Skip to Content
You are currently on the new version of our website. Access the old version .
MetalsMetals
  • Article
  • Open Access

23 June 2026

21 Pages

Microsegregation of Si, Cu, Mn, P, and Sn in Graphitic Cast Irons

and
Department of Materials and Manufacturing, Jönköping University, Gjuterigatan 5, 551 11 Jönköping, Sweden
*
Author to whom correspondence should be addressed.

Abstract

Microsegregation in cast materials is important to their solidification, solid-state transformation, microstructure and material properties. This work studies quantitatively the microsegregation of Si, Mn, Cu, Sn, and P in graphitic cast irons using an electron microprobe with wavelength dispersive spectrometry. The alloys contain [mass%] C: 3.86, Si: 2.59, Mn: 0.64, P: 0.03, S: 0.01, Sn: 0.098, Cu: 0.84, Mg: 0.065, include graphite morphologies ranging from ductile iron to compacted graphite iron and solidified with a solidification time of 10 min. Concentration maps are presented, showing that microsegregation patterns provide detailed information about the solidification chronology of the metal matrix. Sequencing the measurements into concentration profiles showed that, despite large differences in microstructure and cooling curve characteristics, the severity of microsegregation was similar in the studied materials. Scheil simulation of concentration profiles provided decent prediction of concentration profiles, given appropriate thermodynamic data. Numerical simulation of isothermal diffusion suggested that, for about 10 min of solidification time, diffusion in austenite mainly affected the last 10% of the matrix to freeze. Effective partition coefficients extracted from the concentration profiles varied slightly through solidification. The estimated mean effective partition coefficients for the first 90% of the alloy to freeze are k ¯ S i γ / L = 1.124 ± 0.006 , k ¯ M n γ / L = 0.696 ± 0.008 , k ¯ P γ / L = 0.15 ± 0.03 , k ¯ S n γ / L = 0.50 ± 0.02 , k ¯ C u γ / L = 1.35 ± 0.01 , where ± indicates standard deviation.

1. Introduction

1.1. Microsegregation

Solute can have non-uniform distribution in solid phases of cast materials, referred to as solute segregation or chemical segregation. This may influence the material’s solidification and response to heat treatment and may come with local variations in physical and mechanical properties [1]. In some alloys, segregation is associated with an excessive amount of eutectic, unexpected phases, and lower solidus temperature relative to thermodynamic equilibrium [1].
Solute segregation occurs at different scales. Microsegregation refers to variations in chemical composition over the scale of the material’s microstructure, while macrosegregation refers to variation of the average chemical composition over the scale of a casting [1]. Macrosegregation is essentially microsegregation combined with macroscopic mass transport.
Microsegregation relates closely to the partitioning of solute elements between liquid and solid phases during solidification. In a binary system, solid is generally less susceptible to solute, meaning solidification involves rejection of solute into the liquid. As the temperature falls, both solid and liquid tend to accept more solute. However, the rise of the solute concentration in earlier frozen solid requires diffusion within the solid, a process referred to as back-diffusion. Depending on diffusivity, distances, and cooling rate, back-diffusion can be slow [2]. In effect, while the most recently frozen solid is in local equilibrium with the liquid, the solute concentration in earlier formed solid falls behind of the rising equilibrium concentration upon cooling, producing solute gradients in the solid. If back-diffusion is sufficiently slow, variations in solute concentration between early and late frozen solid may remain at room temperature, resulting in solute microsegregation.

1.2. Prediction of Microsegregation

At thermodynamic equilibrium, the concentration of solute is uniform in every phase, meaning no microsegregation. Commercial and open-source software are available for equilibrium calculations of multi-component systems based on thermodynamic data bases [3,4].
However, thermodynamic equilibrium is rare in practical metal casting scenarios. The most extreme case of microsegregation can be estimated using the so-called Gulliver–Scheil equation which assumes that no diffusion of solute occurs in the solid, solute mixes instantaneously in the liquid, and that the ratio of concentrations in solid and liquid over their interface is constant [5]. The equation describes how the mass fraction of solute in the liquid w L changes as the mass fraction of solid f S rises:
w L = w 0 1 − f S k − 1
where w 0 is the mass fraction of solute in the liquid at onset of solidification and the partition coefficient k is defined as the ratio of mass fraction of solute in solid over liquid k = w S / w L across their interface.
Variants of the equation have been developed, describing intermediate cases where diffusion is neither complete nor negligible in solid or liquid [2,6,7,8]. Such equations require input about the diffusivity of solute in the phases, the solidification time or rate, and the scale and shape of the microstructure over which diffusion takes place. The equations also rely on simplifying assumptions, for example, regarding the evolution of the mass fraction of solid over time.
More advanced numerical methods are also available which include more complete thermodynamic descriptions, multiple solute elements, multiple solid phases, chemical interactions among diffusing elements, and more close coupling with microstructure formation [3,8].

1.3. Microsegregation in Graphitic Cast Irons

Cast iron remains the most produced cast alloy [9]. In essence, only iron and carbon are necessary to produce cast iron. However, in practice, additional elements are included, either as tolerated biproducts from raw materials and melt preparation, or as deliberate ingredients to modify its solidification characteristics or properties. Early observations of microsegregation in cast irons were based on etching of polished cross sections, where certain regions etched darker than others [10]. While this was indicative of microsegregation, it did not show clearly which elements segregate and to what extent. It was later shown using quantitative methods that last to freeze (LTF) regions tend to contain an excess amount of Mn, Cr, Ti, Mo, P, and a depletion of Si, Cu, and Ni [10,11,12,13,14,15,16,17]. Several authors noted that excess elements in LTF regions tend to be carbide stabilizers while depleted elements in LFT regions tend to be graphite stabilizers [10,11].
Equilibrium partition coefficients between austenite (γ) and liquid (L) k 0 γ / L   = w γ / L / w L have been assessed in Fe-C-X systems and were found to apply with good approximation to the austenite-graphite eutectic [18,19,20]. Equilibrium partition coefficients k 0 γ / L show strong dependence on the concentration of C in the liquid [18,19,20]. This explains why depletion of elements in LTF regions ( k 0 γ / L > 1 ) is rarely observed in carbon steels with austenitic solidification [21,22]. Another consequence of the strong dependence of partition coefficients on the concentration of C in liquid is that it changes considerably through the primary solidification of hypoeutectic irons as the carbon rises towards eutectic composition [23,24]. Relations between the partition coefficients of elements and their effect on graphitization and eutectic temperature were found to be closely tied to their interaction with carbon [18,19,20,23,25].
Partition coefficients are also affected by interactions among substitutional alloying elements. The addition of 1.2 mass% Al has, for example, been shown to change partitioning of Si from k e f f >   1 to k e f f   <   1 [26].
Near-eutectic cast irons are often assumed to feature constant partition coefficients during solidification [27,28]. However, others have shown that microsegregation itself causes partition coefficients to drift during eutectic solidification, proposing they are better expressed as functions of temperature and chemical composition of the liquid [29,30,31].
Slow diffusion in the liquid relative to the velocity of the solid–liquid interface may cause concentration gradients in the liquid which lead to departure of effective partition coefficients from equilibrium partition coefficients [32]. Using zone-melting with a flat solid–liquid interface, the k e f f for Si in Fe-C-Si was found to rise from 1.22 to 1.82 as the velocity was increased from 0 to 1 mm/s [33]. However, the velocity of the flake–graphite eutectic is rarely more than 0.1 mm/s and commonly in the order of 0.01 mm/s [34]. Growth velocities are known to be considerably slower in ductile irons (SGI) due to dependence on diffusion of carbon through austenite [35]. The convex interface of eutectic cells also accelerates diffusion in the surrounding liquid.
There are also indications that k e f f is affected by diffusion in the austenite. Homogenization heat treatment has been found to reduce microsegregation in SGI [17]. An investigation of SGI using an electron probe microanalyzer (EPMA) line scans found that microsegregation of Mn tends to be more severe in thick sections, arguing that this relates to the coarser solidification structure [10]. This is in line with a later study finding more severe segregation in SGI for 64 min than for 10 min of solidification time [14]. Similarly, microsegregation of Si and Mn in SGI has been found to be reduced by increased solidification rate and this was associated with a refinement of the solidification structure [17,36]. These reports all suggest that refinement of solidification structure is more important for microsegregation than the shorter time for diffusion when the cooling rate is increased. However, for a small solidification time of 30 s and rapid solidification rate 17 mm/s, diffusion of substitutional elements in austenite has been reported to be negligible [16,30]. This seems to suggest that refinement of diffusion distances is more important at slow cooling rates, whereas the reduced diffusion time may be more important at very high cooling rates.

