Contributions of James E. Mauser, chemist, Albany Metallurgy Research Center, Albany, Oreg., in sampling and making qualitative analyses of byproducts are appreciated. Also, contributions of Darrell G. Burt, chemical engineering technician, Albany Metallurgy Research Center, Albany, Oreg., in perfecting new devices and in the construction and operation of the chlorinator are acknowledged. EQUIPMENT, PROCEDURE, AND ORES Equipment A schematic drawing of the unit used for chlorination of ilmenite is given in figure 4. It does not include the reactors used to dechlorinate titanium tetrachloride or iron chlorides that are indicated on the flowsheet in figure 3. Commercial-grade chlorine, oxygen, and nitrogen were used in most chlorination tests. In the chlorination unit, measured amounts of chlorine and powdered ilmenite were reacted in a heated fluidized coke bed, and the products were separated by the removal of dust in a cyclone and fractional condensation of iron chloride powder and TiC14 liquid. The chlorination reactor was a 10.2-cm-ID vitreous silica tube 107 cm long. A flared ground flange at the top and a reduced inlet with a tapered ground joint were used to connect the reactor tube to the feeder and condensing sections of the system. A removable 1.7-cm-OD Vycor5 feed loop and distributor that extended 2 cm into the bed was fitted into the bottom of the reactor and was connected to the ilmenite feed screw and chlorine inlet. This permitted chlorine-entrained feeding of ilmenite powders. An external electric furnace heated the reactor tube to temperatures up to 1,150° C. An Acrison hollow-helix screw feeder (model 105-XA) was used to feed ilmemite or ilmenite-coke mixtures into the feed loop inlet. Weights of material fed were measured using a scale equipped with a jack mounted beneath the feeder. The reactor was sealed at the top to a water-cooled, refractory-lined, Inconel housing with a Buna-N flat gasket. The housing was lined with A. P. Green castable Greencast number 12, with an inner coating of cured Fiberfrax putty, also used for metal-to-graphite junctions. Three flexible Inconel connectors, somewhat protected by inner Vycor tubes, were used in the system. All tube junctions had four-way crosses. The extra inlets were used for insertion of thermocouples and immersion heaters, which were used to prevent clogging. The 10.2-cm-ID Inconel cyclone used to collect ilmenite and coke dusts was heated with an outer, insulated Nichrome element to temperatures of 500° to 700° C. This allowed condensation of high-boiling-point chlorides and passage of FeCl3 and TiCl4, which have lower boiling points. Iron chlorides were then condensed in a 45-cm-ID by 92-cm-deep metal condenser that was 5Reference to specific materials or equipment is made for identification only and does not imply endorsement by the Bureau of Mines. heated by an external, insulated Nichrome element to temperatures of 100° to 250° C. Offgases and TiCl4 vapor were cleaned in a bed of heated salt pellets at 200° to 400° C, and the liquid NaCl-FeCl3 eutectic that formed was collected. Some of this minor byproduct may furnish the NaCl catalyst used in the dechlorination of iron chloride powders. Most of the salt-column filter and TiCl4-condensing system was constructed from pyrex double-tough glass tubing and fittings. In the primary TiCl4 condenser (15.2-cm-ID), vapors were cooled by an internal hollow plate with fins and with an externally wound coil, both cooled with tapwater. Secondary condensers of the tube-and-shell type were cooled to minus 11° C by circulated brine from an ice-salt mixture. Liquid Tic14 was collected in graduated tubes and then discharged into preweighed storage containers. Vapor and gas mixtures leaving the final TiCl4 condenser were scrubbed with tapwater in a countercurrent packed column to collect soluble chlorides. to evaluate condensation. The resulting solution was neutralized with sodium carbonate before disposal. Final scrubbing of gases was done with sodium carbonate solution sprayed into the top of a countercurrent packed column. The clean CO-CO2 offgas could be recycled and burned to heat systems in pilot or industrial plants. Elements heating the reactor, cyclone, and iron chloride condensers were controlled by thermocouples and Wheelco controllers, with a second thermocouple and controller to guard against override. Other heaters were manually controlled using powerstats with ammeters. A six-point recorder was used to continuously record temperatures in the fluidized bed, crossarms, cyclone, iron chloride condenser, and salt column. During chlorination tests, a continuous sample of offgas was aspirated through a calibrated automatic chlorine analyzer. A photocell in this analyzer sensed the preferential adsorption of utltraviolet light by chlorine in the offgas. The output was amplified and recorded on a Leeds and Northrup strip chart recorder. This allowed ease in sensing suitable ilmenite-chlorine ratios and facilitated maintenance of positive but low chlorine concentration (>5 vol-pct) in the system. Fairly high chlorination efficiency was attained, but it also insured that most of the iron would form FeCl, rather than the high-boiling-point FeCl2. Samples of the offgases were also analyzed periodically by Orsat