CHAPTER 12

Melting, Casting, and Hot Processing

Summary
THE PRIMARY PRODUCTION PROCESSES of melting, casting, and hot processing are invisible to the end user. The vast majority of stainless steel is made by arc furnace melting followed by argon oxygen decarburization (AOD) rening and continuous casting. It is not normal, and it is seldom benecial for the end user to specify processing paths. The end user should, however, be knowledgeable and require the producer to document the process and the producers control of it. austenitic alloys, impossible to produce. The advent of AOD, continuous casting, ladle metallurgy, and powerful hot rolling mills has led to stainless steels of much higher quality produced at lower cost. Ironically, the low processing cost of stainless steel has spurred demand and made some of its ingredients, such as molybdenum and nickel, which are relatively scarce and expensive commodities, even more costly, forcing the cost of many alloys to spike even higher than in earlier years.

Melting and Rening

The arc furnace is nearly universally used for the rst step in the production of stainless steel. The arc furnace is quite exible in the types of charge materials it can accept. Since the charge materials for stainless steel are typically carbon steel and stainless steel scrap, this exibility allows scrap of all types to be used. The necessary chromium is added as ferrochromium, whose cost is inversely related to its carbon content. The carbon content of the heat of steel is roughly 1.5 to 2.5% when it is melted and ready to charge into the separate rening vessel. It is this carbon whose removal is the primary focus of rening. In the 1960s, Union Carbide engineers perfected a method, the previously mentioned AOD process, of removing nearly all the carbon from molten stainless steel without signicant loss of chromium. This process is based on the following chemical reaction:
Cr3O4 (Solid) + yC = yCO (gas) + Cr (Eq 1)

Introduction
The manner in which stainless steel is made at the producing mill can have a great impact on its nal properties. These production methods have undergone a major evolution over the last 50 years and are mainly responsible for stainless steels becoming the practical, widespread engineering materials they are today. Traditional carbon and alloy steel-making methods are not suitable for stainless steels. The fundamental difference is that the basic decarburization step, which is common to all steel making, is thermodynamically very difficult in stainless steel because the essential element, chromium, reacts more strongly with the purifying agent, oxygen, than does carbon. Thus, early stainless steel making, done in an arc furnace, was a lengthy process that necessarily involved high chromium losses to the slag as carbon was removed. This process was not only very expensive, the carbon levels that could be achieved were not much below 0.10%, making most of todays stainless steels, whose carbon levels range from 0.010% in stabilized ferritic alloys to about 0.07% in normal

The equilibrium for this reaction is:

Ln (K ) = G 4575T (Eq 2)

156 / Stainless Steels for Design Engineers

where K is the equilibrium constant, and G is the Gibbs free energy. Working through the thermodynamics yields the relationship that summarizes the important relationship among carbon, chromium, and CO (Ref 1):
Log = %Cr 13,800 = + 8.76 0.925 log p CO T %C (Eq 3)

Thus, increasing the temperature works to increase the elimination of carbon as CO, which evolves from the melt. This is similar in principle to the basic oxygen furnace (BOF) process for carbon steel in which oxygen is injected into molten steel to remove carbon by oxidizing it. The key to the AOD process, though, is the injection of oxygen and argon into the bath to keep the partial pressure of CO (pCO) very low. This is done at a temperature consistent with economic refractory life. The injection is done through tubes called tuyeres in the bottom of the barrelshaped vessel. The injection and the reaction cause extremely thorough mixing, which would never happen in the at, stagnant, arc furnace bath. This mixing not only allows the CO-producing reaction to reach equilibrium, but also the mixing of the slag and metal also permits desulfurization. By increasing the ratio of argon to oxygen in the injected gas as the rening proceeds, the carbon is selectively oxidized without concurrent chromium oxidation. A typical starting ratio is 3 to 1 oxygen to argon/nitrogen by volume. The ending ratio can be as low as 1 to 9, oxygen to argon/nitrogen. The choice of which inert gas to use, argon or nitrogen, is based on cost and nal nitrogen content desired. Stabilized stainless steels require low carbon and nitrogen levels, for instance, so the more expensive argon must be used. It is possible to use a vacuum system to keep the partial pressure of CO low when rening with injected oxygen. This is the vacuum oxygen decarburization (VOD) process. The VOD process can achieve slightly lower carbon levels but does not achieve cleaner steel as some believe. In both processes, after nal carbon content has been achieved ferrosilicon is added to reduce the chromium in the slag and have it return to the molten steel. The excellent mixing of the slag and metal in the AOD permits this to be done efficiently. The silicon plus the manganese in the