1.4. Aim of the Article

This study seeks to clarify the microsegregation of Si, Mn, Cu, P, and Sn in graphitic cast irons by quantitative electron probe microanalysis (EPMA) with wavelength dispersive spectroscopy (WDS). The scale of the solidification structure is varied by comparing SGI and compacted graphite iron (CGI) as well as intermediate mixes of graphite morphologies solidified under equal cooling conditions. Besides visual concentration maps, the concentration data are filtered, sequenced, and processed to estimate concentration profiles and effective partition coefficients for quantitative comparison among maps and with simulation results. Discussion of the microsegregation is supported by Fourier thermal analysis (FTA), equilibrium calculations, simulation of Scheil solidification, and diffusion in austenite using Thermo-Calc 2022a.

2. Materials and Methods

2.1. Material and Experiment

Details regarding production of the studied materials have been published elsewhere [37,38], but a summary is provided here along with complementary details.
From a single hypereutectic SGI casting, 400 ± 0.5 g cylinders were machined and remelted, one at a time, in a cylindrical alumina crucible with a height 60 mm and inner diameter of Ø40 mm resting on a column of graphite inside a vertical tube furnace. The chemical composition for the material is given in Table 1, measured using an optical emission spectrometer (OES) (AMETEK, Kleve, Germany, SPECTROMAXx CCD LMXM3) on quenched samples collected from the melt prior to initial casting. The furnace chamber was protected by 99.999 vol pct purity Ar-gas introduced at a rate of 5 L/min from the bottom and vented through a hole in the top. The furnace was programmed to ramp up temperature to 1450 °C over 75 min, dwell for a period, then turn off, allowing the molten cast iron sample to cool and freeze. Exploiting the phenomenon of fading [37,38], the duration of the dwell at 1450 °C for 10, 20, 30, 40, 50, and 60 min, identified as H10 to H60, resulting in graphite mainly of forms III and VI (ISO 945) ranging from SGI in H10 to CGI in H60. Examples of graphite morphologies for the materials are presented in Figure 1. The samples solidified in crucibles are cylinders with diameter Ø40.5 ± 0.1 mm and about 45 mm tall. Percent nodularity (ISO 16112) for the material cross-sections are given in Table 2, taken from the earlier publication [38]. Nodule count was measured exhaustively in the region selected for EPMA, excluding particles smaller than 10 µm. The metal matrix is a mixture of ferrite and pearlite.
Table 1. Chemical composition in mass percent (mass%) measured using OES on quenched samples taken from the melt prior to casting, CEV = C + (Si + P)/3.
Figure 1. Examples of microstructure on polished cross-sections used for graphite characterization.
Table 2. Percent nodularity (ISO 16112) for material cross-sections, from earlier publication [38], as well as nodule count for the mapped region of H10.
About 5 min before the furnace turned off, two S-type thermocouples protected by sealed quartz glass tubes were introduced into the melt. One thermocouple was placed against the inside of the crucible wall, the other at the center, secured horizontally by a hole in the alumina lid. The thermocouples were secured vertically 20 mm above the bottom of the crucible by placing O-rings around the quartz tubes above the furnace lid. Thermocouple temperature readings were logged at a frequency of 33 Hz. Cooling curves suggested that the solidification time was consistently around 10 min.
The solidified samples were sectioned horizontally 20 mm above the bottom using abrasive cutting. The bottom half of the sample was first prepared for light optical microscopy (Olympus GX71F, Jönköping, Sweden) by mounting in Ø50 mm Bakelite with carbon filler, grinding on silica paper to a grit of P2000, polishing with 3 μm diamond suspension on satin woven acetate cloth and finally with an oxide slurry on porous neoprene cloth. The polished section was color etched, providing a qualitative preview of the silicon microsegregation [13,39]. The etchant was 50 g distilled water, 40 g KOH, 10 g NaOH, 10 g picric acid, heated to 94 °C and applied to polished sections by face-down immersion for 8–12 min, incrementally checking coloring until satisfactory. Color-etched sections for H10 and H60 are presented in Figure 2. A region to be mapped with EPMA was selected on each cross-section, about 7 mm away from the circumference, and marked with Vickers hardness indentations arranged in an L-shape. The cross-section was then re-polished to remove the color etching oxide layer while keeping the indentations. The mounting material used during sample preparation was removed to maximize conductivity during EPMA and the sample was finally ultrasonically cleaned in ethanol.
Figure 2. Regions (3 × 3 mm) selected for EPMA for (a): H10 and (b): H60. Motz color etching provides a qualitative preview of microsegregation, blue indicative of higher Si and yellow of lower Si. The color spectrum is disturbed slightly by ferrite in vicinity of graphite.

2.2. Fourier Thermal Analysis (FTA)

Fourier thermal analysis was employed to estimate the release of latent heat during solidification. The analysis is detailed elsewhere [40], but a brief summary is provided here. Rearrangement of the heat Equation (2) suggests that the rate of latent heat release Q S [Wm−3] equals the rate of enthalpy change [Wm−3] (first term RHS) minus the rate of enthalpy dissipation by heat conduction [Wm−3] (second term RHS).
Q s = ∂ T ∂ t ρ C p − α ∇ 2 T ρ C p
The rate of temperature change ∂ T / ∂ t [Ks−1] can be calculated from measured temperature curves. The Laplacian of temperature ∇ 2 T [Km−2] can be calculated from temperatures curves measured at two distances from the thermal center, given that the temperature field is quasi-steady state and one-dimensional (planar, cylindrical, or spherical). Further knowledge or assumptions about thermal diffusivity α , density ρ , and specific heat capacity C p , then allows for calculation of the rate of latent heat release Q S . Assuming the thermal diffusivity is a linear combination of the volume fraction of solid and liquid, the evolution of volume fraction of solid may be found by iteratively updating the evolution of volume fraction solid and thermal diffusivity until consistent within a small error.
In practice, the temperature curves are smoothed using a central moving average filter to obtain meaningful values of ∂ T / ∂ t and ∇ 2 T . Calculation of ∇ 2 T also involves measurement of the distance of thermocouples from the central axis in the frozen sample.

