Transition metal trifluoroacetates (M = Fe, Co, Mn) as precursors for uniform colloidal metal difluoride and phosphide nanoparticles

We report a simple one-pot synthesis of uniform transition metal difluoride MF2 (M = Fe, Mn, Co) nanorods based on transition metal trifluoroacetates (TMTFAs) as single-source precursors. The synthesis of metal fluorides is based on the thermolysis of TMTFAs at 250–320 °C in trioctylphosphine/trioctylphosphine oxide solvent mixtures. The FeF2 nanorods were converted into FeF3 nanorods by reaction with gaseous fluorine. The TMTFA precursors are also found to be suitable for the synthesis of colloidal transition metal phosphides. Specifically, we report that the thermolysis of a cobalt trifluoroacetate complex in trioctylphosphine as both the solvent and the phosphorus source can yield 20 nm long cobalt phosphide nanorods or, 3 nm large cobalt phosphide nanoparticles. We also assess electrochemical lithiation/de-lithiation of the obtained FeF2 and FeF3 nanomaterials.

Synthesis of nanoscale inorganic materials remains an active research area in inorganic chemistry, owing to the unique and improved material properties that emerge with respect to their bulk counterparts [1][2][3][4][5][6][7] . Downsizing is particularly important for conversion-type cathode materials for Li-ion batteries such as transition metal fluorides. These compounds offer one of the highest capacities among conventional cathode materials (e.g., 571 mAh g −1 for FeF 2, 237 mAh g −1 for FeF 3 vs. 148 mAh g −1 for LiCoO 2 ) 8,9 , but suffer from reduced rate capabilities and cycling stabilities, which are associated with low electronic conductivity and considerable structural reconstruction of the electrodes during cycling [10][11][12][13][14][15][16] . In this context, reducing the primary grain size of materials has become the main strategy for reducing polarization and improving the overall kinetics of Li + insertion. Nanosized materials have a considerably shorter in-solid diffusion path with less mechanical stress during phase conversion.
The synthesis of uniform transition metal difluoride nanocrystals (NCs) and nanoparticles (NPs) is generally complicated by the highly reactive and hazardous nature of commonly used fluorine sources, e.g., F 2 17 , HF 18 , and NH 4 F 19,20 . Several liquid-phase chemical approaches, including co-precipitation [19][20][21][22][23] , hydrothermal [24][25][26][27] , and solvothermal 28,29 syntheses, have been used to synthesize transition metal fluorides. However, these methods yield NCs of irregular shape and broad size distributions. In 2005, an alternative approach was reported by Yan et al. 30 for the synthesis of LaF 3 nanoparticles (NPs) based on thermolysis of a single-source lanthanum trifluoroacetate precursor in high-boiling point organic solvents. The proposed surfactant-assisted synthetic pathway enabled control over nucleation and growth of LaF 3 NPs by adjusting the capping ligands and the reaction temperature. In subsequent studies, a variety of different colloidal NPs, including NaYF 4 31 and GdF 3 32 , were reported by Murray et al. based on single-source precursors. The reports of Yan et al. 30,33,34 , Murray et al. 31,32 and others [35][36][37][38][39][40] have guided this work, wherein we report on applications of transition metal (Fe, Mn, Co) trifluoroacetate complexes (TMTFAs) as precursors for the synthesis of uniform colloidal NPs of Fe, Mn, and Co difluorides. In these precursors, both transition metal and fluorine are integrated into one compound providing a high control over the size of transition metal (Fe, Mn, Co) fluoride NPs.
