Understanding 'rock candy' fracture in steel castings.

This review article offers insight into this distinct crack defect and provides tips for eliminating this occasional, but severe, production worry.

Eversince aluminum (Al) has been used as a deoxidizer for steel, foundrymen have seen the occasional occurrence of a brittle, intergranular fracture mode in castings. Known as "rock candy," this peculiar form of fracture and its threat of catastrophic consequences have resulted in a topic worthy of much investigation.

After aluminum nitride precipitation was first identified as the defect's main culprit by C.H. Lorig and A.R. Elsea in 1947, considerable attention has been paid to the phenomenon over the last 50 years, and other causes for this type of fracture have come to light.

In 1975, the Steel Founders' Society of America (SFSA), Des Plaines, Illinois, published the results of its research (Special Report No. 12) on this subject, and listed six possible causes for this type of fracture:

* complex aluminum nitride precipitates;

* carboboride precipitates;

* boride precipitates;

* alloy carbide precipitates;

* ferrite network, containing a carbide or carboboride;

* intergranular sulfides and a precipitate.

The characteristics of rock candy fracture are well known - large, smooth facets in the fracture face. A poor tensile ductility and low impact value are often associated with this type of failure. Figure 1 shows a typical rock candy fracture. The appearance of these particular fractures is typical of what is seen during the visual examination of a failed part. There is little evidence of ductility in the fracture surface, and there is a definite shiny pattern to the fracture. Individual grains are easily distinguished. Often, there is evidence of the original dendrites that formed during solidification.

The SFSA research identified several factors that increase susceptibility to intergranular fracture. These include composition of the steel; process variables; relative strength of the matrix with respect to the grain boundary; and size, distribution and morphology of the embrittling constituent.

Another factor - the mass and thickness of the casting section - should be added to the list. Aluminum nitride precipitates have been detected in a 36-in. thick casting despite the fact that the bulk Al in the heat was less than 0.02% (total spectrographic), and the nitrogen (N) level was less than 60 ppm. These aluminum nitride precipitates led to cracking in the casting during the cleaning process.

In revisiting the original 1975 SFSA research, this article reviews this rock candy type of fracture defect and how it may be reduced in production. Further, it may cause foundries to reevaluate their deoxidation and melting practices. Following is an explanation of each of the causes of rock candy and some recommendations for curing the problem on the shop floor.

Aluminum Nitride Embrittlement

The primary cause of rock candy in steel castings is aluminum nitride precipitation. Both high Al and high N contents increase the probability of encountering aluminum nitride embrittlement. Figure 2 shows the limits of N and Al that can be tolerated before rock candy is developed.

In most cases, the problem is limited to quench-and-tempered, high-strength steels. However, it also has been seen in martensitic stainless steels (specifically CA15, which is a 12% chromium grade) that were normalized and tempered. There is at least one report of this problem occurring in a WCB casting. The analysis of this part required quenching from 1650F (899C) in order to reveal the pattern by acid etching.

Susceptibility of a steel heat to rock candy can be tested in several ways. The easiest test is the simple breaking of a test bar in three-point bending [ILLUSTRATION FOR FIGURE 3 OMITTED]. The test bar (which should be at least 2.5-in. thick) must be in the quenched or quenched-and-tempered condition to detect rock candy, since SFSA work concluded that other conditions of heat treatment do not show the same degree of rock candy features. An abrasive saw cut is needed to start the fracture. It also is strongly recommended that the testing be done in an enclosure that will contain the broken pieces.

There is an ASTM A 703, Supplementary Requirement S 23 that provides an etching test with comparison photographs. The test method consists of immersing the specimen in 1:1 HCl in water at 160-180F (71-82c) for 15-30 min. Hendrix uses 1-1.5 hr for this test in order to clearly define any grain boundary precipitates. Temperature control is critical in order to prevent the formation of a black "smut" on the surface that masks results. It is strongly recommended that this testing be done under a chemical hood, due to the odor produced.

Recommendations - According to another SFSA report on the occurrence of aluminum nitride embrittlement, "There is no guarantee of preventing rock candy fracture due to aluminum nitride formation if one controls the Al and N alone." To prevent this cause of rock candy:

* keep the N content as low as possible;

* restrict Al additions to the minimum for complete deoxidation;

* maintain a high cooling rate after solidification;

* pour at 2900F (1593C);

* add titanium in combination with Al;

* homogenize at 2400F (1316c) for 3-4 hr;

* change the morphology of the nitride constituent by time-temperature manipulation.

