Separation Technologies for the Industries of the Future (1998)

The U.S. steel industry produces about 100 million tons annually. In 1994, this constituted 12.6 percent of world production. In addition, the United States imports about 25 million tons of steel each year. The value of U.S. steel production is approximately $50 billion per year, and the industry employs about 170,000 workers (AISI, 1997).

Steel is the fourth largest energy-consuming industry in the United States, using 1.944 quadrillion Btus in 1994. This was 2.3 percent of the energy consumed in the United States and 20 percent of the energy consumed in U.S. manufacturing. Also in 1994, the industry produced 32 million tons of solid waste and by-products. In general, the industry recycles 95 percent of the water used in steel production and processing (AISI, 1997).

The steel industry comprises two major parts. The first is called the ''integrated'' segment because iron ore and coke (made from coal) are combined in blast furnaces to produce molten iron (called hot metal), which is then refined and alloyed to produce various types and grades of steel. The oxygen-based refining process, which is the one most widely used today, is sometimes called the basic oxygen process and may be carried out in a basic oxygen furnace (BOF) or a basic oxygen converter. Oxygen-based steelmaking accounts for about 60 percent of the steel produced in the United States each year.

The second, and fastest growing, part of the U.S. steel industry is based on the electric arc furnace (EAF), which can melt steel scrap to make liquid steel without

going through the ore-coke-blast furnace cycle. Fluxes and oxygen are used to further refine the EAF output, usually by ladle treatment. Facilities that use EAF steelmaking are popularly known as "minimills" because they are cost-effective and much smaller than integrated plants. Minimills now produce about 40 percent of U.S. steel output.

Although there are broad similarities between integrated steel mills and minimills, each have unique requirements. Among integrated steel mills, there are also broad similarities between separation processes for raw materials and for products that are recycled and reused.

Traditional integrated steelworks use iron ore, coal, limestone, and oxygen or air to produce steel. The first step in this process is the conversion of coal to coke, a porous, strong, carbon-rich material used in blast furnaces to provide most of the reducing power and heat required to reduce iron from iron oxides. Coke also maintains porosity in the furnace charge while the iron and slag are being melted.

Coke is produced by the destructive distillation of coal, and valuable organic compounds are separated as by-products. When the coke is finished, however, it is a red-hot mass that must be pushed from the coke oven and quickly quenched with water, which emits copious amounts of gases, steam, and particulates that pose severe air and water pollution problems. Although the industry has tried to solve these problems for many years, better methods of separation are continuously being sought, particularly for the removal of particulates from gases and water. Methods of treating quench water so it can be recycled and reused are also of interest to the industry. Over the years, the industry has tried to cut down or even eliminate the use of coke, and today the "coke rate" (pounds of coke per ton of hot metal) is much lower than it was even a few years ago. Steelmakers have also lowered coke rates by adding fuel oil or powdered coal through the blast furnace tuyeres. Anything that lessens coke requirements lessens the environmental problems associated with coke production.

As supplies of the best natural ores (ores that do not require beneficiation) dwindled, the chemistry and physics of ironmaking in the blast furnace were becoming better understood. By the time the industry was forced to turn to much lower grade ores, which had to be finely ground, beneficiated, and reconstituted as pellets, increased productivity in the blast furnace had more than redressed the balance. Today, the production records belong to sintered ores (partially agglomerated ores strong enough to permit efficient gas flow in the blast furnace). Sintering

also provides opportunities to incorporate fluxing agents into the mix to make their use more efficient and to reintroduce mill scale (the iron oxide waste produced during rolling). The sintering machine can take ore as coarse as 10mm and as fine as "all passing through 100 mesh," provided that efficient air flow through the bed is maintained. Although some iron ore producers are still motivated to increase product grades, ironmakers must take other factors into account. For example, a high grade iron oxide pellet would not be acceptable unless it could stand up to reduction in the blast furnace.

