1 Introduction

Brass is a commonly used engineering material for components where high corrosion resistance is desired, such as couplings and pipe fittings in water handling systems. Traditionally, brass has been alloyed with lead (Pb) to improve machinability and castability of the material [1, 2]. Due to its detrimental effects on living organisms and the environment, Pb is restricted in applications with a risk of Pb leaching, e.g. drinking water applications [3, 4]. Since the use of Pb-containing alloys will be further restricted in the near future [5], Pb-free alloys are gaining popularity in the manufacturing industry.

Machining Pb-free brass is associated with higher cutting forces, severe tool wear and less chip control as compared to machining leaded brass [6, 7, 8, 9]. Since tool wear is a known contributing factor to increased sub-surface deformation [10], and more aggressive tool wear is observed when machining Pb-free brass alloys, a higher degree of sub-surface deformation is expected in Pb-free brass alloys. It is therefore of interest to investigate the influence of cutting data and tool geometry variation with wear on sub-surface deformation and its link to stress corrosion cracking (SCC). Sub-surface deformation is part of surface integrity, as introduced by Field and Kahles [11, 12], and can have large influence on a components service life.

Stress corrosion cracking is a severe problem that can lead to components underperforming or even catastrophic failure. For brass products in drinking water systems, SCC causes leakage of water and creates risks of extensive damages to surrounding structures [13]. SCC involves crack formation due to simultaneous effects of static tensile stresses and corrosion. The factors behind SCC can be divided into environmental factors, metallurgical factors and mechanical stress and strain [14]. Stress corrosion cracking is primarily associated with aqueous ammonia when it comes to brass alloys. Cracking of copper components after exposure to ammonia was reported already in the beginning of the twentieth century [15]. However, SCC can be caused by a wide variety of species including sulphates, nitrates, nitrites, chloride containing species and pure water [16]. Alloy composition, especially zinc concentration, and material heat treatment practices have shown to impact the material susceptibility to SCC, [16]. The tensile stress may originate from external load or as residual stresses induced by cold working, machining, welding or heat treatment [16]. Initiation sites are often either pits formed by pitting, or deposit corrosion or as defects caused by forming and machining [16].

In machining, the removal of material is associated with three separate and distinct deformation zones around the cutting tool [17]. While primary and secondary deformation zones are associated with the chip formation and tool-chip interface, the tertiary deformation zone is formed by the contact between the machined surface and the tool clearance side. This tertiary deformation zone, directly related to surface integrity, remains in the finished machined surface and defines its performance in an application. Assessment of a relationship between machining-induced surface integrity and product performance is a routine procedure for components operating in critical applications, and is well documented for Ni- and Ti-based alloys [18, 19], and stainless steels [20, 21]. Literature about surface integrity of machined copper alloys [19] and brass alloys is available [22, 23], and knowledge regarding SCC in brass also exist [24, 25, 26]. However, no studies on the connection between machining induced surface integrity and SCC could be found in the case of brass. Therefore, this paper investigates the surface integrity response to different machining conditions and tool wear conditions for two common Pb-free alloys and correlates the sub-surface deformation to stress corrosion cracking.

2 Experimental setup

Two brass alloys that are approved to be used in drinking water applications, CuZn38As and CuZn21Si3P, are chosen for the study, and their chemical composition is shown in Table 1. The materials have different microstructures; CuZn38As is a single-phase material that mainly consists of the α-phase, with only trace amounts of the β-phase from the binary Cu–Zn system. The microstructure for CuZn21Si3P is more complex; it contains the same α-phase as CuZn38As, as well as a so-called κ’-phase. The κ’-phase is similar to the κ-phase found in the binary Cu–Si phase system with regard to atomic arrangement but has slightly different lattice parameters, and it is stable at room temperature. These aspects of microstructure for the materials were revealed using Electron Backscatter Diffraction (EBSD) and X-Ray Diffraction (XRD) as seen in Fig. 1.

