Synthesis of rare-earth metal and rare-earth metal-fluoride nanoparticles in ionic liquids and propylene carbonate

Decomposition of rare-earth tris(N,N′-diisopropyl-2-methylamidinato)metal(III) complexes [RE{MeC(N(iPr)2)}3] (RE(amd)3; RE = Pr(III), Gd(III), Er(III)) and tris(2,2,6,6-tetramethyl-3,5-heptanedionato)europium(III) (Eu(dpm)3) induced by microwave heating in the ionic liquids (ILs) 1-butyl-3-methylimidazolium tetrafluoroborate ([BMIm][BF4]), 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([BMIm][NTf2]) and in propylene carbonate (PC) yield oxide-free rare-earth metal nanoparticles (RE-NPs) in [BMIm][NTf2] and PC for RE = Pr, Gd and Er or rare-earth metal-fluoride nanoparticles (REF3-NPs) in the fluoride-donating IL [BMIm][BF4] for RE = Pr, Eu, Gd and Er. The crystalline phases and the absence of significant oxide impurities in RE-NPs and REF3-NPs were verified by powder X-ray diffraction (PXRD), selected area electron diffraction (SAED) and high-resolution X-ray photoelectron spectroscopy (XPS). The size distributions of the nanoparticles were determined by transmission electron microscopy (TEM) and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) to an average diameter of (11 ± 6) to (38 ± 17) nm for the REF3-NPs from [BMIm][BF4]. The RE-NPs from [BMIm][NTf2] or PC showed diameters of (1.5 ± 0.5) to (5 ± 1) nm. The characterization was completed by energy-dispersive X-ray spectroscopy (EDX).


Introduction
Rare-earth (RE) elements gain increasing importance in materials science and modern chemistry [1][2][3]. Special attention has been paid to nanoscaled rare-earth metal particles [4][5][6]. In addition to the oxido and nitrido compounds, the rare-earth fluorides have interesting photo physical and electrochemical properties. An important representative of this category are AREF 4 compounds (A = alkali metal), with unique optical, magnetic and piezoelectric properties [7]. They are applied in solid-state lasers, three-dimensional flat-panel displays, and low-intensity IR imaging [8]. Syntheses of these AREF 4 -type compounds are based on the liquid precipitation reaction between soluble rareearth metal salts and alkaline fluorides. A co-thermolysis of Na(CF 3 COO) and RE(CF 3 COO) in oleic acid/oleylamine for the synthesis of NaREF 4 (RE = Er(III), Tm(III)) has also been described [8]. One problem of these syntheses is that the obtained rare-earth fluoride particles were not phase-pure [8].
However, in the absence of fluoride donors we obtained oxidefree rare-earth metal nanoparticles (RE-NPs) of Pr, Gd and Er (Scheme 1). By using either the chemically more inert IL 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([BMIm][NTf 2 ]) or propylene carbonate (PC) as reaction media RE-NPs were produced. This is an unusual and quite interesting result that we wish to highlight at this point. There exist only few reports on the wet-chemical synthesis of nanoparticles of rare-earth metals or rare-earth metal containing alloys or intermetallic phases. Solution synthesis of any oxide-free RE-NP was first reported by Wagner and co-workers using alkalide reduction of GdCl 3 in THF solution [6]. Scalable airand water-stable core-shell Gd@Au-NPs were obtained by using the same strategy [23]. Rare-earth metal containing intermetallic nano-phases have been suggested as novel materials for various catalytic applications [24]. For example, Pt 3 Y and Pt 5 Gd were predicted to be more active as Pt in the oxygen reduction reaction (ORR) [25]. Nevertheless, the high reduction potentials of rare-earth metal ions (typically below −2.0 V vs NHE) cause much difficulties regarding the chemical reduction of any chosen precursor and the prevention of post-synthesis oxidation or contamination of the RE-NPs. Recently, Alivisatos and co-authors reported on the synthesis of Pt 3 Y and other so-called early-late intermetallic nanoparticles by a solvent-free route employing a melt of the reducing agent (Na/K)BEt 3 H [26]. We like to put our results into this context. Herein we demonstrate that RE(amd) 3 releases the RE component by selective thermolysis even in the absence of additional reducing agents. Microwave heating and employing suitable ILs and PC as inert reaction media proved to be crucial.

