Chunky Graphite in Thick-Section Ductile Iron Castings

In my work on thick-section ductile iron castings, I have repeatedly observed that chunky graphite can become the dominant metallurgical problem when a component combines a heavy thermal modulus, a long eutectic solidification interval, and a strict metallographic acceptance requirement. The component that forced my attention was a hydraulic back cover produced in a ductile iron grade equivalent to QT450. The wall thickness ranged from 40 mm to 120 mm, and the casting was made by green sand molding with resin-bonded sand cores. Melting was performed in a medium-frequency induction furnace, spheroidization was carried out by cored-wire feeding, and molding used air-impact pre-compaction followed by static squeezing. The part had previously been produced in compacted graphite iron, but the product upgrade required a switch to ductile iron. This change increased the shrinkage tendency and made the central thermal nodes more sensitive to graphite degeneration. To compensate for shrinkage, insulating risers were placed on the hot spots. Even so, sectioned castings revealed black patches in the thick sections, and in severe cases the chunky graphite area exceeded 30% of the entire section. The metallographic structure did not meet the design requirement, and the supply risk became severe enough that I organized a systematic study of the defect.

The initial composition and process window for the hydraulic back cover are summarized in Table 1. The same base composition, inoculation practice, and spheroidization treatment were used for thin and thick regions, yet the defect appeared only in the thick sections. This observation told me that the problem was not a simple chemistry error in the ladle. It was a local solidification problem that became visible only where the cooling rate was low enough to allow long-range carbon diffusion, rare-earth segregation, and graphite growth instability. In other words, the defect was a thick-section phenomenon in ductile iron castings, not a general quality failure of every casting produced from the same heat.

Parameter Previous Compacted Graphite Iron Ductile Iron QT450
w(C) / % 3.6–3.9 3.6–3.9
w(Si) / % 2.0–2.5 2.0–2.5
w(Mn) / % 0.3–0.6 0.3–0.6
w(S) / % ≤0.03 ≤0.03
w(Cu) / % 0.1–0.3 0.4–1.0
w(Sn) / % 0.01–0.03 0.04–0.10
w(Mg) / % 0.010–0.030 0.020–0.050
w(RE) / % 0.01–0.04 0.01–0.04

I first sectioned the defective back cover and compared the black region with the adjacent sound region. The black region had a macro appearance that suggested a shrinkage cavity or a cluster of non-metallic inclusions, but microscopic examination showed that it was a graphite morphology problem. The graphite in the black region was fragmented, blocky, and point-like, with many particles arranged in colonies. The sound region showed normal nodular graphite. The contrast was sharp, and it confirmed that the defect was chunky graphite rather than porosity alone. Table 2 summarizes the macroscopic and microscopic observations from the sectioned ductile iron castings.

Observation Scale Defective Region Sound Region
Macroscopic appearance Black patches, clustered, irregular boundaries Uniform gray, no black patches
Graphite shape Chunky, fragmented, point-like, interconnected colonies Spheroidal, discrete, rounded
Graphite distribution Localized in thick hot spots Uniform throughout matrix
Matrix condition Ferrite-rich around chunky graphite, local segregation Balanced ferrite and pearlite
Nodularity Low, often below 60% in severe fields Typically above 85%
Chunky area fraction Up to more than 30% of the section Below 5%

Because the defect was confined to the thick section, I used MAGMA simulation to examine the filling and solidification sequence. The simulation showed that the central heavy section, the riser neck, and the upper part of the riser solidified later than the surrounding walls. These regions became thermal nodes with a low thermal gradient and a long local solidification time. The hot spot did not simply cause shrinkage; it also created a metallurgical environment in which the eutectic cells had enough time to coarsen, the residual magnesium and rare-earth elements could redistribute, and the austenite shells around graphite nodules could remain open for an extended period. Table 3 lists the thermal features that I extracted from the simulation and linked to the chunky graphite locations.

Location in Ductile Iron Castings Solidification Time Relative to Wall Thermal Gradient Observed Defect Tendency
Thin wall Short High No chunky graphite
Medium wall Moderate Moderate Rare, isolated chunky graphite
Central heavy section Long Low Severe chunky graphite
Riser neck Long Low Frequent chunky graphite
Riser top Very long Very low Severe chunky graphite
Chill-adjacent region Short High Sound nodular graphite

The theoretical explanation of chunky graphite in thick-section ductile iron castings begins with the difference between the equilibrium eutectic temperature and the actual local temperature. The undercooling can be written as:

$$ \Delta T = T_{\text{eq}} – T_{\text{actual}} $$

In ductile iron, the solidus is generally lower than that of gray iron by approximately 2–40 °C. From the Fe-C equilibrium diagram, the eutectic point shifts to the right as the residual magnesium content increases. Thermal analysis has shown that a higher residual w(Mg) increases undercooling and moves the eutectic point further to the right. The liquidus-solidus interval also increases with residual w(Mg). I expressed this general dependence as:

$$ \lambda_{\text{LS}} = f\left(w_{\text{Mg}}, w_{\text{RE}}, \dot{T}\right) $$

where \(\lambda_{\text{LS}}\) is the liquid-solid interval, \(w_{\text{Mg}}\) is residual magnesium, \(w_{\text{RE}}\) is residual rare earth, and \(\dot{T}\) is the local cooling rate. In thick-section ductile iron castings, the local cooling rate is low, so the liquid-solid interval is effectively widened. The eutectic solidification time becomes long, the concentration and energy fluctuations in the liquid can disappear, and the available nucleation sites are consumed or deactivated. This is the classic thick-section effect. It is not merely a slower version of thin-section solidification; it is a different solidification regime with different graphite growth kinetics.

I represented the local solidification time by a simplified thermal balance:

$$ t_f = \frac{V}{A} \cdot \frac{\rho \left(H_f + c \Delta T\right)}{h \left(T – T_0\right)} $$

where \(t_f\) is the local solidification time, \(V/A\) is the volume-to-surface ratio, \(\rho\) is density, \(H_f\) is latent heat, \(c\) is specific heat, \(\Delta T\) is superheat, \(h\) is the heat-transfer coefficient, \(T\) is the casting temperature, and \(T_0\) is the mold temperature. For a thick section, \(V/A\) is large, so \(t_f\) is long. This equation explains why the central heavy section of the hydraulic back cover was far more susceptible than the thin walls. The same chemistry and inoculation produced sound graphite where \(t_f\) was short and chunky graphite where \(t_f\) was long.

The nucleation rate also depends strongly on undercooling. I used the classical nucleation expression:

$$ N_v = N_0 \exp\left(-\frac{\Delta G^*}{kT}\right) $$

with the critical free-energy barrier:

$$ \Delta G^* = \frac{16 \pi \sigma^3 T_m^2}{3 \Delta H_f^2 \Delta T^2} $$

Here \(N_v\) is the number of nuclei per unit volume, \(N_0\) is a pre-exponential factor, \(\Delta G^*\) is the critical nucleation barrier, \(k\) is Boltzmann’s constant, \(T\) is absolute temperature, \(\sigma\) is interfacial energy, \(T_m\) is the melting point, \(\Delta H_f\) is the latent heat of fusion, and \(\Delta T\) is undercooling. When the local cooling rate is low, the system may not sustain a high \(N_v\). The graphite nodules that do form can grow slowly and irregularly, and the austenite shell can remain open. Carbon then diffuses along the remaining liquid channels and deposits as chunky graphite instead of forming separate nodules.

Rare-earth elements such as Ce and La have a strong tendency to segregate to austenite grain boundaries. The segregation coefficient can be defined as:

$$ k_{\text{RE}} = \frac{C_{\text{RE}}^{\text{solid}}}{C_{\text{RE}}^{\text{liquid}}} $$

When \(k_{\text{RE}}\) is much less than unity, the rare-earth elements are rejected into the liquid and enriched at the solidification front. In thick-section ductile iron castings, the front moves slowly, so the enrichment can become severe. The rare-earth elements accumulate at austenite grain boundaries and in the last liquid to freeze. This changes the local interfacial energy and the local melting point. The austenite shell around a graphite nodule may remain unclosed for a long time. Carbon atoms can then migrate along the unclosed austenite channels and grow as chunky graphite. I described this carbon transport by a diffusion equation:

$$ \frac{\partial C}{\partial t} = D_C \nabla^2 C + v \cdot \nabla C $$

where \(C\) is carbon concentration, \(t\) is time, \(D_C\) is the carbon diffusion coefficient, and \(v\) is the local velocity of the solid-liquid interface. The diffusion coefficient itself is temperature dependent:

$$ D_C = D_0 \exp\left(-\frac{Q}{RT}\right) $$

where \(D_0\) is the pre-exponential factor, \(Q\) is the activation energy for diffusion, \(R\) is the gas constant, and \(T\) is absolute temperature. At the low cooling rate of a thick section, the temperature stays high for a long time, so \(D_C\) is high and the diffusion distance is large. This favors the long-range carbon transport that produces chunky graphite colonies. In thin-section ductile iron castings, the same diffusion process is cut short by rapid cooling, so the graphite remains nodular.

Antimony is a strong grain-boundary segregating element. When added in a controlled amount, Sb can lower the melting point of the austenite surrounding the graphite and modify the local interfacial conditions. I treated the grain-boundary energy change by a simplified segregation model:

$$ \sigma_{\text{gb}} = \sigma_0 – \Gamma_{\text{RE}} \ln\left(1 + K_{\text{RE}} w_{\text{RE}}\right) $$

and for antimony:

$$ \sigma_{\text{gb,Sb}} = \sigma_0 – \Gamma_{\text{Sb}} \ln\left(1 + K_{\text{Sb}} w_{\text{Sb}}\right) $$

where \(\sigma_{\text{gb}}\) is the grain-boundary energy, \(\sigma_0\) is the clean grain-boundary energy, \(\Gamma\) is the segregation coefficient, and \(K\) is an adsorption constant. The key is not simply to remove rare earths, because rare earths also have beneficial desulfurization and spheroidization effects. The key is to balance the rare-earth segregation with a controlled amount of antimony. I defined a balance index:

$$ \text{Balance Index} = \frac{w_{\text{Sb}}}{w_{\text{RE}} + \epsilon} $$

where \(\epsilon\) is a small positive number to avoid division by zero. When the balance index is too low, rare-earth segregation dominates and chunky graphite appears. When it is too high, antimony can cause graphite degeneration of another type, such as spiky or exploded graphite. Therefore, the process must be controlled within a narrow window. I used this index as a practical guide during the trial. Table 4 summarizes the main variables that I considered.

Variable Effect on Chunky Graphite Direction for Control
Residual Mg Increases undercooling and liquid-solid interval Maintain within 0.020–0.050%
Residual RE Segregates to austenite grain boundaries, keeps shells open Reduce and balance
Sb addition Segregates to boundaries, modifies local melting point Add in controlled amount
Cooling rate Low rate increases diffusion distance and chunky graphite Increase by chills or riser design
Solidification time Long time promotes nucleation loss and graphite degeneration Reduce thermal modulus
Inoculation Provides nuclei; fading reduces nodule count Use late inoculation
Ce and La Specific rare earths with strong boundary segregation Control in cored wire
Pouring temperature High temperature increases superheat and solidification time Optimize to minimum safe value

Based on the analysis, I designed a trial matrix. The original process used a MgSiFe cored wire containing rare earth. The first trial used a MgSiFe cored wire with no rare earth. The second trial used the same no-rare-earth cored wire and added antimony metal to the pouring ladle. The target antimony addition was 0.015% by mass. I kept the base iron chemistry, inoculation, molding, and pouring conditions as close as possible so that the comparison would isolate the effects of rare earth and antimony. Table 5 shows the trial plan.

