Evaluating sources of inclusion defects: SEM-EDS reference data can greatly assist metalcasters in evaluating green sand cast iron inclusion defects and, thus, help create countermeasures to decrease the rate of inclusion errors.

Inclusion defects are among the most frequently observed errors in castings produced in green sand molds. Sand and slag are the principal generators of inclusion defects, but inoculants, refractories, smelting aids or slag removers also can be at fault. In the molten metal, the defect sources undergo chemical changes that tend to obscure the identity of their source. Because these inclusions form slag when heated by the melt, visually identifying the defect producer is difficult. In addition, gas released from the inclusions may create holes very similar to pinholes formed in a solidification film.

In order to address this situation, a series of model inclusion defects was formed by immersing material samples that may cause defects in molten metal. Scanning electron microscopy with energy dispersion spectrometry (SEM-EDS) evaluations were used for analyzing sample defects in cast iron from green sand molds, and the data obtained served as a reference for identifying defect sources. This, in turn, allowed for the development of measures to reduce inclusion defects.

This article investigates how SEM-EDS test data can be used for identifying inclusion defects in green sand iron castings and for evaluating potential defect sources.

Include Reference Data

To obtain the reference data, a number of possible sources of defects were chosen to be placed in a molten metal bath of class 25 gray iron (Tables 1 and 2). The green sands and silica sands were selected so as to have different silicon dioxide (Si[O.sub.2]) contents. Samples of these materials weighing about 5 g each were immersed for 30 sec. in a 25 kg bath and fused and maintained at 1,673K (1,400C; 2,552F) in an induction furnace.

Figure 1 shows a schematic representation of the immersion process. Output of the furnace was attenuated during the immersion periods of the samples. Samples of the two inoculants, which also weighed 5 g each but did not float on the melt, were placed at the bottom of a hollow mold measuring 86 nun by 120 mm, and their residue was collected after pouring.

[FIGURE 1 OMITTED]

Once the testing procedure was complete, it was found that all the samples had large amounts of silicon (Si), aluminum (Al) and oxygen (O), and some of the samples' physical appearances were similar to others. Data was established as a reference in evaluating defects found in cast components.

Know the Source

To utilize the previous findings as reference data, two different investigations were completed in a metalcasting facility--one focused on sand defects (Product 1), and the other focused on oxide defects (Product 2). Both sample casts were bottom fed with class 25 gray iron into a coreless, vertically parted mold.

Sand Inclusions (Fig. 2)--In this test casting, more defects were formed in the items located in the lower portions of the mold (B, C, F, I, L). The defects. which formed on the casting surface, contained inclusions; however. they were not visually identifiable.

[FIGURE 2 OMITTED]

SEM-EDS data of the defects indicated the inclusions had a size similar to that of the silica sand grains, and comparison with the reference data revealed that the inclusions were actually sand grains. The manganese (Mn) content also was similar to that of Green Sands 2 and 3, but partial fusion of the grains and slag formation prevented visual identification of the inclusions.

The defects in product 1 were attributed to green sand grains in the spree that were caught in the melt by an automatic casting machine equipped with a stopper nozzle used for high-speed casting. However, this problem was solved by widening the runner under the sprue to allow the sand grains in the melt to float up, resulting in a large decrease in the defect rate.

Oxides--In this test (Fig. 3), the components that had the most defects (B, C, G. K, L, U, W, Y) were located at the outside of the top two stages. The Product 2 defects, which formed on the casting surface, appeared to he similar to those in Product 1, having a smooth surface and differing in color from the skin.

[FIGURE 3 OMITTED]

The defects here were found to not contain any inclusions, but the oxides [Si.sub.2], ferrous oxide (FeO), and alumina ([Al.sub.2][O.sub.3]) gave them a similar appearance to inclusions. It was concluded that these defects were pinholes formed by oxidation of the so-lidified skin of the casting.