The process of ironmaking takes place in the blast furnace, a highly productive unit capable of turning out several thousand tons daily of molten iron, known as "hot metal" or "pig iron" (if cast). Inside the blast furnace, a high-temperature reducing atmosphere is maintained while fuel and ore are fed into the furnace, more or less continuously, through the top. Pressurized preheated air is blown in near the bottom, providing oxygen for coke combustion and carbon monoxide formation. In the shaft (stack) of the furnace, the mix of coke, limestone, and iron-bearing material progresses downward as solid-state reduction proceeds. Near the bottom of the stack, temperatures reach 1,000°C and slag formation commences, involving the added limestone, silica, and alumina from the ore, as well as other impurities from both the ore and coke. In the fusion zone, temperatures reach 1,200°C to 1,600°C. This portion of the furnace, called the bosch, terminates at the tuyere line, where everything except coke is molten. Near the tuyeres there is a limited oxidation zone in which coke can still be burned, raising the temperature to 1,900°C. In the hearth of the furnace, below the tuyere line, hot metal and slag can collect. Some separation of impurities may still take place, but the molten layers are relatively quiescent and are frequently tapped. Some formation of refractory compounds (e.g., TiN) may take place above the hearth. These are solids and may be troublesome in the downstream steelmaking operations.

As a separator, the blast furnace is only effective for removing silica, alumina, and other gangue minerals and the oxygen in iron minerals. The burden of removing impurities is therefore placed on the steelmaking process. Blast furnace hot metal is far too high in carbon to be directly useful except as pig iron in certain foundry operations. The various processes for making steel are fairly effective for separating impurities, and so are used for this purpose.

The integrated steelmaking industry relies almost entirely on oxygen processes to make steel from hot metal. A typical heat yields about 200 tons of finished steel in about 38 minutes from a charge of 80 percent carbon-saturated hot metal and 20 percent heavy steel scrap. About 10 tons of flux, mostly lime, is also added to the furnace to form a slag for the capture of impurities. During the "blow," large quantities of oxygen forced through the molten bath oxidize carbon, silicon, and other impurities.

Although oxygen steelmaking is a mature process, improvements could still be made in the removal of nitrogen and phosphorus, a requirement for many higher-quality steels. Nitrogen can only be swept out as a gas by carbon dioxide generated in the process, and phosphorus must be oxidized to the slag. Neither separation is very efficient as presently practiced. The furnace slag itself contains recoverable metallic iron, but its separation requires that the slag be crushed and magnetically separated, an inefficient, labor-intensive process. Finally, the oxygen furnace produces a fine dust that must be separated and collected. This dust has a very low bulk density, is difficult to handle, and is usually deposited in a landfill, even though it may contain potentially valuable constituents, such as zinc.

Further refining downstream of the oxygen process may be required to control residual impurities and prevent inclusions. Better knowledge of the relationships between steel properties and the specific costs and benefits of better separation technologies could help prioritize research.

A typical EAF produces 150 to 200 tons of steel from scrap in a heat (about one hour). Fluxes and oxygen may be used, but to a lesser degree than in an integrated shop. These may be used in the EAF itself, but it is more likely to be done in a ladle. Because of the relative purity of the scrap charge, less slag builds up in the EAF operation than in a BOF, but the EAF still has the same problems with the removal of nitrogen and phosphorus. Improved separations here would benefit both segments of the industry.

Dust from the EAF is a serious problem, partly because 1 to 2 percent of the charge by weight appears as dust, but mostly because the EAF dust contains substantial quantities of iron oxide, zinc oxide, and lime, as well as smaller, but very troublesome, quantities of lead oxide, cadmium oxide, and other materials. Like the BOF dust, EAF dust is very light and difficult to handle. For EAF dust to be economically attractive, separation technologies must be found to remove the zinc oxide from the iron oxide, which also contains some zinc ferrite. It is possible that, if a zinc-rich fraction could be separated, zinc could be produced from it. The iron-rich fraction could then be recycled.

After oxygen or EAF steelmaking, the steel is further refined prior to casting. Ladle, or secondary, refining processes include deoxidation, desulfurization, and vacuum degassing. The primary purpose of secondary refining is to produce clean steels from which second phase inclusions, such as alumina (Al₂O₃) and other oxides, have been removed. These inclusions are small (1 to 100 μm) and are

usually captured in a slag on top of the steel. Little is known about the separation mechanism from liquid metals. The problem includes the transport of the inclusion to the interface and its dissolution in the liquid top slag. A better understanding of these separation processes would be useful to the industry.

Both segments of the steel industry have trouble obtaining enough clean scrap of known composition. The integrated steel industry is probably in somewhat better shape because it controls its own "home" scrap, but it also needs heavier scrap as a coolant in the BOF. The minimills, obviously, would suffer quality problems as well as growth constraints if they could only use obsolete scrap, especially automotive shredder scrap.

