US3761246A - Process of preconditioning a chamber and deoxidizing a metal therein - Google Patents

Abstract

SECOND STAGE VESSEL WITH THE PRELIMINARILY DEOXIDIZED FERROUS METAL WHICH IS DEOXIDIZED WITH ALUMINUM TO OBTAIN A FINAL PRODUCT WITH 0.006% OR 0.002% OR LESS OXYGEN. THE PROCESS CAN BE PERFORMED CONTINUOUSLY AFTER PRECONDITIONING. WITH HYDROGEN AS A PRECONDITIONING AGENT AND A METAL HEEL CONTAINING LESS THAN 0.40% CARBON, AFTER PRECONDITIONING THE HEEL CAN BE DEOXIDIZED WITH ALUMINUM, WITHDRAWN AND COMPRISE PART OF THE FINAL PRODUCT.

A TWO STAGE PROCESS FOR PRODUCING FERROUS METALS SUBSTANTIALLY FREE OF OXYGEN-CONTAINING IMPURITIES BY PRELIMINARY DEOXIDATION IN A FIRST STAGE FOLLOWED BY SECOND STAGE DEOXIDATION WITH ALUMINUM IN A REFRACTORY LINED VESSEL. A MOLTEN FERROUS METAL HEEL IS SUPPLIED TO THE SECOND STAGE VESSEL AND THE VESSEL REFRACTORY LINING IS PRECONDITIONED BY DEOXIDATION WITH HYDROGEN AND/OR CARBON AND REMOVING REACTION PRODUCTS UNTIL THE AVAILABLE OXYGEN IS DETERMINED BY AN EQUILIBRATION TEST TO BELOW A LEVEL AT WHICH HERCYNITE FORMS ON SUBSEQUENT ADDITION OF ALUMINUM. THE METAL HEEL IS THEN REPLACED IN THE

Description

Sept. 25, 1973 w. E. MAHIN ETAL 3,761,246

PROCESS OF PRECONDITIONING A CHAMBER AND DEOXIDIZING A METAL THEREIN Filed March 16, 1972 5 Sheets-Sheet 1 POUNDS PER T 0N ALUMINUM ADDED 0.2 0.4 0.5 0.6 L0 12 1.4 L6 L8 20 2.2 2.4 2.6 28 3.0

PE RCENT' OXYGEN PERCENT ALUMINUM -T1 Q J 1 0 o SS+LIQUID\ Fe A|\2 3 A1223 7 FeO -A120;-; 55- 800 LIQUID o '7L/ "0 A1203 ss 3 I600 Y n: E' 1 0 ss 0 Pea/[203 Fe0-A1203 ss 1 I I400 I L u0u|0+ Fe f reos Fec ss I 200 Fe0'iA123 Feo 20 40 60 00 A1203 INVENTORS WILLIAM E. MAHIN NORMAN PARLEE ATTORNEY 7 1g '57 BY 5 Sheets-Sheet 3 w. E. MAHIN ET AL A METAL THEREIN PROCESS OF PRECONDITIONING A CHAMBER AND DEOXIDIZING Filed March 16, 1972 Sept. 25, 1913 QQQm Se t. 25,1973 7 w. E. MAHIN ETAL PROCESS OF PRECONDITIONING A CHAMBER AND DEOXIDIZING A METAL THEREIN 5 Sheets-Sheet 4 Filed March 16, 1972 Sept. 25, 1973 w, E, MAH|N ET AL 3,761,246

PROCESS OF PRECONDITIONING A CHAMBER AND DEOXIDIZING A METAL THEHEIN Filed March 16, 1972 5 Sheets-Sheet 5 N 0mm United States Patent 3,761,246 PROCESS OF PRECONDITIONING A CHAMBER AND DEOXIDIZING A METAL THEREIN William E. Mahin, Ashland, Greg, and Norman A. D. Parlee, Los Altos Hills, Calif., assignors to Kaiser Industries Corporation, Oakland, Calif. Continuation-impart of application Ser. No. 65,742, Aug. 20, 1970, which is a continuation of application Ser. No. 658,162, Aug. 3, 1967, both now abandoned. This application Mar. 16, 1972, Ser. No. 235,237

Int. Cl. C21c 7/10 US. Cl. 75-46 20 Claims ABSTRACT OF THE DISCLOSURE A two stage process for producing ferrous metals substantially free of oxygen-containing impurities by preliminary deoxidation in a first stage followed by second stage deoxidation with aluminum in a refractory lined vessel. A molten ferrous metal heel is supplied to the second stage vessel and the vessel refractory lining is preconditioned by deoxidation with hydrogen and/or carbon and removing reaction products until the available oxygen is determined by an equilibration test to be below a level at which hercynite forms on subsequent addition of aluminum. The metal heel is then replaced in the second stage vessel with the preliminarily deoxidized ferrous metal which is deoxidized with aluminum to obtain a final product with 0.006% or 0.002% or less oxygen. The process can be performed continuously after preconditioning. With hydrogen as a preconditioning agent and a metal heel containing less than 0.40% carbon, after preconditioning the heel can be deoxidized with aluminum, withdrawn and comprise part of the final product.

CROSS REFERENCE TO RELATED APPLICATIONS The present application is a continuation-in-part of copending US. application Ser. No. 65,742, filed Aug. 20, 1970, which in turn is a continuation of Ser. No. 658,162, filed Aug. 3, 1967, both now abandoned.

BACKGROUND OF THE INVENTION This invention relates to a process for producing very clean steels in which aluminum is added to obtain extremely low oxygen contents.

It has long been known that metals, particularly ferrous metals, when melted in contact with air or other oxygencontaining atmospheres, absorb oxygen to cause several undesired conditions which alfect further processing of these metals and reduce the quality of the metal product produced from such metal. Examples of harmful effects of oxygen in rolled steel products are oxide streaks, nonmetallic inclusions and blow holes formed near the surface of the ingot where they burst open during rolling to produce seams.

The most common practice heretofore in dealing with absonbed oxygen has been to introduce into the metal one or more metallic elements such as manganese, silicon or aluminum which, though present in relatively small amounts, have a greater affinity for oxygen than the base metal and form oxides which tend to float into a separate slag layer. In the manufacture of low carbon steels (e.g., for rolled strip and sheet), aluminum is a preferred de- 3,761,246 Patented Sept. 25, 1973 oxidizer because of its high afiinity for oxygen. Thus, small amounts of aluminum deoxidizer may be employed to obtain a low level of residual aluminum (e.g. 0.10% or less). At this low level, aluminum does not seriously harden the steel and does provide improved ductility. However, aluminum addition by prior art methods tends to form hercynite oxide clusters in the surface layers of the cast ingot or slabs and which, unless removed by expensive surface conditioning methods, normally become rolled into the finished product. Such clusters are generally 8 to 20 mm. in diameter in the as-cast state and are composed of many microscopic particles which tend to collect in large numbers and become entrapped near the surface of a slab or ingot. They cause scabs and seams which impair the quality of the rolled product.

Three major sources of such hercynite oxide clusters are: (1) hercynite oxides formed when aluminum is added to the liquid metal containing oxygen and which are not removed prior to the casting of the metal, (2) hercynite oxides formed when oxygen from the environment reenters the previously deoxidized liquid metal containing aluminum, and (3) hercynite oxides formed during solidification of the metal.

The words hercynite and hercynite-type oxide as used herein mean the compound FeO-Al O or Eco-A1 0 containing some manganese oxide (MnO), or FeO-Al O associated with some aluminum oxide (A1 0 containing iron oxide (FeO) in solid solution in a range of about 0.0 to about 3.5% or combinations of any or all of the above or A1 0 in combination with other oxides of elements (such as silicon, chromium, vanadium, molybdenum or nickel) similar to FeO or MnO, the combination of which, while in contact with liquid steel, tends to increase the oxygen activity of the liquid steel to a level over and above that which would exist in the steel when the steel is in contact only with the compound A1 0 Compounds of various compositions may exist and can be described generally by the formula (Z),,(Al 0 where Z is one or more other oxides as mentioned above.

As used herein, the term hercynite changeover point refers to the oxygen level at which molten ferrous metal containing dissolved aluminum can be in equilibrium with both hercynite and alumina (A1 0 containing not more than about 3.5% Fe in solid solution) as stable oxides. At oxygen levels greater than the changeover point, hercynite is the stable oxide form, while at oxygen levels less than the changeover point A1 0 is the stable form although hercynite may also exist metasta'bly under various conditions. This has been determined to be controlled for iron in the absence of manganese by the following equation:

At 1600 C., the value is 0.058% oxygen.

As used herein, the term environmental control refers to a family of processes for improved metal making in a controlled atmosphere environment.

As used herein, the term ceramic alloy refers to a combination of two or more ceramic oxides in solid solution in each other, or in one or more solid solutions combined with one or more chemical compounds composed of one or more metallic oxides.

As used herein, the term heel refers to either a body of metal placed in a furnace vessel specifically for vessel pretreatment purposes, or a body of metal left in a vessel after a portion has been poured out either to act as a susceptor for induction heating or as a reservoir to contain dissolved deoxidizer, or for some other purpose.

As used herein, the term innocuous gas refers to a gas which may include hydrogen and which is essentially free of available oxygen.

As used herein, the term available oxygen refers to oxygen capable of combining with the molten metal or dissolved deoxidizer (e.g. aluminum or carbon) or gaseous deoxidizer (e.g. hydrogen or carbon monoxide) whether it is oxygen from the gaseous environment, dissolved oxygen or chemically combined oxygen capable of being released for purposes of such combination.

The process herein described may be applied in conjunction with any metal-casting procedure through which liquid ferrous metal is converted into a solid metal prodnot. However, this process is particularly useful and in many cases essential for the so-called continuous casting of low carbon steels containing aluminum where it is believed that a method for control of oxygen at a lower and more uniform level in the liquid metal has heretofore been a needed but unavailable improvement.

SUMMARY OF THE INVENTION AND OBJECTS It is a general object of the present invention to provide a method for producing liquid steels and employing aluminum as a deoxidizer which avoids the presence of hercynite in the final stages.

It is a further object of the invention to produce steels which are essentially free from surface oxide defects, non-metallic oxide inclusions or other defects attributable to oxygen.

It is another object of the invention to provide a method for continuous casting of low carbon steels having the above high-quality characteristics which, due to the use of aluminum as a deoxidizer, is capable of refining the grain structure and preventing aging caused by dissolved nitrogen.

It is a further object of the invention to provide a process which substantially improves the life of the refractory linings of the employed vessels.

These and other objects and features of the present invention will appear from the following description taken in conjunction with the accompanying drawings.

