Demand for pressure vessel steels for use in power generation and energy transport applications is increasing in the United States. Coupled with the increased demand is the requirement for higher quality steels that are better able to withstand temperature extremes and mechanical stresses, and that possess greater tensile ductility, impact strengths, and isotropy.

One important method for achieving higher quality pressure vessel steels is to decrease the sulfur content. Low sulfur levels decrease inclusions, and therefore minimize lamellar tearing susceptibility, improve notch toughness, improve other mechanical properties, decrease surface cracking, and improve hot forgeability. These parameters affect both the required degree of conditioning and the yield. Accordingly, economic production of low-sulfur steels will minimize the requirements for mineral commodities by yielding material with longer service lives under adverse conditions, and hence efficient use of mineral resources will be realized. In addition, productivity will be maximized by decreasing the number of reject ingots and forgings, owing to anisotropy and cracking.

The objectives of the research reported herein include (1) testing the feasibility of preparing very low-sulfur steels (<0.005 wt-pct S [sulfur]) by the electroslag and electric-arc-furnace processes, (2) evaluating the possibility of eliminating an external desulfurization step in the electric-arc-furnace process to produce such steels, (3) comparing the quality of low-sulfur pressure vessel steels made by the two processes, and (4) estimating whether the extra cost for electroslag remelting is justified in terms of improved properties.

Four pressure vessel steel compositions were selected on the basis of commercial usage (16)2 and because a wide compositional variation was represented. Both the electric-arc-furnace and electroslag processes were used to produce ingots having the selected compositions. The electroslag process is a consumable-electrode technique whereby metal from the electrode is melted through a slag to form an ingot in a water-cooled crucible. The resulting ingots were forged and rolled, and mechanical test specimens were cut from the resulting bar. In addition, physical properties and composition were determined and the results were compared.

Numerous techniques were employed by others to desulfurize steel. Most of these involve external desulfurization of hot metal after it was removed from the blast furnace. Desulfurizers include calcium carbide; magnesium, either as powder embedded in coke (magnesium-coke), or mixed with steel turnings (DeSulf3); lime-magnesium; lime; sodium carbonate (soda ash); calcium cyanamide and sodium hydroxide. Calcium carbide and magnesium desulfurizers are most widely used commercially. Typical sulfur levels in the treated hot metal range from 0.010 to 0.060 wt-pct. Sulfur contents in the final steels average approximately 25 pct less, owing to the basicity of the slags used. Another desulfurization technique involves blowing calcium or magnesium into molten steel in a covered ladle using argon gas as a carrier. The molten metal is vacuum degassed and killed with aluminum prior to desulfurization. An average of 0.006 wt-pct sulfur is obtained in the steel. Tapping the contents of an electric-arc furnace into a ladle containing either calcium compounds or fluorspar and lime also was used for desulfurization. Minimum sulfur levels of 0.010 wt-pct can be expected. Fluorspar and lime additions do not effectively modify the sulfide inclusions, as does the calcium treatment. Other injected materials can include CaSi powder, as well as CaC2, to obtain sulfur levels near 0.012 wt-pct. Stirring techniques are often used during powder injection to improve reaction rates. Present applications require pressure vessel steels that possess mechanical and physical properties commensurate with sulfur levels lower than 0.005 wt-pct.

Suitable flux-metal reactions provided low sulfur contents of steels remelted by the electroslag process. A basic flux in the system CaF2-CaO-Al2O3 has been used by previous investigators to obtain sulfur levels as low as 0.002 wt-pct in common steels.

All of the aforementioned processes involve at least one step outside the primary melting furnace to desulfurize steel below approximately 0.010 wt pct sulfur.

Experimental

Materials

Preliminary tests in the electric-arc furnace used iron ore pellets prereduced by the Midrex process, shredded automobile scrap, a combination of scrap and Midrex pellets, or prereduced Midrex and FIOR materials. The charge material for the final melts consisted of iron ore briquets prereduced by the FIOR process. This material required the least adjustment in composition and the fewest additions to prepare suitable pressure vessel steels.

Experiments in the electroslag furnace used either prereduced Midrex iron ore pellets isostatically pressed into a suitable consumable electrode, FIOR briquets, iron ore pellets prereduced at the Twin Cities (Minn.) Metallurgy Research Center of the Bureau of Mines, mild steel bars, combinations of the above, or sections of ingots melted in the electric-arc furnace. Use of prereduced materials generally produced somewhat porous ingots with unacceptable sidewall surfaces. Although the FIOR briquets could be pressed to produce satisfactorily strong consumable electrodes, the electrical resistance was sufficiently high to prematurely heat the electrode, with a resulting loss of heat at the slag-electrode interface. Accordingly, the final compositions were electroslag remelted from electrodes consisting of mild steel bars, ingots melted in the electric-arc furnace, or combinations of both materials. Compositions of starting materials used in both processes are summarized in table 1.

