K L R r
= reaction equilibrium constant
irradiated reactor length, cm reactor radius, cm = radial distance in reactor, cm TA = fraction of light of wavelength A transmitted through filter solutions t = time, sec u = axial velocity, cm/sec ii = average velocity across radius of reactor V = volume of batch recycle system, ema VR = volumeof reactor z = conversion z = axial distance, cm TOC = concentration of total organic carbon in solution, g mol/cm3 or mg/l.
GREEKLETTERS a,, = absorptivity a t wavelength A, cm2/g mol
parameter in Equation 25, cma/g mol average rate for whole cross-section of reactor tube ~ 1 ,= reaction rate at radial position r , g mol/(cm3) (see) A = wavelength, cm +cl = quantum yield for disappearance of free available chlorine, g mol/Einstein E = function of the average rate fi as defined b y Equation 29, g mol/(cm3) (sec) wA = attenuation coefficient at wavelength A, cm-‘ K
SUPERSCRIPT
= =
= $2 =
[PA
= “A
c1
0
=
initiai condition before chlorine has been added
literature Cited
Bulla, C. D., Edgerly, E., Jr., J . Water Pollution Control Federation, 40, 546 (1968). Cassano, A. E., Smith, J. hl.,A.I.Ch.E. Journal, 12, 1124 (1966). Cotton, F. A., Wilkinson, G., “Advanced Inorganic Chemistry,” First Edition, p. 446, Interscience Publishers, New York (1962). Engelhard Hanovia, Inc., Hanovia Lamp Division, Newark, New Jersey, Bulletin EH-223 (1968). Handbook of Chemistry and Physics, 47th Edition, The Chemical Rubber Co. Publishers, Cleveland, Ohio (1966). Henderson, G. L., Crosby, D. G., Bull Environmental Contamination Tozicology, 3, 3 (1968). Henderson, G. L., Crosby, D. G., J . Agric. Food Chem., 15, 5, 888 (1 967 ).
bIatsnura, T., Smith, J. AI., A.I.Ch.E. Journal, 16, 1064 (1970a). Matsuura, T., Smith, J. M., Ind. Eng. Chem., 9, 252 (1970b). Meiners, A. F., hlorris, F. V., J . Organic Chem., 29, 449 (1964). bIeiners, A. F., Lawler, E. A., Whitehead, M.E., Morrison, Y. Y., “An Investigation of Light-Catalyzed Chlorine Oxidation for Treatment of Wastewater,” Rept. TWRC-3, C . S. Dept. of Interior, FWPCA, Dec. 1968. Ralston, Anthony, Wilf, H. S., “Mathematical blethods for Digital Computers,” Wiley, New York (1960). Schorr. V.. Boval. B.. Hancil. V.. Smith, J. AI.. Ind. Ena. Chem. Process Des. Develop., 10, 509 11971). Standard Methods for the Examination of Water and Wastewater Including Bottom Sediments and Sludges, Twelfth Edition, American Public Health Association Inc., New York (1967).
SUBSCRIPTS
RECEIVED for review January 8, 1971 ACCEPTEDApril 20, 1971
A, B, C
contributions t o rate of pollutant removal rate for unsensitized reaction free chlorine Dollutant iota! initial or entrance conditions, after chlorine has been added 0 = oxygen r = at radial distance r tu = reactor wall
= nc = C1 = P = tit = 0 =
This work was supported by the Federal Water Quality Administration, Grant li020 EVQ. Tables V and V I will appear following these pages in the microfilm edition of this volume of the Journal. Single copies may be obtained from the Reprint Department, American Chemical Society, 1155 Sixteenth St., N.W., Washington, D. C. 20036, by referring to author, title of article, volume, and page number. Remit $3.00 for photocopy or $2.00 for microfiche.
Kinetics of Selective Chlorination of Ilmenite Using Hydrogen Chloride in a Fluidized Bed Arun S.
