The importance of ferrous scrap as a resource in the domestic iron and steel industry has increased significantly during the last two decades owing to the growth of steel production in electric arc furnaces. In 1984, approximately 55 million tons of scrap was consumed in domestic steel mills, of which 56 pct was used in electric furnaces, 28 pct in basic oxygen furnaces, 7 pct in open hearth furnaces, and 7 pct in blast furnaces. Scrap presently is the most important source of iron, with more of the iron used in steelmaking coming from scrap than from the ore-blast furnace- basic oxygen furnace route.
Steelmaking technology has changed considerably within the same timeframe, with substantial shift of steel production from large integrated producers to regional minimills. These highly productive and energy-efficient minimills use electric furnaces, scrap charges, and continuous casting technology to provide nearly a 4-to-1 primary energy advantage over integrated producers. The integrated steel producers, on the other hand, have more flexibility in that they can add pig iron, hot metal from blast furnaces, direct-reduced iron, or home scrap as a diluent for tramp elements in purchased scrap used in the charge mixture. However, it is not normal practice for electric furnace steelmakers to use hot metal or pig iron as a diluent.
The principal tramp elements of concern to producers of plain-carbon and low-alloy steels are Cr, Cu, Mo, Ni, and Sn, of which Cu presents the most problems. In a 1978 survey of the steel industry made by Luria Brothers and Co., Inc., Cleveland, OH, 71 pct of the 87 responding steel companies reported that Cu contamination in scrap caused them the most problems in steel processing. Cu and Sn cause hot shortness in steel by forming a liquid layer in the grain boundaries under the oxide surface.
Inland Steel Co. recently reported that purchased scrap usually contains 0.20 to 0.27 pct Cu, while most steel products must contain 0.15 pct or less Cu. In addition, Oregon Steel Mills, Portland, OR, reported that purchased scrap contains between 0.2 and 0.35 pct Cu. Cascade Rolling Mills, Inc., McMinnville, OR, has encountered problems with off- grade steel heats containing Cu levels of 0.5 pct and higher. Luria Brothers reports that the Cu problem in scrap has worsened since the firm’s last survey, published in 1978. Nucor Corp. reported that copper levels in purchased scrap have increased steadily during the past few years and presently are approaching 0.40 pct.
Several factors have promoted increased Cu levels in ferrous scrap. The current low prices for ferrous scrap and red metal have lessened the economic incentive to handpick the Cu in the form of electric motors, wiring, etc., from scrap such as that derived from obsolete automobiles. The increased usage of high-strength, low-alloy (HSLA) steels, which may contain up to 1.25 pct Cu, 1.80 pct Cr, and 5.25 pct Ni, eventually affects scrap quality. Another significant source that lowers the quality of ferrous scrap is the carbon steels that are purposely alloyed with 0.2 to 0.4 pct Cu, such as U.S. Steel COR-TEN.
Another factor that impacts on scrap quality is that many minimills have switched to higher quality steel products in order to compete favorably with imported steel. As a result, more individual steel mills are placing stricter specifications on purchased scrap used to produce high-quality steel products. This practice may eventually lead to shortages of premium grades of ferrous scrap and result in higher scrap prices.
The scrap processors currently use hand sorting, mechanical shredding, and magnetic separation techniques to separate the nonferrous metals, glass, and other non-iron-bearing materials from ferrous scrap. About 200 shredders process approximately 60 pct of the discarded automobiles to produce a product containing 0.2 pct Cu, 0.02 pct Sn, 0.16 pct Cr, and 0.04 pct Ni as the principal tramp elements. The various grades of processed scrap are summarized by Swager.
Another physical separation technique, known as cryogenic processing, is employed to a limited extent in Europe to produce a premium grade of scrap, but in the authors opinion, this practice appears to be too expensive for U.S. use.
Several physical and chemical techniques have been investigated on a laboratory and pilot scale to remove Cu from ferrous scrap. Evaporation of Cu from molten scrap under vacuum has been studied by several investigators. Thermal treatment and selective oxidation of Cu in scrap mixtures were investigated by Bureau researchers. Treatment techniques using either sulfide slags or molten Pb baths effectively lowered the Cu levels in molten, C-saturated iron, but several problems remain to be solved before the processes could be applied to molten steel.
Sulfide slags, which have been investigated extensively, are not effective in removing Cu from molten ferrous scrap having low C contents. The activity of both Cu and S in molten iron is enhanced significantly as the C level is increased to saturation. In addition, a large slag volume is required as Cu exhibits a relatively high activity in fused Na2S. In the case of a Pb bath to remove Cu from molten scrap, Langenberg reported the high vaporization rate of Pb from the bath is a major problem.
An alternate approach to physical-chemical treatment of ferrous scrap is neutralization of the effects of tramp elements such as Cu in steel. Copeland and coworkers at the Bureau of Mines found that additions of Al, B, Ni, and Si, to steel melts reduced or eliminated the hot shortness problem, depending on the Cu content. The effects of the steel additives on the structures and properties of the castings have not been established.
To be most useful, a process to remove tramp elements from ferrous scrap should be applicable to steel as well as cast irons, should not unduly extend the total refining time or create serious offgas or effluent problems, and must be effectively inexpensive. The ideal process would be pyrometallurgical in nature and would consist of simple partitioning of Cu into a slag that could be recycled following Cu recovery. In the present work, numerous pyrometallurgical processes were considered, both conceptually and experimentally, to remove Cu from molten steel scrap. The objective of this limited and somewhat qualitative test work was to evaluate some concepts that had not appeared previously in the literature. Positive results were needed to justify a future comprehensive study.