Table 1 Chemical composition of the used materials (wt. %)
Full size table
Fig. 1
figure 1

a EBSD image showing phase distribution and crystal orientation for CuZn21Si3P. b EBSD image showing phase distribution and crystal orientation for CuZn38As. c XRD pattern for CuZn21Si3

Full size image

The cutting conditions chosen for generating the machined samples were similar to conditions at brass component manufacturing facilities. The samples were produced in a longitudinal turning operation. The depth of cut, ap, and cutting speed, vc, were kept constant at 2 mm and 200 m/min, respectively, while the feed rate, f, was varied in three levels: 0.05, 0.15 and 0.4 mm/rev. Since tool wear is a major contributor to sub-surface deformation, a worn tool with flank wear, VB = 0.37 mm, as well as an unworn tool with small edge radius and a large positive rake angle (γ = 22.5°), was used to produce high and low levels of deformation, see Fig. 2. The tool holder DCLNL3225P12-M was used for both the sharp insert, which was of ISO CCGT120408F-AL, KX geometry and the worn tool, which was originally of ISO CNMG120408-MF1 geometry. Both inserts were of uncoated cemented carbide. The profiles of the edge lines, shown in Fig. 2, were measured using a 3D-optical microscope, Alicona Infinite Focus G4, and each edge profile in Fig. 2 represents an average of 500 measurements on the tools in the study. As can be seen in the cutting edge profiles in Fig. 2, both new and worn tool have similar rake angle with the flank wear being the main difference.

Fig. 2
figure 2

SEM image of the unworn (a) and worn (b) tools together with the cutting edge profile of the unworn (c) and worn (d) tool. For the worn tool, the blue line represents the original edge line profile, and the red line shows the actual worn edge profile

Full size image

Nano-indentation and Electron Backscatter Diffraction (EBSD) were used to quantify the depth of the sub-surface deformation and the degree of the deformation. In nano-indentation, a diamond Berkovich tip is indented into the material with a small load, in this case 10 mN, and the hardness is evaluated based on the load–displacement curve [27]. The small size of the imprints, approximately 0.2 – 0.5 µm deep, enables high spatial resolution. Prior to indentation the samples were sectioned, ground and fine-polished. The NanoTest Vantage system by MicroMaterials was used for the hardness measurement. Nano-indentation of the material bulk was done prior to the analysis of sub-surface to understand local variations related to multi-phase brass composition. As can be seen in Fig. 3, the hardness distribution was different for the used materials. Alloy CuZn21Si3P contains two different phases and shows a wider, non-symmetric hardness distribution as compared to alloy CuZn38As that only contains one phase. The mean hardness value for CuZn21Si3P was 2.6 GPa, and for CuZn38As 1.5 GPa.

Fig. 3
figure 3

Hardness distribution in the bulk of the materials (a) and (b) measured in a 10 × 10 matrix with 30-µm spacing between indents and colormaps showing the hardness distribution in the samples (c) and (d). Note the difference in hardness scale

Full size image

To characterize the sub-surface deformation using nano-indentation, 15 indentations were made in depth below the surface, all spaced 15 µm apart. Five indentations were performed for each depth level, spaced 50 µm apart. Thus, a matrix of 5 by 15 indentations was done for each sample as seen in Fig. 4.

Fig. 4
figure 4

Example of hardness measurement imprints captured using a light optical microscopy in polarized light and mapping of measured hardness for a sample machined using a worn insert at f = 0.4 mm/rev in alloy CuZn38As

Full size image

For visualization of the deformation zone, further ion beam polishing of the samples followed by Ion Channeling Contrast Imaging (ICCI) [28] was performed in a FEI Nova 600 Focused Ion Beam/Scanning Electron Microscope (FIB/SEM). ICCI is an imaging technique that provides excellent grain contrast in brass. A JEOL 6490LV SEM with an EBSD detector by Oxford Instrument NordlysMax3 was further used to quantify the microstructure variation in the superficial layer. To evaluate the susceptibility to SCC, accelerated corrosion testing in accordance with SIS 117102 [29] was performed. This test method was chosen because it was developed to evaluate the effect of processing-induced internal stresses in materials without the influence of externally applied stress. For brass and other copper alloys, the susceptibility to SCC is commonly evaluated in accordance with standard ISO 6957 [30], in which an external applied stress is used. That standard is thus not suitable for evaluation of the impact of machining-induced surface integrity only. Before exposure, test specimens were degreased in acetone and pickled in 5 vol. % sulfuric acid solution. After pickling specimens were rinsed in water and dried in forced hot air convection. The test solution was prepared by adding concentrated ammonia to deionized water. Copper sulphate was dissolved in the ammonia solution followed by dilution with deionized water to final test volume to the concentration 1 mol NH4OH and 0.05 mol CuSO4 per liter of test solution. At the start of exposure, test specimens were totally immersed in the test solution at room temperature. The different samples were exposed in separate exicators. Test specimens were taken out from the test solutions for inspection after 30 min to enable evaluation of crack initiation.

