2019 Volume 60 Issue 6 Pages 1011-1017
High chromium cast irons show superior abrasion resistance due to their chromium carbides. Their abrasion resistance is improved by insert casting with cemented carbide. The effects of high-temperature exposure during insert casting on the microstructures of cemented carbide were investigated in this research. The high chromium cast iron (2.7%C–27%Cr) and the cemented carbide round bars (WC–13.7%Co) were prepared. The round bars were dipped in molten high chromium cast iron at 1596 K. The dipped round bars were pulled up after the elapse of 30–180 s. Microstructures of dipped round bars were changed from homogeneous sinter structure to three-layer structure, cemented carbide, diffusion layer, and reaction layer. The thicknesses of the diffusion layer and the reaction layer were increased with increasing of dipping time. FE-EPMA analysis revealed that the diffusion layer was formed by the elution of Co from the cemented carbide and diffusion of Fe and Cr from the molten high chromium cast iron into the cemented carbide round bar. In addition, rectangular particles were randomly distributed in the diffusion layer. The equivalent circular diameter of the rectangular particle was increased with increasing dipping time. The Vickers hardness of the diffusion layer decreased about 30% relative to the cemented carbide but higher than that of high chromium cast irons. The inserted cemented carbide is thought to have contributed to improving abrasion resistance. It was suggested that thin diffusion layers are more effective for improving abrasion resistance.
This Paper was Originally Published in Japanese in J. JFS 90 (2018) 217–223. The abstract, background, experimental procedures, results and discussion have been revised.
High chromium cast irons (HCCIs), containing 2.5–3.5% carbon and 15–30% chromium, show superior wear resistance because of their chromium carbide content and are used for crushers and steel-rolling rolls.1–3) To improve the wear resistance of HCCIs, insert casting (IC) with cemented carbide (CC) has been researched and developed.2–7) It has been reported that using IC to make crusher teeth increases their lifetimes by more than fivefold.6) It is expected that HCCIs with IC will be used in applications involving severe abrasion environments.
Previous studies have mainly been concerned with the interface between the HCCI and the inserted CC. Aso et al. reported that reaction layers (RLs) form between the CC tip and the HCCI, and Matsubara et al. reported that an abnormal phase forms between the CC ball and the HCCI.5,6,8) However, little attention has been paid to the microstructure of the inserted CC inside the aforementioned RLs or abnormal phase. The hardness of the inserted CC might also change when the CC microstructure changes. Therefore, in the present research, we investigated the effects of IC on CC microstructures.
An HCCI and a commercially available WC–Co CC round bar (CCRB) were prepared. The chemical composition of the HCCI is given in Table 1. The CCRB comprised WC particles and Co as a binder. The WC particles had a rectangular shape of ∼1 µm, and the Co content was 13.7%. The CCRB had a diameter of 4 mm, a length of 50 mm, and a weight of 9 g.
Approximately 150 g of the HCCI was melted and maintained at 1596 K in a vertical type electric furnace as shown schematically in Fig. 1. High-purity Ar flowed into the furnace tube from the bottom at a flow rate of 2 L/min to form an inert atmosphere. The molten-metal temperature was measured with an R-type thermocouple penetrating the stage and contacting the bottom of the ceramic crucible.
Schematic illustration of experimental system for dipping.
The CCRB hung with an R-type thermocouple strand and was dipped quickly into the molten metal for times in the range of 30–180 s. After being dipped for the prescribed time, the CCRB was pulled up by a motor at a speed of 200 mm/min.
2.3 Microstructure, elemental analysis, and hardnessDeformation considered to be due to being pressed against the crucible was observed up to 5 mm from the bottom side of the dipped specimens. 10 mm from the bottom side of the dipped specimens, where no such deformation was observed, was taken as the position at which to observe the microstructure. The dipped specimens were cut, embedded with resin, polished, and etched, and their microstructures were observed using an optical microscope. Marble liquid was used as the etching solution; its composition was 4 g of copper sulfate, 20 cm3 of hydrochloric acid, and 20 cm3 of water.
For high-magnification microstructure observation and elemental analysis, a field-emission electron probe micro-analyzer (FE-EPMA) was used with an acceleration voltage of 20–25 kV. In the line and surface analyses, the specimens were coated with carbon to prevent the charging phenomenon. In contrast, no such a phenomenon did not occur in the point analysis of the specimens without coating.