Like the integrated mills, the EAF plants try to use any home scrap they produce because it is heavy to transport and its composition is known. They also compete for prompt industrial scrap generated in the manufacture of steel products, such as automobiles, because this is also quality steel with low residual impurity levels. However, the advent of continuous casting has reduced the quantities of home scrap available, and more efficient manufacturing methods are reducing the availability of prompt industrial scrap.

Postconsumer scrap can be in almost any form. Typically, it is shredder scrap contaminated with both metallic and nonmetallic impurities. For example, automobiles contain significant quantities of nonferrous metals, alloying elements, such as nickel and chromium, and combustibles, mostly plastics. Because these impurities are hidden in the automobile's structure, they cannot be recovered before shredding, and magnetic separation after shredding is only partially effective. Scrap is usually washed to separate nonmetallics, but shredder scrap is relatively coarse, and the remaining metallic contaminants cannot be effectively removed by highintensity magnets (for paramagnetic materials) or eddy current separators (for conductive nonmagnetics).

The most troublesome impurities in EAF steelmaking today are copper and, to a lesser extent, zinc. For higher-quality steel products, such as flat rolled sheet and bar-quality steels, copper and other residual elements must be below specified levels in the EAF feed because they are not removed in the process and their presence may make it impossible to achieve the desired properties. A flat rolled steel, for example, may have a maximum allowable copper level of 0.1 percent, as opposed to 0.25 percent copper in typical automobile shredder scrap. Some plants that run on scrap try to use dilution with other lower-copper scrap, and some do not attempt to produce certain grades of steel, but competitive pressures are forcing EAF operators to expand into a broader spectrum of products. Obviously, better separation methods for removing copper and other residual impurities from scrap is one of the most critical problems facing the steel industry today. In the future, however, changes in the design of

automobiles and the materials that go into them, such as the use of plug-in electrical components that can be recovered before shredding, may have a greater impact than improved methods for separating shredder scrap.

From the beginning of the "minimill revolution" more than 30 years ago, a great deal of consideration was given to direct reduced iron (DRI). The various processes that were studied were all aimed at removing oxygen from iron ores in the solid state. By the 1960s, a few plants had been built using coal-based and natural gas-based processes. Some of these were in locations where a small steel mill could be justified, and the scrap for a more conventional EAF plant was not available (New Zealand, for example). In the United States, scrap was preferred for reasons of cost, both capital and operating. In the 1970s, reformed gas became the fuel/ reductant of choice, although some coal-based plants were in operation and some process development involving coal continued.

There is a strong positive incentive for using high grade feed for direct reduction because slag-forming gangue minerals in the DRI impose additional burdens on the EAF. Suitable feeds ideally contain no more than 3 percent silica plus alumina, but very few natural ores approach this level of purity. Therefore, the feed is usually finely ground, beneficiated by magnetic separation, flotation, or both, and reconstituted as fired oxide pellets before reduction. After reduction, the product typically contains 88 percent or more metallic iron, 8 percent or less unreduced iron oxide, and about 2 percent silica and other constituents.

A finished DRI pellet is very reactive. It has a large internal surface area, will reoxidize in contact with moist air, and may even burn if it contacts water. DRI pellets obviously present problems for storage and shipping, for which producers have tried many different remedies. The most widely used remedy today is hot briquetting, a process in which the DRI pellets, under a protective atmosphere, are briquetted to reduce their surface area and seal off internal surfaces. The DRI briquettes can be handled, stored, and shipped, even across the oceans.

A more recent alternative that has its own advantages is producing iron carbide (Fe₃C). The raw materials for this method must also be high-grade ore or pellets, and the fuel is reformed natural gas, but the product is stable and does not require briquetting. The first carbide plants are now being established, but it remains to be seen how well iron carbide will fare in competition with more conventional DRI.

DRI (and carbide) have several advantages over scrap. First, they can dilute impurities, such as copper. Second, they are steady, predictable sources of highquality feed. Third, they are a means of "importing" low-cost energy from places where natural gas is readily available. Finally, they can act as hedges against rising scrap prices. They do, however, require a capital investment that is not incurred by using scrap.

Many EAF plants in the United States are located where they are because of the availability of scrap, and cheap natural gas may not be available at the same sites. The circumstances for DRI plants are entirely different. Very few U.S. ores are suitable as DRI feeds, and DRI plants in the United States must therefore subsist on imported pellets. The trend toward overseas sites, where cheap gas is available and high-grade pellets can be brought in easily, is probably irreversible. By the same token, gas-based DRI technology has moved ahead so far and so fast that coal-based processes will probably be relegated to areas where available coal can be used or where only incremental quantities of DRI are needed. Therefore, it is doubtful that a DRI process using coal would be of great value to the U.S. steel industry. Just as the iron ore industry became internationalized several decades ago, the DRI industry as a whole will develop according to the supply of ores, the availability of natural gas, and transportation. The cost of DRI will be the summation of these costs.

