The tool temperature measured with the IR imaging setup showed that for all tested cutting conditions the temperature when machining CuZn38Pb3 is much lower than the temperature when cutting lead-free CuZn42 alloy, see Fig. 6. For vc = 200 m/min, f = 0.2 mm/rev, and ap = 2 mm (Fig. 7) the highest measured tool temperature for CuZn38Pb3 is around 180°C, which is significantly lower than the melting temperature of lead (Tmelt, Pb = 328°C) [34]. It is also interesting that the highest temperature is observed in the vicinity of the cutting edge or directly on the edge radius. The lead-free brass CuZn42 has nearly double the tool temperature, around 360°C, under identical cutting conditions. In addition, the hottest region is located behind the cutting edge, indicating a longer tool–chip contact length compared to machining CuZn38Pb3. The observed higher tool temperature can, in part, explain the shorter tool life reported for lead-free brass.
Figure 7 reports tool and tool–chip interfacial temperature. However, it is more difficult to measure the chip temperature because it relies on the emissivity of the brass alloy. For brass alloys, emissivity ranges ε = 0.03–0.5 [35] is reported. The lowest values occur for polished surfaces, and strongly increase with increasing surface roughness, degree of oxidation, and the temperature itself [36]. During machining the side of the chip simultaneously experiences a varying degree of change in roughness and oxidation, thus making the thermal data from chips drastically unreliable.
As already indicated by the thermal imaging, the tool–chip contact length appears to be different. High-speed filming shows the same trend as thermal imaging, but at more accurate temporal and spatial resolution. Figure 8 shows that for the leaded brass CuZn38Pb3, the chip is separated from the workpiece as discrete segments, unlike the chips produced from CuZn42 where the chip is segmented but continuous. As for the contact length, for CuZn38Pb3 brass it is comparable with the theoretical chip thickness and ranges lr = (0.6–1.0)∙h1, while for CuZn42 brass it is more than double, namely lr = (1.7–2.2)∙h1.
More remarkable is that the segmented chip formation for CuZn38Pb3 is independent of cutting speed, as seen in Fig. 9. Even at cutting speeds as low as 12 m/min, the chip formation follows the typical segmentation sequence of compression by the tool followed by deformation along the shear plane [37]. However, in this case, segment separation takes place. Such chip segmentation at low speed indicates that the short chipping of leaded brass is not controlled by temperature in the primary deformation zone and precludes the melting of lead on the interface between the chip segments. A more likely concept for the chip formation is presented by Doyle [17], who investigated the plastic instabilities that can arise in heterogeneous materials. Lead has nearly a 10 times lower Young’s modulus and 20 times lower strength than the matrix brass alloy [38].
From the high-speed footage for CuZn38Pb3, it appears that the tool practically loses contact with the workpiece on the rake side when a chip segment separates from the workpiece. By measuring the dynamics of the cutting forces with a high sampling frequency (200 kHz) for a cutting dataset of vc = 150 m/min, f = 0.4 mm/rev, and ap = 2 mm, the differences between the machining of the two brass alloys become apparent, see Fig. 10. While the peak force for the lead-free material is slightly higher than for the CuZn38Pb3 alloy, the variation in the cutting forces is dramatically different between the alloys. For the leaded brass, the cutting force is almost sinusoidal and is Fc = 560 ± 558 N, whereas for the lead-free brass it is Fc = 1016 ± 29 N, and only small peaks related to the chip formation process are observed. The force pattern for the lead-free CuZn42 alloy follows the traditional sequence of a segmented (saw-tooth) chip formation cycle [39]. Here the material ahead of the cutting edge is compressed, resulting in increased forces, followed by shear initiation, shear localization, adiabatic heating, and segment formation, all resulting in a force reduction. Apparently, in the case of leaded CuZn38Pb3 alloy, the end result of cyclically increased and reduced forces is similar, but the causes might not be identical. The analysis of the force and high-speed footage reveals that the compression stage is the same, but upon shear initiation a plastic instability promoted by the lead rapidly localizes the shear process to the degree that a segment detaches from the material bulk. Each individual segment is then ejected by the oncoming tool and stored elastic strain, thus producing discontinuous chips.
