Introduction

Nanoparticles offer one possibility for the breakthrough needed to advance the conversion of substances to materials [1]. Toward the materialization of nanoparticles, the advantages of nanoparticles may be described from two points of view. One is the so-called “nanosize effect” expected in physical properties, such as luminescence, or chemical properties such as catalytic activity. The other is the resource effect, which reduces the volume of materials required for functionalization. In both cases, the primary requirement is to homogeneously produce highly functional substances in large volumes. If this is achieved, the materialization of nanoparticles is possible.

The current situation, however, is that the properties of nanoparticles have not yet attained to their best. To effectively utilize the properties of nanoparticles, it will be important to control crystallinity and surface properties as well as the homogeneity of crystallite size and chemical composition.

Powder synthesis should be carried out to precisely control particle size, morphology, chemical composition, etc.… A monodispersed powder is defined to have a standard deviation of below 10%. Conventionally synthesized nanopowders, however, hardly satisfy this requirement. The morphology and chemical composition of particles are closely related to particle assembly and functionalization. Control of surface properties is essential to the application of nanoparticles in functional ordered-arrays. Dispersed structures also require controlled surface properties.

Nanoparticle synthesis of ceramic materials has conventionally been achieved by solution processes. Taking the example of luminescence properties, major factors that have hindered the functionalization of solution-synthesized oxide nanoparticles include low crystallinity, quenching by surface hydroxyl groups, incomplete composition control (doping concentration, surface segregation), etc.. Moreover, the existence of surface hydroxyls is known to have a large impact on the dispersion of nanoparticles in a solvent.

Thermal plasma of high temperatures exceeding 10,000 K has been effectively used to synthesize nanocrystalline powders. The synthesis began through an important contribution by Emeritus Professor Pfender, University of Minnesota.

In this review paper, thermal plasma materials processing and its characteristic, and then its advantages in nanoparticle synthesis are first described. Next, research achievements thorough the liquid precursor injection into radio frequency (RF) induction plasma are discussed, with the aim of materializing highly functional nanoparticles synthesized with plasma.

Synthesis of Ceramic Nanoparticles in Reactive Thermal Plasmas

Thermal plasmas possess many advantages and provide unique reaction fields for materials processing. The improved controllability of the reaction fields, in which chemical reactions proceed to yield materials, should lead to the synthesis of unique materials with respect to morphology, crystal structure, and chemical composition [2, 3]. Arc discharge, which is a thermal plasma, proves to be a useful tool in the synthesis and surface modification of a wide range of metals and inorganic materials. Its advantages as a reaction field are as follows. (1) Thermal plasmas have quite high temperatures (up to <15,000 K). As the plasmas are generated at relatively high pressures (close to atmospheric pressure), they have large densities and considered to be at equilibrium (local thermal equilibrium). That is, the temperatures of heavy particles (such as atoms, molecules, and ions) are almost equal to that of light species (i.e., electrons). (2) High concentrations of chemically reactive radicals enhance the reactivity of thermal plasmas. (3) Superfast quenching (<105 to <106 K/s) takes place in the plasma tail flame region.

A representative industrial use of thermal plasma processing is the spheroidization of materials with high melting temperatures and large particle sizes (e.g., 50–100 μm). Figure 1 shows the spheroidized micron-size particles of titanium carbide, which has a melting temperature higher than 3500 K [4]. In fact, 30 kg of refractory tungsten metal particles can be treated per hour, which corresponds to 1 ton per week [5]. As thermal plasma processing has inherent characteristics, i.e., processing at an industrially acceptable production rate, ongoing studies on nano-structure control of materials using thermal plasma technology would also make it possible for extended practical applications of nano-sized functional particles in a short time.

Fig. 1
figure 1

Spherical TiC particles prepared by in-flight melting and spheroidization in RF thermal plasma [4]

Full size image

Nano-sized particles are formed by a process of rapid quenching and condensation after precursors are supplied to thermal plasma and instantaneously evaporated. The rapid temperature decrease in the tail flame gives rise to supersaturation in a vapor phase, after which the nucleation process takes place. In the same way, the rapid cooling suppresses grain growth; therefore, nanoparticles below 100 nm in size are produced without difficulty. Advanced technologies should be pursued to control the plasma reaction fields to yield (a) high production rates, (b) reduction of particle size and narrowing of size distribution, and (c) high crystallinity and phase control. Practical applications of nano-sized particles would be possible as long as such technologies exist.

Synthesis of ceramic nano-sized powders, called ultra-fine powders in the 1980–1990s, using thermal plasmas gained much attention in the fabrication of more easily sinterable non-oxide powders, such as SiC and Si3N4. High-density sintered bodies of these materials were expected for engineering ceramic materials, which could be used at higher temperatures than metals. Pioneering studies headed by Professor Pfender demonstrated the synthesis of SiC and Si3N4 in direct current (DC) arc plasmas [6,7,8,9]. Studies on ultra-fine-powder synthesis of Si-based materials followed using RF-DC hybrid [10, 11] and RF [12,13,14,15] thermal plasmas. Also, hypersonic plasma particle deposition system, in which modified DC plasma torch was employed, showed in situ fabrication of free-standing bodies composed of nano-size grains [16].

