Fatigue Life of 7475-T7351 Aluminum After Local Severe Plastic Deformation Caused by Machining

The fatigue properties of thermo-mechanically treated and machined aluminum alloy 7475-T7351 have been studied. The applied advanced machining strategy induced intensive plastic deformation on the machined surface under defined cutting conditions. Therefore, a detailed study of 3D surface topography was performed. Advanced characterization of the material structure and electron back scattered diffraction mapping of selected chemical phases were performed, as well as energy dispersive X-ray analysis of the surface. Advanced mechanical properties of the material were investigated in situ with a scanning electron microscope that was equipped with a unique tensile fixture. The fatigue results confirmed an evident dispersion of the data, but the mechanism of crack nucleation was established. Fracture surface analysis showed that the cracks nucleated at the brittle secondary particles dispersed in the material matrix. The surface topography of samples that had been machined in wide range of cutting/deformation conditions by milling has not proved to be a decisive factor in terms of the fatigue behavior. The incoherent interface and decohesion between the alumina matrix and the brittle secondary phases proved to significantly affect the ultimate strength of the material. Tool engagement also affected the fatigue resistance of the material.


Introduction
Aluminum and its alloys are used in a wide range of industrial applications. Duralumins, are employed in the aerospace industry. Due to an ideal combination of low density, high strength, good corrosion resistance, and high resistance to fatigue crack propagation, these special alloys take precedence over other structural materials [1]. Aluminum alloys of the 2000 and 7000 series are widely used for primary and secondary aircraft structures, such as frames, spares, and ribs, where any damage has a crucial impact on safety.
The high strength Al-Zn-Mg-Cu 7475 is an alloy with controlled toughness made in the form of sheets and plates, that has an ideal combination of high strength, good fracture toughness, and resistance to fatigue crack propagation. The 7475 alloy has almost 40% greater fracture toughness than the previous version, 7075 [2]. This progress in mechanical properties is a result of the reduction of the content of iron, silicon, and magnesium, and application of thermo-mechanical and heat treatments which achieve a refined grain size [3]. The 7475 alloy, in the form of plates, is usually available in different tempered conditions such as T651, T7351, and T7651 [4]. However, among modifying the chemical composition of the alloys, the mechanical properties of Al-based alloys can also be enhanced by elevated temperature, which allowed relaxation of imposed residual stresses. However, a study of progressive milling technology on the surface topography and fatigue life of aluminum alloy 7475 of Piska et al. [20] showed that in the case of the presence of material inclusions or secondary phases larger than standard topography parameters (such as average roughness of the profile, maximum depth of the valley of the roughness profile, and others), the effect of the surface topography is usually suppressed.
Regarding ever-increasing safety requirements, it is necessary to carefully analyze the effect of the surface quality, including material structure and surface topography together with residual stresses and severe plastic deformation, imposed by the machining process before releasing components into operation.
This study is, therefore, focused on the influence of the different cutting conditions and tool inclination of the face milling strategy applied on the bottom wing panel made of alloy 7475-T7351 with regard to its fatigue life during operational use. The main goal of this study is therefore to define the milling condition range that allows maintaining the optimum balance between the productivity of the production process, the quality of machined surface, and the required fatigue properties.

Material
The aluminum alloy 7475-T7351 in the form of 70 mm thick plates was used in this study. The heat treatment with designation T7351 denotes solution heat treatment at 470 • C, water quenching, controlled stretching, and artificial ageing (over-aged in two stages: first at 121 • C for 25 h, second at 163 • C for a period of 24-30 h). The average chemical composition of the alloy is presented in Table 1 and its basic mechanical properties are shown in Table 2. Table 1. Chemical composition of the 7475-T7351 alloy (in wt.%) [4].  Table 2. Mechanical properties of the 7475-T7351 alloy [4]. The electron back scattered diffraction mapping (EBSD) study showed a heavily deformed structure with high anisotropy and texture of the grains (and very fine subgrains), as shown in Figure 1. During STEM (Scanning Transmission Electron Microscopy) lamella analysis, three different secondary phases have been observed in the material matrix, as indicated in Figure 2. Coarse intermetallic particles Al-Cu-Fe (possibly Al7Cu2Fe) [21,22] and Al-Cr-Fe-Cu-Si in the range from 2 μm up to 20 μm were formed during solidification phase. Precipitated Al-Fe-Si and Al-Mg-Cr dispersoids (possibly Al12Fe3Si; Al12Mg2Cr) were formed by solid state precipitation in the grain boundaries. Third, observed secondary phases can be described as fine metastable precipitates in the During STEM (Scanning Transmission Electron Microscopy) lamella analysis, three different secondary phases have been observed in the material matrix, as indicated in Figure 2. Coarse intermetallic particles Al-Cu-Fe (possibly Al 7 Cu 2 Fe) [21,22] and Al-Cr-Fe-Cu-Si in the range from 2 µm up to 20 µm were formed during solidification phase. Precipitated Al-Fe-Si and Al-Mg-Cr dispersoids (possibly Al 12 Fe 3 Si; Al 12 Mg 2 Cr) were formed by solid state precipitation in the grain boundaries. Third, observed secondary phases can be described as fine metastable precipitates in the material matrix (sizes from 2 nm up to 0.6 µm) and these are responsible for strengthening of the alloy (via GP,ή or η) [22,23]. During STEM (Scanning Transmission Electron Microscopy) lamella analysis, three different secondary phases have been observed in the material matrix, as indicated in Figure 2. Coarse intermetallic particles Al-Cu-Fe (possibly Al7Cu2Fe) [21,22] and Al-Cr-Fe-Cu-Si in the range from 2 μm up to 20 μm were formed during solidification phase. Precipitated Al-Fe-Si and Al-Mg-Cr dispersoids (possibly Al12Fe3Si; Al12Mg2Cr) were formed by solid state precipitation in the grain boundaries. Third, observed secondary phases can be described as fine metastable precipitates in the material matrix (sizes from 2 nm up to 0.6 μm) and these are responsible for strengthening of the alloy (via GP, ή or η) [22,23]. Energy dispersive X-ray spectroscopy (EDX) was used for elemental analysis of the large intermetallic particles, as shown in Figure 3. Energy dispersive X-ray spectroscopy (EDX) was used for elemental analysis of the large intermetallic particles, as shown in Figure 3.