3 Results and discussion

Longitudinal turning is an anisotropic process where observed characteristics such as topography and sub-surface deformation strongly depend on the viewing direction. When inspecting a cross section of the workpiece seen along the axial direction, the surface is curved according to the workpiece diameter, but no ridges are seen, as in Fig. 5a. When inspecting a cross section of a machined workpiece along the tangential direction, the ridges and valleys created by the tool’s nose radius are visible. The spacing between the ridges is approximately equal to the feed rate, Fig. 5b. Although the surface topography appears different depending on viewing direction, the depth of the severely deformed material (εIII,severe closest to the machined surface) is independent from viewing direction.

Fig. 5
figure 5

Ion Channeling Contrast Images (ICCI) after FIB polishing of the deformation zone with the severely deformed part marked by red arrows, for a sample cut in the tangential direction, (a), and a sample cut in the axial direction, (b). The samples of CuZn38As alloy were machined using a sharp insert with f = 0.15 mm/rev

Full size image

Formation of sub-surface deformation—i.e. its depth and degree of work-hardening—is dependent on the thermo-mechanical state of the generated surface. For a given material, it is controlled by cutting conditions, tool geometry and tool wear. It is known that the higher mechanical load caused by a higher feed produces a deeper and more work-hardened layer [31]. Thermal softening under high cutting speeds can facilitate localization of deformation to near-surface layer. Still, tool wear is among the most defining factors influencing sub-surface deformation and sub-surface damage [32]. This rather predicable and linear behavior becomes much more complicated when machining with cutting tools having a nose radius, rε [33]. Sub-surface deformation that is generated during the current cut affects not only machined surface but also the temporary surface, Fig. 6. During the sequential cut, when the tool has moved one feed forward and the workpiece makes one revolution, the cutting tool might cut through already work-hardened layer; thus, multiple deformation might take place. The magnitude of such multiple deformations is controlled by three parameters: tool nose radius—rε, feed—f and depth of sub-surface deformation from a single cut—εIII.

Fig. 6
figure 6

Adopted from Bushlya et al. [33]

Schematic of the interaction between sub-surface deformation from single and sequential cuts in longitudinal turning with tools having a nose radius.

Full size image

3.1 Quantification of machining induced sub-surface deformation

Involvement of so many parameters makes the surface integrity response to the machining operation rather difficult to predict analytically. While it is expected that machining with low feed should generate lower plastic deformation, in the case of a worn tool a rather comparable sub-surface deformation εIII is found for f = 0.05 mm/rev and f = 0.15 mm/rev, see Fig. 7. For both machining conditions, the depth of work-hardening is approximately 220 µm. Still, for low feed of f = 0.05 mm/rev the surface hardness is higher because the cutting tool passed over the surface several times. When the feed is very large and is bigger than the original εIII, the tool will not re-engage with previously deformed layer and will generate lesser overall deformation, as seen in Fig. 8 for f = 0.4 mm/rev.

Fig. 7
figure 7

Hardness distribution for two samples machined with a worn insert and different feed. The red lines show max and min hardness, and the black line shows the average hardness for the unaffected bulk material

Full size image
Fig. 8
figure 8

The quotient between measured sub-surface hardness and the average hardness of unaffected bulk material for all used machining conditions

Full size image

The influence of feed rate was not significant in this study; however, the wear condition of the tools was important. An increase of both depth of εIII and the severity of work-hardening were seen for the worn tool. The two alloys showed different response to the tool conditions. In Fig. 8, a quotient between the local sub-surface hardness, H, to the unaffected bulk hardness, Hbulk, is presented for all cases. For unworn tools, machining of alloy CuZn38As generated nearly twice as deep plastic deformation zone and approx. 20% higher degree of work-hardening compared to the silicon brass CuZn21Si3P. It is worth noting that the depth of work-hardened layer is, as expected, small for low feed condition (f = 0.05 mm/rev), yet the degree of work-hardening (H/Hbulk) is nearly equivalent to the high feed (f = 0.4 mm/rev) case. As demonstrated in Fig. 6, the case of low feed is accompanied by multiple consecutive cuts over the same sub-surface region, and such multiple deformation then results in a severely deformed but shallow εIII layer. Machining with a worn tool gave larger differences between the two alloys. The depth was somewhat comparable for the alloys, yet the work-hardening for CuZn38As was more than 40% higher. The severity of deformation, H/Hbulk, for CuZn38As was close to 2, as compared to CuZn21Si3P, where H/Hbulk was well below 1.5. In this case, the worn tools induce comparable depth of εIII irrespective of feed, since deformation caused by tool geometry dominates over the deformation caused by feed. However, due to multiple deformations in the cases of low feed a slightly more severe work-hardened layer for f = 0.05 mm/rev and f = 0.15 mm/rev is seen in Fig. 8d.