Synthesis of Nano-structured Ceramic Materials Thorough Liquid Precursor Injection Into RF Induction Plasma

Thermal plasmas are mainly generated in DC arc torches and RF induction torches. Direct current arc discharges with high energy density has been widely used for various industrial applications, such as plasma spraying, plasma metallurgy, as well as the powder synthesis process [2, 3]. Compared with DC arc discharge, the RF induction plasma has more advantages. The plasma is inductively heated by an RF electrical power of tens of kW and a frequency in MHz. The diameter is 30–40 mm; therefore, the plasma volume is relatively large. The axial plasma velocity is as low as 10 m/s, and the residence time in the high-temperature plasma flame is as long as tens of ms, which is long enough for chemical reactions to proceed. The most important characteristic comes from the fact that the generation of RF induction plasma does not involve electrodes, and any gas can be used to generate the plasma, regardless of its oxidative, reductive, or reactive property. The above-mentioned characteristics of thermal plasma provide unique reaction fields for materials processing.

Flame pyrolysis [17,18,19,20,21] is a straightforward way to generate well-dispersed crystalline nanoparticles at an acceptable scale. It is believed that P25 powder is produced via flame oxidation of TiCl4 vapor. Radio frequency thermal plasma processing has similarities to flame pyrolysis but is characterized by its much higher temperature (~104 K) and fast quenching (105–6 K/s) at the tail part of the plasma. Nanoparticle production by using the advantages of thermal plasmas, i.e., high-temperature heat source with tremendously large enthalpy, began with the use of solid-sate precursors. Relatively coarse particles were introduced into thermal plasmas to synthesize nano-sized particles through evaporation and the subsequent coagulation processes. With this production method, problems arise from the residual coarse particles. When precursor powders are injected into a thermal plasma, in which the temperature and flow distributions exist both radially and axially, some of the precursor particles are not evaporated completely. The non-evaporated particles, though a small number, necessitate an additional post-separation process [22,23,24].

Injection of liquid precursors into RF induction plasma has advantages in synthesizing nano-sized functional ceramic powders. When a precursor mist with a droplet size of tens of μm is fed into a plasma having a high temperature exceeding 10,000 °C, the mist vaporizes instantaneously, enabling nanoparticle formation. Oxides with the prescribed cation ratio of the liquid precursor can be synthesized with this technique, and it is possible to adjust the chemical composition, which is linked to the appropriate functions of ceramic materials. Compositional control, such as doping content into host materials and the variable chemical composition in complex oxides, is essential to enable these functions.

The use of liquid precursor is also relatively easy compared to the mixing of gas precursors. Gas-phase synthesis of Al-doped TiO2 was reported using gas-phase precursors, TiCl4 and Al2Cl6 gases [25]. Mixed Ar–TiCl4 gas was generated by bubbling of Ar gas in liquid TiCl4 maintained at room temperature, while mixed Ar–Al2Cl6 was generated by Ar gas flowing over solid AlCl3 heated at 378 K. A simplified process was conducted for the thermal plasma synthesis of Al-doped TiO2 nano-sized powders by injecting an AlCl3-dissolved methanol solution into Ar-O2 plasma together with Ar–TiCl4 gas [26].

Precise synthesis of functional oxide was reported with the injection of aqueous solution precursors into Ar RF induction plasma generated at atmospheric pressure [27,28,29]. Variation in particle size and morphology was reported from the synthesis of nano-sized particles of single-component oxides and binary oxide systems. Synthesized particles of single-component oxides were spherical for ZrO2, Y2O3, Sm2O3, La2O3, δ-Al2O3, TiO2(atnatase), β-Bi2O3, and CuO; plate-like for Nd2O3, Cr2O3, and Pr2O3; polyhedral for PrO2, CeO2, and γ-Fe2O3; cubical for NiO, MgO, CaO, Co3O4, and Mn3O4; bar-like for SnO2 and ZnO; and foil-like for β-PbO and MoO3. In binary oxide systems, (Cr2O3, Fe2O3, SnO2)–Al2O3, the distribution of component oxides was carefully investigated. For the Cr2O3–Al2O3 system, each component independently constituted its particles that were homogeneously distributed in the powder. With Fe2O3–Al2O3, the two components formed solid solutions. In SnO2–Al2O3, the powders were composed of each component, where SnO2 had a strong tendency to occupy the particle surface. For complex oxides, this technique was applied to the synthesis of nano-sized particles of perovskite-type oxides, RFeO3(R: Eu, Yb), thin films of garnet-type oxides, Y3Al5O12 and Y3Fe5O12, barium ferrites, Ba(Fe,Co,Ti)12O19, barium hexaaluminate, BaAl12O19, and superconducting oxides of the Bi–Pb–Sr–Ca–Cu–O system [30,31,32,33,34].