steel combine to reduce the oxygen content of the steel to around 100 ppm. This could be further reduced by aluminum, but aluminum-based inclusions are generally undesirable. The thermodynamic activity of aluminum is considerably reduced in iron as chromium levels increase, so its role as a deoxidizer is less valuable in stainless steels. Titanium, on the other hand, is enhanced as a deoxidizer in chromium-iron alloys, and consequently small amounts of it are sometimes used as a supplementary deoxidant in alloys even though an alloy specication may not call for any. Titanium is believed to reduce hot working defects. More active deoxidants, such as calcium and magnesium, can be used when required. Also note that even if no intentional addition of metallic calcium is made, strong deoxidation with aluminum or titanium can reduce small amounts of calcium from the CaO in the slag, producing measurable calcium content in the metal. Besides carbon and oxygen, other impurities can be removed from the molten stainless. Once the steel has been deoxidized, sulfur can be readily removed by contact with a basic slag. Sulfur can be reduced to less than 0.001% in the AOD, and this excellent purity level is commercially furnished without additional price premium. Sulfur, although a harmful impurity from a corrosion standpoint, is often deliberately kept at moderate levels (0.008 to 0.015%) for tungsten inert gas (TIG) welding penetration (see Chapter 17) and at high levels (0.15%+) for machinability (see Chapter 15). These trade-offs, which are benecial to processors, should be viewed with skepticism by end users, whose product integrity is compromised. There are processing methods for which higher levels of sulfur are not necessary that are preferable to the end user while not compromising welding or machining costs. For example, machinability can be improved by calcium additions that produce malleable oxides to replace the deleterious suldes (see Chapter 15), and welding methods, such as laser welding, can be used in many cases to eliminate the need for the weld penetration enhancement of sulfur while increasing welding speeds. Phosphorus is an impurity for which no practical removal technology exists in stainless steel. Any known process to remove it rst removes chromium. Thus, it exists in almost all stainless steel at levels close to its normal specication limit, about 0.030% in austenitic alloys and 0.020% or less in ferritic alloys, which are

Chapter 12: Melting, Casting, and Hot Processing / 157

made from a higher percentage of low-phosphorus carbon steel scrap. The deleterious effects of phosphorus on corrosion are not avoided unless much lower levels are achieved. Consequently, its presence is tolerated since it has no differential effect over the range in which it is found. Heavy metals are eliminated by high-temperature AOD blowing, as is hydrogen. Care must be taken not to reintroduce such impurities after rening, which is a risk when using damp or contaminated scrap for coolant. Alloy adjustment can be done in the AOD or preferably in a treatment-and-transfer ladle. The tapped molten steel generally has excess heat from the highly exothermic rening process. This allows the composition to be measured and adjusted before it must be cast. This can be done very precisely by wire feeding of alloying elements through the slag into the heat, which can be stirred by argon bubbling via porous plugs. This technique is very effective for the netuning of reactive elements such as titanium. The rening treatments used for carbon steel and stainless steel are very similar, but there are subtle differences because of the difference in the thermodynamics of dilute solutions like carbon steel and highly alloyed, nondilute solutions like stainless steel. Table 1 shows the factors by which additions of various elements to stainless steel (j) alter the thermodynamic activity of other alloying elements (i). Equation 4 is used to calculate the activity of elements in steel. The activity coefficient varies with the concentration of alloying element x by:
RT ln i = RT ln i0 +

on interstitial solubility. The higher solubility of carbon, nitrogen, and oxygen in stainless steels is signicant. A manganese/silicon deoxidized stainless steel will still have about 100 ppm of dissolved oxygen at the freezing temperature as opposed to the less than 10 ppm of oxygen found in aluminum-killed carbon steel. This oxygen precipitates as oxides in the solid state. Vacuum induction melting (VIM) is another method of melting stainless steels. This is a nearly slag-free process, and little rening is possible. Melt purity is largely controlled by the purity of the starting material, and use of AOD master melt stock for VIM remelting is common. Limited decarburization is possible via injection of oxides such as Fe3O4 or SiO2 to create CO evolution inside the vessel. Using this technique, very low carbon levels (less than 50 ppm) are achievable commercially. Use of VIM is generally limited to high-value, high-purity, or low-tonnage melts.

