WO1994011304A1

WO1994011304A1 - Process for removing metal impurities from calcined magnesite

Info

Publication number: WO1994011304A1
Application number: PCT/CA1992/000497
Authority: WO
Inventors: Wendell E. Dunn, Jr.
Original assignee: Deloro Magnesite Venture Fund
Priority date: 1992-11-12
Filing date: 1992-11-20
Publication date: 1994-05-26
Also published as: AU2937592A

Abstract

A process of removing metal impurities, such as iron, manganese, and nickel, from impure magnesium oxide, e.g., calcined magnesite, at high temperatures is provided. The process involves contacting the material with a reducing gas capable of partially reducing oxides of the metal impurities, and then with chlorine to produce volatilizable metal chlorides, while limiting the formation of MgCl2. The reduction and chlorination steps are cyclicly repeated until at least about 75 % of the metal impurities are converted to volatilizable metal chlorides.

Description

PROCESS FOR REMOVING METAL IMPURITIES

FROM CALCINED MAGNESITE

Cross Reference to Related Applications The present application is a Continuation-in-Part of U.S. application Serial No. 07/706,636, filed May 29, 1991.

Background of the Invention

Magnesite, or magnesium carbonate (MgCO₃), is an important industrial source of magnesium. It is often referred to as "natural magnesium carbonate" or "crude magnesite" and can be readily decomposed upon heating, i.e., calcination, to magnesium oxide. Magnesite is a major industrial source of various grades of magnesium oxide (MgO), which is often also referred to as "magnesia" or "calcined magnesite." The chemical activity of the various grades of magnesium oxide generally depend upon the temperature and duration of calcination.

Talc, silica, and other siliceous constituents are typically associated with magnesite in nature. These constituents can be present in amounts up to about 60% by weight. Iron is also present in magnesite, generally in the form of iron carbonates. It can be present in amounts up to about 10% by weight. Small amounts of other metals, such as nickel and manganese, generally in the form of carbonates, are also present. Significant amounts of talc, silica, and the siliceous constituents can generally be removed from the magnesite by a flotation process. The resulting magnesite concentrate typically still contains iron and other metal impurities, however.

In a typical transformation process of magnesite to magnesium oxide, i.e., magnesia, the magnesite that has been subjected to a flotation process, i.e., the flotation magnesite concentrate, is calcined. That is, it is fired in an oxidizing atmosphere to drive off carbon dioxide and form "calcined magnesite" (MgCO₃ -- > MgO + CO₂). The firing temperature used is generally below about 1450°C. The resulting magnesium oxide can contain a small amount of CO₂, e.g., about 2-10%. This grade of magnesite is often referred to as "caustic-calcined magnesite," "calcined magnesite," or "calcined magnesia." It typically displays absorptive capacity and moderate to high chemical

reactivity. If a firing temperature above about 1450°C is used, the resulting magnesium oxide contains less than 2% CO₂. This grade of calcined magnesite is a refractory grade magnesium oxide and is often referred to as "dead- burnt magnesia."

Upon calcination of the magnesite to calcined magnesite, i.e., magnesium oxide (MgO), the metal carbonate impurities, e.g., iron impurities, are converted to oxides. The iron oxides are generally in the form of ferrites, such as the magnesium ferrite MgO·Fe₂O₃. which

have proven to be very difficult to remove. The presence of the metal oxide impurities, particularly the iron, detrimentally affects the refractory properties of the calcined magnesite. For example, the iron oxide impurities lowers the melting point of the refractory. Thus, it is desirable to remove the iron oxide and other metal oxide impurities from the calcined magnesite.

One method by which the metal impurities might be removed from calcined magnesite (MgO) involves direct chlorination of the metal oxide impurities to form volatile metal chlorides. For example, treatment of calcined magnesite with chlorine gas at elevated temperatures requires reduction of the iron oxides along with

chlorination to form volatile FeCl₃. Similarly, the treatment of natural magnesite (MgCO₃) with chlorine gas in the presence of a reducing agent results in the formation of volatile FeCl₃.

Chlorination processes such as these, however, typically also result in the chlorination of MgO to MgCl₂, thereby disadvantageously decreasing the amount of pure MgO formed. Furthermore, the formation of even small amounts of liquid MgCl₂ can be detrimental to certain processing systems and cause interference with the reactor operation. For example, the presence of liquid MgCl₂ can destroy the fluidizability of the material. This is particularly true when the process is conducted within the temperature range in which the chlorination and volatilization of the metal impurities occurs at an economical and acceptable rate. Fluidized beds are particularly sensitive to liquids, even tiny amounts, which can cause stickiness and destroy the fluidizability of the bed material.

Direct chlorination is not generally an

economically efficient method of removing metal impurities, particularly iron impurities. For example, although thermochemical calculations indicate that at very high temperatures it is possible to directly chlorinate a fully valent iron oxide, these temperatures are not economically efficient, and not suitable for an iron oxide impurity in an MgO system.