Pebble lime containing 96 wt-pct CaO, and quartz containing 97 wt-pct SiO2 were used as slag constituents in the electric-arc-furnace tests. In addition, metallurgical-grade fluorspar was employed to condition the slage, and a variety of additions such as SiC, SiMn, FeMn, FeSi, Mn, Mo, and Ni were made to achieve desired steel compositions. These materials contained 0.07 wt-pct S or less. For the electroslag tests, fluxes in the system CaF3-CaO-Al2O3 were used exclusively.

Four compositions were employed. Those compositions were deemed sufficiently basic for adequate desulfurization, and possessed suitable low liquidus temperatures. The two most basic fluxes were used to prepare the final low-sulfur pressure vessel steel ingots for working and testing. The fluxes were prepared by blending preheated powders in the appropriate proportions and compacting the blended materials. The crushed compacts were then fused in a graphite resistor furnace at temperatures ranging from 1,235° to 1,375° C for 1 hour under 1/3 atm of helium. The fused mass was crushed, sampled, and stored under vacuum until used. Typical analyses of flux and slag constituents are given in table 2. A number of different desulfurizers were used in the course of this research. For the electric-arc-furnace tests, magnesium turnings, magnesium-coke, and calcium carbide were plunged into the bath for desulfurization. In later tests, the calcium carbide was rabbled into the bath. A basic slag served to improve desulfurization. Desulfurizers and/or deoxidizers used in the electroslag tests included magnesium turnings, calcium carbide, or misch metal. The magnesium turnings and misch metal were either pressed into the consumable electrode or added in mild steel tubes welded to the electrode. The calcium carbide was added to the flux. However, optium desulfurization was achieved by using only a basic flux in the electroslag melts. Compositions of the desulfurizers are given in table 3.

Equipment

Initial electric-arc-furnace tests were conducted in a single-phase ac tilting furnace with a nominal 227-kg (500-lb) capacity. The furnace was lined with magnesite brick, and the lid consisted of a rammed high-alumina refractory. The two graphite electrodes were 7.6 cm (3 in) in diameter, and were operated manually. The furnace was powered by a 1,000-kva transformer, with six voltage taps ranging from 100 to 215 volts. Subsequent tests were conducted in a nominal 0.9-metric-ton (1-ton) capacity tiltable, three-phase ac electric-arc furnace, lined with magnesite brick. The larger furnace was used so that a deeper bath could be achieved. Details of this furnace have been described previously. In the early tests, a graphite plunging bell was used to introduce the desulfurizing agents into the bath. The desulfurizer was contained in a stainless steel can inserted into a 10.5-cm ID (4-1/8-in) graphite cylinder, that was attached to a graphite holder equipped with a 5.4-cm-diameter (2-1/8-in), 183-cm-long (72-in) rod.

The electroslag furnace consisted of a 10.2-cm-diameter (4-in) copper crucible inserted into a water jacket, supported on a movable base suitable for making small adjustments to center the crucible during the test. The consumable electrode was supported by a water-cooled copper rod. The end of this rod was fitted with a worm gear that was motor driven. The feed rate was adjusted manually during the tests. Two single-phase ac transformers provided power to the furnace. Power input, electrode feed rate, amperage, and voltage were monitored during the melts.

As-cast and worked specimens were prepared and tested according to procedures recommended by the American Society for Testing and Materials (1). Dye-penetrant examinations to detect cracks, porosity, and/or other defects were made on room-temperature tensile specimens using a Magnaflux SKL-HF/SKL-S penetrant for at least 20 minutes, and a Magnaflux XP-9 Formula B developer for 15 to 20 minutes. Round specimens for tensile testing were machined to a gage length of 2.54 cm (1.0 in) and a standard diameter of 0.635 cm (0.250 in). The specimens were tested on a 60,000-lbf Baldwin Universal Test Machine at a strain rate of 0.05 cm/cm min (0.05 in/in min). The standard Charpy V-Notch specimens were 5.5-cm (2,165 in) long and 1.000 by 1.000 cm (0.394 by 0.394 in) in cross section. The centered notch was 0.200-cm (0.079 in) deep with an angle of 45° and a radius of 0.25 cm (0.10 in). Charpy specimens were tested on a Riehle impact tester model (R-21314) using a 0- to 220-ft-lbf-capacity range with a velocity at impact of 552 cm sec -1 (18.1 ft sec -1), Impact tests were conducted at room temperature, and at higher temperatures (50° to 240° C) using an isothermal (±5° C) bath of peanut oil. Reduced temperature tests were made using liquid nitrogen passed through a coil in an isothermal bath of ethyl alcohol.

