Machining Duplex Stainless Steel: Comparative Study Regarding End Mill Coated Tools

Ronny M. Gouveia 1, F. J. G. Silva 1,*, Pedro Reis 1 and A. P. M. Baptista 2 1 ISEP (School of Engineering, Polytechnic of Porto), Porto 4200-072, Portugal; 1060593@isep.ipp.pt (R.M.G.); pedro21reis@gmail.com (P.R.) 2 INEGI (Instituto de Ciência e Inovação em Engenharia Mecânica e Engenharia Industrial), Porto 4200-465, Portugal; amb@fe.up.pt * Correspondence: fgs@isep.ipp.pt; Tel.: +351-228-340-500


Introduction
Machining remains one of the most employed processes in the world's metalworking context. The aeronautics, naval and automobile industries are, probably, the economic sectors that most demand this kind of process through the engineering of parts included directly and indirectly in aircrafts, ships, trucks, buses and cars. Machining processes are necessary when surface quality is one of the main requirements demanded by customers or designers. However, competitiveness is a key factor in such industries, requiring accurate and extremely efficient equipment. Indeed, the available equipment on the market has continued to grow, becoming increasingly sophisticated and allowing an improved accuracy [1], answering to the need of the market for complex organic shapes and high surface quality.
Researchers have contributed significantly to this development, devoting their efforts to explaining many of the machining related phenomena, anticipating market needs and studying the best ways to increase production and improve products through tooling improvements, more accurate machining parameters, optimized machining trajectories, and so on.
Coatings have proven to be an attractive way to extend tool life by increasing surface hardness, acting in addition to other crucial parameters concerning tool work such as: decreasing the friction between tool and part; creating a thermal barrier between the surface and the hard metal substrate and allowing a better surface temperature distribution by dissipating the heat generated during tool contact with chips and machined part [16,17]. Thus, tool coating requirements are large and of high importance. Tool manufacturers started using the conventional TiN, however, this coating was not sufficient to solve all the previously mentioned problems, hence, multilayered coatings started to be developed in order to improve tool performance. Nowadays, tools recently introduced on the market present three or more layers, each one with specific functions, in order to satisfy all the requirements.

Materials and Methods
The raw material chosen to be machined for this study was a CD4MCuN duplex stainless steel (material specification: ASTM A890, Werkstoff-Nr. 1.4517 or EN 10283, ARSOPI, Vale de Cambra, Portugal). It is composed of a dual phase microstructure of both austenite and ferrite in similar amounts. The chemical composition given by the supplier for this batch of steel was (wt %) C 0.03%, Si 0.41%, Mn 1.48%, P 0.02%, Cr 25.41%, Ni 6.08%, Mo 2.91%, Cu 3.30%, Nb 0.01%, V 0.02%, W 0.04%, N 0.19%, Co 0.08% and Fe 60.02% [18]. It is sold in a heat treated state (quenched at 1135 °C and water cooled) and according to the supplier has a yield strength (0.2%) and an ultimate tensile strength of 489 MPa and 797 MPa, respectively (in accordance with ISO 6892-1 [19]), as well as hardness of 267 HB (ISO 6506-1 [20]). Among several different material properties, some of the most important are its high mechanical strength (approximately twice as much as the more common ASTM 304 and 316 stainless steels), good corrosion and pitting resistance, good toughness, good weldability and ease of fabrication. Typical applications are the construction of chemical equipment and tubbing, pressurized tanks and vessels, heat exchangers, cellulose and paper fabrication, sea water processing, etc. This duplex stainless steel is thought to be of interest due to the fact that numerous flanges and accessories, which require machining, are made from this material and information surrounding this topic is scarce. A round bar with a diameter of 60 and 300 mm in length was sufficient for all testing trials.
The machining operations were performed on a 3-axis CNC machine (brand/model: Haas VF2 vertical machining center, Hass Automation, Inc., Oxnard, CA, USA). A 3-jaw self-centering chuck (brand/model: Bison 3575, BISON-BIAL, S.A., Bielsk Podiaski, Poland) was mounted to the machine's work table (in accordance with standard DIN 6350 [21]), guaranteeing the exact same positioning of the round bar between each trial. The use of a hydraulic tool holder (brand/model: WTE DIN 69871-AD/B [22], WTE Präzisionstechnik GmbH, Kempten, Germany) was necessary as this type of system minimized unwanted vibrations during testing. Figure 1 illustrates several components utilized during testing. The comparative study is based on the wear analysis of four different milling tools advertised as being appropriate for milling duplex stainless steels. The milling tools used all have in common a cutting diameter of 4 mm, a 6 mm shank diameter and a coated hard metal substrate. Milling tools The comparative study is based on the wear analysis of four different milling tools advertised as being appropriate for milling duplex stainless steels. The milling tools used all have in common a cutting diameter of 4 mm, a 6 mm shank diameter and a coated hard metal substrate. Milling tools with a 4 mm shank diameter were initially tested, however, they proved to be too fragile, suffering premature breakage. Table 1 lists the characteristics of all four different selected end mills. For easier reading and referencing, tools will be referred to by their manufacturer's name, followed by the number of cutting flutes and composition of coating (e.g., Dormer Z2 AlCrN) instead of their complete designation.
Besides analyzing the influence of coatings on wear behavior, the number of cutting flutes was also changed to investigate its effect on tool longevity. Usually, the amount of cutting flutes influences the ease of chip removal, so it is common sense to use tools with less cutting flutes in slotting operations and tools with more flutes in side milling operations. When chip removal is difficult, a buildup edge may appear on the tool.
In order to evaluate the wear of each tool, a simple roughing machining strategy was created to ensure repeatability of test conditions. Based on the round format of the bar, a spiral path was created to optimize cutting length, as can be seen in Figure 2. The tool begins side milling on the outer side of the stainless steel bar making its way to the center. A ramp plunge movement is used in the initial portion of the machining stage to create a gradual tool approach to the material, avoiding potential collisions. with a 4 mm shank diameter were initially tested, however, they proved to be too fragile, suffering premature breakage. Table 1 lists the characteristics of all four different selected end mills. For easier reading and referencing, tools will be referred to by their manufacturer's name, followed by the number of cutting flutes and composition of coating (e.g., Dormer Z2 AlCrN) instead of their complete designation.
Besides analyzing the influence of coatings on wear behavior, the number of cutting flutes was also changed to investigate its effect on tool longevity. Usually, the amount of cutting flutes influences the ease of chip removal, so it is common sense to use tools with less cutting flutes in slotting operations and tools with more flutes in side milling operations. When chip removal is difficult, a buildup edge may appear on the tool.
In order to evaluate the wear of each tool, a simple roughing machining strategy was created to ensure repeatability of test conditions. Based on the round format of the bar, a spiral path was created to optimize cutting length, as can be seen in Figure 2. The tool begins side milling on the outer side of the stainless steel bar making its way to the center. A ramp plunge movement is used in the initial portion of the machining stage to create a gradual tool approach to the material, avoiding potential collisions. All machining parameters remain the same throughout testing with the exception of the feed speed. As the number of cutting flutes changes from tool to tool, feed speeds are adjusted in order to ensure that feed per tooth speeds are similar. This allows similar cutting conditions on each cutting tooth, promoting equal working and wearing conditions. Spindle speed was established based on tool manufacturers' recommendations and initial testing, ensuring a surface finish capable of being analyzed by the profilometer. Depth of cut and working engagement values were determined based on trial and error. Initial trials using 1 mm depth of cut were tested, however, this depth proved to be excessive for the combination of small tool diameter and high hardness steel, needing to be reduced to 0.5 mm. Recommended working engagement values are usually 60%-70% of tool diameter, so 3 mm was chosen for this parameter as smaller values may promote rougher surface finishing and decrease tool life span.
The surface wear of each tool was evaluated using a SEM microscope provided with an EDS (Energy-Dispersive X-ray Spectroscopy) system (EDAX X-ray micro-analysis). For this study, the equipment chosen was a FEI Quanta 400 FED SEM (FEI, Hillsboro, OR, USA). Several global and close up images of the worn edges were registered. Per tool, two global images are shown, one taken with secondary electron imaging and the other with retro diffused electron imaging. Whenever areas All machining parameters remain the same throughout testing with the exception of the feed speed. As the number of cutting flutes changes from tool to tool, feed speeds are adjusted in order to ensure that feed per tooth speeds are similar. This allows similar cutting conditions on each cutting tooth, promoting equal working and wearing conditions. Spindle speed was established based on tool manufacturers' recommendations and initial testing, ensuring a surface finish capable of being analyzed by the profilometer. Depth of cut and working engagement values were determined based on trial and error. Initial trials using 1 mm depth of cut were tested, however, this depth proved to be excessive for the combination of small tool diameter and high hardness steel, needing to be reduced to 0.5 mm. Recommended working engagement values are usually 60%-70% of tool diameter, so 3 mm was chosen for this parameter as smaller values may promote rougher surface finishing and decrease tool life span.
The surface wear of each tool was evaluated using a SEM microscope provided with an EDS (Energy-Dispersive X-ray Spectroscopy) system (EDAX X-ray micro-analysis). For this study, the equipment chosen was a FEI Quanta 400 FED SEM (FEI, Hillsboro, OR, USA). Several global and close up images of the worn edges were registered. Per tool, two global images are shown, one taken with secondary electron imaging and the other with retro diffused electron imaging. Whenever areas of discontinuities (promoted either by loss of coating or adhesion of external materials) were detected, an EDS analysis was made to determine the chemical composition of that unknown area. Close up images of each cutting flute were taken to evaluate and measure flank wear present on the cutting edges. These measurements are designated by the abbreviation VB. Table 2 sums up all the machining parameters used with each tool. Cutting fluid emulsion was used during all machining operations, being composed of a mix of 5% oil in water. The analysis of post-machined surface roughness was achieved using a surface profilometer (brand/model: Mahr Perthometer M2, MAHR, GmbH., Gottingen, Germany) provided with a diamond stylus tip with a 2 µm radius (ISO 4288:1996) [23]. Two different roughness readings were taken, one in a radial orientation and the other in a tangential orientation, utilizing a cutoff value of 0.8 mm. Arithmetic mean surface roughness (R a ), surface roughness depth (R z(DIN) ) and maximum height of the roughness profile (R max ) were registered for each machined surface after each experimental run. R a is an arithmetical mean average and may not truly represent the roughness of the surface. R z is the sum of the height of the largest peak plus the depth of the deepest valley, inside the measured length. R max is the height of the largest peak or the depth of the deepest valley encountered in the measured length. With radial readings it is possible to evaluate the surface roughness left by the tool feed marks (boundary lines between tool passes) while with tangential readings it is possible to verify the surface roughness in between these lines. For each tool trial, six measurements were taken (three in a radial direction and three in a tangential direction), allowing to determine a mean-value that better represents the overall roughness of the machined surface.

