Flame Synthesis of Complex Fluoride-Based Nanoparticles as Upconversion Phosphors

Recent improvements in precursor chemistry, reactor geometry and run conditions extend the manufacturing capability of traditional flame aerosol synthesis of oxide nanoparticles to metals, alloys and inorganic complex salts. As an example of a demanding composition, we demonstrate here the one-step flame synthesis of nanoparticles of a 4-element non-oxide phosphor for upconversion applications. The phosphors are characterized in terms of emission capability, phase purity and thermal phase evolution. The preparation of flame-made β-NaYF4 with dopants of Yb, Tm or Yb, Er furthermore illustrates the now available nanoparticle synthesis tool boxes based on modified flamespray synthesis from our laboratories at ETH Zurich. Since scaling concepts for flame synthesis, including large-scale filtration and powder handling, have become available commercially, the development of industrial applications of complex nanoparticles of metals, alloys or most other thermally stable, inorganic compounds can now be considered a feasible alternative to traditional top-down manufacturing or liquid-intense wet chemistry.


Introduction
Nanoparticle powder technology is a widely applied industrial process for the preparation of advanced functional materials (Hosokawa, 2008). Predesigned and engineered nanostructures attract the interest of research and industry communities. Various methods of synthesis and control of these objects are constantly being developed and appear in high-impact journals, symposia and discussions. Despite a dramatic gap between academic methods for laboratory synthesis and the technological implementation on an industrial scale, various materials based on nanotechnologies are already available from large-scale production. Among the most promising methods are: CVD process for carbon nanotubes (Cassell et al., 1999); emulsion methods for polymer-ic nanoparticles (Muller et al., 2006) in drug delivery applications (PLGA, (Musyanovych et al., 2008)); solgel, e.g. LiFePO 4 (Lee et al., 2010); precipitation, e.g. BaSO 4 , (Adityawarman et al., 2005); and hydrothermal methods (Chen et al., 2009) for various nanoparticles. However, the aforementioned methods are often inapplicable for metastable phases which often possess high functional characteristics.
Gas-phase synthesis stands aside as an independent process to fabricate nanopowders, particularly with aerosol methods. This technique allows a multiscale and cheap production of nanoparticles. It is a relatively flexible technology by which also metastable phases can be obtained. Flame pyrolysis is currently applied in the large-scale production of SiO 2 and TiO 2 nanoparticles. The method is also promising for the production of nanostructured carbonates, e.g. CaCO 3 (Huber et al., 2005) and SrCO 3 (Strobel et al., 2006), sulfides, e.g. PbS and ZnS (Athanassiou et al., 2010), highly reactive metals, e.g. Co  and Ni (Jung et al., 2005), alloys, e.g. Cu-Ni , glasses, e.g. SiO 2 -CaO-P 2 O 5 -Na 2 O (Brunner et al., 2006), and halides, e.g. NaCl, BaF 2 , and CaF 2 , (Grass and Stark, 2005). An overview of nanoparticles prepared by flame-spray pyrolysis is shown in Fig. 1. The established materials, namely spherical oxide nanoparticles, a straightforward product of gas-phase synthesis, stand at the origin of the graph. This group typically includes titanium, silicon, and aluminum oxides. There has been evidence for more complex materials prepared by flame pyrolysis, e.g. Sr 5 (PO 4 ) 3 Cl:Eu 2 (Kang et al., 2003). A reducing atmosphere helps to step away from oxidic nanoparticles towards metallic compounds. Acetylene-fed flames form carbon shells on the surface of metallic nanoparticles (Grass et al., 2007). Such carbon coatings allow chemical functionalization strategies of the nanoparticle surface to use them in filtration and purification (Rossier et al., 2011), magnetic chemical reagents (Wittmann et al., 2010) and catalysts (Zeltner et al., 2011). Sulfides derived from flame pyrolysis belong to another group of widely applied chalcogenide salts. Recently, Athanassiou et al. successfully prepared doped ZnS:Mn 2 nanoparticles (Athanassiou et al., 2010).
