The effect of vacuum melting on the elimination of gases and nonmetallic inclusions for six low-alloy steels was investigated. The purpose of these studies was the analysis and comparison of differences in gas content and inclusions in air-melted and vacuum-melted steel.

Data obtained from melts of 1-¼ to 20 pounds indicate that vacuum melting decreases the number and size of nonmetallic inclusions and at the same time increases the degree of dispersion of the remaining inclusions. These improvements are undoubtedly related to the elimination of approximately 80 percent of the nitrogen and 90 percent of the oxygen as well as most of the undesirable metallic trace impurities.

Impurities may be present in metals either dissolved or combined. Dissolved gases such as oxygen, nitrogen, and hydrogen may be removed by diffusion if their partial pressures above the melt are reduced. Residual graphite, carbide carbon, or interstitial carbon can react with dissolved oxygen and possibly some of the metallic oxides. Carbon monoxide thus formed is evolved during the initial degassing of the melt. Oxygen may be reduced further through reaction with hydrogen to form water vapor, which is subsequently removed from the system by the roughing pump. Perhaps the nitrogen content is largely diminished by the flushing action of the carbon monoxide boil and not just by the reduction in pressure above the melt. The solubility of hydrogen is proportional to the square root of the external hydrogen pressure down to 10 -4 millimeter. It can largely be removed from metals by reducing the pressure above the metal and holding the vacuum for sufficient time. Askoy states that oxygen increases the transition temperature in steel and forms various types of inclusions. Nitrogen affects aging, fatigue, and stress-rupture properties. Hydrogen causes embrittlement and flaking and decreases fatigue properties. Fatigue, impact, and ductility are strongly influenced by inclusions.

Vacuum melting is a relatively old process used in the refining of metals; however, only recently has it come into extensive use for refining large quantities of metals and alloys. Vacuum refining removes dissolved gases and volatile metallic impurities and prevents formation of nonmetallic impurities, particularly oxides.

Gaseous elements are retained in air-melted stock as impurities both in the free and combined state. In the free state, gaseous elements are located interstitially in the metal lattice and in intergranular pockets. The latter is far more deleterious to physical properties than the interstitial location. Oxygen forms nonmetallic inclusions when it combines with metals. The size, distribution, and number of nonmetallic inclusions, however, are difficult to control quantitatively in atmospheric refining, and the usefulness in controlling grain refinement is overcome by the discontinuities in the metal caused by these inclusions. The object of this investigation was to determine the amount of gas removed and the degree of inclusion reduction in steel from melting and casting at reduced pressures.

Laboratory Equipment

Vacuum refining of air-melted stock was conducted in both vacuum resistance and vacuum induction melting units. The steels listed in table 1 were chosen as typical low-alloy constructional steels that are normally produced in an open hearth. Silicon and sulfur contents deviate slightly from specifications in a few samples. Samples of these steels constituted the air-melted specimens used in this study and were the charges for the vacuum heats.

Melting and casting samples up to 1 ¼ pounds in weight was done in a vacuum-resistance furnace (fig. 1), consisting of a tilting crucible, tungsten heating element, molybdenum radiation shields, and water-cooled copper shell

assembly, which was enclosed in a stainless steel bell jar. Although the system was capable of pressures less than 1 micron (10 -3 mm. Hg) all vacuum resistance work reported was in a range of 5 to 10 microns.

Samples as much as 20 pounds in weight were melted and cast in a vacuum-induction furnace (fig. 2), consisting of a bottom-pour crucible with pouring valve assembly, a melting chamber that was constructed of shaped K-28 insulating brick, and induction coils that were mounted coaxially outside a stationary quartz tube. This system was evacuated by a two-stage roughing pump capable of reducing the system pressure to the range of 40 to 50 microns; however, all melting and casting was performed in a range of ½ to 1 millimeter.

Charges for both vacuum-resistance and vacuum-induction refining consisted of specially prepared stock that had been air-melted in an induction furnace. No additions were made to the melts during any of the vacuum heats to adjust the chemical composition. The vacuum-refining and pouring temperatures were within the usual steel-making range of 1,571° to 1,610° C. (2,860° to 2,930° F.), and the heats were cast into split ingot molds of machined graphite that had been painted with a stabilized zirconia wash. Cast in-gots for the vacuum-resistance heats were 1-inch-diameter by 5-inch cylinders. For the vacuum-induction heats they were truncated cones, 2-inch-diameter base by 13 inches with a taper of ½ inch in 13 inches. Initial vacuum-resistance heats were allowed to solidify in their melting crucibles. Button ingots from these heats were 2 1/8 inches in diameter by 1 ¼ inches.