Melting Procedures

A typical electric-arc-furnace test consisted of (1) a meltdown period, (2) a period during which the feed (prereduced iron ore) and slag constituents were continuously charged into the furnace, and (3) a refining period. When scrap was melted, the entire feed was charged into the furnace at one time. For tests in the 227-kg and 900-kg furnaces, approximately 14 and 25 wt-pct, respectively, of the total charge was initially melted down. Sufficient lime was included with the prereduced materials to provide a slag basicity ranging from 1.5 to 1.8. After meltdown, prereduced materials and lime were continuously charged from a belt feeder mounted above the furnace. This material was melted and then the slag was removed. This procedure provided better conditions for desulfurization than those prevailing when the initial slag was left in the furnace, as was the case in the early tests. Further treatment required well-rabbled additions of Ferrocarbo (SiC) for carburization and deoxidation, and/or FeSi, and/or SiMn for deoxidation. Slagging constitents were added to the bath prior to desulfurization. Calcium carbide was used as the desulfurizer in all of the later melts. This material was rabbled during the addition, and no plunger was employed. After the desulfurization reactions were deemed complete, the slag was removed, and suitable amounts of ferroalloys were added for deoxidation to prevent sulfur reversion and to bring the metal composition to within the desired specifications. The metal and remaining slag were tapped from the furnace into a heated side-pour ladle. After the slag had been decanted, the metal was poured into preheated 23-kg (50-lb) ingot molds during which aluminum pellets were used to kill the metal. The ingots averaged 9.5-cm (3.75-in) square by 33-cm (13-in) long. Considerable research was required to devise this optimum procedure. Earlier tests had involved desulfurization with magnesium turnings, magnesium coke, or calcium carbide in a graphite plunger without deoxidation. Metal sulfur levels ranging from 0.006 to 0.017 wt-pct were achieved when the deoxidation step was not used, whereas typical sulfur contents ranged from 0.002 to 0.004 wt-pct when the bath was deoxidized. Ladle desulfurization was not attempted, owing to the relatively small size of the equipment and the probability of premature freezing. Operating parameters for one electric-arc-furnace test are summarized in table 4.

Tests in the electroslag furnace were started by striking an arc between the electrode and a steel base plate, that was covered with 50 g (0.1 lb) of iron turnings and 300 g (0.66 lb) of flux; the fusing flux quenched the arc. Initial power was supplied by one transformer, and as soon as the flux was fused, the second transformer was turned on and the electrode began to consume. Fresh flux and suitable ferroalloys and/or desulfurizers were added to the furnace throughout the melt. When the electrode was approximately 75 to 90 pct consumed, the second transformer was shut off, and power output from the first transformer was decreased to minimize piping and porosity in the top of the ingot. The ingots typically measured 9.5 cm (3.75 in) in diameter by 23 cm (9 in) in length. Operating parameters for one electroslag remelt are given in table 5.

Ingot Testing

Both the electroslag and electric-arc-furnace ingots were forged and rolled according to commercial procedures. Prior to working, the ingots were radiographed with a 35-curie source for 55 minutes to determine the degree of porosity and piping. Two of the most sound ingots for each composition were selected for forging and rolling. Ingots prepared in the electric-arc furnace required surface conditioning and some grinding of remaining pits and slag inclusions. The electroslag-remelted ingots were prepared for forging and rolling by cutting off the base plate, cropping the tops of a few, and cambering the edges 0.6 cm (0.25 in) from the bottom and top. No surface conditioning was required. For each composition, one as-cast ingot prepared by both melting processes was sectioned longitudinally. Metallographic and analytical samples were taken from a 1.3-cm-thick (0.5-in) slice cut from the top of the ingot, and a similar slice cut 3.8 to 4.5 cm (1.5 to 1.75 in) from the bottom of the ingot. In addition, samples for tensile testing were obtained from the ingot, as illustrated in figure 1. The other half of the ingot was macroetched with a 4:1 HCl, (by volume) 30 pct H2O2 solution.