Tool Geometry
The end-milling whole carbide tool ⌀16×55-115 mm JHF 980 Special provided by SECO Tools company, with (Ti, Al)N coating was used for advanced high feed face milling of the specimens. To exclude any potential impact of inaccurate tool geometry, optical 3D tool geometry analysis was

Tool Geometry
The end-milling whole carbide tool 16×55-115 mm JHF 980 Special provided by SECO Tools company, with (Ti, Al)N coating was used for advanced high feed face milling of the specimens. To exclude any potential impact of inaccurate tool geometry, optical 3D tool geometry analysis was performed using a special software subprogram of the ALICONA-IF G5 optical microscope called "Alicona Edge Master". The principle of this analysis was a gradual positioning of the reference plane perpendicular to the cutting edges of the tool. The measured results of the intersection of the reference plane with the cutting edges were statistically processed to obtain final results of the mean radius of the mean cutting edges and information about the mean cutting angles.

Surface Topography Analysis
Complex measurement of the surface topography was performed on a set of samples machined by different cutting parameters of face milling, as presented in Table 3, with the tool positioned perpendicular to the machined surface. Measurement of the surface topography for a set of samples with tool inclination of 1 • was also performed. The high-resolution optical microscope ALICONA IF-G5 was used for analysis of roughness parameters ("R"), waviness parameters ("W"), Firestone-Abbott parameters, and other advanced 3D surface texture parameters ("S"). The measurement methodology was based on the combination of the small depth of focus of the optical system with vertical scanning. In order to perform complex detection of the surface, the high-precision optics moved vertically along the optical axis and continuously captured data from the surface. A corresponding algorithm converted the acquired sensor data into 3D information and true colour images with full depth of field [24]. Nonmeasured points in the datasets were not taken into consideration for further processing or for calculation of corresponding parameters due their low ratio (flat surfaces of samples, good fits of data with the Gaussian distribution, very low occurrence of nonmeasured points in the whole dataset). The measurement methodology was in accordance with the standard EN ISO 25178-606 [25]. Cutting experiments for various cutting speeds and cutting feeds were carried out with a five-axis milling center MCV 1210/Sinumerik 840D. A stationary KISTLER 957B/SW dynamometer was used for measurement of the force loading during high feed face milling for the different cutting conditions, as presented in Table 4. The results have been analyzed with DynoWare software (type 2825A, Kistler, Wintherthur, Switzerland), where mean values of the maximal instantaneous force loading in the X, Y, and Z directions were used for graphical determination of the resultant force F 1M and its vector decomposition to the cutting force F C and the force perpendicular F CN . The cutting force and non-deformed chip cross section A D was used for calculation of the specific cutting energy k c for a given material characterized by the constants c o , the axial depth of cut a p , the radial with of cut a e , a angular tooth engagement φ and the effect of chip thickness on specific force loading expressed with the parameter mc: The basic model of continuous chip formation and the individual parameters for shear deformation and rate of the deformation can be seen in Figure 4.  The angle of the shear plane  is defined as where the orthogonal cutting angle o is sum of the orthogonal flank angle and orthogonal cutting edge angle o, i.e. o = o + o, and ⋀ means the chip thickness coefficient, which is expressed as The shear deformation in the primary zone can be derived as function of shear angle  and orthogonal rake angle o and the rate of shear deformation sequentially as The average thickness Ts and thickness hDC of the material lamella can be measured and calculated statistically by electron microscopy, as seen in Figure 4, and the orthogonal rake angle γo can be measured with the Alicona G5 microscope. The parameter hD corresponds to the feed per tooth.