The present invention is predicated upon the discovery that in current practice the refractory linings of vessels for containing liquid ferrous metals tend to be contaminated by large quantities of oxygen from sources such as (a) sorption from prior melts or the gaseous atmosphere, or (b) moisture present in the refractories. The oxygen may be present in such forms as FeO which tends to form solid solutions or solid compounds rich in available oxygen. Such contamination becomes a reservoir of oxygen for absorption by subsequent ferrous metals contained in the same vessel. With aluminum employed as a deoxidizer, it has been found that the oxygen contamination in conventional refractory walls is sufficient to cause the formation of hercynite resulting in the aforementioned disadvantages. In accordance with the present invention, it has been discovered that the refractory linings of the vessel may be preconditioned with a deoxidizing agent prior to addition of aluminum so that, during the deoxidation, the oxygen level of the vessel is maintained below that at which hercynite will form in the liquid melt or in the solid steel during casting.

In accordance with the present invention, a steel with a carbon content of no greater than 0.40% carbon is produced substantially free of oxygen-containing impurities by two stage deoxidation carried out in first and second refractory lined vessels. For pretreatment of the refractory lining of the second stage vessel, a charge of a molten ferrous metal heel preferably having an oxygen concentration of about 0.015% or less is provided for the vl either by supplying in liquid form or by in situ melting. In one embodiment, hydrogen is employed as the deoxidizing agent. It is circulated in an innocuous gas stream through that vessel and reacts with oxygen in the metal and refractory lining to form water vapor until the refractory lining is preconditioned.

One method for determining refractory preconditioning is an equilibration test in which a hydrogen-containing atmosphere within the vessel is sealed at operating temperature (e.g. 1600 C.) and about one atmosphere pressure for one day with molten ferrous metal containing less than 0.0137% oxygen and less than 0.02% carbon. If the volumetric ratio of water vapor to hydrogen concentration (pH O/pH- in the gas within the vessel is no greater than about 0.05 following the equilibration test, the lining is sufficiently preconditioned so that the available oxygen content is at a level below that which will produce hercynite during the subsequent addition of aluminum as a deoxidant. This 0.05 value corresponds to a dissolved oxygen content in the molten metal of 0.0l37%, safely below the hercynite changeover points (0.058% in the absence of manganese and other elements which affect the changeover point, and 0.018% in the presence of about 0.3% manganese). It should be understood that the equilibration test describes the refractory lining condition but forms no part of the present process per se.

In another equilibration test, a molten ferrous metal heel containing about 0.20% carbon and no more than 0.010% oxygen is sealed in the vessel at an operating temperature of about 1600" C. and an initial carbon monoxide partial pressure of about one atmosphere for one day. If the carbon level following the equilibration test has undergone no substantial decrease, this is indicative that the refractory has been preconditioned to an extent that there is insufficient available oxygen in the vessel to cause the formation of hercynite upon the addition of aluminum.

A body of molten ferrous metal with a carbon content no greater than 0.40% is the metal to be treated (hereinafter the subject molten ferrous metal). The process is particularly beneficial for treating a low carbon (e.g. less than 0.15% carbon) aluminum-containing ferrous metal. It is subjected to a preliminary deoxidation in a first stage vessel so as to lower its dissolved oxygen level to less than 0.015% oxygen. The molten ferrous metal heel is withdrawn from the second stage vessel while maintaining the pHgO/pH ratio at not greater than 0.05. The subject molten ferrous metal charge is deoxidized in that vessel by the addition of a stoichiometric excess of aluminum to obtain a deoxidized product containing not more than 0.006% oxygen. Finally, the deoxidized metal is cast. Prior to casting, the hydrogen in the vessel is removed until it constitutes not more than about 4% by volume at one atmosphere total pressure so as to avoid complications described hereinafter.

In employing the above hydrogen preconditioning process for continuous or intermittently continuous casting, the preliminarily deoxidized subject molten ferrous metal is continuously or intermittently fed to the second vessel (after it has been preconditioned) and withdrawn therefrom after deoxidation to a level of not more than 0.006% oxygen while maintaining the atmosphere of the vessel at a hydrogen level of 4% by volume or less at one atmosphere total pressure and a ratio of water vapor to hydrogen of not more than 0.02. In this manner, the refractory lining may be preconditioned by the passage of an innocuous gas containing a relatively high proportion of hydrogen (e.g. 40 to 50% up to by volume or more) as for a duration of about 5-22 hours while maintaining a pH O/pH ratio of no greater than 0.008 in contact with the metal. Thereafter the hydrogen level can be reduced to 4% or less, thus avoiding the formation of bubbles in the steel during casting. The ability to precondition at a high hydrogen level followed by reuction to 4% or ess for conti uous p ation the eafter is an important economic feature of the present invention since it would take an excessive amount of time to precondition the refractory while maintaining the hydrogen at a desired final concentration to avoid bubbles (4% or less by volume of hydrogen at one atmosphere total pressure).

In another embodiment of the present invention, the refractory lining may be preconditioned by maintaining a high carbon content (e.g. at least 1.5%) in the molten ferrous metal heel. The carbon may serve as the sole deoxidizing agent or may be employed in combination with hydrogen. It may be initially present in the heel fed to the second stage vessel and, in any event, would normally be supplemented by addition of the carbon in pulverulent form as the initial carbon is consumed. The carbon reacts with the available oxygen in the refractory lining and molten ferrous metal heel to form carbon monoxide and carbon dioxide as gaseous reaction products. These products are removed as by venting, and, if necessary, swept away with an innocuous gas stream until the refractory is preconditioned.

The process as outlined above is particularly applicable to the treatment of low carbon, aluminum-containing steels. Low carbon steels are those steels containing not more than 0.15% carbon and usually less than 0.12% carbon. If the metal is solidified after the second stage deoxidation, it is underpoured from the vessel through a passageway into the mold without being exposed to available oxygen. Some of the aluminum may be added to the stream as the metal is introduced into the second stage vessel. A preferred procedure is to have at least some of the aluminum present in or added to the vessel prior to the introduction of the preliminarily deoxidized molten metal thereto.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a series of graphs plotting the percent oxygen dissolved in iron against the dissolved aluminum content in the metal at various temperatures.

FIG. 2 is a phase diagram for the system iron oxidealuminum oxide.

'FIG. 3 is derived from FIG. 1 showing the oxygenaluminum relationships in iron at 1600 C.

'FIG. 4 is a further expanded portion of FIG. 3 showing aluminum-oxygen-iron relationships more clearly.

FIG. 5 is also a further expanded portion of FIG. 3 showing aluminum-oxygen-iron relationships more clearly.

'FIG. 6 is a graph of water vapor content in the total gas leaving the cabinet as a function of time.

FIG. 7 is a graph of the ratio of partial pressure of Water vapor to the partial pressure of hydrogen in the portion of the gas having contacted the liquid metal and refractory as a function of time.

FIG. 8 is a series of graphs of the percent oxygen versus the percent aluminum present in the final product.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The overall process to which the improvement of the present invention is directed is generally described in U.S. Pat. 3,467,167, incorporated herein by reference. It includes the steps of (a) partially deoxidizing a charge of molten ferrous metal in a first stage vessel and (b) completing the deoxidation in a second stage vessel utilizing aluminum as the deoxidizer. A very high purity final steel is produced by maintaining the second stage and final casting operations in a hercynite-free state. To accomplish this, it has been discovered that there is a relatively large quantity of available oxygen in conventional refractory linings which must be lowered, in a preconditioning step, to a value below the hercynite changeover point prior to the addition of aluminum to the second stage vessel.

In the above process, maintenance of the second stage and final casting operation in a hercynite-free state is accomplished by the use of gaseous hydrogen, carbon, carbon monoxide, or combinations thereof as preconditioning agents. The system will be first described completely with reference to hydrogen.

In a first step, a heel of molten ferrous metal is provided in the refractory-lined vessel employed for second stage deoxidation, hereinafter the second stage vessel. The temperature of the vessel is maintained above the melting :point of the ferrous metal, typically at 1600 C. Thus, the molten ferrous metal heel may be supplied in either the molten state or in the solid state with subsequent melting.

In hydrogen treatment of the refractory lining of the second stage vessel, a hydrogen-containing innocuous gas stream is circulated through the vessel. This stream contacts both the surface of the first molten metal, which preferably should be well stirred, and the refractory lining. The available oxygen in the refractory lining from such sources as contained water or FeO along with a portion of the oxygen content in the liquid ferrous metal reacts with the hydrogen to form water vapor. The formed water vapor is removed from the second stage vessel in an exit gas stream which, for economy, may be recycled to the second stage vessel after the removal of water vapor therefrom. This process is continued until the refractory lining is fully preconditioned as indicated by the formation of an innocuous gas atmosphere in the vessel which is indicative of a sufficently low level of impuritie in the lining to avoid the formation of hercynite upon the subsequent addition of aluminum.

One technique which is suitable for determining whether or not the refractory lining is preconditioned is by carrying out an equilibration test. A hydrogen-gas-containing atmosphere is sealed within the vessel, containing molten ferrous metal with less than about 0.0l37% oxygen and less than about 0.02% carbon, at a typical deoxidation temperature (e.g. 1600 C.) for a full day. Thereafter, the desired degree of preconditioning is indicated by a volumetric ratio of water vapor to hydrogen concentration (pH O/pH of no greater than about 0.05 measured at the interface of the liquid metal. This may be approximated by measuring the exit gas stream. This value is in equilibrium with a dissolved oxygen level of 0.0137% in the molten ferrous metal. This is less than the preferred maximum value of 0.015% for the first stage deoxidation and is chosen to be sufficiently below the hercynite changeover point of 0.018% oxygen for a ferrous metal containing about 0.3% manganese.

Another suitable equilibration test is described hereinafter with respect to preconditioning with carbon.

The above refractory lining preconditioning is complete for a typical refractory lining by pretreatment with an innocuous gas stream containing at least 40-50% hydrogen in from 5-8 hours while maintaining a pH O/pH ratio of no greater than 0.008 in the gas in contact with the metal. For certain highly contaminated refractories, hydrogen pretreatment may require 22 hours or more.

The operating parameters to obtain the aforementioned degree of preconditioning are dependent upon a number of factors such as the degree of contamination of the initial refractory lining and of the molten ferrous metal heel. In general, the hydrogen-containing innocuous gas may be of the same type as described in U.S. Pat. 3,516,819 issued to the present inventors. At the beginning, an inert gas such as argon is normally employed to flush the vessel. Thereafter, hydrogen is swept through the vessel in proportions up to hydrogen. It is economically impor tant to accomplish refractory pretreatment in a comparatively short period of time, such as on the order of 5 to 22 hours. For this purpose a hydrogen gas concentration of at least 40 to 50% is employed with recycling of the gas stream after removal of water vapor.

Subject molten ferrous metal is treated by preliminary deoxidation in the first stage prior to being supplied t the second stage vessel pretreated as above-described. The carbon level of subject molten ferrous metal is on the order of 0.15 or 0.12 or less percent in Order to best utilize the process of the present invention. But, the process can be beneficial to steels of considerably higher carbon (e.g. up to 0.40% carbon).