One approach to removing iron impurities from natural magnesite involves heating magnesite and MgCl₂ in an oxygen atmosphere (Canadian Patent No. 1,093,280, H.G. Brandstatter). This process results in the formation of volatile iron chloride. Because of the use of MgCl₂, this process cannot be used in a fluidized bed reactor.

Another approach to removing iron impurities from natural iron-containing magnesium compounds, such as "crude magnesite," involves a two step process of reduction and chlorination of the crude magnesite, i.e., MgCO₃, itself (Austrian Patent No. 265100, G.L. Mortl). In this process. naturally occurring magnesium carbonate, of 4-40 mm

particle size, is fired in a reducing atmosphere. This is done to prevent the formation of ferrites, such as

magnesium ferrite (MgO-Fe₂O₃), which are very difficult to remove from calcined magnesite. This single reduction step of the MgCO₃ starting material results in the formation of magnesium oxide and reduction of the iron carbonate

impurities to elemental iron. This mixture of reduced iron and magnesium oxide is then treated with chlorine gas, with the exclusion of substances that have a reducing or

oxidizing effect, at a temperature above 800°C. The product formed from this step is contaminated with liquid magnesium chloride, as much as 2.23%. Thus, a process such as this prevents the use of a fluidized bed of fine MgO because of the formation and interference of significant amounts of liquid MgCl₂. Furthermore, the product formed from this two-step process is still contaminated with iron impurities, as much as 0.26%.

A method of purifying a titaniferous ore is known that uses carbon monoxide and chlorine (U.S. Patent No. 3,699,206, W.E. Dunn, Jr.). The titaniferous ore, which contains TiO₂, however, typically contains much higher amounts of iron than does magnesite. Thus, this method has not been considered practicable for materials containing very small amounts of metal impurities, such as iron, nickel, and manganese. Furthermore, the formation of liquid MgCl₂ has also been considered an undesirable side effect of the process.

It is therefore an object of the present invention to provide reducing conditions for metal impurities, such as iron oxide, residing in calcined magnesite flotation concentrate using a chlorination process, while not

allowing the chlorination step to chlorinate the MgO. It is another object of the present invention to provide chlorination conditions that are favorable to the

chlorination of iron oxide, and other metal oxide

impurities, but not to the chlorination of MgO to MgCl₂. Yet another object of this invention is to provide the above conditions while removing a substantial amount of the iron, nickel, and manganese impurities.

It is also an object of the present invention to remove a substantial amount of the iron and other metal impurities in calcined magnesite flotation concentrate within a temperature range that does not exceed the

calcination temperature of the incoming MgO. It is yet another object of the present invention to avoid the production of levels of liquid MgCl₂ that would

detrimentally affect fluidizability of the product in a low temperature calcination system, in which the MgO has a relatively large surface area.

Another object of the present invention is to achieve low levels of iron and other metal impurities in the final product in an economically efficient manner.

That is, an object of the present invention is to conduct the process in a reactor of reasonable size, and under operating conditions that are economically efficient and minimize the use of reactants. Summary of the Invention

The present invention is directed to a process for the removal of metal oxide impurities, such as iron, nickel, and manganese, from calcined magnesite without producing significant amounts of magnesium chloride.

Calcined magnesite, i.e., magnesium oxide (MgO), is a magnesite ore that has been subjected to a process, e.g., a flotation process, to remove silica, talc, and other siliceous components; and has been calcined under oxidizing conditions to remove volatile components and form calcined magnesite, i.e., magnesium oxide (MgO). Although the invention is directed to the use of calcined natural magnesite, the process is also envisioned to include the removal of metal impurities from magnesium oxide obtained from any other source and in any other manner. The method of the present invention involves the use of a reducing gas capable of reducing oxides and ferrites of the metal impurities contained in the calcined magnesite to a lower valent state, in an alternating and cyclic manner with the use of a reducing gas followed by chlorine gas.

It has been discovered that if a reduction step were cycled with a chlorination step, e.g., using CO as a reducing agent, and then chlorinating the metal oxide impurities with Cl₂, MgO is typically not placed in an environment or condition where the Cl₂ can preferentially attack it and make MgCl₂. The metal impurities, e. g., iron, which are in a reduced metal oxide form after the reduction step, react preferentially with the Cl₂ to form metal chlorides, which are either volatile, such as FeCl₃, or can be readily volatilized.

The process for removing metal impurities from contaminated, i.e., impure, magnesium oxide, preferably calcined magnesite, includes the steps of: (1) contacting the impure magnesium oxide with a reducing gas capable of partially reducing oxides of the metal impurities to form magnesium oxide and partially reduced metal oxides, e.g., calcined magnesite containing partially reduced metal oxides; and (2) in the absence of a reducing gas,

contacting the magnesium oxide and the partially reduced metal oxides with chlorine so that the formation of magnesium chloride is substantially prevented and

volatilizable metal chlorides are formed. Although the reducing gas could totally reduce the impurity oxides if used in a great amount to zerovalent metals, it is

advantageous that the oxides are only partially reduced in each cycle. The process preferably involves using a calcined magnesite flotation concentrate, i.e., a still impure magnesium oxide obtained from the flotation and calcination of magnesium carbonate. Calcined magnesite flotation concentrate typically contains about 65-80% material of less than 200 mesh, i.e., having a particle size of less than about 0.074 mm. With such a small particle size, the formation of more than trace amounts of liquid MgCl₂ can form a sticky mess that is very difficult to contend with in a fluidized bed reactor.

