As a graduate student specializing in mechanical engineering, I focused my research on improving the mechanical performance of ductile iron crankshafts by embedding continuous carbon fibers. The crankshaft is one of the most critical components in an automobile engine, often subjected to high tensile stresses and alternating bending moments. Traditional methods to strengthen ductile iron, such as alloying with copper or molybdenum and heat treatment, have reached a practical limit in terms of strength-to-cost efficiency. My work explored a novel route: reinforcing ductile iron with continuous carbon fibers, which are known for their outstanding specific strength, high elastic modulus, low density, and relatively low cost. However, the main challenge lies in the poor wettability and interfacial bonding between carbon fibers and the iron matrix, as well as the severe degradation of carbon fibers at high temperatures in the presence of oxygen. This thesis systematically investigated the surface metallization of carbon fibers, the casting process for embedding them in ductile iron using hand-made sand molds, the resulting mechanical properties, and the interfacial microstructure and element distribution.
I began by reviewing the classification and characteristics of ductile iron. Ductile iron, also called nodular cast iron, has graphite nodules that reduce stress concentration and provide higher strength and ductility than gray cast iron. Its composition typically ranges from 3.8 to 4.0 % carbon, 2.0 to 2.8 % silicon, 0.6 to 0.8 % manganese, low sulfur and phosphorus, with magnesium and rare earth elements added as nodulizers. For automotive crankshafts, pearlitic ductile iron is widely used because of its good fatigue resistance and wear properties. Yet under severe service conditions, the crankshaft can still fail by fatigue fracture. Therefore, enhancing the tensile strength of the crankshaft material is of great importance. Carbon fiber, especially the PAN-based T700 grade, offers an impressive tensile strength of around 4.9 GPa and a modulus of 230 GPa, with only 1.8 g/cm³ density. If these fibers can be successfully incorporated into ductile iron without sacrificing their intrinsic properties, they could significantly strengthen the composite. But the feasibility of a casting process is far from trivial.
The first part of my experiments focused on the surface treatment of carbon fibers. Since carbon fiber surfaces are chemically inert and often coated with an organic sizing layer that weakens bonding, I designed a three-step preparation procedure: high-temperature degumming, chemical coarsening, and electroplating nickel. I used T700 carbon fiber bundles with 6000 filaments per bundle. The degumming step was conducted in a heat treatment furnace at 400 °C for 5 minutes. This temperature allowed the sizing to volatilize without causing significant oxidation of the fiber, and the resulting surface appeared cleaner and slightly roughened under scanning electron microscopy (SEM). Further coarsening was performed using a mixed strong acid solution of H₂SO₄/HNO₃ with a volume ratio of 3:7. To avoid splashing during preparation, I placed the solution container in an ice bath first, then slowly added the acids. The coarsening temperature was controlled at 30–40 °C, and the best treatment time was found to be 80 minutes. At this duration, the fiber surface developed a uniform distribution of small pits and grooves that increased the surface area and promoted mechanical interlocking with the metallic nickel coating. If the coarsening time was extended to 100 minutes, the pits merged into larger cavities and caused noticeable surface peeling, which would harm the mechanical strength of the fibers. Therefore, exactly 80 minutes was selected for all subsequent plating trials.
The electroplating process was carried out in a typical Watts-type nickel bath. The bath composition was as follows: nickel sulfate (NiSO₄) 225 g/L, nickel chloride (NiCl₂) 65 g/L, sodium chloride (NaCl) 55 g/L, boric acid (H₃BO₄) 40 g/L, and sodium dodecyl sulfate (C₁₂H₂₅SO₄Na) 0.05 g/L. The pH was maintained between 3 and 4.5, temperature at 25 °C. I varied the plating time from 2 to 10 minutes and the current density from 0.1 to 0.5 A/dm². The cathode was the pretreated continuous carbon fiber bundle, connected to the working electrode with conductive adhesive and an electrode clip. A rectifier supplied direct current. I systematically evaluated the coating morphology using a QUANT FEG 250 field emission scanning electron microscope. The results revealed that nickel deposition proceeded in stages. Initially, nickel nanoparticles nucleated preferentially on the grooves and active sites of the fiber surface. With increasing time, these particles coalesced into islands and eventually into a continuous film. When the coating thickness reached about 0.3 μm, the coating was smooth, uniform, and defect-free. Thicker coatings, particularly above 0.5 μm, contained dendritic growths and unevenness, which reduced adhesion. I established a linear relationship between coating thickness and both plating time and current density, as shown in Table 1.