On the basis of JIS Z 2244: 2009, the Vickers hardness test was performed at 0.9807 N.
The microstructure of the specimen dipped for 30 s is shown in Fig. 2. As can be seen from the cross-sectional microstructure (Fig. 2(a)), the homogeneous sintered structure was changed to a three-layer structure. From the center of the specimen, the three layers are CC, a diffusion layer (DL), and an RL. Both the CC and the DL were similar to the sintered structure with no appearance of melting. In the DL, rectangular particles (RPs) as shown in Fig. 2(b) were distributed randomly. Figure 2(c) shows the microstructure observed from a vertical section; the RPs did not differ in shape from those observed in cross section, and they were not connected with the CC or the RL. Figure 2(d) shows the vertical section of the RL, which exhibits the dendritic shape, indicating that RL is a solidified structure.
Optical micrographs of 30 s dipped specimen, (a) cross section, (b) rectangular particles, (c) vertical section, and (d) reaction layer.
The microstructures of specimens dipped for 30–180 s are shown in Fig. 3. A three-layer structure and RPs were observed in each specimen, and there was hardly any change in the specimen diameter. The relationships between the dipping time and the thicknesses of the DL (lD) and the RL (lR) are shown in Fig. 4. Both lD and lR were calculated from the distances from the specimen center to the DL, the RL, and the outermost surface by using the cross-sectional photographs (Fig. 3). Both lD and lR increased with dipping time, reaching 0.8 and 0.1 mm, respectively, in the specimen dipped for 180 s. The relationships between the dipping time and (i) the equivalent circular diameter d of the RPs and (ii) their number N are shown in Fig. 5. The value of d for RP was obtained by image analysis using the photographs (Fig. 3), whereupon the average value was calculated. The value of d increased with dipping, whereas N increased initially but then decreased for dipping times greater than 60 s.
Optical micrographs of 30–180 s dipped specimens.
Increase in diffusion layer thickness (lD) and reaction layer thickness (lR) with increasing dipping time.
Change of equivalent circular diameter d and number N of rectangular particle with increasing dipping time.
Figure 6 shows the microstructure of the specimen dipped for 180 s as observed using the FE-EPMA. Compared with the CC shown in Fig. 6(a), the WC particles in the DL shown in Fig. 6(b) were widely spaced. In the RL shown in Fig. 6(c), the WC particles were observed with white contrast, shown as rounded and irregularly sized. The RP microstructure shown in Fig. 6(d) was similar to that of the RL.
SEM micrographs of 180 s dipped specimen, (a) cemented carbide, (b) diffusion layer, (c) reaction layer, and (d) rectangular particle.
The X-ray intensities of W, Co, Fe, and Cr measured by FE-EPMA line analysis are shown in Fig. 7. The dotted line in the microstructure photograph indicates the analysis position. The gray region marked “a” indicates the vicinity of the boundary between the CC and the DL, and those marked “b” and “c” indicate RPs. The W X-ray intensity changed only in the “a” region; it remained constant in the CC and the DL, including where there were RPs. The Co X-ray intensity increased initially in the “a” region and then decreased monotonically toward its outer periphery. However, the Co X-ray intensity in the RPs (“b” and “c”) was significantly higher than that in the surrounding DL. Fe was detected in the DL, RL, and RPs with an X-ray intensity that tended to be opposite to that of Co. The Cr X-ray intensity was also generally opposite to that of Co, but remained constant where there were RPs.
FE-EPMA line analysis of W, Co, Fe, and Cr in 180 s dipped specimen.
The DL, RL, and RP element mappings are shown in Fig. 8. In the DL, we found Co, Fe, and Cr distributed around the WC particles. In the RL, we found the WC particles dispersed in the phase containing Co, Fe, and Cr. Co was distributed on the inner peripheral side (lower left-hand side), whereas Fe and Cr were distributed on the outer peripheral side (upper right-hand side). We detected W from the WC particles and the surrounding phase containing Co, Fe, and Cr. The number and size of WC particles in RP are different than those in RL; however, their microstructure is similar to those of the RL.
FE-EPMA elemental mapping of W, Co, Fe, and Cr in 180 s dipped specimen.
