Table of Contents
- Procedure and Equipment
- Experimental Results and Discussion
- Single Stage Leaching
- Two-Stage Countercurrent Leaching
- Recovery of Manganese from Solution
- Proposed Leaching-Recovery Procedure
- Treatment of Tailings
Manganese serves two functions in the steelmaking process; it is an alloying agent improving such properties as strength, toughness and hardenability, and it is also used to control sulfur. Manganese controls sulfur by forming manganese sulfide instead of iron sulfide, which would make the steel brittle. The average manganese content for all types of steel is 8.7 lb/st or 0.44 pct. The United States has no high-grade manganese deposits (containing 35 pct or more manganese) and no economical methods of recovering manganese from its low-grade deposits exist. Consequently, there is no domestic production of manganese, and the United States is completely dependent upon foreign sources for manganese.
In 1987, the United States imported 340,000 st of manganese ore, 360,000 st of ferromanganese and 191,000 st of silicomanganese to produce 89.3 million st of steel. The principal sources of high-grade manganese imports are the USSR, Republic of South Africa, Brazil, and Gabon. The United States will continue to be dependent upon world resources unless domestic resources can be utilized. One such domestic manganese resource is BOF steelmaking slap.
The trend in iron and steel production is away from the older, less efficient open-hearth furnaces to the BOF furnace. Currently, about 60 pct of the steel manufactured in the United States is produced in BOF furnaces. By the 1990’s, BOF furnaces will account for over 80 pct of the production of steel from iron ore.
In the BOF process for making steel, hot metal from the blast furnace is directly charged to a steelmaking furnace where the carbon and silicon content of the melt is adjusted. Fluorspar (CaF2) is also added to the melt to control viscosity. A slag phase is formed on top of the molten steel by the addition of lime. Impurities are oxidized and report to the slag.
Approximately 40 pct of the slag is recycled to the blast furnace, but the remainder is discarded because of its high phosphorus content. Phosphorus is a highly corruptive substance in steel, increasing brittleness and lowering strength and corrosion resistance in concentrations of less than 0.05 pct. Attempts to remove phosphorus from BOF slags by slow crystallization and physical separation have been only minimally successful.
Discarded iron and steel slags are consumed mainly by the construction industry as substitutes for natural aggregates and other construction materials. Slags have been used in place of other materials because of their lower costs, superior performance for many applications, or shortages of natural aggregates. However, using slag as aggregate does not utilize the manganese values still present.
Discarded slag represents a loss of manganese value. It is estimated that 3.8 million st of BOF slag containing 133,000 st of manganese are discarded annually. BOF slags range in manganese content from 1 to 7 pct and average 3.5 pct. BOF slap are a possible domestic source of manganese the United States could use if economical recovery methods were found.
The U.S. Bureau of Mines examined manganese recovery from steelmaking slags as early as the 1950’s. The high silica content of open-hearth slap made them refractory to hydrometallurgical treatment. To liberate the manganese, the slag was clinkered with lime, reduced with hydrogen gas, and leached with ammonium carbamate solution. With the switch from the open-hearth furnace to the BOF for producing iron, a different type of slag exists today. BOF slags are composed primarily of lime, with very little silica present. Direct ammonium carbamate leaching of BOF slap, without clinkering, has met with only limited success. Since phosphorus removal from BOF slap and older methods of manganese recovery have met with limited success, a new process for recovering manganese from BOF slags would be advantageous.
This investigation studied hydrometallurgical methods to separate and recover manganese from BOF slags. The recovered manganese is to be used as a substitute for manganese ore fed into the blast furnace. Fluosilicic acid byproduct from the fertilizer industry was used as a leachant. The major advantage of this process is that a by-product, fluosilicic acid, is used to leach manganese from another byproduct, BOF slags. The primary concern was to maximize manganese recovery by optimizing such parameters as: agitation rate, temperature, fluosilicic acid content, pulp density, particle size, and number of leaching stages.
The slag samples tested were from four different BOF furnaces: high silicon (Ohio), silicon (Ohio), carbon (Ohio), and silicon (Pennsylvania). Each slag sample was ground to pass 100 mesh. The chemical compositions of the slag samples are presented in table 1. Different size fractions of minus 14 plus 28, minus 28 plus 48, minus 48 plus 100, minus 100 plus 200, minus 200 plus 400, and minus 400 of the silicon steel slag sample (Ohio) were prepared to study the effect of particle size in the extraction of manganese.