| Plating time (min) | Thickness (μm) | Current density (A/dm²) | Thickness (μm) |
|---|---|---|---|
| 2 | 0.08 | 0.1 | 0.05 |
| 4 | 0.15 | 0.2 | 0.10 |
| 6 | 0.22 | 0.3 | 0.16 |
| 8 | 0.30 | 0.4 | 0.23 |
| 10 | 0.38 | 0.5 | 0.30 |
I decided to use a plating time of 8 minutes and a current density of 0.5 A/dm², giving a thickness of approximately 0.3 μm. This thin but dense nickel layer served as an ideal transition coating. Visually, the carbon fiber color changed from black to silvery gray after plating, confirming the successful metal deposition. The SEM images showed that the nickel layer completely covered the fiber surface, with some nodular features but no cracks or peeling. This nickel layer was expected to improve the wettability between carbon fibers and molten iron, act as a barrier against harmful chemical reactions, and form a strong metallurgical bond through diffusion.
After preparing the nickel-coated carbon fibers, I had to design a casting procedure to embed them into ductile iron. I chose sand casting because of its flexibility, low tooling cost, and ability to produce complex shapes. Moreover, the sand mold could be easily modified to position the continuous fibers in a precise location. The key challenge was to prevent the fibers from floating, shifting, or clustering in the high-density molten iron. My initial approach was to first create the sand mold and then insert the fibers into the cavity through small holes with steel pins. However, this approach failed because the hot molten iron softened the steel pins quickly, allowing the fibers to move and become misaligned. I then adopted a better strategy: during the sand mold compaction process, I pre-embedded the three-dimensionally braided carbon fiber bundles across the two ends of the sand box, pulling them tight to keep them at the desired height. This approach successfully locked the fibers in place. The schematic of the sand mold with pre-embedded fibers is described by the following steps:
Step 1: Place a 400×250×200 mm wooden or metal flask on a level baseplate. Put the Y-block pattern (designed according to standard dimensions) at the center, with the larger portion facing upward. The pattern had a gating system design typical for tensile test blocks. Step 2: Fill the flask with clay-bonded silica sand in layers of about 5 cm. Before compacting, attach the carbon fiber tows to the mold ends using nails or wires. Ensure the fibers are parallel to the bottom plate and pass through the future centerline of the tensile specimen. The fibers must be tensioned gently to avoid slack. Step 3: Ram each sand layer manually with a rammer, first using the pointed end, then the flat end, to achieve uniform density. Avoid hitting the pattern directly. Step 4: Turn the mold over, cut the pouring cup (about 60° taper, 60 mm large diameter), and make a sprue. Use a vent rod to create air vents. Step 5: Remove the pattern carefully with a draw screw, being sure to keep vertical alignment. Step 6: The cavity now contains the carbon fiber bundle spanning the entire length of the Y-block cavity. To protect against oxidation, purge the mold cavity with argon gas for 10 minutes prior to pouring. Step 7: Pour ductile iron melt at about 1350 °C. The Y-block was poured as part of a series of test specimens. After pouring, cover the mold surface with insulating sand to control cooling rate. Allow the block to cool to room temperature for about 4 hours before shake-out.
