Porosity in Iron Castings from Mold-Metal Interface Reactions.

Last year's Molding Silver Anniversary presentation updates research on iron casting porosity with a focus on current urethane nobake technology and eliminating defects.

Editors Note--Each year at the AFS Casting Congress, presenters from 25 years ago are asked to revisit their research, updating findings as prestigious Silver Anniversary speakers. In this case, 1999 presenter Rod Naro also was honored with the Molding Methods & Materials (Div. 4) Best Paper Award for his work to uncover causes of and remedies to iron porosity defects attributed to nobake binders.

Since the original presentation on nobake binder-related porosity was made in 1974, innovations in synthetic binder technology have resulted in the widespread use of nobake molding and core-making systems, but casting defects continue to pose a problem.

The chemistry of phenolic urethane binders is essentially the same as in 1974, but the Part I resin (a poly-benzylic-ether-phenolic resin diluted 50% by solvents) has less free formaldehyde to reduce odor. In addition, the solvent system has a higher boiling point for improved environmental properties.

The phenolic urethane binder family contains four basic elements: 72% carbon (C), 8.5% hydrogen (H), 3.9% nitrogen (N) and 15.5% oxygen (O). The gases responsible for subsurface porosity in iron castings are N and H--because the high silicon (Si) content of gray iron suppresses carbon monoxide porosity. C and O from the binder usually present no problem.

The original research was aimed at determining how coremaking parameters and melting variables influence porosity and developing remedial techniques to alleviate these problems. This update examines the same questions in current formulations.

POROSITY STUDY

A cylindrical test casting (Fig. 1) was developed to observe porosity formation under various conditions. This stepped-cone configuration was selected because core decomposition gases would be generated rapidly, while the casting was still molten. Also, this design was conducive to studying section size, re-entrant angle (hot spot) and other geometric effects.

Molds were made with a zero-N nobake furan binder. The base core sand used for the experiments consisted of the phenolic urethane nobake binder (PUNB) mixed with a high-purity, washed-and-dried, round-grained silica sand. Coremaking consisted of adding Part I and the catalyst to the sand and mixing for 2 min, followed by the addition of Part II (a polymeric di-isocyanate resin diluted with 25% solvents) and mixing for another 2 min. The mix was hand-rammed into the corebox, and the stepped-cone cores were stripped within 5 min.

Gray and ductile irons were utilized, as well as a high-carbon equivalent iron (4.3 CE) inoculated with standard foundry grade [0.75% minimum Calcium (Ca)] FeSi in the ladle. Inoculant addition levels were 0.25% Si, based on the pouring weight. All heats were prepared with virgin charge materials to ensure low initial gas content.

The extent of porosity was determined by sectioning castings at several locations. Scanning electron microscopy (SEM) was utilized to observe the internal surfaces of gas porosity.

Binder Ratio

The effect of the ratio of Part I to Part II resin for PUNB binders on porosity is shown in Table 1. The results designated "1998 version" refer to casting tests performed on current formulations.

Binder ratios of 60:40 (Part I:Part II) provided sound test castings. As this ratio became balanced (50:50), trace amounts of porosity were found in a few test castings, but the majority made with balanced ratios were sound. In cases in which porosity was found, a substantial portion was surface porosity or semi-rounded holes (pockmarking). Because the binder ratio was unbalanced in favor of excess Part II(40:60 and 35:65), greater amounts of subsurface porosity formed.

The types of defects varied in intensity from nil to very severe. Although the recommended PUNB binder ratio varies between a 55:45 and 60:40, in actual practice, ratios favoring excess Part II often result from worn or defective binder pumps, air in binder lines, changes in binder viscosity from temperature and inefficient mixing. In the early 1970s, foundries would run binder ratios favoring excess polyisocyanates to facilitate the stripping of difficult cores or to increase fully cured core strengths.

New resin formulations showed little difference in casting performance compared to 1974 versions.

Binder Level

To determine the effect of binder level on porosity, cores were made with levels ranging from 1.25-3%. Although these higher levels may never be encountered in actual practice, they were intentionally selected to magnify the effect of binder level or the effect of reclaimed sands having high loss-on-ignition values. As the binder level increased at the same Part I:Part II ratio, the severity of the porosity defects likewise increased.

