Unraveling Compacted and Nodular Cast Iron Porosity: Case Studies Approach

Abstract

Porosity is the culprit for a large fraction of scrap in cast iron foundries, resulting in significant environmental and productivity losses. The present work focuses on characterizing and explaining porosity defects in industrial compacted and nodular graphite cast iron components, utilizing current literature for reference. The goal is to identify existing knowledge gaps in the field, fostering further research work. Complex-shaped castings were sampled from three foundries, weighing between 100 and 300 kg. These were carefully selected to capture recurring defects during stable production. The mechanisms behind these defects were discussed, and the findings were compared to the literature. Scanning electron microscopy (SEM) was used to investigate the inner surfaces of the pores with secondary electron imaging. The surrounding microstructure was captured with optical microscopy in combination with image analysis, where panoramic images and nodularity maps were built. Ultimately, etching based on Si segregation was employed. The results suggest that the understanding of pore surface film formation remains limited, particularly regarding graphite film formation. Notably, the observations reveal a multitude of previously unreported graphite structures within the pores, some with particles in their centers containing Ce, Ca, La and S. These novel structures can provide additional insights regarding pore formation chronology.

Introduction

Recent political developments seek to cut greenhouse gas emissions by 50% in 2030 and reach net zero by 2050. For heavy-duty trucks, this means more efficient powertrains are required. Traditionally, engine blocks and cylinder heads were cast in lamellar graphite iron (LGI) due to its ease of casting and high thermal conductivity, which provided great resistance to thermo-mechanical fatigue. However, since the combustion efficiency increases with the peak firing pressure, LGI struggles to perform in higher-pressure engine designs.¹ With improvements in process control, compacted graphite iron (CGI) becomes a viable alternative, where its strength enables higher firing pressures, as well as adequate thermal conductivity for engine applications.² Compacted graphite can be seen as the intermediate shape between spheroidal and lamellar graphite, but as its production is quite a novel affair, current knowledge lacks in some respects, particularly regarding porosity defects.

Taking into account graphite’s lower density when compared to the surrounding matrix, solidification shrinkage should not occur, according to the Fe-C phase diagram, on carbon contents above 3.8wt%; however, considering that spheroidal graphite iron (SGI) solidifies as a divorced eutectic (where graphite is completely enveloped by austenite), graphite expansion can contribute to an expansion of the casting cavity volume, instead of having a feeding effect.³ Using rigid molding materials helps mitigate this phenomenon by resisting deformation caused by the expanding casting. Furthermore, with an increase in carbon content beyond the eutectic composition, the likelihood of primary graphite formation intensifies. In such cases, the expansion contributes exclusively to increasing the casting volume and not to feeding.⁴ On the other hand, LGI solidifies in eutectic cells where the graphite tips are kept in direct contact with the melt. As a result, the expansion pressure has a greater chance of feeding liquid to zones in need.⁵ Alternatively, Chen et al.⁶ has shown with quenched samples that CGI has direct contact with the melt during the initial solidification stage, being encased later by austenite. These results could be caused by quenching artifacts, but it is plausible to consider that trace element segregation may help maintain contact with the liquid, decreasing the solidus temperature and locally generating liquid channels. Considering these factors, ensuring proper contact between the graphite and the remaining liquid becomes crucial in facilitating expansion and pressurizing it toward shrinking regions.

In instances where feeding is inefficient, microporosity can be found in the casting, presenting a complex shape due to its formation at a late stage of solidification. It usually involves a nucleation and growth stage in the remaining interdendritic liquid; however, other mechanisms may skip nucleation altogether, e.g., a permeable columnar zone enabling gas aspiration from the surrounding atmosphere because of the volumetric shrinkage of the primary austenite. This mechanism is more common in LGI due to its prevalent casting in hypoeutectic composition, granting a greater amount of primary phase.^7,8 Overall, microshrinkage is generally associated with both volume changes and the decreasing solubility of gases such as hydrogen and nitrogen during cooling—especially during phase transformations, where the decrease can be abrupt. If the dissolved gases exceed the solubility limit, pores may nucleate, significantly decreasing the casting’s mechanical properties. In fact, nitrogen’s reported limit for pore nucleation has been below its solubility limit of 150 ppm, with pores being detected with nitrogen content between 110 and 150 ppm.^9,10 This limit for pore nucleation varies according to the cooling rate within the casting, as thick sections are more affected—with microsegregation playing a role in the local nitrogen content—a role that is still unclear, as the partition coefficient’s dependence on chemical composition and segregation of other elements has not been properly determined.¹¹ Kagawa et al. observed that both carbon and silicon contribute to decreasing nitrogen solubility, both in the austenite and liquid phases. Nevertheless, owing to the difference in the rate at which solubility decreases in each phase, it was found that silicon contributed the most to nitrogen segregation toward the liquid, facilitating reaching critical contents for pore nucleation.¹²