I cast a total of eight tensile test bars: three bars contained uncoated carbon fibers (specimens 1#, 2#, 3#) and five bars contained nickel-coated carbon fibers (specimens 4#, 5#, 6#, 7#, 8#). Before machining, I marked the fiber positions by radiography or by visual inspection of the Y-block cross sections. The Y-block castings were then cut into standard round tensile specimens with a gauge diameter of 10 mm and a gauge length of 50 mm. The machining plan removed the upper portion (the feeder head) to eliminate shrinkage porosity and inclusions. The fibers were located along the centerline of the gauge section, allowing a valid tensile test.
The composition of ductile iron used in this study is listed in Table 2, both before and after nodulizing treatment.
| Element | C | Si | Mn | S | P | Mg | RE | Cr | Cu | Sn |
|---|---|---|---|---|---|---|---|---|---|---|
| Base iron | 3.6–4.0 | — | 0.3–0.5 | 0.008–0.025 | ≤0.05 | — | — | ≤0.08 | — | — |
| Nodulized iron | 3.5–3.8 | 1.8–2.0 | 0.3–0.5 | 0.008–0.025 | ≤0.05 | 0.035–0.055 | 0.006–0.03 | ≤0.08 | 0.2–0.3 | 0.015–0.025 |
The tensile tests were performed using an MTS Exceed E45 electronic universal testing machine. The test speed was set to 12.70 mm/min, and the original gauge length was 50 mm. The results are summarized in Table 3. The average ultimate tensile strength was 437 MPa for the bars containing uncoated carbon fibers, while it was 535 MPa for the bars containing nickel-coated fibers. This represents an average improvement of 98 MPa. In the uncoated fiber bars, the fibers were poorly bonded to the metal matrix and were simply pulled out during tensile loading, as observed from the fracture surfaces. In contrast, the nickel-coated fibers fractured together with the matrix, indicating effective load transfer.
| Specimen | Fiber type | Tensile strength (MPa) |
|---|---|---|
| 1# | Uncoated | 441 |
| 2# | Uncoated | 432 |
| 3# | Uncoated | 438 |
| 4# | Nickel-coated | 540 |
| 5# | Nickel-coated | 531 |
| 6# | Nickel-coated | 536 |
| 7# | Nickel-coated | 529 |
| 8# | Nickel-coated | 539 |
The observed increase in tensile strength can be attributed to the improved interfacial bonding and the elimination of large voids that typically form when molten iron cannot penetrate the fiber bundles. In the uncoated fiber bars, the carbon fibers were clustered together, and the metal matrix failed to fill the spaces between individual filaments. This resulted in long, slit-like pores along the fiber bundles, which acted as crack initiation sites and significantly reduced the effective load-bearing cross-section. In contrast, the nickel-coated carbon fibers were well wetted by the liquid iron. The molten iron penetrated the fiber bundle and formed a dense composite with no visible gaps. This demonstrates the positive role of the nickel coating in enhancing wettability and infiltration during sand casting parts production.