If sufficient amounts of evolved H and/or N decomposition gases are available to solidifying irons, porosity will generally occur, even with favorable binder ratios and relatively high pouring temperatures. Excessive amounts of dissolved gases stemming from inappropriate charge materials or liquid metal processing will likewise be more susceptible to core gas defects from absorption of H and/or N.

Casting Temperature

While both binder ratio and level affect porosity formation, their effect is temperature-dependent. Results obtained from castings poured at several temperatures and incorporating unbalanced binder ratios (favoring excess Part II) are shown in Table 2.

Pouring temperatures of 2700F (1482C) and higher (as measured in the pouring ladle) produce severe subsurface defects when unbalanced ratios are used. This is not observed with balanced ratios or excess Part I. Reducing pouring temperature at both binder levels resulted in less porosity until, at the lowest temperature, sound castings were achieved.

Pouring temperature effects also were demonstrated by pouring step cores coated with the Part II component. For these tests, pouring temperatures were 2500F (137 IC), and cores were bonded with an unbalanced (35:65 ratio) binder system containing 3% total resin. Sectioned castings were entirely sound.

The porosity-temperature dependency is illustrated in Fig. 2, in which pouring temperature is plotted against binder ratio. There is a region in which porosity forms and one where sound castings are obtained. In between these areas, porosity may or may not occur, depending on other liquid metal processing factors.

Section Size

Deep-seated, subsurface porosity usually was located adjacent to the 90[degrees] re-entrant angle or "step," and most often occurred in section thicknesses from 0.875-1.375 in. (2.22 and 3.5 cm). These locations act as localized hot spots, since a small volume of the core is heated from both sides by the solidifying iron. In thinner sections, varying degrees of surface porosity or pockmarking were found.

These defects probably were formed by gaseous decomposition products pushing away the semi-skinned-over casting surface. Since these bubbles are formed late in solidification at the mold-metal interface, there was not enough time for them to dissolve. Consequently, a depression is left in the surface, when final solidification commences. This surface porosity varied between somewhat large, semi-rounded holes extending 0.125 in. (0.3 cm) to small surface pores having no appreciable depth.

Sand Effect

The sand used in cores had a significant effect on porosity formation. Severe subsurface porosity was prevalent with silica sand, while castings made with washed-and-dried lake sand were entirely sound. The behavior of lake sand may be attributed to either its larger quantity of surface impurities, bulk impurities or greater permeability.

To determine the effect of surface purity on influencing gas porosity, an acid treatment was administered to the lake sand to remove trace impurities. The sand was soaked in a 10% solution of sulfuric acid for 24 hr, followed by a 24-hr water wash and dry. Still, no porosity was observed with the treated sand.

Because of the known effect of permeability on porosity and the potential chemical effect of sand type, several other sands having a range of compositions, permeabilities and AFS grain fineness distributions were tested for relative porosity susceptibility.

There isn't a correlation between AFS grain fineness and permeability and porosity sensitivity. The lower the impurity level and, particularly, the iron oxide content of the sand, the greater the sensitivity of the system to porosity. Although pure, round-grained sands offer outstanding core and moldmaking properties, they may not produce the best castings, as lower purity sands do.

Iron Oxide Additions

Additions of as little as 0.25% 200-mesh red iron oxide ([Fe.sub.2][O.sub.3] or hematite) inhibited the formation of all porosity in castings. Since commercial foundry grades of red iron oxide occur naturally, not all grades may work like the grades used in the experiments. Further, additions of [Fe.sub.3][O.sub.4] (magnetite) were not as effective as hematite (Table 3).

Although two of the iron oxides had similar mesh sizes (325 mesh and down), the 325-mesh red iron oxide outperformed the black iron oxide, as well as a coarser (100-mesh) European hematite of relatively high purity. Iron oxide purity doesn't affect performance or eliminate porosity. Black oxides are commonly used today because they provide reduced surface area allowing for reduced resin consumption and improved coremaking economics.

Although some iron oxides may contain various percentages of [TiO.sub.2] (titanium dioxide) in their mineralogy, it is doubtful whether sufficient Ti could be available to react with N during casting.

To determine the effect of iron oxide granularity, other grades of hematite were tested. At 1.5% and 4% levels, a coarser-grained hematite was effective in eliminating defects, even though it was randomly distributed in the core due to its large particles.

The role of iron oxide in preventing porosity has long been linked with its ability to react with silica to form fayalite, which forms a "physical" barrier, preventing gas solution. Iron oxide may somehow affect the kinetics of gas absorption by the solidifying metal, but such small additions could not form effective barriers.

To determine if a slag-type barrier at the mold-metal interface is the mechanism responsible for porosity elimination, sodium fluoroaluminate (cryolite) was added. Cryolite has a melting point of 182SF (996C) and does not react with silica, as does iron oxide, to form a slag; cryolite will liquefy in-situ to form a barrier. Additions of 0.5%, 1% and 2% were evaluated.

In these castings, veining defects were minimized but considerable burn-on was present, apparently due to severe sand fluxing. Severe subsurface porosity was found in all castings.

Binder Dispersion/Mixing

Proper dispersion of the liquid binder components on sand surfaces is necessary to produce high-quality cores and molds. Mixers that were prevalent in the early to mid-1970s often provided relatively poor blending of binders and coating of sand grain surfaces.

Although high-speed, high-efficiency sand mixers and advanced resin-metering systems were developed in the 1990s resulting in dramatically improved mixing, maintaining and cleaning the equipment is as important as it was in the 70s.

Cores were prepared in a high-intensity batch mixer--each component was mixed for 5, 10, 20, 30 and 60 sec (double for actual total mix cycle). All cores were prepared with silica sand at balanced ratios (50:50).

Cores mixed for 10, 20 and 30 sec exhibited pronounced non-uniform binder dispersion and looked "spotty." Mixing times of 80, 120 and 240 sec provided uniform results. Physical properties, such as scratch and tensile strengths of sand mixed for 40 sec and longer were not impaired, even though traces of inadequate mixing were apparent on the 40-sec mix.

Short mix cycles of 10-40 sec promoted surface and subsurface porosity. Trace subsurface microporosity was found in the remaining castings made with cores mixed for 60-80 sec. In castings with pronounced defects, porosity was formed where the solidifying casting was in contact with binder-rich areas and, particularly, those containing excess polyisocyanate. Sound castings were obtained when total mixing times ranged from 2-4 min.

Metal Composition

The type and composition of the castings poured had a significant effect on porosity formation. Porosity potential was greatest for the low-CE iron and least for ductile iron. Porosity defects in all gray iron castings formed readily with unbalanced binder ratios favoring excess polyisocyanate. Porosity in low-CE irons formed predominantly as fissure-type defects, although some rounded and irregular holes also formed.

Although it is commonly accepted that ductile iron is more susceptible to porosity defects, the tests show ductile iron castings were less susceptible to defects than the gray iron. However, most of these previous findings have been with ductile iron containing aluminum poured in green sand molds. It also is generally held that ductile irons are more prone to H defects arising from interactions with water vapor and magnesium (Mg). This is probably related to residual Mg influencing H solubility or reducing water vapor.

Since the chemistry and gaseous thermal decomposition products for PUNB binders are more complex than interactions with green sand molds, the performance of ductile iron with these binders may differ considerably. However, porosity in ductile iron should be more difficult due to the higher melt interfacial surface energy. In addition, the bubbling of Mg vapor through the metal during the nodularizing process effectively purges most dissolved gases from the metal, allowing for possible absorption of core gases without supersaturation.

Core Age

The effect of core age within the first 24 hr after strip had no effect on porosity formation. Castings poured with cores used immediately afterstrip or after overnight aging performed similarly.

If cores made with unbalanced systems were aged several days under ambient conditions, the severity of the defects increased slightly. This is related to atmospheric humidity combining with unreacted material in the polyisocyanate and forming urea structures. These substances readily break down into ammonia derivatives at high temperatures and later dissociate into nascent H and N.

ELIMINATING POROSITY

Phase two of the experiment involved investigating remedies to defects in castings poured under adverse conditions.

Ti and Zr Additions

Titanium (Ti) has long been recognized as helpful in reducing subsurface N-related porosity. Varying levels of 70% FeTi (20 mesh and down) were added to the ladle before pouring. In addition, two commercial gray iron inoculants containing Ti were examined. The effect of zirconium (Zr) on porosity reduction was evaluated by adding 0.05% Zr as FeSiZr, as well as incorporating Zr into a high-potency inoculant.

In almost all cases, adding small amounts of Ti to the ladle was effective in eliminating subsurface porosity in castings made with cores bonded with excessive Part II. In the case of 70% FeTi additions, Ti additions of 0.05% were effective in removing subsurface porosity defects; however, a considerable amount of surface porosity or small pores remained.