2.3. Electron Probe Microanalysis (EPMA)

To study microsegregation, a microprobe (ARL-SEMQ, Corr-control, Avesta, Sweden) modified for rapid quantitative mapping of concentrations over large regions was employed [41]. The stationary electron beam was accelerated with 25 kV and had a sample current around 6 μA. Six wavelength dispersive spectrometers (WDS) counted selected bands of characteristic X-rays emitted from the excited region of the sample. In each selected region, a 3 × 3 mm concentration map was captured, consisting of a 300 × 300 raster of 10 × 10 µm subregions (pixels). The map size was chosen to obtain a representative region, significantly larger than eutectic cells in H60 (Figure 2b). Concentration maps were captured by scanning one pixel at a time. During the scan, the beam wiggles over the width of the pixel, perpendicular to the stage movement. As the beam passes into the next pixel on the raster, the X-ray count is stored and reset. Counts were translated into mass% using proportionality factors obtained through calibration on eight iron-base certified reference standards, which extrema concentrations are listed in Table 3. The same mapping procedure and pixel size were used during calibration, although on 2 × 2 mm areas. The average pixel X-ray count for each area was compared to the certified mass% of the reference materials, showing good proportionality. Proportionality factors were obtained and later applied to translate pixel-wise counts into concentrations for mapped regions in the cast irons. No ZAF correction was applied.
Table 3. Concentration ranges of certified reference materials used for calibration of WDS-EPMA.

3. Results and Discussion

3.1. Microsegregation and Microstructure

In accordance with Table 4, 12 out of the 30 maps are presented in Figure 3, illustrating the spatial distribution of concentrations and their relationships to the solidification structure. The similarity of patterns in the concentration maps in Figure 3 to the patterns revealed by color etching provide crude cross-validation. While single-phase ferrite portions of the matrix were easily distinguished from pearlite (lamellar mix of ferrite and cementite) due to the much lower carbon content, they are indistinguishable in maps of substitutional solutes. This is in line with previous research on SGI indicating no significant effect of the solid-state transformation of austenite on the long-range distribution of substitutional solutes [42]. As will be shown in later sections, the distributions of substitutional solutes are well-explained as a result of partitioning during solidification with indications of slight homogenization in LTF regions due to diffusion in austenite during cooling.
Table 4. Overview of concentration maps presented in Figure 3.
Figure 3. Examples of concentration maps (a–l) for specimens and elements in accordance with Table 4.
The patterns observed in Figure 3 are further validated by recognition of well-known features of the solidification structure of cast irons. The characteristic shapes of spheroidal graphite (SG) and compacted graphite (CG) can be recognized by regions absent of Si, Cu, and Mn, appearing black in concentration maps. High concentrations of Si and Cu are found in halos around SG and near the center of compacted graphite-austenite (CG-γ) eutectic cells as well as in the core of dendrite arms. The latter appears as elongated shapes or periodic spots, perhaps most prominent in Figure 3b,c. Similarly, depletion of Si and Cu is observed in LTF regions between dendrite arms, SG-halos and CG-γ eutectic cells. Mn, P, and Sn display the opposite segregation tendency, found in high concentration in LTF regions and low concentration in dendrite arms, SG-halos, and the centers of CG-γ eutectic cells. The concentration of P is very low in most of the metal matrix but approaches concentrations in LTF regions which are well above the nominal value of 0.098 mass% for the alloy measured using optical emission spectrometry. The highest concentrations of Mn, P, and Sn measured in LTF regions should be interpreted with caution as they exceed the calibration range (Table 3).
These observations are qualitatively consistent both with recent understanding of the solidification process of cast irons and investigations of the partitioning of alloying elements in cast irons [15,20,43]. The earliest crystallized austenite tends to grow as dendrites, especially in hypoeutectic melts [15,44,45,46,47]. Austenite may also grow dendritically below eutectic temperature, even in hypereutectic melts [48,49,50,51,52], given that it is able to compete with eutectic growth modes [50,51,52,53]. In both SGI and CGI, graphite grows as spheroids in the liquid before making contact with dendritic austenite [54,55]. Upon contact, the two solid phases begin to grow in more or less close cooperation by exchange of solute. In SGI, austenite engulfs the spheroidal graphite (SG), which subsequently grows by interstitial diffusion of carbon through the growing austenite layer [35]. In the case of CGI, the two phases grow in a more collaborative and branched manner with both phases in contact with the liquid, forming roughly spherical cells [54,55]. As G and γ may exchange solute through the liquid in front of the cells, where solute diffusivity is higher than in γ, they tend to grow faster and larger than SG-γ. The earliest austenite to crystallize is thus found in the core of dendrite arms, the center of CG-γ eutectic cells, and the proximity of SG which contain above average concentrations of elements which partition with k > 1, such as Si and Cu and below average concentration for k < 1 such as Mn, P, and Sn, assuming that partition coefficients k are fairly constant through solidification.
A common method to characterize microsegregation in ductile irons is to measure concentrations along a line between graphite nodules, quantitatively comparing highs and lows along the profile [17,26,56]. Figure 3a,b show that the severity of segregation between graphite nodules depends considerably on which pair of nodules are chosen. The chance of capturing the most severe microsegregation in the profile is small and would require many samples to approach a representative value. This is in line with solidification theory as graphite nodules are not necessarily separated by LTF regions but tend to attach to dendrite arms and grow in clusters. While statistical treatment can reduce this issue [36], the advantages of assessing microsegregation over a larger representative area have been discussed in more detail elsewhere [57].
Moreover, it should be noted that the microsegregation of elements in austenite may be affected by precipitation of other phases, such as Mn forming MnS particles in the presence of sufficient S [58]. However, the activity of S is very low in CGI and SGI by the time γ becomes stable because it is largely scavenged by the graphite modification process.
Figure 3 also indicates that the graphite morphology ranging from spheroidal to compacted graphite mainly affects the spatial distribution of concentrations, resulting in LTF regions that are larger and further apart (compare for example Figure 3f and Figure 3l). The minimum concentrations of Si and Cu and maximum concentrations of Mn, P, and Sn in LTF regions are similar for all microstructures, despite large differences in their spatial distribution.