One critical issue I investigated was the high-temperature oxidation of carbon fibers during the casting process. The pouring temperature of ductile iron is around 1350 °C, far above the oxidation threshold of carbon fibers (approximately 700 °C). Even though I purged the mold with argon gas for 10 minutes before pouring, the atmosphere inside the cavity was not completely oxygen-free. Therefore, the fibers experienced some degree of oxidation. To understand the degradation mechanism, I collected fibers from the fractured specimens after tensile testing and examined their morphology and chemical composition using SEM and energy-dispersive X-ray spectroscopy (EDS). The EDS analysis of uncoated fibers after exposure to high temperature showed high oxygen content (about 21 wt. %) and also some iron (6.43 wt. %). In comparison, nickel-coated fibers after high temperature showed lower oxygen (about 7.77 wt. %) and almost no iron. This indicates that the nickel layer partially protected the carbon fibers from direct contact with oxidizing gases and prevented the formation of harmful iron carbides. The carbon fibers can undergo the following oxidation reactions:
$$ \mathrm{C + O_2 \rightarrow CO_2} $$
$$ \mathrm{2C + O_2 \rightarrow 2CO} $$
$$ \mathrm{2CO + O_2 \rightarrow 2CO_2} $$
$$ \mathrm{C + CO_2 \rightarrow 2CO} $$
Above 700 °C, carbon atoms on the fiber surface react with oxygen to form CO and CO₂. The CO₂ molecules can further diffuse to the fiber surface and react with carbon to form CO, which then reacts with oxygen to produce CO₂ again. This cyclic reaction consumes the carbon atoms, leading to weight loss, surface pitting, and a decrease in fiber strength. I developed a simplified gas-solid reaction model for carbon fiber oxidation. Assuming the fibers are cylindrical with an initial radius \(r_0\), and the oxidation is controlled by chemical reaction at the fiber surface, the reaction rate in terms of molar change is:
$$ -\frac{dn}{dt} = k A c $$
where \(n\) is the number of moles of carbon, \(k\) is the rate constant, \(A\) is the surface area, and \(c\) is the oxygen concentration. Substituting \(n = \pi r^2 l \rho / M\) and \(A = 2\pi r l\), where \(\rho\) is the density and \(M\) is the molar mass, we obtain:
$$ -\frac{dr}{dt} = \frac{k c M}{\rho} $$
Integrating from \(r=r_0\) at \(t=0\) to \(r=0\) at \(t=t_f\) gives the total time for complete consumption:
$$ t_f = \frac{\rho r_0}{k c M} $$
This model confirms that lower oxygen concentration and larger fiber radius help extend the life of the fibers. In practice, I suggested three strategies: (1) casting under vacuum conditions; (2) using a fully inert gas environment; (3) applying a heat-resistant oxygen-barrier coating on the fiber surface. These strategies would minimize the oxidation damage and preserve the reinforcing effect of the carbon fibers in sand casting parts.
The most substantial part of my thesis was the investigation of the interfacial behavior between the fibers and the ductile iron matrix. The interface is the region where stress transfer occurs from the weak metal matrix to the strong carbon fibers. If the interface is weak, cracks will propagate along the fiber-matrix boundary, and the fibers will debond and pull out without contributing to the strength. Conversely, a strong interface allows the fibers to carry a significant portion of the load, improving the overall tensile strength. The interface can be classified into mechanical interlocking, physical wetting, diffusion bonding, and chemical reaction bonding. In my composite, several mechanisms coexisted.
I first observed the fracture surfaces and polished cross-sections using optical microscopy and SEM. The uncoated carbon fibers showed clear separation from the iron matrix. Large gaps were visible between the fiber tows and the surrounding metal, indicating poor infiltration. The fibers were often clustered together, and the dense iron melt could not penetrate between filaments due to the poor wettability and the high surface tension of liquid iron on carbon. This led to a porous and defective composite. On the other hand, the nickel-coated fibers were uniformly distributed, and the metal matrix completely encapsulated each filament. The interface was continuous, without any visible cracks or pores. This was evidenced by SEM images of polished sections: for uncoated fibers, the fiber bundles appeared as dark holes with a smooth boundary; for nickel-coated fibers, the fibers were embedded in a metallic matrix with no discontinuity.
To further analyze the element distribution across the interface, I used EDS mapping and point analysis. Figure 5.8 (not shown in this HTML) provided EDS elemental maps for a nickel-coated fiber specimen. The map showed carbon located in the fiber cores, iron surrounding the fibers, and nickel concentrated between the fiber and matrix. Interestingly, the nickel did not remain as a distinct layer but diffused into the iron matrix, forming a composition gradient. This diffusion occurred because the casting temperature was high enough to promote solid-state diffusion of nickel into austenite. The EDS point analysis at three locations (a: iron matrix, b: interface region, c: carbon fiber) gave the following trends:
| Location | C (wt. %) | Ni (wt. %) | Fe (wt. %) |
|---|---|---|---|
| Point a (matrix) | 2.1 | 0.4 | 97.0 |
| Point b (interface) | 8.5 | 5.2 | 84.6 |
| Point c (fiber) | 91.3 | 0.8 | 1.3 |
The nickel concentration at the interface region was significantly higher than in the matrix, indicating that the nickel coating acted as a diffusion source. The formation of a nickel-rich zone promoted the creation of a solid solution between nickel and iron, which enhanced the chemical bonding between the fiber and matrix. Additionally, the nickel layer may also react with carbon to form stable carbides such as Ni₃C at the interface, which can further improve adhesion. Consequently, the interface can be described as a Fe-Ni-C metallurgical bond with a transition layer. This is an ideal interface for load transfer in sand casting parts.