The process of the present invention involves cyclicly repeating the reducing and chlorinating steps until at least about 75% of the metal impurities are converted to volatilizable metal chlorides without the formation of MgCl₂ in more than trace amounts. The

volatile metal chlorides, e.g., iron chloride, and

volatilizable metal chlorides, e.g., nickel and manganese chlorides, can then be removed or separated from the magnesium oxide by vaporization, to form substantially pure magnesium oxide. Preferably, the reducing and chlorinating steps are repeated until at least about 90% of the metal impurities have been removed, and most preferably until at least about 95% have been removed.

The chlorination step of the process of the present invention is conducted at a temperature within a range of about 700-1100ºC. Preferably, both the reducing step and the chlorinating step are conducted at a

temperature within a range of about 700-1100°C. More preferably, both steps are conducted at a temperature within a range of about 750-1000°C. For certain

embodiments of the present invention, such as when highly active MgO is desired, both steps are advantageously and preferably conducted at a temperature within a range of about 750-850ºC, and more preferably at a temperature within a range of about 750-800°C. Detailed Description of the Invention

The present invention is directed toward further purification and improvement of contaminated, i.e., impure, magnesium oxide, preferably a calcined magnesite product, or a calcined magnesite concentrate produced from a

flotation process. Although the flotation and calcination processes remove silica, talc, other siliceous

contaminants, and numerous volatile materials, the calcined magnesite typically still contains iron impurities, as well as small amounts of nickel and manganese impurities. These metal impurities, particularly the iron impurities, can be effectively and efficiently removed by the method of the present invention.

The present invention involves chlorination at relatively high temperatures but in such a way that the chlorine does not react to a significant extent with the desired product of calcination, i.e., MgO, to produce

MgCl₂. Thus, using the process of the present invention the problem of degradation of the MgO is reduced, if not eliminated. Furthermore, the operation of a fluidized bed reactor, the preferred reactor system, is not hindered by the presence of any signficant amount of liquid MgCl₂.

Using the method of the present invention, it is possible to decrease the amount of metal impurities to a significantly low level in calcined magnesite. Preferably, at least about 75% of the metal impurities can be removed, more preferably at least about 90% can be removed, and most preferably at least about 95% can be removed. For example, the iron content of calcined magnesite flotation

concentrate can be reduced from about 5% by weight to less than about 0.5%, and preferably less than about 0.1%, by the method of the present invention. These advantageous results can be achieved by the alternate use of a reducing gas and chlorine gas, i.e., multicyclic contact. In order to react the metal impurities,

particularly the iron impurities, with chlorine directly, the metal should be in a reduced oxide state, such as

Fe₃O₄, rather than Fe₂O₃, but not in a completely reduced metallic iron state. This can be accomplished by first contacting the material with a reducing gas capable of partially reducing oxides of the metal impurities in calcined magnesite.

This reducing gas can be any that is capable of at least partially reducing the metal oxide impurities, i.e., the iron, nickel, and manganese impurities, within the calcined magnesite. This includes, but is not limited to, carbon monoxide (CO), hydrogen, and C₁-C₈ hydrocarbons, which can include natural gas and producer gas. Producer gas is a low quality gas formed from hot carbon in a water- gas shift reaction. The preferred reducing gas is CO.

The reduced metal oxides are only partially reduced oxides before the chlorine reaction step. The metal oxides are generally never reduced to the zerovalent metal. For example, upon contact with the reducing gas, Fe₂O₃ can be partially reduced to Fe₃O₄ and then potentially further reduced to FeO, but FeO is not reduced further to zerovalent Fe, i.e., elemental iron, under the conditions of the present invention.

Although not intending to limit the process of the present invention, it is believed that the reducing gas facilitates selective reduction of the metal oxide

impurities as opposed to the calcined magnesite. For example, the reducing gas reacts with Fe₂O₃ to form Fe₃O₄ but does not substantially react with the calcined

magnesite to reduce the magnesium oxide. In other words, the reducing gas generally only reduces the iron oxide impurities and not the calcined magnesite. This is

important because chlorine gas preferentially reacts with the reduced iron oxide impurities rather than with the calcined magnesite, thereby avoiding the production of MgCl₂ during the chlorination step.