During solidification, austenite dendrites extend and coarsen, forming a cohesive network that hampers feeding. Thus, the graphite expansion may instead be directed toward the surrounding austenite, which deforms and increases the total casting volume. Vazehrad¹³ has proposed a method of characterizing pores regarding their formation interval during solidification. If the pore is formed very early in the solidification, it would showcase a smooth, inner surface, commonly associated with gas-induced pores. Alternatively, pores regularly associated with shrinkage porosity contain dendrites on the surface. The degree of coarseness of these dendrites can be used to assess when in the solidification the pore forms, since coarsening occurs due to an extended contact with the liquid. Pores formed immediately after austenite nucleation will contain fine dendrites, whereas pores formed at later stages will contain coarsened dendrites.

The process by which pores nucleate and grow is rather complex and has been approached by several authors. The long-standing perspective that pores are either exclusively gas or shrinkage related endures in current practice. However, it is open to challenge. Stefanescu¹⁴ has described porosity as a function of the combined pressures induced by the different elements in effect. On one side, gas content and shrinkage provide the driving force for porosity—on the other, the metallostatic and atmospheric pressures, the pressure induced by the surface tension of the pore, as well as the pressure resulting from the graphite’s expansion. It forms an elegant way to view porosity, as it avoids allocating each pore to a specific gas or shrinkage categorization. Instead, it highlights how both can contribute to the same pore. Campbell¹⁵ then contributes by discussing the high tensile strength of liquids—far higher than the hot solid structure. From both, one can reach the conclusion that pores either nucleate heterogeneously on inclusions from gas or result from gas aspiration, bypassing nucleation altogether. Even though shrinkage may be the root cause for generating the low-pressure zone, it must be supported by gas to generate a pore, because in the absence of gas, the liquid will prevent discontinuities from being formed, caving in the casting skin instead.¹⁴ In the author’s view, it is then difficult to attribute pores to only one of the two causes, as it is very likely a combination of both, where the later the formation, the more the shrinkage contributes and vice-versa. Other authors have likewise called for a better description of the role of gas in shrinkage porosity.¹⁶

In LGI, previous experiments using Direct Austempering After Solidification (DAAS), a technique that preserves the austenitic solidification structure, have shown that the primary austenite has a large influence on defect formation.¹⁷ In fact, through Electron Back Scattered Diffraction (EBSD) austenite grain boundaries were identified and microporosity was located between grains.^18,19 Two important details were taken from these experiments: microporosity can wrongly be assumed to appear between eutectic cells while it is, in fact, between primary grains; the thermal center in the casting can shift during solidification, a phenomenon called hotspot migration, allowing remelting of columnar zones, or hampering its growth. This effect would then either provoke yielding during shrinkage or the introduction of surrounding gases by an increased permeability caused by columnar coarsening. Coarsening of the columnar skin decreases the specific cross-sectional area of the dendrites, leading to a facilitated fluid flow through the dendritic region.²⁰

The aim of this study is to identify opportunities for future research which show potential to advance the understanding of porosity defects in compacted and nodular cast irons. Our approach involved a metallographic examination of industrial castings, focusing on features associated with pore formation which lack comprehensive explanations in existing literature. Interpretation of novel features is discussed, and suggestions are made regarding the need for future research.

Materials and Methodology

The samples used in this case study were supplied by three different industrial partners—cast iron foundries producing components between 100 and 300 kg of slightly hypereutectic composition. Some castings were CGI (EN-GJV-450), while others were SGI, all containing defects that warranted scrapping the component. Most pores were found between 20 and 40mm from the mold walls. The samples were selected based on frequency, encompassing solely commonly occurring defects. Collected samples are subdivided by location in the casting and listed in Table 1.

Table 1 Sample Table, Where Each ID Represents a Unique Location within a Casting

Both a bandsaw and an abrasive disk were used to collect the samples. In the case of porosity analysis, resin mounting is often ill-advised if the inner surface is of interest. Thus, the freshly cut surface was used for analysis in a TESCAN Lyra 3 SEM, with a secondary electron detector operating between 15 and 20 kV. Whenever applicable, semi-quantitative chemical composition analysis was conducted using Energy-Dispersive X-ray Spectroscopy (EDS).