Results
Having concluded all experimental trials, the wear suffered by each tool was evaluated by measuring the surface roughness on each machined part and by a microscopically analyzing the cutting edges of each tool. Average roughness values were calculated to represent the overall level of roughness present on each surface. These results allow evaluating the specific performance of each tool.

Machining Strategy
The adopted machining strategy was based on the premises discussed previously. For every trial, each end milling tool will repeatedly machine a pre-determined quantity of metal (cycle). One cycle corresponds to a completion of a machining pass, with a depth of cut of 0.5 mm, starting on the outer side of the stainless steel bar and making its way to the center. Each tool will be evaluated after completing one run of eight machining cycles, which equals to a total machining distance of 7.5 m. Another set of tools will be tested during 16 cycles, totalizing a machining distance of 15 m.

Map of Experiments
In order to pursuit the main goals of this work, a map of experiments was drawn, which can be seen in Table 3, showing the different tools used in the comparative study as weel as the selected machining parameters.

Walter 4Z TiAlN-Roughness and SEM Results
The 8 cycles run trial for the Walter 4Z TiAlN end mill produced unexpected surface roughness results. When compared to the values obtained by the 16 cycles run, the shorter trial returned higher roughness values. There seems to be no significant explanation for these results (values can be seen in Table 4). The obtained radial R z roughness values for the 8 cycles trial are lower when compared to the 16 cycles run, meaning that the machined surface has a smaller profile height between surface peaks and valleys. This translates in to a smoother and stronger machined surface with better dimensional and mechanical properties.
Due to its configuration (4 cutting flutes), coating and toughness, it is an ideal tool for machining harder materials, as the one used in this experiment.
Although radial roughness values increased from 8 to 16 cycles (as expected), the difference between both is of low significance. This leads to the assumption that wear evolution is slow, when compared to other tools, translating into a longer tool life.
The SEM analysis of the tool surfaces reveals that, for the shorter machining cycle (shown in Table 5), tool wear is very low with a small VB value (flank wear), showing edge chipping as the only type of wear present. Its influence is of low importance as shown by the small roughness values discussed earlier. The images shown in Table 5 demonstrate the existence of several surface impurities.         When analyzing the results of the SEM analysis after 16 cycles (referring to Table 6), it is possible to observe VB values have increased, indicating further tool wear. When analyzing the results of the SEM analysis after 16 cycles (referring to Table 6), it is possible to observe VB values have increased, indicating further tool wear.