For the synthesis of Ti-containing materials, titanium alkoxides were used as precursors because an appropriate water-soluble precursor was not available. As alkoxides are susceptible to water and even water vapor in air to undergo hydrolysis, a stabilizing reagent, tri-ethanol amine, was added to the liquid precursors [35,36,37,38,39]. Films of Ti-based materials, TiN, TiC, TiN–AlN, TiC–TiN, SixN–TiN, and Ti–Si–B–C systems, were prepared by injecting liquid precursors containing alkoxides and a stabilizing reagent. The procedure overcame the difficulty of processing by using water-sensitive alkoxides included below. In the synthesis of TaC nano-sized powders, a liquid precursor, tantalum ethoxide, was fed into Ar–H2 RF induction plasma to form TaC through pyrolytic decomposition. As tantalum ethoxide is more reactive to water than titanium alkoxides, the liquid feeding system was set in an Ar-containing glove box [40].

Aqueous TiCl3 solution was also used as a Ti source [41, 42]. Such a solution as a precursor is advantageous over widely used titanium(IV) compounds such as tetrachloride and alkoxides, in that it is stable under ambient conditions; thus, easily manipulated. The use of TiCl3 led to the simultaneous incorporation of chlorine into the TiO2 lattice, which exhibited enhanced photocatalytic properties and degraded luminescent properties.

Size Control of Nanoparticles in Thermal Plasma Processing

Nano-scale size control of particles is crucial to the appearance of functions, in view of the quantum-size effects in catalytic, electric, magnetic, and optical properties. Also, narrowing the size distribution should contribute to the extension of the application area. The easiest and most effective way to synthesize uniform nano-sized particles is to allow chemical reactions to proceed under dilute conditions. On the contrary, practical process-control should be done under high precursor loading conditions.

The effects of processing parameters were investigated in the synthesis of TiO2 nanoparticles via Ar/O2 RF thermal plasma oxidation of liquid precursor mists [43]. Nucleation from a gas phase and the subsequent grain growth take place in the plasma tail. Therefore, size control should be done by modifying the temperature and flow distribution in the downstream of the plasma. We have demonstrated the particle size controlled synthesis of well-dispersed TiO2 nanopowders via Ar/O2 thermal plasma oxidation of liquid precursor mists through manipulating the cooling rates by injecting quench gases (Ar, He) [44]. In this case, the TiO2 nanoparticles were synthesized via Ar/O2 thermal plasma oxidation of atomized liquid precursors containing titanium tetrabutoxide (TTBO), the titanium source, and diethanolamine, a chelate preventing the hydrolysis of TTBO. The RF power at a frequency of 2 MHz and power level of 25 kW was introduced to generate the plasma at a pressure of 53.3 kPa. The precursor was atomized into mists at the tip of the atomization probe, which was used to inject apatite suspensions for plasma-spraying coatings [45], by Ar carrier gas flowing through the probe.

Quench gas was injected two in different ways: transverse (Fig. 2a) and counter (Fig. 2b). Figure 2 compares the morphologies of the TiO2 powders synthesized with and without Ar quench gas injections. X-ray diffractions showed that the synthesized powders consisted of rutile and anatase phases of TiO2. The majority of particles were rounded, with sizes ranging from a few to about 200 nm. The plasma-generated particles showed weak agglomerations, which is in sharp contrast to those synthesized via most wet-chemical techniques.

Fig. 2
figure 2

Experimental setups for a transverse swirl-flow injection and b counter-flow injection of quench gases [44]

Full size image

The overall morphology of the powder does not alter significantly by injecting 100 l/min of Ar in the transverse mode (Fig. 3b), but the powder becomes appreciably finer by injecting the same amount of Ar in the counter flow mode (Fig. 3c), revealing the significance of the quenching method on powder properties. The size distributions of the powders were evaluated through an image analysis of photos taken using a scanning electron microscope (SEM). It was found that the powders synthesized without quench gas and with Ar transverse flow have similar median diameter, d 50, of ~55 and 53 nm, respectively, while that made with Ar counter flow has a considerably smaller d 50 of ~35 nm. In addition, the size distribution of the powder substantially narrowed by applying the counter flow Ar quench gas. The standard deviation σ for the powders synthesized without quench gas was evaluated to be 22 nm. The Ar transverse flow increased σ to 30 nm, while the Ar counter flow decreased σ to 12 nm.

Fig. 3
figure 3

SEM micrographs showing morphologies of TiO2 powders synthesized a without use of quench gas, b with transverse swirl-flow injection of 100 l/min of Ar, and c with counter-flow injection of 100 l/min of Ar [44]

Full size image

Modeling was conducted to better understand the effects of quench gases and optimize the process parameters and reactor design [44]. The plasma flow and temperature fields were simulated using a two-dimensional and turbulence fluid model. Since the particles are synthesized from the vapor phase of the precursors, their trajectories should follow the streamlines of the fluid due to the small masses of the particles. As a simplified approach, the motion of a test particle with negligible mass and the infinite melting point in the plasma flow were traced. Figure 4 illustrates the calculated streamlines and temperature fields in the reactor. The dots in Fig. 4 indicate the predicted trajectories of a test particle released at the tip of the central probe. Comparing Fig. 4a, b, we can find that transversely injecting Ar does not significantly alter the temperature fields of the plasma and trajectories of the test particle. When Ar is injected counter to the plasma plume, however, it profoundly affects the temperature field and trajectories of particles (Fig. 4c). The greatly decreased size of the plasma plume allows the particle to remain for a shorter duration in the high temperature zone; hence, favoring finer particle formation. The above numerical analysis results support the experimental data given in Fig. 3.