We achieved a colloidal synthesis of highly uniform Fe, Mn, and Co difluoride nanorods (NRs) through thermolysis of inexpensive TMTFAs in the solvents trioctylphosphine (TOP) and trioctylphosphine oxide (TOPO). Synthesis of FeF 2 NRs. "Fe 3 OTFA" (562 mg, 0.5 mmol), was mixed with TOP (10 mL, 22.4 mmol) and OA (0-0.952 mL, 0-3 mmol) in a 50 mL three-neck flask. Afterward, the reaction mixture was dried under vacuum at 110 °C for 1.5 h followed by heating to 320 °C at a heating rate of 6-18 °C min −1 . Finally, the reaction was quenched to 200 °C with compressed air followed by cooling in an ice-water bath with concomitant injection of anhydrous toluene (20 mL) into the crude solution at approximately 120 °C. Different amounts of OA and different heating rates controlled the length of the FeF 2 NRs (see SI for details, Table S1). The FeF 2 NRs were washed two times with a toluene/ethanol mixture and separated by centrifugation. After the second washing step, the NRs were redispersed in toluene (2-4 mL) and stored under ambient conditions. Synthesis of MnF 2 NRs/NPs. Colorless "Mn(TFA) 2 " powder (509 mg, 0.5 mmol) was mixed with TOPO (8.8 g, 23 mmol) or TOP (10 mL, 22 mmol) in a 50 mL three-neck flask under a N 2 flow to obtain short (15-20 nm) or long (25-35 nm) MnF 2 NRs, respectively. Then, the reaction mixture was dried under vacuum at 110 °C for 1.5 h followed by slow heating (6 °C min −1 ) to 250 °C under a N 2 flow (see SI for details, Table S2). Afterward, the reaction mixture was quenched following the washing procedure as described above for FeF 2 NRs. Synthesis of CoF 2 NRs/NPs. "Co(TFA) 2 " (0.5 g) was mixed with TOPO (8.8 g, 23 mmol) and OA (0.5 mL, 1.6 mmol) in a 50 mL three-neck flask under N 2 flow. The reaction mixture was then dried under vacuum at 110 °C for 1.5 h under stirring (1400 rpm) followed by slow heating (6 °C min −1 ) to 300 °C (see SI for details, Table S3). Afterward, the temperature was maintained for 20 min followed by the quenching and washing steps as described above for the FeF 2 NRs.
Synthesis of Co 2 P NPs and Co 2 P NRs. In a typical synthesis of ~3 nm Co 2 P NPs, "Co(TFA) 2 " (0.5 g), TOP (10 mL, 22 mmol), and OA (0.5 mL, 1.6 mmol) were loaded into a 50 mL three-neck flask and dried under vacuum at 110 °C for 1.5 h. The reaction mixture was slowly heated to 300 °C at a heating rate of 6 °C min −1 and maintained at this temperature for 1.5 h. The synthesis of Co 2 P NRs were performed in the same was as that of the Co 2 P NPs; however, OA was added to the reaction mixtures and the heat-treatment time was prolonged to 2 h. Afterward, the reaction mixtures of Co 2 P NPs and Co 2 P NRs were quenched followed by the washing procedure, as described above for the FeF 2 NRs.

Synthesis of FeF 3 NRs.
In a typical synthesis of FeF 3 NRs, a ~150 mg portion of the 120 nm long FeF 2 NRs in powder form was placed in a closed Al 2 O 3 tube and Al 2 O 3 crucible as a container. The tube was then dried by applying a vacuum for 10 min followed by purging with N 2 for 20 min at room temperature. Under an N 2 flow, the sample was heated in a tube furnace (Across International, STF1200) to 500 °C. At this temperature, a flow of F 2 / Ar (Linde, 9.9% of F 2 in Ar) gas was set for 10 min. Then, an N 2 purge was applied for 30 min, and the tube was cooled down to room temperature. The obtained FeF 3 NRs were stored under vacuum. A schematic diagram of the oven set-up is shown in Figure S2.
www.nature.com/scientificreports www.nature.com/scientificreports/ HR-, TEM, and SAED. High-resolution (HR) and low-resolution transmission electron microscope (TEM) images and selected area electron diffraction (SAED) patterns were taken with a JEOL JEM-2200FS microscope operating at 200 kV. EDX maps in scanning transmission electron microscopy mode were recorded on a probe aberration-corrected FEI Titan Themis operated at 300 kV using a SuperEDX detector with a beam current of about 1 nA. EDX mapping was stopped when degradation of the particles due to radiation damage was observed. Samples for TEM analysis were prepared on carbon-coated Cu grids (Ted-Pella).
EDX. The energy-dispersive x-ray (EDX) spectroscopy measurements were performed on a nanoSEM 230 Electrochemical characterization of FeF 2 , and FeF 3 NRs. FeF 2 electrodes were prepared by mixing FeF 2 NRs (58%), carbon black (CB, 21%), graphene oxide (GO, 13%), and poly(vinylidene) fluoride binder (pVdF, 8%). First, the ~120 nm long FeF 2 NRs were mixed with CB in a ball mill (Pulverisette 7, Fritsch, ZrO 2 balls and beaker) for 1 h at 300 rpm followed by a heat treatment at 500 °C for 0.5 h under a N 2 atmosphere to remove the ligands and to improve the carbon embedding. Then, the FeF 2 /CB powder was ball-milled with pre-milled GO (synthesized by the Brodie method 42 ), pVdF, and n-methyl-2-pyrrolidone (NMP) for 2 h, at 500 rpm. The final slurry was brushed directly on Al foil and dried under a vacuum at 80 °C for at least 24 h. FeF 3 electrodes were prepared following the procedure described above for FeF 2 electrodes but without the addition of GO. The composition of the FeF 3 slurry was 58% of FeF 3 NRs, 34% of CB, and 8% of pVdF. Coin-type cells were assembled in a glovebox with the use of a one-layer glass fiber (Whatman) separator and 1 M LiPF 6 in ethylene carbonate/ dimethyl carbonate (1/1 by wt.) electrolyte (300 µL per cell). Li metal served as both a reference and counter electrode. Electrochemical measurements were performed on an MPG2 multi-channel workstation (BioLogic). Prior to electrochemical cycling, the FeF 2 and FeF 3 electrodes were tested by cyclic voltammetry at scan rates of 0.2 and 0.1 mV s −1 for 5 and 2 cycles, respectively. For galvanostatic measurements of the FeF 2 electrodes, constant current-constant voltage (CCCV) mode was applied for discharge and charge steps at 1.5 and 4.0 V vs. Li + /Li. The constant voltage was maintained until the measured current was equal to 1/5 of the initial current value. The obtained capacities were normalized to the mass of the FeF 2 or FeF 3 NRs.

Results and Discussion
Synthesis of transition metal (Fe, Co, Mn) difluoride NPs. Monodisperse MF 2 (M = Fe, Mn, and Co) NRs were obtained by thermolysis of corresponding TMTFAs ["Fe 3 OTFA", "Mn(TFA) 2 ", and "Co(TFA) 2 "] at temperatures of 250-320 °C with the use of TOP or TOPO as solvents (Fig. 1, see methods and Tables S1-S3 for reaction conditions). As indicated by Blake et al. 43 , the thermal decomposition of trifluoroacetic anions (TFA − ) proceeds via decarboxylation, formation of the trifluoromethyl anion (CF 3 − ), and its subsequent dissociation into a fluoride ion (F − ) and difluoromethylene (CF 2 ). Then, F − can couple with a metal ion, resulting in the formation of the corresponding metal fluoride. When the decomposition is performed in a suitable solvent and in the presence of surfactant molecules, the resulting fluorides can adopt the form of uniform colloidal NRs (Fig. 2). In our experiments, the use of TOP and TOPO as solvents was essential for preparing monodisperse MF 2 NRs because these act as neutral L-type ligands 44 , which coordinate to Lewis acidic surfaces. This mechanism might explain the preferred rod morphology of the MF 2 NRs; i.e. growth in the [001] direction of the rutile-type crystal structure. The growth in other directions is likely prohibited by TOP/TOPO molecules covering the Lewis acidic, metal-rich (010) and (100) facets of MF 2 . TOPO likely binds more strongly to the transition metals than does TOP owing to its highly polarized phosphor oxygen bond. Generally, compared with other solvents, such as long-chain alkanes/ alkenes, nitrogen and sulfur containing solvents (see example for FeF 2 NRs, Figure S3), TOP and TOPO are markedly better solvents for growing MF 2 NRs.
Highly crystalline FeF 2 NRs were obtained by heating the "Fe 3 OTFA" precursor solution in TOP solvent to 320 °C (Fig. 1a). The length of the FeF 2 NRs was tunable from 10 to 200 nm by addition of OA ( Fig. 2a-f, Table S1), presumably because of its stronger bonding to the transition metal-rich facets, such as (010) and (100), than that of TOP solvent. At a low OA content, we also observed twinned, boomerang-shaped FeF 2 NRs (Fig. 2c). The heating rate also affected the length of the FeF 2 NRs, yielding longer NRs at lower heating rates.
As illustrated in Fig. 2g,h and S4a,b, the powder X-ray diffraction (XRD), high-resolution transmission electron microscope (HRTEM) images, and selected area electron diffraction (SAED) results confirmed the formation of highly crystalline FeF 2 NRs having a tetragonal rutile-type crystal structure with the space group P4 2 /mnm (a = 4.7035 Å, c = 3.3056 Å, V = 73.13 Å 3 , PDF 045-1062). The FeF 2 NRs showed preferred orientation in the powder XRD pattern (Fig. 2g), as indicated by narrower (101) and (002) peaks in comparison with the (110), (111), and (211) reflections; the former being parallel to the growth direction of FeF 2 NRs. Additionally, we the (101) and (002) peaks of the FeF 2 NRs were more intense than those of bulk FeF 2 owing to alignment of the NRs. As the length of the FeF 2 NRs decreased, the preferential orientation and texture effects disappeared. We note that TOP is known to be an efficient phosphor source and thus might contaminate FeF 2 NRs. From EDX measurements of the FeF 2 NRs (Figure S4c), only a tiny amount of phosphorus was detected.
Synthesis of metal phosphides NRs/NPs. We also found that "Co(TFA) 2 " might also be used as a precursor for synthesizing cobalt phosphides, when combined with the solvent TOP as a phosphorous source and a solvent with heating to 300 °C (Fig. 1c). TEM analysis and EDX mapping (Fig. 3) point to the following formation mechanism (Fig. 3g): First, "Co(TFA) 2 " starts to decompose at 250 °C forming CoF 2 NRs (Figs 3a-c and S10), which then react with TOP at 300 °C. At this point, we observed a mixture of both cobalt fluoride and cobalt phosphide ( Figure S11a,b), followed by the formation of phase-pure 20 nm long cobalt phosphide NRs after 90 min (Fig. 3d-f). The obtained NRs assembled in large 3D clusters owing to their hexagonal shape and narrow size distribution. Our XRD ( Figure S12) measurements indicated that the cobalt phosphide NRs crystallized in two different phases: Co 2 P as the main phase with some CoP, both having the same orthorhombic structure (space group Pnam; for Co 2 P: a = 5.6465 Å, b = 6.6099 Å, c = 3.513 Å, PDF 032-0306; for CoP: a = 5.077 Å, b = 3.281 Å, c = 5.587 Å, PDF 029-0497).
When OA was added to the reaction mixture of "Co(TFA) 2 " in TOP, ultra-small 3 nm cobalt phosphide NPs were obtained at 300 °C after 2 h (Fig. 3j,k). The thermolysis of "Co(TFA) 2 " in TOP with OA for a short reaction time of 0-10 min yielded the mixture of cobalt fluoride and phosphide NRs/NPs ( Figure S11c,d). Our XRD and SAED measurements showed that the obtained NPs were amorphous. In addition, small crystalline regions were also visible from high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images ( Figure S13). In accordance with the EDX measurements ( Figure S13b), the chemical composition of the cobalt phosphide NPs, denoted as Co 2 P, corresponded to a 2:1 Co:P molar ratio.
We note that the cobalt phosphides have attracted considerable attention over the last three years for their ability to act as a bifunctional electrocatalyst for hydrogen [45][46][47][48][49][50][51] and oxygen reduction/evolution reactions 52-57 www.nature.com/scientificreports www.nature.com/scientificreports/ with low overpotentials used for water splitting [58][59][60][61][62][63][64] . These compounds are also used to catalyze substitution reactions of functional groups by hydrogen [65][66][67][68][69] , for heavy metal removal 70,71 , and for hydrogenation of CO 72 . The synthesis of Co 2 P NRs/NPs based on "Co(TFA) 2 " as a precursor has not been reported. Importantly, our synthesis of Co 2 P NRs/NPs yielded a much narrower size distribution than that achieved in NPs synthesized with Co(acac) 2 47,57,73-75 , ε-Co NPs 76 , and others precursors 77-85 . Synthesis of FeF 3 NRs. We tested fluorination of FeF 2 NRs with fluorine gas as a route to FeF 3 NRs. Briefly, ~120 nm long FeF 2 NRs were fluorinated in a powder form at 500 °C with an Ar/F 2 gas mixture for 30 min in an aluminum tube (see Experimental Section for details). Our XRD and SAED measurements showed that FeF 2 NRs were fully converted to FeF 3 NRs (Fig. 4a,b); the latter crystallized in a ReO 3 -type structure of FeF 3 (R-FeF 3 ) with the space group P-3c (167) (a = 5.2 Å and c = 13.323 Å, V = 312 Å 3 , PDF 033-0647). The rod shape of the initial FeF 2 NRs was retained. However, the FeF 3 NRs were four times as large as the FeF 2 NRs (Fig. 4c) owing to the volume difference between the crystal structures of the FeF 2 and R-FeF 3 . Additionally, agglomeration of R-FeF 3 NRs was observed during fluorination caused by concomitant ligand removal from the FeF 2 NRs surface. We note that the fluorine gas flow had considerable effects on the complete fluorination of the FeF 2 NRs. The temperature was also an important parameter. The formation of R-FeF 3 started at a low temperature of 350 °C; however, highly crystalline and phase-pure R-FeF 3 NRs were only obtained at 500 °C.

Electrochemical performance of FeF2 and FeF3
NRs. For the electrochemical measurements, we prepared electrodes by mixing a powder of FeF 2 or FeF 3 NRs with carbon additives [graphene oxide (GO) and/or carbon black (CB)], polyvinylidene fluoride (PVdF), and N-methylpyrrolidone (NMP) solvent (see Experimental Section for details). First, the FeF 2 or FeF 3 NRs were dry ball-milled with GO/CB or CB. Then the resulting powder was subjected to one more step of ball-milling with the pVdF polymer as a binder and NMP as a solvent, followed by casting the slurry onto Al foil. Afterward, the electrodes were dried under vacuum at 80 °C for 24 h.
We tested the FeF 2 NRs vs. Li + /Li in the voltage range of 1.5-4.0 V, which includes both conversion and insertion regions. From the CV curves (Fig. 5a), lithiation of electrodes composed of FeF 2 NRs during the first discharge was characterized by the appearance of a peak at 2 V vs. Li + /Li associated with reduction of graphene oxide followed by the intensity increase of the negative current at 1.5 V vs. Li + /Li indicating a conversion process of FeF 2 NRs (FeF 2 + 2Li + + 2e − → Fe + 2LiF). The formation of Fe and LiF during lithiation of FeF 2 has been reported in www.nature.com/scientificreports www.nature.com/scientificreports/ numerous studies based on in situ 86 and ex situ 19,[87][88][89] methods. In the reverse scan, the FeF 2 electrode displayed two peaks at 2.8 and 3.4 V vs. Li + /Li, which are associated with the formation of Li 0.5 FeF 3 and Li 0.25 FeF 3 phases. The third peak at a higher potential of 4.2 V was related to a pronounced and unknown irreversible reaction. Upon further cathodic cycling, a peak at 3.0 V vs. Li + /Li appeared, which we attributed to the formation of a Li 0.25 FeF 3 phase. We note that the intensity increase of CV curves during the initial cycles might be attributed to restructuring processes in the electrode caused by the formation of metallic Fe, which lowers the resistivity of the electrodes. As shown in Fig. 5b, the discharge voltage profiles of the FeF 2 NRs were similar to CV curves representing distinct insertion (3.4-2.6 V vs. Li + /Li) and conversion (1.5-2.0 V vs. Li + /Li) reactions. However, upon charging, the galvanostatic curves were rather smooth, which suggested a slow gradual de-lithiation processes. Figure 5c shows cyclic voltammetry curves of the electrodes composed of FeF 3 NRs at a scan rate of 0.1 mV s −1 . In the first cathodic cycle (lithiation step), the first two peaks appear at approximately 3.26 and 3.88 V vs. Li + / Li. We attribute these features to electrochemical lithiation of the FeF 3 NRs. We note that the mechanism of Li + intercalation into FeF 3 has been examined experimentally and theoretically; however, many of the details remain unclear. It is believed that the initial ReO 3 -type structure of FeF 3 transforms into a tri-rutile-like structure with a composition of Li 0.25 FeF 3 and Li 0.5 FeF 3 , which is followed by the formation of FeF 2 and LiF (for an overall one-electron process) upon further reduction. In the reverse scan, two distinct reversible de-lithiation reactions of FeF 3 91 , and Li et al. 9 Fig. 5d shows the typical voltage profiles of the Li-ion half-cells based on FeF 3 NRs as an active material at a current density of 50 mA g −1 . The shape of the voltage profiles and the cyclic voltammetry (CV) curves, were . For HAADF-STEM elemental mapping, the following color code was used: cobalt (red), fluorine (yellow) and phosphorous (blue). (g) Synthetic scheme presenting the reaction conditions for obtaining CoF 2 and Co 2 P NRs and Co 2 P NPs. (h-k) TEM images of CoF 2 NRs (h,i) and Co 2 P NPs (j,k) synthesized with OA.
www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ relatively sharp with a low polarization compared with previously reported data on FeF 3 cathodes 13,92,93 . These results point to an intermittent mechanism of lithiation/de-lithiation of the FeF 3 NRs through the formation of intermediate Li 0.5 FeF 3 and Li 0.25 FeF 3 phases. The galvanostatic measurements of FeF 2 and FeF 3 NRs are presented in Figures S14 and S15, respectively.

Conclusions
This work presents a simple synthetic route to prepare high-quality Fe, Mn, and Co difluoride NRs via thermolysis of transition metal (M = Fe, Mn, and Co) trifluoroacetate complexes in TOP and TOPO solvents. We show that the use of TOP or TOPO solvents is essential for synthesizing monodisperse MF 2 NRs, which act as neutral L-type ligands that coordinate to Lewis acidic surfaces leading to the preferred rod morphology (i.e., growth in the [001] direction). The length of the FeF 2 NRs was tunable by the addition of oleic acid owing to its stronger bonding to transition metal-rich facets such as (010) and (100). A bottom-up synthesis of high crystalline phase-pure FeF 3 NRs by fluorination of our FeF 2 NRs by fluorine gas is also reported.
We show that the cobalt trifluoroacetate complex can be thermally decomposed in a TOP solvent system to yield cobalt phosphide NRs/NPs. The reaction mechanism includes: thermolysis of "Co(TFA) 2 " with the formation of CoF 2 NRs following their reaction with TOP at higher temperatures of 300 °C leading to highly monodisperse 20 nm long Co 2 P NRs or 3 nm Co 2 P NPs (in the presence of oleic acid).
We also assessed the electrochemical storage of Li + ions in FeF 2 and FeF 3 NRs. Our studies are currently underway towards optimization of electrodes composed of FeF 2 and FeF 3 NRs (e.g., carbon encapsulation) and their electrochemical performance (choice of voltage intervals, and electrolytes).