Observable Issues Analysis Damage Issues Probable Causes Possible Solutions
Notch Wear [24] Coatings 2016, 6, 51 8 of 30      In Figure 3 it is possible to observe distinct areas, such as regions where the tool has lost its coating. This phenomenon is due to the natural occurring wear caused by friction and abrasion between the tool and the material being machined. It is also possible to observe white specs which are caused by the collision of metal shavings with the tool surface during machining operations. The area marked as Z1 shows a section with intact coating composed of titanium aluminium nitride. The marked Z2 area contains adhered material from the machined part, tool substrate material and several other impurities, being composed mainly of carbon, titanium, chromium, iron and manganese. The adhesion of these impurities may be related to the geometry of this end mill, as 4 cutting flutes hinder the ease of chip removal, promoting collisions between the tool and the built up material. The area marked as Z3 represents the tool substrate as the EDS analysis displays elements coherent with the composition of hard metal.
In Figure 3 it is possible to observe distinct areas, such as regions where the tool has lost its coating. This phenomenon is due to the natural occurring wear caused by friction and abrasion between the tool and the material being machined. It is also possible to observe white specs which are caused by the collision of metal shavings with the tool surface during machining operations. The area marked as Z1 shows a section with intact coating composed of titanium aluminium nitride. The marked Z2 area contains adhered material from the machined part, tool substrate material and several other impurities, being composed mainly of carbon, titanium, chromium, iron and manganese. The adhesion of these impurities may be related to the geometry of this end mill, as 4 cutting flutes hinder the ease of chip removal, promoting collisions between the tool and the built up material. The area marked as Z3 represents the tool substrate as the EDS analysis displays elements coherent with the composition of hard metal.

HPMT 3Z AlCrN-Roughness and SEM Results
Roughness results shown in Table 7 demonstrate no significant variation in terms of Ra roughness from one trial to the other. Rz roughness values for 8 cycles are relatively low, although worse than when compared to the previous Walter 4Z TiAlN end mill. The evolution of roughness values from 8 to 16 machining cycles clearly demonstrates an increase of surface roughness and loss of tool cutting quality. Tangential Ra and Rz values are in fact the highest values obtained across all tested tools, creating a coarser surface finish which can lead to loss of dimensional tolerance. Radial values position this tool in third place, which translates to a machined surface with noticeable feed marks.

HPMT 3Z AlCrN-Roughness and SEM Results
Roughness results shown in Table 7 demonstrate no significant variation in terms of R a roughness from one trial to the other. R z roughness values for 8 cycles are relatively low, although worse than when compared to the previous Walter 4Z TiAlN end mill. The evolution of roughness values from 8 to 16 machining cycles clearly demonstrates an increase of surface roughness and loss of tool cutting quality. Tangential R a and R z values are in fact the highest values obtained across all tested tools, creating a coarser surface finish which can lead to loss of dimensional tolerance. Radial values position this tool in third place, which translates to a machined surface with noticeable feed marks. Looking at the SEM analysis for 8 cycles, shown in Table 8, it is possible to observe foreign material adhered to the tool surface. As shown in Figure 4, the analysis of the adhered material shows that it is mainly composed of iron, nickel and chromium, consistent with the composition of stainless steel, confirming that the origin of this material is the machined part. This type of phenomena may lead to a large build up on the tool surface and consequently coating detachment.
In Figure 4, it is possible to closely view several distinct areas representing wear and material adhesion, as well as the EDS analysis for those areas.  Looking at the SEM analysis for 8 cycles, shown in Table 8, it is possible to observe foreign material adhered to the tool surface. As shown in Figure 4, the analysis of the adhered material shows that it is mainly composed of iron, nickel and chromium, consistent with the composition of stainless steel, confirming that the origin of this material is the machined part. This type of phenomena may lead to a large build up on the tool surface and consequently coating detachment.
In Figure 4, it is possible to closely view several distinct areas representing wear and material adhesion, as well as the EDS analysis for those areas.  Looking at the SEM analysis for 8 cycles, shown in Table 8, it is possible to observe foreign material adhered to the tool surface. As shown in Figure 4, the analysis of the adhered material shows that it is mainly composed of iron, nickel and chromium, consistent with the composition of stainless steel, confirming that the origin of this material is the machined part. This type of phenomena may lead to a large build up on the tool surface and consequently coating detachment.
In Figure 4, it is possible to closely view several distinct areas representing wear and material adhesion, as well as the EDS analysis for those areas.  Looking at the SEM analysis for 8 cycles, shown in Table 8, it is possible to observe foreign material adhered to the tool surface. As shown in Figure 4, the analysis of the adhered material shows that it is mainly composed of iron, nickel and chromium, consistent with the composition of stainless steel, confirming that the origin of this material is the machined part. This type of phenomena may lead to a large build up on the tool surface and consequently coating detachment.
In Figure 4, it is possible to closely view several distinct areas representing wear and material adhesion, as well as the EDS analysis for those areas.  Looking at the SEM analysis for 8 cycles, shown in Table 8, it is possible to observe foreign material adhered to the tool surface. As shown in Figure 4, the analysis of the adhered material shows that it is mainly composed of iron, nickel and chromium, consistent with the composition of stainless steel, confirming that the origin of this material is the machined part. This type of phenomena may lead to a large build up on the tool surface and consequently coating detachment.
In Figure 4, it is possible to closely view several distinct areas representing wear and material adhesion, as well as the EDS analysis for those areas.   When comparing flank wear between both trials, an increase of values is noticeable after 16 cycles of machining. For instance, the highest VB value for the 8 cycles run was 117.50 µ m, while the 16 cycles run present a value of 380.00 µ m. Nevertheless, the performance of this tool is acceptable bearing in mind its low acquisition price. These values can be seen in Tables 8 and 9. When comparing flank wear between both trials, an increase of values is noticeable after 16 cycles of machining. For instance, the highest VB value for the 8 cycles run was 117.50 µ m, while the 16 cycles run present a value of 380.00 µ m. Nevertheless, the performance of this tool is acceptable bearing in mind its low acquisition price. These values can be seen in Tables 8 and 9. When comparing flank wear between both trials, an increase of values is noticeable after 16 cycles of machining. For instance, the highest VB value for the 8 cycles run was 117.50 µ m, while the 16 cycles run present a value of 380.00 µ m. Nevertheless, the performance of this tool is acceptable bearing in mind its low acquisition price. These values can be seen in Tables 8 and 9. When comparing flank wear between both trials, an increase of values is noticeable after 16 cycles of machining. For instance, the highest VB value for the 8 cycles run was 117.50 µm, while the 16 cycles run present a value of 380.00 µm. Nevertheless, the performance of this tool is acceptable bearing in mind its low acquisition price. These values can be seen in Tables 8 and 9. When comparing flank wear between both trials, an increase of values is noticeable after 16 cycles of machining. For instance, the highest VB value for the 8 cycles run was 117.50 µ m, while the 16 cycles run present a value of 380.00 µ m. Nevertheless, the performance of this tool is acceptable bearing in mind its low acquisition price. These values can be seen in Tables 8 and 9.