Fig. 4
figure 4

Streamlines and temperature distribution for: a no quench gas, b transverse swirl-flow injection of Ar at 100 slpm, and c counter flow injection of Ar at 100 slpm [44]

Full size image

The effect of mass number of quenching gas was also observed in the synthesis of TaC nano-sized powder by injecting radially atomized mists of tantalum ethoxide into Ar-H2 RF induction plasma at the bottom of the plasma torch [40]. The atomized precursor rapidly dissociated in the high-temperature region of the plasma flame and recombined as the temperature decreased. Nano-sized tantalum carbide was easily prepared, as shown in Fig. 5. The particle size decreased with the increase in the plasma cooling rate by injecting quenching gas into the tail flame. This technique can be used to decrease the temperature locally in the injected region and control the chemical reactions there. According to Soucy et al. [46], injecting Ar at 20 l/min should have an adequate cooling effect. However, the particles synthesized with the tail injection of Ar were not smaller than those synthesized when it was not injected. The broadening of size distribution shown in Fig. 6 indicates the turbulence of flow induced by the Ar gas radial injection. When He, which has a thermal conductivity higher than that of Ar, of small atomic weight was injected as the tail quenching gas, the synthesized TaC particles were much smaller than those synthesized when no tail quenching gas was injected. The average size of the synthesized TaC nanoparticles decreased to 8 nm when He was injected at the plasma tail.

Fig. 5
figure 5

TEM image of plasma-synthesized TaC nano-sized particles [40]

Full size image
Fig. 6
figure 6

Size distributions of TaC nano-size particles synthesized with and without tail injection gases (pressure, 500 Torr). Number after ±is SD of particle size, dp [40]

Full size image

Synthesis of Titanium Oxide Nanoparticles: Phase Selection and Formation of Non-equilibrium Chemical Composition

Non-equilibrium composition and structure were also observed in the nanoparticles as a result of rapid cooling at the plasma tail flame and are extremely interesting phenomena from the scientific viewpoint.

Titanium dioxide, which has classically been used as a white pigment, exhibits interesting properties, such as high transparency in the visible wavelength region, high refractive index, and remarkable chemical and thermal stabilities, allowing it to be applied in photocatalysis, solar cells, semiconducting gas sensors, bio sensors, electronic and magneto-electronic devices, and as a building block for photonic crystals.

Titanium dioxide is known to have two common polymorphs: anatase and rutile. Interestingly, thermodynamically metastable anatase predominates undoped TiO2 nanopowders, which can be explained from a kinetic viewpoint based on classical homogeneous nucleation theory [47, 48], while decreasing the oxygen partial pressure in the vapor phase resulted in pure rutile nanoparticles [49].

The nucleation of rutile and anatase from TiO2 melt was analyzed by thermodynamic calculation using estimated values of the interfacial energy between the liquid and solid phases in the pure TiO2 system [47]. Based on classic homogeneous nucleation theory, estimated interfacial energies were used to evaluate critical nucleation energies for homogeneous nucleation in melt. Figure 7 shows the temperature dependence of the ratio ΔG rutile*/ΔG anatase*, where ΔG rutile* and ΔG anatase* are the critical free energies for stoichiometric rutile and anatase TiO2, respectively. It can be seen that the value of ΔG rutile*/ΔG anatase* is greater than unity over a wide range of temperatures, indicating that metastable anatase TiO2 nucleates preferentially. This phase formation behavior agrees with the experimental observation that, under a high cooling rate, anatase solidifies from the melt directly in high-temperature syntheses with a high degree of undercooling, while rutile solidifies at a temperature close to the melting point of TiO2.

Fig. 7
figure 7

Temperature dependence of critical free energy ratio, ΔG rutile*/ΔG anatase*, where ΔG rutile* and ΔG anatase* are critical free energies for stoichiometric rutile and anatase TiO2, respectively [48]

Full size image

The structural characteristics in the two TiO2 phase are apparently related to the formation of the defective rutile structure (with an oxygen deficiency) that forms from doping of metal ions with lower valence than Ti4+ or under reduced conditions, which alter the densely arranged lattice plane in the tetragonal structure of the crystals. The tolerance of anatase and rutile against the presence of defects can be qualitatively correlated with the local environment of titanium in anatase and rutile. In both structures, Ti is sixfold-coordinated, but the number of shared octahedral edges increases from two in rutile to four in anatase. Traditional crystal chemical theory argues that shared edges should lead to cation–cation repulsion and structural destabilization, in accordance with the relative stability of both phases. Rutile has greater tolerance than anatase toward oxygen vacancies due to fewer shared edges in the crystal structure of rutile. Substitution of Ti4+ in TiO2 with trivalent ions or divalent ions, such as like Fe3+ and Co2+, introduces oxygen vacancies to maintain charge neutrality. Synthesis of TiO2 under reduced atmosphere also gives rise to the formation of oxygen vacancies. Therefore, rutile TiO2 is preferentially formed [48, 49].

