja et al., 1995;Pourroy and Poix, 1989). The XRD results in Figure 5 indicate that a new phase of NH₄FeF₄ (#20-0503) forms at 270°C. Furthermore, NH₄FeF₄ (#20-0503) becomes white, and the morphology changes to fine spherical (Figure 6g). From these results, it may be concluded that (NH₄)₃FeF₆ releases NH₄F to form NH₄FeF₄ at 259.2−259.9°C according to the following reaction: [5] Theoretical calculation show that NH 4 FeF₄ residues at 280°C for mass ratios of 2.5 and 3.5 are 55.45% and 43.13%, in fair agreement with the measured values of 58% and 47% from the TG curve after consideration of the measurement error, as listed in Table II.
With increasing temperature, NH₄FeF₄ (#20-0503) will lose NH₄F to form FeF₃ at temperature above 330°C (Alexeiko et al., 2008;Juneja et al., 1995;Kraidenko, 2008;Laptash and Polyshchu., 1995;Pourroy and Poix, 1989;Shinn, Crosket, and Haemdler, 1966;Sophronov et al., 2016). However, Figure 4 shows that there exists an another peak at 327-327.6°C. Figures 7a and 8a suggest that the fluoride at 330°C is not FeF 3 but (NH 4 )0.₁₈FeF 3 (#47-0646) with minor N (Bentrup and Menz, 1990). The fluoride of (NH 4 )0.₁₈FeF₃ is a different colour (green) to that formed at 270°C (white). Furthermore, the particle size increases up to 150 nm. with significant agglomeration (Figure 7a). From the above results, it could be concluded that NH₄FeF₄ lost only part of its NH₄F to form a new intermediate phase of (NH₄)0.₁₈FeF₃ at 327°C. According to theoretical calculation, the residue masses at 330°C for mass ratios of 2.5 and 3.5 are 49.26% and 33.42%. These values are lower than the measured values of 52% and 41%, as seen in Table II. There are three reasons for this. The first is the measurement error, as seen in Table II. The second is the release of NH 4 F at 327°C, which results in the further fluorination of Fe₃O₄ residues according to follow reaction: There is no direct evidence for this. However, the XRD results in Figure 5 indicate that the complete fluorination of Fe₃O₄ is a lengthy process Therefore, the assumption that fluorination of Fe₃O₄ is incomplete before 327°Cduring TG-DTA analysis is reasonable (Wang et al, 2021). The third reason is the slower release rate of NH 4 F from NH 4 FeF 4 at 327°C during TG-DTA analysis. To confirm this assumption, a Fe 3 O 4 /NH 4 HF 2 mixture with mass ratio of 2.5 was prepared and heated at 330°C for 10 minutes, then analysed using SEM/EDS and XRD. The results are shown in Figure 11. Clearly, only minor (NH 4 )0.₁₈FeF₃ (#47-0646) was detected, with major NH₄FeF₄ (#20-0503). The results indicate that the complete release of NH₄F from NH₄FeF₄ (#20-0503) takes a long time, even at 327°C.
With a further increase in temperature, (NH₄)0.₁₈FeF₃ will gradually lose all its NH4F to form FeF₃ with minor FeF₂ due to the partial reduction of Fe (II) to Fe (III) by ammonia at 400°C (Alexeiko et al., 2008;Bentrup and Men., 1990;Pourroy and Poix, 1989;Laptash and Polyshchuk, 1995;Laptash et al., 2000;Wang et al., 2021), as seen in Figure 8b. After 400°C, the mass loss is negligible, as seen in Figure 4. Therefore, the measured residue at 600°C is closed to the theoretical value after consideration the measurement error and the adsorption of F, N, and NH₃ by FeF₃ during TG-DTA analysis.
At 550°Cwithout gas protection, Fe oxides ( Figure 9) form due to the oxidation of FeF₃ (Alexeiko et al., 2008;Juneja et al., 1995;Sophronov et al., 2016). In order to analyse the oxidation progress, a Fe₃O₄/NH₄HF₂ mixture with mass ratio of 2.5 was prepared and heated at 550°C for 10 minutes, then analysed using SEM/EDS and XRD. The results are shown in Figure 12. Clearly, the fluorides contain a major component of FeF₃ with minor FeOF (#18-0648). FeF₂ is either absent or below the detection limit of XRD. Combined with the results in Figure 9, it can be concluded