All of the prepared ingots were forged to a cross section of 8.9 by 8.9 cm (3.5 by 3.5 in). Ingots melted by both processes were worked using the same heating schedule for a given composition. Hot rolling was accomplished in several passes, depending on the desired bar thickness, using a 10-pct reduction-per-pass schedule. The schedule used for forging and rolling is

summarized in table 6. For the ASTM A537 composition, the values given in table 6 refer to a second set of ingots that were forged and rolled to thinner bar. Peculiar fracturing and erratic results on mechanical test specimens obtained from thicker bar (5.76 cm [2.27 in] thick), derived from ingots prepared previously, were believed due to insufficient working. Specimens for metallographic observation, and for mechanical testing were taken from the worked bar, according to the diagram given in figure 2. Impact tests were not conducted for the A285 ingots . This alloy is used primarily for fusion-welded pressure vessels, and impact strength specifications were not available. As noted in figure 2, tensile specimens were taken in two directions from near the two ends of the bar. For the remaining alloys, tensile specimens also were taken in orientations longitudinally and transverse to the rolling direction from near the two ends of the bar in accordance with ASTM A20 specifications, Charpy V-Notch impact specimens were taken with the notch facing in six directions near the end of the bar, corresponding to the bottom end of the ingot, and in three directions (CTL, CLT, CSL) at the other end. The orientations are illustrated in figure 2, and correspond to those used by Wilson.

Results

Electric-Arc-Furnace-Melted Ingots

Initial melts in the smaller electric-arc furnace in which calcium carbide and magnesium-coke were plunged into the bath, provided insufficient desulfurization to achieve the desired levels. This is illustrated in table 7 (tests 3 and 5). In tests 4-5, a more basic slag was used, and the slag was removed from the furnace prior to desulfurization. Desulfurization results of selected tests conducted in the larger furnace also are given in table 7 (tests 9-12, 17-18). Magnesium did not perform as well as calcium carbide, because it prematurely oxidized despite encasement in a plunging bell. In the case of higher alloy steels , greater amounts of ferroalloy additions, required to achieve the desired compositions, decreased the amount of calcium carbide necessary to attain the desired low sulfur levels , A deoxidized bath prior to the addition of desulfurizers is preferred. This can be accomplished by the addition of ferroalloys early in the test.

When sufficiently low sulfur levels were achieved, ingots having four pressure vessel steel compositions were prepared for evaluation (table 8). All of the steel compositions were within the tolerances allowed for the products.

Upon macroetching, all of the ingots melted in the electric-arc furnace displayed a band of equiaxed grains that ranged from 0.13 to 0.70 cm (0.05 to 0.28 in) in thickness around the periphery. Essentially, radially oriented columnar grains ranging in length from 1 to 2.5 cm (0.4 to 1 in) comprised the remainder of the ingot. These columnar grains impinge on a central core consisting of more equiaxed grains. The diameter of the central core is larger at the top and bottom of the ingots. Figure 3 illustrates a macroetched section of a low-sulfur pressure vessel steel ingot having the A533, type B (class 2) composition.

Metallographic examination of the as-cast ingots revealed a pearlite structure in a ferrite matrix for the A285, grade C, and A516, grade 65 compositions. More pearlite was present in samples taken from the ingot with A533, type B composition. The ingot having the A537, class 2 composition possessed lath-like ferrite grains in a matrix of pearlite. None of the metallographic samples exhibited any evidence of grain boundry precipitation of aluminum nitride. One sample taken from the ingot having the A285, grade C composition was subjected to examination on the electron microprobe. The results showed randomly distributed Al2O3 and AlN particles; however, no AlN inclusions were present at the grain boundaries.

Electroslag-Remelted Ingots

Early tests in the electroslag furnace used Midrex prereduced iron ore pellets pressed into bars to form the consumable electrodes, Although acceptable homogeneity and structures were realized, target ingot compositions could not be obtained

consistently. Typically, the carbon and/or sulfur contents were too high, and the Mn and Si levels were too low. Managanese recoveries ranged from 32 to 70 pct, and silicon recoveries were between 8 and 42 pct. When prereduced pellets were used as the electrode material, CaC2 was a more effective desulfurizer than magnesium impregnated coke, although carbide additions to the flux presented difficulties in completely melting the base of the ingot, and often resulted in rough sidewalls. Use of magnesium coke often resulted in pitting near the top of the ingot. Addition of increasing amounts of a deoxidizer (such as aluminum) as the melt progressed, ameliorated this problem to some extent. Misch metal additions were relatively ineffective. Prereduced iron ore briquets with low electrical conductivity could not be used as electrode feed stock, because the current flowed along the more conductive welds and the electrode melted apart in the weld areas. When prereduced iron ore pellets from several sources (table 1) were blended and pressed into a consumable electrode, the resulting ingots contained carbon levels within specifications, and the sulfur contents were lower. However, the ingots typically possessed rough sidewalls. Further desulfurization could be attained with CaC2 additions to the flux, but the melts were typically erratic. Smoother tests were achieved when the CaC2 was replaced with SiC, but desulfurization was less evident.