The next key point in this chapter is the use of liquid mixtures, in which some metal elements exist as ions in aqueous solutions or as constituent atoms of metallorganic compounds. The use of liquid precursors enables uniform nanoparticles to be formed and has the advantages of high production rate. Precise control of chemical composition also provides the opportunity to obtain particles of non-equilibrium doping, in which the doping quantity is larger than the solubility limit under equilibrium condition.

Doped TiO2 nanoparticles have been synthesized via Ar/O2 thermal plasma oxidation of atomized liquid precursors. In the case of iron doping, mixtures of TTBO and ferrocene dissolved in ethanol were used as precursors [48]. The liquid precursor for uropium doping was made in the following procedure [50]. Titanium tetraisobutoxide was added to diethanolamine to stabilize the TTBO against hydrolysis. Separately, uropium nitrate and citric acid were dissolved in water, and the pH of the solution was adjusted to 9.0 with ammonia solution. Mixing the above two solutions yields a stable clear solution to be used as the liquid precursor. For Er3+ and Co2+ doping, the same procedure as Co2+ doping was used to prepare clear liquid solution without any precipitate [51, 52].

In iron-doped TiO2 nanoparticles, no other phases except anatase and rutile TiO2 were identified for a wide range of iron to titanium atomic ratios (R Fe/Ti ) ranging from 0 to 0.2, though the solubility of iron in TiO2 was reported to be around 5 at.% by conventional wet processes [48]. Iron doping strongly promoted the formation of rutile because rutile is more tolerant than anatase against defects such as oxygen vacancies resulting from the substitution of Fe3+ for Ti4+ in TiO2. The concentration of oxygen vacancies reached its maximum at R Fe/Ti  = 2%, above which the excessive oxygen vacancies tended to concentrate. As a result of this clustering, an extended defect, such as crystallographic shear structure, was established (Fig. 8). The highly iron-doped TiO2 nanocrystals showed paramagnetic properties, as Fe3+ substitutes Ti4+ and is uniformly distributed in the TiO2 lattice [53].

Fig. 8
figure 8

HRTEM images of highly iron-doped nanopowder with RFe/Ti = 0.2. d spacing of 0.324 nm corresponds to rutile {1 1 0} planes [48]

Full size image

The solubility of Co2+ in the TiO2 lattice was determined to be around 2.0 at.% [52]. The solubility limit is substantially smaller than that (~20 at.%) for Fe3+ in TiO2, which might be understood by considering the following facts. (1) For sixfold coordination, Co2+ (0.0745 nm, high spin state) is much larger than Fe3+ (0.0645 nm, high spin state), whereas the later has an ionic size close to that of Ti4+ (0.0605 nm). (2) Every two Fe3+ ions, while only one Co2+, are needed to create one oxygen vacancy for charge compensation; hence, within the tolerance of the TiO2 lattice, the amount of Co2+ that can be accommodated into TiO2 is smaller. (3) Cobalt oxide is more alkaline than its iron counterpart from the Lewis definition of acid and base; therefore, cobalt oxide is more readily able to form compounds (CoTiO3, Co2TiO4) with TiO2, instead of Co2+ incorporation into TiO2 via Ti4+ substitution by Co2+.

The amount of Eu3+ that can be doped into the TiO2 lattice was limited up to 0.5 at.%, above which Eu2Ti2O7 pyrochlore was formed. Such a phenomenon is conceivable from the quite large size discrepancy between Ti4+ (0.0605 nm for sixfold coordination) and Eu3+ (0.0947 nm for sixfold coordination) ions. Figure 9 illustrates typical emission spectra of TiO2:Eu3+ nanoparticles as well as those of pure Eu2O3 and Eu2Ti2O7 under 325-nm He-Cd laser excitation [50]. It is the 5D0 → 7F2 transition that gives a sharp red emission peak. The Eu3+-doped samples exhibit emissions clearly different from those of Eu2O3 and Eu2Ti2O7, in terms of peak positions and peak shapes, implying different local environments. Efficient non-radiative energy transfer from the TiO2 host to Eu3+ ions, which was seldom reported in the wet-chemically derived nanoparticles or thin films of the same system, was confirmed by combined studies of excitation, UV–Vis absoption, and photoluminescence (PL) spectroscopy. Single-particle fluorescence spectroscopy was also used to investigate the defect-mediated PL dynamics of Eu3+-doped TiO2 nanoparticles [54]. The PL spectra and time traces of individual nanoparticles or aggregates revealed that the PL band originating from defects at the TiO2 surface appears in the visible region. Energy transfer from defects to surface-located Eu3+ ions is experimentally suggested.