A two-stage process in which aluminum is employed as a deoxidizer in both the first and second stages is illustrated generally in Mahin Pat. 3,467,167. As shown in that patent, it is best to underpour the ferrous metal from the first stage while removing oxides that float on the surface of the metal in that stage. In order to avoid hercynite formation in the second stage, it is desirable to deoxidize the subject molten ferrous metal in the first stage so as to lower its dissolved oxygen content to less than 0.015% oxygen prior to second stage introduction. This low oxygen level is substantially less than the hercynite changeover point of 0.058% oxygen, in the absence of manganese and other elements which affect the changeover point, and 0.018% in the presence of about 0.3% manganese, at a typical operating temperature of 1600 C. By employing aluminum as a deoxidizer for the subject molten ferrous metal, insoluble hercynite formed in the first stage is sep arated and withdrawn as by skimming from the surface of the metal and underpouring to the second stage.

In the next step, the molten ferrous metal heel is Withdrawn from the second stage vessel and replaced with the subject molten ferrous metal which has been deoxidized in the first stage. During this transfer of metal, the atmosphere in the second stage vessel is characterized by a pH O/pH ratio of no greater than 0.05. Otherwise, it would be necessary to further deoxidize the vessel at some time prior to the addition of aluminum in order to avoid the formation of hercynite.

For final deoxidation according to the present invention, an excess of aluminum over the quantity stoichiometrically required to convert all oxygen remaining in the metal to aluminum oxide is preferably added either simultaneously with the subject molten ferrous metal or to the remaining portion of the metal heel. The invention is particularly applicable to reducing the oxygen content in the final ferrous metal to 0.006% or less and preferable as low as 0.002% or even 0.001%. For deoxidation on this order, aluminum in at least 25% excess may be added. A typical residual aluminum concentration in the cast steel for the above oxygen levels may range from about 0.04 to 0.08%.

During the final period of second deoxidation to obtain a ferrous metal containing 0.006% by weight or less of oxygen, an atmosphere of gas in contact with the low carbon ferrous metal is maintained in which the pH O/pH is no greater than about 0.02. correspondingly, for a final metal product of 0.002% or less oxygen, the pH O/pH ratio should be no greater than 0.008.

After deoxidation in the second stage vessel, the subject molten ferrous metal may be underpoured to a mold for continuous or batch casting without exposure to an available oxygen-containing atmosphere. A protective atmosphere during casting is necessary to avoid hercynite formation. As a safety precaution, against oxygen leakage into the system through the refractory, a superatmospheric pressure may be employed within the system.

During second stage deoxidation of the subject molten ferrous metal, it may be desirable to continuously sweep from the vessel any formed gaseous reaction products with innocuous gas containing hydrogen. It should be understood that the amount of water vapor produced as a reaction product decreases substantially as the oxygen content in the steel approaches the final low desired level. When the water vapor production has decreased to a relatively small amount, it is possible to seal the vessel and remove it intermittently, if desired, rather than continuing to circulate the gas.

Low carbon steels having very low oxygen levels are obtained according to the present invention by avoiding the presence of hercynite in the metal or in the refractories. For this purpose, the low carbon steel removed from the second stage vessel contains no greater than 0.006% oxygen, substantially less oxygen during casting than the 0.015% maximum oxygen level obtained from the first stage vessel in which hercynite conditions may exist. This oxygen content is employed because the casting process itself normally causes a segregation of oxygen and aluminum in the liquid core of the partially frozen ingot or slab so that the percent oxygen in the remaining liquid may become higher than the hercynite changeover point. This could cause precipitation of hercynite, particularly at the lower portion of the core, and formation of clusters in the mold which tend to float to the surface. To avoid this, it is preferred that the dissolved oxygen content of the poured steel not exceed about 0.003%, for steel containing only small quantities of manganese, or 0.001% for steel containing about 0.30% manganese.

Prior to removal of a batch of steel from the second stage, it is important to maintain the hydrogen content at not more than about 4% by volume hydrogen (preferably about 3.2% by volume or less) whereby excess hydrogen is removed from the liquid steel. This avoids any harmful effects of hydrogen such as brittleness or bubbles as fully described in the aforementioned US. Pat. 3,516, 819 which deals with hydrogen deoxidation of liquid steel.

The two-stage process of the present invention in which hydrogen is employed as the deoxidant for preconditioning the refractory lining is particularly well suited to continuous casting operations. After the aforementioned treatment of the refractory lining in the second vessel to a sufiicient extent to avoid the presence of hercynite, subject molten ferrous metal which has been deoxidized to an oxygen level of 0.015 or less, may be continuously fed to the second stage vessel and withdrawn therefrom after a deoxidation to a level of 0.006% or less oxygen. Continuity of the operation could be accomplished by employing two preliminary deoxidation vessels with alternate feeds to the second stage vessel. During continuous or intermittently continuous operation of this type, the second stage vessel is maintained at a hydrogen level of four percent or less to avoid the aforementioned harmful effects of a high hydrogen content in the steel. The ability to maintain the system at such a low hydrogen level after pretreatment is predicated upon the discovery that it is unnecessary to repeat initial pretreatment by flushing with a relatively high concentration of hydrogen gas for an extended period of time for each new quantity of metal to be deoxidized. That is, after a single treatment of the refractory lining, subject molten ferrous metal having an oxygen level of 0.015% or less may be continuously fed to the second stage vessel for an extended period of time while maintaining an innocuous gas atmosphere.

The molten ferrous metal heel employed during the initial pretreatment of the refractory lining in the second stage vessel may either be discarded or itself cast to form a final product. A molten metal heel comprising a high carbon steel, in which the carbon functions as a refractory lining preconditioner as described hereinafter, cannot, of course, be employed to form a final product of low carbon steel. On the other hand, a ferrous metal having a carbon content of 0.40% or less may be subject to preliminary deoxidation as in a first stage of the above-described type, may be employed as the molten metal heel. After deoxidizing of such metal to an oxygen level of 0.006% by weight or less and by lowering the hydrogen content in the gaseous atmosphere to 4% or less, the metal may be underpoured into a mold to form a final cast steel of the above high quality.

The preceding discussion relates to the employment of hydrogen as the deoxidizing agent for preconditioning of the refractory lining of the seco d s g vessel. In another embodiment, carbon may be employed as the sole deoxidizing agent or in conjunction with hydrogen. For this purpose, carbon in an amount from 1.0 or 1.5% to 4% is either initially present in the molten ferrous metal heel supplied to the second stage vessel or is subsequently added to the metal in pulverulent form. The carbon reacts with available oxygen in the refractory lining and with the molten ferrous metal heel to form carbon monoxide and carbon dioxide as gaseous reaction products. These products are removed as by venting and, if necessary, swept away with an innocuous gas stream until the refractory is preconditioned.

The following equilibration test is preferably employed to determine preconditioning of the refractory lining for carbon as the deoxidant. A molten ferrous metal heel, containing about 0.20% carbon and no more than 0.01% oxygen, is sealed in the vessel for one day at a temperature in the vicinity of 1600 C. and an initial carbon monoxide partial pressure of one atmosphere. A total pressure of about one atmosphere or more should be maintained in the vessel to avoid the infusion of gas from the surrounding environment. For this purpose, make-up inert gas, such as argon, may be added, if necessary. If there has been no substantial decrease in the carbon level, this is indicative that the refractory is preconditioned so that there is insufficient available oxygen to cause the formation of hercynite upon the subsequent addition of aluminum.

To precondition the refractory lining to a desired extent as indicated by an equilibration test of the above type, the partial pressure of carbon dioxide to carbon monoxide (pCO /pCO) should be no greater than 0.01 following the equilibration test. This corresponds to a dissolved oxygen content in the molten metal at equilibrium of about 0.011% which is below the hercynite changeover point.

The equilibration test described with respect to hydrogen preconditioning may also be employed for carbon preconditioning.

In order to achieve preconditioning of the refractory lining, the reaction products are removed from the second stage vessel as by sweeping with an innocuous gas stream. If it is desired to recycle the gas stream for economy, carbon dioxide should be removed, particularly if the pCO /pCO ratio exceeds the desired value. However, carbon in the ferrous metal heel may be suflicient to produce ample CO so as to lower the ratio to the desired value by venting of the gas produced and thus it would not be necessary to use an innocuous gas such as argon nor to remove CO by chemical means.

It should be noted that a high carbon content in the molten metal heel may be employed in conjunction with a certain amount of hydrogen in the innocuous gas stream. If so, a method of reducing the carbon dioxide present in a recirculated gas stream is by the removal of water vapor therefrom. This results in driving the following equation to the right:

An independent source of gaseous carbon monoxide may also be added to the system as a supplementary deoxidizing agent. This, of course, would tend to reduce the pCO /pC ratio.

Carbon is preferably present in the molten ferrous metal heel in a quantity of about 1.5 to 4%. Although it is possible to employ 1% carbon, there is insuflicient carbon at levels below about 1.5 to fully deoxidize the refractory lining with a relatively short residence time in the absence of the addition of other deoxidizing agents (hydrogen or carbon monoxide). If the carbon content exceeds about 4%, a certain amount of violent reaction and attack on the surface of a new refractory lining may result.

After preconditioning of the refractory lining, the high carbon containing molten ferrous metal heel is replaced with a subject molten ferrous metal (containing less than 0.40% carbon) which has been preliminarily deoxidized in the first stage to lowerits dissolved oxygen content to les than 0.015% oxygen. During this transfer of metal, a pCO /pCO of no greater than about 0.01 is maintained in the gaseous atmosphere within second stage vessel. This value approximately corresponds to the aforementioned pH O/pH ratio of 0.05 and also to an equilibrium dissolved oxygen content slightly less than that in the entering low carbon ferrous metal of 0.015%. These values have been chosen to be sufiiciently below the hercynite changeover point as to avoid the presence of hercynite in the second stage vessel.

In the next step, the subject molten ferrous metal in the second stage vessel is deoxidized by the addition of an excess of aluminum over the stoichiometrically required quantity to convert to A1 0 all oxygen remaining in the metal to obtain a metal containing 0.006% by weight or less of oxygen. During the final period of deoxidation to obtain this product, an atmosphere of gas in contact with metal is maintained characterized by a pOO /pCO of no greater than about 0.005. For a final metal product of less than 0.002% oxygen, a pCO /pCO of less than 0.002 is maintained.

It should be understood that specific pH O/pH ratios and pCO /pCO ratios directly correspond to an equilibrium value for the dissolved oxygen content in the molten ferrous metal. The choice of which ratio is to be employed as an indicator of the dissolved oxygen in the system is dependent upon the presence of sufiicient amounts of hydrogen and water vapor of carbon monoxide and carbon dioxide in the system for convenient measurement.

Following deoxidation, the final subject molten ferrous metal is withdrawn from the second stage vessel and cast in the manner described above.

The hercynite equilibrium relationship provides a basis for the aforementioned maximum preferred oxygen contents in the steel. It has been shown that under certain ideal conditions the reaction between dissolved aluminum and dissolved oxygen in steel produces A1 0 instead of hercynite and that the aluminum disposes of the oxygen present in the metal effectively and for all practical purposes, completely. Although such ideal conditions have been obtained heretofore in small scale laboratory experiments, it has not been known heretofore that such ideal conditions could be produced in a large scale commercial steel process so that the presence of hercynite and its undesirable effects could be avoided in the manufacture of solid steels.