HPMT 4Z AlCrN-Roughness and SEM Results
Even though all machining and testing parameters where kept equal for all trials, the results obtained by this tool are somewhat unexpected. Seen in Table 10, the surface roughness produced after 8 cycles was the worst of all measured results. Regarding the 16 cycles run, the Ra radial roughness is, as well, the highest of the batch of experiments, delivering a machined surface of inferior quality. However, in terms of tangential Ra values, this tool delivered a surprisingly low surface roughness, being tied with the WALTER 4Z TiAlN end mill (best values measured across all experiments). When comparing these two tools, the HPMT 4Z AlCrN has a major price advantage, with the downside of having a shorter working life.

HPMT 4Z AlCrN-Roughness and SEM Results
Even though all machining and testing parameters where kept equal for all trials, the results obtained by this tool are somewhat unexpected. Seen in Table 10, the surface roughness produced after 8 cycles was the worst of all measured results. Regarding the 16 cycles run, the Ra radial roughness is, as well, the highest of the batch of experiments, delivering a machined surface of inferior quality. However, in terms of tangential Ra values, this tool delivered a surprisingly low surface roughness, being tied with the WALTER 4Z TiAlN end mill (best values measured across all experiments). When comparing these two tools, the HPMT 4Z AlCrN has a major price advantage, with the downside of having a shorter working life.

HPMT 4Z AlCrN-Roughness and SEM Results
Even though all machining and testing parameters where kept equal for all trials, the results obtained by this tool are somewhat unexpected. Seen in Table 10, the surface roughness produced after 8 cycles was the worst of all measured results. Regarding the 16 cycles run, the Ra radial roughness is, as well, the highest of the batch of experiments, delivering a machined surface of inferior quality. However, in terms of tangential Ra values, this tool delivered a surprisingly low surface roughness, being tied with the WALTER 4Z TiAlN end mill (best values measured across all experiments). When comparing these two tools, the HPMT 4Z AlCrN has a major price advantage, with the downside of having a shorter working life.

HPMT 4Z AlCrN-Roughness and SEM Results
Even though all machining and testing parameters where kept equal for all trials, the results obtained by this tool are somewhat unexpected. Seen in Table 10, the surface roughness produced after 8 cycles was the worst of all measured results. Regarding the 16 cycles run, the Ra radial roughness is, as well, the highest of the batch of experiments, delivering a machined surface of inferior quality. However, in terms of tangential Ra values, this tool delivered a surprisingly low surface roughness, being tied with the WALTER 4Z TiAlN end mill (best values measured across all experiments). When comparing these two tools, the HPMT 4Z AlCrN has a major price advantage, with the downside of having a shorter working life.

HPMT 4Z AlCrN-Roughness and SEM Results
Even though all machining and testing parameters where kept equal for all trials, the results obtained by this tool are somewhat unexpected. Seen in Table 10, the surface roughness produced after 8 cycles was the worst of all measured results. Regarding the 16 cycles run, the Ra radial roughness is, as well, the highest of the batch of experiments, delivering a machined surface of inferior quality. However, in terms of tangential R a values, this tool delivered a surprisingly low surface roughness, being tied with the WALTER 4Z TiAlN end mill (best values measured across all experiments). When comparing these two tools, the HPMT 4Z AlCrN has a major price advantage, with the downside of having a shorter working life. As referred to previously, the results obtained after 8 cycles indicate poor performance. When observing Table 11, it is possible to see that most cutting flanks suffered severe damage. The registered VB wear values after 8 machining cycles are the highest of all tools (exceeding in some cases 1 mm in length). The most plausible hypothesis for this phenomenon is the fact that the material being machined is not completely homogenous, containing localized areas with higher hardness values. These areas are essentially microstructures with different metallurgical phases. The existence of these heterogeneities can be seen, in Figure 5, being also possible to observe a dragging pattern on the surface of the machined part. This pattern was formed by the impact between the tool and the higher hardness areas and seems to be the main reason of wear in tools, being particularly noticeable during this trial.
When viewing the SEM images for the 16 cycles run (Table 12), it is possible to observe the loss of the flank edges. This type of wear caused higher radial R a roughness values and led to more noticeable feed mark embossments.  As referred to previously, the results obtained after 8 cycles indicate poor performance. When observing Table 11, it is possible to see that most cutting flanks suffered severe damage. The registered VB wear values after 8 machining cycles are the highest of all tools (exceeding in some cases 1 mm in length). The most plausible hypothesis for this phenomenon is the fact that the material being machined is not completely homogenous, containing localized areas with higher hardness values. These areas are essentially microstructures with different metallurgical phases. The existence of these heterogeneities can be seen, in Figure 5, being also possible to observe a dragging pattern on the surface of the machined part. This pattern was formed by the impact between the tool and the higher hardness areas and seems to be the main reason of wear in tools, being particularly noticeable during this trial.
When viewing the SEM images for the 16 cycles run (Table 12), it is possible to observe the loss of the flank edges. This type of wear caused higher radial Ra roughness values and led to more noticeable feed mark embossments.  As referred to previously, the results obtained after 8 cycles indicate poor performance. When observing Table 11, it is possible to see that most cutting flanks suffered severe damage. The registered VB wear values after 8 machining cycles are the highest of all tools (exceeding in some cases 1 mm in length). The most plausible hypothesis for this phenomenon is the fact that the material being machined is not completely homogenous, containing localized areas with higher hardness values. These areas are essentially microstructures with different metallurgical phases. The existence of these heterogeneities can be seen, in Figure 5, being also possible to observe a dragging pattern on the surface of the machined part. This pattern was formed by the impact between the tool and the higher hardness areas and seems to be the main reason of wear in tools, being particularly noticeable during this trial.
When viewing the SEM images for the 16 cycles run (Table 12), it is possible to observe the loss of the flank edges. This type of wear caused higher radial Ra roughness values and led to more noticeable feed mark embossments.  As referred to previously, the results obtained after 8 cycles indicate poor performance. When observing Table 11, it is possible to see that most cutting flanks suffered severe damage. The registered VB wear values after 8 machining cycles are the highest of all tools (exceeding in some cases 1 mm in length). The most plausible hypothesis for this phenomenon is the fact that the material being machined is not completely homogenous, containing localized areas with higher hardness values. These areas are essentially microstructures with different metallurgical phases. The existence of these heterogeneities can be seen, in Figure 5, being also possible to observe a dragging pattern on the surface of the machined part. This pattern was formed by the impact between the tool and the higher hardness areas and seems to be the main reason of wear in tools, being particularly noticeable during this trial.
When viewing the SEM images for the 16 cycles run (Table 12), it is possible to observe the loss of the flank edges. This type of wear caused higher radial Ra roughness values and led to more noticeable feed mark embossments.                      In Figure 6a, which is a close up of cutting edge 1, it is possible to observe several white specs which indicate lack of coating. This phenomenon occurs when machined chips collide with the end mill and chisel away at the surface coating. EDS analysis in Figure 6b shows the composition of area Z1 as being perfectly coherent with the composition of the tool substrate (hard metal), indicating loss of coating. In the same manner, the Z2 area is composed mainly of aluminum and chromium, which are the elements that form the AlCrN coating. In Figure 6a, which is a close up of cutting edge 1, it is possible to observe several white specs which indicate lack of coating. This phenomenon occurs when machined chips collide with the end mill and chisel away at the surface coating. EDS analysis in Figure 6b shows the composition of area Z1 as being perfectly coherent with the composition of the tool substrate (hard metal), indicating loss of coating. In the same manner, the Z2 area is composed mainly of aluminum and chromium, which are the elements that form the AlCrN coating. In Figure 6a, which is a close up of cutting edge 1, it is possible to observe several white specs which indicate lack of coating. This phenomenon occurs when machined chips collide with the end mill and chisel away at the surface coating. EDS analysis in Figure 6b shows the composition of area Z1 as being perfectly coherent with the composition of the tool substrate (hard metal), indicating loss of coating. In the same manner, the Z2 area is composed mainly of aluminum and chromium, which are the elements that form the AlCrN coating. In Figure 6a, which is a close up of cutting edge 1, it is possible to observe several white specs which indicate lack of coating. This phenomenon occurs when machined chips collide with the end mill and chisel away at the surface coating. EDS analysis in Figure 6b shows the composition of area Z1 as being perfectly coherent with the composition of the tool substrate (hard metal), indicating loss of coating. In the same manner, the Z2 area is composed mainly of aluminum and chromium, which are the elements that form the AlCrN coating. In Figure 6a, which is a close up of cutting edge 1, it is possible to observe several white specs which indicate lack of coating. This phenomenon occurs when machined chips collide with the end mill and chisel away at the surface coating. EDS analysis in Figure 6b shows the composition of area Z1 as being perfectly coherent with the composition of the tool substrate (hard metal), indicating loss of coating. In the same manner, the Z2 area is composed mainly of aluminum and chromium, which are the elements that form the AlCrN coating. In Figure 6a, which is a close up of cutting edge 1, it is possible to observe several white specs which indicate lack of coating. This phenomenon occurs when machined chips collide with the end mill and chisel away at the surface coating. EDS analysis in Figure 6b shows the composition of area Z1 as being perfectly coherent with the composition of the tool substrate (hard metal), indicating loss of coating. In the same manner, the Z2 area is composed mainly of aluminum and chromium, which are the elements that form the AlCrN coating.