3.2. Concentration Profiles

To compare and analyze the microsegregation patterns more systematically, the concentrations in the maps are rearranged into so-called concentration profiles. While maps indicate spatial distribution of elements, these concentration profiles indicate how concentration varies from the first to the last portion of the matrix to freeze.
Measurements on graphite are first removed from the dataset using an upper bound of 5 wt.% carbon, limiting the analysis to the metal matrix which froze as austenite. Sorting the measured concentrations in order of solidification sequence has shown to be useful for characterization and analysis of microsegregation in cast alloys [31,57,59,60]. The simplest sorting method is by ascending (k < 1) or descending (k > 1) concentration. This works as long as the concentration monotonously rises or falls as the material freezes, which is not always the case in cast irons due to variable partition coefficients [22,60]. An alternative approach is to sort the measurements by descending equilibrium eutectic temperature calculated from the local concentrations [31]. In this work, all measurements are sorted in the order of ascending Mn concentration. Some sorting errors occur due to statistical error of the measured concentration of Mn, but the smooth gradients of Mn concentration in Figure 3b,h indicate that the statistical noise is small relative to its range of concentrations. It is thus reasonable to assume that measurements are sorted from the earliest crystallized part of the metal matrix to the last. This implies that, for any measurement in the sequence, there was a point in time when the locations of measurements, prior and after it in the sequence, were solid and liquid, respectively. By normalizing sequence indices ( i ) by the last ( N ), a solid fraction of the matrix ( f S M   =   i / N ) is obtained, where 1/N (near 0) and 1 correspond to the first and last pixel to freeze, respectively. Associated concentrations are plotted against f S M   in Figure 4. From here on, these curves are referred to as concentration profiles. All concentration profiles besides Mn are smoothed with a central moving average filter with a width of 200 (0.25% of the horizontal axis) for the sake of readability.
Figure 4. Concentration profiles for Si, Mn, P, Cu, and Sn. All profiles except Mn are smoothed.
All concentration profiles either decrease (Si, Cu) or increase (Mn, P, Sn) more or less monotonously with rising f S M . This is contrary to the depletion of Si observed in hypoeutectic irons [24], which can be attributed to the near-eutectic composition of the current alloy. Again, despite large differences in microstructure among the six cast irons, differences among the concentration profiles are small. This implies that the around 10 min of solidification time in the present castings is insufficient for the scale of the solidification structures to have significant impact on back-diffusion or homogenization of the microsegregation. This appears to contradict an earlier finding that 10 min solidification time resulted in less microsegregation than at 63 min for an SGI, implying that significant homogenization had occurred in the former [14]. The apparent discrepancy may relate to the larger diffusion distances in this work, implied by lower nodule count of 51 mm−2 for H10 (Table 2) compared to 216 mm−2 in the referenced work.
The variation among concentration profiles in Figure 4 appears to be largest for Si in terms of absolute concentration. For Cu and Mn, differences among the curves, albeit small, appear to form a rough pattern, where H10, H20, and H30 tend to display lower concentrations of Cu and higher concentrations of Mn compared to H40, H50, and H60, particularly in the range f S M   =   [ 0.3 ,   0.7 ] . This pattern is not as consistent for Si.

3.3. Effective Partition Coefficients

The profiles can be treated further to extract effective partition coefficients ( k e f f ). This is achieved by dividing each measured concentration on the profile by the mean of measurements later in the sequence as shown in Equation (3):
k e f f , i = w i 1 N − i ∑ j = i + 1 N w j
where w is the concentration of an element, i is the index of a given measurement in the sorted sequence, j are indices after i , and N is the number of measurements in the sequence. By sweeping the equation over indices i   =   1 to i   =   N   −   1 , a new profile is obtained for each element, representing the evolution of its k e f f as a function of f S M . The numerator Equation (3) represents the concentration of solute in solid near the solid–liquid interface, while the denominator represents the average solute concentration in the liquid. The resulting curves are presented in Figure 5. Curves for Si and Cu are smoothed with a central moving average filter with a width of 500 (0.64% of horizontal axis). The noise observed near f S M =   0 relates to the low Mn measured on the edge of graphite, which slipped through the filter before sorting. The growing noise near f S M   =   1 relates to the diminishing population size for the average in the denominator of Equation (3).
Figure 5. Effective partition coefficients. All profiles but Mn are smoothed.
While the concentration profiles of Si in Figure 4 showed most variance, k e f f of Si in Figure 5 showed the least. This indicates that differences among concentration profiles of Si in Figure 4 reflect differences in the average measured concentration rather than differences in its partitioning or microsegregation behavior. Effective partition coefficients for Si initially display a roughly constant value around k e f f =   1.10 , then a gradual rise to about 1.17 by f S M =   0.9 , followed by a decline and growing noise.
The k e f f for Cu scatters around 1.35, with a variance that is larger than for Si. Its k e f f curves appear to form two categories. For H10, H20, and H30, k e f f rises gradually up to f S M =   0.6 , after which it falls and noise grows. However, for H40, H50, and H60, k e f f falls to a local minimum around f S M =   0.45 , followed by a rise until f S M =   0.75 . This is followed by a dip and growing noise.
Some variation among the partition coefficients is also observed for Mn and can similarly be separated into two categories. For H10, H20, and H30, k e f f is relatively constant until f S M =   0.4 , followed by an accelerating rise to near f S M =   1 . For H40, H50 and H60, on the other hand, k e f f initially rises, flattens out around f S M =   0.6 , then rises again at accelerating pace to near f S M =   1 .
The k e f f for Sn rises roughly in proportion to f S M from 0.4 to around 0.55 by f S M =   0.8 , followed by an accelerating rise towards 1. The variance among the curves seems to be relatively constant throughout.
The k e f f for P shows an accelerating rise up to f S M =   0.95 after which it declines under growing scatter. For H30, the calculated k e f f for P falls below zero. This relates to small negative concentrations measured in a large part of the metal matrix of H30, which are obviously invalid. Since P is very low in most of the metal matrix, the measured concentration is very sensitive to error. This uncertainty transfers onto the numerator of Equation (3). For this reason, the k e f f calculated for P are relatively unreliable. There is no doubt, however, that the k e f f of P is considerably lower than of Si, Cu, Mn and Sn.
A curious observation is that all k e f f display tendency to approach unity near the end of solidification. It has been suggested before that sharp concentration gradients in LTF regions (see Figure 3) are vulnerable to homogenization, which may bring the k e f f in these regions closer to unity [61]. This is discussed in a later section in more detail and with support of diffusion simulation.
With these observations in mind, the means of each partition coefficient profile presented in Figure 5 up to f S M =   0.9 were calculated and are presented in Table 5. Moreover, Table 5 includes the mean partition coefficients over all materials H10–H60 along with standard deviation and standard error. The k e f f profile of P belonging to H30 is excluded from the means due to considerable amount of unphysical negative values.
Table 5. Mean effective partition coefficients based on curves in Figure 3 for f S M <   0.9 , followed by partition coefficients measured by others. Asterisk (*) indicates average excluded due to negative k e f f P in parts of the profile.
Partition coefficients measured in this work are between Morita [16] and Boeri [16] for Si, very close to Boeri [16] and Kagawa [20] for Mn and close to Boeri [16] for Cu. The proximity is rather unexpected since there are considerable differences in terms of alloy composition. For example, Boeri measured on quenched Fe-C-Si and Fe-C-Si-X cast irons of unclear carbon equivalent with 2.45 (vs. 2.59) mass% Si, 1.34 (vs. 0.64) mass% Mn and 1.36 (vs. 0.84) mass% Cu, respectively, with concentrations for the present multi-component alloy given in parenthesis. While this implies that partition coefficients are relatively insensitive to minor alloying of cast irons, they have well known sensitivity to C [20] and recent research indicates that Al alloying can considerably alter partitioning of Si [26,62].