Fig. 9
figure 9

Emission spectra of Eu3+-doped TiO2 nanopowders compared with those of Eu2O3 and Eu2Ti2O7 [50]

Full size image

Highly crystalline, highly luminescent nanopowders of Er3+-doped TiO2 have been also synthesized [51]. Results were similar to those for the Eu3+-doped TiO2. The large size mismatch between Ti4+ (0.065 nm) and Er3+ (0.089 nm) made it difficult for Er3+ to enter the TiO2 lattice. Approximately 0.5 at.% of Er3+ could be doped into the TiO2 lattice, above which Er2Ti2O7 pyrochlore appeared as an impurity phase. Bright PL at ~1.53 μm was observed from the nanopowders either by directly exciting the Er3+ activator or exciting the TiO2 host lattice. A comparative study showed that the nanopowder of the same system made using a wet-chemical coprecipitation method did not show bright PL due to the lack of energy transfer from the TiO2 host to Er3+ activators.

High-concentration niobium (V)-doped TiO2 nanoparticles of non-equilibrium chemical composition have been synthesized with various Nb5+ concentrations (Nb/(Ti + Nb) = 0–25.0 at.%), in which niobium pentaethoxide was used as a niobium source [55]. Solubility as high as ~25.0 at.% has not been achieved with wet-chemical techniques. The preferable anatase formation was attained in the plasma-synthesized powders and enhanced by niobium doping. All the powders were heated at high temperatures (600–800°C) to investigate their phase transformation, band gap variation, inter-particulate binding behavior, and photocatalytic stability. The transformation from anatase to rutile was effectively inhibited by increasing the Nb5+ content. The Nb5+ doping prevented the band gap of TiO2 from narrowing after heating. At high temperatures, Nb5+ doping could not only preserve particle size but also prevent inter-particulate binding.

In (Eu3+–Nb5+)-co-doped TiO2 nanopowders, doping Nb5+ cannot have an appreciable effect on Eu3+ solubility (0.5 at. %) in the TiO2 host lattice, but can significantly inhibit the increase in the rutile weight fraction for TiO2. The 617-nm PL intensity at 350-nm indirect excitation through energy transfer is considerably weaker than that at 467-nm direct excitation, indicating that a defect state level in the TiO2 host lattice might be lowered below the excited state of Eu3+ by doping Nb5+, which is conceivable from a relatively large amount of oxygen deficiencies yielded in the TiO2 host lattice [56].

Photocatalystic Properties Of Nano-sized Titanium Oxide Particles Doped with Transition Metals

Since the discovery of the Honda–Fujishima effect in 1972 [57], semiconductor photocatalysts have drawn enormous attention due to their potential application in water and air purification, anti-virus, deodorizing, self-cleaning, as well as photocatalytic hydrogen generation and CO2 reduction. Although different types of photocatalysts have been developed over the past 40 years, TiO2 is still regarded as the most promising due to its high chemical stability, low toxicity, and suitable band position. However, TiO2 can only respond to UV light, which greatly inhibits its utilization under visible light. Therefore, significant efforts have been made to make TiO2 exhibit activity under visible light, such as doping and noble metal supporting.

Well-crystallized iron(III)-doped TiO2 nano-size powders with controlled doping concentration and uniform dopant distribution prepared using a thermal plasma method were applied to photocatalytic measurement [58]. The photocatalytic reactivity of as-synthesized TiO2 nanopowders with iron doping concentration (R Fe/(Fe+Ti)) up to 16.4 at.% was quantified in terms of the degradation rates of methyl orange (MO) in aqueous TiO2 suspension under UV and visible light irradiation. The doping concentration was larger than that at the optimal iron doping concentration under UV irradiation; therefore, the nominally undoped TiO2 had the highest photocatalytic reactivity, suggesting that iron doping in TiO2 with a particle size of 50–70 nm is detrimental. Under visible light irradiation, however, TiO2 with an intermediate iron doping concentration at ~1 at.% had the highest photocatalytic reactivity due to the narrowing of the band gap so that it could effectively absorb the light with longer wavelengths. This suggests that iron doping with an optimal concentration holds promise for finding photocatalysts that are active in the visible light region.

The photocatalytic properties of TiO2-based nano-size powders were further evaluated: Nb-doped [55], Nb- and Fe-codoped [59], Nb- and Eu-codoped [60], and Cl-doped [41] TiO2. The properties were discussed in terms of defect chemistry to understand what effect oxygen defects have on photocatalytic performance and extend the potential applications of TiO2 as a photocatalyst into the visible light region. More interestingly, Nb- and Eu-codoped powder exhibited improved photocatalytic performance under visible light illumination with the 600–700-nm wavelength range, which is relatively long in visible light wavelength range (400–700 nm) [60].