Experimental Work
Scrap charges comprising 25- to 30-kg Ingots of premelted shredded automobile scrap were melted within an alumina crucible in a 45-kg-capacity induction furnace. Materials to be evaluated were injected as fine particles below the surface of the molten scrap through a nominal 3/8-in (0.9525-cm-ID) Fe lance using air, O2-enriched air, or Ar carrier gas. Injection required about 10 min, during which time 2 to 4 m of Fe lance (1.2 to 2.4 kg) was melted into the steel. Following injection of the material, a sample of slag was recovered on a cold steel rod, and a spoon sample of metal was cast into a 2.54-in-diam by 2.54-cm-high alumina mold. The slag sample was pulverized, and any occluded metal was removed by a hand magnet prior to analysis by atomic absorption. The metal sample was ground flat on the bottom for X-ray emission analysis and drilled on the side to provide material for C and S analyses by the combustion method. Most of the difference between the initial and final Cu concentrations in the steel is the result of dilution of the steel by the Fe lance.
Results
A number of pyrometallurgical processes were evaluated to remove Cu from molten steel. The initial effort was designed to exploit the stability of Cu compounds under oxidizing conditions. These tests depend upon the formation of oxide species in regions of high O2 potential obtained by sparging with air or O2- enriched air. Numerous complex oxides containing Cu, as listed in table 1, were investigated using two techniques:
- Copper and the other metal in the complex oxide were dissolved in the steel, and air or air-50 vol pct O2 was sparged into the steel to effect oxidation. Slag and metal were removed immediately for analysis before general equllibrium between slag and metal was attained, because the complex oxides are thermodynamica1ly unstable in the presence of Fe.
- Copper was dissolved in the steel, and the metal oxide of interest was injected under the surface of the melt with air, air-50 pct O2, or Ar. Slag and metal again were removed immediately for analysis.
The complex oxides and their melting or decomposition temperatures are listed in table 1, and the results of the experiments are given in table 2. The formation of complex oxides containing Cu by gas injection of an existing oxide is mechanistically simpler than the multi-body reaction involving the oxidation of two metals in solution, which may explain the observation that injection of the oxides was generally much more effective than in situ oxidation. The oxides V2O5 and Cr2O3 were more effective than Al2O3 and MnO2. Note that the higher S content of the steel for the oxide injection tests, where comparisons can be made (Cr2O3, NiO, Fe2O3, and V2O5), resulted in less copper reporting to the slag. Copper removal is better for the low-S steel; however, for the method to be economically viable, the Cu level in the slag must be increased 20 to 80 times above the results obtained thus far, even in the best cases.
Numerous alkali silicates and complex alkali-alkaline earth silicates and aluminates were studied by injecting the finely divided material below the surface of the molten steel with air,
air-50-vol pct O2, or Ar. Justification for studying these materials lies in the stability of CuSiO3 and the propensity for Cu to substitute in complex alkali metal silicates. A case in point is the mineral litidionite (NaKCuSi4O10), which is a stable liquid phase at the melting point of Fe. The Fe analog of these silicates may also be stable, and in fact may be more stable than the Cu-containing compound. Hence, a local environment of high alkali or alkaline earth oxide concentration may be advantageous, as provided by injecting the silicates and aluminates below the surface of the molten metal. The results, given in table 3, are not encouraging, although some interesting observations are apparent. For example, based on slag analyses, 2K2O·SiO2 is 3 times more effective than 2Na2O·SiO2 and 10 times more effective than 2Li2O·SiO2.
The effectiveness of sulfides to remove Cu from C-saturated iron has been recently reaffirmed by Liu and Jeffes, who studied Na2S slags containing FeS and CaO. They concluded that 70 wt pct Na2S-30 wt pct FeS was the best
composition and that the addition of CaO improved the transfer of Cu to the slag while impeding the transfer of S to the metal. The present study extended their work to evaluate other metal oxides where the metal either forms very refractory sulfides (Ca, Ce, Mg, Y) or forms strong bonds with S but forms a sulfide with a lower melting point. The C content of the iron in the present study was greater than 1 pct, but significantly less than C saturation. The cover slag for the test series, a slightly basic CaO-SiO2, was present to retard the reversion of Cu.
The test results, given in table 4, are the most encouraging to date, with K2O producing 0.27 pct Cu in the slag and both the MnO2- and BaO-moderated slags producing 0.18 pct Cu in the cover slag. However, even in the best case, a 10- to 30-fold increase in Cu concentration in the cover slag is needed for a viable process.
Injection of CaC2 and Fe-Si-Mg was designed to exploit the stability of the intermetallic compounds Ca4Cu, CaCu5, and MgCu2. The latter is a very stable phase melting congruently at 815° C. The CaO- SiO2 cover slag was modified with K2CO3 or with K2SO4 in separate tests to evaluate the effects of K and S in the cover slag on the reversion of Cu to the metal phase.
The test results, given in table 5, indicate a negative effect for K in the cover slag and a definite positive effect for S in the cover slag. The last two tests In table 5 were designed to evaluate the coremoval of Cu and S from iron using CaC2 injection. Both tests indicate the effectiveness of CaC2 to desulfurize the metal and indicate an improvement in Cu removal. However, in all cases the Cu concentration in the slag is over 100 times too small for a viable process.
Summary and Conclusions
Numerous pyrometallurgical processes were considered both conceptually and experimentally to remove Cu from molten steel scrap. Over 60 semiquantitative tests were conducted in searching for a positive indication of Cu removal. The resulting information, if any of the tests had been successful, would have formed the basis for future research.
The most successful test produced nearly 0.3 pct Cu in a CaO-SiO2 slag using Ar injection of K2O-moderated Na2S-FeS below the surface of the molten steel. However, in the authors opinion, a viable process must provide at least 5 pct Cu in the slag, a level that seems unattainable by the methods employed in this work.