After an initial SEM analysis, the samples were hot-mounted in 50 mm cylinders, using a Struers Citopress-1 with a mix of Durofast for edge retention and Polyfast resin. A Struers Tegramin-30 automatic polisher was used for grinding and polishing, under the following sequence: SiC abrasive foils (#120–#320–#500–#800–#1000), Largo 9 µm, Dac 3 µm, Dur 1 µm, Nap ¼ µm. The polishing steps were alternated with ethanol ultrasound cleaning, as well as gently rubbing any debris off with an ethanol-soaked cotton pad.

Sample etching was performed using a mixture of picric acid, distilled water, NaOH and KOH.²¹ The etchant was heated up to 110 °C and the samples were submerged for 3–8 min, depending on chemical composition, matrix, and mass. This etchant is based on Si microsegregation patterns. As the liquid solidifies, alloying elements partition between the liquid and the austenite. For cast iron, the high C content potentiates inverse Si segregation, where it accumulates in the solid austenite. Si exhibits a higher affinity for bonding with the solidifying iron when compared to C, hence having a graphitizing effect.²² Kagawa et al.²³ has shown that Si starts segregating toward the austenite well below the eutectic content of C. Depending on local Si contents, the film generated by the etchant will vary in thickness, where higher Si will be colored light blue and low Si will be colored orange, indicating early and late solidification respectively. Vazehrad et al.²⁴ has demonstrated experimentally that Si composition has a close correlation to the color scale obtained in etched samples.

Microstructural analysis was done using an Olympus DSX1000 optical microscope, where panoramas were built by taking automated images along the grid of interest with a 10% overlap. Then, using in-house custom Python code, nodularity was mapped along the samples. The mapping involves subdividing the panoramas into squares of a predetermined size—200 by 200 µm. Afterward, the nodularity for each subdivided square was calculated in agreement with ISO 16112:2017. The particles were identified, and their roundness was calculated by dividing their area A by the area of a circle A_m with a diameter corresponding to their maximum Feret diameter, l_m:

$$\begin{array}{*{20}c} {{\text{Roundness}} = \frac{A}{{A_{m} }} = \frac{4 \times A}{{\pi \times l_{m}^{2} }}} \\ \end{array}$$

(1)

Then, the particles were classified as nodular (> 0.625), intermediate (0.525–0.625) or compacted (< 0.525) according to their roundness. The nodularity was calculated by dividing the area of the nodular particles A_nod plus half the area of the intermediate particles A_int, by the total area of all the particles, A_G:

$$\begin{array}{*{20}c} {{\text{Nodularity}} = \frac{{\sum A_{{{\text{nod}}}} + 0.5 \times \sum A_{{{\text{int}}}} }}{{\sum A_{G} }}} \\ \end{array}$$

(2)

Following the calculation for all the sub-squares, the panoramas were rebuilt with each sub-square being assigned a color corresponding to a nodularity color scale.

Results and Discussion

Different topics of discussion were identified in our observations, warranting subdivision of the results into the following sections: “Graphite Structures”, “Pore Atmosphere and Graphite Films”, “Pore Chronology”, “Devious Microstructure” and “Casting Skin”. Thus, in these sections, different samples will be highlighted and examined concerning the literature.

Graphite Structures

In location A1, two distinct types of pores were identified, each displaying different characteristics that suggest formation at different stages of solidification—one with a dendritic surface, indicating formation at a later stage, and the other with a smooth surface, indicating early formation in connection with gas ( Figures 1 and 2 respectively).

Figure 1

Picture and SEM image of A1; The micrograph for pore (a) is on the right-note the patchy graphite film, likely removed due to sample preparation.

Figure 2

Pore in location A1 with smooth graphitic interior and graphite structures—corresponding to the pore in Figure 1 (b).

The earlier formed round pore presents a smooth graphite-coated interior surface. A curious phenomenon is also visible under higher magnification (Figure 2), on the right side: several graphitic structures with raised bases. In hypereutectic alloys, primary graphite may nucleate within the liquid phase, where it floats until encountering austenite dendrites, leading to the rapid formation of an austenitic shell. Considering both the pore’s early formation and the sample being slightly above the eutectic, the structures are hypothesized to be primary graphite spheroids, where flotation led to encountering the pore. Moreover, the raised bases could be a surface tension-related phenomenon.