Because the inclusions in cast steel are mainly globular in form and are not subjected to elongation deformation as in rolled and forged steel, a modification of ASTM designation, E 45-51 Microscopic Method A, for rolled and forged steel was used. Oxides in steel remain globular in shape after rolling or forging, but sulfide, silicate, and alumina inclusions are elongated to form either continuous or discontinuous stringers. In this modification, comparison charts for only the globular type were employed.

Gas analyses by the vacuum fusion method were made to determine oxygen and nitrogen. A literature search indicated that analysis for hydrogen was not necessary because of its high diffusion rate from steel. When other researchers have found comparable levels of oxygen (O2) and nitrogen (N2) in prepared vacuum melted steel, they have found the hydrogen content to be less than 2 parts per million. Condensable metallic vapors from the furnace roof and vacuum port were analyzed spectro-graphically for trace elements. Microetching with 3 percent nital proved to be the most satisfactory for revealing microstructural constituents.

Experimental Melting and Casting Procedure

Preliminary tests to determine the operating characteristics of the vacuum-resistance furnace proved that when heat was applied too quickly to the melt uncontrollable outgassing would splatter metal on the furnace roof, and the vacuum system pressure would rise above the range of capability (greater than 100 microns) of the diffusion pump. This problem was eliminated by holding the charge temperature just below the melting point for 20 minutes and then slowly raising the temperature above the melting point. Degassing occurred with a minimum of splattering when this procedure was used.

The study of ingot mold design during this initial phase of work revealed that one-piece, uncoated, machined graphite molds were inadequate because carbon pickup (0.15 to 0.84 percent) was excessive and ingots were difficult to remove. These problems were remedied by using a stabilized zirconia wash on the inside of the graphite mold and by using split molds which could be parted to remove ingots. Further experiments showed that piping in the ingots could be kept to a minimum by using a slightly tapered mold (big end up).

The knowledge gained with the vacuum-resistance furnace operation was employed in the design of the vacuum-induction unit and in subsequent heat campaigns. Before each heat, the insulating liner and crucible of the vacuum- induction unit were baked with an inductively coupled cylindrical graphite tube. The temperature in the melting compartment was gradually brought up to 1,500° C. in order to drive the moisture out of this section.

Some difficulties were encountered with the pouring mechanism in which the thermocouple assembly (fig. 3) included the pouring valve. Impervious mullite, magnesia, and recrystallized alumina tubes were used as combination pouring valve and thermocouple protection tube assemblies. Impervious mullite was the most satisfactory with respect to symmetry, seating, and thermal shock; recrystallized alumina might have been equally satisfactory had its shape been more nearly symmetrical. Magnesia was unsatisfactory because the valve fused to the crucible taphole seat during melting making it impossible to open the valve and cast the metal.

Progress of a vacuum-induction heat was indicated by the recorded time-temperature curve. Distinct arrests were noted in the curve as heat was absorbed when the solid charge passed through the alpha-to-gamma and gamma-to-delta thermal critical zones and a sharper arrest when heat was absorbed during melting. These points conformed with those found in source material thus checking the accuracy of the thermocouple temperature measurements.

Comparison of Air- and Vacuum-Melted Steel

Nonmetallic Inclusions

The prevention of nonmetallic inclusions in air-melted steel is a major problem, and quantitative control of the size, distribution, and type of inclusions is difficult to maintain at a fixed level in successive heats
and even in all ingots from the same heat. Vacuum refining of

steel will partially eliminate undesirable inclusions and increase the degree of dispersion of those remaining. Photomicrograph A (fig. 4) shows the original air-melted stock. The decrease in size and number of nonmetallic inclusions in vacuum-resistance and vacuum-induction melted steel is illustrated in photomicrographs B, C, and D.

In this study determining the inclusion content of samples requires a Jernkontoret Chart for comparing visual microscopic inclusions. The fields of the chart are numbered 1 to 5 in order of increasing frequency of inclusions. The globular-type oxide fields of the chart were compared with the fields for specimens under investigation but the types of inclusions were not differentiated. All fields viewed were recorded so that the resulting numerical inclusion data represent the averages of all fields.. The sole purpose of this tabularized count (table 2) is to show that vacuum melting (as opposed to air melting) decreases the number and size of the inclusions.