Remelting
Some stainless steels and related alloys are remelted to rene composition or ingot structure. There are two principal remelt processes: vacuum arc remelting (VAR) and electroslag remelting (ESR). In VAR, the material to be remelted is cast into a cylindrical electrode and placed inside a cylindrical water-cooled vacuum chamber. A high-current direct current (dc) arc is established between the electrode and a starter plate at the bottom of the chamber. The end of the electrode is melted, and the molten drops fall through the intervening vacuum. Volatile constituents escape from the molten drops, and the puried drops collect to form a molten pool on top of the starter plate. VAR parameters are adjusted to maintain a shallow pool, which solidies in a bottom-up fashion. The shallowness of the molten pool produces a rened grain

j =1... n

RT

ln i xj

(Eq 4)

This calculation is best left to computer programs such as Thermo-Calc that have been perfected for these lengthy procedures. It should be noted that chromium, which is always present in nondilute quantities, has a powerful effect

Table 1 Inuence of alloying elements on the thermodynamic activity of carbon, nitrogen, sulfur, and oxygen
J i Al C Cr Mn Mo N Ni O S Si Ti W

O N S 0

.04 .03 ... .39

.14 .13 .11 .45

.02 .05 .01 .04

.01 .02 .03 .02

.01 .01 .003 .003

.11 0.0 .01 .06

.01 .01 0.0 .006

.34 .05 .27 .20

.05 .01 .03 .13

.08 .05 .06 .13

... .53 .07 .6

.005 .001 .01 .01

158 / Stainless Steels for Design Engineers

structure with less solidication segregation than found in typical cast product. In ESR, the material to be remelted is cast into an electrode of similar shape, but slightly smaller than the water-cooled mold. A gap between the electrode and a starter plate at the bottom of the mold is lled with a prepared slag. Typically, this slag is calcium uoride-based with high lime (CaO) content. Additional ingredients control the basicity, uidity, oxidizing potential, and other properties of the slag. A high current is used to melt the slag, which in turn melts the end of the electrode, and the molten drops fall through the slag. Reaction of the molten drops with the slag removes sulfur and some other impurities, and the puried drops collect to form a molten pool on top of the starter plate. ESR melting typically is done at a higher rate than VAR, and the molten pool is deeper. This deeper pool produces a grain structure between that of VAR and typical cast product, with commensurate intermediate segregation patterns.

Casting
Continuous slab, billet, and bloom casting have become the standard methods of making stainless steel primary products, replacing the obsolete ingot method. There are some alloys that cannot be continuously cast, but these represent a miniscule percentage of stainless production. Continuous casting produces slabs directly, thus removing the costly soaking and slabrolling processes. In a well-executed continuous casting operation, slabs are of sufficient quality that they require no surface conditioning before being hot rolled. Slabs range in thickness from 13 to 63 cm (5 to 15 in.). The segregation in continuous casters is less than in ingots because of the smaller section size. It is not eliminated, however, and certain alloying elements concentrate at the centerline, where they defy homogenization. Carbon and molybdenum are examples of alloying elements with this tendency. In properly executed continuous casting, the ladle feeds by a slide gate, or preferably a stopper rod gate, into a ceramic tube into the large tundish situated over the caster mold. The metal in the tundish is covered with a protective slag cover, and ow patterns within the tundish are designed to minimize dead spots and encourage removal of inclusions by impingement with the slag cover. The metal feeds through another