Another group of industrially valuable non-oxidic salts are halides. Chlorides and fluorides of alkaline and alkaline-earth metals have been successfully obtained by gas-phase synthesis yielding nanoparticles of various morphologies and structures. Fluorides have unique optical and electronic properties which are widely used in biomaterials and electronic applications, though the need of more complex phosphors requires the preparation of glasses and mixed fluorides.
Recently, upconversion (UC) phosphors based on the rare earth and alkaline-earth materials have been investigated. In a UC process several low energy photons are absorbed, their energy is converted and finally a photon of higher energy is emitted. UC is also known as anti-Stokes emission (Auzel, 2004). It is known for f-elements such as the lanthanide ions Er 3 and Tm 3 , as well as U 3 , and several d-element ions embedded into specific matrices (Auzel, 2004).
Upconversion phosphors such as -NaYF 4 : Yb, Tm or Yb, Er are regularly synthesized as bulk microcr ystalline materials by high-temperature solidstate synthesis: NaF (1-x-y)YF 3 xErF 3 yYbF 3 NaYF 4 : (xEr 3 ,yYb 3 ) Targeting molecules in biological applications and thin coatings for solar cells imply size limits of the embedded upconversion particles from tens to hundreds of nanometers. For the same purposes, it is preferable to produce upconversion nanoparticles with high luminescence intensity. Few approaches in the production of nanocrystalline UC phosphor materials are known: decomposition of multiprecursors (Yi et al., 2004), co-precipitation (Martin et al., 1999), hydrothermal and solvothermal methods (Zeng et al., 2005). Nonetheless, those methods are limited to predictable cr ystallite morphology, low production rates and complicated synthesis schemes. Hexagonal sodium yttrium fluoride is one of the most efficient host matrices for NIR-to-visible upconversion phosphors (Sommerdijk, 1973 (Lim et al., 2009) successfully produced sub-10-nm polydisperse UC nanophosphors by flamespray pyrolysis in a one-step continuous synthesis. The host matrix used for their study was cubic Y 2 O 3 which is less efficient than hexagonal NaYF 4 (Auzel, 2004). Grass et al. (Grass and Stark, 2005) proposed a fluoride doping with rare earth elements by flame synthesis. As a result, it appeared feasible to synthesize upconverting sodium yttrium fluoride nanoparticles doped with Yb-Tm or Yb-Er, although the cubic phase might be obtained instead of the hexagonal phase which shows better UC properties. Due to the limited thermodynamic stability of the hexagonal low-temperature phase, see the phase diagram in Fig. 1b, the cubic high-temperature phase of NaYF 4 might be obtained from the synthesis as metastable kinetic product.
In the present work, flame pyrolysis is used to prepare nanoparticles of non-oxidic, doped rare earth fluorides. The upconversion emission during excitation with NIR laser and the thermal behavior of the derived phosphors are analyzed. The cubic-to-hexagonal phase transition of NaYF 4 will be examined for various syntheses and thermal treatment conditions.
The prepared stoichiometric mixtures of rare earth ethylhexanoates, sodium 2-ethylhexanoate, and fluorobenzene (ABCR-Chemicals, 99%) were combusted to produce NaYF 4 nanoparticles, doped with 25 mol.% Yb and 0.3 mol.% Tm, i.e. NaY 0.747 Yb 0.25 Tm 0.003 F 4 , and 20 mol.% Yb and 2 mol.% Er, i.e. NaY 0.78 Yb 0.2 Er 0.02 F 4 . The solution was pumped through a 0.4-mm-diameter capillar y at rates of 3, 5, 7, and 9 l min -1 into fuel flames. Alternatively, the liquid solution was dispersed into an aerosol and burnt in oxygen (99.8%, Pan Gas) at rates of 7, 5, and 3 l min -1 with a pressure drop at the capillary tip of 1.5 bar. A steady combustion was achieved by an oxygen (99.8%, Pan Gas) sheath gas flow of 230 l h -1 through a concentric sinter metal ring. A Teflon filter was used to collect the prepared particles of UCNP. The particles remained stable at ambient conditions. In the case of the reducing flame pyrolysis, the precursors were burned in a nitrogen-rich atmosphere using a nitrogen (5N, PanGas) glove-box with gas flow (Grass et al., 2007). This flow was circulated by a vacuum pump (Busch, Seco SV1040CV). The oxygen concentration was fixed below 100 ppm (volumetric) for the reducing flame pyrolysis.