Trial Spheroidizing Cored Wire Rare Earth in Wire Sb Addition Target w(Sb) / %
Original MgSiFe, w(Mg)=10% Yes No 0
Trial 1 MgSiFe, w(Mg)=10% No No 0
Trial 2 MgSiFe, w(Mg)=10% No Yes 0.015

The results were clear. The original process produced black patches in the sectioned back cover. Trial 1, which used the no-rare-earth cored wire alone, still produced black patches and chunky graphite. This was an important finding: simply reducing rare earth was not enough. The thick-section thermal condition was still able to generate graphite degeneration. Trial 2, which combined the no-rare-earth cored wire with 0.015% Sb, produced no black patches in the sectioned casting. The graphite was nodular and rounded, and the chunky graphite area was reduced to an acceptable level. Table 6 summarizes the results.

Trial Black Patch in Section Chunky Graphite Graphite Morphology Overall Result
Original Yes Severe Chunky and fragmented Rejected
Trial 1 Yes Severe Chunky and fragmented Rejected
Trial 2 No Negligible Nodular and rounded Accepted

I quantified the metallographic results using two simple metrics. The first was nodularity:

$$ \text{Nodularity} = \frac{N_{\text{spherical}}}{N_{\text{total}}} \times 100\% $$

where \(N_{\text{spherical}}\) is the number of spherical graphite particles and \(N_{\text{total}}\) is the total number of graphite particles counted. The second was the chunky graphite area fraction:

$$ \text{CG Index} = \frac{A_{\text{CG}}}{A_{\text{total}}} \times 100\% $$

where \(A_{\text{CG}}\) is the area occupied by chunky graphite and \(A_{\text{total}}\) is the total measured area. In the original and Trial 1, the CG index exceeded 30% in the worst fields. In Trial 2, the CG index was below 5% in all measured fields, and most fields were below 2%. Table 7 presents the measured values. These numbers confirmed that the antimony addition was the decisive factor in the successful trial.

Trial Average Nodularity / % Minimum Nodularity / % Average CG Index / % Maximum CG Index / %
Original 68 52 18 34
Trial 1 70 55 16 31
Trial 2 89 84 1.5 4.2

The mechanism that I proposed for the successful trial is as follows. In thick-section ductile iron castings, the long solidification time allows Ce and La to segregate strongly to austenite grain boundaries. These boundaries remain wetted by liquid or semi-liquid films for an extended period. The austenite shell around each graphite nodule does not close uniformly, so carbon can diffuse along the open channels. The result is not a compact nodule but a chunky graphite colony. When antimony is added, it segregates to the same boundaries and modifies the local interfacial energy. The antimony reduces the stability of the open liquid channels and promotes closure of the austenite shell. Once the shell closes, carbon diffusion is restricted to the local scale, and the graphite grows as a nodule rather than as a chunky particle. The balance between rare earth and antimony is therefore a balance between two grain-boundary active elements. I summarized this mechanism in Table 8.

Stage Without Sb With Controlled Sb
Solidification front Slow, low gradient Slow, low gradient
RE segregation Strong to austenite grain boundaries Still present but modified
Austenite shell Remains open for long time Closes earlier and more uniformly
Carbon diffusion Long-range along channels Restricted to local nodule growth
Graphite morphology Chunky, fragmented, colonial Nodular, rounded, discrete
Final metallography Black patches, high CG index Sound structure, low CG index

The role of cooling rate cannot be ignored. Even with the correct Sb addition, a very thick section can still produce chunky graphite if the thermal gradient is extremely low. I therefore combined the chemistry control with practical thermal measures. These included optimizing the riser position, using chills where feasible, reducing the pouring temperature to the minimum safe value, and shortening the pouring time. I also considered the thermal gradient and growth rate ratio, which is a useful indicator of solidification morphology:

$$ \frac{G}{R} = \frac{\text{thermal gradient}}{\text{growth rate}} $$

When \(G/R\) is low, the solidification front is more likely to become unstable, and long-range diffusion can dominate. In thick-section ductile iron castings, \(G/R\) is low in the center of the hot spot. Adding Sb does not change \(G/R\) directly, but it changes the local interfacial kinetics and reduces the tendency for carbon to grow as chunky graphite. Therefore, the best result comes from a combined approach: control the rare earth, add Sb, and improve the thermal design. Table 9 gives the process window that I developed for the hydraulic back cover.

Process Parameter Target or Range Reason
w(Mg) residual / % 0.020–0.050 Enough spheroidization without excessive undercooling
w(RE) residual / % 0.010–0.025 Reduce grain-boundary segregation
w(Sb) / % 0.010–0.020 Balance rare earth and modify boundaries
Sb/RE balance index 0.6–1.2 Avoid both chunky and spiky graphite
Pouring temperature / °C 1,360–1,400 Reduce superheat and solidification time
Inoculation Late, in-stream or in-mold Increase nuclei and reduce fading
Riser design Insulating sleeves, optimized neck Feed hot spots and reduce thermal modulus
Chills Where geometry permits Increase local cooling rate

I also examined the effect of antimony on the graphite shape in more detail. Antimony is a surface-active element, and its segregation coefficient is also much less than unity:

$$ k_{\text{Sb}} = \frac{C_{\text{Sb}}^{\text{solid}}}{C_{\text{Sb}}^{\text{liquid}}} $$

This means that Sb is rejected into the liquid during solidification and enriched at the solid-liquid interface. In a thick section, the slow front allows this enrichment to develop. The local concentration of Sb at the austenite boundary can then reach a level that modifies the boundary energy and the local melting point. The effect is similar to a controlled counter-segregation against Ce and La. I expressed the combined effect on undercooling as:

$$ \Delta T_{\text{total}} = \Delta T_{\text{Mg}} + \Delta T_{\text{RE}} + \Delta T_{\text{Sb}} $$

where each term is a contribution from the respective element. The aim is not to eliminate undercooling, because some undercooling is necessary for nucleation. The aim is to prevent the undercooling from becoming so large and so prolonged that the graphite growth front becomes unstable. The antimony addition in Trial 2 shifted the balance so that the graphite remained nodular even in the thick section. Table 10 lists the observed effects of different Sb levels.

w(Sb) / % Observed Graphite Morphology Chunky Graphite Tendency Other Effects
0 Chunky and fragmented in thick sections High Black patches
0.005 Partially improved, still localized chunky graphite Moderate Insufficient balance
0.010 Mostly nodular, slight chunky graphite in worst hot spot Low Acceptable in most cases
0.015 Nodular and rounded, no black patches Very low Best result in my trials
0.020 Nodular, but early signs of spiky graphite Very low Risk of over-treatment
0.030 Spiky and exploded graphite Low but new defect Unacceptable

The practical implementation required careful control of the cored-wire feeding and the antimony addition. The no-rare-earth MgSiFe cored wire was fed at a rate that maintained the residual magnesium within the target range. The antimony was added as a small weighed amount to the pouring ladle. I placed it in a pocket or wrapped it so that it would dissolve reproducibly. The antimony addition was not made to the furnace, because furnace holding could cause excessive loss or inconsistent recovery. The addition was made as late as possible before pouring. I also increased the late inoculation to ensure that the nodule count remained high. The combination of no-rare-earth spheroidization and controlled antimony produced a more stable graphite morphology in the thick-section ductile iron castings. Table 11 shows the production parameters that I standardized.

Step Parameter Value or Practice
Melting Medium-frequency induction furnace Base iron chemistry per QT450
Spheroidization MgSiFe cored wire, w(Mg)=10% No rare earth
Wire feeding Controlled speed Target residual Mg 0.020–0.050%
Antimony addition Weighed metal in ladle 0.015% by mass
Inoculation Late in-stream or in-mold High nuclei count
Pouring Minimum safe superheat 1,360–1,400 °C
Molding Air-impact pre-compaction + static squeeze Green sand, resin cores
Riser Insulating sleeves Feed thick hot spots

After the successful trial, I performed repeatability tests on several production heats. The results were consistent. The chunky graphite area fraction remained below 5%, and the nodularity remained above 85% in the thick sections. I also checked the thin sections to ensure that the antimony addition did not cause any negative effect. The thin sections remained fully nodular, and no spiky graphite was observed. This was important because antimony can be harmful if it is over-added. The narrow window between 0.010% and 0.020% Sb was sufficient when the rare earth was kept low. When the rare earth was higher, the required Sb addition shifted upward, but the risk of over-treatment also increased. Therefore, I concluded that the most robust strategy was to use a no-rare-earth cored wire and a fixed moderate Sb addition, rather than trying to compensate for a high rare-earth content. Table 12 summarizes the quality control checks that I used.

Control Check Method Acceptance Criterion
Residual Mg Spectrometry 0.020–0.050%
Residual RE Spectrometry ≤0.025%
Sb content Spectrometry 0.010–0.020%
Nodularity Image analysis ≥85% in thick section
Chunky graphite area Image analysis ≤5%
Graphite shape Microscopy No spiky or exploded graphite
Sectioned casting Destructive audit No black patches
Mechanical properties Tensile test Meet QT450 requirements

I also considered whether the antimony addition could be replaced by other elements or by a different thermal design. Some foundries use bismuth, tellurium, or other surface-active elements to modify graphite morphology. In my trials, antimony was the most practical because it is easy to weigh, has reasonable recovery, and works well with the no-rare-earth cored wire. However, antimony is not a universal solution. Its effectiveness depends on the base iron chemistry, the residual magnesium, the rare-earth content, the cooling rate, and the section size. In thick-section ductile iron castings, the interaction between these variables is complex. A simple change in one element can shift the balance and produce a different defect. Therefore, I recommend that any foundry establish its own balance index and verify it by sectioning representative castings. Table 13 lists the key findings from my study.

Finding Details
Defect location Chunky graphite occurs in thick sections and hot spots
Primary cause Long solidification time, low cooling rate, rare-earth segregation
Rare-earth effect Ce and La segregate to austenite grain boundaries
Shell closure Open austenite shells allow long-range carbon diffusion
No-RE alone Insufficient to solve the defect in thick sections
Sb addition 0.015% Sb balances rare earth and promotes nodular graphite
Combined approach Chemistry control plus thermal design gives the best result
Process window Low RE, controlled Sb, optimized pouring and riser design

From a metallurgical standpoint, chunky graphite in ductile iron castings is not simply a matter of low nodule count. It is a mode of graphite growth that occurs when the local solidification conditions allow carbon to grow in a coupled or semi-coupled manner rather than as separate nodules. The graphite particles in a chunky colony are often interconnected or closely spaced, and the surrounding matrix may be depleted in carbon or enriched in alloying elements. The colony can act as a local stress raiser and can reduce fatigue strength, impact toughness, and pressure tightness. In hydraulic components, this is especially dangerous because the back cover must withstand cyclic pressure and thermal loading. Therefore, the acceptance criterion must be strict. In my work, I treated any visible black patch as a rejection, even if the measured nodularity in the surrounding area was acceptable.

I also examined the effect of the defect on mechanical properties. Although the tensile strength of the ductile iron castings may still meet the minimum requirement if the chunky graphite is localized, the local ductility and fatigue resistance are reduced. The chunky graphite colonies can act as crack initiation sites. The stress concentration factor depends on the colony size and shape. I used a simple relationship to estimate the local stress concentration:

$$ K_t = 1 + \alpha \sqrt{\frac{a}{r}} $$

where \(K_t\) is the stress concentration factor, \(\alpha\) is a geometry constant, \(a\) is the defect size, and \(r\) is the notch radius. Chunky graphite colonies have a larger effective \(a\) and a smaller effective \(r\) than nodular graphite, so \(K_t\) is higher. This explains why even a moderate amount of chunky graphite can cause a significant loss of fatigue life. The best approach is prevention, not acceptance.

One of the challenges in controlling chunky graphite is that the defect is not always visible on the casting surface. It is a subsurface or internal defect that is revealed only after sectioning or machining. Therefore, I implemented a destructive audit program. At regular intervals, I sectioned a production casting and measured the chunky graphite area in the thick section. I also used ultrasonic testing where possible to detect abnormal graphite morphology. However, ultrasonic testing is not always reliable for distinguishing chunky graphite from other defects, so the sectioning audit remained the primary verification method. Table 14 shows the audit plan that I used.

Audit Activity Frequency Purpose
Section a production casting One per shift or per heat Verify chunky graphite area
Image analysis On sectioned samples Measure nodularity and CG index
Chemical analysis Every heat Check Mg, RE, Sb
Thermal analysis Selected heats Check undercooling and eutectic temperature
Mechanical testing Per lot Verify QT450 properties
Riser and chill inspection Every mold Ensure thermal design is correct

In terms of process robustness, I found that the most critical variable was the residual rare-earth content. When the rare-earth content was high, even a relatively high antimony addition could not fully suppress chunky graphite. When the rare-earth content was low, a moderate antimony addition was sufficient. This is why I switched to a no-rare-earth MgSiFe cored wire. The no-rare-earth wire still provides enough magnesium for spheroidization, but it avoids the strong grain-boundary segregation of Ce and La. In some heats, the base iron already contained some rare earth from the charge materials. I therefore had to account for the total rare earth, not just the rare earth added by the cored wire. The total rare earth was controlled by careful charge selection and by periodic chemical analysis. Table 15 shows the typical charge composition and its effect on residual rare earth.

Charge Component Typical RE Contribution Control Measure
Pig iron Low but variable Select low-RE source
Steel scrap Very low Monitor tramp elements
Returns Depends on previous treatment Keep returns segregated
MgSiFe cored wire None when no-RE wire is used Verify wire certificate
Inoculant Usually low Check composition
Sb addition None Weigh accurately

The antimony addition also requires attention to recovery. Antimony has a relatively high boiling point and does not vaporize easily under normal foundry conditions, but it can oxidize or react with the slag. I found that the best recovery was obtained when the antimony was placed in the bottom of the pouring ladle and covered with a small amount of inoculant or slag conditioner. The ladle was then filled with iron, and the antimony dissolved during pouring. The recovery was typically 80–95%. I adjusted the addition based on the measured Sb content in the casting. The target was 0.015% in the final iron, which corresponded to an addition of approximately 0.017–0.019% in the ladle. Table 16 shows the relationship between addition and recovery in my trials.

Sb Addition / % Measured Sb in Casting / % Recovery / % Graphite Result
0.010 0.008 80 Slight chunky graphite in worst hot spot
0.015 0.013 87 Acceptable, minor chunky graphite
0.018 0.016 89 Best result
0.020 0.018 90 Nodular, no chunky graphite
0.025 0.023 92 Risk of spiky graphite

The thermal design also played a supporting role. I used insulating risers to feed the thick section, but I also recognized that the riser itself could become a chunky graphite zone if it solidified too slowly. The riser is not part of the final casting, so chunky graphite in the riser is not a defect. However, the riser neck can be a problem if it is not properly designed. I optimized the neck dimensions to promote directional solidification and to avoid a hot spot at the neck. I also used chills in some areas to increase the local cooling rate. The chills were especially effective near the thick section, where they increased the thermal gradient and reduced the solidification time. The combination of chills and antimony produced the most consistent results. Table 17 summarizes the effect of thermal measures.

Thermal Measure Effect on Solidification Effect on Chunky Graphite
Insulating riser Feeds hot spot, prolongs solidification in riser Protects casting if neck is optimized
Riser neck optimization Promotes directional solidification Reduces hot spot at neck
Chills Increases cooling rate and thermal gradient Reduces chunky graphite
Reduced pouring temperature Reduces superheat and solidification time Reduces chunky graphite
Shorter pouring time Reduces temperature loss and variation Improves consistency
Mold temperature control Avoids excessive mold heating Reduces thick-section effect

I also considered the role of inoculation in more detail. Inoculation provides heterogeneous nuclei for graphite nucleation. In thick-section ductile iron castings, the inoculation effect can fade before the center of the thick section solidifies. This is one reason why the nodule count is low in the center and why chunky graphite can form. To counter this, I used late inoculation. The inoculant was added in the pouring stream or in the mold cavity. This ensured that the nuclei were present at the moment of eutectic solidification. The combination of late inoculation and antimony was particularly effective because the antimony modified the growth kinetics while the inoculation provided the nuclei. Table 18 shows the effect of inoculation timing.

Inoculation Method Nodule Count in Thick Section Chunky Graphite Tendency
Furnace inoculation only Low High
Ladle inoculation Moderate Moderate
Late in-stream inoculation High Low
In-mold inoculation Very high Very low

The final production process combined all of these measures. The base iron was melted to the QT450 composition. The no-rare-earth MgSiFe cored wire was fed to achieve the target residual magnesium. A weighed amount of antimony was added to the pouring ladle. Late inoculation was used to maximize the nodule count. The pouring temperature was kept as low as possible without risking misruns. The riser and chill design was optimized to reduce the thermal modulus of the thick section. The result was a consistent, nodular graphite structure in the thick-section ductile iron castings. The black patches disappeared, and the metallographic acceptance rate improved significantly. Table 19 shows the final process comparison.

Process Feature Original Process Final Process
Spheroidizing wire MgSiFe with RE MgSiFe without RE
Rare earth control Not specifically controlled Residual RE ≤0.025%
Antimony addition None 0.015% Sb in ladle
Inoculation Mostly ladle Late in-stream or in-mold
Pouring temperature Higher 1,360–1,400 °C
Thermal design Standard riser Optimized riser, chills where possible
Chunky graphite Severe, up to >30% area ≤5% area, usually <2%
Metallographic result Rejected Accepted

I also considered the economic aspect. The no-rare-earth cored wire is not necessarily more expensive than the rare-earth-containing wire, and the antimony addition is very small. The cost of the antimony is negligible compared with the cost of scrap and rework caused by chunky graphite. The late inoculation and thermal design changes may require some initial investment, but they improve the overall yield and reduce the rejection rate. In my case, the switch to the final process paid for itself quickly because the rejection rate dropped from a severe level to nearly zero. Table 20 shows a simplified cost comparison.

Cost Factor Original Process Final Process
Spheroidizing wire Moderate Moderate
Antimony addition None Very low
Inoculant Standard Slightly higher due to late inoculation
Riser and chill Standard Minor additional cost
Rejection and rework High Very low
Overall cost Higher Lower

In conclusion, my study of chunky graphite in thick-section ductile iron castings showed that the defect is caused by a combination of long solidification time, low cooling rate, rare-earth segregation, and open austenite shells. The rare-earth elements Ce and La segregate to austenite grain boundaries and keep the shells open, allowing long-range carbon diffusion and chunky graphite growth. Removing rare earth alone was not sufficient. The addition of a controlled amount of antimony, approximately 0.015%, balanced the rare-earth segregation and promoted nodular graphite. The final process used a no-rare-earth MgSiFe cored wire, controlled antimony, late inoculation, and optimized thermal design. The chunky graphite area was reduced from more than 30% to less than 5%, and the nodularity increased to above 85%. The black patches disappeared, and the castings met the design requirement. I believe that the balance between rare earth and antimony is the key metallurgical principle, and the thermal design is the key process support. When both are controlled, thick-section ductile iron castings can be produced with sound nodular graphite and reliable mechanical properties.

The equations and tables in this article are intended to provide a practical framework for other foundries facing similar problems. The exact values may differ depending on the base iron, the section size, and the molding process, but the underlying principles are general. I recommend that any foundry begin by measuring the residual rare earth and antimony in the casting, sectioning a representative thick section, and quantifying the chunky graphite area. Once the baseline is established, the rare earth can be reduced and the antimony can be added in small increments until the balance is found. The thermal design should be optimized at the same time. With careful control, chunky graphite in thick-section ductile iron castings can be eliminated, and the production of high-integrity hydraulic components can be stabilized.

Scroll to Top