Reference should now be made to FIG. 1, in which curves (1) and (2), in conforming to the relationship Log [(percent Alp-(percent Q) represent hercynite-type oxide formation in liquid iron containing oxygen and aluminum, and curves (3) and (4) represent A1 0 formation in liquid iron containing oxygen and aluminum. In each case, the relationship is shown between dissolved aluminum (Q) and dissolved oxygen (Q) in the liquid metal at the temperature indicated.

In FIG. 1 a horizontal line has been drawn at a level of 0.065% Q which is about the amount of oxygen in solution normally found in an undeoxidized low carbon steel of about 0.06% carbon. Lines A and B have been drawn to show what happens when the process of this invention is carried out and a total of 1.5 and 2.0 pounds per ton of aluminum respectively are added to such a steel, ignoring any aluminum which might be consumed in actual practice by reaction with the vessel environment such as the gaseous atmosphere or refractories.

Beginning at S on line A, 1.5 pounds per ton of aluminum are added to metal at 1600 C. and this results in a precipitation of hercynite until the aluminum and oxygen remaining in solution are at the point S and at this point the metal appears to be substantially in equilibrium with solid hercynite at 1600 C.

It may be noted that at S the liquid steel contains an appreciable amount of aluminum and oxygen in solution at 1600 C. and most of the oxygen would normally form hercynite during further cooling and freezing which would result in oxide clusters, nonmetallic inclusions or oxide streaks in rolled products.

Under normal operating conditions where the liquid metal continues to be held in contact with hercynite, the relationship between oxygen and aluminum will follow curve (2) and only a comparatively modest amount of further deoxidation can be achieved even by adding considerably more aluminum.

In the present invention as described above, during the deoxidation in the second stage vessel, the composition of the final product generally follows curve (4) of FIG. 1. Thus additional aluminum (e.g., 0.5 pound per ton of iron) is introduced at S and the composition momentarily shifts to L. After further deoxidation the composition shifts to L' with about 90% or more of the oxygen remaining after the hercynite step being precipitated as A1 and the resulting deoxidation is much more complete.

Further explanation of the method should commence with an understanding of the FeO-Al O equilibrium diagram which is illustrated in FIG. 2. At the right-hand side of this diagram, it may be noted that at a temperature of 1600 C. there is a solid solution containing up to about 3.5% FeO in A1 0 If there is more than about 3.5% FeO in such a ceramic alloy, the excess will be entirely found in the form of hercynite until FeO exceeds about 41% of the ceramic alloy. It follows, therefore, that the refractory vessel lined with A1 0 in contact wth liquid metal must contain less than about 3.5% FeO in order to avoid the presence of hercynite in the refractory.

Oxygen contents in pure liquid Fe of 0.058% at 1600 C., which is reported to be the hercynite changeover point at this temperature, tend to be in equilibrium with solid hercynite at a temperature of 1600 C. and may be in equilibrium with both hercynite and the solid solution of A1 0 containing 3.5% FeO. Thus a refractory lining, which initially was pure A1 0 of a vessel holding liquid Fe containing slightly more than 0.058% Q would tend to be converted below the liquid level to hercynite and saturated solid solution of A1 0 containing 3.5 FeO. The relative amounts of these two phases might vary up and down so long as the total of the two equals 100%. At times when more than 0.058% O is present, the relative amount of the hercynite would increase while at times when there is less than 0.058% Q, the relative amount would decrease.

FIG. 1, curve (2) shows a relationship between Q and g which extends below 0.058% Q. Curve (4) also shows an ;A l, Q relationship which overlaps that of curve (2) for the same A l contents. As will be demonstrated, the overlap can be explained through a careful study of the ceramic alloy diagram of FIG. 2. In FIG. 3, curve (4) of FIG. 1 has been extended upward from about E to B which is the changeover point. In FIG. 3, lines ALO-O and HERO represent the ratios of oxygen to aluminum in the compounds A1 0 and FeO-Al O respectively. Precipitation of A1 0 and FeO-Al O from liquid Fe would follow lines parallel to these lines.

As shown in FIG. 3, along the curved line A-B there is only one relationship between A} and Q, i.e., added Q or Q forms FeO-Al O Beginning at point B there can be two relationships. Below point B in an ideal system where the environment is completely innocuous, i.e., neither the atmosphere nor the refractories (for example, of pure A1 0 nor any solids suspended in the metal are able to contribute oxygen to the metal, addition of aluminum theoretically would cause the Q to drop along the line B-E and there would be no FeO-Al O formed by the reaction.

Now, instead of beginning at point B in such an ideal system, consider the situation where one starts at point A. Here the vessel tends to become coated or infiltrated with the presence of FeO-Al O by absorbing some Q and Fe from the metal. Also, any aluminum added reacts with Q to form FeO-Al O which remains as suspended solid matter floating in contact with the metal or adhering to the vessel walls. So upon reaching the point B, the system has become contaminated with hercynite and a supply of available oxygen and when further Al is added at B, the Al, Q relationship tends ot follow the curve BCD.

For example, at point S aluminum added to point T would precipitate more hercynite along the line T-W, as shown in FIG. 4. If there were nucleation of pure A1 0 in such a system, it would quickly be converted over to hercynite by the excess Fe and Q that are always available from within the system. So long as there is any appreciable amount of hercynite present in the system, the relationship between aluminum and oxygen tends to follow the curve A-B-C-D even though the atomsphere itself may contribute no available oxygen.

To understand the nature of the relationship occurring along the line B-E, let us further consider the earlier statement that at B the 3.5% FeO solid solution is in equilibrium with hercynite and also with liquid Fe con taining 0.058% Q at 1600 C. On the other hand, not far below point B, nearly pure A1 0 is in equilibrium with liquid Fe containing of the order of, say, 0.0025 Q in the liquid Fe. It follows that between B and B there is a Whole series of solid solutions of FeO and A1 0 in equilibrium with the metal and the right-hand scale of FIG. 3 shows the theoretically calculated composition of these.

Now, let us consider what happens when we have such a solid solution as shown at point U and add aluminum to point V (see FIG. 5). At this point, initial precipitation is of the solid solution containing FeO at the U level and as precipitation continues down along the line V-X, the concentration of FeO in the precipitate decreases to the X level. Meanwhile, the refractory walls tend to remain at the U level of FeO until diffusion can occur. As diffusion occurs, the concentration of the system tends to reach equilibrium at point Y, which varies up and down depending upon the relative amounts of the liquid metal and of the solid refractories.

The end result of continuing to add aluminum to a system in the B-E range is to convert the oxygen in the metal and in the refractory to precipitated aluminum-oxide which ultimately is nearly pure at the point E. However, this high purity theoretically is only possible to achieve completely when the relative amount of the liquid metal is great compared to the amount of oxygen that is available in the refractory. If the refractories had been allowed to soak up oxygen as FeO almost up to but not actually reaching the point B, there would be a reservoir of oxygen available which would continue to diffuse into the system for perhaps several hours before it could finally be nearly completely precipitated as A1 0 of relatively low FeO content.

The situation just described is still far better than the situation in which the refractories are saturated with hercynite because with hercynite the reservoir would contain at least 10 times as much available oxygen, and, furthermore, the presence of the hercynite of course essentially guarantees that the line A-B-C-D will be followed rather than the line A-B-E-F when Al is added to the system.

There are two ways in which a system having refractories and precipitated oxides saturated with hercynite virtually guarantees the continued existence of the hercynite relationship along the line A-BC-D and, therefore, makes it substantially impossible to follow the line A-B-E-F. One of these has to do with the nucleation surface energy at the interface between solid hercynite and liquid metal. The crystal geometry of hercynite at the solid-liquid interface tends to cause the Al and O atoms to conform to it rather than A1 The other is that hercynite continues to feed available oxygen into the system to cause it to follow the AB-C-D curve even when aluminum is added in excess.

One important conclusion from the above analysis is that the presence of any appreciable amount of hercynite in refractories or precipitated oxides, or both, tends to perpetuate itself even when more and more aluminum is added and no additional oxygen is available from external sources. Adding aluminum merely precipitates Q from the metal to form more hercynite. Unless a large excess is added, there probably will be very little tendency to remove much oxygen from the hercynite already present. However, too much aluminum used in the preliminary first stage deoxidation leads to excessive deoxidation of slag and refractories with waste of aluminum.

Even when there is a tendency to remove oxygen from the hercynite already present, this takes an appreciable length of time because of the solid state diffusion process involved. Tests show that it takes 8 to 22 hours or more to reach equilibrium even in a very small laboratory crucible.

So far as the source of oxygen to form the original hercynite is concerned, it may be readily seen that low carbon steels in which 0.065% Q is the normal amount, may tend to contribute the difference between 0.065 and 0.058% Q (or 0.018% Q with 0.3% manganese) to the refractories in the form of dissolved FeO and hercynite. As aforementioned, a new refractory lining normally contains H 0 and other unstable oxides capable of contributing available oxygen to the system. Also, under practical conditions of steel plate operations, there will always be some FeO-rich slag material which will enter the vessel and supply additional available oxygen to it. In any system except a system of high vacuum and with at least some carbon in the metal or a fully environmentally controlled system, oxygen will also be available from the atmosphere at some stage. A refractory lined vessel which has contained liquid Fe and is emptied while in contact with air will invariably tend toward becoming saturated with hercynite due to the droplets of Fe left embedded in or adhering to the lining while still at high temperatures, or during preheating prior to re-use, which will quickly be converted to FeO and absorbed by the refractories as FeO in solid solution or hercynite.

A further significant factor to be reckoned with is that even when aluminum is added to a system containing no hercynite, the surface energy requirement for nucleation of A1 0 is higher than it is for hercynite. Therefore, some driving force is needed to get over the hump" of extra energy required to nucleate A1 0 Also, the hercynite changeover point becomes lower, with or without manganese, as the temperature approaches the freezing point. Preferably, the Q level of Fe should be well below the hercynite changeover point and the system must be free of hercynite compounds in the refractories, or suspended, or floating in the metal. To obtain added driving force for the nucleation of A1 0 and to be below the hercynite changeover point at a temperature just above the freezing point of the metal, a pure Fe-C alloy desirably should be deoxidized to below about 0.015% Q. A steel containing 0.30% manganese should be deoxidized to about 0.010% Q. These levels of percent Q are far below the 0.065 Q level normally found in low carbon steels.

At temperatures significantly above 1600 C., the dissolved oxygen concentration at the hercynite changeover increases significantly. For instance, at 1650" C. the her- 14 cynite changeover point in the absence of manganese is about 0.074% Q (28% greater than at 1600 C.). Therefore, for operation at temperatures in the vicinity of 0 C. the control limits for first-stage deoxidation can be increased to about 0.019% Q for a pure Fe-C alloy, and to about 0.013% Q for steel containing 0.03% manganese.

