PROSPECT
- A Continuous Process for Titanium
Diffusion Flames for Producing Titan um Titanium is still an expensive metal, partly because it is made in a batch process. This article describes experiments which strongly indicate that titanium can be made in a continuous process by establishing a diffusion flame of liquid sodium in an atmosphere of titanium tetrachloride vapor
PRELIMINARY
calculations of burning rates for heterogeneous diffusion flames have been carried out using theoretical equations previously reported ( 3 ) . Because of large uncertainties in transport properties for sodium vapor and titanium tetrachloride, it seemed desirable to perform some crude experiments which would establish orders of magnitude for the most important parameters. Theoretical estimates of transport properties for the multicomponent gas mixture in the region between the sodium droplet and the flame front are limited to order-of-magnitude accuracy. Experimental
The apparatus used in preliminary experiments (Figure 1) consisted essentially of a spherical glass vessel in the center of which sodium (solid or liquid) could be supported and exposed to gaseous titanium tetrachloride diffusing into the reaction chamber. The glass flask had a diameter of about 23 cm. and was provided with flat extension windows through which the flame surface and burning (evaporating) sodium could be photographed by means of a 35-mm. Arriflex movie camera. Preliminary experiments with solid sodium supported on quartz fibers showed that ignition was not spontaneous. For this reason in subsequent experiments two wire filaments were used both as supports and as heating coils to melt the sodium and initiate burning. The entire apparatus was flushed with helium and evacuated prior to admitting titanium tetrachloride vapor. Vapor entered the reaction flask, and a stable luminous flame was established. As the burning rate of heterogeneous diffusion flames is notably insensitive to pressure, no attempt was made to introduce corrections for the variable pressure within the closed system.
Results From the photographs obtained during burning it was a simple matter Present address, Northwestern University, Evanston, Ill.
to estimate the effective size of the flame shape. Unfortunately, liquid sodium was clearly visible only at the beginning of the experiments because the intensely luminous flame front hid it from view during the remainder of the burning period. For this reason it was not possible to follow the rate of decrease of the sodium pellet with time directly and thereby determine the evaporation constant K. However, K is easily obtained. Thus, the present work did not verify the conventional burning rate law for heterogeneous diffusion flames. Actual shape of the liquid sodium corresponded roughly to a cylindrical section 5 mm. high, 2 mm. in radius, and subtending an angle of 105' (Figure 2). Total initial mass of the liquid sodium was 16.7 f 8.4 mg., and the initial radius of an equivalent sphere was 1.6 mm. Total observed burning time, tb, was 2.9 =!c 0.3 seconds. The equivalent sphere corresponding to the flame surface had a radius of 6.4 mm. and changed little during burning. Hence the experimentally observed ratio of r l / r o was less than 0.25, a result used subsequently to obtain a significant simplification of the burning rate expression. Principal deficienciesof the experiment are the following: The liquid droplet was not spherical. On the basis of previous work it appears unlikely that this will produce a significant error. The evaporation constant, K , although obtained from a valid expression, is defined only if the square of the drop diameter decreases linearly with time. This burning rate law is known to hold for all heterogeneous diffusion flames studied previously but has not been established for the sodium-titanium tetrachloride system. The wires supporting the sodium acted as a heat sink during burning. Hence observed values of K must be lower than the value which would be obtained in a steady-state experiment in which similar conduction heat losses are nonexistent. Burning Rate Calculations for
A. E. FUHS' Daniel and Florence Guggenheim Jet Propulsion Center, California Institute of Technology, Pasadena, Calif. Heterogeneous Diffusion Flames. For the following approximate calculations, a simplified relation (7, 2) is used:
In Equation 1, 7 j t F denotes mass burning rate of the liquid fuel droplet, corresponding to liquid sodium in the present case. Temperature of the sodium is taken equal to the normal boiling point of sodium; this approximation has no important influence on the calculated values for mass burning rate. Flame temperature, T,, is assumed to be equal to adiabatic flame temperature. In an actual diffusion flame T,may well exceed adiabatic flame temperature ; however, if proper allowance is made for dissociation, it is unlikely that T , will be significantly larger than the adiabatic flame temperature. Theoretical calculation of r1/rerequires use of the complete diffusion flame theory ( 3 ) and is complicated by lack of knowledge concerning appropriate numerical values of the transport parameters. I t is preferable and simpler to utilize the experimental results which indicate
1 (2) From Equations 1 and 2 it follows that rl/rc