Since 70% FeTi may be difficult to dissolve below 2700F (1482C) resulting in erratic recoveries, two Ti-containing gray iron inoculants also were investigated. Inoculant A, based on 50% FeSi, was effective in eliminating porosity when the Ti addition level was 0.03%.

Inoculant B is based on 75% FeSi, and since these inoculants dissolve more rapidly than those based on 50% FeSi, Inoculant B was more effective at lower Ti addition rates. No porosity was found when Ti addition levels of 0.025% were employed with Inoculant B. FeSiZr was almost as effective in eliminating porosity, but somewhat higher levels of 0.05% Zr were necessary.

Inoculant C is a potent gray and ductile iron inoculant containing 30-33% oxy-sulfide-forming elements that was modified by the addition of 9% Zr (in the form of FeSiZr). With Zr additions of 0.025%, trace-to-no subsurface porosity was found. Since Zr forms more stable nitrides than Ti, more Zr must be added due to its higher atomic weight. Therefore, higher levels of Zr must be added to Inoculant C for complete porosity elimination. Although Inoculant C did not entirely eliminate porosity, it was the most effective of the three inoculants, reducing chill, and produced the most uniform microstructure consisting of 100% Type A graphite flakes.

Metallographic inspection of the castings made with 70% FeTi showed that higher addition rates of Ti (0.05% and greater) were effective in tying up N as Ti compounds (TiCN or TiN) and preventing re-precipitation as gas holes during solidification. The inoculants produced similar results. The FeTi additions were not, however, effective in preventing surface reactions associated with lustrous carbon pockmarking reactions from high pouring temperatures.

In-the-Mold Additions

Adding selenium (Se) to stainless steel castings poured in green sand molds is effective in eliminating porosity. To evaluate the effect of controlled amounts of Zr and Se on porosity elimination in gray iron, small amounts were added (8% Zr as FeSiZr) to a 9-g in-the-mold inoculating tablet. A second experiment was run with an addition of 3.3% Se to an 8% Zr-modified in-the-mold inoculating tablet.

Castings made with either Zr or with both Se and Zr still contained some subsurface porosity, probably the result of insufficient treatment alloy. The microstructures of both castings treated with the 9-g inoculating tablets were somewhat improved, containing 100% Type A graphite, compared to the standard ladle inoculation with 75% foundry grade FeSi containing 0.75% Ca, which contained some Types B and D graphite.

Core Washes

Experimental core washes were applied to cores to determine effectiveness as porosity inhibitors. Red iron oxide-bearing washes provided slight or no reduction in porosity defects. Experimental washes of aluminum and Ti powder provided similar performance.

A 100% red iron oxide wash and another prepared with sodium silicate and iron oxide completely prevented porosity, however these castings exhibited severe surface finish degradation. The sodium silicate red iron oxide wash deteriorated the casting surface slightly, but the 100% red iron oxide wash had a deleterious effect.

Results indicated that adequate amounts of iron oxide were not employed in washes; however, in experimental washes with red iron oxide, too much was added with a resultant loss in surface smoothness.

Core Post-Baking

To determine the effect of core baking on porosity elimination, several cores were subjected to post-baking or curing for 1, 2 and 4 hr.

Castings made with cores baked at 450F (232C) for 1 hr contained severe porosity defects. Intermediate times of 2 hr significantly reduced porosity. Baking for 4 hr at 450F (232C) produced a distinctive core color change to chocolate brown and had a significant effect on porosity elimination. Although such lengthy times may be impractical, higher baking temperatures or short times at high temperatures might be effective in reducing overall binder level in the core surface layers. Baking also demonstrates that some free hydrocarbons are volatilized, and N components from the Part II resin may undergo further reactions to form more stable compounds.

DISCUSSION

Pouring, solidification and high casting temperatures enhance the breakdown rate of both H and N, increasing gas solubility in the liquid metal. High pouring temperatures also have a significant effect on liquid metal surface tension, which affects porosity formation.

Chemical analyses showed that considerable pickup of both H and N occurred in the immediate subsurface layers, when conditions favoring porosity were employed. At depths 0.25 in. (0.6 cm) below the cored surface, H and N levels were low and representative of the base metal. Just before solidification, momentary supersaturation of both H and N may exist just under the casting surface.