Results and Discussion
Decomposition of RE(amd) 3

and Eu(dpm) 3
Thermogravimetric analysis (TGA) revealed decomposition of the rare-earth metal(III) tris(N,N′-diisopropyl-2-methylamidinate) (RE(amd) 3 ; RE = Pr(III), Gd(III), Er(III) and tris(2,2,6,6tetramethyl-3,5-heptanedionato)europium(III) (Eu(dpm) 3 ) at temperatures between 160 and 230 °C (Table S1 and Figure S2, Supporting Information File 1). To keep the formation of by-products as low as possible and to achieve complete decomposition of the precursors, a temperature of 230 °C was selected on the basis of these TGA measurements for all microwaveassisted thermal NP syntheses. As reaction media we used the fluorous IL 1-butyl-3-methylimidazolium tetrafluoroborate The rare-earth metal amidinates and Eu(dpm) 3 were suspended under an argon atmosphere in dried IL or in PC. The compounds were decomposed by microwave irradiation for 20 min at a power of 50 W at a temperature of 230 °C (Scheme 1). The size distributions of the obtained nanoparticles were determined by transmission electron microscopy (TEM) and highangle annular dark-field scanning transmission electron microscopy (HAADF-STEM). The crystalline phases and the absence of impurities (oxides) in RE-NPs and REF 3 -NPs were identified by powder X-ray diffraction (PXRD) and selected area electron diffraction (SAED). The results are summarized in Table 1.

REF 3 -NPs from RE(amd) 3 and Eu(dpm) 3 in [BMIm][BF 4 ]
The microwave-induced decomposition of the rare-earth metal amidinates RE(amd) 3 with RE = Pr(III), Gd(III), Er(III) and Eu(dpm) 3 [12,42]. It is also known that the [BF 4 ] − anion decomposes to fluoride at high temperature [43]. According to ion chromatographic (IC) analysis, a fluoride source other than the [BF] − anion can be excluded since the IL contains only a very small amount of fluoride ions (below 1 ppm, see Experimental section for IC Pr(amd) 3 PrF 3 11 ± 6 Eu(dpm) 3 EuF 3 23 ± 7 Gd(amd) 3 GdF 3 38 ± 17 Er(amd) 3 ErF 3 Gd e 1.5 ± 0.5 Er(amd) 3 Er e 3.0 ± 0.5 PC Pr(amd) 3 Pr e 2 ± 1 Eu(dpm) 3 d d Gd(amd) 3 Gd e 1.5 ± 0.5 Er(amd) 3 Er The average particle sizes are 11 ± 6 nm for PrF 3 , 23 ± 7 nm for EuF 3 , 38 ± 17 nm for GdF 3 , and 14 ± 5 nm for ErF 3 (Table 1). STEM/TEM images show small, nearly spherical and partially agglomerated particles. Close-up TEM images show interference patterns for EuF 3  The crystallinity of the particles was confirmed and the crystal phases were determined as pure rare-earth fluorides REF 3  The characterization was completed by energy-dispersive X-ray spectroscopy (EDX, in combination with TEM) for the qualitative element composition. EDX spectroscopy ( Figure 3 and Figures S5b, S6c, Supporting Information File 1) show the expected signals for Eu, Gd or Er and fluoride besides the bands for carbon and copper of the carbon-coated copper grid. The oxygen peak can largely be attributed to air contamination when the sample was introduced into the TEM device. A quantification of fluoride against rare-earth metal was not done, because matching of the F Kα 1 binding energy against the Lα 1 or Lβ 1 binding energies for Eu, Gd and Er is not very accurate.   The measured oxidation state 3+ of the rare-earth metals in the fluorides was corroborated by high-resolution X-ray photoelectron spectroscopy (HR-XPS) ( Figure 4 and Figures S4c, S5c, S6d, Supporting Information File 1) through comparison to the reported binding energies of metal(III) fluorides/oxides, metal(0) and organic fluorine/oxygen (Table 2) [48,49]. The measured metal and fluorine XPS values are in good agreement with the values of metal(III) fluorides that are given in litera-ture and thereby exclude the formation of metal(0) and the presence of organic fluoride (from residual IL). In addition, the formation of metal oxides can be excluded, since the measured binding energies of oxygen match very well with the literature values of organic oxygen [48,49]. For EuF 3 , no oxygen peak was seen in the XPS analysis. Therefore, SAED and PXRD data in combination with HR-XPS exclude any contamination of the REF 3 -NPs with metal(III) oxides.  Metal fluorides are used, for example, as cathode materials in lithium-ion batteries [6]. The lithium-ion battery is one of the most important rechargeable energy storage devices in modern electrical appliances such as mobile phones and laptops, but also in electric and hybrid vehicles [51]. The increasing performance of modern lithium-ion batteries is of great interest in cur-rent research [52][53][54]. Grey et al. showed that the use of FeF 2 nanoparticles as electrode material leads to a significant increase in the performance of the batteries compared to the macroscopic LiFeF 3 [55]. Therefore, we investigated the electrochemical properties of ErF 3 -NPs by galvanostatic charge/discharge profiles ( Figure 5). Until now there has been no report on ErF 3 applied as electrode material for Li + /Li. Cyclic voltammetry was performed to address general aspects of the redox behaviour ( Figure 5). In the range from 1.0 to 5.0 V, during the reduction, a reduction process takes place starting at 1.5 V, which may be attributed to the transformation of Er 3+ to Er 2+ , which can be reversibly oxidized. During the oxidation process, oxidation of the electrolyte was observed, starting around 3.4 V. Hence, such standard electrolytes cannot be applied to this redox couple. In the range from 1.0 to 0.05 V, there is an overpotential electrodeposition process of Al 3+ , stemming from the Al collector, indicating that the used potential cannot be lower than 0.  Figure S3, Supporting Information File 1).
Rare-earth metal nanoparticles (RE-NPs) were obtained for RE = Gd and Er. For Pr(amd) 3 and Eu(dpm) 3 no particles were seen in TEM investigations. The size and size distribution of the Gd-NPs and Er-NPs were determined by TEM ( Figure 6 and Figure S7a, Supporting Information File 1) to values of 1.0-2.5 nm for Gd (average diameter of 1.5 ± 0.5 nm) and 2.0-3.5 nm for Er (average diameter 3.0 ± 0.5 nm) ( Table 1). A close-up of the TEM images shows interference patterns, indicating crystallinity of the RE-NPs.
The crystallinity of the RE-NPs was confirmed by SAED and gave the expected reflections for elemental Gd and Er (Figure 7 and Figure S7b, Supporting Information File 1). Due to the very small size of the Gd and Er particles a meaningful PXRD pattern could not be obtained.
Characterization by EDX (Figure 7 and Figure S7b, Supporting Information File 1) gave the expected bands for Gd and Er. The small fluorine and sulfur peaks are due to residual IL around the nanoparticles ( Figure S3, Supporting Information File 1). We suggest that the residual IL coverage of the RE-NPs also prevents their oxidation during the short air contact upon transfer from a Schlenk flask into the TEM and XPS.
The oxidation state zero of gadolinium and erbium, i.e., the formation of Gd(0) and Er(0) metal NPs were indirectly supported by the measured XPS binding energies of oxygen and fluorine ( Figure 8, Table 3 and Figure S7c   RE = Gd(0) and Er(0) [48,49]. Thus, HR-XPS data in combination with SAED exclude any significant contamination of Gd(0)-NPs and Er(0)-NPs with oxides or fluorides. We note that the amount of metal oxides is below the detection limit and that small amounts of water in the ILs of up to 30    The microwave-assisted thermal decomposition gave rare-earth metal nanoparticles (RE-NPs). Again, for Eu(dpm) 3 no parti-cles were seen in TEM investigations. The size and size distribution of the RE-NPs were determined by TEM ( Figure 9 and Figures S8, S9a, Supporting Information File 1) to values of 1.5-3.5 nm for Pr, 1.0-2.5 nm for Gd and 4.0-7.0 nm for Er giving average diameters of 2 ± 1 nm for Pr, 1.5 ± 0.5 nm for Gd and 5 ± 1 nm for Er (Table 1). Interference patterns for Pr-NPs and Gd-NPs in close-up TEM images ( Figure 9 top and Figure S8, Supporting Information File 1) indicate crystallinity  of the RE-NPs. For the crystalline Er-NPs SAED gave the expected reflections of elemental erbium ( Figure S9b, Supporting Information File 1). The very small size of the particles failed to yield meaningful PXRD patterns. EDX (Figure 9 and Figure S9b, Supporting Information File 1) gave the expected bands for Gd and Er. No fluorine was detected by EDX analysis. The oxygen peak can be attributed mainly to air contamination when the sample was introduced into the TEM device.
Again, the formation of Pr(0), Gd(0) and Er(0) metal NPs were indirectly supported by the measured XPS binding energies of oxygen ( Figure 10, Table 4 and Figures S8b and S9c, Supporting Information File 1). The measured RE (RE = Pr, Gd, Er) binding energies had again shifted due the small NP size so that no clear assignment to metal(0) or metal(III) could be made. The measured XPS binding energies of oxygen are in very good agreement with the binding energies of organic oxygen that are given in literature and thereby exclude the formation of RE(III) oxide for RE = Pr(0), Gd(0) and Er(0) [48,49]. Understandably, the non-fluorous solvent PC ( Figure S3, Supporting Information File 1) gave no signals for fluorine.