Characterization of UC phosphors
The specific surface area was calculated by measuring the nitrogen adsorption at 77 K on a Tristar (Micromeritics Instruments) following the Brunauer-Emmett-Teller (BET) method. Prior to the surface area determination, samples were preheated in vacuum at 150 with p < 0.1 mbar during 1 hour.
The prepared NaYF 4 nanoparticles with Yb-Tm and Yb-Er rare earth dopants were sintered at various temperatures (500 -800 ) in air or nitrogen flow during 2 or 3 hours at heating rates of 10 min -1 . A "fast" heating of the NaYF 4 :Yb, Er upconversion phosphors was achieved by placing the powders directly in an 800 preheated furnace.
The relative upconversion luminescence was measured according to the following procedure: Powders of the UC phosphors were filled in glass tubes of 1 mm inner and 1.5 mm outer diameter and then fixed in a sample holder. The powder densities were the same for all samples, as the powders were pressed into the glass tubes with glass tips. The samples were excited by a 980-nm IR laser diode coupled to a 1-mm-diameter fiber. The non-focused laser beam illuminated the sample in a spot of about 1 mm 2 at a distance of 2-3 mm from the surface. The UC emission was collected by a Y-fiber, i.e. parallel to the excitation light, and measured by an Ocean Optics SD1000 spectrometer. The reflected IR laser light was blocked by a filter in front of the spectrometer. The laser power was measured by a power meter. The luminescence spectra were corrected for the spectral response of the detection system. The integrated emission peaks yielded the relative UC efficiencies of the powder samples with a reproducibility of 5%. The size and shape of the as-prepared upconversion nanoparticles were characterized by transmission electron microscopy (TEM) with a Philips CM30 ST (LaB 6 cathode, operated at 300 kV, point resolution 4 Å). Morphologies were also analyzed by scanning electron microscopy (SEM) with a Zeiss LEO 1530 Gemini. The phase composition was characterized by X-ray diffraction (XRD) patterns recorded with a PANalytical XPert PRO-MPD (CuKα radiation, X Celerator linear detector system, step size of 0.033 , ambient conditions). The mean cr ystallite size was estimated from X-ray diffraction patterns by the Scherrer equation.

Flame-spray powder morphologies and phase analysis
NaYF 4 : Yb, Tm powders were prepared by flamespray synthesis. The obtained nanopar ticles had diameters of 20-40 nm. The fuel/oxygen flow rates of 3/7, 7/3, and 7/9 L/min resulted in the formation of cubic α-NaYF 4 particles, see Fig. 2a. Equation (2) shows the reaction scheme, Liquid NaF NaYF 4 (2) where liquid corresponds to the melt in the flame before cooling down at the Teflon filter. The medium 5/5 l min -1 feeding rate led to a mixed product with hexagonal( ) and cubic(α) phases of sodium yttrium fluoride, see Fig. 2a.
The TEM analysis of this powder revealed the presence of hexagonal particles. The particles have sizes of about 10 nm with agglomerates up to 50 nm, see Fig. 2b.
By the BET method, specific particle surface areas from 29.5 to 31 m 2 g -1 were determined for nanoparticles prepared in oxidic or reduced atmospheres.
The calculated par ticle diameters according to equation ( The SEM micrographs of flame-sprayed NaYF 4 : Yb, Tm also demonstrated the homogeneous distribution of nanoparticles without any visible micron-sized agglomerates, see Fig. 3a. These particles are stable at room temperature under atmospheric conditions. Images of the as-prepared NaYF 4 :Yb,Er nanoparticles show agglomerated clusters in the 100-nm range with hexagonal crystallites of less than 50 nm in size, see Fig. 3b. It correlates to previously measured BET and TEM particle size estimations. During sintering, the grains grow and micron-sized cubic crystals of Y 2 O 3 form, see Fig. 3c. The image of a fragment in Fig. 3d shows narrow grain boundaries of -NaYF 4 crystals after 3 hours sintering at 700 under constant nitrogen flow.