Because only small changes in composition can be obtained with electroslag remelting, the ingot to be remelted must have a composition very close to that of the desired product. Electrodes consisting of metal from several electric-arc-furnace tests with variable compositions were found to be most satisfactory, especially for low-alloy steels. In addition, ingot quality was related to the degree of deoxidation present in the electrode material. Oxygen levels above approximately 0.025 wt-pct should be avoided, as well as oxide inclusions, which should represent less than 1 wt-pct of the ingot mass. Sulfide inclusions are less harmful with respect to melting operations and ingot quality.

Although optimum desulfurization was achieved by using only a basic flux (compare tests 23-24 with test 25), electroslag remelting experiments were made to determine the relative effectiveness of common desulfurizers. These experiments showed that CaC2 is a good desulfurizer, but that calcium and magnesium metal are better. This result is illustrated in table 9 (compare tests 22 and 29 with test 27). Compositional control also could be achieved more readily with the metal desulfurizers. However, the melts were typically erratic, and considerable metal was distributed as prills in the slag. Sometimes, it was difficult to remove the ingot from the crucible. The degree of desulfurization at these low sulfur levels was relatively insensitive to the amount of desulfurizer added. This result appeared independent of the type of desulfurizer used, as shown in table 9 (compare tests 22 and 29; and tests 25 and 27). Better alloying efficiency was obtained when the desulfurizers and alloying elements were added at a controlled rate from crimped tubes welded to the side of the electrodes (compare tests 20-21 in table 9). Efficiencies also were improved when more reducing conditions were used and/or the electrode material contained either less oxygen and/or more residual aluminum.

As expected, a more basic flux provided better desulfurization during electroslag melting, although this has a somewhat higher liquidus temperature. A comparison of tests 23-24 in table 9 illustrates this result.

Sound ingots having the same four pressure vessel steel compositions as those produced in the electric-arc furnace were prepared by the electroslag process. Very low sulfur levels were achieved, and the ingots possessed smooth sidewalls, which required no conditioning prior to working (table 10). Typically, the oxygen and hydrogen contents of the electroslag-melted ingots were lower than those for the ingots melted in the electric-arc furnace. Nitrogen levels in ingots melted by both processes were essentially equal. Typical ingots melted by both processes are shown in figure 4.

The macrostructure of the electroslag-remelted ingots, consisted of large columnar grains up to 2.5 cm (1 in) in length emanating essentially radially from the outside toward a central core of equiaxed grains. Axially oriented columnar grains were visible in the bottom 5 cm (2 in) of the ingots.

Metallographic examination of samples from the ingots having A285, grade C; A516, grade 65; and A537, class 2 compositions revealed a lath-like ferrite structure in a pearlite matrix. For the A533, type B composition, a martensitic structure was observed. All of the samples examined possessed clean grain boundaries with no evidence of precipitates.

Forged and Rolled Bar

All of the prepared ingots were forged and rolled without difficulty. For the A285, grade C composition, bar rolled from material melted in the electric-arc furnace exhibited some surface roughness, and one end exhibited separation. Bar from electroslag-remelted steel had no visible defects. No surface flaws were observed on A516, grade 65 bar made from ingots melted by both processes. A small crack and small folds were visible on bar rolled from electric-arc-furnace-melted steel. No visible defects were noted on any of the A533, type B bars. For the A537, class 2 composition, bars produced from steel melted in the electric-arc furnace possessed a small crack at the end of each. One bar had a slight tear on one edge. Bars rolled from

electroslag-remelted steel exhibited one small end crack on each. Two examples of defect-free 5.7-cm-thick (2.2-in) bar rolled from ingots produced by both melting processes are shown in figure 5.

Microcleanness and Microporosity

Metallographic specimens taken from the bottom section of each as-cast ingot and the section of bar derived from the bottom of each ingot for each of the four compositions were examined for nonmetallic inclusions and microporosity. The number of globular oxide inclusions was determined at 100X and subsequently rated using ASTM Designation E45-76, Method D. Percentages of rated fields, which were equal to or less than a given rating for each composition in the as-cast and worked condition, were computed from the data (table 11) . The numerical ratings are proportional to the number of inclusions of each type observed in each field. For example, 4 pct of the fields surveyed in a specimen taken from an electroslag-remelted ingot having the A285, grade C composition had ≤3 type D inclusions, each <8 µm in diameter (DIT in table 11). Higher percentages at lower ratings signify a cleaner sample. Accordingly, the table shows that electroslag remelting improves microcleanness for all of the four compositions studied except for the thin inclusions for ASTM A537 steel. This result was confirmed in a compilation of worst field ratings. Other types of inclusions, such as sulfides, alumina, and silicates, were rarely observed in any of the specimens except as noted in table 11.