3.4. Thermal Analysis

The next section explores whether the observed variance among k e f f may be related to differences observed in cooling curves.
Thermal analysis allows for estimation of the evolution of latent heat of fusion in the system. The fraction of latent heat released f ∆ H at a given moment of solidification is approximately proportional to the fraction of the mass of the material which has frozen fS, thus analogous to f S M in Figure 4 and Figure 5. Figure 6 shows the fraction of latent heat released, estimated using FTA versus the average of temperatures measured near the center and periphery of the cylindrical castings T a v g . Equivalent plots using either of the two thermocouples represented by T a v g shows qualitatively the same pattern.
Figure 6. Temperature versus fraction of latent heat released.
Interestingly, the curves in Figure 6 suggest that H10 and H20 freeze at lower temperatures than H40, H50, and H60 in the range f Δ H   =   [ 0.02 ,   0.4 ] , corresponding to early solidification. This coincides roughly with the range of f S M in which differences in the trends of k e f f were most pronounced in the respective materials for Cu and Mn, indicating a possible relationship between undercooling and effective partition coefficients. Given the sensitivity of partition coefficients to the carbon concentration in the liquid [18], the skewed coupled zone of cast irons may play a role by raising the carbon content of the liquid upon undercooling [49,50], in turn causing changes in partition coefficients. This highlights a need for better understanding of the impact of undercooling on microsegregation in cast irons.

3.5. Equilibrium Partition Coefficients at the Eutectic Point

Partition coefficients were further analyzed by comparison to calculations using Thermo-Calc 2022a. Partition coefficients were calculated at the eutectic point of the T-C isopleth of the Fe-C-Si-Mn-P-Cu system with concentrations in accordance with Table 1 and a pressure of 100 kPa. Results of the calculations are presented in Table 6, listing the temperature, carbon content, and partition coefficients at the eutectic point for the Thermo-Calc Software Steel and Fe-alloys Database (TCFE) versions TCFE7, TCFE10, and TCFE12 along with the mean measured effective partition coefficients from Table 5. A comparison of columns in the table indicates that TCFE7 agrees far better with experimental results than the later versions TCFE10 and TCFE12 of the database. Given the sensitivity of partition coefficients to temperature and the concentration of C in the liquid, the worse agreement of the newer databases may relate to the shift of the eutectic to higher temperature and carbon concentration, also indicated in Table 6. While the three databases predict an identical binary Fe-C phase diagram, the ternary Fe-C-Si system shows a shift of the eutectic similar to Table 6, suggesting that the thermodynamic description of this ternary interaction may be an important source of the disagreement.
Table 6. Comparison of measured and calculated partition coefficients at eutectic carbon content and temperature for databases TCFE7, TCFE10, and TCFE12.

3.6. Simulation of Concentration Profiles and Effective Partition Coefficients

To interpret the results further, concentration profiles and effective partition coefficients were calculated using the Scheil simulation module of Thermo-Calc 2022a with C set as fast diffuser, now with only the TCFE7 database. In this calculation, concentrations were the average of the profiles for H60 in Figure 4 and the pressure was 100 kPa.
The Scheil solidification simulation is based on the traditional Scheil assumptions, meaning no diffusion in the solid, instantaneous mixing in the liquid, and partitioning over their interface according to equilibrium, but allows for variable partition coefficients and multiple solid phases.
A comparison of measured and calculated concentration profiles of H60 is provided in Figure 7. In order to exclude the graphite in the same way as for the measured profiles, the solid fractions of the calculated profiles were obtained by normalizing fractions of austenite by the last.
Figure 7. Measured (color) and simulated concentration profiles (gray) for H60. Shown here are only results for 0 min diffusion time, indicated by “0”.
Simulated effective partition coefficients were calculated by dividing the simulated concentrations of austenite by the contemporary concentrations in the liquid. A comparison between simulated and measured effective partition coefficients is provided in Figure 8.
Figure 8. Measured (color) and simulated (gray) effective partition coefficients for H60. Numbers 0, 1, and 10 indicate the simulated diffusion time in minutes, for the adjacent curve.
While the profiles agreed relatively well with the Scheil simulation results, some diffusion may have occurred in austenite during solidification and cooling. To estimate the significance of this effect, the Scheil simulation concentration profiles calculated using TCFE7 were subjected to a diffusion simulation using the Diffusion Controlled Transformation (DICTRA) module of Thermo-Calc. Unfortunately, the TCFE10 database had to be used in the diffusion calculation due to its compatibility with the available Thermo-Calc Software Steels/Fe-Alloys Mobility Database (MOBFE) version MOBFE5. This may be a source of error in chemical potential gradients which drive the diffusion. The concentration profiles were placed over a 100 µm region of austenite divided into a mesh of 100 nodes with insulated boundaries. Unidirectional diffusion was subsequently simulated at constant pressure of 100 kPa and temperature of 1150 °C. Concentration profiles were extracted after 1 min, 10 min and 60 min of diffusion, from which effective partition coefficients were calculated according to the same principle as for the measured profiles. Results of the diffusion simulations are presented in Figure 7 and Figure 8, along with measurements and Scheil simulation results. Redundant curves were removed for clarity, showing only the 0 min diffusion time reference as well as the diffusion time in best agreement with experimental data, where applicable.
With no diffusion in the solid (0-curves), the simulated concentration profiles in Figure 7 are already in decent agreement with measured profiles for Si, Mn, and Cu. Diffusion simulations seemed to only worsen the agreement and are thus excluded from the figure. Simulated concentration profiles appear to be shifted relative to the measured. Given that the input to the simulation was the mean of the measured, this indicates a discrepancy. A solute mass balance for the system at the end of the simulation was employed for further analysis, expressed in Equation (4):
w = ∑ i = 1 N ( w A i Δ f A i ) + w L N f L N + w G N f G N
where w is the conserved mean concentration, w A i and Δ f A i are the simulated concentrations and changes of mass fraction of austenite at simulation steps i. w L N and f L N are the concentration and mass fraction of liquid at the last simulation step N and w G N and f G N are the corresponding values for graphite.
Application of Equation (4) to the simulation results reproduced the input concentrations perfectly, confirming that solute was conserved. The apparent discrepancy between simulation and measurements in Figure 7 relates to neglect of the remaining 1 mass% of liquid and the ~2 mass% graphite at the end of the simulation. Neglect of the residual liquid causes overestimation of the mean concentration of Si and Cu and underestimation of the mean concentration of Mn and P. While the product w G N f G N is zero for non-carbon concentrations, the balance 1 = ∑ i = 1 N f A i + f L N + f G N means that f G N comes at the expense of ∑ i = 1 N ( w A i Δ f A i ) and f L N . This demands a rise in concentrations w A i and w L N to conserve w . The latter can be interpreted as a relative rise in the concentrations of remaining elements in liquid and γ as carbon evacuates their solutions to precipitate graphite.
The shape of the concentration profiles is for this reason more important to compare than their absolute or mean values. In this regard, the almost parallel Scheil simulation (0) of the concentration profile of Si in Figure 7 is considered in good agreement with the measured concentration profile, consistent with the good agreement of the simulated and measured effective partition coefficient of Si for f S M <   0.95 in Figure 8. Scheil simulation of the concentration profile for Cu provided worse agreement than for Si in terms of shape, failing to reproduce the rapid decline measured around f S M   =   0.25   in Figure 7. The simulation underestimated the effective partition coefficient for Cu through most of the solidification, particularly for f S M <   0.3 where an elevated coefficient was measured. Scheil simulation of the concentration profile for Mn in Figure 7 underestimated the concentration in the early crystallized austenite but appears to agree relatively well with the measured concentration profile beyond f S M   =   0.3 . However, the simulated effective partition coefficient for Mn in Figure 8 agreed relatively poorly with the measured coefficient in the sense that it predicted a monotonous decline up to f S M   =   0.97 , while the measured coefficient rose steadily up to f S M   =   0.99 . The decent prediction of the concentration profile of Mn for f S M   >   0.3 is for this reason regarded as a coincidence of the late error of the predicted partition coefficient happening to compensate well for its earlier error. The Scheil simulation underestimated the sharp rise in P observed for f S M >   0.8 . Comparison of simulated and measured k e f f for P in Figure 8 suggests that the underestimation of the rise of P by Thermo-Calc with TCFE7 relates mainly to an overestimation of its k e f f for f S M <   0.7 . While the k e f f for P was underestimated for f S M >   0.7 , this did not make up for the former overestimation.
While diffusion simulation generally worsened agreement between simulated and measured concentration profiles, it tended to bring partition coefficients at high f S M closer to unity in Figure 8, in a manner reminiscent of what is observed for measured partition coefficients. This suggests that diffusion of substitutional solutes is a plausible explanation for the deviations in effective partition coefficients in LTF regions.
In summary, the Scheil simulation with database TCFE7 provided decent prediction of concentration profiles and effective partition coefficients, particularly for Si. This suggests that effects of diffusion in austenite and build-up of gradients in the liquid are small. Diffusion simulation results suggest that slight homogenization of the sharp concentration gradients found in the LTF region is a plausible explanation for the tendency of effective partition coefficients to approach unity near the end of solidification. Figure 5 indicates nevertheless that this effect was not sensitive to the scale of the eutectic. The inconsistent diffusion times required to reproduce the observed features on the curves warrant suspicion against the reliability of diffusivities of Si, Mn, Cu, and P in FCC implied by the TCFE10 and MOBFE5 databases. Given that the TCFE10 database predicted partition coefficients in poor agreement with the experiment, it is likely also an important source of errors in the diffusion calculations by providing erroneous chemical potential gradients which drive the diffusion.