Fatigue Testing and Frature Surface Analysis
The objective of the fatigue testing was to examine the influence of the defined cutting conditions of the face milling on the fatigue life, as summarized in Table 5. The effect of tool inclination of 1° has The plastic flow of the material is defined with the condition of a constant volume of machined material V passing through the first deformation zone and converted to the chip: where A D is the cross section of the undeformed material entering with cutting speed v c and A DC is the cross section of the material converted to a chip, leaving with speed of chip v ch . The other variables can be understood according to Figure 4a-f. The angle of the shear plane ϕ is defined as where the orthogonal cutting angle δ o is sum of the orthogonal flank angle and orthogonal cutting edge angle β o , i.e., δ o = α o + β o, and Λ means the chip thickness coefficient, which is expressed as The shear deformation γ in the primary zone can be derived as function of shear angle ϕ and orthogonal rake angle γ o γ = cos γ o sinφcos(φ − γ o ) (6) and the rate of shear deformation sequentially as The average thickness T s and thickness h DC of the material lamella can be measured and calculated statistically by electron microscopy, as seen in Figure 4, and the orthogonal rake angle γ o can be measured with the Alicona G5 microscope. The parameter h D corresponds to the feed per tooth.

Fatigue Testing and Frature Surface Analysis
The objective of the fatigue testing was to examine the influence of the defined cutting conditions of the face milling on the fatigue life, as summarized in Table 5. The effect of tool inclination of 1 • has also been examined. The geometry of the fatigue specimens was chosen to achieve the best match with final operational use of the bottom wing panel. The main criteria for appropriate specimen geometry are defined as follows: • Specimen must allow performance analysis of the effect of the high feed face milling on the fatigue life. Therefore, the flat specimen with largest possible functional area must be chosen.

•
Specimen must allow simulation of the tensile cyclic loading during operation.
All specimens were oriented in the L-T direction of the rolled plate. Specimens were machined at the five-axis MCV 1210/Sinumerik 840D milling center (TAJMAC ZPS, share company, Zlin, Czech Republic/Siemens AG, Erlangen, Germany). Specimens were specially protected against bending or torsion during machining, and excess heating was limited by use of CIMSTAR 597 coolant (Cimcool Industrial Products B.V., Vlaardingen, the Netherlands) of 10% volume concentration, 20 bar pressure, and 20 L/min flow rate. All functional areas of the specimens have been protected against scratches, all sharp edges rounded to 0.3 mm radius, and all other stress concentrators have been removed.
Fatigue testing has been performed at special axial testing machine BISS and parameters of the testing are defined as follows: • Fluctuating tensile cycle with stress ratio, R = 0.1. The source of fatigue crack nucleation was examined with the scanning electron microscope (TESCAN ORSAY HOLDING share company, Brno, Czech Republic) TESCAN MIRA 3 operating in both secondary and backscattered electron mode. The fatigue crack initiation and propagation mechanism and the integrity of the adjacent surfaces were investigated.

In Situ Testing
A specimen with special geometry was designed for in situ tensile mechanical testing. Profiles of the specimen were cut by EDM wire cutting (Electrical Discharge Machining) and polished. Flat functional surface areas were face milled with a special high feed end-mill (SECO tool JHF 980 Special, f z = 0.05 mm, v c = 200 m/min, a p = 0.2 mm), and all sharp edges were rounded and polished.
Testing was performed on a special in situ tensile stage MT1000 made by NewTec (10 kN, a tensile stage) and all analyses were carried out with the SEM TESCAN MIRA 3, equipped with the NewTec SoftStrain software (version 1, NEWTEC, Nîmes, France), as shown in Figure 5.
 Frequency, f = 10 Hz.  Stress levels: 180 MPa, 220 MPa, 250 MPa, and 300 MPa. The source of fatigue crack nucleation was examined with the scanning electron microscope (TESCAN ORSAY HOLDING share company, Brno, Czech Republic) TESCAN MIRA 3 operating in both secondary and backscattered electron mode. The fatigue crack initiation and propagation mechanism and the integrity of the adjacent surfaces were investigated.

In Situ Testing
A specimen with special geometry was designed for in situ tensile mechanical testing. Profiles of the specimen were cut by EDM wire cutting (Electrical Discharge Machining) and polished. Flat functional surface areas were face milled with a special high feed end-mill (SECO tool JHF 980 Special, fz = 0.05 mm, vc = 200 m/min, ap = 0.2 mm), and all sharp edges were rounded and polished.
Testing was performed on a special in situ tensile stage MT1000 made by NewTec (10 kN, a tensile stage) and all analyses were carried out with the SEM TESCAN MIRA 3, equipped with the NewTec SoftStrain software (version 1, NEWTEC, Nîmes, France), as shown in Figure 5.