DORMER Spectrum S812HA-Roughness and SEM Results
This tool returned low surface roughness values for the 8 cycles run (Table 13). Radial and tangential Ra and Rz values for this test were the lowest of all tested tools. This combination of factors leads to almost imperceptible feed marks on the surface of the machined part. On the other hand, the evolution in roughness values from 8 to 16 cycles indicates a rapid tool degradation, which may translate into low tool life/durability.
Bearing in mind that this mill just presents two cutting edges, the roughness values obtained after 16 cycles of machining can be considered reasonable. Lesser cutting knives signify higher working loads per knife (twice as much when compared to a 4-edge cutting mill), so the observed values should take into account this factor.

DORMER Spectrum S812HA-Roughness and SEM Results
This tool returned low surface roughness values for the 8 cycles run (Table 13). Radial and tangential R a and R z values for this test were the lowest of all tested tools. This combination of factors leads to almost imperceptible feed marks on the surface of the machined part. On the other hand, the evolution in roughness values from 8 to 16 cycles indicates a rapid tool degradation, which may translate into low tool life/durability.
Bearing in mind that this mill just presents two cutting edges, the roughness values obtained after 16 cycles of machining can be considered reasonable. Lesser cutting knives signify higher working loads per knife (twice as much when compared to a 4-edge cutting mill), so the observed values should take into account this factor.
When taking an overall look at Table 13, it is possible to say that this tool creates good machined surface quality, especially for shorter machining paths. However, due to its high degradation rate, its longevity may be reduced. In accordance with the previously discussed roughness results, the SEM analysis for an 8 cycles run (Table 14) illustrates that this tool does not present any major damage to its cutting edges. It is also noticeable that one cutting edge returns a higher VB value than the other edge. This situation is mainly caused by the fact that one of the cutting flanks executes an initial roughing cut, with a larger material removal rate, while the other edge executes a type of finishing pass, rectifying the previous cut. The loads applied to the secondary edge are smaller and consequently less severe.
As stated previously, in spite of this having a superior wear after 16 cycles, the work load completed by each cutting edge is higher than any other tool present in this experiment. Regardless of this fact, the tool generated a fairly satisfactory machined surface quality.
No EDS analysis was made as this tool did not show any major differences from the previous ones tested. Another small note that can be made is that due to its geometrical configuration, chip extraction is more easily accomplished, which consequently minimizes collision/adhesion of material to the tool's cutting edges. This can be confirmed by the SEM imagery, showing a relatively low presence of white specs across the end mill surface, as can be seen in Table 15, where the tool wear can be observed and the probable causes of that wear are listed, as well as some possible solutions. When taking an overall look at Table 13, it is possible to say that this tool creates good machined surface quality, especially for shorter machining paths. However, due to its high degradation rate, its longevity may be reduced. In accordance with the previously discussed roughness results, the SEM analysis for an 8 cycles run (Table 14) illustrates that this tool does not present any major damage to its cutting edges. It is also noticeable that one cutting edge returns a higher VB value than the other edge. This situation is mainly caused by the fact that one of the cutting flanks executes an initial roughing cut, with a larger material removal rate, while the other edge executes a type of finishing pass, rectifying the previous cut. The loads applied to the secondary edge are smaller and consequently less severe.
As stated previously, in spite of this having a superior wear after 16 cycles, the work load completed by each cutting edge is higher than any other tool present in this experiment. Regardless of this fact, the tool generated a fairly satisfactory machined surface quality.
No EDS analysis was made as this tool did not show any major differences from the previous ones tested. Another small note that can be made is that due to its geometrical configuration, chip extraction is more easily accomplished, which consequently minimizes collision/adhesion of material to the tool's cutting edges. This can be confirmed by the SEM imagery, showing a relatively low presence of white specs across the end mill surface, as can be seen in Table 15, where the tool wear can be observed and the probable causes of that wear are listed, as well as some possible solutions. When taking an overall look at Table 13, it is possible to say that this tool creates good machined surface quality, especially for shorter machining paths. However, due to its high degradation rate, its longevity may be reduced. In accordance with the previously discussed roughness results, the SEM analysis for an 8 cycles run (Table 14) illustrates that this tool does not present any major damage to its cutting edges. It is also noticeable that one cutting edge returns a higher VB value than the other edge. This situation is mainly caused by the fact that one of the cutting flanks executes an initial roughing cut, with a larger material removal rate, while the other edge executes a type of finishing pass, rectifying the previous cut. The loads applied to the secondary edge are smaller and consequently less severe.
As stated previously, in spite of this having a superior wear after 16 cycles, the work load completed by each cutting edge is higher than any other tool present in this experiment. Regardless of this fact, the tool generated a fairly satisfactory machined surface quality.
No EDS analysis was made as this tool did not show any major differences from the previous ones tested. Another small note that can be made is that due to its geometrical configuration, chip extraction is more easily accomplished, which consequently minimizes collision/adhesion of material to the tool's cutting edges. This can be confirmed by the SEM imagery, showing a relatively low presence of white specs across the end mill surface, as can be seen in Table 15, where the tool wear can be observed and the probable causes of that wear are listed, as well as some possible solutions.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection.