While the peak value of the cutting force component Fc is similar for both alloys, the peak value of the feed force Ff is almost halved for the leaded CuZn38Pb3 alloy. It seems that a compression cycle, which is similar for both alloys, contributes to the similarity in the cutting force component Fc. Feed force Ff, on the other hand, is to a higher degree controlled by the tribology of the tool–chip contact on the rake. The tool temperature at high cutting speeds is evidently lower for the leaded brass (Fig. 7). Several explanations have been offered for this lower temperature, including the low friction on the rake. Some authors [22; 25] suggest that melting lead has a significant lubricating effect that enhances the machinability. Thermal softening and melting of lead, and thus its lubricating efficiency, are temperature dependent. Varying the cutting speed is a controllable process parameter that has a direct effect on the process temperature.
A series of machining experiments within the speed range of vc = 2 m/min to 100 m/min were conducted to evaluate the effect of lead on cutting forces and the friction coefficient. All tests were performed using a constant depth of cut, ap = 2 mm. The results can be seen in Fig. 11b–d show that the cutting forces for both brass alloys are largely independent of the cutting speed. However, the force level is higher for CuZn42 than for CuZn38ZnPb3 under all conditions. It is worth noting that machining CuZn38Pb3 alloy resulted in a consistent artifact in force behavior. This artifact consists of a strong variation in forces at particular speed ranges, which depended on the feed used. For feed f = 0.1 mm/rev, this variation was observed at vc = 30 m/min, it increased to vc = 60 m/min at f = 0.2 mm/rev, and further shifted to vc = 80 m/min at f = 0.3 mm/rev. The most likely source of the variation is vibration or chatter. To eliminate this effect, the experiments were repeated with different setups, and even on another lathe with another force measurement system, yet all yielded similar results. An alternative explanation is that the formation of a built-up edge (BUE) might influence the forces. However, analysis of the machined surfaces and the cutting tools revealed an absence of BUE. The consistency of the observed phenomenon calls for further investigation.
Estimation of the coefficient of friction on the tool-chip interface requires separating the force components acting on the rake and the flank side of the tool. It is sometimes, often in the cases of rough machining, considered that forces on the flank can be neglected and the coefficient of friction is simply a ration of feed to cutting forces: µ = Ff / Fc [40]. In practice, the workpiece material flowing against the tool is separated into two separate flows by the cutting edge. The point where the material flow is separated is located in the vicinity of the edge radius and is called the stagnation point (see Fig. 12). The material below stagnation point moves toward the tool’s clearance side, is ploughed under the edge and is not removed from the workpiece. Ploughing forces involved in this action do not contribute to the normal and frictional forces on the tool rake side. Similarly, the rubbing action of the tool clearance against the machined workpiece material adds to the clearance forces.
These forces are easily found by extrapolating the feed or chip thickness values for h1 = 0 [42]. Therefore, the total force can be described by, where C2, D2, and E2 are the forces acting on the clearance side.
\({F}_{c}={{C}_{2}+{C}_{1}\bullet {h}_{1}=T}_{r}+{T}_{cl}\)
\({F}_{f}={{D}_{2}+{D}_{1}\bullet {h}_{1}=A}_{cl}+{A}_{r}\)
\({F}_{p}={{E}_{2}+{E}_{1}\bullet {h}_{1}=R}_{r}+{R}_{cl}\)Eq. 3
Figure 11a shows the linearization of the cutting forces necessary to determine C2, D2, E2, and their values. From Fig. 12 and Eq. 4 it is possible to derive and calculate an estimate of the coefficient of friction based on the measured cutting forces, by slightly alter the model presented by Schultheiss [41] to accommodate the contribution to friction from Fp, Eq. 4. The apparent coefficient of friction will be an indicator of how much lead contributes to reducing friction and thereby the cutting forces.
