Adjustments in additives and binders can lead to significant differences in mold properties and quality in steel castings.

Ralph Showman and Eric Scheller, ASK Chemicals, Dublin, Ohio

(Click here to see the story as it appears in the December issue of Modern Casting.)

When molten steel is poured into a chemically bonded sand mold to produce a casting, heat moves rapidly from the liquid metal into the surface of the sand and produces steep thermal gradients. As a result, the sand grains will expand and the organic binder in the mold will pyrolyze, producing carbon monoxide and other gasses that can cause defects.

Sand additives are commonly used in molds and cores to reduce defects like veining, metal penetration and unacceptable surface finish. Historically, iron oxides have been the additives of choice, primarily red iron oxide (hematite, Fe2O3) and black iron oxide (magnetite, Fe3O4).  However, these options can negatively impact the casting in terms of cost and quality. Engineered sand additives (ESAs) have been developed that are promoted to lower costs, improve veining resistance and/or reduce gas defects, but little comparative data is currently available to the steel casting industry. 

A recent research project compared additives and their effects on both the mold/core properties and the final casting’s quality, exploring which sand additive should be used for specific casting results and what, if any, tradeoffs can be expected on mold/core properties. Experiments were conducted to measure the performance of several ESAs and red and black iron oxides. The impacts of each additive on mold/core properties and casting quality were measured in order to provide some insight on which sand additive should be used to achieve particular casting results.

Common Sand Defects

Veining: Veining defects result from the expansion and contraction in silica sand when it comes in contact with molten steel. Sand additives can modify these reactions, but they also can affect negatively mold/core strength, permeability and benchlife. The challenge has been finding an additive with the best anti-veining qualities with minimal impact on other properties.

Metal Penetration: Also known as burn-in/burn-on in less severe cases, metal penetration occurs when liquid metal fills the small voids between the grains of a sand mold or core. This bonds a layer of sand to the surface of the casting, increasing cleaning room time and cost.   

Penetration, which can be influenced by sand additives, can occur by both mechanical and chemical mechanisms. An additive with a particle size smaller than the sand grains will tend to fill voids and reduce mechanical penetration. Other additives, such as iron oxide and fluxing ESAs, may promote the formation of fayalite (Fe2SiO4) and increase chemical penetration.

Surface Finish: Surface roughness of a steel casting depends on the sand particles of the mold/core. Larger particles will produce a rougher surface than finer grains. The surface tension of the liquid metal and its ability to smooth over the small imperfections on the mold/core will impact the eventual finish. Sand additives will affect the surface finish similarly to metal penetrations. Fine material will tend to fill in imperfections on the mold/core surface. Carbonaceous additives may positively impact metal surface tension while oxides and fluxes may reduce surface tension.

Carbon Pickup: Steel castings can pick up surface carbon created by the decomposition of an organic binder. Carbon pickup can be controlled through a number of different process variables, including binder type and percentage. Sand additives also can have an effect.  Additives like iron oxide that release oxygen at casting temperatures can remove some of the carbon, while others may promote carbon pickup if they contain carbon or if they reduce mold/core permeability and retard the escape of the carbonaceous gasses.

Gas Defects: Gas defects can occur in steel castings by two mechanisms. If gas pressure at the mold/metal interface is higher than metallostatic pressure, gas can push into the liquid metal to create blow defects. If the mold atmosphere contains gasses that are soluble in the liquid metal, the steel can absorb additional gas like hydrogen or nitrogen that may come out of solution during solidification.

Testing Procedures

Testing was conducted using low carbon steel (ASTM A216 WCB) to produce step-cone castings (Fig. 1). Cores were made using two different sands (W410, AFS GFN of 51 and W460, AFS GFN of 34) and phenolic urethane nobake (PUNB) binder levels of 1% (considered typical) and 2% (to simulate higher LOIs as would be expected with mechanically reclaimed sand). No additives or coatings were used for the baseline cores.

Castings were poured at 2,950F (1,510C). Following cooling and shakeout, the risers and gating were removed and the castings were sectioned and sand blasted. The castings were then rated on a five-point scale for veining, penetration and surface finish (Table 1).

Differences between the surface structures of the casting were immediately noticed. The two castings produced with the W460 coarser sand had much rougher surfaces and contained burn-in or trapped sand grains.  The two castings produced with 2% binder showed more pearlite in the surface layer, indicating surface carburization. 

Experiments were conducted to compare red and black iron oxide to six other materials and engineered sand additives (Fig. 2) at two different levels.

• RIO: Red iron oxide or hematite (Fe2O3) with a fine powder.
• BIO: Black iron oxide or magnetite (Fe3O4) with a relatively fine powder, but coarser than RIO.
• SX: A synthetic ESA consisting primarily of iron oxide, with a grain size similar to sand and a rounded particle shape.  Typical addition rates are in a range of 2-8% and it reportedly has low impact on mold/core strength. 
• Additive V2: A mineral ESA with some red iron oxide that is a fluxing material with a lower impact on mold/core strength than RIO or BIO.  
• Additive I900: A mineral ESA with some red iron oxide. 
• V400: A mineral ESA based on red iron oxide with a mixture of hematite, magnetite and other minerals, though its particle size is much larger than RIO or BIO. 
• V2003: A mineral ESA containing spodumene, a lithium ore.
• IL: A naturally occurring mineral called ilmenite containing primarily iron and titanium oxides. 

Core Test Results

Work Time/Strip Time Analysis: Work time/strip time (WT/ST) analysis was measured on each mix and cores were produced and tested in triplicate and averaged. The data was analyzed from the DOE results for level sums for each variable (Figure 3).  The V400 additive had the greatest effect, extending the WT/ST of the mix very significantly and the RIO and V2 additives had a lesser, but significant effect. The sand additives type had the greatest overall effect on the WT/ST with binder level having a smaller, but significant effect. The sand type had minimal effect on the WT/ST, high binder levels reduced WT/ST, and the additive levels had little effect. The pooled error was relatively large at 13%. This may indicate variation from the test method using the “B” hardness scale rather than effects of the variables.

Tensile Test Analysis: Level average charts (Figure 4) also were used to analyze the tensile data. The RIO, V2 and V400 additives reduced tensile strength while the other additives had lesser effects. Higher binder levels greatly increased strength, which suggests the negative effects of some sand additives could be offset with somewhat higher binder levels.

Additive type had the greatest impact on the one-hour tensile strength, but a reduced effect on the 24-hour strength.  This may have to do with related effects on WT/ST.  Higher binder levels had the expected significant impact on strength with only minor effects from sand type and additive levels.  The humidity resistance was impacted by binder level, with all other factors having only minor effects.

Permeability Analysis: Level average charts for core permeability are shown in Fig. 5. Sand additives had a minimal effect on permeability.  While sand additives certainly can affect the permeability of a mold/core, usage amounts of only 2-4% of the weight of the sand produce only small effects. The sand type or GFN had the overwhelming effect, with the coarser sand yielding much higher permeability.

Smoke Opacity Analysis: The level average chart for smoke opacity (Figure 6) showed a surprising result.  Several of the sand additives reduced the opacity somewhat, but the V400 reduced the opacity about 50%.  As expected, the % binder had a major impact.

Casting Results

Following shakeout and cooling, the castings were sectioned through the parting line of the cores and sand blasted to remove sand and oxidation. The cored surfaces were then evaluated for veining, penetration and surface finish using the 1-5 scale (1 being preferred), shown in Table 2. None of the castings showed signs of gas or penetration defects, so all castings were rated a “1.” 

Level averages for the effects on veining were plotted for the different variables (Figure 7).  BIO appeared to have the greatest effect on veining with I900, V400 and V2003 close behind. The RIO and IL appeared to have the least effect. Lower binder levels produced less veining as did higher additives levels.

The effects on surface finish were plotted for the different variables (Figure 8). The relative differences between castings and level averages were small, but the BIO, V400 and IL seemed to produce the best ratings. The finer sand also produced a better surface finish.

Surface Carburization Analysis: Sections were cut from each casting to compare casting microstructures to the baseline castings. Quantitative evaluation was difficult, but several differences were noted.  The castings shaken out in an hour had smaller grain sizes than those cooled for 24 hours. This would have been the result of faster cooling through the eutectoid transformation. The photomicrographs also confirmed the surface finish results with the casting from the cores with the coarser sand showing higher peaks and valleys and more retained sand grains. 

Surface carburization as measured by the percent pearlite was much more difficult to evaluate. Many of the castings showed increased ferrite on the surface, indicating decarburization. However, the decarburization appeared to be more prevalent on the castings produced with cores with the lower binder levels.

Only the castings produced with V2003 and illmenite showed appreciable surface carburization. The interior structures were similar to that of the other castings, but these two additives contained only very low levels of any iron oxide materials. These castings also showed more carburization on the castings made with the cores with higher binder levels.

The results of these experiments show that different sand additives should be used for specific mold/core properties.  However, the effects of the sand additives on mold/core properties outlined by the results of this study are limited to the PUNB systems, as other binder systems have different mechanisms of reaction.