In Situ Tensile Testing
The main scope of the in situ tensile testing was to observe the crack initiation and propagation mechanism in the 7475-T7351 alloy. Two main analyses were performed: (a) Observation of the entire functional area of the specimen; and (b) detailed observation of selected intermetallic particles located at the free machined surface of the specimen. Engineering strain distribution under tensile loading in selected particle was analyzed by DIC (Digital Image Correlation) in MERCURY real-time tracking software.

In Situ Cyclic Testing
The aim of the in situ tensile cyclic loading was to simulate cyclic loading under the operation mode and to observe crack nucleation and short crack propagation. Detailed observation of twenty selected large intermetallic particles was performed in parallel. Parameters of the fatigue loading were defined as follow:

Tool Geometry
The geometry of the tool and profile roughness of the cutting edges were analyzed and the results are presented in Tables 6 and 7. All parameters complied with manufacturer's specifications. The standard deviations were in the range of 3-4% of the mean values. Table 6. SECO end-milling tool 16×55-115 mm JHF 980 Special geometry.

Cutting Edge
Cutting Edge Radius r n (µm) Ra-average roughness of the profile, Rq-root-mean-square roughness of the profile, Rz-mean peak to valley height of the roughness profile, Rp-maximum peak height of the roughness profile, Rv-maximum valley depth of the roughness profile.

Surface Topography Analysis
The effect of the defined cutting conditions and tool inclination on R-parameters of the surface topography was evident. For the set of samples machined with tool positioning perpendicular to the machined surface, a digression of the average values of the profile roughness parameter (measured perpendicularly to the cutting speed, along the feed speed and longitudinal axis of the samples) was observed: • The average roughness (Ra) and root-mean-square roughness (Rq) dropped by 35% for the highest cutting speed (v c = 400 m.min −1 ).

•
No linear function was observed between the cutting speed and the profile roughness parameters, as indicated in Figure 6. • Standard deviations of the repeated measurements varied between 5% and 8% of the average values for all measurements and conditions.
On the other hand, the increase of the profile roughness parameters was observed with the increase of the feed speed (increase of the feed per tooth, f z , from 0.05 to 0.90 mm). Some examples of the profile roughness values increasing while increasing the feed speed are mentioned below: • The average roughness (Ra) increased from 2.71 to 4.30 µm (an increase of 37%), as indicated in Figure 7.

•
The maximum peak to valley height of roughness profile (Rt) increased from 28.59 to 33.78 µm (an increase of 15%), as indicated in Figure 7.
• The maximum valley depth of roughness profile (Rv) increased from 13.23 to 16.92 µm (an increase of 22%), as indicated in Figure 7.
perpendicularly to the cutting speed, along the feed speed and longitudinal axis of the samples) was observed:  The average roughness (Ra) and root-mean-square roughness (Rq) dropped by 35% for the highest cutting speed (vc = 400 m.min −1 ).  No linear function was observed between the cutting speed and the profile roughness parameters, as indicated in Figure 6.  Standard deviations of the repeated measurements varied between 5% and 8% of the average values for all measurements and conditions. No linear function was observed between the cutting speed and the waviness parameters, or between the feed speed and the waviness parameters (R-squared parameter varied in the range from 0.1 to 0.7 for different waviness parameters), as shown in Figures 8 and 9.
The surface parameters under different cutting parameters of the face milling were examined (with tool positioning perpendicular to the machined surface). The results of the measurement can be summarized as follows: • The increase of the feed speed (increase of the feed per tooth from 0.05 to 0.90 mm) caused an increase of surface topography parameters, as indicated in Figure 10.

•
The average height of the selected area (Sa) increased from 2.60 to 5.30 µm.

•
The root-mean-square height of the selected area (Sq) increased from 3.40 to 6.43 µm.

•
The maximum valley depth of the selected area (Sv) increased from 26.77 to 35.35 µm.