The resultant reduced metal oxides are then contacted with chlorine gas to chlorinate the resultant reduced metal oxides to chlorides, which are volatilizable within the temperature range of the reaction. Iron

chloride (FeCl₃) is volatile at all temperatures within the temperature range of the process. Although nickel and manganese chlorides are not typically volatile at all temperatures within the temperature range of the process, under the conditions of the process of the invention they are capable of being volatilized, and thereby removed.

The chlorination step generally involves chlorination and volatilization of the metal impurities, e.g., iron oxides or ferrites, with the oxygen being left behind in the crystalline lattice. The remaining metal impurities then combine with the oxygen and form the fully valent state. For example, upon chlorination of the iron oxides Fe₃O₄ and FeO, volatile FeCl₃ and the fully valent Fe₂O₃ are formed. Thus, the process of the present

invention generally involves the reduction of Fe₂O₃ to Fe₃O_A and the chlorination of Fe₃O₄ to FeCl₃ and Fe₂O₃·

These two reactions are cyclized to reduce the iron content to preferably less than about 0.5%, and more preferably less than about 0.1%.

The steps of contacting the material with a reducing gas and chlorine gas can be continuously repeated as necessary to obtain the desired purity of the

contaminated magnesium oxide, e.g., calcined magnesite. Preferably, the contaminated magnesium oxide is purged with an inert gas, such as N₂ or CO₂, between the steps in which a reducing gas and Cl₂ are used.

The calcined magnesite is in a fine powdered form generally having a particle size of not greater than about 0.1 mm, and preferably not less than about 0.05 mm. More preferably the calcined magnesite has a particle size of not greater than about 0.074 mm, i.e., that which passes through a 200 mesh screen. Most preferably, the particle size of the calcined magnesite is about 0.05 mm to about 0.74 mm.

In a typical fluidized bed system, gas entrance concentrations of the reactants typically are about 100% to about 5%, more preferrably about 90% to 10%, and most preferrably about 75% to 15%. These percentages are based on volume. Gas flow times for each portion of the cycle may vary for each gas, preferably with flow durations of about 1-10 minutes. More preferrably the gas flow times are about 2-8 minutes. The inert gas purge time, i.e., flow duration, is typically less than about 5 minutes, if an inert gas purge is used. More preferably, the inert gas purge time is about 1-3 minutes, and most preferrably about 1 minute. Gas concentrations, flow rates, and tiroes can be adjusted as the number of cycles increases to ensure total input to be above the stoichiometric demand for reduction and chlorination of the reduced metal oxides for that cycle.

By using a gaseous reductant, as opposed to carbon in the fluidized bed, the necessity of physically

separating the carbon from the MgO in each cycle is

eliminated. Furthermore, a gaseous reductant accurately limits the time of the reduction step, and provides for a more precise separation of it from the chlorination step. This type of process is designated herein as a "phased reaction," i.e., one in which the reducing and chlorinating functions are separate steps.

The chlorination step of the cycling process is carried out at a temperature of about 700-1000°C.

Preferably, both cycling steps, i.e., the reduction and chlorination steps, are carried out at a temperature of about 700-1100°C, more preferably at a temperature of about 750-1000°C, and most preferably at a temperature of about 750-850ºC, depending upon the calcination temperature and the heat balance requirements. For certain embodiments of the present invention, the cycling process steps are preferably carried out at a temperature of about 750-800°C. This temperature range is advantageous when highly active MgO is desired. Such active MgO is useful in coal-fired power plant scrubbers to remove sulfur from coal gases, and for acid neutralization in animal feed.

In these temperature ranges, essentially no MgCl₂ is formed . Any trace amounts of MgCl₂ formed would

generally be in the liquid phase, however, such amounts would be so small as to not detrimentally affect the fluidization. In these temperature ranges, the .impurity metal chlorides formed are removed with the exiting gases.

In cycling the reactants, i.e., the reducing gas and Cl₂, it is possible to achieve higher ratios of metal impurity removal than might be expected from the

thermochemical reduction equilibria. Although not

intending to be limiting in any manner, this might be because chlorine is adsorbed onto, and desorbed from, the surface of the solid material, allowing reactivity of the iron with adsorbed Cl₂ even during the reduction phase of the process with the reducing gas, e.g., CO.

The reverse reaction of the volatile metal chlorides, e.g., FeCl₃, with MgO, would typically be expected to occur; however, this reaction does not appear to occur, at least to any significant extent, in the process of the present invention. Furthermore, the metal impurity removal, e.g., iron removal, per cycle seems to be a constant percentage of the impurity remaining so that a relatively constant coefficient can be applied to express the impurity remaining per cycle. A "cycle" is used herein to mean a period of reduction followed by a period of chlorination, with or without a purge between reduction and chlorination and vice versa.

Cyclical operation can be employed in any of a number of ways. A batch reactor can be used, or the process can be carried out with a continuous solids feed to a set of reactors in series. The process can also be carried out using a shaft furnace with pelletized calcined magnesite.