To quantify the mechanical contribution of the fibers, I estimated the theoretical tensile strength using the rule of mixtures:
$$ \sigma_c = \sigma_f V_f + \sigma_m (1 – V_f) $$
where \(\sigma_c\) is the composite strength, \(\sigma_f\) is the average fiber strength, \(\sigma_m\) is the matrix strength, and \(V_f\) is the fiber volume fraction. In my Y-block test bars, the fiber volume fraction was low (approximately 1–2% based on the cross-section). Therefore, the observed increase of 98 MPa cannot be solely explained by the rule of mixtures. Instead, the improvement was mainly due to the elimination of casting defects (porosity) and the crack-arresting effect of the continuous fibers. Even a small amount of well-bonded continuous fibers can restrict crack opening and delay fracture by bridging the cracks. This explains why a small fiber volume fraction resulted in a substantial strength increase. In future work, I aim to increase the fiber volume fraction and optimize the fiber architecture to achieve a more significant strengthening effect. Based on the rule of mixtures, if the fiber volume fraction were 10%, the theoretical composite strength would be:
$$ \sigma_c = 4900 \times 0.10 + 437 \times 0.90 = 890 \text{ MPa} $$
This estimate ignores oxidation effects and assumes perfect interfacial bonding, but it demonstrates the enormous potential of continuous carbon fibers in ductile iron composites for sand casting parts.
Another important aspect was the effect of nickel plating on the protection of carbon fibers against harmful reactions with iron. The EDS results showed that uncoated fibers after high-temperature exposure contained iron, which suggested the formation of iron carbides (Fe₃C) or iron oxides. Fe₃C is a hard but brittle phase that can degrade the toughness of the composite. If formed at the interface, it acts as a stress concentrator and promotes brittle fracture. The nickel coating prevented direct contact between carbon and iron, thereby suppressing the formation of Fe₃C. In addition, the nickel layer reduced the oxygen diffusion to the fiber surface, as evidenced by the lower oxygen content in the nickel-coated fibers after casting. Thus, the nickel coating served both as a barrier and as a bonding promoter.
The experimental observations also revealed that the microstructure of the ductile iron in the vicinity of the fibers was affected by the presence of the nickel coating. Since nickel is an austenite stabilizer, it promotes the formation of pearlite or even martensite in the matrix near the fibers, depending on the cooling rate. This can alter the local hardness and residual stress distribution around the fibers. While this aspect was not deeply studied in this thesis, it is worth investigating in the future because it can influence the overall mechanical behavior of the composite. The coefficient of thermal expansion (CTE) mismatch between carbon fiber (about 1.5×10⁻⁶/°C) and ductile iron (about 12×10⁻⁶/°C) generates thermal residual stresses at the interface upon cooling. The nickel coating, with an intermediate CTE, helps to alleviate these stresses and prevents interfacial cracking. This is another reason why nickel-coated fibers resulted in better mechanical properties than uncoated fibers.