AthaValel Asian Paints (India) Private Limited, Bhandup, Bombay-7S, India
Vishwanath A. Altekar National Metallurgical Laboratory, N M L , Jamshedpur-7, India
Ilmenite and rutile ores constitute the main raw materials for the production of titanium and its compounds. Since rutile resources are very limited, efforts are made to replace it by upgraded ilmenite. Although ilmenite can be considered as FeO.TiOs, most of the naturally occurring deposits contain a major percentage of Fez03 and a higher percentage of Ti02 than the theoretical value of 52%. The higher ferric oxide con-
To whom correspondence should be addressed.
tent has been attributed to the presence of arizonite (FezOa3TiOs). Investigations b y Overholt e t al. (1950) tend to show that the naturally occurring ilmenite is a mixture of haematite, ilmenite, and rutile. I n total chlorination of ilmenite, both iron and titanium are chlorinated in the presence of a reducing agent. Attempts have been made to suppress the total chlorination by adding a controlled amount of reducing agents or by a proper choice of a chlorinating gas. The two-step process for Tic14 first involves the chemical beneficiation of ilmenite by preferential chlorinaInd. Eng. Chem. Process Des. Develop., Vol. 10, No. 4, 1971
523
Table 1. Analysis of Kerala Ilmenite Concentrate
Chemical Analysis Component
% '
by w t
Ti02 61.5 F e as FezO, 36 Si02 1.27 A1203,MgO, MnO, etc. 1.68 ao = initial analysis = 3.51 X lo-* g mol of iron oxide/g inerts Sieve Analysis IMM standard
-60 -72 -100 - 120
+ 60 + 72 + 100
+ 120
% by w t 3 21 53.2 22.7 3 16
tion of its iron oxides, yielding a titanium-rich residue which in a second step is chlorinated for the production of TiC14. T h e bulk of reported literature on this subject deals with chlorination in a conventional boat in tube setup or a static bed using a briquetted charge. These processes have the usual difficulties of channeling, sintering, hanging, stratification, and poor temperature distribution. Further, the binding agent may consume a certain amount of chlorine, and the briquettes have a tendency to disintegrate in the reactor with the loss of binder. These difficulties can be largely overcome by application of a fluidized bed. Considerable literature on the fluidized bed chlorination of ilmenite refers to chlorine as a fluidizing gas with or without C, CO, Cc14, and TiC14, etc. The use of chlorine alone requires a very high temperature of reaction. Use of CC14 and TiC14 is costly, without any marked advantage. Application of a selective gaseous chlorinating agent such as hydrogen chloride gas may be promising because of the following reasons: Selective chlorination with HC1 results in a n uncontaminated product It can be carried out a t a considerably lower temperature (900' C) than required with chlorine alone (1200' C) If HC1 is synthesized in the reactor itself, engineering difficulties of attaining high temperatures can be solved Gaseous raw materials, namely H2 and Clz, are cheap byproducts of the caustic chlorine industry Process economics is likely to favor downgrading of HZand Clz to HCl Literature available on the use of HC1 as chlorinating agent is either qualitative or in the form of patents. Work was, therefore, undertaken for the process development and kinetic studies in fluidized bed reactors. Thermodynamic data have been reported by Kellogg (1950) for the reactions of individual constituents of ilmenite with the chlorinating agent. By comparing the free energy values, selective chlorination of iron oxide in ilmenite could be expected. Hence, components other than iron oxides, viz., TiO2, SiO,, -k12O3,and MgO are referred to as inerts. The principal selective reaction of oxidized ilmenite around 800' C is: Ti02
a purer grade. The chemical composition and the sieve analysis are presented in Table I. Ilmenite sand is suitable for fluidization because of its free flowing nature. The classical fluidization characteristics have been studied by Krishnamurthi (1964). The working range of gas velocities used here is above the incipient fluidization and within the limiting transportation velocities. Chlorination was carried out in a batch and in continuous fluidized bed reactors as described below. Batch Process. T h e experimental setup is shown in Figure 1. Hydrogen and chlorine from the cylinder were led to the burner after recording the flows by using capillary flow meters. Nitrogen gas, purified in a pyrogallic acid scrubber, was used for initial fluidization, final purging, and for dilution of HC1 in some experiments. The burner is simple in design, consisting of a n outer '/Z-in. stainless steel tubing and inner I/h-in. copper tubing. It is capped by 200-mesh monel wire gauze and wrapped and covered with afbestos packing. This burner was replaced later, b y the one fabricated completely in refractory materials and shown in Figure l a (See Figure 1). The Hz and Clz gases from separate silica tubes were led to the graphite mixing chamber wherein they could be ignited to produce HCl. The reactor, a vitreosil silica tube 2.1 cm i.d. or a graphite tube 2.45 cm i.d. and about 30-35 cm long, was heated by an electric furnace having a capacity of