Chip Analysis for DORMER 2Z AlCrN and WALTER 3Z AlCrN End Mills
Chips generated by mill tools HPTM 3Z AlCrN and DORMER 2Z AlCrN were collected after the 8 cycles machining trials and observed under the SEM microscope. By observing Figure 7, it is possible to perceive differences between the chips created by each tool.
The chips created by the HPMT 3Z AlCrN end mill demonstrate a rougher and more irregular edge/rim. This geometry may be explained by the higher rake angle (less sharp) of this tool, which consequently generates higher friction loads between cutting edges and chips, and by the additional difficulty of chip extraction due to the increased number of cutting flutes. That said, the chip surface (seen in Figure 8) displays greater surface irregularities, indicating the existence of larger loads during extraction/ejection. The chips originated by the DORMER 2Z AlCrN end mill demonstrate a cleaner and more uniform surface and rim edge. This is mainly due to its facilitated extraction and lower friction loads. The chips originated by the DORMER 2Z AlCrN end mill demonstrate a cleaner and more uniform surface and rim edge. This is mainly due to its facilitated extraction and lower friction loads. The chips originated by the DORMER 2Z AlCrN end mill demonstrate a cleaner and more uniform surface and rim edge. This is mainly due to its facilitated extraction and lower friction loads. In Figure 9, it is possible to observe the surface of the machined chip, demonstrating a lesser crushed surface mainly due to the smaller rank angle (sharper) and spaced out cutting flutes.
Coatings 2016, 6, 51 24 of 30 In Figure 9, it is possible to observe the surface of the machined chip, demonstrating a lesser crushed surface mainly due to the smaller rank angle (sharper) and spaced out cutting flutes.

Discussion
In this section, a global analysis was done in order to establish which tools demonstrated superior performance and what lead to such results.
Depending on which tool was used and the length of the trial, several different roughness results were measured on each machined surface. The shorter 8 cycles trials presented lower roughness values, as expected. As seen in Figure 10, the tool which produced the lowest surface roughness for the shorter cycle run was the DORMER 2Z AlCrN end mill, having, as well, the least difference between radial and tangential roughness values. This indicates a more uniform surface without noticeable feed marks. The Rz values were also the smallest of all tests, indicating a low variation between surface peaks and valleys, which ensures longer part life with continued dimensional accuracy.
The WALTER 4Z TiAlN gave slightly higher roughness values for the 8 cycles when compared to the previous mentioned tool, however, it returned the most consistent global performance.
Regarding the HPMT 3Z AlCrN tool, reasonable roughness (including Rz) values were obtained, being the third best tool for the 8 cycles in terms of machining length.
End mill HPMT 4Z AlCrN performed poorly, returning very high values. This outcome was somewhat unexpected and, perhaps, may be a consequence of the excessive wear/damage suffered by the tool on the account of hardened material phases possibly present in the raw material. A possible manufacturing defect may also be speculated as the reason for such a rapid downfall. Figure 10 also shows, in terms of radial roughness, an end mill which negatively stands out among the rest (HPMT 4Z AlCrN). While all other end mills showed similar Ra results, HPMT 4Z AlCrN mill revealed higher Ra radial roughness. As previously discussed, this tool suffered damage during the experimental trials, leading to a loss of material and edge sharpness of the cutting flutes. This phenomenon may explain the more noticeable feed mark embossments, leading to the larger radial Rz and Rmax roughness values.
Tangential roughness values for the 16 cycles trials, shown in Figure 11, clearly demonstrate that WALTER 4Z TiAlN and HPMT 4Z AlCrN obtained the best results. The remaining tools present similar and satisfactory tangential Ra values, however, they exhibit higher Rz values which lead to a more irregular surface in terms of peak height and valley depth.