Summing up, it may be seen that in prior art systems using aluminum as a deoxidizer at least some of the refractories of the vessel containing liquid low carbon steel tend to become saturated with hercynite. Afortiori; such prior art systems do not form pure A1 0 as the reaction product with the curve (4) relationship prevailing.

It thus becomes apparent that a practical system for avoiding hercynite must provide a means whereby liquid metal goes through a multi-stage process in which the first stage accomplishes the following: (1) preliminarily deoxidizes the metal so as to remove excess oxygen down to well below the hercynite changeover point in FIG. 3 and so that it bears a relationship to any aluminum present that approximately conforms to curve B-D in FIG. 3; (2) separates out essentially all suspended or floating hercynite-type materials and restrains these from entering the second stage. The method of separating the suspended or floating slag or hercynite-type materials from the first stage is to allow time for these to rise into the surface layers and then to underpour clean, preliminarily deoxidized liquid metal into a second, or stage (2), vessel or portion of the same vessel.

In the second stage of the process there must always be sufficient g to react with essentially all of the remaining Q and precipitate this as nearly pure A1 0 Sufficient time is given for nearly all the latter to float upward and be restrained from that metal which is underpoured into any further stage or into the casting mold.

The second stage of the process must be fully environmentally controlled in the following ways: (1) the gaseous atmosphere must not be capable of contributing any significant amount of oxygen to liquid Fe at the low oxygen contents which are desired; (2) the refractory walls of the vessel in the second stage must be maintained at all times essentially out of contact with available oxygen and therefore free of any hercynite infiltration; (3) the refractory walls must be composed of stable oxides such as A1 0 which are themselves not decomposed by Fe of low oxygen activity in the presence of Q; (4) a new refractory lining must be given a preconditioning treatment to eliminate serious sources of available oxygen; and (5) some excess of Al over and above the stoichiometric proportions of Q in the metal must be maintained at all times.

Thee environmental control as specified above must be maintained through any subsequent stages and the casting process itself so that a significant amount of oxygen cannot re-enter the metal while in the liquid state.

The first stage can be carried out in several ways including carbon deoxidation with or without reduced partial pressure of carbon monoxide, use of hydrogen, use of other controlled gas environments, or use of aluminum or other suitable deoxidizers.

If a carbon, vacuum or controlled gas method is used for the first stage, it may be desirable to add some aluminum shortly before the end of the first stage deoxidation. In this case, time must be allowed for precipitation and separation of hercynite. Adding aluminum in this manner will help to offset refractory attack.

If aluminum is used as a deoxidizing agent for the first stage, the first stage may or may not be fully environmentally controlled. It is recognized that in the first stage much hercynite will be present and that the main benefit of an innocuous environment will be to conserve some aluminum.

The hercynite changeover point and control curve in FIG. 3 is for the pure Fe, 0, Al system and is subject to variation depending on the presence of other oxides in the system which, of course, are usually present. For example, with manganese content of the metal at around 0.25 to 0.50%, it is probable that point B (0.058% Q) may be as low as 0.018% Q or lower depending upon temperature and, therefore, the rest of the control curve should be adjusted accordingly. Examples of other oxide forming elements, small quantities of which may alter the hercynite changeover point and control curve, are titanium, zirconium, silicon, chromium, magnesium, vanadium, molybdenum, nickel and copper. Some of these may be expected to lower the changeover point B while others may raise it. In the presence of manganese, it may be necessary to reduce the Q level in the liquid metal at 1600 C. in the first stage to below about 0.010% Q and the A}, Q relationship would, of course, be altered also to a lower Q for a given Al content. A complex oxide such as (B -(A1 would now be the product of deoxidation in the first stage.

In practice, low carbon steels nearly always contain some manganese and in view of its lowering effect on the hercynite changeover point and the extra driving force that may be required for effective nucleation of A1 0 in the second stage, Q in the first stage should be reduced to 0.010% Q or below.

It is interesting to note that the hercynite problem is related to carbon contents in steels and may become a major problem only at the lowest carbon levels. For example, a steel containing 0.30% carbon normally would contain no more than 0.02% Q. Such a steel, in the absence of manganese, would not be expected to produce hercynite in the refractory lining of a vessel and if hercynite were already present, the Q (dissolved carbon) in the steel would tend to react with Q produced by the hercynite and thus at least partially deoxidize the refractories. Thus, the instant invention is especially advantageous for the treatment of low carbon steels. However, for steels containing considerable manganese, the instant invention may be beneficial even for steels containing up to 0.40% carbon.

It may be noted that carbon causes a change in kind rather than merely in degree with respect to hercynite. Increased carbon tends to lower Q concentration but an abrupt significant change occurs when the Q reaches the hercynite changeover point and control curve for example, at point B and this latter is influenced by the presence of other alloying elements such as manganese.

By properly carrying out the second and any subsequent stages of the process, the metal can be brought into a highly purified state with very little oxygen present in the metal or available to the metal elsewhere in the environment of the system. Therefore, it is possible to introduce any desired oxidizable alloying elements into the metal in the second stage and obtain essentially 100% addition to the metal or recovery of these added elements. Obviously, this not only avoids waste of the materials themselves but enables very close control of chemical analysis of the finished metal. Thus, in making low carbon tonnage steels, such oxidizable elements as manganese, silicon, chromium and vanadium preferably are added in the second stage. Also, a final carbon addition may be made after the second stage aluminum has been added to obtain close control of carbon content. Also, other oxidizable elements may be added most efiiciently in the second stage for special purposes. For example, calcium, magnesium or rare earths may be added for desulfurization, or titanium or zirconium may be added to obtain vairous benefits such as denitrification or exceptionally low oxygen contents. Digestion of these additions is assisted by stirring as by induction.

One reason that exceptionally close chemical control of metal is obtained is that skulls or coatings of hercynitetype materials which are attached to vessel refractory linings are avoided. Such skulls tends to be produced in prior art methods by chemical reactions at the solid refractory-liquid metal interface, between FeO in the refractories and metal oxide forming elements such as manganese in the metal. Manganese hercynite skull may be formed in one heat giving poor manganese recovery, yet this lost manganese can be recovered back into the metal in a later heat. This condition leads to unpredictable manganese recovery.

With regard to the design of the vessel or vessels for the system, instead of having two or more vessels each having an underpouring arrangement, one or more elongated vessels having refractory dams or baffles to separate the stages, or any other suitable system, might be used. One suitable design of system would consist of a ladle with a bottom tapping nozzle or a teapot spout arrangement for the first stage, with aluminum addition being made to the stream as the vessel is filled. The first stage vessel is underpoured into the second stage vessel which is a channel-type induction furnace with an arrangement for pouring from below the liquid surface through a teapot spout. The channel-type induction heating provides some stirring, but this may be enhanced by magnetic induction stirring. The second stage vessel underpours into a channel-type induction heated tundish, which in turn underpours into one or more continuous casting molds. The second stage vessel is so constructed as to provide a seal at the receiving opening so that no air can enter the vessel during filling. The second stage vessel, the tundish and the liquid metal connections between the second stage vessel and the tundish, and between the tundish and the continuous casting mold, are surrounded by interconnected gas-tight envelopes containing argon or other innocuous gas containing no more than 4% hydrogen.

With regard to selection of the best refractories, this is more important in the second stage and any subsequent stages. In the first stage, the refractory will tend to become coated with hercynite and therefore any reasonably good refractory material should be satisfactory. As pointed out earlier, the second stage refractories should be very stable, for example, A1 0 or possibly a MgO-Al O spinel, so that after conditioning the refractory material itself will not yield available oxygen to the liquid iron. If magnesium oxide is used in the first stage it will tend to become coated with hercynite.

With regard to temperature control, it is evident that this is important in the first stage because the lowest temperature commensurate with requirements of the subsequent stages in the process will insure maximum removal of oxygen in the first stage. Under some circumstances, it may be desirable to reheat the metal to a slightly higher temperature in the second stage and any subsequent stages to aid in A1 0 particle size growth and to insure proper conditions for casting. Some particle size growth in the second stage will assist in the floating upwards of A1 0 particles into an upper stratum of concentrated particles and will thus make for best separation of oxide and for the purest metal passing to subsequent stages.

If desired A1 0 nucleation in the second stage may be aided by adding small particles of pure A1 0 (e.g., minus 200 or smaller mesh corundum or alpha alumina) to the metal at the same time that aluminum is added. This probably should be compounded or mixed with the aluminum additive.

Considerable study has been necessary to determine the amount of aluminum that should be maintained in the second stage in stoichiometric excess of the amount of Q present in the metal as it enters this stage. As earlier mentioned, there are a number of considerations which must be taken into account.

First is the question of whether aluminum had been used for the first stage deoxidation. If aluminum had been used, normally there would be already in the metal as it enters the second stage some stoichiometric excess of aluminum present. One reason is that there may be practical advantages, such as reduced amount of oxides to be disposed of in the second stage, for separating as much as possible of the oxides from the metal in the first stage where it can be done more easily. This is true, for example, where a transfer ladle with aluminum added is used for the first stage. It is also true where vacuum or gaseous deoxidation are practiced for the first stage.

However, a reason for not adding excessive aluminum in the first stage is to avoid over-deoxidation of any slag present or of the refractories, which leads to waste of aluminum and unnecessary refractory attack. Such waste of aluminum will also be minimized if the time allowed for the first stage is not excessive because unnecessary time will lead to diffusion of oxygen from the solid oxides and refractories into the metal. Time allowance in the first stage normally should not be in excess of 20 minutes for more eflicient use of aluminum and for best refractory life. Also, as previously stated, it is desired to add some of the aluminum in the second stage so if an excessive amount is used in the first stage there would then be a total amount which would be unnecessarily large. A further reason for avoiding use of excessive aluminum in the first stage is so that there will be sufiicient Q left in the second stage to give some driving force for nucleation and growth of A1 particles so that best separation of oxide particles will be achieved.

If aluminum had not been used in the first stage, then of course there must be an addition of aluminum at the beginning of the second stage treatment, preferably as the metal is being introduced and/or previously added to the remaining heel of liquid metal, the amount of which is at the very least in stoichiometric excess of the oxygen present.

Even if a stoichiometric excess of aluminum is already present as the metal enters the second stage, some additional aluminum preferably should be added in the second stage to aid in nucleation of A1 0 by furnishing nucleation sites as the aluminum diffuses into the newly added liquid iron containing some oxygen.

Perhaps the most important consideration for determining the amount of stoichiometric excess of aluminum to be maintained in the second stage is to provide driving force so that deoxidation will take place to such an extent that the dissolved oxygen remaining in the metal will be very low and will remain low during solidification by having sufficient excess aluminum to take care of segregation of oxygen in the liquid core. This is particularly important if the metal is to be cast continuously and the degree of stoichiometric excess will also depend upon the temperature. The higher the temperature, the greater the stoichiometric excess should be in order to obtain lowest oxygen contents.