Carboborides

The precipitation of a complex carboboride phase has been identified in some steels that exhibit rock candy features. Boron (B) is added to steel for the purpose of increasing its hardenability. Figure 4 shows the effect of 0.0025% B and the austenitizing temperature on the asquenched hardness of a 0.25% carbon steel. Very low amounts of B have a major effect, particularly in low-carbon steels. B forms stable carbides and nitrides in steel.

Recommendations - The following guidelines can prevent carboboride precipitation in the grain boundaries:

* B content should be in the range of 0.001-0.003%;

* B precipitates increase with increasing carbon (C) or B content;

* high pouring temperatures [3000F (1649C)] increase the formation of B compounds;

* high austenitizing temperatures increase carboboride precipitates;

* cool castings rapidly from the austenitizing temperature to prevent carboboride precipitation.

Boride Precipitates

Semicontinuous boride precipitates observed metallographically were identified as complex borides of the [Fe.sub.2]B type. The precipitation of this phase is not well understood. The largest contributor seems to be the B content.

Recommendations - To prevent the formation of this phase:

* maintain a B content below 0.003%;

* cool castings rapidly after solidification;

* minimize alloying elements in the steel, as the segregation of these elements become favorable sites for boride formation;

* homogenize at high temperature [2000F (1093C)] to dissolve the precipitate and quench to avoid reprecipitation.

Alloy Carbides

At least one low alloy steel that exhibited rock candy fracture had precipitates identified as complex alloy carbides of the [(FeMnCr).sub.x][C.sub.y] type. Complex chromium carbides are reported to form at temperatures of 900-1400F (482-760c) and dissolve at 1600-2100F (871-1149C).

Recommendation - Proper heat treatment to prevent these carbides would entail rapid cooling from the austenitizing temperature.

Ferrite Network Plus Precipitate

Two steels with medium manganese (Mn) (1.65 and 2.20%) exhibited rock candy features in the as-cast condition. It was reported that these alloys also exhibit this problem in the quenched and tempered condition, but not in the normalized and tempered condition. Metallographic examination revealed a ferrite network with an intergranular precipitate. The precipitate was identified as a carboboride in one case and a carbide in the other.

Recommendation - The formation of the ferrite network is a function of the hardenability and the cooling rate after solidification. Heat treatment will cure the problem.

Intergranular Sulfides and a Precipitate

Two steels [one an 8630 and the other a Mn-molybdenum (Mo) type] showed rock candy features in the quenched condition. Type II manganese sulfide inclusions were found in both samples, and carboborides were present in one sample. The prevention of Type II sulfides lies in the deoxidation of the steel. Figure 5 shows the various common deoxidizers and their effect on the sulfide shape. This chart is only applicable to steel with 0.3% C, 0.6% Mn, 0.35% silicon (Si) and 0.013% phosphorus (P). Other grades show different ranges for the formation of the sulfides.

The relative affinity of deoxidizers for the elements of interest is shown in the table on p. 49. Unfortunately, no reference exists that can quantify the effects of C, Mn and Si levels in the equation. Oxygen content, which is of prime importance, is often an unknown quantity.

Other Embrittlement Phenomena

When a failed part is examined, it is often unclear if the fracture is of a rock candy type. The following other problems can also show up as brittle fractures.

Temper Embrittlement - This problem occurs on tempering in the range of 700-1100F (371-593C). It is caused by segregation of Ni, Mn and Cr with the impurity elements P, Sb, Sn and As. Mo is added to delay this embrittlement. The best practice is to quench the steel from the tempering temperature.

Hydrogen Embrittlement - This type of embrittlement is caused by the absorption of hydrogen during melting, welding, heat treatment or in service. High-strength steels, especially those containing nickel, are more prone to this problem. Aging the steel at 400-600F (204-316C) for 20 hr/in. will restore its properties. Severe hydrogen embrittlement will produce the characteristic "fisheye" flakes in the steel. (See sidebar.)

Quench Cracking - This is cracking that occurs during quenching. It has been related to the martensite start (Ms) temperature of the alloy, as well as the heat removal characteristics of the quench tank, loading of the parts, formation of steam pockets, etc. Part geometry plays an integral part in the location and direction of those cracks.