\({\mu }_{r}=\frac{\sqrt{{A}_{r}^{2}+{R}_{r}^{2}}}{{T}_{r}}=\frac{\sqrt{{\left({F}_{f}-{D}_{2}\right)}^{2}+{\left({F}_{p}-{E}_{2}\right)}^{2}}}{{F}_{c}-{C}_{2}}\) Eq. 4
The calculation results for the apparent coefficient of friction are shown in Fig. 13. For the lead-free alloy CuZn42 (on the right), the apparent coefficient of friction µr is between 0.39 and 0.57 and is relatively constant over the considered range of cutting speed and the feed. The same cannot be said about machining CuZn38Pb3. While the change in µr over the speed range can be related to the force artifact mentioned above, there is also a strong dependency on the feed. At feed f = 0.1 mm/rev, the apparent coefficient of friction is µr = 0.45, and it reduces to µr = 0.24 at feed f = 0.3 mm/rev. On the one hand, an expected increase in process temperature with an increased feed should facilitate softening of Pb, increased lubrication efficiency, and thus a lower friction coefficient. On the other hand, an increase in the cutting speed should also increase the process temperature and thus produce similar effects, yet the data shows that increasing speed does not reduce µr.
The surface chemistry within the contact region was analyzed in an attempt to detect the impact of lead on the friction and tribology, because the mechanistic explanation appears to be contradictory. An uncoated cemented carbide tool was used to machine CuZn38Pb3 for three minutes at vc = 200 m/min, f = 0.2 mm, and ap = 1.5 mm. SEM and XEDS data (see Fig. 14) do show the presence of lead on the rake face, but not in the contact zone between the tool and the chip. The Lα energy level for lead is 10.55 keV, and thus XEDS maps were made using an acceleration voltage of 20 kV for accuracy of quantification. Using such high voltage for XEDS analysis means that the excitation volume also includes the bulk material and probably underestimates the data for thin surface layers such as the lead monolayer reported by Stoddart et al. [20].
To further study the chemical composition of the tool–chip interface, time-of-flight secondary ion mass spectrometry (ToF-SIMS) was used (see Fig. 15) because the method is sensitive to the top-most layers of a tool sample. The same tool was examined using both XEDS and ToF-SIMS. As with the XEDS, high levels of lead were found outside the theoretical contact area, but ToF-SIMS reveals that the copper from the workpiece material covers nearly the entire contact zone. Lead is detectable in the part of the contact area furthest from the edge line. Using the analogy of the common sticking and sliding regions found on cutting tools [43; 44], lead is not present in the sticking region, but is slightly present in the sliding region. As shown in the high-speed footage, Fig. 8 and Fig. 9, the tool–chip contact length for CuZn38Pb3 is very short and approximately equal to the theoretical chip cross-section profile. The footage also reveals that there is no sliding region in its conventional form. Therefore, it can be assumed that the lead within the contact zone stems from rubbing of chips after separation from the workpiece, and the lead outside the contact zone is from the particles ejected from the chips. Since the area affected by lead is small (approximately 10%) compared to the entire tool–chip contact zone, the effect of Pb on the frictional forces and its likely lubrication effect can be considered minor.
As there is no evidence supporting either melting of lead or lead’s lubricating effects on the tool–chip interface, attention turned to the workpiece to find the mechanisms behind the high machinability of the alloy. Quick-stop samples were prepared as shown in Fig. 5 in order to follow the path of the lead through the cutting zone. The quick-stop samples for both alloys were generated in longitudinal turning at vc = 70 m/min, f = 0.4 mm/rev, and ap = 2 mm and are shown in Fig. 16 before cross-sectioning. As can be seen, the CuZn38Pb3 quick-stop sample has chip segments formed but partly attached, while the other segment is in the compression stage. For the lead-free case, the attached chip is continuous, with clear saw-tooth morphology.
