Introduction

Material microstructure serves as a bridge between process parameters and material properties. Traditionally, the microstructure of cast irons is divided into the carbon-rich phase and the metallic matrix. However, some researchers have noted the potential importance of dendrites, an often overlooked feature of cast iron microstructure. Dendrites are branched microstructures of austenite which develop mainly during the early solidification due to instabilities of the austenite-liquid interface. While dendrites later seed or merge with eutectic austenite, their early reign leaves traces in the final microstructure. Dendrites have shown correlations with properties of cast irons such as tensile strength and thermal conductivity. Dendrites have also been associated with various casting defects such as shrinkage pores, hot tears and aligned graphite flakes or nodules.

As most alloys, cast irons tend to freeze dendritically. An example is shown in Figure 1, presenting the microstructure of a hypoeutectic iron quenched during primary solidification. These dendrites grow normal to the six sides of the face-centered cubic austenite crystal lattice. The crystallographic orientation of austenite is therefore evident from directions of dendrite arms and gives rise to grain boundaries, an example of which can be observed in Figure 1. While mainly associated with the primary solidification of austenite in hypoeutectic irons, dendrites are often found also in eutectic and hypereutectic irons.1,2 This has been explained using the coupled zone concept, which postulates that coupled eutectic growth depends on a balance of supersaturations and growth kinetics of the solid phases.3 Peculiarities of the iron-carbon phase diagram and the growth kinetics of graphite and austenite provide an unusually narrow window for coupled growth which is skewed toward high carbon contents. In effect, solidification without dendrites is rare in cast irons, possible only under very slow cooling conditions.

Figure 1
figure 1

Etched section of a hypoeutectic gray iron quenched during primary solidification. Dark interdendritic areas are fine ledeburite which were liquid before quench.

Full size image

This paper reviews research work related to dendrites in cast irons, with focus on its relevance for material properties and defect formation, its evolution through solidification, its interaction with graphite and eutectic and the traces it leaves behind in the microstructure. The review is supplemented with micrographs of our own work where relevant, which have not been published elsewhere. Finally, prospects of improving cast iron production and performance by manipulation of the dendrite structure are discussed.

Effects of the Dendrite Macrostructure

A columnar zone is often observed in connection with the mold, containing dendrites which are elongated along the temperature gradient, commonly normal to the mold surface.4 An equiaxed zone is also often observed in the interior of the casting, recognized by randomly oriented dendrites which are not elongated along any one axis.4 Columnar and equiaxed dendrites are occasionally described as exogenous and endogenous, respectively, referring to their apparent external and internal origin.5

The columnar and equiaxed zone of a hypoeutectic cast iron is shown in Figure 2. While the steel die in the example certainly influenced the density of austenite grains in the columnar zone, research has shown that iron cast in sand molds tends to present similar mixtures of columnar and equiaxed grains.6 The grain structure of a hypoeutectic gray iron cast in a sand core is presented in Figure 3. The photograph also illustrates that each dendrite hosts a large number of eutectic cells, which boundaries appear as a network of dark lines. The austenite grain structure is unusual to observe in cast irons due to transformation into ferrite or pearlite but may be preserved for observation by direct austempering after solidification.6

Figure 2
figure 2

Backscatter electron image a hypoeutectic gray iron frozen in a thin sheet steel die. The apparent boundary between columnar and equiaxed dendrites is indicated with a black line. Interdendritic areas are filled with fine flake graphite, too small to resolve.

Full size image
Figure 3
figure 3

Photograph of the grain structure of a hypoeutectic gray iron cast in a sand core, preserved by austempering and revealed using Nital etch.