Thermal treatment of UC phosphors
The thermal behavior of the synthesized powders was studied by differential thermal analysis (DTA), see Fig. 4a. At low temperatures, the adsorbed water was released from the sample according to a weight loss of 0.3 wt% In addition, the DTA signal indicates a minor exothermal peak at 380 It may correspond to the formation of traces of cubic yttrium oxide. At 400-500 , the TG curve shows a step correlated to an endothermic peak on the DTA curve.
Analysis of the phase composition of upconversion phosphors sintered at 500 showed decreasing halfwidths of the Bragg peaks which correspond to an increased size of the crystallites. The NaYF 4 : Yb, Tm powders were sintered at temperatures from 500 to 800 . The XRD curves up to 700 give no hint for a phase transition, see Fig. 4b. The sintering at 800 shows a partial transition of sodium yttrium fluoride from its cubic to the hexagonal phase. According to the NaF-YF 3 (Thoma et al., 1966) phase diagram, see Fig. 1b, the hexagonal -NaYF 4 phase is stable at room temperature whereas the high-temperature cubic α-phase is stable above 691 . Thus cubic NaYF 4 is obtained in the synthesis as a metastable, kinetically stabilized phase. Only at high enough temperatures, close to the αphase transition, can the α-phase overcome the activation energy barrier and transform into the -phase. The 800 curve in Fig. 4b shows a partial αphase transition. The sample temperature had obviously not been 800 in this experiment, because then no -phase would be obtained at all, but it had been high enough to partially activate the transition.
The thermal treatment at 700 in an air flow during two hours led to the formation of Y 2 O 3 . This hydrolysis results in HF gas evolution and leaves Y 2 O 3 and NaF behind, cf. equation (4).  In order to transform the cubic phase towards the more favorable hexagonal phase, the powders were sintered in vacuum. The X-ray diffraction patterns of samples treated at 500 demonstrate the growth of crystals, see Fig. 4b, but no formation of hexagonal NaYF 4 . Cr ystallite sizes derived from the Scherrer Equation show a growth from 40 nm to 80 nm. Sobolev et al. (Sobolev et al., 1963) earlier demonstrated that the NaYF 4 hexagonal phase undergoes a phase transition into NaYF 4 with cubic symmetry at 600 . Further examinations of the flame-spray synthesis conditions and thermal treatment modes such as cooling rate are required. Flame pyrolysis allows the synthesis of nanocrystalline upconversion phosphors with a high surface area which show luminescence (Fig. 5b, 6a and b).

Thermal behavior and luminescence of NaYF 4 :Yb, Er
NaYF 4 :Yb, Er samples were prepared using a 7/9 (L /min) feeding rate. The particles revealed green and red luminescence on excitation with a 980-nm IR laser. After sintering at 500 and 700 , the luminescence intensity increased because the crystallites grew in size. Fast heating, i.e. placing powders into a preheated furnace at 800 , resulted in rapid formation of hexagonal sodium yttrium fluoride, see Fig. 4b. As a result, the green emission intensity increased, see Fig.5a. Phosphors sintered under an inert atmosphere formed oxide phases, which in turn intensified the red emission in the UC luminescence spectrum, see Fig. 5a. Yttrium oxide formation in the inert atmosphere is based on phase kinetics and prior built-in oxygen. This embedding happens during flame pyrolysis in an oxygen-containing environment. The formation of micron-size cr ystallites is observed on the SEM micrographs, see Fig. 3c. A slow heating rate of 10 min -1 yielded a mixture of yttrium oxide and cubic NaYF 4 ; the fast heating leads to hexagonal NaYF 4 formation.
However, for the "fast heating" at 800 sample, the green-to-red emission ratio becomes higher because hexagonal -NaYF 4 formed which has a significantly stronger green emission. Photographs of the UC emission of the respective samples are shown in Fig.  6.

Conclusions
Pre-designed complex nanostructures with characteristic properties are built up by assembling their functional units step by step. In this regard, gasphase methods have a great potential for industrially realizing some of the promises of nanotechnology. Their scaling potential and access to various types of precursors covers a plethora of nanostructures such as metals, oxides, salts, and complex compounds. with Yb-Er (a) and Yb-Tm (b) rare earth dopant couples. The relative peak intensities increase after sintering from 500 to 800 .
In the present work, a successful bottom-up access through a refined flame-spray technique provides access to complex non-oxide particles such as rareearth-doped sodium yttrium fluorides. The co-doping with Yb-Tm and Yb-Er ion couples leads to blue and green upconversion luminescence, respectively. Particles of less than 50 nm in size were obtained under reducing conditions as cubic NaYF 4 . It was possible to tune a cubic to hexagonal phase transition by thermal treatment of the nanomaterial. A strongly enhanced UC luminescence intensity was observed for the -NaYF 4 . Upconversion luminescence spectra showed a correlation between crystallite size (i.e. low surface area) and luminescence intensity. Oxide impurities reduced the green and increased the red UC emissions in Yb 3 , Er 3 -doped phosphors.