ceramic tube, called the submerged entry nozzle, into the mold, which is covered with a consumable protective and lubricating slag cover, called a mold powder. The mold powder, which melts in the mold as it is added, contains ceramics, uxes, and carbon. The level of the molten metal should be carefully controlled by ultrasonic measurement, or other methods, to prevent uctuations in level that may entrap slag in the slab surface. The entire water-cooled, copper alloy mold oscillates in a precise pattern as the solidifying strand of steel is withdrawn from the mold bottom by pinch rolls and sprayed with water to cool it. The pinch rolls apply enough pressure to slightly deform the slab. This deformation has a crucial, seldom-recognized effect. It causes a benecial recrystallization that improves hot working characteristics of austenitic and duplex alloys. In ferritic alloys, it can cause excessive grain growth, which detracts from hot workability. The initial portion of slab cast in a sequence is seldom of adequate quality to be used because of exogenous inclusions, entrapped mold powder, and non-steady-state solidication structure. The defective portion must be identied and scrapped or diverted to low-quality requirement end uses. The strand is bent from an initial slightly curved shape to at and cut into slabs. More than one heat of steel may be cast sequentially without restarting the process. This is ideal economically and for quality reasons since initial and nal segments of a casting can contain more inclusions and aberrant structure. Some end users stipulate that no rst slabs be applied to their orders. Producers generally apply rst slabs to less-critical uses or discard suspect sections of them. If casting conditions are not optimal, the result can be slabs with poor surface quality that must be surface ground. Slabs are sometimes quenched to avoid precipitation of phases; however, they may be held at high enough temperatures prior to hot rolling to stay above the temperature range in which embrittlement can occur or to stay above the temperature at which an embrittled slab can fracture. Ferritic and martensitic alloys are especially prone to these problems. There has been great interest for decades in producing stainless steel coils directly from the melt in so-called strip casters. Elimination of hot rolling could be quite valuable in stainless steel, whose hot rolling from slab can be both expensive and problematic. There are a

Chapter 12: Melting, Casting, and Hot Processing / 159

number of such machines in pilot or limited production. They have not had sufficient commercial or technical success to have become a factor in the industry. Since their development is only being undertaken by those large stainless steel producers who already have the hot rolling assets that strip casting would replace, it seems unlikely that strip casting will soon become a major factor even if it is perfected technically. Another method of shortcutting the casting/ ingot step has been perfected: the powder metallurgy approach. In powder metallurgy, the rened molten metal is atomized by gas or liquid and made to freeze into small particles. These particles, having been quenched extremely rapidly, are quite homogeneous. Powder technology methods allow for the design of alloys that would otherwise freeze with too much segregation and too coarse a structure with conventional production methods. Traditional powder metallurgy production methods are used to make small near-net shape components, avoiding most of the costly machining steps. More impressively, powder technology is also used to produce massive components. For example, very high carbon/vanadium stainless tool steel components can be made by encapsulating powder in an evacuated canister in which it can be sintered and hot worked to 100% density and virtually complete homogeneity. Chapter 9 on martensitic alloys discusses these materials.

Hot Rolling
Hot rolling remains an essential process for the vast majority of stainless steel used. Hot rolling characteristics of stainless steels vary greatly. Ferritic stainless steels are extremely easy to hot roll since they have a soft, singlephase structure at hot rolling temperatures. Martensitic stainless steels roll like their carbon and alloy steel counterparts since their microstructure during hot rolling is a moderately alloyed austenite similar to alloy steels. The microstructure during hot rolling is the crucial factor. Austenitic stainless steels have high strength at hot rolling temperatures. Furthermore, the low diffusion rates in austenite slow recrystallization so that the steel does not always soften between stands in tandem mills. This increases mill loads, and lower reductions must be taken than for alloy steels. Powerful hot strip tandem