Phase Formation in Other Nano-sized Oxide Particles Composed of Three Metal Elements

Cubic-structured Y2O3:Eu3+ solid solutions (optimal Eu3+ content: ~5 at.%) exhibit sharp red emissions at ~610 nm arising from the 5D0–7F2 electronic transition of Eu3+. The material is currently the most widely used red phosphor for frequent applications in areas such as fluorescent lamps, white LEDs, plasma display panels, flat-panel displays, field emission displays, and cathode-ray tubes. The Y2O3:Eu3+ phosphors are classically made via solid state reaction at ~1500 °C and are usually characterized by large particle size (up to tens of μm), irregular particle shape, and significant aggregation. Current advances in higher solution displays, however, require finer phosphors of better dispersion to improve resolution by decreasing pixel size. Yttrium oxide, Y2O3, is generally cubic structured under ambient conditions, but will transform into a monoclinic polymorph upon being heated to close to its melting point (~2430 °C). The monoclinic polymorph of Y2O3 is nonetheless frequently observed in nanopowders made via flame pyrolysis, and Skandan et al. argued that this is due to a ‘‘size effect’’, i.e., the very tight curvature of the nanoparticles can increase the internal pressure (ΔP = 2σ/r, s: surface energy and r: radius of particle) within the particle [61]. In flame pyrolysis and in thermal plasma processing, particles are formed via nucleation/growth through gas–solid and/or gas–liquid–solid processes. Upon cooling of the species generated at a high temperature close to or above the melting point of Y2O3, the monoclinic polymorph would nucleate with priority then quenched to the final product by subsequent rapid cooling.

Well-dispersed ultrafine particles of Y2O3:Eu were synthesized in one step via RF thermal plasma oxidation of atomized alcohol solutions of the component nitrates [62]. Powders collected at the reactor wall and filter were examined by X-ray diffractometry (XRD). The powders of pureY2O3, collected from either the reactor wall (Fig. 10, line a) or filter (Fig. 10, lineb), are mixtures of the monoclinic and cubic phases. For pure Y2O3, the powders from the reactor wall and filter have ~54.8 and ~44.5 wt% of the monoclinic phase, respectively. Doping with only 5at % of Eu3+ raised the monoclinic content to ~63 wt% (Fig. 10, line c) while codoping with 20 at.% of Sc3+ (relative to total cations) turned the product into an almost pure cubic phase (monoclinic content: ~10 wt%, Fig. 10, line d). The enhanced monoclinic crystallization by Eu3+ doping arises from the much lower temperature of the cubic-monoclinic phase transition of Eu2O3 (~1100 °C under atmospheric pressure and ~900 °C under 2.5 GPa). The low transition temperature allows the monoclinic polymorph to be stable over a much wider temperature range; thus, enhanced nucleation of the monoclinic phase can be expected. The significantly suppressed monoclinic crystallization by Sc3+ codoping is due to the fact that Sc2O3 (melting point ~2400 °C) has no reported cubic → monoclinic phase transition; thus, its presence retards monoclinic nucleation and stabilizes the cubic phase at the same time. Shifting of the C(440) diffractions (Fig. 10) to the lower-angle side by Eu3+ doping (lattice expansion) and to the higher angle side by Sc3+ codoping (lattice contraction) suggests the direct formation of solid solutions, and the shifting is understandable from the differing ionic sizes of Y3+ (0.0900 nm), Sc3+ (0.0745 nm), and Eu3+ (0.0947 nm).

Fig. 10
figure 10

XRD patterns of oxidation products with nominal compositions of a pureY2O3 collected from reactor wall and b pureY2O3, c (Y0.95Eu0.05)2O3, and d [(Y0.8Sc0.2)0.95Eu0.05]2O3 collected from filter. ‘‘M’’ and ‘‘C’’ denote monoclinic and cubic structured phases, respectively [62]

Full size image

Barium titanate exhibits excellent dielectric properties such as high dielectric constant, small dielectric loss, low leakage current, and high dielectric breakdown strength. It has a wide range of applications in electronic and electro-optic devices, such as thermistors, multilayer ceramic capacitors (MLCC), microwave absorbers, and transducers. The progress in electronic technologies has led to a demand for MLCCs of higher density and capacity, but to fabricate them, we need finer BaTiO3 powders with particle sizes of below 50 nm. Barium titanate is used in combination with special additives (Sr2+) that modify and enhance its dielectric properties. It is a ferroelectric perovskite, while SrTiO3 is a quantum paraelectric material. The Curie temperature can be adjusted by varying the Ba/(Ba + Sr) ratio. Barium titanate is noncentrosymmetric tetragonal, and adding Sr changes it into a centrosymmetric cubic phase. The properties of barium strontium titanate ((Ba, Sr)TiO3) are usually dependent on the particle size, purity, dispersability, and homogeneity of powders.

Barium strontium titanate nano-sized powders were synthesized via spray pyrolysis of a liquid precursor mist [63]. To adjust the cation stoichiometry of the complex oxide, liquid precursors were prepared by mixing titanium butoxide stabilized with diethanolamine and aqueous solutions of barium nitrate and strontium nitrate stabilized with citric acid. The molar ratios of titanium, barium, and strontium ions (Ba/(Ba + Sr) and (Ba + Sr)/Ti ratios) in the liquid precursor were varied. Changing the metallic ratio in the liquid precursors had a strong effect on the formation of impurity BaCO3 and Ba2TiO4 phases. Figure 11 shows the XRD patterns of (Ba, Sr)TiO3 with Ba/(Ba + Sr) ratios from 0.0 to 1.0, in which almost no diffraction from impurities was observed. The Ba/(Ba + Sr) ratio was varied to prepare impurity-free powders at 0.7, 0.8, and 0.9 with the (Ba + Sr)/Ti ratios of 0.98, 0.98, and 0.97, respectively. The peak positions of (Ba, Sr)TiO3 shifted as a result of changing the Ba/(Ba + Sr) ratio in the liquid precursor. The increase in lattice spacing from SrTiO3 to BaTiO3 can be explained on the basis of the ionic radius of the two cations. Ba2+ (0.161 nm) has a larger radius than Sr2+ (0.140 nm); hence, substitution of Ba2+ for Sr2+ will surely enlarge the lattice.

