The Production of Some Ultrafine Particles by Gas Phase Reactions in Aerosol Reactors

This paper summarizes the authors' recent studies on properties of Ti0 2 and nickel powders produced by gas phase reactions in aerosol reactors. The weight-average diameter and polymorphism of Ti0 2 particles at temperatures lower than 1200° C were virtually unchanged 5-cm downstream from the nozzle mouth, and the effects of the residence time of gas and the reaction temperature were not dominant. However, the mean diameter of nickel powders produced by the hydrogenation of NiCl2 at 1380° C was close to the theory assuming instantaneous fusion of coagulated particles. The fraction of rutile was maximum at ca. 1000° C. These results imply the importance of the fusion rate of particles and the heterogeneous deposition rate from the gas phase.


Introduction
Gas phase reactions have been used to produce ultrafine particles with novel functions. Although overall stoichiometric reactions are normally simple, complicated kinetic steps such as nucleation, deposition and fusion are involved. The synthesis of ultrafine particles have been investigated by many authors4-6, 10-12, [15][16][17][20][21][22][23][24], but there are many aspects which require further research.
Titania usually exists in two polymorphic forms; anatase and rutile. The anatase phase is preferentially obtained at lower temperatures, while the rutile phase is the most stable at any temperature and pressure. Matsumoto   The present paper summarizes studies 1 1 • 1 2 ) on the production of Ti0 2 and nickel particles by gas phase reactions. In particular, the effects of various operating conditions on the particle size and the polymorphic composition of Ti0 2 produced by the oxidation of TiC1 4 without additives have been investigated 11 ). The growth mechanism of nickel particles produced by the hydrogenation of NiC1 2 12 ) is also cited in relationship to Ti0 2 particles.
2. Experiment apparatus and procedure 2. 1 Ti0 2 production The first series of the experiment~ 1 ) was carried out by using a reactor consisting of two mullite tubes with an inside diameter of 30 mm and length of 600 mm connected in series. The temperature inside the tubes was measured with 0.1 mm diameter platinum/platinum rhodium thermocouples. The reactor was heated by means of a resistance heater. The heated length of each furnace was 390 mm. Nitrogen and oxygen were dried over silica gel and in an ethanol-dry ice trap. TiC1 4 was of reagent grade quality. The TiC1 4 vapor was fed into the center of the reactor through an alumina nozzle of inside diameter 4 mm and outside diameter 6 mm, and it reacted with the oxygen  which entered through the annular inlet to the reactor. The reaction was carried out at temperatures above 850°C where the TiC1 4 conversion was 100%. The particles produced were trapped on a filter paper in a dust collector. The concentration of TiC1 4 was determined by absorbing Cl 2 formed in the reaction in an aqueous potassium iodide solution.
The following investigation was performed: (1) the effects of the reaction temperature and the TiC1 4 and 0 2 concentrations on the Ti02 particle size and the weight fraction of rutile; (2) the transformation from anatase to rutile when the powder produced in the first furnace was aged in the second furnace (mode A); (3) the growth and transformation of the crystallites when not only the particles produced in the first furnace but also TiC1 4 was fed into the second furnace (mode B).
The experiments of mode A and B are illustrated in Fig. 1 11 ). The flow rates of N2 and 0 2 in the two modes were constant at 3.3 cm 3 / s and 10 cm 3 /s, respectively, based on 25° C. The particle size was measured by X-ray diffraction (Cu-Ko:) and transmission electronmicroscopy. The weight fraction of the anatase and rutile particles was calculated from the ratio of the X-ray diffraction intensities by Spurr and Myers' 19 ) method.
The second series experiment was done by using a reactor with mullite tubes having an inside diameter of 52 mm arranged vertically or horizontally. The TiC1 4 vapor diluted with nitrogen was fed into the center of the reactor via an alumina nozzle of inside diameter 6 mm Qn(TiCI 4 ) = 0.24 cm 3 s-1 : t = 0.12 s and outside diameter 10 mm. In some runs, particles were sampled by evacuation through a 1 mm i.d. platinum tube from different parts of the reactor, and they directly impinged on the electron microscope grid mounted in a small holder. Figure 2 shows the direct sampling system. The lateral distribution of the reaction zone was visualized by inserting alumina rods of 1 mm in diameter across the flame.

2 Production of nickel particles
Nickel powders were produced in a vertical reactor of inside diameter of 3 5 mm 12 ). The nozzle was a mullite tube with an inside diameter of 9 mm connected to a mullite tube of 24 mm i.d. The nickel vapor was generated by heating pelletized anhydrous nickel chloride in the evaporating chamber of the nozzle side. The nickel chloride vapor was blown into the reactor through the nozzle using argon as a carrier gas, and it reacted with the hydrogen introduced through the annular space enclosing the nozzle. The rate of vaporization of the nickel chloride was determined by measuring the amount of HCl which was absorbed in aqueous NaOH at the outlet of the reactor.

Results and discussion
3. 1 Reaction zone Figure 2 illustrates a map of the reaction zone in the 52 mm i.d. reactor. Since the concentration of TiC1 4 was low, the reaction was controlled by the diffusion of oxygen toward the core of the jet. The length of the unreacted core decreased with increasing reaction temperature. The lateral expansion of the jet was not remarkable because the diffusion of particles was very slow compared to that of gas.