mills that routinely roll carbon steel to 1.5 mm (0.06 in.) can struggle to attain 4.5-mm (0.18-in.) thickness for 316 stainless. The high separating forces on the hot rolling mill stands also cause greater roll deection and compression, which if not countered by roll bending or roll shifting schemes can lead to signicant variation in thickness across the sheet, as much as 0.25 mm (0.01 in.). This variation as a percentage of thickness is not reduced by cold rolling and is a major cause of tolerance loss in sheet and strip. Hot-rolled bands vary in thickness along the length of the coil because the tail end of the slab is colder and harder to roll. Coil boxes (on reversing mills) address this problem to a degree by permitting the semirolled coil to equalize in temperature. Hot strip tandem mills powerful enough to successfully roll high-quality stainless steel hotrolled bands are massively expensive and are seldom justied for the tonnage of stainless steel rolling a given melt shop produces, although rolling stainless on hot tandem mills used primarily for carbon steel can be an excellent production method. Hot Steckel mills have become the favored method of hot rolling stainless steel because their throughput better matches stainless steel melt shop production outputs. This permits the melt shop and caster to be adjacent to the hot mill, which permits energy-saving hot charging of slabs. In hot Steckel mills, typically a fourhigh reversing rougher rolls slabs to about 3-cm (1.2-in.) thick. Then, the transfer band is rolled to nal gauge on a separate reversing four-high nishing mill with coil boxes to preserve temperature. The economy of having only two mill stands makes these mills ideal for typical stainless production quantities and permits the cost of sophisticated mill capabilities, such as roll shifting, roll crossing, or roll bending, not to have to be duplicated among many stands. This is the same justication for using Sendzimir mills to cold roll stainless. In both cases, the logic applies more to austenitic alloys than to the easily rolled ferritic stainless alloys. In either case, the hot-rolled band carries a heavy, embedded scale that must be removed from the surface before further processing in most cases. Some alloys can be cold rolled in the black band state at a cost of coarser surface nish and greater rolling mill roll wear. If normal cold rolling or use as hot-rolled coil is foreseen, the hot-rolled band must then be annealed and pickled since the as-rolled hot-rolled band

160 / Stainless Steels for Design Engineers

has poor corrosion resistance, poor mechanical properties, residual cold work and hardness variations, as well as a heavy oxide layer.

Defects
Stainless steel hot-rolled bands can contain many types of defects. These are seldom seen by the end user because they are removed when they are not prevented. They do have repercussions on delivery. The major categories are: Hot mill defects Inclusion-related defects Hot ductility-related defects Stainless steel is less forgiving of hot mill faults because its surface is not removed by oxidation to the degree carbon steels surface is. Thus, a skid mark from a slab-heating furnace will remain through the hot rolling, annealing, and cold rolling processes. This is true of all hot mill scratches, gouges, digs, etc. Rolling stainless requires a different mindset than rolling carbon steel, which argues against the benets of rolling stainless on a mill built and primarily used for carbon steel. Inclusion-related defects are all essentially avoidable by using state-of-the-art technology. Protecting metal from reoxidation and keeping precise mold-level control in the continuous caster prevents all inclusions of a size that can produce a defect. Hot ductility defects are more subtle. They arise from many causes and are manifest prima-

rily as edge cracks and slivers. Edge cracks are simply a lack of ductility at the colder strip edge. Stainless hot ductility often has a narrow temperature window, and many factors can affect the size of that window depending on alloy type. The most inherently challenging alloys for hot working are the duplex alloys and the alloys that solidify in the fully austenitic state. The former has a mixed-phase structure, and the phases can exhibit mechanical incompatibility at certain temperatures. The latter alloys reject sulfur and oxygen during solidication and slab reheating to the grain boundaries, where they form very weak lms. But, even alloys such as 304 and 316 can have very poor hot ductility if they contain much sulfur and oxygen or if they are reheated for long times or at temperatures above 1250 C (2280 F), which facilitates diffusion of sulfur and oxygen to the grain boundaries and also encourages very large grains. This poor hot ductility manifests itself as slivers, which can require grinding of the entire hot band surface. These tendencies are fought by low oxygen and sulfur levels and minimal slab-reheating temperatures and times, as well as slab surface working in the caster pinch rolls. Sometimes, very poor hot working alloys are given a single hot reduction pass on a hot mill to produce a full recrystallization that disperses grain boundary-weakening elements on subsequent reheat.
REFERENCE

1. D. Peckner and I.M. Bernstein, Handbook of Stainless Steels, McGraw-Hill, 1977, p 313