4. Conclusions

Pieces from a ductile iron casting were remelted under protection of Ar in a laboratory furnace with isothermal holding times of 10, 20, 30, 40, 50, and 60 min, followed by cooling and solidification inside the furnace, resulting in cast irons with varying proportions of spheroidal and compacted graphite. On the cross-section of each resolidified casting, the microsegregation of Si, Mn, P, Sn, and Cu was studied quantitatively using an electron probe microanalyzer with the results as follows:
  • For a solidification time of approximately 10 min, the proportions of compacted and spheroidal graphite had no notable effect on the severity of microsegregation, but the larger eutectic cells of compacted graphite tend to leave last to freeze zones larger and further apart. This is expected to hold true for irons with coarser solidification structure than the investigated ductile iron or for shorter solidification time than 10 min;
  • Diffusion simulation using DICTRA suggested that diffusion in the solid during and after solidification was likely very limited in most of the metal matrix, but caused slight homogenization of last to freeze zones, which brought effective partition coefficients in these regions closer to unity. However, this effect was similar in all studied materials, despite variance of scale of the eutectic;
  • Measured concentrations were sequenced followed by calculation of effective partition coefficients through solidification. All effective partition coefficients varied slightly during solidification, particularly towards its end, but may be represented approximately by a mean;
  • Neglecting the last 10% of the matrix to freeze, the mean effective partition coefficients over all investigated materials with graphite morphology ranging from ductile iron to compacted graphite iron were k ¯ S i = 1.124 ± 0.006 , k ¯ M n = 0.696 ± 0.008 , k ¯ P = 0.15 ± 0.03 , k ¯ S n = 0.50 ± 0.02 , k ¯ C u = 1.35 ± 0.01 , where errors are standard deviation between the cast irons;
  • The effective partition coefficients of Si, Mn, P, and Cu were in reasonable agreement with equilibrium partition coefficients calculated using Thermo-Calc 2022a with database TCFE7. Later versions of the database TCFE10 and TCFE12 provided substantially worse agreement;
  • Scheil simulation connected to a thermodynamic database TCFE7 predicted experimental concentration profiles reasonably;
  • Early variations of effective partition coefficients for Cu and Mn showed correlation with variations of undercooling during the early part of solidification. The effect of undercooling on microsegregation appears to be important for cast irons and is not well understood.