about 800 W. Uniform distribution of the gases was achieved with the help of two alumina distributors snugly fitting in the reactor. Requisite support to the bed was made available by putting inert silica granules, 15-20 mesh size, between the distributors. T h e upper portion of the reactor was connected to the glass column, which increased in diameter by a factor of two. Jointing was made with a suitable asbestos packing. Temperature was recorded by chromel-alumel thermocouples covered with a silica sheath. Effluent gases were fent through the condenser and a buffer flask, to the set of scrubbers. At the end, a water jet ejector was provided to ensure smooth flow of gases and to absorb unreacted HCl. Procedure. Temperature of t h e reactor was first raised in the vicinity of the predetermined reaction temperature. The water jet ejector was then started. Nitrogen flow was also started, its rate being adjusted a t about the minimum fluidization requirement. ilfter about a 10-min interval, a weighed quantity of ilmenite was dropped in the reactor from the top.
, G L A S S PIPE
S ILlCA TUBE
FURNACE
PRESSURE RECORER
FLOWMETER
+ Fez03+ 6HC1 = 2FeC1, + 3H20 -t Ti02
Experimental
Ilmenite was obtained from Kerala as concentrates, but was refined on a cross belt type magnetic separator to obtain 524
Ind. Eng. Chern. Process Des. Develop., Vol. 10, No. 4, 1971
Figure 1 . Batch fluidized bed reactor
Beneficiation of ilmenite i s industrially important because of limited resources of naturally occurring rutile. Gaseous hydrogen chloride can be used to upgrade ilmenite as successfully as chlorine. Experiments were carried out in a batch and continuous fluidized bed reactors to study the effect of process variables and reaction mechanism. Experimental results could be correlated by using a penetration model of the unreacted core. Overall kinetic and geometric constants when related to temperatures 600-850" C yielded activation energy of 9.5 kcal per mole. The data from batch and single-stage continuous experiments were used to determine the optimum number of stages and the compositions of the streams leaving the stages. Velocity voidage data, determined in a separate column, have been used to ascertain the expanded volume of ilmenite chlorinated to a known extent. In a commercial reactor from the engineering point of view, hydrogen chloride gas may be generated in situ.
Fluidization with nitrogen was cont~iiiuedfor about 20 t o 30 min t,o achieve constant isothermal conditions in the bed. Next H2 and Cls were almost simultaneously started, adjusting to their predetermined ratio. Unless otherwise mentioned, 98y0 pure HC1 was produced b y supplying H2 in 4% escess. Iron chlorides were collected in the condenser and scrubbers. At the end of the predetermined chlorination time, the heating as well as Hz and CIS were switched off. Xitrogen was again used, prolonging the fluidization by about 30 min. During this period, the temperature dropped down to about 150200' C, and the residual charge was found to have free flowing characteristics a t the time of its removal. All the parts were dismaiitled. Residue was carefully collected, weighed, and analyzed. All parts of the assembly were washed, aiid washings and the scrubber contents were collected, made to standard of 3- or 5-liter volume and analyzed for F e content. Material balances showed an experimental accuracy of i2%. T h e Jones reductor was used to reduce the Ti4+t o T i 2 +aiid Fe3+ to Fez+. Oxidation titration was done with ferric ammonium sulfate solution for T i and KMiiO4 solution for (Ti Fe) using the internal indicator. Preliminary runs were made to: Obtain steady control over gas rates and proportions. By keeping Hz/Cl2 = 1.04, a constant purity of 98% HC1 could be maintained throughout the runs. This was found b y the analysis of the stream gases after a 30-min run Obt,ain an accurate iron balance b y complete absorption of chlorides in the bubblers, and to compare t'he same obtained b y residue analysis Ensure t,hat the materials of const'ruction undergo a negligible amount of react'ioii Select the working velocity a t which film resistance could be assumed to be low Continuous Process. Figure 2 shows a single-stage, coiltinuous, fluidized-bed graphite reactor of 4.48 c m i.d. I t was heavily lagged b y (10 em thick) asbestos magnesia powder and foam glass bricks, Arrangement was made to affis the hydrogen chloride gas burner a t the bottom and also to remove the solids overflow through a 1.25-cm silica downcomer pipe. The overflow lip could be adjusted to a n y desired level. Temperature was recorded at, two places b y introducing a silica sheath serving as a well for t,he chromel-alumel thermocouple. T h e upper half of the reactor was made of borosilicate glass pipe having a side arm with a ?crew feeding arrangement. The feeding rates of ilmenite were adjusted by using the gear bos and changing the pulley:'. -1water jet ejector was used a t the end of the gas effluent line t,o overcome the resiFtance to the smooth flow of the gases through the pipelines and scrubber. Hydrogen and chlorine cylinders were coilnected through flow meters and pressure manometers to the burner.