Another reason for maintaining a residual concentration of aluminum in the metal is for the purpose of combining with nitrogen to make the steel non-aging.

Another purpose for residual aluminum is to provide the well-known effect on grain structure and physical properties.

Although improved control of dissolved oxygen in liquid ferrous metals at a uniform and very low level constitutes an important improvement for any casting process, such a method is particularly needed for the continuous casting of low carbon steels. This is because the continuous casting of low carbon steels brings into combination a solidification system which continuously tends to concentrate, i.e. segregate, impurities into the remaining liquid as freezing continues, and such an impurity is oxygen, which has an unusually high tendency to segregate into the remaining liquid.

Although the maximum segregation is only achieved with true equilibrium and this takes time, the fact remains that with oxygen content of the starting liquid metal at the higher levels corresponding to the hercynite conditions as in prior art, there will be a significant and harmful buildup of increased oxygen content in the pool of liquid metal in a continuous casting solid-liquid system. If carbon is present and aluminum or some other deoxidizer is not used properly and in sufficient excess, a violent rimming action may result since carbon also tends to segregate in the pool of liquid metal remaining. If, on the other hand, residual aluminum is available, hercynite conditions such as in prior art plus the segregation of oxygen in the remaining liquid tend to form serious concentrations of hercynite reaction products and formation of serious defects in the continuous cast product.

As earlier discussed in connection with FIG. 1, curve (2), the hercynite aluminum-oxygen relationship makes for relatively incomplete removal of oxygen by aluminum from liquid steel. Even with a considerable excess of residual aluminum, oxygen concentration remaining in the metal is high. Since oxygen solubility in solid metal is very low, nearly all of the oxygen remaining in the liquid metal after the hercynite reaction as in prior art will appear as precipitated hercynite oxides during the cooling and freezing which occur in the casting process.

Table I has been prepared to show the greatly improved completeness of deoxidation by aluminum that can be achieved by the environmental control method as compared With the prior art method. Not only is less aluminum required but a very much lower oxygen content is obtained.

TABLE I [Effectiveness of aluminum deoxidation of low carbon steels with 0.20% available oxygen] Aluminum added Cumu- Final chemical lative, analysis, percent Lbs./ lbs./ Temp., C. ton ton 0 Al Prior art method:

1700 first stage-.." 4. 40 5 0. 028 0. 09 Additional 1. 30 0. 020 O. 15 1600 first Stage" 3. 5.90 0. 010 0. 05 Additiona 2. 00 5. 90 0. 005 0. 15 Enviromfrlrental control method: 4 4 4 9 0 028 0 09 0.5 4.9 0.001 0.085 1600 first stage 3.6 4. 1 0.013 0. 027 Second stage- 0. 5 4. 1 0. 0005 0. 04

There are at least three types of undesired conditions that exist in the prior art methods during the continuous casting process which tend to maintain the undesired curve (2) conditions. These are: passing metal through an atmosphere containing available oxygen, passing metal through a tundish lined with oxidized refractories, and having the metal itself containing particles of solid hercynite such as will happen if the freezing interface in the mold of a steel containing 0.25% manganese contains more than about 0.02% Q.

The formation of gross amounts and sizes of oxide clusters tends to lead to serious oxide streaks, particularly near the surfaces of rolled products. This type of condition frequently is alleviated in prior art steel plant practice by scarfing slabs or billets to remove the surface layers where such gross oxides often are mainly found.

However, non-metallic inclusions which may consist of rather well distributed oxide particles have long been known to have various harmful effects in steel products. Practice of this invention is of great benefit in the early complete elimination of such inclusions.

It may be expected that none whatsoever of the massive types of oxides (i.e. hercynite oxide clusters) will be formed and, therefore, no scarfing will 'be required through the practice of this invention. Furthermore, the

19 amount of non-metallic inclusions will be of the order of or less by weight of the inclusions present from other methods, and will consist of well-dispersed particles of nearly pure A1 0 and of very small size.

In the practice of this invention a further advantage gained is with respect to refractory life. There are at least five ways in which hercynite adversely influences refractory life.

First, in forming a ceramic alloy with A1 0 hercynite reduces the refractoriness of the A1 0 because the ceramic alloy has a lower melting point than the pure Al O Second, the dissolving of FeO in A1 0 causes a volume change to occur which could lead to cracks and a general distintegration of the material.

Third, when oxidized refractories are reduced to a lower level of oxygen through contact with liquid metal in which a deoxidizer has just been added, the deoxidizer and metal tend to attack the refractory wall by reduction of the FeO back to Fe. Fe particles left in the refractory can subsequently reoxidize and this will cause expansion stresses and cracking.

Fourth, as previously mentioned, a hercynite skull or coating can be formed in prior art methods against the lining of a vessel and this may require premature replacement of refractories in the vessel.

A fifth possible adverse effect upon refractories is evident by noting in FIG. 1 that liquid metal at L on curve (2) is capable of dissolving an appreciable amount of aluminum and oxygen from the refractories if the temperature rises to curve (1). Under curve (3) and (4) conditions, on the other hand, this solution effect is very slight.

One way to restrain the fifth effect mentioned above is to maintain a residual amount of aluminum always in solution in the metal. The mass action of the aluminum atoms tends to restrain the solution of A1 0 The first four effects will tend to exist to some extent in the first stage of the process, but since the conditions of the refractory, including temperature, will be more nearly constant, it is expected that refractory life will be improved over conventional systems for handling low carbon liquid Fe. The refractory life in the second and further stages will be especially good.

When all of the factors are considered which are affected by the aluminum-oxygen relationship in the first stage, in the second stage and those which relate to the quality of the steel product, a stoichiometric excess of aluminum to the amount of available oxygen present for A1 0 should be present at the beginning of the second stage. Normally, the Q in the metal should be reduced in metal containing manganese in the first stage to 0.010% or below. In the second stage, there should be additional aluminum added to the extent of about 0.025% or more. After the deoxidation reaction in the second stage is completed, the preferable chemical analysis would be: aluminum 0.04% to 0.08%, oxygen less than 0.003%.

The following may help to clarify the reasons for the requirements as set forth above. If in the second stage the refractories are not pretreated and if the atmosphere is not maintained at a pH O/pH ratio below the level at which there is a tendency to nucleate and grow solid hercynite particles in the metal in the second stage, the basic purpose of the process will not be accomplished. Furthermore, if a stoichiometric excess of aluminum over the amount of oxygen present in the metal is not maintained in the second stage, then it follows that there may fail to be sufficient driving force provided to maintain the A1 0 relationship in this and any subsequent stages, and the desired result may again fail to be achieved.

Two different test series were performed upon 12-24 kg. melts. Series I (tests 36-48) was performed in one induction furnace crucible and with an argon atmosphere. Series H (tests 4969) was performed in one crucible for tests 49-56 and n anoth r fo t sts 57-69 a d y g was added to the argon atmosphere for all of the Series II tests. The melting stock was primarily iron having the following analysis: carbon 0.02%, manganese 0.04%, silicon 0.02%, sulfur 0.015%, aluminum 0.005%, oxygen 0.080%, and nitrogen 0.007%. To this primary stock was added in some of the tests small proportions of so-called carbon steel having the following analysis: carbon 0.16%, manganese 0.32%, silicon 0.024%, sulfur 0.024%, aluminum 0.008%, oxygen 0.013%, and nitrogen 0.004%. The innocuous atmosphere used in the tests was comprised of varying proportions of 99.9% pure argon and 99.995 pure hydrogen. The argon was certified to contain not more than 8.5 parts per million (p.p.m.) water and not more than 5 p.p.m. oxygen, and the hydrogen was certified to contain less than 10 p.p.m. water and less than 2 p.p.m. oxygen. The refractory lining of the vessels employed for both Series I and Series II comprised 98% magnesia dry granules mixed with 2% water which was predried and sintered about a thin steel inner cylinder which was melted and removed in the first melt. For tests 57 et seq. of Series II, this lining was used in conjunction with a refractory cylinder forming the upper half of the crucible having the following composition: A1 0 (96.64%), CaO (2.71%), SiO (0.07%), Fe O (0.05%), MgO (0.06%), alkalies (0.12%), and TiO (trace). In all tests, power was turned on in the melting furnace only after oxygen had been removed from the vessel by displacement with argon as the circulating innocuous gas stream. Temperature of all tests was controlled by a two color radiation pyrometer. The gas pressure in tests 57 through 69 of Series II was maintained at slightly greater than atmospheric pressure by the addition of make-up gas to offset potential small leaks. In those tests where an aluminum addition was made, the aluminum was added by plunging a packet of aluminum into the molten melt at the end of a rod of iron of the same composition as that already molten in the melting furnace. In the Series II tests, hydrogen level in the circulating gas was dropped to no greater than about 3-6% during pouring of the melt at the end of each test so that the cast metal would be substantially free of gas bubbles as explained aove. All metal was cast into copper molds contained in the same environmentally-controlled cabinet as the furnace. For tests 58 et seq. of Series II, the stream of poured metal was directed into the mold by use of a copper funnel to avoid contamination from available oxygen in, say, a ceramic funnel. The results of the tests in Series I in all of which aluminum was employed as a deoxidizer in the absence of hydrogen pretreatment of the refractories are summarized in Table II below.

TABLE II [Summary ofSerles I tests made with aluminum deoxidizer and in the absence of hydrogen Al added (percent) Test No.

99999999999 2saaaa88888 999999999 8H co Q 2882222828 rrrrrrrr-- \IQQQQQQQQQQ 88282288828 1 In test 48a small amount of hydrogen was added.

The results of tests in Series II in which some hydro- 1gen treatment was used are summarized in Table III be- TABLE III [Summary of Series II tests 1 made in argon and hydrogen] Gas conditions when metal was poured Gas Hydrogen added Melt analysis (percent) leaving Total gas Total gas Total gas crucible Max. vol. Hours Al added emp., H20 Hg pH O/pHz pHzO/pH, Test No (percent) Liters added (percent) Mn Al C. (p.p.m.) 2 (p.p.m.) 2

7 42 0. 7 0. 10 0. 004 0. 01 0. 03 0. 15 1, 700 2, 200 16, 000 0. 137 3. 3 50 149 1. 0 None 0. 001 0. 05 0. 035 0. 129 1, 700 1 ,350 32, 000 0. 042 0. 70 7 2 l1 0. 6 None 0. 001 0. 05 0. 008 0. 127 1, 700 i 164 2. 1 None 0. 001 0. 05 0. 004 0. 084 1 650 175 1. 75 None 0. 001 0. 04 0. 007 0. 085 1, 700 481 2. 7 None 0. 001 0. 04 0. 015 0. 067 1, 700 20 537 2. 6 None 0. 001 0. 03 0. 023 0. 032 1, 620 20 555 2. 8 0. 10 0. 001 0. 03 0. 054 0. 009 1, 620 2, 060 8. 4 None 0.001 0. 03 0. 027 0. 103 l, 600

10 440 2. 2 None Trace 0. 05 0. 004 0. 080 1, 700 7, 500 93, 000 0. 081 0. 225 8 400 2. 6 None Trace 0. 02 Trace 0. 041 1, 650 3, 100 40,000 0. 078 0. 200 40 2,600 3. 5 None 0. 006 0. 03 0. 005 0. 015 1, 650 1,000 60, 000 0. 017 0. 033 55 2, 900 2. 7 None 0. 001 0. 03 0.014 0. 008 1, 600 480 90, 000 0. 0053 0. 0103 55 16, 000 20. 0 None 0. 001 0. 03 0. 014 O. 002 1, 650 165 36, 000 0. 0046 0. 0088 52 4, 400 4. 2 None 0. 004 0. 04 0. 007 0. 0025 1, 650 205 85, 000 0. 0024 0. 0047 60 4, 900 5. 6 0. 10 0. 002 0. 02 0. 077 0. 0055 1, 650 170 54, 000 0. 0031 0. 0061 57 4, 100 5. 6 0. 05 0. 002 0. 06 0. 041 0. 002 1, 650 230 56, 000 0. 0041 0. 0131 61 3, 400 4. 4 0. 05 0. 038 0. 05 0. 043 0. 0025 1,630 115 47,000 0. 0024 0. 0045 60 2, 800 2. 6 3 0. 05 0. 002 0. 10 0. 027 0. 0015 1 650 65 40, 000 0. 0016 0. 0027 l Tests 49 through 56 made in same MgO crucible; Tests 57 through 69 made in same combination MgO, A1203 crucible.

2 By volume.

5 A1 added was contained in Ca-Al alloy of 50% Aland 50% Ca composition.

A comparison of the tests of Series I and those of Series 11 illustrates the importance of pretreating the refractory lining with hydrogen prior to the addition of aluminum. The presence of hercynite throughout Series I was indicated by petrographic examination, the most obvious examples of which were tests 36, 40 and 44 in which massive hercynite oxide clusters appeared in the cast metal.

One apparent reason for the large quantities of hercynite formation in the tests of Series I may be found by referring to Table IV below. It is seen from Table IV, derived from the literature, that even extremely small amounts of water vapor in argon are oxidizing to iron at 1600" C. in the absence of hydrogen. Dew points of the gas in the system during the Series I tests varied from 5 C. to l+5 C. (4000-8600 p.p.m. water). It is apparent from the above, that all tests of Series I were contaminated by some source of water having a highly oxidizing tendency and that argon, though dry when introduced into the system, became contaminated by oxygen in the form of water in substantial amounts.

TABLE IV [Equilibrium level of oxygen dissolved in iron at 1,600 O. at different levels of dew point and hydrogen content of argon] 1 Very low.

In tests 49-56 of Series II concentrations and total quantities of hydrogen were gradually increased to obtain lower oxygen levels in the metal. However, in tests 49-2 and 55-3, with the addition of 0.10% aluminum to the melt, the oxygen levels of 0.15% and 0.009% in the metal were still in accordance with curve (2) of FIG. 1. This indicates that the hercynite relationship continued.

Test 56-3 of Series II indicates that even the increase of hydrogen addition fourfold over an extended period of time (8.4 hours) of continuous recirculation over a molecular sieve to remove water still produced an unacceptable level of water vapor in the effluent gas at the end of the test (1150 p.p.m. water and a dew point of 21 C.). It was hypothesized that the source of available oxygen was the refractory lining itself. It was surmised that the refractory was chemically unstable to the extent that it yielded available oxygen to the metal which reacted with the hydrogen maintained at elevated temperatures to form water vapor over extended periods of time.

The above theory of chemical instability of the refractory walls was substantiated by an examination of a crosssection of the lining employed in tests 49-56 of Series II. It was determined that infiltration of the MgO crucible wall by FeO to a depth of about 10 mm. occurred with the formation of a ceramic alloy consisting of a solid solution of FeO and MgO as indicated by the presence of an ebony black glassy zone to that depth. At a magnification of 100, a band of grayish metallic white material at a depth of 4 to 5 mm. within the lining was found to contain particles of metallic iron. It was found that the hydrogen had partially reduced the iron oxide from the above solid solution to that depth.

For the performance of the tests 57-69 of Series II summarized in Table III, the furnace was relined using rammed MgO for the lower half of the crucible and a sintered A1 0 cylinder for the upper half thereof. All tests in both Series I and II were run sequentially by num ber. Thus, for a sequence of tests in which the same crucible was employed, the condition of the crucible at the beginning of each subsequent test was approximately the same as that which existed at the end of the immediately preceding test, except for oxygen contamination introduced when the cabinet was opened to remove samples and recharge the furnace with new melting stock after each test.

Referring to FIG. 6, a graph is illustrated in which water vapor content in parts per million of the total gas leaving the cabinet was plotted as a function of time. The results of test 61 illustrate that after an initial peak of water vapor at 2800 p.p.m. (dew point -9 C.) at about ten minutes, there was a steady decrease in concentration with time. It is evident that in test 61, at least 22 hours were required for depletion of available oxygen from the ceramic crucible. In tests 62, 63, 65 and 68, the time for preconditioning the crucible to remove the available oxygen to an acceptable level was decreased. The moisture content of the total effluent gas, as plotted in FIG. 6, is not as accurate an indication of the refractory conditioning as the ratio of water vapor concentration to hydrogen concentration by volume in the gas in contact with the stirred liquid metal. A reasonable approximation of this ratio in the gas in contact with the liquid metal is the ratio in the leaving gas which has been in contact with the liquid metal and crucible wall. Referring to FIG. 7, the ratio of water vapor concentration by volume to hydrogen (pH O/pH in the gas leaving contact with the liquid metal and crucible is plotted against time at the melting temperature. These values of the pI-I O/pH ratio are greater than existed in the total gas leaving the cabinet and were computed taking into account the fact that only a portion of the hydrogen-rich gas contacted the metal and crucible. The balance of the gas flow bypassed through the outer regions of the cabinet and was effective in preventing air in-leakage, but not in conditioning the refractory. In test 57-1, the first one in a new crucible, it is seen that after 2.2 hours of hydrogen addition and 60 minutes at melting temperature there was very little reduction in the ratio of water vapor to hydrogen, which ratio was still at the high level of 0.225. Furthermore, the percentage oxygen in the steel was 0.08% or about the same as the original melting stock prior to treatment. However, the ratio was consistently improved with subsequent tests. When the pH O/pH ratio was decreased to below 0.008 (the ratio in gas contacting well-stirred metal should not exceed 0.008 value if liquid ferrous metal containing about 0.002% oxygen or less is to be produced continuously) by hydrogen pretreatment of the refractory for a sufiicient period of time, a stable condition of the refractory developed in which the ratio thereafter fairly consistently remained below that value. Apparently the hydrogen added in tests 59 and 60-1 was sufiicient to lower this ratio to below 0.008 but insufficient to maintain that ratio upon cessation of the hydrogen conditioning. Thus the refractory lining was only partially conditioned and still capable of yielding available oxygen upon lowering the percent hydrogen in the atmosphere. In contrast, the lengthy hydrogen treatment in test 61 was sufficient to precondition the refractory lining for subsequent tests involving less extensive treatment. For test 65, the computed ratio of pH O/ pH (0.0131) in the portion of the gas stream leaving contact with the liquid metal and the crucible, is unexpectedly high considering the 0.002% dissolved oxygen in the metal poured. This may be attributable to errors in recording the flows of makeup gas and total gas. All melts in Series I and II deoxidized with aluminum including runs made with and without refractory lining pretreatment by hydrogen are summarized in Table V below.

TABLE V curves are of the type described fully with respect to the same curves in FIG. 1. It is apparent that the plot of the tests in Series I, using aluminum as a deoxidizer without hydrogen pretreatment of the refractory, indicates the formation of hercynite. At the beginning of Series II, in the plot designated El the hercynite relationship holds indicating insufiicient pretreatment to deoxidize the refractory. The plot designated (9 is between the hercynite and A1 curves which indicates partial refractory conditioning, i.e., sufficient to avoid hercynite formation but insutficient to produce iron having desired low oxygen content. Finally, the plot designated EB is essentially on the A1 0 curve, taking experimental error into account, which means that the refractory had been conditioned.

It is apparent from the foregoing that the conventional refractory linings of vessels containing molten ferrous metal at about l600-1700 C. contain large amounts of available oxygen which may be yielded to the molten ferrous metal. Such available oxygen may be removed by pretreatment with hydrogen for say 5-22 hours or more with a heel of molten ferrous metal in the vessel. Preconditioning suflicient to prevent the formation of hercynite is indicated when the ratio of water vapor to hydrogen in contact with the liquid metal is maintained below about 0.008 during preconditioning at a metal temperature in the vicinity of 1600 C. At this level, the curve (4) relationship in FIG. 1 will be obtained and aluminum oxide rather than hercynite will be produced upon the addition of aluminum. Otherwise, there is a substantial chance that the curve (2) relationship of FIG. 1 will be obtained upon the addition of aluminum. The maintenance of this condition in the refractory does not require pretreatment for each new batch fed to the vessel or for continuous feed of a batch. Thus, subsequent to hydrogen pretreatment such as in test 61, a proper pH O/pH ratio may be maintained with only a small proportion of hydrogen in the vessel atmosphere (say 4% or less). This ability is essential for continuous casting of high purity aluminum deoxidized steel as explained above.

During the final period of second deoxidation to ob- [Summary of melts deoxidized by All Unreeovered Al Analysis (percent) Avallalbo Test A1 added Temp., Al Percent O Equiv. oxygen Oxide N0. (percent) 0. A1 0 (percent) recovery (percent) (percent) type 36 0, l, 700 0. 018 0. 054 0. 182 9. 0 0. 215 0. 269 Hereynite 37 0,25 1, 700 0.032 0. 020 0. 218 12. 8 0. 258 0. 278 D0. 38 0,25 1, 700 0. 090 0. 007 0. 160 36. 0 0. 190 0. 197 Do. 0,30 1,700 0. 087 0. 008 0. 213 29. 0 0. 252 0. 260 D0. 40 0,31 1 700 0. 024 0. 032 0. 286 7. 8 0. 340 0. 372 Do. 42 0,20 1,700 0. 095 0. 007 0. 105 47. 5 0. 124 0. 131 Do. 44 0, 29 1, 690 0. 091 0. 024 0. 199 31. 3 0.235 0. 259 Do. 45 0, 25 1,690 0. 120 0. 0065 0. 130 48. 0 0. 152 0. 158 Do. 46 0, 15 1, 700 0. 098 0. 006 0.052 65. 0 0. 062 0. 068 D0. 47 0, 14 1, 700 0. 090 0. 007 0.050 64. 0 0. 059 0. 066 D0. 48 0,25 1,700 0.11 0. 006 0.140 44. 0 0. 166 0. 172 Do. 49 0, 10 1, 700 0. 03 0. 15 0. 070 30. 0 0. 003 0. 243 Do. -3 0,10 1, 620 0. 054 0. 009 0. 046 54. 0 0. 054 0. 063 Do. 03 0, 10 1,650 0. 077 0.0055 0. 023 77. 0 0. 020 0. 0255 A120; 0, 05 1,650 0. 041 0. 002 0. 009 82. 0 0. 0080 0. D0. 68 0, 05 1, 630 0. 043 0. 0025 0. 007 86. 0 0. 0062 0. 0087 Do.