The slag samples were leached using fluosilicic acid (H2SiF6), a byproduct of the fertilizer industry. Fluosilicic acid is recovered when stack gases are scrubbed during the processing of phosphate fertilizers. A chemical profile of the acid byproduct used is presented in table 2. All other chemicals used were reagent grade.
Procedure and Equipment
Leaching was carried out in a 500-mL resin kettle glass reactor fitted with a reflux condenser and a thermocouple. The reactor was heated with a mantle controlled by an energy regulator and agitated by a 3- by ¾-in Teflon fluorcarbon polymer anchor glass shaft stirring assembly. Agitation speed was controlled by a voltage regulator.
A weighed amount of slag sample was placed in the reactor and the fluosilicic leach solution added while stirring. After completion of a leach, the slurry was immediately filtered through a 0.2-µm porosity paper filter. The slag tailings were washed thoroughly with deionized water and the filtrate stored in a Teflon fluorocarbon polymer container.
Experimental Results and Discussion
Single Stage Leaching
When slag is mixed with fluosilicic acid, calcium, iron magnesium and manganese are all extracted as fluosilicates;
MO + H2SiF6 → MSiF6+ H2O…………………………………………………..(A)
where M = Mn, Fe, Ca, and Mg. Selective extraction of manganese and iron is pH dependent. As fluosilicic acid is neutralized by lime contained in the slag, the pH of the solution rises and calcium and magnesium fluorides are precipitated:
MSiF6 + 2MO → 3MF2 + SiO2………………………………………………..(B)
where M = Ca and Mg. At a pH of 4.5, most of the calcium and magnesium have been precipitated. Manganese and iron remain in solution as soluble fluosilicate salts at pH 4.5.
To find the conditions that would yield maximum manganese extraction, tests were conducted in which agitation rate, temperature, acid content, mesh size, pulp density, and leaching time were varied. To assure that evaluation of variables was not influenced by slag composition, all single stage leaching tests were done using the silicon (Ohio) steel slag sample.
Effect of Agitation Rate
The effect of agitation rate on manganese extraction was studied using slag samples of minus 100 mesh size. A pulp density of 9.9 pct, an acid content of 0.94 g of H2SiF6 per gram of slag, a temperature of 95° C, and a leaching period of 5 h were used. Under these conditions, 693 pct of the manganese was extracted at an agitation rate of 160 rpm. Increases of the agitation rate to 320 and 640 rpm gave no increase in extraction. In subsequent experiments, an agitation rate of 320 rpm was used, and the above conditions were maintained while only one parameter at a time was varied.
Effect of Temperature
The effect of temperature over the range 50° to 95° C was investigated, and manganese extraction was plotted as a function of time (fig. 1). Manganese extraction rate increased as the temperature increased from 50° to 95° C, but total of extraction did not. At 50° C, extraction reached a maximum of 69 pct after 5 h at a final pH of 4.4, while at 95° C, this maximum and final pH were reached after only 3.5 h of leaching.
Temperature had a profound effect on calcium and magnesium concentration in the pregnant liquor. Initially, the pH of the slurry is low and calcium and magnesium are soluble. With time, fluosilicic acid is neutralized by lime contained in the BOF slap, and the pH of the leach solution increases, precipitating calcium and magnesium as insoluble fluorides. With an increase in temperature, the reaction rate of this neutralization is increased and calcium and magnesium are precipitated more rapidly and more completely. Since at higher temperatures both manganese extraction and calcium and magnesium precipitation rates were increased, 95° C was selected as the standard leaching temperature.
Fluosilicic Acid Requirements
The acid requirements for leaching manganese from silicon steel slag were evaluated. Leaching tests were performed using 0.75 to 1.32 g of H2SiF6 per gram of slag (66 to 117 pct of stoichiometric as defined by reactions A and B). From figure 2, increasing the acid content above 0.94 g of H2SiF6 per gram of slag (83 pct of stoichiometric) did not increase total manganese extraction. However, at higher acid content, the extraction rate increased.