•
No statistically significant linear function (probability 95%) was found between the cutting speed and the 3D surface topography parameters. perpendicularly to the cutting speed, along the feed speed and longitudinal axis of the samples) was observed:  The average roughness (Ra) and root-mean-square roughness (Rq) dropped by 35% for the highest cutting speed (vc = 400 m.min −1 ).  No linear function was observed between the cutting speed and the profile roughness parameters, as indicated in Figure 6.  Standard deviations of the repeated measurements varied between 5% and 8% of the average values for all measurements and conditions. No linear function was observed between the cutting speed and the waviness parameters, or between the feed speed and the waviness parameters (R-squared parameter varied in the range from 0.1 to 0.7 for different waviness parameters), as shown in Figure 8 and Figure 9.  No linear function was observed between the cutting speed and the waviness parameters, or between the feed speed and the waviness parameters (R-squared parameter varied in the range from 0.1 to 0.7 for different waviness parameters), as shown in Figure 8 and Figure 9.    The machined surface can by described by the Firestone-Abbott curve, which indicates the percentage of the material of the profile elements at a defined height relative to the evaluation profile length (R) or surface area (S). This specific surface criterion is characterized by several parameters. The parameters of the core roughness depth (Rk and Sk) indicate the volume of the material above the core material which can be worn during operational use. The parameters of reduced peak height (Rpk and Spk) describe the mean height of peaks above the core material. Furthermore, the parameters of reduced peak height (Rpk and Spk) express the amount of the material that will be removed during the initial operational wearing process. The parameters of reduced valley height (Rvk and Svk) describe the mean depth of the valleys below the core material. Therefore, Rvk and Svk parameters indicate the ability of the machined surface to retain liquids. The parameter Rmr1 indicates the fraction of the surface which consists of peaks above the core material, and the parameter Rmr2 indicates the fraction of the surface which will carry the load [28]. An examples of the Firestone-Abbott curve are presented in Figures 11 and 12. The machined surface can by described by the Firestone-Abbott curve, which indicates the percentage of the material of the profile elements at a defined height relative to the evaluation profile length (R) or surface area (S). This specific surface criterion is characterized by several parameters. The parameters of the core roughness depth (Rk and Sk) indicate the volume of the material above the core material which can be worn during operational use. The parameters of reduced peak height (Rpk and Spk) describe the mean height of peaks above the core material. Furthermore, the parameters of reduced peak height (Rpk and Spk) express the amount of the material that will be removed during the initial operational wearing process. The parameters of reduced valley height (Rvk and Svk) describe the mean depth of the valleys below the core material. Therefore, Rvk and Svk parameters indicate the ability of the machined surface to retain liquids. The parameter Rmr1 indicates the fraction of the surface which consists of peaks above the core material, and the parameter Rmr2 indicates the fraction of the surface which will carry the load [28]. An examples of the Firestone-Abbott curve are presented in Figure 11 and in Figure 12.   Figure 13. Therefore, less material will be removed during the initial wearing process of operational use while implementing higher cutting speeds.  Rk, Sk and Rpk, Spk increase linearly with increasing feed speed (while increasing feed per tooth from 0.05 mm to 0.90 mm, Rpk increased by 16% and Spk increased by 26%), as shown in Figure 14. Therefore, a higher volume of the material will be removed during the initial wearing process of operational use when implementing higher feed speeds. The machined surface can by described by the Firestone-Abbott curve, which indicates the percentage of the material of the profile elements at a defined height relative to the evaluation profile length (R) or surface area (S). This specific surface criterion is characterized by several parameters. The parameters of the core roughness depth (Rk and Sk) indicate the volume of the material above the core material which can be worn during operational use. The parameters of reduced peak height (Rpk and Spk) describe the mean height of peaks above the core material. Furthermore, the parameters of reduced peak height (Rpk and Spk) express the amount of the material that will be removed during the initial operational wearing process. The parameters of reduced valley height (Rvk and Svk) describe the mean depth of the valleys below the core material. Therefore, Rvk and Svk parameters indicate the ability of the machined surface to retain liquids. The parameter Rmr1 indicates the fraction of the surface which consists of peaks above the core material, and the parameter Rmr2 indicates the fraction of the surface which will carry the load [28]. An examples of the Firestone-Abbott curve are presented in Figure 11 and in Figure 12.   Figure 13. Therefore, less material will be removed during the initial wearing process of operational use while implementing higher cutting speeds.  Rk, Sk and Rpk, Spk increase linearly with increasing feed speed (while increasing feed per tooth from 0.05 mm to 0.90 mm, Rpk increased by 16% and Spk increased by 26%), as shown in Figure 14. Therefore, a higher volume of the material will be removed during the initial wearing process of operational use when implementing higher feed speeds. The machined surface under the defined cutting conditions of the face milling showed the following results:

•
Rpk and Spk decrease linearly with the increase of the cutting speed (while increasing the cutting speed from 90 m.min −1 to 400 m.min −1 , Rpk dropped by approximately 58%, and Spk dropped by 25%), as shown in Figure 13. Therefore, less material will be removed during the initial wearing process of operational use while implementing higher cutting speeds. • Rk, Sk and Rpk, Spk increase linearly with increasing feed speed (while increasing feed per tooth from 0.05 mm to 0.90 mm, Rpk increased by 16% and Spk increased by 26%), as shown in Figure 14. Therefore, a higher volume of the material will be removed during the initial wearing process of operational use when implementing higher feed speeds. Analysis of the effect of the tool inclination on the surface parameters can be summarized as follows: • Roughness parameters are similar for both types of strategies; however, a greater increase of roughness parameters while using higher feed speeds was observed for samples machined with tool inclined by 1 • . • Roughness parameters were affected by the feed per tooth, not by the tested cutting speeds.