During the experimental design, I encountered several practical problems related to the handling and positioning of continuous carbon fibers in sand casting parts. Because the density of carbon fiber is much lower than that of molten iron, the fibers tend to float and align perpendicularly to the gravity direction. I overcame this problem by pre-tensioning the fibers across the sand mold cavity. The fibers were held in place by the compaction of the sand, which provided enough mechanical anchoring to resist the buoyancy and flow forces during pouring. In the initial approach, when I inserted fibers after the mold was prepared, the steel pins could not withstand the high temperature and got softened, causing the fibers to shift. The pre-embedding method proved superior. I also learned that the fibers should be arranged along the centerline of the tensile specimen, which is the location of maximum axial stress. This ensures that the reinforcing effect is directly measured in the tensile test. For actual crankshaft production, the fibers could be placed in the regions of maximum stress, such as the fillets or the journal areas. The concept of using a preform made of carbon fibers and metallic powder could be adapted to complex geometries. For example, a viscous pasty preform containing short or continuous fibers could be placed in the mold cavity before closing the mold. During pouring, the metal infiltrates the preform, creating a localized reinforced zone. This approach is currently under development in my research group.
I also analyzed the microstructure of the fractured surfaces to understand the failure modes. The uncoated fiber specimens failed in a brittle manner with a flat fracture surface. The carbon fibers appeared as long pull-out fibers protruding from the metallic matrix, with lengths of several millimeters. This pull-out consumed little energy and contributed little to the tensile strength. The nickel-coated fiber specimens showed a more irregular fracture surface, with fibers broken at the fracture plane or slightly pulled out. The presence of nickel improved the frictional sliding resistance during pull-out, allowing the fibers to break instead of being simply extracted. This indicates that the interfacial shear strength was high enough to transfer the load to the fiber, causing fiber fracture. The fracture surfaces exhibited small dimples on the metal side, typical of a ductile fracture in the iron matrix, while the fiber tips were covered with a thin layer of iron, suggesting strong adhesion.
In summary, my research demonstrated that continuous carbon fiber reinforcement of ductile iron via sand casting is feasible, provided that the carbon fibers are surface-modified with a nickel coating and properly positioned within the sand mold. The nickel coating significantly improves the interfacial wettability between the carbon fibers and the molten iron, leading to complete infiltration and a defect-free interface. The resulting composite exhibits an average tensile strength increase of 98 MPa compared to bars with uncoated fibers. The nickel layer also mitigates the oxidation of carbon fibers during the high-temperature casting process, reducing the formation of harmful byproducts such as iron carbides. The interface is characterized by a diffusion layer of nickel into the iron matrix, forming a Fe-Ni-C metallurgical bond. This bond enhances the load transfer efficiency and enables the fibers to contribute to the composite strength.
For future work, I propose several directions to further improve the performance of this composite. First, the fiber volume fraction should be increased while maintaining uniform distribution and complete infiltration. This might require the use of pressure-assisted casting or vacuum-assisted infiltration. Second, the oxygen content inside the mold should be minimized by using a vacuum chamber or by introducing an inert gas in a closed-loop system. Alternatively, the carbon fibers can be coated with a ceramic barrier layer, such as SiC or Al₂O₃, in addition to the nickel layer. Third, the use of three-dimensional woven fiber preforms could provide reinforcement in multiple directions, which is beneficial for crankshaft applications where loads are complex. Fourth, numerical simulations of the casting process, including flow, heat transfer, and thermal stress, could help optimize the fiber placement and predict the residual stresses at the interface. Fifth, long-term fatigue tests and wear tests should be conducted to evaluate the durability of the composite under cyclic loading conditions, which are relevant to crankshaft operation.
To conclude, this thesis contributes to the understanding of continuous carbon fiber reinforced ductile iron composites produced by sand casting. The findings show that careful surface modification and proper mold design are the keys to successfully embedding carbon fibers in iron matrices. The nickel coating acts as a dual-functional layer: it enhances wettability and bonding while protecting the fibers from chemical degradation. The improvement in tensile strength demonstrates that carbon fibers can be used as effective reinforcements in ductile iron, opening new possibilities for lightweight, high-performance sand casting parts in automotive and other engineering applications. The results provide a solid foundation for future research and industrial implementation of fiber-reinforced metal matrix composites in sand casting parts.