Fig. 11
figure 11

Variation in XRD patterns of (Ba, Sr)TiO3 powders made by plasma pyrolysis of liquid precursor with Ba/(Ba + Sr) ratio of precursors. (Ba(Ba + Sr) = 0.0–0.6 of (Ba + Sr)/Ti = 1.00), (Ba/(Ba + Sr) = 0.7–0.8 of (Ba + Sr)/Ti = 0.98), (Ba/(Ba + Sr) = 0.9 of (Ba + Sr)/Ti = 0.97), and (Ba/(Ba + Sr) = 1.00 of (Ba + Sr)/Ti = 1.00) [63]

Full size image

Highly Dispersed Behavior of Thermal Plasma-Synthesized Nanoparticles in Liquid Media

Nanopowders agglomerate more strongly than submicrometer powders, and some of the particles coagulate strongly. Agglomerated nanoparticles do not fragment easily and are difficult to disperse perfectly. Therefore, easily dispersive nanoparticles are needed to extend the application area of nano-size powders. Spherical TiO2 nanoparticles, with mean primary particle sizes of 29 and 53 nm, denoted as Plasma A and B powders, respectively, were synthesized in Ar/O2 RF induction thermal plasma pyrolysis of atomized liquid precursors containing titanium tetrabutoxide [44]. The plasma-synthesized nanoparticles at 1.0 vol% in solid fraction were dispersed by ultrasonic irradiation in the aqueous solution of polyacrylic acid [64]. Figure 12 shows transmission electron microscope (TEM) images of the particles of Plasma A and B powders and of the commercial P25 powder. Nanoparticles were taken from each suspension, which had all been ultrasonically irradiated, and put on a carbon grid. Particles aggregated automatically as they dried, and the commercial P25 powder showed the strongest coagulation among the primary particles (Fig. 12c). However, the particles of Plasma A and B powders did not appear to have coagulated very much (Fig. 13a, b). The mean aggregate size of the plasma-synthesized powders in aqueous suspension measured using a particle size analyzer corresponded exactly to the mean primary particle size evaluated by examining surface area data and analyzing SEM images, which suggests that the plasma-synthesized TiO2 nanoparticles were individually dispersed in water [65].

Fig. 12
figure 12

TEM images of TiO2 nanoparticles taken from suspensions of a plasma B b plasma A, and c P25 powders [65]

Full size image
Fig. 13
figure 13

Aggregate size of TiO2 nanoparticles in aqueous suspension. “Sediment” under “measurement method” denotes measurement using X-ray particle size analyzer. “SEM image” denotes measurement by analyzing SEM images. “Mean particle size from BET” denotes measurement from surface area and powder density data [65]

Full size image

Future Perspective on Applications of Functional Nanoparticles Prepared by Thermal Plasma Processing

When aiming at the application of highly functional nanoparticles, it is considered necessary to structure nanoparticles in the form of (1) dispersoids, (2) fillers, and (3) patterns. For example, in the case of red, green, and blue fluorescent particles used in plasma displays, fluorescent particles of several micrometers in diameter produced using a solid-phase method by comminution are currently used. However, when using even smaller fluorescent nanoparticles of several tens of nm with excellent dispersion, the resolution markedly improves and light scattering decreases, as a result, high energy efficiency can be expected. Similarly, the absorption properties of GHz and electromagnetic wave absorbers used in mobile devices are significantly improved with the use of magnetic nanoparticles completely dispersed in plastic matrices. Consequently, whether using the liquid phase or vapor method for synthesis, the most important tasks at present are to improve the crystallinity of nanoparticles, create additional functions through composition control, prevent aggregation through surface property control, and prepare complete dispersoids. In situ dispersion is expected to become key, regardless of the synthesis method used. In thermal plasma processing, nanoparticles are formed through coagulation from the vapor phase. The thermal plasma method, although similar to the spray-pyrolysis and flame-spray methods, can produce highly crystalline nanoparticles with little surface residue since the synthesis temperature is far higher. Because high crystallinity is related to functionalization, and the surface control is indispensable to dispersion, thermal plasma processing is expected to make a significant contribution to nanoparticle applications. In fact, plasma-synthesized nanoparticles are spherical and hardly agglomerated, and through an appropriate dispersion procedure, particles are individually dispersed in water. We expect the possible synthesis of functional nanoparticles, whose prominent functions can be given only by thermal plasma processing.