Comparing the features found in Figure 2 with the micrograph in Figure 3, coming from the same sample, provides additional support to the previously mentioned flotation hypothesis. Exploded graphite is visible all throughout the panoramic micrograph, and those that have collided with the pore show up on the pore’s surface. While these observations seem to be in line, further investigation would be required to establish conclusive proof.

Figure 3

Polished section micrograph for a sample in location A1.

Different mechanisms have been proposed for the formation of exploded graphite. Considering that nodularizers facilitate nucleation into basal planes with growth mainly occurring in the prismatic direction, recent work by Ghergu et al.²⁵ refers that, from a given critical particle radius, the growth rate in the a-axis is insufficient to keep up with the nucleation rate on the basal faces. The sectors would then decouple from each other, and form exploded morphologies. Since exploded nodules tend to appear at sizes three times smaller in RE-treated melts than in Mg-treated melts—and Ce is more commonly associated with exploded graphite than Mg—one can reason that Ce either performs better at basal nucleation or it slows graphite growth in the a-axis.^26,27 Exploded graphite is increasingly more likely at higher carbon equivalents, as it is a primary graphite phenomenon, usually observed paired with flotation.²⁸

As previously stated, nodules may degenerate into exploded graphite from a critical radius. However, it is possible to find nodules larger than exploded graphite particles in Figure 3, suggesting a variation in the critical radius for degeneration due to heterogeneities in local Mg and RE content.

A novel observation was made in a sample in location A2 (Figure 4), unveiling pores containing cylindrical graphite protrusions that, to best knowledge of the authors, have not been described in previous works. Additionally, a particle can be seen in their central regions, where the graphite assumedly nucleated. Similar phenomena have been described in ductile iron by Regordosa et al.,²⁹ where graphite nodules were suggested to grow into the pore, growing near dendrites in hypoeutectic alloys. Moreover, the nodules show fractured austenite atop the protrusions, suggesting full coverage at some point during solidification, being fractured later during solid-state growth through diffusion. The findings presented in this paper can be distinguished by their flat top surfaces, containing what is very likely to be their nucleating precipitate.

Figure 4

Cylindrical graphite protrusions in a pore in location A2, with a picture of the porosity on the top left; the arrow points toward the pore containing the protrusions.

The authors hypothesize that these particles grew from the pore’s surface into the liquid unimpeded until an austenite shell was formed. Then, as the growth is restricted by said shell, the easiest growth path was to be extruded into the pore, where there is no resistance to growth. This pore was located close to the surface, and the fact that no graphite film was found suggests that there was a connection to outside gases. During solid-state cooling, the iron matrix contracts to a greater extent than the graphite, while the decrease in carbon solubility of the matrix would provide carbon for the graphite to grow—a fact extensively proven by other authors, and yet another factor for these protrusions.^30,31 Overall, the particle growth behavior after austenite encapsulation is similar to an extrusion process, where the addition of carbon to the inner shell pushes the graphite surface further out, keeping the inoculating particle in plain view, as visible in Figure 5.

Figure 5

Proposed hypothesis timeline for the extruded graphite particles.

Proof that the particles are the inoculating agents can be seen in Figure 6 through EDS characterization. Those particles were found to contain higher contents of S, La, Ce, and Ca, which are typically found in commercial inoculation products.

Figure 6

Graphite extrusion and selected EDS maps, showing the elements with the strongest signal in the central particle.

Small precipitates are visible all throughout the pore’s surface in Figure 4. An additional EDS characterization was performed ( Figure 7) on another protrusion containing a precipitate right beside it. In this case, the composition agrees with the precipitates found on top of the protrusions – except for Ca not being found on this instance.

Figure 7

Graphite protrusion and selected EDS maps, showing the elements with the strongest signal. Note the particle with the arrow, laying on top of the pore's surface.

Pore Atmosphere and Graphite Films

This chapter intends to address the relation between the gas content of pores and their characteristics. The driving forces for porosity have been commonly divided into two closed-off groups: gas and shrinkage. As per the introductory chapter, this is likely not the case, as both could contribute to the same pore. While performing a thorough pore analysis, understanding the dynamic interaction between both could help assess whether pores originate within the liquid or are aspirated from the metal–mold interface. Thus, the surface characteristics of pores could be related to their gas composition, thereby providing insights into their source. This information could be used to develop more effective strategies to address the core issue behind pores.