Discussion
In this section, a global analysis was done in order to establish which tools demonstrated superior performance and what lead to such results.
Depending on which tool was used and the length of the trial, several different roughness results were measured on each machined surface. The shorter 8 cycles trials presented lower roughness values, as expected. As seen in Figure 10, the tool which produced the lowest surface roughness for the shorter cycle run was the DORMER 2Z AlCrN end mill, having, as well, the least difference between radial and tangential roughness values. This indicates a more uniform surface without noticeable feed marks. The R z values were also the smallest of all tests, indicating a low variation between surface peaks and valleys, which ensures longer part life with continued dimensional accuracy.
The WALTER 4Z TiAlN gave slightly higher roughness values for the 8 cycles when compared to the previous mentioned tool, however, it returned the most consistent global performance.
Regarding the HPMT 3Z AlCrN tool, reasonable roughness (including R z ) values were obtained, being the third best tool for the 8 cycles in terms of machining length.
End mill HPMT 4Z AlCrN performed poorly, returning very high values. This outcome was somewhat unexpected and, perhaps, may be a consequence of the excessive wear/damage suffered by the tool on the account of hardened material phases possibly present in the raw material. A possible manufacturing defect may also be speculated as the reason for such a rapid downfall. Figure 10 also shows, in terms of radial roughness, an end mill which negatively stands out among the rest (HPMT 4Z AlCrN). While all other end mills showed similar R a results, HPMT 4Z AlCrN mill revealed higher R a radial roughness. As previously discussed, this tool suffered damage during the experimental trials, leading to a loss of material and edge sharpness of the cutting flutes. This phenomenon may explain the more noticeable feed mark embossments, leading to the larger radial R z and R max roughness values.
Tangential roughness values for the 16 cycles trials, shown in Figure 11, clearly demonstrate that WALTER 4Z TiAlN and HPMT 4Z AlCrN obtained the best results. The remaining tools present similar and satisfactory tangential R a values, however, they exhibit higher R z values which lead to a more irregular surface in terms of peak height and valley depth.  All tools used in the experimental trials demonstrated similar types of wear, having some suffered more wear than others. In a broad way, it is possible to say that the main issues encountered were flank wear, chipping, cracking and breakage of the cutting edges. These issues manifested in   All tools used in the experimental trials demonstrated similar types of wear, having some suffered more wear than others. In a broad way, it is possible to say that the main issues encountered were flank wear, chipping, cracking and breakage of the cutting edges. These issues manifested in Figure 11. Roughness values observed on machined surfaces after 16 cycles.
All tools used in the experimental trials demonstrated similar types of wear, having some suffered more wear than others. In a broad way, it is possible to say that the main issues encountered were flank wear, chipping, cracking and breakage of the cutting edges. These issues manifested in different levels of severity. Tools with higher wear demonstrated poorer machined surfaces, while tools that suffered breakage created surfaces with noticeable feed marks. The depth of cut used for all trials (0.5 mm) was more in tune with a finishing operation rather than a conventional roughing operation. The small material removal rate, aided with cutting fluid, allows a reduction of temperatures at the tool's tip. The use of cutting fluids gives way to a large variation of temperature during machining, creating heat cycles which may lead to thermal fatigue. This fatigue may perhaps be the cause of, for instance, cutting edge fractures or other encountered defects. When adding the inherent vibration (created by the machining of tough materials) to the equation, the potential for tool chipping and crack propagation is increased, which then leads to a higher tool failure probability.
A comparison of flank wear suffered on each tool can be seen in Figures 12 and 13. The WALTER 4Z TiAlN is the end mill with the lowest flank wear in the group having, as well, the smallest wear evolution when compared to the rest. different levels of severity. Tools with higher wear demonstrated poorer machined surfaces, while tools that suffered breakage created surfaces with noticeable feed marks. The depth of cut used for all trials (0.5 mm) was more in tune with a finishing operation rather than a conventional roughing operation. The small material removal rate, aided with cutting fluid, allows a reduction of temperatures at the tool's tip. The use of cutting fluids gives way to a large variation of temperature during machining, creating heat cycles which may lead to thermal fatigue. This fatigue may perhaps be the cause of, for instance, cutting edge fractures or other encountered defects. When adding the inherent vibration (created by the machining of tough materials) to the equation, the potential for tool chipping and crack propagation is increased, which then leads to a higher tool failure probability.
A comparison of flank wear suffered on each tool can be seen in Figures 12 and 13. The WALTER 4Z TiAlN is the end mill with the lowest flank wear in the group having, as well, the smallest wear evolution when compared to the rest.    different levels of severity. Tools with higher wear demonstrated poorer machined surfaces, while tools that suffered breakage created surfaces with noticeable feed marks. The depth of cut used for all trials (0.5 mm) was more in tune with a finishing operation rather than a conventional roughing operation. The small material removal rate, aided with cutting fluid, allows a reduction of temperatures at the tool's tip. The use of cutting fluids gives way to a large variation of temperature during machining, creating heat cycles which may lead to thermal fatigue. This fatigue may perhaps be the cause of, for instance, cutting edge fractures or other encountered defects. When adding the inherent vibration (created by the machining of tough materials) to the equation, the potential for tool chipping and crack propagation is increased, which then leads to a higher tool failure probability.
A comparison of flank wear suffered on each tool can be seen in Figures 12 and 13. The WALTER 4Z TiAlN is the end mill with the lowest flank wear in the group having, as well, the smallest wear evolution when compared to the rest.   DORMER 2Z AlCrN came in second in terms of flank wear for the smaller cycle run, however, when compared to the results after 16 machining cycles, it is noticeable that this tool suffered severe cutting edge wear. As this tool only has two cutting flutes, it is at a disadvantage when compared to the others, demanding that each flute do extra work to achieve the same 15 m machining distance.
HPMT 3Z AlCrN achieved similar low VB values for the initial trial indicating low tool wear, and demonstrated the second best result after 16 cycles of machining. Although obtaining reasonable results, a large variation of flank wear between cutting flutes is present. Despite the fact that a low flank wear is present, this tool gave the highest surface roughness of all machined parts, delivering an inferior machined surface finish.
As stated previously, the HPMT 4Z AlCrN end mill suffered extended wear and damage during the first trial, exhibiting an overall significant degree of wear. SEM results for the lengthier trial show a large wear variation between cutting flutes. This variation may support the statement given previously which speculates that this tool dealt with different levels of hardness during its machining operation due to the presence of heterogeneities in the material.
Considering Figures 12-15, it is possible to state that the WALTER 4Z TiAlN end mill: • Showed the overall best performance of the experiments; • Gave the second best/lowest radial and tangential surface roughness results for the 8 cycles trial and the best/lowest for the 16 cycles trial; • The quality of the machined surface is maintained independently of the increase in machining distance/working life of the tool. This can be seen by the registered low VB values of the cutting flutes and by the small slope of the line that correlates surface roughness between the first and second trials. This slope is the smallest among all the tools indicating a slower wear evolution and consequently longer tool life.
As can be seen in the previous figures, the coatings are valid essentially during the initial stage of the trials. Posteriorly, it is possible to observe deeply worn cutting edges, displaying coating detachment and requiring the hard metal substrate to work without any of the advantages conferred by the coating. Thus, it is important to observe the wear evolution from the 8 cycles to 16 cycles trials. The TiAlN coating displays better results in terms of flank wear, both in 8 and 16 cycles trials. All the AlCrN coatings demonstrate premature detachment and severe wear, mainly after 16 machining cycles. Hence, it can be stated that, when machining Duplex Stainless Steel, a TiAlN coating displays better results in terms of machining work than a AlCrN coating.
Coatings 2016, 6, 51 27 of 30 DORMER 2Z AlCrN came in second in terms of flank wear for the smaller cycle run, however, when compared to the results after 16 machining cycles, it is noticeable that this tool suffered severe cutting edge wear. As this tool only has two cutting flutes, it is at a disadvantage when compared to the others, demanding that each flute do extra work to achieve the same 15 m machining distance.
HPMT 3Z AlCrN achieved similar low VB values for the initial trial indicating low tool wear, and demonstrated the second best result after 16 cycles of machining. Although obtaining reasonable results, a large variation of flank wear between cutting flutes is present. Despite the fact that a low flank wear is present, this tool gave the highest surface roughness of all machined parts, delivering an inferior machined surface finish.
As stated previously, the HPMT 4Z AlCrN end mill suffered extended wear and damage during the first trial, exhibiting an overall significant degree of wear. SEM results for the lengthier trial show a large wear variation between cutting flutes. This variation may support the statement given previously which speculates that this tool dealt with different levels of hardness during its machining operation due to the presence of heterogeneities in the material.
Considering Figures 12-15, it is possible to state that the WALTER 4Z TiAlN end mill:  Showed the overall best performance of the experiments;  Gave the second best/lowest radial and tangential surface roughness results for the 8 cycles trial and the best/lowest for the 16 cycles trial;  The quality of the machined surface is maintained independently of the increase in machining distance/working life of the tool. This can be seen by the registered low VB values of the cutting flutes and by the small slope of the line that correlates surface roughness between the first and second trials. This slope is the smallest among all the tools indicating a slower wear evolution and consequently longer tool life.
As can be seen in the previous figures, the coatings are valid essentially during the initial stage of the trials. Posteriorly, it is possible to observe deeply worn cutting edges, displaying coating detachment and requiring the hard metal substrate to work without any of the advantages conferred by the coating. Thus, it is important to observe the wear evolution from the 8 cycles to 16 cycles trials. The TiAlN coating displays better results in terms of flank wear, both in 8 and 16 cycles trials. All the AlCrN coatings demonstrate premature detachment and severe wear, mainly after 16 machining cycles. Hence, it can be stated that, when machining Duplex Stainless Steel, a TiAlN coating displays better results in terms of machining work than a AlCrN coating.   Hard metal grade seems to play an important role in wear evolution as some tools showed substrate wear and corresponding degradation in terms of flank wear.
The number of flutes is also an important issue to consider because when using the same feed speed on a tool with less cutting flutes, the number of tool rotations and corresponding impact increase, leading to a premature cutting edge wear.
The overall best tool (WALTER 4Z TiAlN) has a coating composed by titanium aluminium nitride. This coating was shown to be generally superior in terms of longevity and quality of machined surfaces. The aluminium oxide promotes high temperature resistance while the combination of elements promotes a hard and tough surface. In fact, the general consensus is that tools that are properly coated improve longevity and optimize cutting parameters.