After extraction and polishing of the quick-stop samples, the differences between the cutting zones of the two alloys can be seen in Fig. 17 and Fig. 18. The chip root of the CuZn42 alloy is characterized by segmentation, with larger segments separated by shear bands alternating with smaller peaks. The shear bands are very thin (5–15 µm), some even underdeveloped. The secondary εII and tertiary εIII deformation zones, on the other hand, are very substantial: εII ≈ 100 µm and εIII ≈ 150 µm. This is indicative of higher contact loads on the tool–chip and tool–workpiece interfaces and is consistent with the higher forces and friction observed for this alloy (see Fig. 11 and Fig. 13).
The chip root of CuZn38Pb3 alloy exhibits a different morphology with a distinct and wide shear band and the expected segment separation. Unexpectedly, secondary and tertiary deformation zones are large: εII ≈ 100 µm and εIII ≈ 50 µm as can be seen by elongation of brass and lead grains. The primary deformation zone is also very wide (εI ≈ 100–150 µm), where the lead grains are strongly textured and parallel to the shear plane, as compared to the unaffected microstructure shown in Fig. 1. A more detailed image of the primary shear zone can be seen in Fig. 19, in combination with XEDS maps for relevant elements. The elongated and flake-like lead grains are clearly visible as bright spots in the backscatter image. Additionally, in the XEDS map, an agglomeration of lead on both sides of the shear plane is clearly distinguished. As previously discussed, high excitation voltage during XEDS analysis will read the signal from sub-surface layers not visible in the backscatter mode. The agglomeration of lead near the shear plane may stem from a high concentration of lead on the shear plane itself.
While the sides of the shear plane are not visible in cross-section as seen in Fig. 19, this information can be found on the chips. Figure 20 shows that lead is indeed elongated and smeared all over the chip side of the shear plane.
The same traces of smeared lead found on the shear plane can be found on the machined surface, see Fig. 21. The layer of lead found on the machined surface is, however, very thin and is mostly visible when using a relatively low acceleration voltage in the SEM. At high acceleration voltage of 20 kV, only the elongated lead grains are distinguished, while at low voltage of 5 kV lead covers nearly the entire surface. Smearing of lead on machined surfaces is documented in the literature [24; 45]. From a product point of view, smearing and redistribution of lead across the entire machined surface is more problematic than the presence of localized inclusions. It is known that the lead may leach into water and create adverse health and environmental effects [2–4]. The extensive surface coverage by the lead (Fig. 21) may create a spike in lead leaching and exposure during the initial stage of product use and cause more harm than estimated for the bulk case.
Our findings suggest that neither melting of lead nor lubrication and thus low friction across the tool–chip interface occur when machining leaded brass. Instead, the observations, as summarized graphically in Fig. 22, show that the low cutting forces are largely due to deformation of the segregated lead inclusions in the primary deformation zone. The highly deformed lead grains stretch along the shear plane, change shape from globular to flake-like inclusions, and act as crack initiation points in the brass matrix, so enforcing the process of discontinuous chip formation. The presented findings supports Doyles [17] conclusion about mechanical instabilities causing discontinuous chip formation. This type of chip formation restricts the contact length between tool and chip (approximately 0.6–1.0∙h1), resulting in low frictional forces and low tool temperature. Thermal measurements confirm that machining CuZn38Pb3 results in an approximately 180°C lower tool temperature than machining CuZn42 (Fig. 7). Such low temperatures explain the extremely long tool life of cutting tools when machining brass with high lead content.
It is common to find the term “internal lubrication” being used to describe the beneficial effects of lead on the machining process [17; 19; 22; 25]. This term is often confusing as induced lubrication implies reduced friction. Our findings reveal that the lower apparent coefficient of friction for leaded brass is more dependent on chip formation, low tool–chip contact length, and subsequently lower feed force Ff, than on the lubricating effect of lead. In particular, atomic mono-layer resolution ToF-SIMS shows that lead is sparse on the tool surface in the contact zone, see Fig. 15. For these reasons, the term “internal lubrication” may be misleading.