To prevent these defects, the melt temperature on tapping and pouring was increased based on the thought that oxides are formed at lower temperatures. Additionally, the coke bed layer thickness and the amount of coke added in the cupola were standardized to lower the oxygen concentration in the melt. These adjustments helped to significantly decrease defect rate, as well.

Augment Inclusion Analysis

Even though the two case studies from the facility led to the improvement of the castings produced, the reference data can assist in analyzing and even decreasing a number of other inclusion defects. The following examples were studied in an actual metalcasting facility.

Green Sand--Sand inclusions were observed in a grade 25 gray iron component cast in a green sand mold with a shell-molded core. The defects were found on the inner side of the central part of the cast body in contact with the core. The SEM image showed inclusions containing sand grains, and the grain surface showed evidence of fusion and had fairly large pores. The EDS data indicated high aluminum content. These results suggest that the defects were caused by green sand grains, while the pinholes were considered to be formed by carbon monoxide (CO) released from fused sand grains.

Green Sand from the Core--Sand inclusions in a grade 25 gray iron component cast in a green sand mold with coldbox and shell mold cores were observed immediately under the casting surface in the lower mold close to the weir. Either the green sand mold or the coldbox, which constitute the runner, may have provided the grains that caused the defects. The SEM image, showing dispersed sand grains with smooth surfaces, suggested that the cores were the source of the grains.

Inoculant--The same casting as the previous example also had slag inclusions on its surface (above the sand inclusions). They were characterized by craters and slag-like substances, (dissimilar from sand) observed in the SEM image. EDS analysis showed the presence of barium (Ba), a component of the inoculant, along with a fairly large amount of calcium (Ca). This indicated that the defects were caused by slag originating from inoculant residue.

Stream Inoculant--Inclusion defects were found on the interface with the lower part of a 65-45-12 ductile iron component cast in a green sand mold with a shell mold core under stream inoculation. Craters observed in the photograph, absence of nonmetallic inclusions as indicated by the SEM image and the presence of Si and O, as shown by EDS, suggested that the defects were due to oxidation of Si in the stream inoculant.

Nodularizer--Inclusion defects in a 60 40-18 ductile iron casting from a green sand mold with a coldbox core were found at the core interface. The SEM image showed nonmetallic grains, partially mined into slag, within the defects. No evidence of gas release was observed, but EDS data revealed plenty of magnesium (Mg), Si, Al, Ca, and O. These findings suggested that nodularizer components were responsible for the defects.

Ceramic Filter--Defects were observed in a grade 25 gray iron component cast in a green sand mold with shell and coldbox cores. The defects were located on the bore surface of the casting. The SEM image showed a number of nonmetallic inclusion grains. There was a significant presence of AL, which most likely originated from the ceramic filter be cause separate analyses showed higher Al content in the filter than in the green sand. These defects were effectively eliminated by replacing the ceramic filter with fine grains with at new filter that consisted of coarser grains.

Pinholes on the Casting Surface Similar to Inclusions--The pinhole detects found on the surface of a grade 25 gray iron casting formed in a green sand mold without a core appeared similar to inclusion defects. The deep pores prevented visual recognition of inclusions, and SEM images showed graphite film and manganese sulfide (MnS) around the deflects. These defects were likely generated by the formation of pinholes by gas bubbles nucleated at the MnS particles and subsequent crystallization, in the pinholes, of carbon contained in the melt.

Data for All

Although these studies were conducted using iron green sand castings, such analysis could greatly assist a facility regardless of its casting or material methods. These examples demonstrate that the SEM-EDS reference data are an effective tool for identifying inclusion defects in castings. If properly observed, countermeasures can be developed to lessen the number of defects and improve the overall quality of your castings.

Table 1. Chemical Compositions of Mineral Types For
Examination (Mass %)

                       Si[O.sub.2]   A[L.sub.2][O.sub.3]

Green Sand-1               92                4.3
Green Sand 2               88                6.8
Green Sand-3               86                8.1
Green Sand-4               87                6.3
Silica Sand-1              99               0.37
Silica Sand-2              88                5.6
Slag (ladle)               31                 12
Slag (stopper)             15                 42
Slag Disposal Agent        75                 14
Ceramic Filter             86                 13
Adhesive for Core          64                9.1

                       [Na.sub.2]O   [K.sub.2]O   MgO    CaO

Green Sand-1              0.59          0.32       1     0.39
Green Sand 2                1           1.3       0.94   0.58
Green Sand-3               1.2          0.69      1.2    0.6
Green Sand-4               1.4           1        1.4    0.61
Silica Sand-1             0.04          0.18       --    0.02
Silica Sand-2              1.1          2.1       0.18   0.63
Slag (ladle)               1.7          0.42       2     4.5
Slag (stopper)            0.46          0.37      0.07   0.62
Slag Disposal Agent        3.6          4.7       0.13   0.89
Ceramic Filter            0.13          0.04      0.03   0.03
Adhesive for Core          14           5.9       1.5    1.1

                       [Fe.sub.2][O.sub.3]   TiO    Ba0    MnO

Green Sand-1                  0.90           0.09    --    0.02
Green Sand 2                  0.94           0.11    --    0.02
Green Sand-3                   1.7           0.6     --     --
Green Sand-4                   1.1           0.13    --    0.03
Silica Sand-1                 0.05            --     --     --
Silica Sand-2                  1.5           0.35    --    0.03
Slag (ladle)                   30            0.48   1.8     12
Slag (stopper)                 12            0.82   0.27    23
Slag Disposal Agent           0.74           0.11   0.11   0.12
Ceramic Filter                0.13            --     --     --
Adhesive for Core              2.2           0.9     --    0.12

Table 2. Chemical Compositions of Inoculants Used For
Examination (Mass %)

               Si     Al      Ca      Ba     Sr     Mn     Mg

Inoculant-A   84.7   1.72    1.6     1.19    --    0.14   0.04
Inoculant-B   88.5   0.15   0.04 -    --    0.64   0.08    --

               Ti     Fe

Inoculant-A   0.07   10.5
Inoculant-B   0.03   11.2

Table 3. Reference Data Found From SEM-EDS Evaluations

SEM

Green Sands: granular shape before immersion
Silica Sands: grains with smooth surface
Ladle slag: foamed and fused textures.
Stopper slag: foamed and fused textures.
Slag remover: foamed and fused textures.
Ceramic filter: partial blistering.
Adhesive for cores: foamed surface.
Inoculant A: blocky feature.
Inoculant S: blocky feature.

EDS

Green sands: presence of Si, Al, 0, Na, Mg, K, and Ca in the component
  minerals; low-silica sands also contain Mn.
Silica sands: presence of Si, Al, 0, and K
Ladle slag: presence of Si, Al, 0, Mn, Ca, and Ba; Mn exists in a
  large amount.
Stopper slag: presence of Si, Al, 0, Mn, Ti, and Cr; Mn exists in a
  large amount.
Slag remover: presence of Si, Al, 0, Na, and K.
Ceramic filter: presence of Si, Al, and 0 only.
Adhesive for cores: large amount of Na, in addition to Si, Al, and 0.
Inoculant A: presence of Si, Al, 0, Ca, Ba, and Fe.
Inoculant S: presence of Si, Al, 0, Ca, Cr, Mn, and Fe.

This article was adapted from a paper (04-017) presented al the 2004 Metalcasting Congress.

For More Information

"Observation of Pinhole Defects in Cast Iron Castings by Surface Analysis, "Y. Kurokawa, H. Kambayashi, N. Ozoe, H. Ota, H. Miyake, 2003 AFS Transactions, Paper No. 03-011.

H. Kambayashi, H. Une and Y. Kurokawa, Tsuchiyoshi Industry Corp., Shimane, Japan and Hiroshima, Japan; T. Ito, S. Mikamoto, Yoshiwa Industry Corp., Shimane, Japan; H. Miyake, Kansai Univ., Osaka, Japan