methods for Cl2, CO, CO2, O2, and COCl2. Procedure Most chlorination tests were performed at close-to-steady-state conditions to obtain material balance data and to permit valid comparison of the effect of operating variables. Most of the tests were conducted for 3- to 5-hr periods, and operating conditions of temperature (±10° C), chlorine flow (±1 1/min), and ilmenite feed rate (±2 g/min) were maintained as constant as possible. Chlorine flows and ilmenite feed rates tested were from 15 to 25 1/min and 25 to 54.8 g/min, respectively. Ilmenite and coke used in all tests were predried at 110° C. Generally the chlorinator system was preheated; then 4,000 g of coke was loaded into the reactor and heated to the selected operating temperature using a fluidizing flow of nitrogen. Upon initiating the chlorine flow, the nitrogen flow was turned off and the ilmenite feed was started. Initial weights of the chlorine tank and ilmenite feeder were recorded. In the majority of tests, the ilmenite fed contained no coke to simplify material balances. Preliminary bed-height tests, and other tests using 15 to 20 pct coke in the ilmenite fed, showed no appreciable differences in processing results with bed depths from 30 to 76 cm, the range encountered in a 5-hr test without a coke feed. However, in some tests, mixtures of coke and ilmenite were used to compensate for the coke used up to maintain a constant bed height. Temperatures were adjusted to those desired for the reactor, cyclone, FeCl3 condenser, and salt-column filter. Flow of water and brine coolants was started during the preheating stage. Chlorine flow and ilmenite feed rates were set using calibration charts. Final adjustments were made depending on the chlorine content in the offgas as indicated by the continuous analyzer and on rates calculated from periodic weights of chlorine and ilmenite supplied. Ilmenite feed rates were generally slightly less than for stoichiometric reaction to produce TiCl4 and FeCl3, thus insuring the presence of chlorine (2 to 5 pct) in the offgas. Records were kept of all operating conditions and of the volume of liquid TiCl4 collected versus time. When tests were terminated, all systems except cooling lines were shut off, and a small flow of nitrogen was started to carry residual chloride vapors to the condensers. When the system was cooled to room temperature, weights of collected TiCl4 and byproducts were recorded. Solu Samples of the TiCl4 and byproducts were analyzed, and the water-soluble, insoluble, and ignited insoluble materials in the byproducts were determined. Soluble materials were itemized as chlorides; ignited insoluble materials, as unreacted ilmenite; and loss on ignition of the insoluble portion, as coke. Spectrographic and chemical methods were used to determine chemical content of the as-collected samples and soluble and insoluble portions of samples. ble chlorides were generally dissolved in distilled water, and then precipitated with NH4OH; the precipitate was ignited at 800° C. This simplified qualitative spectrographic analysis and gave impurity results in terms of the end-product oxides (TiO2 and Fe203) rather than chlorides. Periodic analyses showed that solutions remaining after filtering hydroxides had insignificant amounts of soluble chlorides. The ilmenite reacted in each test was calculated from the ilmenite fed less the weight of ignited insoluble material in the bed residue, cyclone, and iron chloride byproducts. Recovery of TiCl4 was calculated in percent of stoichiometric recovery based on the ilmenite reacted rather than on the ilmenite fed. This is reasonable because it gives recoveries more representative of continuous operation in which most of the unreacted ilmenite collected in cyclones would be recycled, thereby increasing Tic14 recovery. Also, this procedure facilitated material balances because the ore left in the bed and cyclone was eliminated. Chemical Analyses and Particle Size of Ores Table 3 shows the chemical analyses of the domestic ilmenites and one slag that were tested. Significant features are the lower TiO2 and higher Fe0 content of New York ilmenite compared with the other ores. Appreciable amounts of Al, Mg, Mn, and Si impurities are found in all of the ilmenites. High-boiling-point chlorides form from Mn, Mg, Ca, and Cr impurities. Vanadium and silicon form low-boiling-point chlorides that may condense with the Ticl,. Part of the loss on ignition (LOI) shown is accounted for by the water, C, and S contents indicated. Some may result from flotation reagents used to prepare these ore concentrates from the original ores. Coarser Ilmenites used for the majority of tests were ground to 90 pct finer than 325 mesh for rapid reaction, thus allowing close control of the ilmenitechlorine reaction ratio and, therefore, of the FeCl3 -FeCl2 ratio. ores might accumulate in the bed allowing preferential iron removal and would produce too much FeCl2, which would adversely affect subsequent processing. Wet screen analyses are given in table 4 for the ores tested. As shown, coarser sizes of New York ilmenite, Tennessee ilmenite, and slag were tested also. A calcined grade of petroleum coke, 98.6 pct of which was minus 20plus 100-mesh screen size, served as the fluidized-bed material and as the reductant. |