Author Contributions

Conceptualization, B.D. and A.D.; methodology, B.D. and A.D.; software, B.D.; validation, B.D.; formal analysis, B.D.; investigation, B.D.; resources, B.D. and A.D.; data curation, B.D.; writing—original draft preparation, B.D.; writing—review and editing, A.D.; visualization, B.D.; supervision, A.D.; project administration, A.D.; funding acquisition, A.D. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Knowledge Foundation, grant number [20180033], This research was funded by Knowledge Foundation, grant number [20210082], co-financed by the Swedish Knowledge Foundation, Jønkøping University, Scania CV AB, Volvo Group Trucks Technology AB, TUPY SA, SKF Mekan AB, Bruzaholms Bruk AB.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors on request.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Kraft, T.; Chang, Y.A. Predicting microstructure and microsegregation in multicomponent alloys. JOM 1997, 49, 20–28. [Google Scholar] [CrossRef] [Scilit]
  2. Brody, H.D.; Flemings, M.C. Solute redistribution in Dendritic Solidification. Trans. Metall. Soc. AIME 1966, 236, 615–624. [Google Scholar]
  3. Sundman, B.; Kattner, U.R.; Palumbo, M.; Fries, S.G. OpenCalphad—A free thermodynamic software. Integr. Mater. Manuf. Innov. 2015, 4, 1–15. [Google Scholar] [CrossRef] [Scilit]
  4. Andersson, J.-O.; Helander, T.; Höglund, L.; Shi, P.; Sundman, B. Thermo-Calc & DICTRA, computational tools for materials science. Calphad 2002, 26, 273–312. [Google Scholar] [CrossRef] [Scilit]
  5. Gulliver, G. The quantitative effect of rapid cooling upon the constitution of binary alloys. J. Inst. Met. 1913, 9, 120–157. [Google Scholar]
  6. Clyne, T.; Kurz, W. Solute redistribution during solidification with rapid solid state diffusion. Metall. Mater. Trans. A 1981, 12, 965–971. [Google Scholar] [CrossRef] [Scilit]
  7. Ohnaka, I. Mathematical analysis of solute redistribution during solidification with diffusion in solid phase. Trans. Iron Steel Inst. Jpn. 1986, 26, 1045–1051. [Google Scholar] [CrossRef] [Scilit]
  8. Kobayashi, S. Solute redistribution during solidification with diffusion in solid phase: A theoretical analysis. J. Cryst. Growth 1988, 88, 87–96. [Google Scholar] [CrossRef] [Scilit]
  9. American Foundry Society. Census of world casting production. Mod. Cast. 2021, 12, 28–30. [Google Scholar]
  10. Jolley, G. Microsegregation in Nodular Cast Iron and Its Effect on Mechanical Properties. Doctoral Thesis, Aston University, Birmingham, UK, 1966. [Google Scholar]
  11. Jolley, G. Segregation During Solidification of Nodular Cast Iron. In Proceedings of the Solidification of Metals, Brighton, UK, 4–7 December 1967. [Google Scholar]
  12. Liu, P.; Loper, C., Jr. Segregation of Certain Elements in Cast Irons. Trans. AFS 1984, 92, 289–295. [Google Scholar] [CrossRef] [Scilit]
  13. Motz, J.M. Microsegregation an easily unnoticed influencing variable in the structural desciption of cast materials. Pract. Metallogr. 1988, 25, 285–293. [Google Scholar] [CrossRef] [Scilit]
  14. Hayrynen, K.; Moore, D.; Rundman, K. Heavy section ductile iron: Production and microsegregation. Trans. AFS 1988, 96, 619–632. [Google Scholar]
  15. Boeri, R.; Weinberg, F. Microsegregation in ductile iron. AFS Trans. 1989, 106, 179–184. [Google Scholar]
  16. Boeri, R.; Weinberg, F. Microsegregation of Alloying Elements in Cast Iron. Cast Met. 1993, 6, 153–158. [Google Scholar] [CrossRef] [Scilit]
  17. Owhadi, A.; Hedjazi, J.; Davami, P.; Fazli, M.; Shabestari, J.M. Microsegregation of manganese and silicon in high manganese ductile iron. Mater. Sci. Technol. 1997, 13, 813–817. [Google Scholar] [CrossRef]
  18. Kagawa, A.; Okamoto, T. Coefficients for equilibrium partition of a third element between solid and liquid in iron-carbon base ternary alloys and their relation to graphitization during iron-carbon eutectic solidification. J. Mater. Sci. 1984, 19, 2307–2318. [Google Scholar] [CrossRef] [Scilit]
  19. Kagawa, A.; Okamoto, T. Theoretical calculation of eutectic temperature and composition in iron-carbon base ternary and multicomponent alloys. J. Mater. Sci. 1987, 22, 643–650. [Google Scholar] [CrossRef] [Scilit]
  20. Kagawa, A.; Okamoto, T. Partition of alloying elements in freezing cast irons and its effect on graphitization and nitrogen blowhole formation. In Foundry Processes; Springer: Boston, MA, 1988; pp. 135–162. [Google Scholar] [CrossRef] [Scilit]
  21. Basso, A.; Toda-Caraballo, I.; San-Martín, D.; Caballero, F.G. Influence of cast part size on macro- and microsegregation patterns in a high carbon high silicon steel. J. Mat. Res. Technol. 2020, 9, 3013–3025. [Google Scholar] [CrossRef] [Scilit]
  22. Ocansey, P.M.; Pourier, D. Equilibrium partition ratios of C, Mn, and Si in a high carbon steel. Mater. Sci. Eng. A 1996, 211, 10–14. [Google Scholar] [CrossRef] [Scilit]
  23. Zhukov, A.A. Thermodynamics of microsegregation and influence of elements on structure of unalloyed and alloyed cast iron. Met. Sci. 1978, 12, 521–524. [Google Scholar] [CrossRef] [Scilit]
  24. Domeij, B.; Diószegi, A. Depletion of Si in the early solidification structure of Gray Cast Iron. In Proceedings of the 12th International Symposium on the Science and Processing of Cast Iron (SPCI-XII), Muroran, Japan, 9–12 November 2021. [Google Scholar]
  25. Morita, Z.-I.; Tanaka, T. Thermodynamics of Solute Distributions between Solid and Liquid Phases in Iron-base Ternary Alloys. Trans. Iron Steel Inst. Jpn. 1983, 23, 824–833. [Google Scholar] [CrossRef] [Scilit]
  26. Franzen, D.; Weiß, P.; Pustal, B.; Bührig-Polaczek, A. Modification of Silicon Microsegregation in Solid-Solution-Strengthened Ductile Iron by Alloying with Aluminum. Int. J. Met. 2020, 14, 1105–1114. [Google Scholar] [CrossRef] [Scilit]
  27. Fredriksson, H.; Thogrimssson, J.T.; Svensson, L. Computer Simulation of Structure Formation and Segregation during the Solidification of Cast Iron. In State of the Art of Computer Simulation of Casting and Solidification Processes; Les éditions de Physique: Ulis, France, 1986; pp. 267–275. [Google Scholar]
  28. Fredriksson, H.; Svensson, I.L. Computer Simulation of the Structure Formed During Solidification of Cast Iron. MRS Proc. 1984, 34, 273–284. [Google Scholar] [CrossRef] [Scilit]
  29. Lundbäck, E.; Svensson, I.L.; Persson, P.-E. Modelling of structure formation and relation to mechanical properties of nodular cast iron. In Modelling of Metal Forming Processes; Springer: Dordrecht, Netherlands, 1988; pp. 37–46. [Google Scholar] [CrossRef] [Scilit]
  30. Selig, C.; Lacaze, J. Study of Microsegregation Buildup during Solidification of Spheroidal Graphite Cast Iron. Metall. Mater. Trans. B 2000, 31, 827–836. [Google Scholar] [CrossRef] [Scilit]
  31. Domeij, B.; Diószegi, A. Inferring the development of microsegregation and microstructure in Spheroidal and Compacted Graphite Iron using EPMA-WDS. In Solidification Processing 2017 (SP17); Fan, Z., Ed.; Old Windsor: London, UK, 2017; pp. 455–458. Available online: https://urn.kb.se/resolve?urn=urn%3Anbn%3Ase%3Ahj%3Adiva-35531 (accessed on 14 June 2026).
  32. Burton, J.; Prim, R.; Slichter, W.P. The distribution of solute in crystals grown from the melt. Part I. Theoretical. J. Chem. Phys. 1953, 21, 1987–1991. [Google Scholar] [CrossRef] [Scilit]
  33. Kagawa, A.; Okamoto, T. Partition of silicon during eutectic solidification of iron–carbon–silicon alloy. Met. Sci. 1980, 14, 519–524. [Google Scholar] [CrossRef] [Scilit]
  34. Lux, B.; Kurz, W. Eutectic growth of Iron-Carbon-Silicon and Iron-Carbon-Silicon-Sulphur Alloys. In Proceedings of the Solidification of Metals, Brigton, UK, 4–7 December 1967. [Google Scholar]
  35. Wetterfall, S.-E.; Fredriksson, H.; Hillert, M. Solidification process of nodular cast iron. J. Iron Steel Inst. 1972, 210, 323–333. [Google Scholar]
  36. Liu, J.; Elliott, R. The influence of cast structure on the austempering of ductile iron Part 2: Assessment of model calculations of microsegregation. Int. J. Cast Met. Res. 1999, 12, 75–82. [Google Scholar] [CrossRef] [Scilit]
  37. Hernando, J.C.; Domeij, B.; González, D.; Amieva, J.M.; Diószegi, A. New Experimental Technique for Nodularity and Mg Fading Control in Compacted Graphite Iron Production on Laboratory Scale. Metall. Mater. Trans. A 2017, 48, 5432–5441. [Google Scholar] [CrossRef] [Scilit]
  38. Domeij, B.; Hernando, J.C.; Diószegi, A. Size Distribution of Graphite Nodules in Hypereutectic Cast Irons of Varying Nodularity. Metall. Mater. Trans. B 2018, 49, 2487–2504. [Google Scholar] [CrossRef] [Scilit]
  39. Vazehrad, S.; Elfsberg, J.; Diószegi, A. Study of microstructure and silicon segregation in cast iron using color etching and electron microprobe analysis. Mater. Charact. 2015, 104, 132–138. [Google Scholar] [CrossRef] [Scilit]
  40. Dioszegi, A.; Hattel, J. Inverse thermal analysis method to study solidification in cast iron. Int. J. Cast Met. Res. 2004, 17, 311–318. [Google Scholar] [CrossRef] [Scilit]
  41. Grönlund, K.; Runnsjö, G. An electron micro/macro probe system for elemental concentration mapping. In CETAS Conference on Progress of Analytical Chemistry; Commission of the European Communities: Luxembourg, 2002. [Google Scholar]
  42. Freulon, A.; de Parseval, P.; Josse, C.; Bourdie, J.; Lacaze, J. Study of the Eutectoid Transformation in Nodular Cast Irons in Relation to Solidification Microsegregation. Metall. Mater. Trans. A 2016, 47, 5362–5371. [Google Scholar] [CrossRef] [Scilit]
  43. Okamoto, T.; Morita, Z.-I.; Kagawa, A.; Tanaka, T. Partition of carbon between solid and liquid in Fe-C binary system. Trans. Iron Steel Inst. Jpn. 1983, 23, 266–271. [Google Scholar] [CrossRef] [Scilit]
  44. Mampaey, F. Modeling and Experimental Validation of Austenite Dendrite Growth. AFS Trans. 1998, 106, 469–476. [Google Scholar]
  45. Lora, R.; Diószegi, A.; Elmquist, L. Solidification Study of Gray Cast Iron in a Resistance Furnace. Key Eng. Mat. 2011, 457, 108–113. [Google Scholar] [CrossRef] [Scilit]
  46. Diószegi, A.; Lora, R.; Fourlakidis, V. Dynamic Coarsening of Austenite Dendrite in Lamellar Cast Iron Part 1―Investigation Based on Interrupted Solidification. Mater. Sci. Forum 2014, 790, 205–210. [Google Scholar] [CrossRef] [Scilit]
  47. Hernando, J.C.; Elfsberg, J.; Dahle, A.K.; Diószegi, A. Evolution of primary austenite during coarsening and impact on eutectic microstructure in Fe–C–Si alloys. Materialia 2019, 7, 100391. [Google Scholar] [CrossRef] [Scilit]
  48. Olen, K.R.; Heine, R.W. A Revision of the Fe-C-Si System. Cast. Met. Res. J. 1968, 4, 28–43. [Google Scholar]
  49. Heine, R.W.; Loper, C.R. On Dendrites and Eutectic Cells in Gray Iron. AFS Trans. 1969, 77, 185–191. [Google Scholar]
  50. Lakeland, K.; Hogan, L. The Coupled Zone Concept Applied to Solidification of Cast Irons. Iron Steel Inst. Publ. 1968, 213–223. [Google Scholar]
  51. Fredriksson, H. The Coupled Zone in Grey Cast Iron. Metall. Trans. A 1975, 6, 1658–1660. [Google Scholar] [CrossRef] [Scilit]
  52. Jones, H.; Kurz, W. Growth temperatures and the limits of coupled growth in unidirectional solidification of Fe-C eutectic alloys. Metall. Trans. A 1980, 11, 1265–1273. [Google Scholar] [CrossRef] [Scilit]
  53. Kurz, W.; Fisher, D. Dendrite growth in eutectic alloys: The coupled zone. Int. Met. Rev. 1979, 24, 177–204. [Google Scholar] [CrossRef] [Scilit]
  54. Pan, E.N.; Ogi, K.; Loper, C.R., Jr. Analysis of the solidification process of Compacted/Vermicular Graphite Cast Iron. AFS Trans. 1982, 90, 509–527. [Google Scholar]
  55. Chen, J.; Wu, D.; Liu, P.; Loper, C., Jr. Liquid metal channel formation in compacted/vermicular graphite cast iron solidification. Trans. AFS 1986, 94, 537–544. [Google Scholar]
  56. Chang, W.-S.; Lin, C.-M. Relationship between cooling rate and microsegregation in bottom-chilled directionally solidified ductile irons. J. Min. Metall. B 2013, 49, 315–322. [Google Scholar] [CrossRef] [Scilit]
  57. Gungor, M. A statistically significant experimental technique for investigating microsegregation in cast alloys. Metall. Mater. Trans. A 1989, 20, 2529–2533. [Google Scholar] [CrossRef] [Scilit]
  58. Muhmond, H.M.; Fredriksson, H. Relationship Between Inoculants and the Morphologies of MnS and Graphite in Gray Cast Iron. Metall. Mater. Trans. B 2013, 44, 283–298. [Google Scholar] [CrossRef] [Scilit]
  59. Lacaze, J. Solidification of Spheroidal Graphite Cast Irons: III. Microsegregation Related Effects. Acta Mater. 1999, 47, 3779–3792. [Google Scholar] [CrossRef] [Scilit]
  60. Domeij, B.; Dioszegi, A. Solidification Chronology of the Metal Matrix and a Study of Conditions for Micropore Formation in Cast Irons Using EPMA and FTA. Mater. Sci. Forum 2018, 925, 436–443. [Google Scholar] [CrossRef] [Scilit]
  61. Avazkonandeh-Gharavol, M.H.; Haddad-Sabzevar, M.; Fredriksson, H. Effect of partition coefficient on microsegregation during solidification of aluminium alloys. Int. J. Miner. Metall. Mat. 2014, 21, 980–989. [Google Scholar] [CrossRef] [Scilit]
  62. Franzen, D.; Weiß, P.; Pustal, B.; Bührig-Polaczek, A. Influence of aluminium on silicon microsegregation in solution strengthened ductile iron. Mater. Sci. Technol. 2019, 35, 687–694. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.