Higher acid content resulted in higher concentrations of calcium and magnesium in the final pregnant liquor. As acid is consumed and the pH increases, calcium and magnesium fluorides are precipitated. In the tests that used 1.13 and 1.32 g of H2SiF6 per gram of slag (100 and 117 pct of stoichiometric), excess acid existed in the final pregnant liquor. Consequently, the pH of the final pregnant liquors was too low (2.2 and 1.8, respectively) to precipitate all of the calcium and magnesium. Since maximum manganese extraction and minimum calcium and magnesium concentration are desired, an acid content of 0.94 g of H2SiF6 per gram of slag was selected as a standard.
Effect of Pulp Density
Leaching tests were performed to determine the effect of pulp density on manganese extraction. Pulp densities of 8.8 to 13.2 pct were evaluated. At pulp densities of
13.2 and 11.3 pct, manganese was precipitated from solution during the leach (fig. 3). This phenomenon did not occur using pulp densities of 9.9 and 8.8 pct. The higher the pulp density, the more quickly calcium and magnesium were precipitated as fluorides. Since maximum manganese extraction and minimum calcium and magnesium solution concentrations are desired, a pulp density of 9.9 pct was chosen.
Effect of Particle Size
The effect of particle size was studied using minus 14 plus 28, minus 28 plus 48, minus 48 plus 100, minus 100 plus 200, minus 200 plus 400, and minus 400 mesh fractions. Leaching time was increased to 6.0 h to compensate for the slower dissolution of the larger mesh sizes. Results are shown in figure 4. There was a 9.2 pct increase in manganese extraction when comparing the minus 28 plus 48 to the minus 48 plus 100 mesh fractions. When silicon steel slag was ground finer than minus 48 plus 100 mesh, manganese extraction rate increased, but manganese extraction was not improved.
Fluosilicic acid was consumed more quickly when finer mesh material was used, due to the greater surface area of the finer material. This increase in the acid consumption rate caused calcium and magnesium to precipitate more rapidly. All further work was done on minus 100 mesh material.
Two-Stage Countercurrent Leaching
During single stage leaching, two distinct leaching steps were identified; solubilization of calcium, iron, magnesium, and manganese followed by precipitation of calcium and magnesium. To increase extraction of manganese and rejection of calcium and magnesium, a two-stage counter-current leach was investigated.
Silicon (Ohio) steel slag was leached countercurrently as shown in figure 5. Forty grams of slag ground to minus 100 mesh was placed in reactor 2 and mixed with pregnant liquor exiting reactor 1, It was leached at a pulp density of 10.7 pct, a temperature of 95° C, for a duration of 2.0 h. After liquid-solid separation, the partially leached slag was transported to reactor 1 where it was mixed with 325 mL of fresh fluosilicic acid containing 37.6 g (equivalent to 0.94 g of H2SiF6 per gram of slag) at a pulp density of 9.9 pct and leached at a temperature of 95° C, for a duration of 2.5 h. After filtration, the tailings were washed with 200 mL of deionized water. The wash water was recycled and used to dilute the fluosilicic acid to the desired concentration.
Calcium, magnesium, manganese, and iron were extracted from the slag in reactor 1 (fig. 6). Examining figure 6, manganese and iron extractions reached maximums of 75 and 70 pct, respectively, after 2.0 h of leaching. Precipitation of calcium and magnesium as fluorides
from the pregnant liquor reached a maximum after 2.5 h. At a final pH of 1.9, 10 pct of the calcium and 40 pct of the magnesium remained in the pregnant liquor. To assure reactions in reactor 1 are completed, a leaching time of 2.5 h is recommended.
In reactor 2, the pregnant liquor was mixed with sufficient slag containing residual lime to raise the pH to 4.5 (fig. 7). At pH 4.5, manganese and iron remained in solution as soluble fluosilicates, while the remaining calcium and magnesium were precipitated as insoluble fluorides. Calcium concentration in the pregnant liquor decreased until a minimum of 8 mg/L was reached after 2.0 h. Magnesium concentration reached a minimum of 16 mg/L after 2.0 h. Iron and manganese remained soluble in reactor 2, and their concentrations remained constant.
