Microstructure and corrosion behaviour of plasma-nitrocarburized sintered steel

Powder characteristics and manufacturing processes determine the microstructure, and therefore, the physical, chemical, and mechanical properties of sintered steels. In particular, porosity and corrosion resistance are intimately related, since the contact area be-tween substrate and electrolyte significantly affects the corrosion resistance of sintered steels. This study addresses the effect of powder characteristics and pressing parameters on the microstructure and corrosion resistance of low-carbon sintered and sintered/plas-ma-nitrocarburized steel. The results indicated that the corrosion resistance increased with increasing density and decreasing specific surface area. Additionally, plasma-nitrocarburizing was highly effective in coating open pores of the material.


Introduction
Traditional applications for sintered steels usually do not require superior corrosion resistance. However, this scenario has gradually changed as novel high-performance products have pushed forward a rapid growth of the powder metallurgy market. Large scale use of sintered stainless steels remains an economic challenge due to the high cost of commercially available pre-alloyed powders [1]. Several solutions in order to improve the performance of powdered steels in corrosive environments have been investigated. Although both the quality of the starting powders and sintering process can be optimized by aiming at improved chemical stability, post-sintering surface treatments, such as nitriding and nitrocarburizing, are far more efficient [8,9,11,13]. Coating layers capable of improving bulk hardness [2], wear [3,4], fatigue endurance [5], and corrosion resistance [6 -14] can be formed in a single-step process. Conventional gas nitriding and salt bath treatments have been widely used in industrial plants, especially to coat steel and cast iron components. However, environmental hazards have driven companies towards replacing conventional nitriding methods, and adopting new approaches such as plasma-nitriding [15]. Furthermore, plasma-nitriding and nitrocarburizing of conventional steels offer a series of additional advantages over other methods, such as shorter treatment times, near-net shape, uniform coating of complex geometries, and tailored microstructures of nitrided layers by suitable adjustment of the plasma-reactor parameters.
The microstructure and properties of sintered components are mainly determined by the characteristics of the starting materials and the metallurgical process [16 -18]. In particular, corrosion resistance is especially affected by the homogeneity of the alloy, its density, and characteristics of residual pores, including pore size distribution and shape factor. Improving alloy homogenization increases the corrosion resistance by preventing the formation of localized galvanic cells. Increasing density and limiting pore size may also improve the corrosion resistance, as the contact area between the substrate and the electrolyte solution is reduced. The propensity for crevice corrosion may also be diminished.
Despite the above mentioned advantages, plasma-nitriding is generally inefficient at sealing relatively large pores, i.e., those with a few micrometers in diameter [6, 7, 10 -13, 19]. This limits the corrosion protection offered by nitrided layers, since the corrosive agent infiltrates into the sintered substrate via interconnected pores. Ideally, the component should be as dense as possible, and the remaining pores should be as small as possible. This can be achieved by using fine powders, optimized compaction, and liquid-phase sintering. The size distribution of secondary pores (shaped during sintering processes) is directly related to the particle size distribution of the master alloy [20,21]. Fine particles are characterized by high surface energies, which activate sintering and enhance mass transport [17,22,23]. As powder consolidation takes place, increasing pressure and multiple re-pressing cycles can be implemented to increase the final density. This approach is only limited by the quality of the die material, the loss of ductility resulting from strain hardening and the economics of increased production cost. In this scenario, the objective of the present study was to investigate the effect of compaction and nitriding parameters on the corrosion resistance of plasma-nitrocarburized sintered steels. Optimizing the process is a step forward in manufacturing cost-efficient powdered components capable of withstanding corrosive environments and applications thereafter.  Table 1.
Mn and C were added to the mixture as constituents of a master alloy of composition FeMnC. The carbon content of the alloys was adjusted by adding 0.08 w/o graphite to the powder mixture. The master alloy was classified in a strainer according to its particle size distribution and the fraction containing particles smaller than 45 lm was chosen.
Fe-powder was mixed to 5 w/o FeMnC, 0.08 w/o C and 0.6 w/o zinc stearate (solid lubricant) for 5400 s at 45 rpm in a Ytype mixer. The powder mixture was pressed into pellets 9.5 mm in diameter using 600 MPa in a double acting uniaxial press.
The pellets were pre-sintered in a horizontal furnace (Heraeus R0KIF7160) at 873 K for 3600 s in an atmosphere of 90% H 2 -10% CH 4 . The average heating rate was 0.5 K/s. After pre-sintering the pellets were re-pressed into pellets of 10 mm in diameter using a pressure of 600 MPa (samples P). Then one half of these samples were pre-sintered again and ground down to compensate the elastic deformation of the compact die and re-pressed one more time under similar conditions (samples W). All samples were then sintered at 1493 K using the same furnace employed in the pre-sintering stage. Finally, the samples were cleaned in acetone using an ultrasonic bath for 120 s and plasma-nitrocarburized. The identification of samples was, P for re-pressed one time and W for sample re-pressed two times.

Nitrocarburizing
Nitrocarburizing was carried out in atmosphere of 75% N 2 -23.75% H 2 -1.25% CH 4 , at different temperatures (783, 813, and 843 K) measured with a cromel-alumel thermocouple (Type K) and different holding times (7,200,14,400 e 21,600 s). The main parameters used in the process are summarized in Table 2. Prior to nitrocarburizing, the samples were cleaned in hydrogen, establishing a plasma discharge of 400 -

Microstructural and physical characterization
The density of the sintered samples was measured using "Archimedes" principle according to ISO 2738-1973 guidelines. Cross-sections of three samples of each set of experimental conditions were cut with a diamond disc, ground down, and polished to a 1 lm finish using diamond paste and alumina slurry. After surface preparation fifteen pictures of each sample were taken using Philips XL30 scanning electron microscope by backscattering electrons detector. These pictures were submitted to image analyzation with software Analysis Pro 2.11.002 in order to estimate representative porosity parameters, such as average pore diameter, size distribution, shape factor, and the real area correction factor. The average pore diameter was taken as the arithmetic average of the distances measured from the geometric center of the pore to its border at slopes ranging from 158 to 1808 at 158 steps. The shape factor was then obtained from In order to obtain the real area of the sample (area in contact with the solution, A), a homogeneous distribution of non-connecting pores was assumed to be representative of any longitudinal section of the sample. The real surface length was obtained from the binary images according to where L1 is the geometrical length of the cross-sectional surface, L2 is the real length, and RCF is the radius correction factor (see Figure 3). To perform the microstructure analyzation, the polished surfaces were etched in 2% Nital for ten seconds and then observed under a Carl Zeiss-Jena Neophot 30 optical microscope and a Philips XL3O scanning electron microscope.
The presence of nitride phases in the compound layer was identified by X-ray diffraction using Cu-Ka radiation (k ¼ 0.15406 nm). A Philips X' Pert diffractometer set to 30 mA and 40 kV was used to scan the angular range 308 2h 1008 at a rate of 0.018/s.

Mechanical characterization
Vickers microhardness profiles were measured in three identical samples of each set using a Shimadzu Hardness Tester. Each point in the profile corresponded to an average of five indentations created using a load of 0.025 kg applied for 15 s. The distance between adjacent measurements was about 30 lm. The hardening depth was taken as the distance from the surface where the hardness matched the bulk value.

Electrochemical characterization
Electrochemical tests were carried out at ambient temperature (295 AE 2 K) using a EG&G-Princeton Applied Research 273 Potentiostat/Galvanostat having a sample as its working electrode. The electrolyte consisted of a 0.5 M KNO 3 solution (pH ¼ 6) prepared from analytical graded KNO 3 and water from a MILLIPORE system. E corr vs. time runs were previously carried out for 3600 seconds in order to estimate the open circuit potential (E oc ) of the materials in the electrolyte solution, as well as to evaluate its corrosion potential. The tests indicated the propensity of the system to dissolution or formation of passivating films on the surface of the working electrode in open circuit conditions (E oc ). In order to reduce oxides originated during the E corr vs. time tests, the electrode was polarized to À 400 mV Â E corr for 120 s. The potentiodynamic curve was obtained scanning the potential range from À 250 mV to 1.6 V vs. E oc , at a rate of 0.8 mV/s. In addition, the occurrence of active dissolution, active-passive transition, passivation, and transpassivation were also investigated. The Tafel extrapolation methods were used to determine the corrosion rate. All the results were the mean of three tests.

Electrochemical cell
The electrochemical cell consisted of a 3-electrode system, having samples with geometric area 0.283 cm 2 as the working electrode, a saturated calomel electrode (SCE) coupled to a Luggin capillary as the reference electrode, and a double graphite bar as the counter-electrode. The system was immersed in a 0.5 M KNO 3 solution. The voltage obtained was relative to a Hydrogen Normal Electrode (HNE). Further details on the cell configuration can be found elsewhere [24 -26].

Statistical analysis
Changing the number of re-pressing cycles in the processing of sintered material it was possible to estimate the effect of this variable (re-pressing). The effect of time and temperature was estimated using a complete factorial experiment with a central point and three repetitions. The central point adopted for the model was: Response surfaces were obtained using the least squares method. Automated variance analysis and surface fitting were employed. Variance analysis was carried out using Pareto's plots for each response surface. A confidence level of 95% was attributed to independent parameters the effect of which was estimated to exceed 0.05, implying a significant action on the response of the system. The effect of meaningful process parameters on the thickness of nitrocarburized layers, hardening depth, and corrosion rate could therefore be established.  Table 3). The effective or real area was estimated for samples re-pressed either once (labeled P) or twice (labeled W), based on a reasonable assumption that no interconnecting pores were present at the density levels obtained (> 7,100 kg/m 3 ). The pore fraction area, distributed according to the mean pore diameter, is depicted in Figure 4. The total fraction of pores (A t ), measured from samples P and W, showed quite distinctly different results, with significant reduction in porosity proportional to the pore size range.
Following sintering, six samples were plasma-nitrocarburized at each condition (see Figure 2). Table 4 summarizes the numerical values obtained for the thickness of the compound layers and corresponding hardening depths as a function of the number of re-pressing cycles. Figure 5 shows that an increase in sample density (from 7,100 to 7,400 kg/m 3 ), related to an increase in the number of re-pressing cycles, resulted in thinner compound layers. Thicker layers were obtained for low substrate densities as a consequence of extended superficial area exposed to the plasma environment. Surface diffusion coefficients were generally higher than the corresponding vo-   Since surface area decreased as density increased, so did the relative importance of the corresponding diffusion coefficient. In addition, variance analyses showed that both nitrocarburizing temperature and time significantly affected the thickness of the compound layer. For samples repressed twice, only the temperature was found to be an important parameter. The fact that both the slope of the plot and the temperature-time interaction were not important parameters confirmed the adequacy of the fitting model employed.

Mechanical analysis
Denser samples (labeled W) exhibited thicker hardening depths at higher temperatures and longer times, contrary to what was obtained for samples P. Figure 6 shows the response surfaces for the hardening depth of those two samples. Although both plots had their maxima at the longest nitrocarburizing time, their appearances were quite distinct, as should be expected from the data presented in Table 4. As the volume of pores increased, lower nitrocarburizing temperatures resulted in deeper hardening depths, contrary to what was observed for denser samples. Variance analyses suggested that the effect of temperature, time, and their interaction were all determining aspects of the resulting hardening depth.
Microhardness profiles as a function of plasma treatment parameters (samples PA and PD) and number of re-pressing cycles (samples PD and WD) are shown in Figure 7. As can be seen, increasing the density of the substrate shifted the profile to lower hardness values. This was a result of the decrease in the contact area and the corresponding lower contents of nitrogen available to react with the substrate material. Figures 8 and 9 show potentiodynamic polarization plots illustrating the effect of the number of re-pressing cycles and nitrocarburizing parameters on the electrochemical behaviour of the material. Denser samples (samples W) were characterized by the presence of a passivating region (Figure 8). The corresponding current decreased as the density of the material increased. The effect of the nitrocarburizing step was to decrease the anodic current of those samples, indicating lower corrosion rates (Figure 9). The main parameters obtained from the electrochemical analyses are summarized in Table 5. Figure 10 shows the surface response for the corrosion rate of samples re-pressed once and twice. The surface corre-   Figure 10a (single re-pressing) was tilted with respect to its reference axes, clearly suggesting the determining aspect of temperature, time, and their interaction on the corrosion behaviour of the material. On the other hand, the surface depicted in Figure 10b (double re-pressing) suggests that, within the error limits, the average corrosion rates for each temperature-time set were virtually identical. The individual influence of those parameters and that of their interaction was negligible. Notwithstanding, all sets of nitrocarburizing parameters applied to the densest material (sample W) resulted in lower corrosion rates compared to plain sintered substrates.

Potentiodynamic and electrochemical analysis
The area correction factor was applied to estimate the effective sample area. The corresponding corrosion parameters are listed in Table 6. The corrosion rate decreased as the number of re-pressing cycles increased, corresponding to a decrease in the effective exposed area, due to the presence of fewer surface pores. As the pore surface area decreased, the electrolyte attack at microscopic regions became less intense, resulting in better corrosion behaviour for samples W (double re-pressing) compared to samples P (single re-pressing). Plasma-nitrocarburizing applied to sintered steel, as a

Surface analysis of the corroded material
Surfaces obtained before and after the potentiodynamic polarization tests performed on sintered/nitrocarburized double re-pressed samples are shown in Figure 11. Traces of uniform corrosive attack can be observed, affecting both interparticle areas, as well as binary regions of the compound layers. Signs of less intense attack can be observed in nitrocarburized samples, in contrast to plain sintered samples, which was in good agreement with the corrosion rates summarized in Table 5.

Conclusions
The following conclusions can be drawn from the results depicted herein: 1. Increasing the green density of the substrate resulted in thinner compound layers. Both nitrocarburizing temperature and time are important parameters in determining the thickness of nitrocarburized layers. Denser samples were mostly affected by the nitrocarburizing temperature.
2. Denser samples developed relatively thick hardening depths at higher temperatures and longer times. Lower nitrocarburizing temperatures were required to produce deeper hardening depths in less dense samples characterized by a larger volume of open pores. Temperature, time, and their interaction determined the resulting hardening depth.
3. Increasing green density decreased the contact area of the substrate and displaced the microhardness profiles towards lower hardness values. 4. Denser samples exhibited signs of passivation. The passivation current and onset potential decreased as the density of the material increased. Nitrocarburizing decreased the anodic dissolution current thus lowering the corrosion rate.
5. Uniform corrosion attacked both interparticle areas and binary regions of compound layers. Nitrocarburized samples were less damaged.
6. Plasma nitrocarburizing is an effective means of improving the corrosion resistance of sintered steel substrates provided the number of open pores is minimum. Coating should be extended to the inner surface of residual pores.