Conclusions
All in all, it is possible to state that end mills with four cutting flutes maintained a better surface roughness, presenting the lowest Ra, Rz and Rmax results for the 16 cycles trials. They are an appropriate choice to perform side milling operations on the grade of Duplex Stainless Steel used in this study. Nonetheless, the other tested end mills showed interesting results (only slightly worse than the four-flute end mills), being more suitable for slotting or down milling operations. In these types of machining operations, having lesser cutting flutes gives the advantage of easily and effectively extracting chips from the tooling edges. This ease of extraction avoids potential material adhesion or build up on cutting edges, phenomena that worsens when machining materials with similar compositions as the ones here tested. On the downside, tools with only two or three cutting flutes tend to have a shorter working life owing to the fact that, to machine the same distance, each cutting knife has a higher work load when compared, for example, to four-flute end mills. The work load per tooth is independent of speed parameters, such as the feed per tooth speed of the machining center.
The duplex stainless steel used in the experiments proved to be very harsh on the cutting tools. Even though the machining distance used in the trials is considered to be relatively short, the tested tools suffered extended damage and wear, exemplifying how difficult it is to work and process these types of materials.
It seems to have been made clear that WALTER 4Z TiAlN end mill delivered the best results in terms of wear and almost always of machined surface quality. Worth mentioning as well is that, in spite of suffering large wear, the surface quality of parts machined with DORMER 2Z AlCrN end mill revealed very good results, being close to those of the other tools. However, this outcome can be attributed essentially to the quality of the tool substrate, made from a very good hard metal grade. Hard metal grade seems to play an important role in wear evolution as some tools showed substrate wear and corresponding degradation in terms of flank wear.
The number of flutes is also an important issue to consider because when using the same feed speed on a tool with less cutting flutes, the number of tool rotations and corresponding impact increase, leading to a premature cutting edge wear.
The overall best tool (WALTER 4Z TiAlN) has a coating composed by titanium aluminium nitride. This coating was shown to be generally superior in terms of longevity and quality of machined surfaces. The aluminium oxide promotes high temperature resistance while the combination of elements promotes a hard and tough surface. In fact, the general consensus is that tools that are properly coated improve longevity and optimize cutting parameters.

Conclusions
All in all, it is possible to state that end mills with four cutting flutes maintained a better surface roughness, presenting the lowest R a , R z and R max results for the 16 cycles trials. They are an appropriate choice to perform side milling operations on the grade of Duplex Stainless Steel used in this study. Nonetheless, the other tested end mills showed interesting results (only slightly worse than the four-flute end mills), being more suitable for slotting or down milling operations. In these types of machining operations, having lesser cutting flutes gives the advantage of easily and effectively extracting chips from the tooling edges. This ease of extraction avoids potential material adhesion or build up on cutting edges, phenomena that worsens when machining materials with similar compositions as the ones here tested. On the downside, tools with only two or three cutting flutes tend to have a shorter working life owing to the fact that, to machine the same distance, each cutting knife has a higher work load when compared, for example, to four-flute end mills. The work load per tooth is independent of speed parameters, such as the feed per tooth speed of the machining center.
The duplex stainless steel used in the experiments proved to be very harsh on the cutting tools. Even though the machining distance used in the trials is considered to be relatively short, the tested tools suffered extended damage and wear, exemplifying how difficult it is to work and process these types of materials.
It seems to have been made clear that WALTER 4Z TiAlN end mill delivered the best results in terms of wear and almost always of machined surface quality. Worth mentioning as well is that, in spite of suffering large wear, the surface quality of parts machined with DORMER 2Z AlCrN end mill revealed very good results, being close to those of the other tools. However, this outcome can be attributed essentially to the quality of the tool substrate, made from a very good hard metal grade.
Thus, despite the large evolution registered by AlCrN coatings, this work shows that TiAlN is yet more effective in cutting tough materials, such as Duplex Stainless Steel grades, in end mill machining operations.