The increase in heavy inclusions of the cast vacuum-resistance heats over the other vacuum-resistance heats (melt solidified in melting crucible) was due to particles of zlrconia mold wash, which had been stripped from the mold wall by the stream of molten metal. The original steel (air-melted and cast) contained both stringer and globular type inclusions that were larger than any shown on the American Society for Testing Materials (ASTM) comparison chart. Because the presence of these large inclusions would not appreciably affect the relative evaluation of melting and casting methods, they were not included in the count. Similar large inclusions were not present in any of the ob served vacuum melted samples.

Gas Analysis

Oxygen and nitrogen in air- and vacuum-melted steel was determined by using the vacuum fusion technique (fig. 5). Duplicate, and in some cases, triplicate analyses were made for each steel sample used (table 3). In one case six analyses were made of a sample of steel that was also submitted to another laboratory for the purpose of checking the apparatus and analytical technique. Both sets of results agreed closely. A comparison of these data is shown in table 4.

Gas was removed principally by lowering the pressure above the melt. However inductive stirring also played a definite role. Oxygen removal by vacuum-induction melting at 0.5 to 1.0 millimeter pressure is essentially the same as vacuum-resistance melting at 0.005 to 0.01 millimeter pressure. Vacuum-induction melting virtually always removes slightly less nitrogen than vacuum-resistance melting. The high oxygen value in vacuum-resistance melted- and-cast 4017 steel was probably due to a reaction between melt and crucible or the melt and the coated mold. This heat boiled violently when it was tapped, and the ingot surface was poor. Usually vacuum refining removed more than 90 percent of the oxygen and more than 80 percent of the nitrogen.

Composition Changes During Vacuum Melting

Condensable Metallic Vapors

Although trace elements may amount to only a few parts per million, they can contribute to low ductility in metals and alloys. They may be introduced in many ways including the impurities in the smelted ore or subsequent alloying elements.

Vacuum melting removes the bulk of the volatile trace elements and unfortunately part of the more volatile alloying constituents as well. Spectrographic analyses were made of metallic powders condensed in the cooler sections of the vacuum-melting units (table 5). The three samples taken could be classified as coming from three temperature zones with respect to their proximity to the surface of the molten steel. The elements are recorded in order of decreasing quantities.

The high tin content of sample 2 indicates that a large portion of baled tin can scrap may have been used to make the plain carbon bar stock, which was the base of the air-induction steels.

Cleaning the vacuum-resistance unit after some heats revealed heavy magnesium deposits that were readily pyrophoric when disturbed. These deposits were noted particularly when blocks of stock metal adhered to the crucible walls above the melt and could be dislodged only by superheating the melt. Virtually all of this magnesium came from carbon reduction of the magnesia crucible.

Loss of Alloying Constituents

Some alloying constituents in steel are partly lost during a vacuum heat. They may enter into reaction either with constituents of the melt and refractory crucible, or they may be eliminated by vaporization. Carbon and manganese are removed to a greater degree than any of the other alloying elements present in steel. Carbon will enter into reaction with dissolved oxygen, with some of the less stable metallic oxides, and with the crucible refractory. In this work more carbon was removed in vacuum-resistance melting than in vacuum-induction melting (table 6). Photomicrographs of the etched steel specimens (fig. 6) show that the fine pearlite network in A has decreased considerably in the vacuum melted specimens B, C, and D with a resulting greater quantity of coarse pearlite and ferrite.

Consider the reaction of carbon with metal oxides,

C + MO → M + CO ↑,

where MO = metal oxides. If M were volatile it would vaporize and condense on cooler sections of the furnace. Otherwise on solidification of the melt M would substitute in the cubic ferrite or orthorhombic cementite lattice.

The manganese content decreases to a lesser extent in vacuum-induction melting than in vacuum-resistance melting (table 6). Pressure seems to be the greater controlling factor in manganese loss with less emphasis on inductive stirring.

Conclusions

The results of research on six low-alloy steels indicate that vacuum melting substantially refines the steel alloys through elimination of undesirable metallic and nonmetallic elements. As a consequence the following improvements are obtained:

1. Decrease in oxygen and nitrogen contents.

2. Decrease in the number and size of inclusions and increase in the degree of their dispersion.

3. Elimination of undesirable trace elements.