Good correlation was seen between the measured microhardness profiles and the crystallographic misorientation, measured using EBSD, see Fig. 9. Kernel average misorientation analysis was used to evaluate the deformation depth with a step size of 0.5 µm. The analysis of deformation zone in alloy CuZn21Si3P was less clear, which may be related to the dual phases and the relatively unknown crystallographic arrangement of the κ’ phase. It was difficult to obtain Kikuchi diffraction patterns in the severely deformed layer closest to the machined surface, due to excessive deformation which resulted in unresolved black areas. Further investigation of εIIIsevere with higher magnification and a step size of 0.05 µm revealed that the black area closest to the machined surface had a high fraction of submicron and nano-sized grains, Fig. 10. Formation of such microstructural features in the near-surface region is typically attributed to either severe plastic deformation (SPD) under conditions of high strain and low homologous temperature, Thom, or dynamic recrystallization (DRX) under high strain and high homologous temperature. Dynamic recrystallization typically requires T/Thom ≥ 0.5 [34], which for brass alloys is equivalent to T ≥ 320 °C. The temperature when machining these two brass alloys was measured with Infra Red thermal imaging in orthogonal turning following the methodology described elsewhere [35]. Figure 11 shows that the cutting tool temperature is slightly higher than this threshold for sharp tools, and likely even higher for worn tools. Therefore, the submicron grains are likely the result of DRX, while presence of microshear bands is the result of SPD [34], which is related to a gradient temperature reduction deeper into the sub-surface layer, Fig. 10. These microstructural features and transformations are representative of the so-called white layer which is well studied for machining Ni-alloys and Ti-alloys [18].

Fig. 9
figure 9

Kernel average disorientation measured using EBSD overlayed with the hardness profile for the sample, measured with micro indentation. a CuZn21Si3P machined with a sharp insert and f = 0.05 mm/rev, b CuZn38As machined with a sharp insert and f = 0.05 mm/rev, c CuZn21Si3P machined with a worn insert and f = 0.4 mm/rev, d CuZn38As machined with a worn insert and f = 0.4 mm/rev

Full size image
Fig. 10
figure 10

EBSD at higher magnification of the severely deformed layer εIIIsevere for CuZn38As machined with a worn tool and f = 0.4 mm/rev and b sharp tool and f = 0.05 mm/rev

Full size image
Fig. 11
figure 11

Tool temperature in orthogonal turning for f = 0.2 mm/rev. ap = 2 mm and vc = 200 m/min. Note the difference in scale between the images

Full size image

3.2 Corrosion

Figure 12 shows optical images of test specimens after corrosion experiments. Compared to the as-machined samples both brass alloys had a more dull appearance after testing. Spots and areas with darker coloration were visible. This appearance was expected since the exposure time was relatively short compared to the maximum exposure time, 48 h, referred to in guidelines to standard SIS 117102 [29]. For alloy CuZn21Si3P, a fraction of the surface had the appearance of corroded striations perpendicular to the cutting speed direction, see Fig. 12. The fraction of corroded surface was higher for a sample machined with a worn tool. For alloy CuZn38As, the corrosion striations were not observed, and the surface was affected uniformly. However, the cracks propagating from the surface into the bulk material were present on the CuZn38As sample machined with a worn insert, and also in this case the cracks propagate perpendicularly to the cutting speed direction. The cracks were spaced rather uniformly, approximately 30–50 µm apart, see Fig. 12. The difference in cracking behavior of the alloys cannot be attributed to a difference in the surface topography. Surface roughness before and after exposure was comparable for equivalent feeds and tool wear conditions, as shown by the surface profiles extracted from 3D images in Fig. 12.