Cracking Under Risers - This is cracking that occurs because of alloy segregation under risers or to the lack of adequate preheating in cutting or arcing. The segregation may be due to inadequate riser volume or contamination from sleeves or hot toppings. In heavy-section castings, it is critical to properly size the risers and eliminate sources of Al, P, S and C. Proper reheat for the alloy also is a must. (See sidebar.)

Relative Affinities of Various Elements for C, N, O and S.

               Carbon      Nitrogen        Oxygen         Sulfur

Aluminum        Weak     Very Strong    Very Strong        Weak
Vanadium       Strong    Very Strong       Weak            Weak
Columbium      Strong       Medium         Weak            Weak
Titanium       Strong    Very Strong    Very Strong       Strong
Zirconium      Strong    Very Strong    Very Strong       Strong
R.E.M.          Weak        Strong      Very Strong    Very Strong

This article was adapted from a presentation at the 1998 Steel Founders' Society of America (SFSA) T&O Conference.

RELATED ARTICLE: Cracking the Case of Steel Casting Fracture Defects

Editor's Note: A entire session at the November 1998 SFSA T&O Conference focused on understanding - and solving - crack-related casting defects. In addition to rock candy, hot tearing, hydrogen-assisted cracking (HAC) and under-riser cracking were examined.

Paul Rudd, Texas Steel Co./Citation Corp., Fort Worth, Texas, examined hot tear defects, which are jagged discontinuous cracks on the surface of castings that occur at high temperatures related to solidification. According to Rudd, two mechanisms occurring in solidification jointly cause hot tears. These are stress areas (caused by a restriction to contraction while the casting is cooling or by high thermal gradients) and a hot spot where the stress is concentrated.

After making sure that the defect is a hot tear ("it should be jagged in appearance, and a close examination of the fracture face should reveal an oxidized condition"), Rudd said that the process for elimination includes:

* look 90 [degrees] from the direction or axis of the hot tear to determine the direction the stress has worked to cause the defect;

* establish what caused the stress;

* find the hot spot.

Then, he said, you can either reduce the stress (by using large radii to join two sections or tapering thin sections, for instance) or reduce the effectiveness of the hot spot (through chills, cracking brackets, using specialty sands, etc.) for the specific tear. "The elimination of either one will eliminate the hot tear," said Rudd.

Rod Duncan, Casteel Technical Service, Fort Worth, Texas, described HAC (or "hydrogen embrittlement") in steel castings. The cases of HAC most often encountered in steel castings are: tensile ductility failures due to HAC "fisheyes" and delayed weld cold cracking in the weld HAZ.

To prevent HAC in steel castings, Duncan said foundrymen should reduce the H in the steel during the melting process and bake the casting to diffuse H out of the part before it can initiate H damage.

Sources of H in melting of steels, he said, include any source of moisture such as wet alloys or lime, improperly cured refractory and ferroalloys that contain dissolved H such as electrolytic chrome and nickel. Any H that is introduced in the initial charge, however, is of little consequence since H is effectively removed during a vigorous carbon boil. He said utmost care should be taken to see that all environmental sources of moisture from refractories are completely eliminated. Secondary refining processes such as argon rous plug stirring will reduce H slightly or not at all. Further, if these processes are not controlled properly, they can actually result in an increase in H.

The final tool, he said, is the H baking or diffusing cycle, which must be accomplished as soon as possible after casting of the part, before the H has had time to cause damage. Duncan agreed with Dutcher's recommendation, noting that it is most often accomplished by soaking the casting at 400-600F (204-316C) for up to 24 hr (greater for massive sections).

Kenneth Murphy, American Cast Iron Pipe Co., Birmingham, Alabama, discussed elemental segregation and the causes of under-riser cracking. Segregation, he said, is the state in which iron rejects alloying elements during solidification, causing a non homogenous distribution of the solute elements.

"Under-riser cracking can have a significant effect on quality, cost, schedule and product liability," said Murphy. "Often, there is not a single cause, but rather a synergistic combination of physical phenomena."

His paper discussed segregation effects and identified tools available to design risers, accurately evaluate costs and prevent under-riser cracking by design.

In providing tips for preventing under-riser cracks, Murphy shared 10 suggestions that can allow the foundry engineer to design economically sound process controls that can consistently prevent under-riser cracks. His checklist is:

* decrease yield about

* minimize S and P;

* avoid breaker cores;

* beware of the S/H interaction;

* control Al and N;

* control sources of contamination;

* reduce H-induced stress;

* reduce thermal stress;

* heat-treat to maximize toughness;

* use a cost analysis spreadsheet to design risers.