Best manganese and iron extractions for the single stage leach were 69 and 63 pct, respectively. The minimum calcium and magnesium concentrations obtained during the single stage leach were 290 and 460 mg/L, respectively. When a two-stage countercurrent leach was performed, manganese and iron extraction improved to 75 and 70 pct, while calcium and magnesium concentrations in the final pregnant liquor were decreased to 8 and 16 mg/L, respectively. Two-stage countercurrent leaching increased manganese and iron extraction and improved selectivity with respect to calcium and magnesium. Therefore, a two-stage countercurrent leach is recommended. Photomicrograph and X-ray diffraction analysis of the slag tailings did not reveal why 75 pct Mn extraction could not be exceeded.
To determine if two-stage countercurrent leaching would work on other BOF slags, tests were performed on the three other BOF slag samples available. Table 3 shows that the leaching technique was effective on all the BOF steelmaking slags tested.
Recovery of Manganese from Solution
Having demonstrated a satisfactory method for leaching BOF slags, efforts were directed toward methods for recovering manganese and iron from the pregnant liquor. Methods for precipitating manganese from leach liquors were investigated. Best results were obtained using sodium chloride and sodium carbonate and are discussed in the following section.
Recovery of Manganese Using Sodium Hydroxide
A simple method to recover manganese from the pregnant liquor would be the addition of a reagent, which would precipitate the manganese. Under basic conditions, Mn²+ forms an insoluble hydroxide Mn(OH)2. Precipitation tests were performed on pregnant liquor using sodium hydroxide.
Although sodium hydroxide was successful in preciptating manganese and iron from the pregnant liquor, a problem occurred. Sodium combined with fluosilicate forming an unwanted precipitate, sodium fluosilicate. Using sodium carbonate to precipitate manganese and iron from the pregnant liquor, also resulted in coprecipitation of sodium fluosilicate. Sodium hydroxide and sodium carbonate treatments were unsatisfactory since the manganese-iron product contained sodium fluosilicate, which cannot be recycled to the steelmaking blast furnace.
By adding a solution of sodium chloride to the pregnant liquor, sodium fluosilicate was selectively precipitated while manganese and iron remained in solution as soluble chlorides. The sodium fluosilicate was filtered from the pregnant liquor, and manganese and iron were precipitated as hydroxide-carbonate salts using sodium carbonate.
Recovery of Manganese Using Sodium Chloride and Sodium Carbonate
Since sodium fluosilicate is used in the water treatment industry for water fluoridation, the step to recover it might be a practical addition. To determine the minimum amount of sodium chloride required to precipitate sodium fluosilicate from the pregnant liquor, experiments were conducted using 1.0 to 1.5 times the stoichiometric requirement (fig. 8).
As shown in figure 8, 91 pct of the fluosilicate was precipitated when the stoichiometric amount of sodium chloride was used. A 25-pct excess was required to precipitate 99 pct of the fluosilicate. Addition of more sodium chloride did not precipitate more sodium fluosilicate.
To determine whether the precipitated sodium fluosilicate represented a viable byproduct, information was obtained from the American Water Works Association (AWWA) defining the criteria necessary for use in water fluoridation. There are five specifications that sodium fluosilicate must meet. The specifications are as follows:
- Chemical composition must be 98 pct Na2SiF6 by weight.
- Moisture content must not exceed 0.5 pct by weight.
- Insoluble matter shall not exceed 0.5 pct by weight.
- The particle size shall be such that at least 98 pct will pass a U.S. standard sieve No. 40 (420 µm) and less than 25 pct shall pass through a U.S. standard sieve No. 325 (44 µm).
- Heavy metals, expressed as lead, shall not exceed 500 ppm by weight.
Analysis of the sodium fluosilicate revealed a concentration of 770 ppm iron, which exceeds the maximum allowable tolerance for metals. To remove impurities, the sodium fluosilicate was washed with 0.5M HCl. Specifications 1, 2, 3, and 5 were achieved by the acid washed sodium fluosilicate. Size specification 4 was met after the sodium fluosilicate was agglomerated with 4 pet water and dried for 2 h at 95° C.
With fluosilicate removed, iron and manganese can be precipitated from the pregnant liquor using sodium carbonate. To determine the minimum amount of sodium carbonate required to precipitate manganese and iron, experiments were conducted using 1.0 to 1.5 times the stoichiometric requirement to precipitate manganese and iron as carbonates (fig. 9). At the stoichiometric amount of sodium carbonate, 95 pct of the manganese and iron were recovered as hydroxide-carbonate salts. A 20-pct excess of sodium carbonate was required to precipitate 99 pct of the manganese and iron.