Full size image

The dendrite structure has importance for the formation of various casting defects. Equiaxed dendrites differ from columnar dendrites; in that they are mobile in the melt and may move due to factors such as buoyancy and melt movement.7 This has importance for mold filling and feeding of solidification shrinkage. Buoyancy-driven convection has been shown to practically halt inside dendrites, meaning their introduction contributes to a decrease of heat transport subsequently dominated by conduction.8,9 Convection of the melt affects the temperature distribution in the casting, for example shifting hotspots upward, and contributing to the local cooling conditions which influence the subsequence graphite structure.8 Columnar dendrites are associated with defects such as hot tearing and can make feeding of solidification shrinkage difficult.5,10 Observations has emerged indicating that the dendrite grain structure may influence the distribution of micropores in cast irons.11,12,13 Dendrites hinder floatation of primary graphite particles.14 This is advantageous to avoid accumulation of graphite in high positions of the casting. However, excessive accumulation of spherical graphite on dendrite arms is considered a defect which weakens ductile irons (SGI).15,16

The characteristics of the dendrite grain structure is often recognized as important for the properties of cast components. Columnar grains are desired in special uniaxially loaded parts such as turbine blades but are in most cases undesired due to loading perpendicular to the elongated direction.10 Fine equiaxed structures are associated with more uniform and isotropic properties.10 To what degree the solidification grain structure impacts the properties of cast irons is not well-understood. Contrary to conventional wisdom about grain refinement, research has indicated that the tensile strength of gray irons is favored by coarse, long, and columnar dendrites.5,17,18 This may relate to the fact that while the solid-state transformation of austenite erases its crystal grain boundaries, neutralizing the Hall–Petch effect, former grain boundaries remain characterized by discontinuities in the network of graphite-free dendrite arms.

Given, the scarce attention the dendrite grain structure has been given by researchers, there seems to remain potential to improve both the production and properties of cast irons by studying the role it plays during the casting process, in the final material, and how it may be manipulated and controlled.

Factors Affecting the Dendrite Macrostructure

The transition between columnar and equiaxed solidification of alloys is a complicated topic which is still developing after decades of research. A 2006 review concludes that equiaxed grains either nucleate heterogeneously on substrate particles in the melt or correspond to fragments of or detached columnar dendrites.10 Which mechanism is dominant depends on the precise alloy, metallurgy, geometry, and cooling conditions, and may also vary across a complex casting.10 As the columnar and equiaxed zones are complementary parts of the casting, it follows that any condition that favors one comes at a cost of the other. It has for example been shown for gray irons (LGI) that whenever the equiaxed dendrite structure is refined, this tends to result in a shorter columnar zone.19

Cooling conditions are important for the development of the dendrite structure. A sharp temperature gradient has been shown to favor columnar dendrites over equiaxed in hypoeutectic gray irons.4 The cooling rate of the casting has been found to promote equiaxed dendrites when achieved by a small casting modulus.5 However, rapid cooling has been shown to favor columnar dendrites when imposed by accelerated heat extraction.4 This suggests that if the cooling rate of a casting is increased without increasing the temperature gradient, for example by reduction in the section thickness or casting modulus, this favors equiaxed dendrites. However, if the cooling rate of a casting is increased by increasing temperature gradients, for example using chills or a mold material with higher thermal diffusivity, then the increase in the gradient is more important, favoring columnar dendrites.

Equiaxed dendrites can be promoted by facilitating nucleation of austenite in the melt. This is reviewed separately in the next section. Observations of refined equiaxed structures in severely undercooled non-treated melts imply that there may be mechanisms other than heterogeneous nucleation that give rise to equiaxed dendrites in cast irons.20

Nucleation of Austenite

The austenite grain structure may be manipulated by providing heterogeneous nucleation sites on or off the mold wall. Extreme undercooling of austenite has been achieved in a study utilizing dispersion of iron droplets in inert slag, indicating that austenite does not nucleate easily in the absence of effective substrates.21 Large undercooling has also been achieved for non-inoculated irons in Al2O3 crucibles.20

Influence of Melting

Holding of the melt at higher temperatures has shown to impede equiaxed dendrites and delay proeutectic release of latent heat as recognized on cooling curves.4,5,22 Some have found that prolonged holding of the melt impedes equiaxed dendrites,4,22 while others have found this to favor them.5 That holding temperature and time after melting affects the equiaxed structure implies that austenite can nucleate heterogeneously in untreated melts on a population of particles which remain or generate in the metal after melting and evolve over time.