The results of the FE-EPMA point analyses of the CC, DL, RL, and RPs are given in Fig. 9, where the analysis positions are indicated schematically. As seen in Fig. 9, the concentration of each element in the CC was roughly constant in the radial direction. In the DL, the W and C concentrations were also approximately constant and together accounted for 75% of the total content. Co, Fe, and Cr accounted for the remainder but in markedly differing proportions. In the RL, although the measurement positions from the specimen center were almost the same as each other, the Co, Fe, and Cr concentrations differed markedly and some Si was also detected. The C concentration was also lower in the RL than in the CC and DL. Although the element concentration of each RP was fairly constant, the C concentration was lower than that in the surrounding DL and was close to that in the RL.
FE-EPMA spot analyses for 180 s dipped specimen.
The Vickers hardness values for the CC, DL, and RPs in the specimen dipped for 180 s are shown in Fig. 10. The test positions were the same as those in the FE-EPMA point analysis. However, because some cracks extended from the corner of an indentation in the RL, its hardness is excluded from Fig. 10. Compared with the hardness of 1900–2002 HV of the CC, that of 1265–1474 HV of the DL is 30% lower and that of 1829–1925 HV of the RP is 5% lower. In the DL, the concentrations of Co, Fe, and Cr (excluding the RPs) changed monotonically in the radial direction (according to the FE-EPMA line analysis), but the hardness did not change radially.
Vickers hardness of 180 s dipped specimen.
Previous studies have not reported about the thick DL observed in the present experiment.5–10) According to Matsubara et al., who used a CC ball and molten HCCI at 1773–1823 K, the mechanism for the three-layer structure can be explained as follows.8) When molten HCCI comes into contact with CC, the Co content of the CC melts, whereupon WC particles and Co elute into the molten metal.8) In the WC–Co pseudo-binary system, the eutectic temperature is 1573–1593 K.11–13) In the present research, elution of WC particles and Co could have occurred. However, in the present experiment, the molten-metal temperature was 1596 K, which limited the melting of Co at the surface of the CCRB. Furthermore, there was little outflow of WC particles. The dominant factors in the microstructural change were (i) increasingly farther-spaced WC particles, (ii) elution of Co into the molten metal, and (iii) diffusion of Fe and Cr from the molten metal. The RL formed by solidification of the melted surface of the CCRB, giving the bar its observed three-layer structure.
Assuming lattice diffusion of Co, Fe, and Cr in solid Co, the calculated diffusion distances in the sample dipped for 180 s are of the order of several micrometers at 1596 K.14) However, in the present experiment, Fe and Cr were found up to 0.9 mm from the surface of the CCRB toward its center, which is nearly three orders of magnitude greater than the calculated values. This indicates the existence of a mechanism for very-high-speed diffusion, considered as follows.
The fact that the CCRB maintained its shape and no solidified structure was seen in the DL shows that very-high-speed diffusion developed despite the Co being in the solid phase. In the sintering process of CC production, Co diffuses to the WC surface at a temperature lower than the WC–Co eutectic temperature.13) In addition, Silva et al. observed that Co particles placed on a WC plate wetted and spread at 1523 K, which is lower than the temperature at which the liquid phase occurs in the WC–Co pseudo-binary system.15) Silva et al. also clarified that Co could spread very rapidly when the WC particles were relatively fine and when high heating rates were used. Investigating the volume shrinkage during sintering indicated that when fine WC particles were used, Co diffusion occurred at temperature less than 1073 K.15) The good wettability (contact angle: 0°) of Co to WC is one reason why Co diffuses at the surfaces of WC particles.16) In the present experiment, the WC particles were as fine as 1 µm, meaning a relatively large surface area as the diffusion path. The heating rate was also relatively high because of the melt dipping. Therefore, in the sample dipped for 180 s, it led to the development of a high-speed diffusion of 0.9 mm.
Fe and Cr diffused into the gaps that appeared because of Co eluting into the molten metal.5) Therefore, before Fe and Cr can diffuse into the CCRB, the Co therein must be eluted into the molten metal. The melting point of Co in CC decreases with the amount of Fe, Si, and C in the molten metal.9) Furthermore, because the specific heat of CC is about one quarter that of cast iron and the thermal conductivity of CC is about four times that of cast iron, the temperature increased rapidly upon dipping, and Co diffusion was promoted.10) Consequently, Co eluted actively from the CC to the molten metal, and Fe and Cr diffused into the CCRB.