•
The effect of the tool inclination cannot be compared properly if no 3D surface topography parameters are used (see example of two machined samples with different tool positioning in Figure 15).

•
The average height of the selected area (Sa) showed higher values for samples machined with the tool inclined by 1 • (by 50-70%) compared to the results with perpendicular tool positioning, which were in the range of 2.00 to 5.30 µm for all tested conditions. Similar relations were found for Sq, Ssk, and Sku.

•
The texture aspect ratios (Str) for the machined surface did not present significant changes (the differences for the same cutting conditions were about 15-20%).

•
Roughness parameters are not sufficient for comparison of the complex topography of the surface after machining because of their inhomogeneity and sensitivity to the measured place, which affect skewness, kurtosis, etc. • Study of the surface parameters revealed that no crucial variable of surface topography was linked to the fatigue results of the studied material.
(b) Figure 13. Mean height of peaks above the core material for different cutting speeds (at feed per tooth fz = 0.05 mm) (a) Rpk -mean height of peaks above the core material for a profile; (b) Spk -mean height of peaks above the core material for a selected area. Analysis of the effect of the tool inclination on the surface parameters can be summarized as follows:  Roughness parameters are similar for both types of strategies; however, a greater increase of roughness parameters while using higher feed speeds was observed for samples machined with tool inclined by 1°.  Roughness parameters were affected by the feed per tooth, not by the tested cutting speeds.  The effect of the tool inclination cannot be compared properly if no 3D surface topography parameters are used (see example of two machined samples with different tool positioning in Figure 15).  The average height of the selected area (Sa) showed higher values for samples machined with the tool inclined by 1° (by 50-70%) compared to the results with perpendicular tool positioning, which were in the range of 2.00 to 5.30 μm for all tested conditions. Similar relations were found for Sq, Ssk, and Sku.  The texture aspect ratios (Str) for the machined surface did not present significant changes (the differences for the same cutting conditions were about 15-20%).  Roughness parameters are not sufficient for comparison of the complex topography of the surface after machining because of their inhomogeneity and sensitivity to the measured place, which affect skewness, kurtosis, etc.  Study of the surface parameters revealed that no crucial variable of surface topography was linked to the fatigue results of the studied material.

Force Loading Analysis of High Feed Face Milling and Induced Severe Plastic Deformation
The parameters of primary shear deformation and its rates were very high, as can be seen in Figure 16. Primary shear deformation decreases with increase of the cutting speed as well as with increase of the feed speed (feed per tooth).

Force Loading Analysis of High Feed Face Milling and Induced Severe Plastic Deformation
The parameters of primary shear deformation and its rates were very high, as can be seen in Figure 16. Primary shear deformation decreases with increase of the cutting speed as well as with increase of the feed speed (feed per tooth).

Force Loading Analysis of High Feed Face Milling and Induced Severe Plastic Deformation
The parameters of primary shear deformation and its rates were very high, as can be seen in Figure 16. Primary shear deformation decreases with increase of the cutting speed as well as with increase of the feed speed (feed per tooth). An explanation of this phenomenon is not easy and other works dealing with dislocations are ongoing. Meanwhile, we tentatively propose that there exists a small region over which there is a sudden proliferation of high angle boundaries in the microstructure of the material as it is deformed into the chip [29].

Fatigue Testing and Frature Surface Analysis
The effect of the milling parameters on the fatigue life was striking even if the surface topography after machining was not the key factor affecting the fatigue crack nucleation.
The important decrease of the fatigue life of specimens machined with higher feed rates while keeping the same cutting speed (an increase from feed per tooth fz = 0.05 mm to fz = 0.90 mm) can be seen in Figure 17. This decrease of the fatigue life may be caused by the severe plastic deformation An explanation of this phenomenon is not easy and other works dealing with dislocations are ongoing. Meanwhile, we tentatively propose that there exists a small region over which there is a sudden proliferation of high angle boundaries in the microstructure of the material as it is deformed into the chip [29].