It is often desirable, however, to conduct chemical reaction or purification of minerals, particularly those with a fine particle size such as flotation

concentrates, in a fluidized bed reactor, with the

resultant product being removed either as underflow or overflow. This is the preferred reactor type for use in carrying out the method of the present invention.

Using a fluidized bed, either batch or multiple reactor continuous flow can be employed, with the batch mode preferable. In a batch reaction mode, two reactors in parallel are convenient and preferred, switching the Cl₂ and CO streams between the reactors, and recycling the excess gas as the iron content falls.

Using continuous flow, it is generally necessary to use several stages to achieve low residual iron values. This can be accomplished using a number of reactors in series. In such a system, the process of the present invention is conducted with a continuous flow of solids through multiple fluidized bed reactor stages.

A comparison of batch and continuous flow methods is presented in Example 1 and Table 2, which compares single reactor continuous flow and multiple reactor

continuous flow with a batch method.

For a fluidized bed reactor, or reactors, velocities are usually limited by the feed particle size and the need for vigorous fluidization in order to fluidize a bed of MgO. Although velocities, as well as concentrations of the gases, can vary during the process depending on the amount of metal impurities present, velocities can range from about 0.05 feet per second to about 0.5 feet per second, preferably from about 0.2 to 0.4 feet per second.

Cycle times can vary from two minutes to twenty minutes, preferably from three to ten minutes. The number of cycles per batch is related to the "coefficient," which is obtained by the equation:

%Fe/%Fe_o = xⁿ

where %Fe represents the amount of Fe remaining after the process of the present invention, %Fe_o represents the amount of Fe in the original sample, n is the number of cycles and x is the "coefficient," which appears to remain substantially constant as the impurity, e.g., iron, content is reduced.

Gas concentrations are preferably varied as the metal impurity, particularly iron impurity, concentration varies. For example, as the concentration of the impurity is lowered, the entering gas concentration can be lowered to avoid waste by the use of excess gas, particularly of Cl₂. Furthermore, gas velocities and flow times can vary as the concentration of metal impurities varies, so as to decrease the amount of waste gases.

The exiting gases from the fluidized reactor can be cooled in flues, preferably separate flues to effectuate gas recycling. Any CO₂ present could either be scrubbed from the CO stream or a portion by-passed to the CO

generator. The method of the present invention is

advantageous over the methods that reduce the iron

impurities in magnesite to elemental iron. This is because the reduction of iron impurities to elemental iron is greatly adversely affected by the presence of CO₂, whereas the reduction of iron oxide impurities to the reduced oxides Fe₃O₄ and FeO is not. If conditions are adjusted such that excess Cl₂ is not used, very little Cl₂ exits the reactor with the volatile metal chlorides, such as FeCl₃. The volatilizable metal chlorides are then preferably condensed and collected as a solid.

The magnesite, i.e., MgCO₃, which has had the talc, silica, and other siliceous constituents removed, is initially calcined, i.e., heated to drive off volatile components and form MgO. The calcined magnesite enters the reactor preferably at or near calcination temperatures, i.e., less than about 1450°C, preferably less than about 1100°C, and more preferably less than about 850ºC. The heat generated from the reduction/chlorination reactions is preferably used to maintain the temperature of the reactor.

Depending on the magnesite calcination. temperature, two regimes of operation are possible. The reactions of low surface area materials, which result from calcination at relatively high temperatures, i.e., above about 800°C, have relatively high reaction rates at high temperatures. Lowering the temperature of the

reduction/chlorination process lowers the amount of metal impurity removal per cycle. At calcination temperatures at or below 800ºC, high surface area calcined magnesite results. Calcination temperatures between about 700°C to 800°C yield MgO, which has a tapped density of about 59.13 lbs/ft³. Calcination temperatures over 800ºC yield MgO with a significantly higher density (therefore lower surface area). For example, a calcination temperature of 1000ºC yields MgO with a tapped density of 96.9 lbs/ft³. The added surface area of MgO calcined at temperatures below about 800°C, and preferably at about 700°C to about 800°C, contributes significantly to reaction rates, such that lower temperatures of the reduction/chlorination process do not result in the expected lower rates of reaction and metal impurity removal .

Using thermochemistry as a guide, which is discussed in further detail below, the stages of the reduction of the metal impurities, e.g., iron reduction, can be characterized. Based on thermochemical values, one would expect cycle "coefficient" values to be 8/9ths;

however, higher metal impurity removal, i.e., lower

"coefficient" values are actually obtained by the process of the present invention.

It has been observed in the chlorination of other reduced oxides that the oxygen stays in the crystal lattice or in the particle while a volatile chloride species escapes. The oxygen is removed only in the reduction step, or during a simultaneous reduction/chlorination reaction. If the oxygen stays in the lattice during the chlorination phase, then only l/9th of the iron can be volatilized as FeCl₃ per cycle. Under conditions, such as low temperature calcination, where the resulting MgO possesses high surface area, the on-off nature of the "phased" reactor concept is modified by adsorption. This allows a reaction situation where CO and Cl₂ are together at least at the surface of either the MgO or FeO_x or both.