The ratios of Fe to Cr in the HCCI as given in Table 1 and in the DL, RL, and RPs as shown in Fig. 9 are almost the same, suggesting that Fe and Cr share almost the same wettability with WC. Although WC–Cr wettability has received little attention, Fe is known to have good wettability with WC, similar to that of Co. Therefore, the WC wettability of Cr is thought to be equivalent to that of Fe.
As shown in Fig. 10, the hardness of the DL considered to be formed by the aforementioned mechanism was reduced by 30% compared with that of the CC, and the dispersion was large. DL hardness changed because the WC particle spacing increased in the DL, allowing higher concentrations of Co, Fe, and Cr and causing sparse/dense distribution of the WC particles. Furthermore, higher Co concentration leads to lower hardness and higher Fe and Cr concentrations were considered to reduce the hardness similarly.17) Although the details of how Cr acts as a binder remain unclear, the hardness was almost the same even if Fe was used instead of Co.18) This was considered the reason why no monotonic change in hardness was found in the DL in the radial direction. Nevertheless, the DL hardness was 1.6–2.6 times higher than the 500–820 HV of the HCCI, thereby improving the wear resistance.1) However, the hardness reduction of the DL is considered to be inferior to the original wear resistance of the CC. To prevent the abrasion resistance from deteriorating, it is necessary to expose the CC to high temperatures for less time, thereby reducing the DL thickness.
Because the RL observed in the present experiment exhibited a dendritic shape, at least part of it was in the liquid phase. Aso et al. observed rounded WC particles in a solidification phase in which W dissolved when the CC tips were inserted with the HCCI.5,6) The above microstructure agrees well with the results of the present experiment. The mechanism for RL formation was that WC dissolved in the solidification phase by the interdiffusion of Co, Fe, and Cr, and any WC particles not dissolved were considered to be dispersed.
4.2 Characteristics of RPsOkada et al. reported that an RL formed between the cast steel and the inserted WC–6%Co CC, and that Fe3W3C RPs crystallized in the RL.19,20) However, Fig. 6 and 8 suggest that the morphology of the RPs observed in the present research was due to the dispersion of WC particles in Co–Fe–Cr–W–C alloy. The RPs were distributed irregularly in the DL, indicating that local melting occurred only where the RPs formed.
Although such a mechanism is yet to be confirmed, we examined its possibility by conducting a dipping experiment with a gray cast iron (3.99%C–2.22%Si–0.03%Mn–0.022%P–0.009%S) to prepare a specimen that had been dipped for 30 s. Figure 11 shows that its microstructure changed to a three-layer structure similar to that of HCCI, but no RPs were observed in the DL. Although contrasting black spots were observed in the DL, these were merely aggregates of places where the WC particle spacing had increased and the density had decreased. The reason why no RPs appeared to form upon dipping in the molten gray cast iron is more likely due to the different melt composition than to any issue with the experimental method.
Optical micrograph of specimen dipped in molten gray cast iron for 30 s.
The variations in equivalent circular diameter d and number N of RPs with dipping time shown in Fig. 5 can be explained by the coalescence of adjacent RPs and the temperature increase. The longer the dipping time, the more the temperature of the CCRB increased, the more that localized melting occurred, and the greater the area of the already molten portion. Although d and N both increased initially, because the thickness of the DL was limited, N then began to decrease because of the coalescence of adjacent RPs. Such behavior is similar to Ostwald growth.
As shown in Fig. 10, the hardness of the RPs decreased by 5% while that of the surrounding DL decreased by 30%. The increased WC particle spacing and the high Co, Fe, and Cr concentrations led to decreased hardness, as described in Section 4.1. However, we reason that the dissolved WC and dense solidified structure increased the hardness. Suzuki reported that WC particles are attracted by flow of the liquid phase and by the liquid generated between WC particles, whereby densification is promoted.16) This increased hardness due to densification apparently limited the reduction in the hardness of RPs to 5%.
In the present research, we clarified the effects of high-temperature exposure during IC on the microstructure of CC as follows:
