The influence of ball milling processing variables on the microstructure and compaction behavior of Fe–Mn–Cu alloys

Table 1 displays the design input parameters for the present study, which includes 27 experiments conducted on the [(Fe–35wt%Mn)_100−x–Cu_x] system, where x = 0 wt%, 5 wt%, and 10 wt%. The processing variables comprise the MT (1 h, 5.5 h, and 10 h), BPMR (5:1, 10:1, and 15:1), and MS (100 rev/min, 200 rev/min, and 300 rev/min). Ball milling of the as-received Fe, Mn, and Cu powders, with purity >99.5% (supplied by Nanografi, Germany), was carried out using Pulverisette 5/2 classic line (Fritsch GmbH - Milling and Sizing, Idar-Oberstein, Germany) through simultaneous milling in two tungsten carbide (WC) containers (250 mL capacity each) using 10-mm-diameter balls made of tungsten carbide. The process was performed continuously to complete the MT by repeating a milling cycle comprising four stages: vial rotation in the forward path for 15 min; a stop for 15 min; vail rotation in the reverse path for 15 min; and one more stop for 15 min. The two stops in each cycle were implemented to inhibit overheating inside the milling containers. For each experiment, the as-received powders were premixed based on the designed composition and then the premixed powders were ball-milled for the designed MT, MS, and BPMR, according to the design of experiments shown in Table 1. The ball milling was conducted in a wet medium of ethanol (pureness >99.9%) to inhibit oxidization and agglomeration of the developed alloy powders.

All processed samples were examined by X-ray diffraction (XRD) to identify the formed phases and to calculate the crystallite size and lattice strain under various processing conditions. An XRD machine (Empyrean, Malvern Panalytical) was used with a radiation source of Cu-kα to investigate the specimens at a rate of 0.6°/min and a step size of 0.01° with a range of 2θ from 20° to 80°. The obtained XRD data were analyzed using the X’Pert High Score Plus software for phase identification and determination of crystallite size and lattice strain. The crystallite size was calculated using the Scherer formula [Eq. (1)], based on the major Fe peak only, and the homogeneous lattice strain was determined from the ratio of the deviation of standard and calculated d-spacing in relative to the standard d-spacing [38]. The Williamson–Hall method can also be used to determine the crystallite size and lattice strain. However, here, only two major XRD peaks were observed and, hence, Scherer formula was used. Furthermore, the instrumental broadening was accounted for according to the basic equation and inbuilt in the software. The XRD machine gave the data after deducting the instrumental broadening [according to Eq. (2)]. Scherer formula [Eq. (1)] was used to determine the crystallite size as follows: (1) t=kλβcosθ t=\frac{k\lambda }{\beta \cos \theta } where t is the crystal size, λ is X-ray wavelength, θ is Bragg's angle, and β is the full-width at half-maximum. The instrumental broadening was corrected using Eq. (2) as follows: (2) βhkl=[(βhkl)Measured2−(βhkl)Instrumental2]1/2 {{\beta }_{hkl}}={{\left[ \left( {{\beta }_{hkl}} \right)_{\text{Measured}}^{2}-\left( {{\beta }_{hkl}} \right)_{\text{Instrumental}}^{2} \right]}^{1/2}}

A field-emission gun high-resolution scanning electron microscope (FEG-HR-SEM; Apreo; ThermoScientific) coupled with an energy-dispersive spectrometer (EDS) was used for investigating the microstructure of the selected samples under varying processing conditions. The size and morphology of the as-received and processed powders and the elemental mapping of the processed powders and green samples were examined. Additionally, particle size analysis was performed on selected samples of the processed alloys by Zetasizer.

Compaction experiments were conducted to study the effect of the processing variables (shown in Table 1) on the densification. The compaction experiment was performed using an MTS universal testing machine (MTS Systems Corporation, USA) and a hardened H13 tool steel die with an inner diameter of 15 mm. All compaction experiments were performed at room temperature by applying CP ranging from 25 MPa to 1,100 MPa at a loading rate of 1 mm/min. RD–CP curves were generated from the compaction experiment.

RSM, which is a tool for the design of experiments, was used to analyze the effect of the input parameters on the compressibility/densification, in terms of RD response of the studied green samples, by identifying the most significant variables affecting the densification response. The central composite design was used to perform the experimentations. Twenty-seven experiments were carried out as given in Table 1. The repeated conditions in Table 1 (Experiments 19, 26, and 27) are essential for the design of the experiment by the central composite design. These repeated experiments are necessary for the accurate determination of the center of the space in the central composite design. The input parameters in the present study include MT, BPMR, MS, and alloy composition (CC). The measured response in this study was the RD. It should be highlighted that the RD listed in Table 1 was obtained at the maximum applied load (1,100 MPa) during compaction. The design of the experiments and the analysis of the results obtained in the present study were performed using Minitab software (version 15). ANOVA test was used to determine the most significant input parameters.

Figure 1 shows the secondary electron images (SEIs) of the as-received Fe, Mn, and Cu powders used in the present study. Figure 1A illustrates the spherical particles of Fe powders with size <10 μm. Figure 1B displays Mn powders with facet shape and particle size <80 μm. Figure 1C illustrates the Cu powders’ dendritic shape with an average size <120 μm.

Fig. 1

Figure 2 shows the SEIs of three selected samples to illustrate the effect of BPMR and MT, as examples, on the size and morphology of the processed alloy powders. Figure 2A displays the BD5-1 alloy with processing conditions of 5 wt%Cu, 1 h MT, 200 rev/min MS, and 5:1 BPMR. Figure 2B exhibits the processed powders of BD5-3 alloy with input parameters of 5 wt% Cu, 1 h MT, 200 rev/min MS, and 15:1 BPMR. Figure 2C reveals the processed powders of BD5-4 alloy with conditions of 5 wt% Cu, 10 h MT, 200 rev/min MS, and 15:1 BPMR. It is obvious from these images that on increasing the BPMR (BD5-1 vs. BD5-3 samples), the powder particles in Figure 2B became more refined (the circled regions) and flattened (arrowed), compared with the particles in Figure 2A. Furthermore, increasing the MT (BD5-3 vs. BD5-4 samples), the powder particles in Figure 2C displayed a large decrease in the particle size, as seen in the circled regions, and an additional flattening in the particles’ shape, as seen in the arrowed particles. Increasing MT and/or BPMR resulted in more energy input during the milling process, which applied more collision action inside the milling vails and produced finer powders, as shown in Figure 2C.

Fig. 2

Figure 3 presents the results of the EDS analysis of the same selected alloys (BD5-1, BD5-3, and BD5-4). Figures 3A, 3D, and 3G show the SEIs for the BD5-1, BD5-3, and BD5-4 alloys, respectively. Figures 3B, 3E, and 3H show the elemental maps for the BD5-1, BD5-3, and BD5-4 alloys, respectively; these maps are in full agreement with the results shown in Figure 2. The EDS spectra in Figures 3C, 3F, and 3I confirm the alloys’ chemical composition. Increasing BPMR from 5:1 in the BD5-1 alloy – shown in Figures 3A and 3B – to 15:1 in the BD5-3 alloy – shown in Figures 3D and 3E – was observed to improve the elemental dispersion, with a notable refining in the metal particles. Moreover, increasing the MT from 1 h in BD5-3 alloy (illustrated in Figures 3D and 3E) to 10 h in BD5-4 alloy (shown in Figures 3G and 3H) resulted in an additional refining of the powder particles and improved the elemental dispersion with a notable flattening/deformation of Fe particles due to milling action.

Fig. 3

The alloys selected for particle size analysis include the following: BD5-1 vs. BD5-2 and BD5-9 vs. BD5-10 to show the effect of MT; BD5-2 vs. BD5-4 to illustrate the impact of BPMR; and BD5-9 vs. BD5-11 to display the influence of MS, as shown in Table 1. Figure 4 and Table 2 display the results of particle size analysis of the selected samples. The average particle size of the selected samples was measured based on the main peak shown in Figure 4. Figure 4 and Table 2 show that increasing the MT from 1 h (BD5-1 sample) to 10 h (BD5-2 sample) resulted in decrease in the average particle size from 1,477 nm to 655.40 nm, respectively. Moreover, raising the BPMR from 5:1 (BD5-2 alloy) to 15:1 (BD5-4 alloy) resulted in an additional decrease in the average particle size from 655.40 nm to 327.10 nm, respectively. Furthermore, increasing the MT from 1 h (BD5-9 sample) to 10 h (BD5-10 sample) resulted in decrease in the average particle size from 1,274 nm to 638.70 nm, respectively. In addition, raising the MS from 100 rev/min (BD5-9 alloy) to 300 rev/min (BD5-11 alloy) resulted in a significant decrease in the average particle size from 1,274 nm to 499.50 nm, respectively. The results presented in Figure 4 and Table 2 confirmed the effect of ball milling input parameters (MT, BPMR, and MS) on the particle size of the processed powders. Increasing one or more of the input parameters resulted in a significant effect in terms of decreasing the particle size of the processed powders due to the increased energy input of milling and the strong mutual collision among powders, balls, and vail walls. These effects activate the common mechanism of mechanical alloying, which includes deformation, welding, breaking, and rewelding of the particles, leading to increased particle refining and alloying of the mixed powders. The polydispersity index (PDI) values listed in Table 2 confirmed the uniformity of the obtained particle size for the processed alloys, and the distribution of these particle sizes lies over a narrow range. Generally, a PDI value <0.7 indicates an even distribution of particle size over a small size range, which is the case for all samples presented in Table 2 and Figure 4.

Fig. 4

XRD analysis was performed on all samples to illustrate the effect of the input parameters of ball milling on the crystallite size and homogeneous lattice strain of the processed powders. The results of the selected samples (from Table 1) are discussed in this section. The selected samples include BD5-1 vs. BD5-2 to display the effect of MT; BD5-1 vs. BD5-3 to reveal the influence of BPMR; and BD5-9 vs. BD5-11 alloys to show the effect of MS. Figure 5 displays the XRD patterns of all samples for XRD analysis. The peaks were identified using X’Pert High Score Plus software, where the Fe [reference International Centre for Diffraction Data [ICDD]: 01-089-4186, space group: Im-3m(229)], Mn [reference ICDD: 00-001-1237, space group: I-43m(217)], and Cu [reference ICDD: 01-085-1326, space group: Fm-3m(225)] peaks were recognized, as illustrated in Figure 5. The estimated lattice strain and crystallite size are listed in Table 1. These features were determined using X’Pert High Score Plus software based on the major Fe peak only in each XRD pattern. With regard to the crystallite size, it was observed that increasing the MT from 1 h to 10 h resulted in a notable decrease in the crystallite size at various BPMRs and different MSs. The determined crystallite size for the BD5-1 sample milled for 1 h was 57.7 nm, compared with 18.2 nm in the case of the BD5-2 sample milled for 10 h. Similar observations when increasing the MT from 1 h to 10 h but at large BPMRs (BD5-3 vs. BD5-4 alloys) led to decrease in the crystallite size from 37.7 nm to 22.3 nm, respectively. The homogeneous lattice strain was noted to increase with increase in the MT, as shown in Table 1. For example, the estimated lattice strain for the BD5-1 alloy milled for 1 h was 0.165%, compared with 0.507% in the case of the BD5-2 alloy milled for 10 h. The effect of BPMR on the crystallite size and lattice strain was also examined, as listed in Table 1. For instance, increasing the BPMR from 5:1 (BD5-1 alloy) to 15:1 (BD5-3 alloy) led to a reduction in the crystallite size from 57.7 nm to 37.7 nm and an increase in the homogeneous lattice strain from 0.165% to 0.248%. Furthermore, the effect of MS on the crystallite size and lattice strain was investigated, as listed in Table 1. The increase in MS from 100 rev/min (BD5-9 alloy) to 300 rev/min (BD5-11 alloy) led to a considerable decrease in crystallite size from 57.6 nm to 21.3 nm and a substantial increase in the homogeneous lattice strain from 0.167% to 0.461%. Furthermore, increasing the energy input of the milling process resulted in elemental dispersion and alloying through the common mechanism of mechanical alloying, including deformation, welding, breaking, and rewelding of the elemental powders, leading to increased elemental alloying/dissolution of the mixed powders. For example, when comparing BD5-9 vs. BD5-11 samples, the energy input of milling for the BD5-11 alloy is higher than that of the BD5-9 alloy due to the high MS of 300 rev/min. Consequently, BD5-11 alloy displays greater peak broadening in the XRD patterns, with the disappearance of some peaks, compared with the XRD pattern of the BD5-9 alloy in Figure 5. These observations match the elemental mapping analysis presented in the next section wherein increased elemental dispersion and low agglomeration were observed for BD5-11 compared with the BD5-9 sample. These are indications of the formation of large extents of solid solution in the BD5-11 alloy compared with the BD5-9 alloy. Therefore, the extent of solid solution formation in the processed samples depends on the input parameters of the process. These findings ascertain that these parameters (MT, BPMR, and MS) play a major role in determining the microstructure features of the processed alloys. Augmenting the MT and/or BPMR and/or MS produced a substantial effect in terms of decreasing the crystallite size, increasing the lattice strain, and increasing the extent of solid solution formation. All these changes directly affect the subsequent material behavior, such as the compressibility and densification of the powders, as elaborated in the following sections.

Fig. 5

The compaction experiments were performed according to the input parameters listed in Table 1. The RDs at the maximum pressure applied at 1,100 MPa are shown in Table 1. Figure 6 displays the RD versus compaction load for selected samples of [(Fe–35 wt%Mn)₉₅–Cu₅]. From Figure 6, it is obvious that the RD for all samples revealed a continuous improvement with increase in the CP. However, the improvement rate in RD varied according to the controlling mechanisms of powder compressibility [32]. From Figure 6, a sharp increase in the RD of the samples up to ∼ 60 MPa is seen due to the particle ordering and rapid reduction in the spacing among the powders. Figure 6 also shows that in the CP ranging from ∼ 60 MPa to ∼ 200 MPa, the RD is still increasing but with a reduced slope due to the predominant mechanism of powder deformation, which led to extra contact areas among the powders and reduced pores [34]. Pressures >200 MPa were observed to produce more improvements in the compressibility until the maximum values of RD were reached at 1,100 MPa; however, the curve in this region revealed a lower slope due to the increased strain hardening of the powders, which resist further deformation and impede additional increase in the contact areas among powders [35]. With regard to the behavior of the selected samples presented in Figure 6, these 13 samples displayed the same general behavior of RD versus CP; however, these samples did not reveal the same RD at the applied loads, specifically, at the maximum CP of 1,100 MPa. The BD5-5 alloy displayed the maximum RD (75.46%), while the BD5-8 alloy exhibited the minimum RD (64.30%). The remaining 11 samples presented in Figure 6 were observed to display various RD values with CP between the maximum and minimum behaviors according to the various input parameters (Table 1). The sample with exceedingly high energy of ball milling involving the highest BPMR (15:1) and MS (300 rev/min) displayed the least value of RD due to the large stored energy, increased lattice strain, and great amount of work hardening in the processed powders. Therefore, the processed powders resist the compaction process and the final RD was reduced. On the other hand, the highest RD in Figure 6 was attained with a moderate energy of ball milling, which included BPMR (5:1), and MS (100 rev/min); in this case, the processed powders were still deformable with minimum strain hardening effect, which permitted more compressibility and enhanced the RD of the powders.