Identical procedures and analyses were conducted to determine microporosity. The results also are presented in table 11. In this case, electroslag remelting was less effective than electric-arc-furnace melting in minimizing microporosity, and the differences observed in specimens taken from metal melted by both processes often were minimal. The data indicate that electroslag remelting was more effective in decreasing microporosity only in the as-cast ASTM A533, type B material, whereas electric-arc-furnace melting was more effective in the as-cast ASTM A285, grade C and A537 steels, and in the worked ASTM A533, type B compositions. In all other cases, there appeared to be no significant differences in the two melting processes with respect to microporosity.

Mechanical Properties

A major difference in material melted by the two processes was expected to be the number, dispersion, and morphology of the nonmetallic inclusions. In addition, microstructural differences caused by the different melting and solidification sequences inherent in the two processes were anticipated. Because inclusions have their predominant effect on steel properties during ductile fracture, the percent reduction in area is considered the most useful measure of tensile ductility. However, the Charpy V-Notch upper shelf energy (value at 100 pct ductile fracture appearance) is a more critical test of the effects of inclusions, as well as processing variables, and chemical inhomogeneities.

Tensile Properties

Tensile properties were determined for the four alloys in the as cast condition, and are shown in table 12. Except for the A533, type B composition, electroslag-remelted ingots showed greater yield and tensile strengths than those melted in the electric-arc furnace. The opposite relationship was evident for ductility, although comparable values were obtained for both processes for the A516, grade 65 composition.

Data for worked bar, which are summarized in table 13, show good tensile strengths and ductility uniformity (isotropy) in both directions for material derived from both melting processes. Bars from electroslag-remelted steels were more isotropic with respect to ductility than those derived from arc-furnace melted steels, but the differences were not statistically significant at the 95 pct confidence level. However, the absolute values of all tensile strengths of the steel bars from electroslag-remelted material were significantly greater than those from electric-arc-furnace-melted steel. An exception was the ultimate tensile strength for electric-arc-furnace-melted A516 steel, which was significantly greater than the electroslag-remelted material. Reduction in area values for tensile specimens from the electroslag-remelted material were also either significantly greater than (A516 and A533B), or equivalent to (A285 and A537) those from the bars derived from electric-arc-furnace melting. Yield strength, ultimate tensile strength, and reduction in area values for the electroslag-remelted material were an average of 13, 8, and 9 pct greater, respectively, than the bars from electric-arc-furnace-melted material. No correlation was found between tensile property isotropy and ingot sulfur content.

Impact Properties

A summary of the most important Charpy V-Notch impact properties for three steels is tabulated in table 14. Transition curves for the four steels are shown in figures 6-11. The symbols on the curves correspond to those given in table 14. For all three steels , there was no statistically significant difference in the anisotropy of the upper shelf energy as a function of melting method. However, plates from the electroslag-melted steels showed a slight improvement in isotropy over the electric-arc-furnace-melted material. There was a significant difference in the absolute values of the upper shelf energy (100 pct ductile behavior) displayed in all three steels tested as a function of melting technique. Upper shelf energy values for the electroslag-remelted bars were ≥126 pct above those from bars derived from the corresponding steels melted in the electric-arc furnace, as noted in table 14 and figures 6-11.

The curves in figures 6-11 show that the upper shelf energies for the A516, grade 65 and A537, class 2 compositions met ASTM specifications within experimental uncertainties. No specifications were available for the A533, type B alloy.

Table 14 also gives data for the fracture appearance transition temperature (FATT) and 40 ft-lb transition temperature (TT). The lowest FATT and TT at 50 pet shear occurred in the higher alloy A533, type B material, followed by A537, class 2 and A516, grade 65 steels.

Tensile reduction in area values and Charpy V-Notch upper shelf energies indicate that there is no significant difference in isotropy of the four steels tested when the ingot sulfur content is less than 0.005 wt-pct.

Microscopy

Optical

Grain size determinations, in accordance with ASTM specification E112-77 at 100X on metallographic specimens cut from bar, gave a range from ASTM 7 to 8 for all but the A533, type B alloy. The grains were larger in the A533 material (ASTM 2 to 4 rating). For the A285 and A516 steel bars, a conventional ferrite-pearlite structure was observed. Tempered martensite was present in the A533 material, and in the A537 alloy bar derived from electroslag-remelted ingots. Apparent ferrite precipitation from the martensite characterized the A537 bar from electric-arc-furnace material.