Further, if a considerable amount of nascent N is dissolved in a casting from unbalanced binder ratios favoring excessive polyisocyanate components, even a small amount of H will lower the overall solubility of N. H may exert a catalytic effect on N to enhance porosity formation. Alloying elements have the same effect on gas solubility.

To further aggravate conditions, if the melt initially has a high gas content resulting from poor charge metallics or carbon additives, then the tolerance for additional solution of nascent mold or core gases is reduced considerably, and porosity formation is extremely likely.

Typical microstructures in sound and porosity-containing castings were taken at the mold-metal interface. In all cases, no differences in matrix structure or graphite morphology were found. Both microstructures contained the same ferritic-type matrix with Type A graphite. Although H and N are carbide stabilizers and favor formation of pearlite and other graphite structures, insufficient time was available during solidification and subsequent cooling through the transformation temperatures forsuch phases to form.

Although most gas holes exhibited a bright or shiny interior of a graphitic nature, no such films were observed. Further examination of these areas by SEM showed distinct layers of a crystalline graphitic coating in the gas holes.

Although gas holes were located just underneath the surface and most extended no more than 0.25 in. (0.6 cm), a few castings contained gas fissures almost 0.5 in. (1.27 cm) long. Because of the sub-surface nature of the defects, large amounts of alloying elements that form stable N compounds may not be needed, since only these sub-surface layers are affected.

Incorporating N-stabilizing elements or "scavengers," which include both Ti- and Zr-based ferroalloys, may offer additional possibilities for treating binder-induced porosity defects. Likewise, in-the-mold inoculating tablets incorporating Nr for N control and small amounts of Se for H control, also show promise for defect elimination.

It is not well understood how small amounts of red iron oxide were so effective in eliminating subsurface porosity in the castings. The red iron oxide may be exerting some type of "catalytic effect" on binder decomposition products that minimize or alter the generation of N and H gases.

When exposed to the sudden high temperatures of iron casting, red iron oxide readily releases oxygen. This oxygen immediately reacts with N from the binder to form stable [NO.sub.x] compounds. Since hematite has a much higher concentration of oxygen compared to magnetite and, based on its performance, this is feasible.

               Effect of Binder Ratio on Porosity Formation
Binder Level         Ratio Pt. I:Pt. II Porosity Extent
1.50% (1998 version) 60:40              nil
1.50%                60:40              nil
1.50%                50:50              nil to trace
1.50%                40:60              traces to moderately
                                          severe
1.50% (1998 version) 35:65              severe
1.50%                35:65              severe
Test conditions: PUNB binder with washed and dried silica sand; Iron
Chemistry, 4.3 CE iron; Pouring temperature,2700F(1482C)
                    Effective of Pouring Temperature on
                            Porosity Formation
Binder  Pouring
Level % Temperature (F) Porosity Extent
1.50    2780            very severe--gross w/some
                        traces of fissures
1.50    2700            severe
1.50    2625            traces
1.50    2550            none
3.0     2700            very severe
3.0     2600            moderate
3.0     2500            none
Test conditions: PUNB binder on W/D silica
sand; 4.3 CE Gray Iron; Pt. I:Part II
ratio held constant at 35:65
             Effect of Iron Oxide Type on Porosity Elimination
Binder % Iron Mesh                     Porosity
Level  Oxide  Size Oxide Type          Extent
1.50%  0.0%   --   none
1.50%              [Fe.sub.2][O.sub.3] severe
       0.25%  325  (red)               none
1.50%              [Fe.sub.3][O.sub.4]
       0.25%  325  (black)             severe
1.50%  0.25%  100  hematite ore        severe
Test Conditions: PUNB binders on W/D silica sands; 4.3 CE
iron; 35:65 Ratio Part I: Part II; Pouring temperature, 2700F
[Fe.sub.2][O.sub.3]assay: 87% [Fe.sub.2][O.sub.3], 8.0%
Si[O.sub.2], 2% A[l.sub.2] [O.sub.3], Balance not reported.
[Fe.sub.3][O.sub.4]assay: 62% [Fe.sub.3][O.sub.4], 1.5% Si[O.sub.2],
4% [Al.sub.2][O.sub.3], Balance not reported.
Hematite ore: 92.5% [Fe.sub.2][O.sub.3], 4.75% Si[O.sub.2], 1% [Al.sub.2]
[O.sub.3], Balance not reported.