Influence of Cooling

The influence of cooling rate on the nucleation of austenite is unclear. The number of grains per unit area has been found to increase slightly with cooling rate in gray and ductile iron.22,23,24 However, one study of ductile iron found that cooling rate has little or no effect.25 A study of directionally solidified white iron found that the number of primary grains depends strongly on the cooling rate at onset of primary solidification.26 However, this relation was determined by sectioning of directionally solidified bars perpendicular to the temperature gradient, thereby most likely do not reflect independently nucleated grains, but perpendicularly sectioned columnar grains. The number of columnar grains is expected to diminish with distance from a cooling source by elimination of unfavorably oriented grains.27

Influence of Carbon Equivalent

There are also contradictory observations regarding the influence of carbon equivalent on nucleation of austenite. A study on gray irons with CE between 3.94 and 4.64 found no significant effect on the number of equiaxed grains.24 However, studies by the same authors of ductile irons with CE ranging from 3.83 to 4.7 found that grains tended to be more numerous at higher CE.23,25 A study on directionally solidified white iron found a strong increase in the number of the primary grains up to a carbon equivalent of 3.9, but again, there is high risk that they counted perpendicularly sectioned columnar grains rather than equiaxed grains.26

Influence of Melt Treatment

An overview is provided in Table 1 reference articles on melt additions versus evidence suggesting it may promote heterogeneous nucleation of austenite. Evidence include increased grain count, more equiaxed or randomly oriented dendrites and reduced primary undercooling. Some also discuss the hypothetical substrate on which austenite nucleate and argue why it is favorable. Given the sensitivity of nucleation conditions and the often limited evidence the findings are based on, the table should be considered with skepticism.

Table 1 Additions Which may Favor Heterogeneous Nucleation of Austenite
Full size table

A study on gray iron found that Fe-powder, SiC and SiO2 all increased the number of equiaxed grains compared to an untreated melt; however, Fe-powder was most effective.28 The effectiveness of iron powder was suggested to relate to its transformation into austenite upon heating, thereby becoming starting points of austenite grains. However, it was never shown that the powder particles survived long enough to play this role. Similarly, Fe-powder in combination with Fe-Si inoculant has been found to promote more numerous and more equiaxed grains compared to Fe–Si alone.30 The refined grain structure was in both studies associated with reduced primary undercooling, another indicator that nucleation was facilitated.

A study focused on SiO2 showed that its potential to promote nucleation of austenite is nuanced.20 Addition of crystalline SiO2 was more effective than amorphous SiO2. Moreover, addition of crystalline SiO2 was more effective if performed below the temperature above which SiO2 is reduced by C (1360 °C according to authors). Extreme undercooling has been achieved for droplets of metal in SiO2-based glassy slag (70wt%SiO2-20wt%Al2O3-10wt%CaO-15wt%FeO), indicating that the suitability of SiO2 as nucleation substrate for austenite may be cancelled when combined with other oxides.21

Ti has been suggested to have a similar grain refining effect in cast irons as in steels.17,18,32 However, only one study was found that demonstrated that TiN may refine the solidification structure of austenitic steels.33 Moreover, another study of austenitic steel suggests grain size increases (not decreases) with increasing mole fraction of TiN.34 Confusion regarding the effect of Ti or TiN may relate to TiN promoting nucleation of δ-ferrite and has a tendency to pin austenite grain boundaries during heat treatment.35,36,37,38

One study suggests that B, Bi, and Zr promote the nucleation of austenite in gray irons.32 Unfortunately, evidence is scarce and descriptions are vague. The article also discusses the supposed effects of elements on dendrite growth, suggesting alternative mechanisms for grain refinement.

The effect of Bi on the dendrite structure of hypoeutectic gray iron was studied in more detail.22 Melts treated with Bi up to 0.02 wt% showed more numerous and more equiaxed dendrites. However, cooling curves indicated that Bi lowers the temperature of primary recalescence slightly. This led the authors to propose that the grain refining effect of Bi may relate to slowing of dendrite growth rather than to facilitate nucleation.