Proposed Leaching-Recovery Procedure
Integrating the countercurrent leach scheme with sodium chloride treatment and manganese and iron precipitation, results in the flowsheet shown in figure 10.
Manganese and iron are selectively leached from BOF slag by countercurrent leaching using fluosilicic acid (reactors 1 and 2). Manganous and ferrous fluosilicates
contained in the pregnant liquor are converted into soluble chloride salts by adding a solution of sodium chloride (reactor 3). Sodium fluosilicate is precipitated and washed with 0.5M HCl. The acidic wash water is combined with the recycle stream from sodium carbonate precipitation. Manganese and iron are precipitated from the pregnant liquor by adding a solution of sodium carbonate (reactor 4). The barren solution exiting reactor 4 contains sodium chloride as well as residual sodium carbonate. When the barren stream and wash water from sodium fluosilicate washing are combined, excess sodium carbonate is converted to sodium chloride. The result is a sodium chloride stream with a pH of 6. This stream is concentrated by evaporation and recycled to reactor 3. By recycling the sodium chloride barren stream exiting reactor 4, only a small makeup amount of sodium chloride will be required. A material balance for the process is presented in table 4.
Most process streams in the integrated flowsheet are recycled. Only four streams (table 5) exit in the process; streams 3 (tailings), 23 (sodium fluosilicate), 27 (manganese and iron hydroxy carbonates), and 31 (excess water). Stream 23 is a sodium fluosilicate byproduct, which meets AWWA criteria for water fluoridation. Stream 27 contains manganese and iron, which can be recycled to the steelmaking blast furnace after the product is pelletized and
blended with iron ore. Stream 31 is water vapor from the evaporator, which can be condensed and used as a water source for the process. Only stream 3, the slag tailings, is a waste stream.
Note: In stream 31, all of these elements were <1 ppm.
Analysis of the slag tailings revealed that they contained 35 wt pct CaF2 and 28 wt pct SiO2. Calcium fluoride (fluorspar) has a market value if appropriate specifications can be met.
Treatment of Tailings
Acid grade fluorspar contains over 97 pct CaF2, while ceramic grade and steel industry require 90 pct and 60 to 85 pct CaF2, respectively. An attempt to concentrate the calcium fluoride by flotation was made using oleic acid dissolved in ethanol as a frother. Flotation of the slag tailings yielded disappointing results. Concentrate and tails were almost identical in chemical makeup, indicating separation did not occur.
A scanning electron microscope (SEM) analysis of the slag tailings, flotation concentrate, and flotation tailings revealed why separation did not occur. Calcium fluoride crystals measured 0.1 to 1.5 /an in size and were interlocked with silica. Rather than discrete particles of homogenous composition, each particle was a clump of different phases. Calcium fluoride appeared to be precipitated on siliceous material as if silica had acted as seed.
An attempt was made to increase the size of the calcium fluoride particles by adding fluorspar as seed during the leach. Fluorspar was ground to minus 400 mesh and added to a sample of minus 100 mesh slag. The sample was leached countercurrently as described previously. SEM analysis of the seeded tailings revealed no difference between the unseeded and seeded tailings.
Methods were investigated to recover manganese from BOF steelmaking slags. Manganese and iron in the slap were leached using fluosilicic acid, which is a byproduct of the fertilizer industry. Manganese and iron formed soluble manganous and ferrous fluosilicate, while calcium and magnesium formed insoluble fluorides. Best extraction and selectivity were achieved using a two-stage counter-current leach. Sodium fluosilicate was precipitated from the pregnant liquor and recovered as a byproduct using sodium chloride. Manganese and iron were precipitated as hydroxide-carbonate salts with the addition of sodium carbonate to the treated liquor. Total recoveries for manganese and iron were 73 to 83 and 67 to 70 pct, respectively.
The major advantage of this process is that a byproduct, fluosilicic acid, is combined with slag, another byproduct to form two potential products, manganese and iron hydroxide-carbonate salts, and sodium fluosilicate. Process waste streams are minimized resulting in only one waste stream, the slag tailings. If calcium fluoride could be recovered from the tailings, the tailings would also be a byproduct; however, no satisfactory technique was found.