Conclusion
Microwave-assisted thermal decomposition [56][57][58] using the rare-earth metal(III) coordination compounds tris(N,N′-diisopropyl-2-methyl-amidinate) RE(amd) 3 (RE = Pr(III), Gd(III), Er(III)) and tris(2,2,6,6-tetramethyl-3,5-heptanedionato)europium(III) (Eu(dpm) 3  The phase purity and the absence of oxide impurities was proven by powder X-ray diffraction (PXRD), selected area electron diffraction (SAED) and high-resolution X-ray photoelectron spectroscopy (HR-XPS). To the best of our knowledge, there have been so far no reports on the synthesis of non-oxidized nanoparticles of any rare-earth element by soft-wet chemical routes from metalorganic precursors. However, access to a simple, reproducible and scalable method to obtain RE-NPs in solution will be the key for developing the nano-chemistry of non-oxide (and nonfluoride) RE materials. Our results on praseodymium, gadolinium and erbium nanoparticles derived from microwaveassisted thermolysis of the respective metal amidinates RE(amd) 3 as precursors may open up new avenues for applications of pure rare-earth metal nanoparticles and nanomaterials derived from these. In particular we are aiming to study the nano-alloying of RE-NPs with late transition metals M and study the catalytic properties of the obtained intermetallic M/RE-NPs in extension of our previous work on Ni/Ga nanophases derived from organometallic precursors by co-thermolysis in ILs and PC [59].
The syntheses of the rare-earth metal amidinates (RE(amd) 3 , RE = Gd, Er, Pr) were performed according to literature procedures [60][61][62][63]. The rare-earth metal amidinates RE(amd) 3 were synthesized by an insertion reaction of methyl lithium into 1,3diisopropylcarbodiimide in THF. The resulting lithium amidinate solution was reacted with the RE halides in a salt metathesis reaction.

Synthesis procedures of rare-earth metal nanoparticles (RE-NPs) and rare-earth fluoride nanoparticles (REF 3 -NPs)
were based on previous literature [12]. HR-X-ray photoelectron spectroscopy: HR-XPS (ESCA) measurements were performed with a Fisons/VG Scientific ESCALAB 200X xp-spectrometer, operating at 70-80 °C, a pressure of 7.0 × 10 −9 mbar and a sample angle of 33°. Spectra were recorded using polychromatic Al Kα excitation (11 kV, 20 mA) and an emission angle of 0°. Calibration of the XPS was carried out by recording spectra with Al Kα X-rays from clean samples of copper, silver and gold at 50 eV and 10 eV pass energy and comparison with reference values.
Powder X-ray diffraction: PXRD data were measured at ambient temperature on a Bruker D2-Phaser using a flat sample holder and Cu Kα radiation (λ = 1.54182 Å, 35 kV). Samples had been precipitated with ethanenitrile from the NP/IL dispersion and washed several times with ethanenitrile. PXRDs were measured for 1 h. Small shifts in PXRD patterns are not uncommon for nanoparticles. A number of effects can be considered for such shifts including the range of stoichiometric composition, partly inhomogeneous element distribution, defects such as stacking and twin faults and nanosized crystalline domains being much smaller than the bulk reference material causing lattice contraction or expansion and strain [65][66][67][68][69].
Transmission electron microscopy: TEM was performed with a FEI Tecnai G2 F20 electron microscope operated at 200 kV accelerating voltage [70]. Conventional TEM images were recorded with a Gatan UltraScan 1000P detector. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) as shown in Figure S4a (Supporting Information File 1) was also performed with this microscope. TEM samples were prepared by drop-casting the diluted material on 200 μm carbon-coated copper grids or gold grids. The size distribution was determined manually or with the aid of the Gatan Digital Micrograph software from at least 50 (if not stated otherwise) individual particles. EDX spectra for composition analysis were recorded with the same instrument using an exposure time of 3 min.
Selected area electron diffraction: SAED patterns ( Figures S4  and S6, Supporting Information File 1) have been recorded with an FEI Titan 80-300 TEM [71], operated at 300 kV accelerating voltage. The area selection was achieved with a round aperture placed in the first intermediate image plane with a corresponding diameter of 0.64 µm in the object plane. For each acquisition a sample region with a significant amount of material was placed inside the aperture. The objected was illuminated with wide-spread parallel beam obtaining focused diffraction patterns. The diffraction images were calibrated with Debye-Scherrer patterns recorded from a gold reference sample (S106, Plano GmbH, Wetzlar, Germany).
Thermogravimetric analysis: TGA was performed with Netzsch TG 209 F3 Tarsus equipped with an Al crucible by using a heating rate of 10 K·min −1 .

Electrochemical measurements:
The ErF 3 working electrodes were prepared by coating an (N-methyl pyrrolidone)-based slurry composed of 75 wt % ErF 3 , 15 wt % conductive agents (Super P active carbon from Temical) and 10 wt % binder (PVDF) on a current collector (aluminium foil). A half-cell was assembled in an Ar-filled glovebox, with lithium foil as a counter electrode and 1 M LiPF 6 in ethylene carbonate-ethyl methyl carbonate (50:50) as the electrolyte. The cyclic voltammetry (CV) data of these half-cells were collected utilizing an electrochemical workstation (Autolab 302) with different cutoff potentials.