Fig. 6

Figures 7 and 8 show the effect of the input parameters (MT, BPMR, and MS) on the microstructure of these compacted green samples on investigation of the selected green samples by EDS analysis. Figures 7, 8A, 8D, and 8G reveal the SEI of the green compacts collected from the selected samples. The chemical composition of the selected samples was confirmed by the EDS spectra in Figures 7, 8C, 8F, and 8I. Figures 7, 8B, 8E, and 8H show the elemental mapping accomplished on the outlined regions of the selected samples. From the maps presented in Figures 7B, 7E, and 7F, it can be observed that increasing the BPMR from 5:1 (BD5-1 alloy) to 15:1 in (BD5-3 alloy) resulted in improvement in the elemental dispersion with a notable refining in the microstructure constituents, as shown in Figures 7B and 7E, respectively. Additionally, increasing the MT from 1 h (BD5-3 alloy) to 10 h (BD5-4 alloy) resulted in an additional refining of the microstructure constituents and enhanced dispersion of the elements, as revealed in Figures 7E and 7H, respectively. From the maps presented in Figures 8B, 8E, and 8F, it can be noticed that a similar effect was observed when increasing the MT from 1 h (BD5-9 alloy) to 10 h (BD5-10 alloy). Increasing the MS from 100 rev/min (BD5-9 alloy) to 300 rev/min (BD5-11 alloy) resulted in an extra refining of the microstructure constituents and an additional enhancement in the dispersion of the elements, as revealed in Figures 8E and 8H, respectively.

Fig. 7

Fig. 8

EDS analysis was used in the present study for qualitative and quantitative analyses of the microstructure of the powders and green samples obtained under different processing conditions. With regard to the accuracy of the quantitative EDS analysis, it depends on the structural homogeneity and density of the examined alloys. In general, quantitative EDS analysis of standard dense alloys with a homogeneous structure provides high accuracy and low scattering. On the other hand, a scattering in EDS results is attained for experimental samples that reveal low density and a relatively segregated microstructure. In the present study, the examined samples were in powder and green forms and were subjected to different processing variables (MT, MS, and BPMR), as shown in Table 1. Accordingly, these examined samples displayed different homogeneity levels and densities. In this case, qualitative analysis was used to perfectly illustrate elemental dispersion and other features such as microstructure refinement. Chemical analysis of the examined samples confirmed the presence of the alloying elements and their dispersion profiles for each sample; however, there was scattering in the quantitative chemical analysis values, which could be attributed to variations in the processing variables, the differences in sample homogeneity levels, and the low densities for the examined samples. The same reasons are responsible for the considerable influence observed on the compaction behavior of the samples, as shown in Figure 6 and the RD values listed in Table 1.

With regard to the input parameters and the experimental results displayed in Table 1, the ANOVA test was performed to identify the input parameters having significant and nonsignificant influence on the crystallite size, lattice strain, and compactibility/densification of the processed alloys in terms of RD. The results of the ANOVA test for all responses are shown in Tables 3–5 for the crystallite size, lattice strain, and RD, respectively. Tables 3–5 list the estimated statistical terms such as sum of squares, degree of freedom (DF), mean square, F-value, and p-value. The p-value is used to decide the significance of the input parameters [39]. According to the ANOVA test, a p-value < 0.0500 indicates the significance of the model terms, while p-value > 0.1000 refers to model terms that are nonsignificant [40]. Based on the ANOVA results for the crystallite size (Table 3), the most significant influencing linear parameter is the MT, followed by BPMR and MS. The Cu content was found to be a nonsignificant parameter. The square term of MT × MS also slightly influenced the crystallite size. Similarly, based on the ANOVA results for the lattice strain (Table 4), the only significant parameter was the lattice strain, and all other input parameters were not much significant. The results of the ANOVA test presented in Table 5 illustrate that all linear terms such as MT, BPMR, MS, and CC are statistically significant. Furthermore, only one interaction term (MT × CC) was observed to display statistical significance.