2 Particle size
TiO 2 particles Figure 3 shows a transmission electron micrograph of the Ti0 2 particles collected at the outlet of the furnace. Most particles were round or cubic. The mean diameter of crystallites, calculated from the X-ray diffraction with Si as an internal standard agreed closely with the value obtained by electron-microscopy. This indicates that the particles were single 1_rystals with few defects. Figure 4 11 ) reveals the effect of the reaction temperature on the size of the Ti0 2 particles. The concentration of TiC1 4 is the value in the gas leaving the nozzle. The size of the particles produced in the first furnace ( o) decreased somewhat with increasing temperature. The results obtained with the 52 mm i.d. reactor show the same tendency. The mean size of Ti0 2 particles sampled 5 em downstream from the nozzle mouth was nearly equal to that at the reactor outlet. The symbols • in Fig. 4 represent the particle size in mode A, in which particles were aged at 1000°C in the second furnace, while the symbols • denote the particle size in mode B, in which the feed gas was also introduced into the second furnace. There was no change in the particle size when the particles issuing from the first furnace were simply aged at 1000°C in the second furnace. Assuming that all of the TiC1 4 reacts at the nozzle mouth (t = 0), and that fresh crystals are produced by instantaneous fusion as a result of interparticle collisions due to the Brownian motion, the mean particle size is given by 7 > dp = 1.88 (6kT/Ps)Vsl/;2fst 2 1 5 (1) where the effective residence time in the furnace, t, is calculated at the reaction temperature. The experimental data of dP is 5 times smaller than Eq. (1) at 1000° C and 3 ~ 4 times smaller at 1100°C, respectively. The above results show that the growth of Ti0 2 particles The calculation indicates that the fusion of Ti0 2 crystals larger than serveral nm in diameter is not significant in the range of T < 1200°C under the present experimental conditions. When a second reaction was carried out at 1000°C, however, the particle diameter increased by about 1.5 times (equivalent to a 3.3fold increase in the volume) as shown in (1) is due to the high temperature of their reaction. Particle growth by simple aging was also observed when the second furnace was kept at 1200° C. The growth mechanism of Ti0 2 particles is summarized in Fig. 6 11  to fuse into larger particles at temperatures below II 00°C. However, when some unreacted titanium chloride was present, filling the space between the attached particles, fusion occurred at temperature of I 000° C. This is also valid for particles larger than a few tens of nm. Any smaller particles would probably have coalesced at lower temperatures. Figure 7 12 ) shows the effect of temperature on the mean size of nickel particles. The mean particle size is virtually equal to the theoretical value of Eq. (1) at 1380°C which is close to the melting point of nickel (1450°C). of this maximum in the rutile content was observed irrespective of reactor dimensions and whether or not an aging or a second reaction took place in the second furnace. The rutile fraction in the particles sampled at 5 em downstream from the nozzle coincides with that at the exit of the reactor. It was reported that the rutile content in a flame reactor operated at 1450 ~ 1650°C was very low 4 >.

3 Fractional rutile content
Shannon and Park 18 ) and others 3 , 8 • 9 ) have indicated that the rutile phase is nucleated on the surface and then spreads in the anatase body. The anatase-rutile transformation rate is given by, where a: is the fraction of transformation completed. Figure 9 shows the coefficient of the transformation on the basis of Eq. (3). Even at 1200° C, the transformation rate is very slow, and the spreading rate of the rutile phase is in the order of 0. stage of the reaction is the preferential formation of clusters of anatase. The clusters of anatase either grow into anatase particles or else are transformed into clusters of rutile. As soon as the anatase clusters grow into anatase particles, the transformation virtually stops. With increasing reaction temperature, the rate of transformation of the anatase particles into rutile decreases because the rate of fusion increases.
(II) Heterogeneous deposition model As shown in Fig. 8, the rate of formation of rutile particles in the second reaction is strongly affected by the polymorphic phase composition of the particles entering the second furnace. This implies the heterogeneous deposition of rutile from gaseous TiC1 4 on the particle surface which is already transformed to rutile is important. At higher temperatures, TiC1 4 reacts with oxygen more quickly to form Ti0 2 particles. Thus, TiC1 4 disappears at an earlier stage of the reaction. After the consumption of TiC1 4 , an increase in the rutile fraction may not be expected.
To specify the mechanism for the formation of rutile particles, it is important to evaluate

Conclusion
Ultrafine Ti0 2 particles were produced by gas phase oxidation of TiC1 4 without additives in aerosol reactors operated in the laminar regime. The weight-average diameter of Ti0 2 particles increased with an increase in the TiCl 4 concentration. The reaction terminated within 5 em downstream from the nozzle mouth, and the effects of the residence time of gas and the reaction temperature were not dominant. The mean diameter of nickel powders produced by the hydrogenation of NiC1 2 at 1380°C was close to Eq. (1 ). The intra-aggregate fusion was strongly related to the heterogeneous deposition and the temperature.
The fraction of rutile took a maximum at ca. 1000°C. Models were proposed for the explanation, but more research is needed to elucidate the exact mechanism.