Fig. 12
figure 12

Optical images of test specimens before (blue) and after 30-min exposure (red) with respective surface profile, measured as indicated by the white arrow. a CuZn21Si3P machined with a sharp insert and f = 0.05 mm/rev, b CuZn38As machined with a sharp insert and f = 0.05 mm/rev, c CuZn21Si3P machined with a worn insert and f = 0.4 mm/rev, d CuZn38As machined with a worn insert and f = 0.4 mm/rev

Full size image

Cross-sectioning of the corroded samples was performed in the cutting speed direction, which was followed by mechanical grinding, polishing and ion beam polishing. ICCI was used to analyze the effect of sub-surface deformation on the stress cracking corrosion, because EBSD was not able to resolve the severely deformed near-surface region (see Fig. 10) where crack initiation and propagation are expected. Despite the very different degree of work-hardening (see Fig. 8) for CuZn21Si3P samples machined with a sharp and worn tools, neither of the samples had cracks, see Fig. 13. The cross section of corroded CuZn38As machined with a sharp tool revealed plastic deformation identical to as-machined samples, but without distinct localized corrosion defects and cracks. On the contrary, CuZn38As sample machined with a worn tool had a series of cracks perpendicular to the surface. Figure 14 shows a detailed view of a crack, which was approximately 45 µm long. The crack originated on the surface and propagated until it met the grain boundary where it was deflected.

Fig. 13
figure 13

a CuZn21Si3P machined with a sharp insert and f = 0.05 mm/rev, b CuZn38As machined with a sharp insert and f = 0.05 mm/rev, c CuZn21Si3P machined with a worn insert and f = 0.05 mm/rev, d CuZn38As machined with a worn insert and f = 0.05 mm/rev showing a crack from the corrosion test

Full size image
Fig. 14
figure 14

Detailed view of the crack formation in CuZn38As with a high degree of deformation after machining. The crack is approximately 45 µm deep, and several similar cracks were found in the sample

Full size image

The observed behavior of the two brass alloys can be attributed to several different factors. Higher zinc content in alloys is known to increase the susceptibility to SCC [16]. CuZn21Si3P contains 21 wt.% Zn, while the Zn content in CuZn38As is 37 wt. %. Thus, the lower zinc content in CuZn21Si3P is a possible factor behind the absence of cracking in these samples. Still, it is likely that more prolonged exposure of silicon brass will cause cracking. As seen from nano-indentation (Fig. 8), the degree of work-hardening for this alloy was lower compared to CuZn38As; thus, more favorable stress state can be expected. At the same time, silicon brass has nearly 3 times lower thermal conductivity [8], and therefore higher tensile residual stresses are expected [31]. Hence, complementing the surface integrity analysis with analysis of residual stresses should be able to further clarify the observed difference in the accelerated corrosion test.

In this study one sample of each machining condition was exposed to the SIS117102 corrosion test, which makes it difficult to draw definitive and wide-reaching conclusions on the materials resistance to SCC. Additionally, experimenting with machined CuZn21Si3P in prolonged corrosion test could have enabled identification of the defects that initiate stress corrosion cracks. However, our investigation clearly shows that the sub-surface deformation is both deeper and more severe for CuZn38As, compared to CuZn21Si3P, under identical machining conditions. It is also shown that cracks were only identifiable in CuZn38As machined with a worn insert and f = 0.4 mm/rev, i.e. the case with the most severe sub-surface deformation. It should be noted that the testing environment according to SIS117102 is dissimilar to real working conditions in order to speed up the reactions. A more accurate impact of surface deformation and residual stresses on SCC is best achieved by extended field exposures. Although very time consuming, such life-time testing can be achieved by comparing test components manufactured with sharp and worn tools installed in active water handling systems.

4 Conclusions

Surface integrity and sub-surface deformation have a significant impact on component performance during the use phase of its life cycle. This paper investigates the sub-surface deformation generated when machining two lead-free brass alloys at three levels of feed rate and under two cutting tool conditions, including a sharp unworn tool and a tool with substantial flank wear.

Analysis of sub-surface deformation revealed only a minor effect of different feed rates but a large influence of the degree of tool wear and cutting edge geometry. Under identical machining conditions, the degree of work-hardening in the CuZn38As alloy was far more severe compared to the CuZn21Si3P alloy but the deformation depth was similar for both alloys. For an unworn tool, the difference in the depth of sub-surface deformation between the alloys was approx. 20%, while for the worn tool the work-hardening variation was as high as 40%.

Stress corrosion cracking of both lead-free brass alloys was assessed by accelerated testing performed according to SIS 117102 standard, in the corrosive environment with 1 mol NH4OH and 0.05 mol CuSO4 per liter concentration. After corrosion testing, cracks were found in the most severely deformed sample of CuZn38As machined with a worn insert and f = 0.4 mm/rev. This was the case of the most severe case of sub-surface deformation, thus pointing to a link between sub-surface deformation and SCC and highlighting the need for close process control in manufacturing of components to be used in corrosive environments. Higher corrosion resistance of CuZn21Si3P can also be related to a more favorable alloy composition with lower Zn content and dezincification corrosion resistance.