Scanning Electron Microscopy

Fracture surfaces from tested Charpy V-Notch impact specimens were examined up to 10,000X by the scanning electron microscope. For ASTM A516, grade 65 steel, the primary failure mode was microvoid coalescence. There were no significant structural differences as a function of either the degree of fibrous fracture or specimen orientation with respect to the rolling direction. Material melted in the electric-arc furnace displayed large euhedral inclusions, suggestive of type III MnS morphology. Some hexagonal Al2O3 inclusions, and areas with stringer-like inclusions, suggestive of type II MnS morphology, also were present. In contrast, electroslag-remelted steel contained only a few small Al2O3-type inclusions.

Samples from ASTM A533, type B pressure vessel steel bars, derived from metal melted in the electric-arc furnace, exhibited only euhedral Al2O3 inclusions, with no evidence of sulfides. The same was true for electroslag-remelted material, which contained fewer inclusions, some of which were elongated into rectangular shapes. Microvoid coalescence was the failure mode, with some quasicleavage failure also observed.

For the ASTM A537 composition, microvoid coalescence was primarily responsible for failure. Some failure along cleavage planes also was noted. The material melted in the electric-arc furnace contained relatively few Al2O3 inclusions, and no sulfides were evident. The electroslag-remelted steel,which displayed a peculiar fracture surface and yielded anomalous impact test results, possessed unusual stringer-shaped Al2O3 inclusions. These inclusions were usually broken, as shown in figure 12. Spherical inclusions, ranging from 5 to 10 µm in diameter, also were present. These were generally mixtures of iron, aluminum, and manganese oxides. Samples cut from plates produced from electroslag remelts, which replicated the above tests and which had normal fracture surfaces, contained only a few hexagonal Al2O3 inclusions as well as a few rectangular inclusions similar to those found in electroslag-remelted ASTM A533, type B steel.

Discussions and Conclusions

One of the objectives of this research involved a comparison of the two melting methods used. During the melting operation, there were a number of inherent differences in the two processes that manifest themselves in ingot quality, composition, and mechanical properties.

It was more difficult to achieve a specific steel composition during electroslag remelting than it was in the electric-arc furnace because not all of the metal and slag are molten at any given time in the electroslag process, as is the case in the electric-arc-furnace process. Therefore, manipulations of metal and/or slag chemistry by changing the amounts of ferroalloy additions, for example, will not be as effective.

During electroslag-remelting operations to prepare ingots for testing, element efficiencies were signigicantly higher than those realized in electric-arc-furnace-melting operations. Efficiencies for carbon were near 100 pct during electroslag processing, and ranged from 24 to 44 pct for electric-arc- furnace melting. For silicon and manganese, efficiencies ranged from 5 to 20 pct and 17 to 83 pct, respectively, for electric-arc-furnace melting, compared with ranges of 61 to 100 pct and 77 to 100 pct, respectively, for electroslag remelting. The differences in ingot microporosity observed in this research might be attributed to inadequate hottopping procedures during electroslag processing.

Fluxes in the system CaF2-CaO-Al2O3 for use during electroslag remelting must be sufficiently basic for adequate desulfurization. This research has shown that at least 40 wt-pct CaO should be used to achieve acceptable sulfur levels in the ingots (table 9, tests 31-33). Quaternary fluxes containing MgO with sufficiently low liquidus temperatures also could be used, wherein the MgO could be substituted for some of the CaO to maintain flux basicity.

This research has demonstrated that deoxidation prior to desulfurization is required to obtain metal sulfur levels less than 0.005 wt-pct in the electric-arc furnace. In addition, prior deoxidation increases the recoveries of such elements as Si and Mn in the metal. Desulfurization reaction kinetics also are favored in a deoxidizing environment. The reaction mechanism has been described in three steps: (1) transfer of sulfur to the slag-metal interface, (2) reaction at the interface, and (3) transfer of products away from the system. The reaction at the slag-metal interface may be written as

at a specific temperature and slag composition. The activities of oxygen and sulfur in the metal are designated by a[0] and a[S] , respectively. Therefore, if a[0] is as low as possible, desulfurization reaction 1 will proceed more readily. Similar reasoning shows that more Si and/or Mn will report to the metal under deoxidizing conditions, and therefore the efficiencies will be greater. If the FeO content of the slag is regarded as a measure of deoxidation, then the reaction

2[Fe] + (SiO2) = [Si] + 2(FeO)……………………………..(3)

whose equilibrium constant at a specific temperature and slag composition is

will proceed more readily.