The available oxygen at the time of the aluminum addition in the above Table V was calculated by taking the stoichiometric equivalent of oxygen in the appropriate form of aluminum oxide corresponding to the aluminum presumably lost in the reaction. Such lost aluminum equals the aluminum added minus the residual aluminum found in the metal. To this oxygen figure was added the percent oxygen actually found in the metal after the reaction. It is apparent that the available oxygen content steadily decreased with increased hydrogen treatment of the refractory.

Referring to FIG. 8, percent oxygen is plotted against percent aluminum in the cast product. For reference purposes, curves (1) and (2) represent hercynite-type oxide formation in liquid iron containing oxygen and aluminum, and curves (3) and (4) represent A1 0 formation. These tain a ferrous metal containing 0.006% by weight or less of oxygen, an atmosphere of gas in contact with the low carbon ferrous metal is maintained in which the pH O/ pH is no gerater than about 0.02. correspondingly, for a final metal product of 0.002% or less oxygen, the pH O/pH ratio should be no greater than 0.008.

What is claimed is:

1. A process for producing ferrous metals substantially free of oxygen-containing impurities by preliminary deoxidation in a first stage followed by second stage deoxidation in a refractory-lined vessel comprising the steps of:

(a) providing a heel of molten ferrous metal in the refractory-lined second stage vessel maintained above the melting point of the metal;

ferrous metal has a low carbon content of no greater than about 0.15% carbon.

num content in the subject molten ferrous metal after step (h) is in the range of 0.04 to 0.08%.

performed by continuous casting.

1y free of oxygen-containing impurities by preliminary deoxidation in a first stage followed by a second stage deoxidation in a refractory-lined vessel comprising the steps of:

(b) maintaining the molten ferrous metal heel-containing second stage vessel with a supply of deoxidation agent selected from the group consisting of carbon, carbon monoxide, hydrogen and combinations thereof to react with oxygen impurities in the refractory lining to form gaseous reaction products selected from the group consisting of carbon monoxide, carbon dioxide and water vapor;

(c) removing the gaseous reaction products from the second stage vessel;

(d) continuing steps (b) and (c) until the refractory lining is preconditioned by lowering its available oxygen to a level below that which will produce hercynite by the addition of aluminum in step (g);

(e) treating a body of subject molten ferrous metal containing less than 0.40% carbon by preliminary deoxidation in the first stage so as to lower its dissolved oxygen level to less than 0.015% oxygen;

(f) withdrawing the forrous metal heel and supplying the subject molten ferrous metal deoxidized in step (e) into the second stage vessel;

(g) deoxidizing the subject molten ferrous metal in the second stage vessel by the addition of an excess of aluminum over the stoichiometrically required quantity to convert to A1 0 all oxygen remaining in that metal to obtain a metal containing 0.006% by weight or less of oxygen, while maintaining a substantially innocuous atmosphere; and

(h) withdrawing the deoxidized subject molten ferrous metal from the second stage vessel while maintaining substantially innocuous atmosphere with a hydrogen content of not more than about 4% by volume at one atmosphere total pressure.

2. A process as in claim 1 in which the subject molten 3. A method as in claim 1 in which aluminum is em- 4. A method as in claim 1 in which hydrogen is employed as a deoxidizer for the subject molten ferrous metal in step (e).

5. A process as in claim 1 in which the residual alumi- 6. A process as in claim 1 in which the oxygen content in the subject molten ferrous metal withdrawn in step (b) is no greater than 0.002% by weight.

7. A process as in claim 1 in which the subject molten ferrous metal withdrawn in step (h) is solidified in a mold without exposure to an atmosphere containing significant available oxygen.

8. A process as in claim 7 in which the solidification is 9. A method as in claim 1 in which a small quantity of A1 0 is added in step (g) to advance the formation and growth of the A1 0 deoxidation product.

10. A process for producing ferrous metal substantial- (a) providing a heel of molten ferrous metal in the second stage refractory-lined vessel maintained above the melting point of the metal;

(b) circulating a hydrogen-containing innocuous gas stream through the second stage vessel and contacting both the surface of the molten metal heel therein and the refractory lining so as to provide a reaction between the hydrogen and the oxygen impurities of the same to form water vapor;

(c) removing the formed water vapor from the gas in the second stage vessel until the refractory lining 1S preconditioned by lowering its available oxygen to a 75 level below that which will produce hercynite by the addition of aluminum in step (f);

(d) treating a body of subject molten ferrous metal containing less than 0.40% carbon by preliminary deoxidation in the first stage so as to lower its dissolved oxygen level to less than 0.015 oxygen;

(e) withdrawing the metal heel from the second stage vessel and supplying the subject molten ferrous metal deoxidized in step (d) into the same while maintaining the atmosphere in said vessel characterized by a volumetric ratio of water vapor to hydrogen of no greater than about 0.05;

(f) deoxidizing the subject molten ferrous metal in the second stage vessel by the addition of an excess of aluminum over the stoichiometrically required quantity to convert to A1 0 essentially all oxygen remaining in that metal to obtain a metal containing 0.006% by weight or less of oxygen while maintaining a substantially innocuous atmosphere and an atmosphere of gas in contact with the subject molten ferrous metal characterized by a volumetric ratio of water vapor to hydrogen concentration of no greater than about 0.02 during the final period of deoxidation; and

(g) withdrawing the deoxidized subject molten ferrous metal from the second stage vesesl while maintaining a substantially innocuous atmosphere with a hydrogen content of not more than about 4% by volume at one atmosphere total pressure.

11. A process as in claim 10 in which a test for said preconditioning comprises equilibrating by sealing a hydrogen gas-containing atmosphere within the vessel at 1600 C. and about one atmosphere pressure for one day with molten ferrous metal containing lesst han 0.0137% oxygen and less than 0.02% carbon to thus form an atmosphere characterized by a volumetric ratio of water vapor to hydrogen concentration of no greater than about 0.05.

12. A process as in claim 10 in which a test for said preconditioning comprises no substantial decrease after equilibration in carbon level in a molten ferrous metal heel initially containing about 0.20% carbon and no more than 0.01% oxygen by sealing said vessel for one day at a temperature in the vicinity of 1600 C. and an initial carbon monoxide partial pressure of about one atmosphere and maintaining a total pressure of about one at mosphere by the addition of a sufiicient quantity of argon to the vessel.

13. A method as in claim 10 in which a quantity of the subject molten ferrous metal is continuously fed to the second stage vessel and withdrawn therefrom after deoxidation to a level of 0.006% or less oxygen While maintaining the atmosphere of said vessel at a hydrogen level of 4% or less and at the ratio of water vapor to hydrogen maintained in step (f).

14. A process as in claim 10 in which prior to withdrawal in step (e) the molten metal heel contains less than 0.15 carbon and is deoxidized by the addition of an excess of aluminum over the stoichiometrically required quantity to convert to A1 0 all oxygen remaining in that metal to obtain a metal heel containing 0.006% by weight or less of oxygen and hydrogen is reduced to the level of step (g).

15. A process as in claim 10 in which preconditioning of the refractory is performed with an innocuous gas stream containing at least 40% hydrogen in from 5 to 22 hours while maintaining a pH O/pH ratio of no greater than 0.008 in contact with the liquid metal.

16. A process for producing ferrous metals substantially free of oxygen containing impurities by preliminary deoxidation in a first stage followed by second stage deoxidation in a refractory-lined vessel comprising the steps of:

(a) providing a heel of molten ferrous metal in the refractory-lined second stage vessel maintained above the melting point of the metal;

(b) maintaining in said metal heel a carbon content of at least about 1.0% by weight based upon the total heel content for a substantial portion of its residence time to provide a reaction between the carbon and the oxygen impurities in the refractory to form gaseous reaction products selected from the group consisting of carbon monoxide and carbon dioxide;

(c) removing the gaseous reaction products from the second stage vessel until the refractory lining is preconditioned by lowering its available oxygen to a level below that which will produce hercynite by the addition of aluminum in step (f);

(d) treating a body of subject molten ferrous metal containing less than 0.40% carbon by preliminary deoxidation in the first stage so as to lower its dissolved oxygen level to less than 0.015% oxygen;

(e) withdrawing the metal heel from the second stage vessel and feeding the low carbon molten metal deoxidized in step ((1) into the same while maintaining the atmosphere in said vessel at a volumetric ratio of carbon dioxide to carbon monoxide concen tration of no greater than about 0.01;

(f) deoxidizing the subject molten metal in the second stage vessel by the addition of an excess of aluminium over the stoichiometrically required quantity to convert to A1 essentially all oxygen remaining in that metal to obtain a metal containing 0.006% by weight or less of oxygen, while maintaining a substantially innocuous atmosphere and an atmosphere of gas in contact with the subject molten ferrous metal characterized by a volumetric ratio of carbon dioxide to carbon monoxide of no greater than about 0.005 during the final period of deoxidation; and

(g) withdrawing the deoxidized subject molten ferrous metal from the second stage vessel while maintaining a substantially innocuous atmosphere.

17. A process as in claim 16 in which the carbon content in step (b) is maintained in a range of 1.5 to 4% by weight of the total metal heel content.

18. A process as in claim 16 in which a test for said preconditioning comprises equilibrating by sealing a hydrogen gas-containing atmosphere within the vessel at 1600 C. and about one atmosphere pressure for one day with molten ferrous metal containing less than 0.0137% oxygen and less than 0.02% carbon to thus form an atmosphere characterized by a volumetric ratio of water vapor to hydrogen concentration of no greater than about 0.05.

19. A process as in claim 16 in which a test for preconditioning comprises no substantial decrease after equilibration in carbon level in a molten ferrous metal heel initially containing about 0.20% carbon and no more than 0.015% oxygen by sealing said vessel for one day at a temperature in the vicinity of 1600 C. and an initial carbon monoxide partial pressure of about one atmosphere and maintaining a total pressure of about one atmosphere by the addition of a sufficient quantity of argon to the vessel.

20. A method as in claim 1 in which the subject molten ferrous metal contains at least 0.30% manganese and the dissolved oxygen content is lowered to no greater than about 0.010% in step (e).

References Cited UNITED STATES PATENTS 3,467,167 9/1969 Mahin 46 UX 3,459,537 8/1969 Hornak 7549 3,375,100 3/1968 Vopel 75-49 3,230,074 l/l966 Roy et al. 7549 3,389,989 6/1968 Finkl 7549 L. DEWAYNE RUTLEDGE, Primary Examiner M. J. ANDREWS, Assistant Examiner U.S. Cl. X.R. 75-49, 58; 16456

Images