Optical microscope analysis was performed for the A1 location visible in Figure 8. These pores are found during the drilling and tapping of holes, which means they have been found after machining, accruing additional scrapping costs. Instances of higher nodularity around the pores are found in some samples, as well as thin graphite-free areas encircling the pores. The latter have been extensively described as a product of a reaction between hydrogen or hydrogen–nitrogen pores with dissolved carbon, forming methane:^32,33

$${\text{C}}_{{{\text{dissolved}}}} + 2{\text{H}}_{2} \left( g \right) \to {\text{CH}}_{4} \left( g \right)$$

(3)

Figure 8

Micrographs of polished pore-containing samples.

These decarburized rings are often accompanied by graphite layers inside the pores, that have been hypothesized to form during cooling:³²

$${\text{CH}}_{{4}} \left( {\text{g}} \right) \to {\text{C}}_{{{\text{graphite}}}} + 2{\text{H}}_{{2}} \left( {\text{g}} \right)$$

(4)

Inside a pore in location A2, there was both a graphite film and a film-free surface, visible in Figure 9. The graphite-covered surface is presumed to have been liquid that was likely forced into the pore due to eutectic expansion-related pressures or recalescence remelting. This assumption is supported by both the different surface features and the abrupt outward bulging geometry, which deviates from the uniform surface. When analyzing both surfaces, it becomes necessary to consider what contributed to graphite film forming in the freshly squeezed liquid and not on the already solidified surface.

Figure 9

Sample in location A2, with a pore containing both a film-free and graphite film-containing surface; A picture of the porosity on the top left; the arrow points toward the pore containing this feature.

To address this pore’s configuration, two hypotheses have been drawn by the authors:

• 1. Since the pore was close to the molding materials, there might have existed a connection to the surrounding gases. This would promote oxidation of the solidified surface, where graphite would not be prone to grow as a film. Then, when the remaining liquid was squeezed into the pore, the environment was insufficiently oxidizing (meaning lower partial oxygen pressure), allowing graphite to freely cover the newly formed surface.

• 2. The previously solidified surface contained graphite particles. These particles may have been the receptacle for carbon deposition instead of forming a film.

The first hypothesis would rule out the graphite film formation mechanism explained in Eqs. 3 and 4. If the atmosphere is, at first, oxidizing, it is unlikely that hydrogen gas is present. Therefore, dissolved carbon would not be able to form methane via reaction with hydrogen. The takeaway is that there is evidence indicating that graphite films, contrary to popular belief, do not indicate the existence of hydrogen in the pore.

To strengthen the idea that graphite films are not generated by hydrogen gas, the variation in volume fractions of hydrogen and methane with temperature was calculated using Thermo-Calc and is shown in Figure 10. Parks et al.³⁴ have likewise plotted the Gibbs free energy of formation of different hydrocarbons compared to the hydrogen molecule plus carbon formation, depicting higher stability for methane at lower temperatures. From both, we can extract that at around 500°C methane becomes more stable.

Figure 10

Volume fraction of CH₄ and H₂ with temperature, calculated using Thermo-Calc with SUBDEMO, a subset of the SSUB database.

An alternative for graphite film formation is through the formation of CO from the reaction between eutectic carbon and entrained slags at high temperatures.^35,36 Generated CO could decompose into CO₂, leaving behind a graphite film by:

$$2{\text{CO}}\left( {\text{g}} \right) \to {\text{CO}}_{2} \left( {\text{g}} \right) + {\text{C}}_{{{\text{graphite}}}}$$

(5)

Slag entrainment is potentiated by sulfur, where manganese sulfides are formed and fluidize the slag, facilitating passage through filters in the gating system.³⁷ However, this effect seems unlikely due to both CGI and SGI’s typical low sulfur content and presence of Mg and rare earths.⁹ Additionally, no evidence of manganese sulfide particles was found in the pores. Independently of this, if an oxidizing atmosphere is present in the mold before solidification, dissolved carbon can promptly react with moisture and oxygen, resulting in CO generation at the surface. If in the previous calculation, we instead plot for an equimolar gas mixture of H, C and O, as seen in Figure 11, the takeaway is that CO and H₂ stabilities decrease with decreasing temperature, while CO₂ and CH₄ stabilities increase. Figure 11 also includes H₂O, where it reacts with carbon to form both H₂ and CO. Fruehan¹⁰ points out how dissolved aluminum can, very easily, even at low concentration, provoke hydrogen gas generation by reacting with moisture. Carbon is likewise capable of generating gas in this manner:

$$\begin{array}{*{20}c} {{\text{H}}_{{2}} {\text{O}}\left( {\text{g}} \right) + {\text{C}}_{{{\text{dissolved}}}} \to {\text{H}}_{{2}} \left( {\text{g}} \right) + {\text{CO}}\left( {\text{g}} \right)} \\ \end{array}$$

(6)

where CO could later react in the same way as described by Eq. 5. Nevertheless, if gases were to be introduced from the mold, one would anticipate finding evidence of a surface reaction or connection to the pore, which was not found.

Figure 11

Volume fraction of different gases with temperature, in an equimolar mixture of C–H–O, calculated using Thermo-Calc with SUBDEMO, a subset of the SSUB database.

While graphite films in cast iron literature have been described in binary terms, indicating either presence or absence, the authors have identified graphite films with different surface characteristics, both in previous works and in the present work, depicted in Figure 12. The wrinkly surfaces have been previously described by Amini et al.³⁸ and are reasoned to be formed under cooling, allowing graphene layers to bend under biaxial stresses from the metallic substrate thermal contraction. Additionally, a distinction can be made between different wrinkling behaviors, as it is possible to find randomly oriented wrinkles^29,39 and wrinkles that seem to follow a preferential direction.⁴⁰ Nevertheless, whenever concave valleys are present, wrinkles appear to align with their surfaces. The various manners in which graphite films can manifest imply their formation takes place under different local conditions. Gaining an understanding of how these film characteristics correspond to different conditions could prove to be invaluable when addressing defects in an industrial context. Diverse film characteristics would potentially serve as indicators for different underlying defect causes.

Figure 12

Graphite films found in pores under SEM analysis. (a) Partial graphite lining in location A2, (b) Graphite lining in location C1, with visible wrinkles and zones with a thinner film, (c) Graphite lining in location B1, with wrinkles and a thinner film on dendrites, (d) Smooth graphite film with graphite particles on top of exuded material in location A1.

The formation mechanism of the graphite films is still in question then, but evidence from detached films provides some insights, as seen in Figure 13. Specifically, the surface textures of the metal and graphite films are strikingly similar, suggesting that the films grow from the metal rather than the pore’s atmosphere. If the films were formed by gases, no metal texture would be expected. Furthermore, this finding supports the idea that films grow from carbon-oversaturated metal. In addition, graphite growth has been shown to occur during solid-state cooling as well.^30,31 Considering these results, further investigation is necessary to elucidate both the mechanisms of film formation and what the characteristics of the film may indicate about formation conditions inside the pore.

Figure 13

SEM micrographs of samples in location A2 where (a) shows imprints left on the metal by the detached graphite film, and (b) the underside of a graphite film. The depression is where a graphitic structure lies.

Pore Chronology

The previous chapters help extend the knowledge front when it comes to the pore chronology, i.e., the interval during solidification when the pore formed. As per the introductory chapter, Vazehrad et al. timed pore formation based on pore geometry: round, fine dendritic or coarse dendritic. By using the color-etching technique, it is possible to extend the timing scope. The following observations provide extra information that helps complete the pore chronology picture.

Figure 14 features a pore, which contained a graphite film, before (a) and after etching (b). Before etching, a graphite-free layer is observed encircling the pore, while on the etched cross-section, a light halo in the graphite-free region. This halo indicates that the silicon content is higher than in surrounding areas, suggesting that the metal around the pore froze at an early stage of solidification. When graphite grows, the immediate vicinity is impoverished in carbon, which favors austenite’s nucleation and growth. This suggests that the pore was lined with graphite from an early stage, serving as a carbon sink and potentiating austenite growth around the pore early in the solidification. This suggestion is comparable to austenite shell formation around graphite nodules, where austenite rapidly encircles the nodules due to the lower carbon content of their surroundings. Moreover, the liquid/gas interface could also facilitate graphite film nucleation. Further research is required to determine the mechanism by which these films form and grow.

Figure 14

Close up on the resin-filled pore in location A1, (a) polished with a graphite-free layer, and (b) color-etched.

Location B1 shows porosity with a very high degree of interconnectivity. Several panoramas were taken with SEM, showcasing a highly dendritic pore structure, with what seems to be identical coarseness, similarly covered with a graphitic film (visible in Figure 15).

Figure 15

Pore surface for sample in location B1.

The sample was split into 5 different pieces, as visible in Figure 16, a large one, spanning across the other 4 smaller ones, where the polished surface is increasingly closer to the mold wall. In this way, it is possible to observe if there are any microstructural gradients in the sample. It is important to note that this porosity was located no more than 20mm from the mold wall; despite that, no entry points for gas were found on the casting surface.

Figure 16

Sample cutting and polished surfaces (a, b, c, d, e): (a) is roughly 30mm away from a resin-bonded sand, and (b, c, d, e) are gradually closer to the green sand, with the last one in partial contact with the green sand in the bottom left corner.

Polished surfaces (b, c, d, e) as well as the respective nodularity mappings are presented in Figure 17. On the polished surfaces, the porosities can be distinguished by the large dark areas, where in the nodularity maps, a variation in nodularity can be seen across the samples where there is also a gradient in the thermal field. Additionally, Figure 17b-e showcases an increasing distance to the mold wall. There is no strong correlation, although zones with higher nodularity (dark red) may be identified in the vicinity of pores. In Figure 17b, there is also a gradient of nodularity decreasing from top to bottom. As expected, higher nodularity is located intercellularly.

Figure 17

Polished surfaces (b, c, d, e) with respective nodularity maps displayed below each one.

By looking at the etched cross-section in Figure 18, it is possible to observe a well-developed dendrite network enveloping a pore of considerable size. Moreover, a light blue halo is visible around the porous areas. Considering that the light blue indicates early solidification, it is reasonable to assume the pore grew constrained by the already solidified dendritic structure. Necking can be observed on some dendrites in Figure 15, a clear indicator of coarsening in contact with the liquid phase.

Figure 18

Sample from location B1 corresponding to the surface on Figure 16(a), color-etched cross-section reveals a light blue halo around pores.

Based on our findings, and considering our hypereutectic composition, we suggest supplementing Vazehrad et al.’s plot with two extra points, visible in Figure 19 (which features a hypereutectic cooling curve drawn based on Chaudhari et al.⁴¹):

• Point (a), representing pores formed well before solidification, showcase primary structures that would necessarily have to nucleate within an established pore. Thus, finding these structures would guide foundries to attribute pores to non-solidification-related phenomena (gas entrainment and mold-related gases).

• Point (b), representing pores with primary graphite on the inside surface, indicating pore formation either before or during primary solidification - which is likely associated with gas porosity.

Figure 19

Suggested pore chronology related to a cooling curve for a hypereutectic cast iron.

Points (c) and (d) have been updated with our observations but are equal in principle, with dendrites of increasing coarseness in relation to the contact duration with the liquid. The authors suggest that even though previous descriptions referred to pores with coarse dendrites as formed later, it may be more accurate to state that these pores keep growing until capturing dendrites with that level of coarseness. Thus, pore chronology can be used both for describing the initial formation interval (points (a) and (b)) and dendrite growth arrest interval (points (Figure 19c and d), depending on the situation at hand.

Devious Microstructure

Microstructure is known to affect propensity toward porosity defects, and analysis of early solidification structures often depicts heterogeneities within the same casting. Whether coincidental or not, these heterogeneities are periodically associated with shrinkage porosities. These zones of heterogeneous microstructure will be hereafter addressed as zones of devious microstructure. While the original grain boundaries are lost in the solid-state transformation, these zones are visible through clear differences in eutectic cell size, without a clear indication of their root cause. Figure 20 showcases these zones associated with shrinkage porosity defects - although one cannot establish a causal relationship; however, if such zones were to be found again in this case study, further research would be justified. Moreover, other authors have also noted their existence.^42,43

Figure 20

Porosity defects within zones of devious microstructure, etched with Motz reagent. Credit is directed toward Francisco Fonseca for capturing these micrographs.

This chapter addresses if this phenomenon is observable in regularly produced industrial castings. For a sample in location A1, a pore was found in direct connection to regions of much higher nodularity than the bulk of the casting, visible in Figure 21. High nodularity is expected intercellularly, however, in this case, it is spread out over a much larger area.

Figure 21

(a) Color-etched panorama with circumscribed high nodularity region, and (b) nodularity map for a sample in location A1.

Considering nodularity increases in newly solidified material along solidification, the authors hypothesize the highlighted region was under tensile stresses stemming from solidification. While under these stresses, already solidified material is pulled apart under shrinkage strain, allowing liquid to flow from elsewhere to fill the space. This liquid would be different in composition from the already solidified material due to microsegregation, and in combination with different cooling conditions results in higher nodularity. The presence of such stresses would point to the formation of a lower-pressure region, facilitating pores resulting from nucleation from dissolved gases or aspiration of surrounding gases. Even though commercially used cast iron alloys present a net expansion, localized tensile stresses may ease the introduction of pores. This phenomenon is yet to be proven or discussed in the literature, indicating further research is warranted to shed light on the problem area of devious microstructures.

Casting Skin

SGI and CGI components risk generating a layer of lamellar graphite on the surface, as seen in Figure 22, which results in worse static and dynamic mechanical properties.⁴⁴ This defect is commonly known as flake skin, and several formation mechanisms have been proposed. Most include that Mg depletes by reacting with both S and O, both present in binders and moisture on the metal/mold interface.^45,46. Mold coating containing desulfurizing elements have been shown to reduce the thickness of this layer.⁴⁷

Figure 22

Flake skin in a CGI component.

Although flake skin is often seen in connection to pores, establishing a definitive correlation between its formation and porosity remains conjectural yet significant to investigate, particularly when considering the heightened sensitivity of CGI to this phenomenon, given the lower Mg content when compared to SGI. Location B2, shown in Figure 23, represents a part of a casting with complex geometry, featuring quick thin-to-thick section variation and complex cores. The most common issue in this part is a defective casting skin, with depressions likely generated from gas pressure from the mold–metal reactions.

Figure 23

Sample from location B2, with rough surface texture and depressions.

After polishing a cross-section of the sample, a panoramic image was captured, and a nodularity map was built, both represented in Figure 24. A flake skin layer of varying thickness was noted. Additionally, it is possible to observe that part of the flake graphite degeneration is within the bulk of the casting, being easily observable in the nodularity map. Since gas intrusions are visible, there is reasonable evidence that the gases developed from the molding materials pushed the skin inward.

Figure 24

Sample in location B2 (a) polished panorama, and (b) nodularity map. Note the (a) close-up on flake graphite cells, and (b) with a circumscribed lamellar graphite region.

Further evidence of this event is visible in Figure 25, where very fine eutectic cells are shown, along with microporosity. The fine compacted graphite eutectic cells seem to be caused due to a faster cooling rate, since the liquid was entrapped between two solid structures, providing greater heat extraction. This suggests that this zone was filled later by the last-to-freeze liquid.

Figure 25

Etched cross-section of a sample in location B2, where three distinct regions are identifiable, (a) a finer cell region, (b) a detached flake layer, and (c) with a larger cell region.

When fresh liquid encountered the surface, it also underwent the flake-skin degeneration reaction. However, since there may have been an exhaustion of the reactants with the detached layer, the reaction extent was limited, resulting in a thinner flake surface. Figure 26 depicts a schematic explanation.

Figure 26

Detached flake skin formation mechanism, (a) initial flake skin formation, (b) skin detachment due to outer pressure, inducing liquid flow into now free zone, and (c) new skin formation, followed by rapid cooling.

Lacaze et al.⁴⁸ recently collected literature data regarding the effect of composition on the surface tension of iron melts. It is noted that elements from group VI (O, S, Se, Te) are strong surfactants and greatly decrease the surface tension of cast iron melts. Both O⁴⁹ and S^50,51 are known to deplete Mg in the melt to form the flake skin—therefore, it is plausible to suggest that their increased activity not only contributes to flake skin formation, but also to a decrease in surface tension, which in turn would facilitate gas intake from molding materials, leading to porosity.

Conclusion

In conclusion, the present work has identified several key knowledge gaps in the study of porosity defects. By examining persistent industrial porosity defects using different characterization techniques new insights were gained into different topics:

• Graphite structures of different morphologies protruding from the pore’s surface have been a common sighting in the studied defects. Their relatively large size suggests growth surrounded by liquid. This approach offers a promising means to gain a better understanding of the exact moment of pore formation and enables a more precise determination of the underlying cause of pores.

• Many of the pores have a graphite-free layer; etching revealed that these had colorings like dendrites and eutectic cells, indicating growth under similar conditions during solidification.

• Graphite films were found in both CGI and SGI castings. While they have been observed before, their characteristics have been largely ignored. The suggested formation mechanisms have been critically assessed and inadequately explain the occurrences in this work. A largely graphite-free surface was found with a graphite-lined exudation in a pore. It is troublesome to explain such an observation with proposed mechanisms.

• In one of the samples, a large pore was found in connection with a broad zone of distinctly higher nodularity. Given that similar phenomena have been reported before, this observation increases the likelihood of a relation between heterogeneities in the microstructure and porosity.

• Flake skin was seen paired with surface-connected porosity. This suggests that the intake of O and S generating the flake skin may have likewise caused the porosity by facilitating gas intake through a decreased surface tension.

Throughout this work, further research into porosity defects was motivated, given their contribution to a significant amount of energy and material waste. Understanding the mechanisms behind porosity formation can provide foundries with valuable insights to diagnose and implement solutions against these defects at different stages of production.