As an example, the correlation of the experimental and predicted results for RD with the corresponding normal probability of residuals is presented in Figure 9. Figure 9A reveals the actual RD experimental values versus the predicted values by the statistical analysis performed in the present study. Figure 9B displays the normal %probability versus residual, which assesses the model validity. Figures 9A and 9B show that almost all data points are adjacent to the straight line, which confirms the accuracy of the statistical analysis results obtained in the present study and the robust relationship between the predicted RD and the experimental results.

Fig. 9

The impact of the input variables on the compactibility/RD of the developed alloy systems is presented in the form of 3D surface plots and 2D contour plots, as shown in Figure 10. Figure 10 (A–L) can be used for forecasting the RDs of the studied alloys. Figures 10A and 10B reveal the effect of the MT and BPMR on the RD. From the surface plot in Figure 10A and the contour plot in Figure 10B, it can be observed that the optimum RD and compressibility can be attained by decreasing the MT and BPMR. This behavior is due to the fact that increasing the MT and the BPMR results in increasing the energy of milling, which leads to increased lattice strain (as discussed in Section 3.3) and excessive plastic deformation of the powders [32]. The increased lattice strain and severe deformation of the powders impede the compressibility of the powders and decrease the RD [35]. Figures 10C and 10D show the impact of the MT and MS on the RD. From Figures 10C and 10D, it can be noticed that the RD decreased on increasing the MT and MS. Increasing the MT and speed led to high energy input during milling, which produced extra work-hardened powders with a substantial amount of lattice strain, which resists the compression and reduces the RD [34]. Figures 10E and 10F reveal the impact of MT and CC on the RD. Concerning Figures 10E and 10F, it can be seen that the RD improved by decreasing the MT and increasing the CC. Increasing the CC led to an improved RD because copper is a ductile element and increasing its content promoted the mechanism of compaction process through providing more deformability of the powders. Figures 10G and 10H illustrate the influence of MS and BPMR on the RD; it can be observed that the powder's compactibility and RD were enhanced on reducing the MS and/or BPMR due to the decreased power input during milling. Figures 10I and 10J show the influence of BPMR and CC on the RD, and it can be concluded that the powder's compactibility and RD were enhanced on increasing the CC and/or reducing the BPMR. Figures 10K and 10L display the influence of MS and CC on the RD; it can be perceived that the RD was enhanced on reducing the MS and/or increasing the CC.

Fig. 10

The abovementioned sections covered the investigation into the effect of the input parameters on the microstructure and compactibility/RD of the developed alloys. Based on this analysis, selecting the input parameters to attain the desired microstructure features and improved densification behavior is a challenge. It was found that increasing the ball milling parameters (MT, MS, and BPMR) resulted in significant enhancements in the microstructure, such as uniform elemental dispersion and reduced segregation of the developed alloys (e.g., Figures 7 and 8). Additionally, increases in the input parameters produced refined particles with a substantial decrease in their crystallite size (see Tables 2 and 3). On the other hand, increasing these parameters exhibited an undesirable influence on the powder compactibility and RD of the developed alloys, as displayed in Figure 10. Increasing MT, MS, and/or BPMR resulted in high amounts of stored energy, increased lattice strain, and large amount of work hardening in the processed powders. Therefore, the processed powders resist the compaction process, and the attained RD was negatively affected. Accordingly, a recommended combination of the milling parameters was selected to provide the optimum compromise among the microstructure features and powder compressibility/RD of the developed alloys. Based on the surface and contour plots shown in Figure 10, the following conditions would provide a reasonable combination of milling conditions to attain an acceptable compromise of enhanced microstructure features and improved compactibility and RD: MT – 5 h; MS – 150 rev/min; and BPMR – 10:1. In addition, as discussed earlier, increasing the CC resulted in a substantial improvement in the RD of the developed alloys; hence, 10 wt% Cu is recommended.