Additions of Al have been reported to promote more equiaxed dendrites in gray irons.5,31

A study employing dispersion of iron droplets in a slag concluded that austenite nucleates with relative ease on primary graphite.21 This is consistent with a later study showing that coating of a mold with graphite powder shows increased number density of equiaxed grains compared to a non-coated mold.28

Remarks About Nucleation of Austenite

Without substrates to nucleate heterogeneously on, austenite may undercool substantially. A population of substrate particles is naturally available after melting of an iron but diminishes with increasing superheat. The nucleation potential of the melt has been reported to either increase or fall during the holding time, suggesting that the evolution of the population is conditional, presumably due to competition between generation of new particles and dissolution or coarsening of existing ones. Additions of various elements have been observed to affect the dendrite structure, but evidence is not so strong. While some studies associate grain refinement to crystallographic similarity of the added substance to austenite, it has not been demonstrated that the additions survive long enough in the melt to act as substrate. The sensitivity of nucleation to minor impurities calls for careful analysis of reproducibility and mechanism. It remains unclear how long additions survive in the melt, what secondary particles generate during dissolution, and which particles are the actual substrates for nucleation of austenite. Moreover, since large numbers of equiaxed grains have been observed also at large undercooling, a correlation between additions and the number of equiaxed grains is not sufficient to demonstrate an effect on heterogeneous nucleation. Correlations should be complemented by a reduction of undercooling, a changed population of particles in the melt and ideally observations of particles at the origin of dendrites.

Dendrite Microstructure

Besides the size, orientation and shape, each dendrite also has an internal microstructure, typically described in terms of primary arms populated by orthogonal secondary branches on which ternary branches in turn may grow. Columnar dendrites tend to contain a single primary branch, originating from a location on or near the mold wall, while equiaxed dendrites tend to feature a primary arm in each of the six preferred growth directions, originating from a central location. The microstructure of dendrites is commonly characterized using a measure of amount (area, volume) and a measure of scale (length).

Volume Fraction of Dendrites

The volume fraction of dendrites observed in as-cast irons, hypoeutectic and hypereutectic, tends to considerably exceed estimations of primary austenite using the equilibrium phase diagram and even more so under assumption of Scheil segregation.31,39 This is generally attributed to dendritic growth of austenite below the eutectic temperature. This is due to the conditions in the residual liquid falling outside of the so-called coupled zone, which is described as a field of carbon and temperature inside which solidification kinetics of the two solid phases allows for coupled growth.3,18,40,41,42 The shape of the coupled zone is sensitive to growth kinetics and thereby to graphite morphology.

In flake graphite irons, both graphite and austenite tend to grow in contact with the liquid.43,44 But the anisotropic growth behavior of flake graphite limits the degree of coupling, leading to a variety of eutectic morphologies and scales.45,46 The growth rate and spacing between graphite flakes are sensitive to undercooling and impurities.44,46,47,48 The coarse flake structure desired for gray iron has been proposed to grow on the carbon-rich side of the coupled zone, while dendrites grow on the iron-rich side along finer graphite.40 Dendrites observed in hypereutectic gray irons are for this reason thought to precede the coarse flake eutectic. This is allowed because undercooling of the eutectic liquid shifts the coupled zone to higher carbon contents, leaving the liquid on its iron-rich side.3 Rejection of carbon from the growing dendrites shifts the liquid to the carbon-rich side of the coupled zone, favoring nucleation and growth of coarse flake graphite. The volume fractions of dendrites has been shown to be predictable with reasonable accuracy, without full consideration of the coupled zone, by allowing the amount of primary austenite to rise below eutectic temperature up to eutectic recalescence.49,50 However, predicting accurately when the eutectic recalescence occurs remains challenging due to the variance and many unknowns of an industrial casting process.

The trajectory of the liquid relative to the coupled zone is less clear for ductile irons. It was early suggested that the spheroidal graphite of ductile irons grows predominantly outside of the coupled zone.3 It has later been proposed that ductile irons do not have a coupled zone;51 however, this is disputed by observations of chunky graphite in slowly cooled ductile irons, featuring finer graphite morphologies growing coupled with austenite in direct contact with the melt.52,53,54 The precise conditions under which coupling occurs remain unclear but has been shown to be sensitive to a variety of elements.52,53,54 Given that the liquid tends to be cut off from spheroidal graphite by austenite shells, the carbon content of the liquid can be expected to maintain equilibrium with austenite, on the carbon-rich side of the coupled zone. However, graphite has been observed to nucleate and grow to some extent in the liquid throughout solidification of ductile irons.55 This could shift the liquid to the iron-rich side of the coupled zone. A computational study suggests, similar to gray iron, that austenite grows mainly dendritically until the onset of recalescence in ductile iron.56,57 The greater eutectic undercooling of ductile irons thereby permits the volume fraction of dendrites to grow larger than in gray irons.

Compacted graphite iron (CGI) shares solidification characteristics both with gray iron and ductile iron. In common with ductile iron, compacted graphite iron features a portion of nodular graphite and is associated with large undercooling prior to recalescence.58 In common with gray iron, on the other hand, vermicular graphite grows coupled with austenite in roughly spherical eutectic cells with both solid phases in contact with liquid. Moreover, vermicular graphite is associated with strong recalescence and growth under moderate to low undercooling.58 Under the assumption that austenite dendrites grow independent of graphite up to onset of recalescence, this implies that CGI is more prone to form dendrites than gray irons, but less so than ductile irons. However, there is not much data available to verify that this is the case.

In line with the influence of growth morphology, some have noted that dendrite growth may be promoted by additions which suppress nucleation of graphite.18

In summary, cast irons tend to contain a higher volume fraction of dendrites than there can be primary austenite under equilibrium when the temperature falls to eutectic. Explanations presented in the literature suggest that the additional growth of dendrites occurs mainly during cooling until onset of recalescence and is thereby sensitive to cooling conditions and solidification kinetics of the eutectic, including growth morphology and nucleation conditions.

Shape and Scale of Dendrites

The most wide-spread measure of the scale of the structure within dendrites is the average spacing between secondary arms λ2, often easily obtained from a cross section. However, while often practical, the measure suffers from the somewhat arbitrary choice of arms which are not easily discerned for certain dendrite morphologies.59 Shape-independent measures of scale have been proposed, such as the reciprocal of surface area per unit volume SV-159 or its product with the volume fraction of dendrites VVSV-1, sometimes referred to as the modulus of dendrites M.60 These latter measures are robustly defined for any dendrite morphology but can be more tedious to measure. SV can be measured on a cross section using the stereological relation SV = 4/π LA = 2PL, where LA is the perimeter length of dendrites per unit area and PL is the number of intercepts of the dendrite perimeter per unit length of randomly oriented lines.61 Beware that the factor 4/π is sometimes neglected.

While the initial scale of dendrites is known to depend on conditions at the solidification front, it has been demonstrated that the subsequent coarsening of the dendrite structure is typically more important to predict the scale of dendrites found in castings.62 The driving force for dendrite coarsening is the excess free energy of the solid–liquid interface and occurs in alloys by solute diffusion between high- and low-curvature interfaces.59

Holding of hypoeutectic cast irons between liquidus and eutectic temperature has shown that the scale of dendrites (λ2, SV-1, M) grow in proportion to the cube root of holding time tC1/3, which is in line with other technical alloys.59,62,63 The effect of this coarsening process on the structure is shown in Figure 4. Since dendrite coarsening is driven by the interfacial energy of the austenite-liquid interface, surface active elements could have an influence on the rate of coarsening. However, the proportionality to tC1/3 has been found to be similar for LGI, CGI and SGI, implying that the scavenging of O and S, known to be surface active, does not have a significant effect on the rate of coarsening.63 The evolution of dendrites through the solidification process has also been studied; however, the growth of the structure in parallel with its coarsening complicates relationships to coarsening time.64 SV-1 and M also change due to rise of solid fraction during cooling by thickening of existing dendrites.65,66 An advantage of λ2 is that if dendrite coherency occurs, λ2 is mainly affected by the subsequent coarsening process, thereby less sensitive to changes of volume fraction. A study on hypoeutectic white irons suggests that λ2 may be estimated as a function of carbon content and cooling rate at onset of primary solidification.26

Figure 4
figure 4

Etched cross sections of hypoeutectic gray irons quenched a: during primary solidification and b: after 180 min isothermal hold over eutectic temperature. Interdendritic areas were liquid before quench.

Full size image

Besides growth of length scales, long coarsening times have been shown to cause fragmentation of the dendrite structure, meaning dendrite branches detach from the trunk and float independently.63,67 Long coarsening times also cause the initially ordered arrays of secondary arms to deteriorate into more globular morphologies.63,67 This emphasizes the advantage of measures such as SV-1 and M, which can be measured despite these morphological changes.

While there are currently no known means to influence the rate of the dendrite coarsening process in cast irons, there are ways to shorten the time over which coarsening occurs, resulting in a refined dendrite microstructure. The most obvious is to accelerate cooling; however, this is limited by the risk of metastable eutectic solidification. The metastable eutectic can to a degree be suppressed by alloying and by inoculation promoting nucleation of graphite, allowing for higher cooling rates.68 For hypoeutectic irons, the coarsening time can be shortened by raising the carbon equivalent, causing an earlier onset of dendrite growth. However, this comes with a decrease of the volume fraction of dendrites. It has also been shown possible to refine the dendrite microstructure by inhibiting the nucleation substrates for austenite.20 While this route for refining the dendrite microstructure is not well-understood, it could relate to a combination of growth at high undercooling and a shortening of the freezing range.

Effects of Dendrite Microstructure

Investigators have also found that zones experiencing penetration of metal into the sand mold, or conversely, penetration of mold gases into the casting, tend to contain locally coarser dendrites.69 This highlights that the dendrite coarsening process contributes to growth of interdendritic channels and thereby facilitates transfer of fluids between metal and mold. The impact of the coarsening process on the interdendritic permeability is well-recognized for other technical alloys.70,71,72 This has importance for metal feeding and thereby for avoiding shrinkage defects.

Several researchers have noted correlation between the area fraction of dendrites on cross sections and the material properties of cast irons. It was early suggested that dendrite arms may act as graphite-free fibers which reinforce the otherwise brittle material.18,32

Studies suggest that tensile strength is enhanced by increased volume fraction of dendrites.17,18,32,73 Some have found that a refined secondary arm spacing also plays a role,18,74,75 while others have been unable to find an influence.49 Another measure of the dendrite microstructure which has shown correlation with tensile strength is the-so called hydraulic diameter D = (1 − VV)SV-1, a length measure of the space between dendrite arms.74 In particular, tensile strength of pearlitic gray iron was found to fall in inverse proportion to the square root of D, akin to the maximum crack length of the Griffith equation.

Attribution of strength to the dendrite structure is complicated by its correlation with the microstructure of graphite and the metal matrix. Any change of the volume fraction of dendrites induced by a change of carbon equivalent also comes with a decreased volume fraction of graphite. Moreover, accelerated cooling rate not only refines the dendrite microstructure, but also affects the nucleation and growth of graphite and the solid-state transformation of austenite. A recent investigation decoupled dendrite microstructure from graphite and matrix by varying the cooling rate through primary solidification independent of cooling through the eutectic and eutectoid temperatures.63,75,76 This showed that the tensile strength of CGI falls with increasing scale of the dendrites, even when the graphite structure is kept relatively unchanged in terms of volume fraction, nodularity and eutectic cell count. The ferrite/pearlite ratio was not measured but was judged to be approximately unchanged.

Interactions Between Eutectic and Dendrites

It is tempting to conceptualize dendrites and eutectic as independent entities. However, many observations suggest that growth of the eutectic is influenced by dendrites and vice versa.

As soon as the graphite structure is fine enough to resolve dendrites, it becomes clear that growth of the graphite-austenite eutectic is restricted to the space between dendrite arms. This is clearer at lower carbon equivalents, where flakes also tend to be increasingly aligned with the diminishing space between dendrite arms (type E flake graphite in standards ASTM 247).77

Low carbon equivalent is also associated with growth of eutectic at larger undercooling.78,79 This can be understood by considering a eutectic cell of a certain diameter, growing at a certain velocity. Since dendrites block a portion of its interface with the liquid, proportional to the volume fraction of dendrites, less latent heat is generated by the growing cell.

Correlations have been found between the size of eutectic cells and the secondary dendrite arm spacing λ2 in gray irons, suggesting that also the scale of the dendrite microstructure may have impact on the subsequent eutectic.19,80,81 However, later experiments on CGI have shown that if the cooling rate through the eutectic is kept constant, the scale of the dendrite microstructure has no noticeable impact on the later eutectic, in terms of eutectic cell count.63 This suggests that correlations between eutectic cell count or size and λ2 are not causal, but a reflection of both depending on cooling rate.

Graphite tends to nucleate heterogeneously on non-metallic particles in the liquid of ductile irons.82,83,84 The nucleation sites for flake graphite iron are not well-understood, but until there is evidence to the contrary, it seems fair to assume it nucleates similarly on non-metallic particles in the liquid. It has been suggested frequently that dendrites may contribute to the nucleation of graphite on such substrates by enriching the surrounding liquid with carbon.1,15,17,85 This was recently confirmed using in situ X-ray tomography for ductile iron, observing graphite spheroids nucleating in the vicinity of growing dendrites.14,86

There is also evidence that eutectic cells are favored to grow in the vicinity of dendrites. Figure 5 presents a micrograph from a cross section of a hypoeutectic gray iron quenched after isothermal treatment between liquidus and eutectic temperature. Small flake graphite cells are observed concentrated to the cluster of coarsened dendrites. The small size of the cells and the graphite they contain imply that they did not grow during the isothermal treatment, but nucleated and grew during the quench in the carbon-enriched liquid surrounding the dendrites, before metastable transformation. Note that the occasional eutectic cells observed elsewhere appear to have instead grown on finer dendrites which grew during the quench, thereby all cells appear to have grown on dendrites.

Figure 5
figure 5

Hypoeutectic gray iron water-quenched from above eutectic temperature after proeutectic coarsening of the dendritic structure for 3 days. Light-optical micrograph, lightly etched. The dark background is fine ledeburite. Courtesy of Juan-Carlos Hernando and Daniel Gonzalez.

Full size image

Figure 6 presents a micrograph of a near-eutectic gray iron with a solidification time of about 100 seconds. Akin to the quenched iron, the micrograph shows dendrites of two distinct scales. Coarse flake graphite has grown preferentially on the coarse dendrites. Fine dendrites, followed by fine flake graphite, have grown in the residual liquid. Such diverse zones of microstructure were dispersed across most of the interior of the casting. The near-eutectic composition in combination with poor conditions for nucleation of graphite is suspected to have caused interruption of dendrite growth before coherency during early solidification.

Figure 6
figure 6

Backscatter electron image of a near-eutectic gray iron, showing regions containing graphite and dendrites of distinct scales.

Full size image

Figure 7 shows a similar structure in an industrially produced CGI. The color etch reveals dendrites and eutectic cells as blue, thanks to their high Si content.87 Similar to the previously presented gray iron, there are two distinct regions of eutectic; however, rather than a set of finer flake graphite, the lower part of Figure 7 shows a region with a set of distinctly smaller cells of compacted graphite. The scale of dendrites in the two regions is not as distinctly different as in the previous two cases; however, one could argue that the volume fraction of dendrites appears to be distinctly higher in the region with smaller cells.

Figure 7
figure 7

Longitudinal section of a fractured tensile bar of CGI. The microstructure is color-etched using Motz reagent (10 g Picric acid, 10 g NaOH, 40 g KOH, 50 ml distilled water). Courtesy of Dharmateja Chalasani.

Full size image

Given that graphite and eutectic show preference to grow on dendrites and show difficulty growing in their absence, the authors would like to propose the hypothesis that the microstructures observed in Figure 5, 6 and 7 all reflect situations where eutectic has grown on an incomplete dendrite framework, leaving large regions without eutectic during early solidification. At a later point in time, the austenite-liquid interface has become unstable again, giving rise to a secondary set of dendrites, which in turn allow for the remaining liquid to freeze. The microstructure constituents of these late frozen regions vary, depending on when secondary dendrites extended into them and allowed for the eutectic to grow.

If this hypothesis is correct, a prerequisite for a homogeneous eutectic is a coherent dendrite framework for the eutectic to grow on. If the dendrite structure is not coherent at the start of eutectic growth, a too strong eutectic recalescence could contribute to a prolonged interruption of dendrite growth, postponing the completion of the framework, thereby aggravating inhomogeneities.

Dendrite incoherence is expected to be more common for near-eutectic or hypereutectic irons, where they face more competition with the eutectic. Poor nucleation conditions for austenite may further hamper dendrite coherence by having delayed nucleation or fewer nucleation sites. Another potential source of dendrite incoherence is tearing of coherent dendrite structure due to stresses, for example by hot tearing or burst feeding.

Better understanding of heterogeneities in the microstructure of cast irons and how they relate to the development of the dendrite framework appears to be important to improve prediction and control of the material properties of cast irons.

Concluding Remarks

A literature review has been presented, detailing the characteristics of the dendrites in cast irons, their interaction with eutectic solidification, their relevance for production and properties of a cast iron, and how they may be manipulated.

Dendrites play important roles in production of cast iron in phenomena such as feeding, the penetration of metal into the mold, blowholes, microsegregation and hot tears. Dendrites also show significant influence on the graphite structure by restricting its distribution, forcing alignment, and offering suitable conditions for graphite and eutectic to grow. Recent work has also indicated that an incoherent dendrite framework may be a source of heterogeneous graphite structure. Dendrite microstructure parameters such as volume fraction and spacing between dendrite arms have shown strong correlation with tensile strength of hypoeutectic pearlitic gray irons and compacted graphite irons. There are reasons to expect this effect to be less for ductile irons, but this has on the other hand not been investigated.

The most straightforward means to control the dendrite structure is through the carbon equivalent and cooling rate. Increasing carbon equivalent decreases the volume fraction of dendrites. While a significant volume fraction typically remains even at hypereutectic compositions, high carbon equivalent may contribute to incoherence in the dendrites while the eutectic is developing, which in turn may be a source of inhomogeneous distribution of graphite. Increased cooling rate mainly refines the dendrite microstructure by shortened time for coarsening, but may also contribute to the volume fraction of dendrites by additional growth beneath eutectic temperature. Cooling rate also affects the dendrite grain structure; however, the effect depends on how the cooling rate changed. Increased cooling rate using chills or modified sand-binder mixture favors columnar dendrites due to the increased temperature gradient. Reduction in section size, on the other hand, appears to increase the number of equiaxed dendrites. This is likely because the temperature gradient is not increased by reduction of section size. Columnar dendrites may help avoiding defects related to exchange of fluids between mold and casting but may on the other hand make feeding more difficult.

A less understood route to control dendrites is by manipulation of conditions for nucleation of austenite. This shows potential for control of the freezing range and coherence of the dendrite structure as well as the balance between columnar and equiaxed grains, all of which has importance for feeding characteristics and potentially for graphite structure and properties. Austenite undercools substantially if nucleation is not facilitated by appropriate substrates. While such substrates appear to be present naturally after remelting in a typical foundry, dependence on process variables such as raw material, temperature, holding time, melt treatment and exposure to atmosphere and surrounding materials and implies that negligence of these nucleation conditions may still be a source of suboptimality or variance in the production. Moreover, there are indications that nucleation of austenite may be promoted by late additions to the melt, such as iron powder or crystalline SiO2 or suppressed by TiN. Understanding the influence of dendrites and how to control them remains an underutilized area which shows potential to improve the production and properties of cast irons.

Research on the influence and control of dendrites faces several challenges which may be addressed by improved experimental methods and analysis. Localization of nucleation sites and associated substrate particles may be facilitated using smaller iron samples combined with rapid quenching during the earliest stage of solidification. The substrate could also be introduced at a known location, rather than being mixed into the melt. Correlations between melt additions and equiaxed grain density should be complemented by analysis of undercooling, including comparison of thermal analysis to the theoretical liquidus temperature, accounting for the changed chemistry of the melt following the added substance. The influence of the dendrite structure on properties should be better isolated by inclusion of all competing microstructure parameters in the analysis and application of statistical methods. In hypoeutectic irons, the influence may be further isolated by control of cooling through the primary stage of solidification independent from cooling through eutectic and the solid-state transformation. While dendrites are notoriously difficult to measure in ductile irons due to interactions with austenite shells, this latter method may provide clues about their potential role.