Fatigue Testing and Frature Surface Analysis
The effect of the milling parameters on the fatigue life was striking even if the surface topography after machining was not the key factor affecting the fatigue crack nucleation.
The important decrease of the fatigue life of specimens machined with higher feed rates while keeping the same cutting speed (an increase from feed per tooth f z = 0.05 mm to f z = 0.90 mm) can be seen in Figure  The effect of the higher cutting speed on the fatigue life of the specimen is evident in the case of combination with lower feed speed, as presented in Figure 18. Use of higher feed speed increases the fatigue life for both low and high cycle fatigue modes. However, this effect is suppressed by combination with a high feed cutting strategy. The effect of the higher cutting speed on the fatigue life of the specimen is evident in the case of combination with lower feed speed, as presented in Figure 18. Use of higher feed speed increases the fatigue life for both low and high cycle fatigue modes. However, this effect is suppressed by combination with a high feed cutting strategy.
Slight inclination of the cutting tool (only 1 • ) resulted in a 29-64% reduction of total cycles (for specimens machined with the combination of the cutting speed v c = 90 m.min -1 and different feed speeds, as indicated in Figure 19).
Regardless of the cutting conditions, the fatigue cracks were always initiated in the large intermetallic particles which occurred in different morphologies (as large particles, elongated particles, or clusters of intermetallic particles). Fatigue cracks were mostly initiated in the intermetallic particles located in the vicinity of the machined surface regardless of the stress level. An example of the fracture surface is presented in Figure 20. The effect of the higher cutting speed on the fatigue life of the specimen is evident in the case of combination with lower feed speed, as presented in Figure 18. Use of higher feed speed increases the fatigue life for both low and high cycle fatigue modes. However, this effect is suppressed by combination with a high feed cutting strategy.  Slight inclination of the cutting tool (only 1°) resulted in a 29-64% reduction of total cycles (for specimens machined with the combination of the cutting speed vc = 90 m.min -1 and different feed speeds, as indicated in Figure 19). Regardless of the cutting conditions, the fatigue cracks were always initiated in the large intermetallic particles which occurred in different morphologies (as large particles, elongated particles, or clusters of intermetallic particles). Fatigue cracks were mostly initiated in the intermetallic particles located in the vicinity of the machined surface regardless of the stress level. An example of the fracture surface is presented in Figure 20. Regardless of the cutting conditions, the fatigue cracks were always initiated in the large intermetallic particles which occurred in different morphologies (as large particles, elongated particles, or clusters of intermetallic particles). Fatigue cracks were mostly initiated in the intermetallic particles located in the vicinity of the machined surface regardless of the stress level. An example of the fracture surface is presented in Figure 20.

In Situ Tensile Testing
Crack propagation from large intermetallic particles to the material matrix was observed even before reaching the tensile strength limit (484 MPa), as presented in Figure 21. Large intermetallic particles were the main source of the local stress concentration regardless of the severe plastic deformation caused by the machining process. Cracks propagated at the angle of 45 • to the direction of the maximum shear stresses.

In Situ Tensile Testing
Crack propagation from large intermetallic particles to the material matrix was observed even before reaching the tensile strength limit (484 MPa), as presented in Figure 21. Large intermetallic particles were the main source of the local stress concentration regardless of the severe plastic deformation caused by the machining process. Cracks propagated at the angle of 45° to the direction of the maximum shear stresses. The evolution of the engineering strain distribution under tensile loading is shown in Figure 22. The average engineering strain at the area of the intermetallic particle at yield strength was in the range of 0.80% to 0.90%, and upon reaching the maximal engineering strain (1.15-1.20%), local crack initiation was observed. Figure 21. Observation of particle cracking: (a) Intermetallic particle before tensile loading; (b) Intermetallic particle at yield strength (415-419 MPa); (c) Intermetallic particle at tensile strength limit (484 MPa). The loading axis was horizontal.
The evolution of the engineering strain distribution under tensile loading is shown in Figure 22. The average engineering strain at the area of the intermetallic particle at yield strength was in the range of 0.80% to 0.90%, and upon reaching the maximal engineering strain (1.15-1.20%), local crack initiation was observed.
The evolution of the engineering strain distribution under tensile loading is shown in Figure 22. The average engineering strain at the area of the intermetallic particle at yield strength was in the range of 0.80% to 0.90%, and upon reaching the maximal engineering strain (1.15-1.20%), local crack initiation was observed. Figure 22. DIC analysis of the engineering strain distribution in the intermetallic particle during tensile loading: at the beginning of the tensile loading, at the yield strength, and at the tensile strength limit.

In Situ Cyclic Testing
During in situ cyclic testing, initial local fatigue cracks were observed in some intermetallic particles before the fatigue life (1000 cycles) was reached, regardless of severe plastic deformation induced by the milling process. Local fatigue cracks were initiated in the core of the intermetallic particles, and with the rising number of cycles, the fatigue cracks propagated locally to the boundary of the intermetallic particles and the material matrix. Fatigue testing was interrupted at the fatigue level of 6000 cycles. Up to this number of fatigue cycles, the short fatigue cracks remained inside the intermetallic particles and did not propagate further. DIC analysis confirmed that once the local engineering strain reached values of 1.15-1.20% (at 1000 cycles, in this case), the local fatigue cracks were initiated, as demonstrated in Figure 23. . DIC analysis of the engineering strain distribution in the intermetallic particle during tensile loading: at the beginning of the tensile loading, at the yield strength, and at the tensile strength limit.