Differences in tensile ductility and Charpy V-Notch upper shelf energy isotropy were statistically insignificant in bar derived from both melting processes. Since bar derived from electroslag remelting generally contained fewer oxide inclusions, isotropic behavior cannot be correlated with such inclusions at ingot sulfur levels lower than 0.005 wt-pct. At higher sulfur levels, this correlation appears to be more evident in A533, type B steels, according to Wilson. The presence of clusters of oxide inclusions also could lead to a closer correlation. Such agglomerates were not evident in the bar prepared in this investigation.

Differences in values between the two melting processes for upper shelf energy could not be correlated with ingot sulfur content This observation also may be a function of the sulfur level in the metal. At very low levels, no correlation may be possible, owing to the absence of sulfide inclusions in bar prepared from ingots melted by both processes.

A comparison of the data from this work with relevant literature data is summarized in table 15. For the A516 steel composition, available data show that mechanical properties for the material from this investigation are slightly lower than those of Wilson. This may be due to somewhat less work in Bureau bars. The opposite result is evident for the tensile properties for A533, type B and A537, class 2 steels, although the tensile ductility and upper shelf energy is lower for Bureau material. Lower fracture appearance transition temperatures (FATT) (50 pct shear) also characterized Bureau bars for A533 steels, but not for A537 material. The lower values also may reflect the thinner bar used in this investigation. A thinner bar cools faster during quenching, and hence a smaller grain size would be expected. The higher FATT for the A537 steel in Bureau material may be related to the peculiar fracture surface exhibited by the impact specimens mentioned previously.

A comparison of the mechanical properties as a function of globular oxide inclusions (type D) shows that these oxides have a minimal effect on tensile strengths and FATT. These oxides may have an effect on tensile ductility and impact strengths, although differences in bar thicknesses between this work and that in the literature may have an overriding effect.

Relative energy, materials, and labor costs for producing low-sulfur pressure vessel steels by the electroslag and electric-arc-furnace processes were estimated, based on the experiences of this investigation. No attempt could be made to account for the differences in furnace sizes used in the two melting processes. The calculations showed that electroslag remelting was from two to three times more expensive for steel than was electric-arc-furnace melting.

In summary, this research has shown that four compositions of very-low—sulfur pressure vessel steels can be produced by the electroslag and electric-arc-furnace processes. Sulfur levels as low as 0.0008 and 0.0028 wt-pct were achieved by electroslag remelting and electric-arc-furnace melting, respectively. Calcium carbide rabbled into a deoxidized bath provide the most satisfactory desulfurizing conditions in the electric-arc furnace. A basic flux containing at least 40 wt-pct CaO provided the optimum desulfurization and ingot quality conditions during electroslag remelting. The electroslag process provided more restrictions with respect to major composition changes in the metal during melting. Essentially defect-free bar was produced from ingots melted by both processes. This investigation has shown that very— low-sulfur pressure vessel steels can be made in the electric-arc furnace without an external desulfurization step. Prior surface conditioning was required for material melted in the electric-arc furnace. Electroslag remelting generally improved microcleanness for the four steel compositions studied. This process was less effective in eliminating microporosity. Good tensile strengths, and ductility isotropy were realized for material melted by both processes. In general, electroslag remelting improved tensile strength and ductility. Ingot sulfur content could not be correlated with tensile property isotropy. Upper shelf energy isotropy was independent of the melting technique. However, values for upper shelf energies were higher for the electroslag-remelted material. At very low sulfur levels (<0.005 wt-pct), no correlation of upper shelf energy isotropy with ingot sulfur content was evident. Microvoid coalescence was the primary failure mode during impact testing. At very low—sulfur levels, globular oxide inclusions were not considered responsible for anisotropic behavior. Electroslag remelting was estimated to be two to three times more expensive for steel than was electric-arc-furnace melting. However, differences in furnace sizes and the quality of material were not evaluated in this estimate.

Bureau of Mines research has shown that very low-sulfur pressure vessel steels with improved mechanical properties can be prepared by two melting processes. Four pressure vessel steels having sulfur contents ranging from 0.0008 to 0.0040 wt-pct, were produced by the electric-arc furnace or by the electroslag processes. Calcium carbide was used as a desulfurizer in the electric-arc furnace. Fluxes containing at least 40 wt-pct CaO provided sufficient desulfurization during electroslag remelting. Element recoveries were higher for electroslag remelting than for metal melted in the electric-arc furnace. Bar rolled from electroslag-remelted ingots generally possessed higher tensile strengths and ductility than bar rolled from ingots melted in the electric-arc furnace. Charpy V-Notch upper shelf energies were also improved as a result of electroslag remelting. The primary failure mode during impact testing was was microvoid coalescence. Sulfide inclusions were minimal, especially in the electroslag-remelted material.