Given the reactivity of fresh DRI, removing impurities from this product is out of the question, so impurities must be excluded from the feed. Therefore, ore beneficiation would be a better target for research related to DRI. The coal separation needs identified here are the same as those for coal that is intended for coke production. Whether for coke or DRI, there are physical requirements (i.e., size and moisture) that make it difficult to consider flotation as a separation method for improving the grade of coal. Liquefaction works on the coal rather than ash and sulfur.

In coke production, gases are produced and large volumes of liquids are used. These gases and liquids carry particulates that complicate separation and treatment. Even so, major amounts of VOCs, which are valuable by-products, are routinely recovered from coke operations. Water is partially recovered and recycled, decreasing the use of fresh water. Although the coke ovens are one of the most obvious sources of gaseous and liquid emissions, all iron and steelmaking processes produce similar emissions. Therefore, research on the separation of fine particulates from gases and liquids could yield broad benefits for the industry as a whole.

Currently, phosphorus is removed from molten metal by oxidation to the slag, and nitrogen (N₂) and residual gases are removed in the large quantity of CO gas generated in the process. These processes could be improved. Considerable research has been done on methods of refining copper from liquid iron using various

fluxes. The sulfidation of copper, where the Cu₂S is dissolved into a sulfide matte, has been investigated in detail by a number of researchers. These studies have shown that it is impractical and uneconomical to refine copper from molten iron and that it is best to separate copper out in the scrap stage.

Separation technologies for upgrading scrap include sorting methods based on physical properties, such as magnetic and electrical properties, density, and melting point (see Chapter 10). Copper should be removed from solid steel scrap before it goes into solution. Promising techniques include cryogenic fragmentation of the scrap into small pieces followed by magnetic or visual separation. Dissolution of the solid copper into a sulfide matte, or even liquid aluminum, is another promising technique, but difficulties include sulfides or aluminum being attached to the scrap and high costs.

Separation methods based on gravity or density can be used to improve the quality of ore for DRI. Particles of different densities are separated in a medium of intermediate density, such as a suspension of fine heavy particles in water; in a solution of a salt, such as calcium chloride and various bromides; or a in a true "heavy liquid," such as a thallium malonate/formate solution. For separation of particles heavier than water, suspensions of fine solids can be used, such as magnetite and ferro-silicon. For separation of coarse particles, such as metals from the nonferrous product of shredded automobiles, relatively quiescent vessels can be used. For separation of finer particles, the application of a second force (e.g., centrifugal force) may be required to hasten the process and sharpen the density difference.

Inorganic membrane modules composed of cordierite honeycomb monolith coated with a microfiltration membrane (e.g., Al₂O₃) and vanadium-based catalysts could be used to remove SO₂ from flue gas. The use of oxygen, as opposed to air, for combustion could decrease gaseous emissions. The principal attraction of oxy-fuel is its intrinsic low NO_x attributes compared with conventional fuel-air burners. However, the relatively high cost of oxygen remains a significant barrier, and lower-cost, hot oxygen would greatly facilitate the implementation of oxy-fuel systems. In addition, the use of oxygen for combustion results in energy savings because it eliminates the need to heat the nitrogen component of air.

The separation needs identified for the steel industry are as follows:

• more efficient recovery of VOCs and better separation of fine particulates from gases in coke production
• better recovery and reuse of quench water in coke production
• separation processes for improving the chemical and physical quality of iron-bearing materials for blast furnace feed
• improved methods of removing impurities, particularly nitrogen and phosphorus, from steel in both oxygen and EAF steelmaking
• improved methods for handling, recycling, and recovering valuable materials from BOF dust
• improved methods for recovering metallic iron from BOF slag
• improved methods for handling EAF dust and technologies for partially recovering the zinc portion and recycling the iron-rich portion of the dust
• improved understanding of the mechanisms that separate inclusions from liquid steel, including transport of the inclusion to the metal-slag interface and dissolution or incorporation of the inclusion into the liquid slag
• better separation methods for dealing with copper and other residual impurities in scrap
• separation processes to produce very high grade DRI feeds
• better understanding of the behavior of fine particulates in gaseous or liquid media