+
Procedure. Hydrogen and chlorine were lighted outside in air, and t'hen t'he burner was fixed in its seat a t the bottom of the reactor. The flows of the hydrogen and chlorine gase3 were adjusted to predetermined values in order to achieve the required temperature of the reactor and t o provide 98% pure HC1 b y using about 4% escess of hydrogen over chlorine. The doivncomer silica pipe was adjusted to a desired level to maintain a known bed height in the reactor. The feeding rate was then adjusted to the predetermined level, so that the solid and the gas streams reached equilibrium working Conditions in about 40 min. This was indicated by the constant composition of the solid product stream. Ffom here 011: the process was continued for about 3 hr. Xt the end of the esperiment, the holdup of the column was measured. The material balance on the basis of iron o d e removal was checked as before. Kinetic Analysis
In recent years, considerable efforts have been made to establish a model of a reaction between a gas stream and granular solid particles or a bed. Thorough understanding of a reaction model of a single particle and its subsequent applica-
2 SCREW F E E D E R
4
',,
;/
I
G A S E S IO SCRUBBER
~
PRODUCT FOR WEIGHING
Figure 2. Continuous fluidized bed reactor Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 4, 1971
525
tion to the granular bed appear to be essential for optimizing operating variables, as suggested b y Osman et al. (1966). According to Hougen and Watson’s model (1943) of gas reacting with solid, the resistance occurs in series when HCl from bulk gas phase gets transported through gas film, diffused through the inert ash layer (Ti02, Si02, etc.) to react with t’heiron oxide a t the surface, producing gaseous products which would diffuse out of the solid and subsequently get transported back to the bulk gas phase across the gas film. Levenspiel (1962) has dealt with the expressions considering first-order kinetics when a single resistance controls the rate of reaction. Lu (1963) has developed a n expression considering first-order irreversible kinetics wherein interface resistance and ash or shell layer resistance are likely to be important. According to Lu (1963), resistance due to the shell is insignificant; but this resistance may become a controlling factor as the reaction moves deep into the solid. These changes are gradual and continuous, and difficult to predict. The above mechanism appears to be applicable to ilmenite reactions. Lakshmanan et al. (1965) in their kinetic study of ilmenite chlorination with CO and Clz mixture in a fluidized bed seem to conclude that the surface reaction is the slowest rate controlling step. Dunn (1 960) studied chlorination of TiOz hearing materials (iiicluding ilmenite) by CO and Clz and showed that the loss of weight of the bed, either fluidized or packed vs. time of reaction, gives a straight line plot. Similar mechanism is likely to hold when HC1 is used as the chlorinating agent,. Aiialysis of the factors concerning selective gaseous reactions in minerals has been reported elsewhere by Athavale and Altekar (1969). Only pertinent equations are considered here. Keeping in view the stoichiometry of the main selective reaction, weight loss rate may be obtained as follo\vs :
-