Although the scope of the invention is not to be bound by any particular theory, the following calculations and observations indicate the thermodynamics of the process of the invention. Also, the following discussion of the thermodynamics involved in the process of the present invention is discussed in terms of iron oxides, although this is not intended to be limiting in any manner.

Calculation of the amount of iron oxides in the form of Fe₂O₃ reduced to Fe₃O₄ by CO, a preferred reductant, which is then chlorinated to FeCl₃, using the JANAF

thermodynamic tables, results in an expected elimination of only about l/9th of the iron present per reduction/

chlorination cycle according to the equations:

9Fe₂O₃ + 3CO ---> 3CO₂ + 6Fe₃O₄

6Fe₃O₄ + 3Cl₂ ---> 2FeCl₃ + 8Fe₂O₃

(with 8/9ths Fe₂O₃ remaining).

The further reduction of Fe₃O₄ to FeO has a much lower ΔG, with an equilibrium constant K represented by

P_co2 /P_co. The ratio of CO₂/CO for the first reduction step

2

of Fe₂O₃ to Fe₃O₄ at 1000 K is 1.1 x 10⁵ :1, which decreases with an increasing temperature to 2.7 x 10⁴:1 at 1300 K.

The second reduction step, Fe₃O₄ to 3FeO, has a CO₂/CO ratio of only 1.43:1 at 1000 K, which increases to only 18.8:1 at 1300 K.

The ratio of the equilibria of the first reduction step to the second reduction step is 8539:1 at 1.000 K.

Even though this ratio decreases with increasing

temperature, the ratio is still relatively high going from 8539:1 at 1000 K to 1451:1 at 1300 K. So, at all

temperatures in the contemplated operating temperature range, there is still a large factor in favor of the first reduction step to Fe₃O₄ over the second reduction step to FeO.

If the kinetic reduction rate is related to the thermodynamic potential, the first step of the reduction of Fe₂O₃ to Fe₃O₄ would be greatly favored over the second step of Fe₃O₄ to FeO. Reduction of FeO to Fe has a positive free energy and would be unlikely for any reasonable CO efficiency. See Table 1.

Table 1

Reduction Thermochemistry

Thus, upon a thermochemical evaluation it is expected that for a cyclic reaction, CO reduction followed by Cl₂ chlorination, the amount of iron remaining would be 8/9ths per cycle.

If kinetics were so fast that substantial equilibria attainment were achieved under the experimental conditions, with large excesses of CO, which increases with decreasing iron content for batch reactions, the iron removal coefficient would fall below 8/9ths, according to the equations:

3Fe₂O₃ + 3CO ---> 6FeO + 3CO₂

6FeO + 3Cl₂ ---> 2Fe₂O₃ + 2FeCl₃

or 4/6ths (0.667) of the Fe remaining. If reduction were to wustite (FeO_.947) one could have a ratio of 0.631. This is the lowest coefficient ever observed.

It is apparent from the data of Table 3 below, which relates the bed iron remaining after several cycles to other experimental conditions, that the rate of iron removal does not meet the thermochemical expectations of reduction to FeO per cycle. Thus, thermochemical equilibria calculated from JANAF thermodynamic tables show the ease of reduction to Fe₃O₄ from Fe₂O₃ but the difficulty of going to FeO. The "coefficients" are frequently below the 8/9ths value.

Considering the data at low temperature

calcination, where equilibrium calculations show the relative equilibria for CO reduction per step, i.e., Fe₂O₃ to Fe₃O₄ and Fe₃O₄ to FeO, to be much lower at low

temperatures, the effectiveness of higher surface area in promoting the iron removal indicates that a different reaction mechanism is involved.

These data, combined with the experimental observation that FeCl₃ continues to come from the reactor after the Cl₂ portion of the cycle, appear to indicate a surface adsorption of Cl₂. In the subsequent CO. reduction step, the adsorbed Cl₂ causes a reaction of Fe₂O₃ + CO + Cl₂, which has a much more favorable equilibrium constant, to form FeCl₃. Again, these comments are merely

theoretical, and not intending to limit the process of the present invention.

It is understood that the amounts of the reducing gas and chlorine gas, the reaction temperature and velocity of the bed reactor can be varied with respect to each other as long as essentially no magnesium chloride is formed. These variables may be adjusted to improve the coefficient of iron removal and/or to improve economic efficiency.

The following examples illustrate various embodiments of the invention. The examples, however, are not meant to limit the scope of the invention which has been fully characterized in the foregoing disclosure.

Example 1

Theoretical Results of Batch vs. Continuous Feed Cycling

The information presented in Table 2 compares single reactor continuous flow with multiple reactor continuous flow, and with a batch method for removal of iron impurities using theoretical calculations. For the data in Table 2, the relative reactor sizes are as follows: Batch = 8.52 ft. ID; 1 continuous reactor = 2.01 ft. ID; 3 continuous reactors in series = 5.0 ft. ID.

The theoretical results, using a constant 0.800 coefficient for the Fe removal per cycle, shown in Table 2 demonstrate that 11.9% of the initial iron would remain after 3 stages of 6 cycles each, i.e., 0.412%, and 21.82% of the initial iron would remain after 18 cycles in 1 continuous flow reactor, i . e . , 0 .755% , whereas 1. 8% of the initial iron content would remain after 18 cycles in a batch reactor, i.e., 0.062%. Thus, batch systems are at least theoretically favored.

Table 2

Calculation of Batch vs. Continuous Feed Cycling

6 CYCLES/REACTOR FOR

3 IN SERIES REACTORS, 18 CYCLES IN A SINGLE REACTOR AND

18 CYCLES PER BATCH

Initial [Fe] = 3.46% Fe

Coefficient of iron removal = 0.800

Fe Remaining = 0.412%

Thus, 11.9% of the

initial iron would be

remaining using 3 continuous flow reactors

in series with 6 cycles

run per reactor. Examples 2-7

Effect of Varying Conditions on the Removal of Iron

From Calcined Magnesite

Example 2

The reactor used in this example consisted of a 4" diameter quartz reactor with a cone bottom, which was 48" tall with a 7" tall cone, and a side arm fitted with a ball joint to which was attached a glass condenser. The reactor was heated with electrical windings so that the reactor zone and cone were held at the desired reaction

temperature, while the upper part of the reactor was held at temperatures well above the condensation temperature of FeCl₃.

Magnesite flotation concentrate (2185 grams) was calcined at 1000ºC to drive off volatile components, cooled, and removed from the reactor. The calcined

magnesite (803.8 grams), which had an iron content of 3.46% Fe, was reloaded into the reactor.

Gas flows were measured by capillary flow meters and the pressure was monitored by a water filled manometer. The condition of the bed fluidization could be judged by the manometer fluctuations and confirmed by the state of the bed after cooling.

Bed samples were taken for iron analysis by ICP. The calcination was accomplished before the run was made by carefully raising the temperature to the desired

calcination temperature, which was to be the reaction temperature of the run.

Seven cycles of CO/Cl₂ at 7.7% CO and Cl₂ were conducted with purging between a 15 minute CO phase and a 15 minute Cl₂ phase. The fluidization velocity was 0.24 feet per second. The bed of MgO was cooled and removed. The parameters and results appear as. run #1 in Table . The iron content of the treated calcined magnesite bed was 0.30% Fe. Using this value and the starting Fe content of calcined magnesite, the coefficient of iron retained can be calculated for the seven cycles by the equation:

Fe/Fe_o = xⁿ

where x is the coefficient and n the number of cycles

0.3/3.46 = x⁷; x = 0.705.

Example 3

Using the same apparatus at the same temperatures with higher CO and Cl₂ concentrations of 60% and 40%, respectively, but with cycle times of only 2 minutes, twenty cycles produced a product of 0.06% Fe. The

calculated coefficient was 0.817. The results of this experiment are shown as run #2 in Table 3. This result shows that thermochemical equilibrium does not prevail.

Example 4

Using the above reactor and a similar bed calcined and reacted at 800ºC, the "coefficient" was 0.783 for the first six cycles. The bed was withdrawn, sampled and twelve more cycles were run. The "coefficient" for the second portion of the run was 0.781. This demonstrated the constancy of the "coefficient." The results of this experiment appear as run #3 in Table 3.

Comparing with Example 3, one notes the

"coefficient" is still quite good at the lower temperature when compared to the following Example 5.

Example 5

Using the same reactor as above, with a bed calcined at 1000°C, and with the reduction-chlorination reaction at 815°C, a much poorer "coefficient" of 0.970, compared to 0.781 for Example 4, was obtained. The results of this experiment appear as run #4 in Table 3 . Example 6

Following the above examples, but with lowered calcination and reaction temperatures to just above the MgCl₂ melting point, a "coefficient" of 0.853 was obtained. The results of this experiment appear as run #5 in Table 3. This shows the effect of reduced temperatures even in the calcination temperature region producing high surface area by comparing the "coefficient" with the Example 4

"coefficient" 0.853 versus 0.781. Example 7

Following the above experiments in equipment and technique, a run at increased Cl₂ concentration, increased time, and decreased velocity, conducted at a temperature well below the MgCl₂ melting point shows an extremely low reactivity and an extremely high "coefficient." The parameters and results appear as entry #6 in Table 3.

Table 3 shows the effect of widely varied conditions on the resulting coefficient. An interesting effect of the calcination temperature is shown in comparing #3 and #4. Run #4, with a 1000ºC calcination temperature and an 815°C reduction/chlorination temperature, was less effective at removing the iron, with a coefficient of

0.970, than was run #3, with 800°C calcination and

reduction/chlorination temperatures, which had a

coefficient of 0.783. It is noted that samples taken after six cycles, and then twelve cycles more, for run #3 have nearly identical coefficients. Note also that the bed velocity was higher for run #3, being 0.44 feet per seoond versus 0.26 feet per second for run #4. Other comparisons of pairs of runs establish the constancy of the "coefficient" cycle by cycle as the iron content changes by nearly two orders of magnitude, the effect of lower calcination temperature, the effect of the reduction/chlorination temperature, and the cycle times.

y

NOTE #1: Balance of gas is Nitrogen

NOTE #2: Bed removed and relntroduced after sampling

NOTE #3: Head sample recalcined at 1000°C has 3.46Z Fe.

NOTE:* Nitrogen purges at same bed velocity between reactant gases.

The foregoing detailed description has been given for clarity of understanding only and no unnecessary limitations are to be understood therefrom. The invention is not limited to the exact details shown and described, for obvious modifications will occur to those skilled in the art.

Claims

WHAT IS CLAIMED IS:

1. A process for removing metal impurities from impure magnesium oxide comprising:

(a) contacting at a temperature of about 700-1100°C the impure magnesium oxide with a reducing gas capable of partially reducing oxides of the metal impurities to form magnesium oxide and partially reduced metal oxides;

(b) contacting the magnesium oxide and the partially reduced metal oxides with chlsrine, at a temperature of about 700-1100ºC, so that formation of magnesium chloride is substantially prevented and volatilizable metal chlorides are formed; and

(c) cyclicly repeating steps (a) and (b) until at least about 75% of the metal impurities are converted to volatilizable metal chlorides.

2. The process of claim 1 wherein the impure magnesium oxide is a calcined magnesite flotation concentrate, and calcined magnesite containing reduced metal oxides is formed in the reduction step.

3. The process of claim 2 further comprising removing the volatilizable metal chlorides from the calcined magnesite to form substantially pure calcined

magnesite.

4. The process of claim 1 wherein step (a) is conducted at a temperature of about 750-800°C.

5. The process of claim 1 wherein the metal impurity is iron.

6. The process of claim 1 where the reducing gas is chosen from a group consisting of CO, C₁-C₈

hydrocarbons, and hydrogen.

7. The process of claim 6 wherein the reducing gas is CO.

8. The process of claim 1 wherein steps (a) and (b) are conducted in a fluidized bed reactor.

9. The process of claim 8 wherein the fluidized bed

reactor is used as a batch reactor.

10. The process of claim 8 wherein steps (a) and (b) are conducted with continuous flow of solids through multiple fluidized bed reactor stages

11. The process of claim 1 wherein the impure magnesium oxide is calcined magnesite obtained from heating natural magnesite at a temperature of about 850ºC or below.

12. The process of claim 11 wherein the calcined magnesite has a particle size not greater than about 0.1 mm.

13. The process of claim 12 wherein the calcined magnesite has a particle size of about 0.05 mm to about 0.1 mm.

14. The process of claim 1 wherein the contaminated

magnesium oxide is purged with an inert gas between the steps of contacting the contaminated magnesium oxide with a reducing gas and with chlorine.

15. The process of claim 1 wherein steps (a) and (b) are cyclicly repeated until at least about 90% of the metal impurities are converted to volatilizable metal chlorides.

16. The process of claim 1 wherein the volatilizable metal chlorides are condensed and collected as a solid.

17. A process for removing metal impurities from calcined magnesite comprising:

(a) contacting at a temperature of about 750-850°C the calcined magnesite with a reducing gas capable of partially reducing oxides of the metal impurities to form calcined magnesite containing partially reduced metal oxides, wherein the calcined magnesite has a particle size of about 0.05 mm to about 0.1 mm;

(b) contacting the calcined magnesite containing partially reduced metal oxides with chlorine at a temperature of about 750-850ºC so that formation of magnesium chloride is substantially prevented and volatilizable metal chlorides are formed;

(c) cyclicly repeating steps (a) and (b) until at least about 75% of the metal impurities are converted to volatilizable metal chlorides; and

(d) removing the volatilizable metal chlorides from the calcined magnesite to form substantially pure calcined magnesite.

18. The process of claim 17 wherein the reducing gas is CO.

19. The process of claim 17 wherein the velocities and

concentrations of the reducing gas and the chlorine are varied over time and cycles as the metal

impurities are removed to decrease the amount of waste gases.

20. A process for removing metal impurities from calcined magnesite flotation concentrate comprising:

(a) contacting at a temperature of about 750-800°C the calcined magnesite flotation concentrate with sufficient CO to partially reduce oxides of the metal impurities to form calcined magnesite flotation concentrate containing partially reduced metal oxides;

(b) contacting the calcined magnesite flotation concentrate containing partially reduced metal oxides with chlorine, at a temperature of about 750-800ºC, so that formation of magnesium chloride is substantially prevented and volatilizable metal chlorides are formed; and

(c) cyclicly repeating steps (a) and (b) until at least about 95% of the metal impurities are. converted to volatilizable metal chlorides.