In Situ Cyclic Testing
During in situ cyclic testing, initial local fatigue cracks were observed in some intermetallic particles before the fatigue life (1000 cycles) was reached, regardless of severe plastic deformation induced by the milling process. Local fatigue cracks were initiated in the core of the intermetallic particles, and with the rising number of cycles, the fatigue cracks propagated locally to the boundary of the intermetallic particles and the material matrix. Fatigue testing was interrupted at the fatigue level of 6000 cycles. Up to this number of fatigue cycles, the short fatigue cracks remained inside the intermetallic particles and did not propagate further. DIC analysis confirmed that once the local engineering strain reached values of 1.15-1.20% (at 1000 cycles, in this case), the local fatigue cracks were initiated, as demonstrated in Figure 23.

Discussion
The observations from the experimental machining, surface analyses, and fatigue testing confirm similar results as those of Ojolo et al. [17] and Novovic et al. [18].
Surface topography analysis confirmed that the roughness parameters increase with the increase

Discussion
The observations from the experimental machining, surface analyses, and fatigue testing confirm similar results as those of Ojolo et al. [17] and Novovic et al. [18].
Surface topography analysis confirmed that the roughness parameters increase with the increase of the feed speed (feed per tooth). The increase of the cutting speed caused a decrease of the surface roughness parameters. This result partially confirms the observation of Ojolo et al. [17].
The increase of the feed speed increases surface topography parameters such as average height of the selected area (Sa), root-mean-square height of the selected area (Sq), or maximum valley depth of the selected area (Sv). The roughness parameters were found to be similar for both strategies (perpendicular and inclined by 1 • ); however, a greater increase of roughness parameters was observed while using higher feed speeds for samples machined with the tool inclined by 1 • . The average height of the selected area (Sa) showed higher values for samples machined with the tool inclined by 1 • . This parameter was not adequate, considering that maximum valley depth of the selected area (Sv) was higher for samples machined by a tool positioned perpendicularly to the machined surface.
The trends of specific cutting force and shear deformation confirm a reduction of plastic deformation with increasing cutting speed, but more intensive deformation with reduction of feed per tooth. In other words, the intensity of plastic deformation is higher for shallow cuts and higher cutting speeds.
The highest fatigue resistance was observed at samples machined with the highest cutting speed (v c =200 m.min −1 ) and lowest feed per tooth (f z = 0.05 mm). A decrease of the fatigue life of specimens machined with higher feed rates while keeping the same cutting speed was observed (increase from feed per tooth f z = 0.05 mm to f z = 0.90 mm). This decrease of the fatigue life may be caused by the severe plastic deformation achieved in the smallest chip cross sections and machined at the highest cutting speeds (v c = 200 m.min −1 ).
Slight inclination of the cutting tool (1 • ) resulted in reduction of the total cycles for specimens machined with cutting speed v c = 90 m.min −1 and different feed speeds. This reduction may be caused by plastic deformation caused by teeth not engaged in the cut, as in the case of face milling perpendicular to the machined surface. This plastic deformation may impose beneficial compressive residual stresses into the machined surface and thus increase the fatigue life.
Therefore, the cutting conditions can affect the material removal rate, but a more serious impact can be seen in terms of the surface quality and the resistance to mechanical loading. The effect of inclusions is very serious, and materials used for dynamic loading should be carefully analyzed not only in view the surface integrity, but also considering the occurrence of the phases, which confirms the results of Piska et al. [20]. The effect of material hardening and thermal softening when cutting should be studied further in terms of the density and arrangement of dislocations, stacking fault energy, and other atomic hardening or softening mechanisms.

Conclusions
The application of very advanced laboratory facilities yielded the following results: • The 7475-T7351 aluminum material was suitable for dynamic mechanical loading with good machinability when milling with the special monolithic cutter SECO JHF 980 Special.

•
The quality of surface parameters and fatigue resistance improved when higher cutting speeds (200 m/min) and low feeds per tooth (0.05 mm) were used, and extreme shear deformation (γ = 2.5) and deformation rates (1.2 × 10 5 s −1 ) were achieved.

•
The measured values of Sa correlated with the Ra parameters in the trends according to the cutting conditions and proved to be featureless multiplications of the Ra values for all tested conditions. Nevertheless, no significant variable of the surface topography linked to the fatigue results was found.
• The fatigue resistance of the samples machined with the standard perpendicular position of the tool to the machined surface (i.e. without any spindle inclination) was greater than the results for samples machined with the inclined tool. Therefore, surfaces with more complex topography seems to be beneficial; however, new studies with a material with minimal intermetallic inclusions are needed.

•
The crucial and decisive factor for crack nucleation can be seen in the coarse intermetallic inclusions (Al 7 Cu 2 Fe; Al-Cr-Fe-Cu-Si) in sizes from 2 µm up to 20 µm, suppressing the effect of the surface parameters after machining.

•
Further study of the dislocation mechanism responsible for deformation hardening and softening are required. Research relating surface layer depths and grain size are ongoing.