Experimental study of micromilling process and deburring electropolishing process on FeCo-based soft magnetic alloys

The micro-milling process was conducted on a precision three-axes machine tool. This machining tool is a computer numerical control (CNC) workstation specifically developed and customized for the microscale (Fig. 1). It can obtain drilling and milling of micrometric features. This machine can perform 3D XYZ movements with submicrometre precision. This CNC station can mill and drill using micrometric cutting tools. It also includes a USB microscope that allows recalibration and referencing procedure using visual feedback from the microscope which is installed on the machine in parallel to the spindle. The design of the micromilling machine tool is oriented to produce small structures. The station basically includes XYZ motion tables and a rigidly fixed spindle head (Fig. 1). Micro-milling CNC has three ultraprecision linear motor stages with built-in controllers and linear encoders provided by Zaber Technologies INC company. XY axes are moved by means of model X-LDM110C-AE54D12 (intended for horizontal displacements) and X-LRQ075AP-DE51C for the Z-axis of the station. XY tables have a travel of 110 mm with a positioning accuracy below 1 μm. Due to this small size, small axes with low travel and low moving masses could be implemented, resulting in high velocities and low energy consumption. The attachment of the workpiece material laminates is done by using a vacuum chuck. This permits to flatten the laminate and assure the parallelism. The spindle is a Sycotec 4033 AC-ER8 model of 420 W of power, with a maximum rotational speed of 100,000 rpm, a maximum torque of 8.4 Ncm and a run-out of 1 μm at a selected speed of 30,000 rpm. The spindle, as well as the spindle mount, is air-cooled. Microparticles and nanoparticles are removed by pressurized air. The micromilling station is protected by a transparent casing with which it is possible to carry out all the micro-milling and prevent dust contamination. The whole micro-milling tool is desktop-sized covering a total volume of 900 mm × 600 mm × 500 mm.

FeCo alloys are generally little ductile and hard to work. To enhance its machinability and stiffness, vanadium additive is included in the composition. With this additive, the magnetic performance is slightly decayed, the magnetic permeability is decreased, and the coercivity increases [55]. In the case of FeCo–2 V alloys, this decay can be down to 10% regarding the equiatomic FeCo [56], depending on the heat treatment. Three commercial FeCo–2 V alloys have been chosen: Vacodur49, Vacoflux48 and Hiperco50 depending on the laminate thickness. These three materials present a composition of 49% cobalt, 49% iron and 2% vanadium. These materials present the highest magnetic saturation and greater mechanical strength than other soft magnetic alloys. Selected FeCo-2 V alloys are purchased in cold rolled laminated, with thickness between 55 and 355 μm. They exhibit yield strength of 1150–1270 MPa, tensile strength of 1230–1310 MPa, a typical Poisson’s ratio of 0.29, Young’s modulus of 200–206 GPa, hardness between 342 and 380 HV, density of 8.12 g/cm³ and expansion coefficient 8.9–9.7 (20 to 200 °C) α [10⁻⁶/K].

Micro-milling was performed by MF208-0005–0008-M with 50-μm-diameter and MF208-0025–0038-M with 250-μm-diameter end mills pro metal, 2-flute stub with an helix angle of 28ª, fishtail and stub end-mill from company FrezyCNC Ingraph (Poland). The tool diameters selected are 50 µm as the minimum available and 250 µm the maximum that allows machining features below 100 µm. The tool material is solid submicrograin carbide. This material has a typical hardness of 1350–1600 HV, a transverse rupture strength of 3100–3650 MPa and a density of 14–14.5 g/m³. These tools were chosen because cutting-edge geometry virtually eliminates chatter at high feed rates. Spiral geometry ensures efficient debris removal, and flute geometry enhances rigidity for accurate, single-pass cutting at high chip loads. Furthermore, the material selected for tools is suitable to cut FeCo–2 V. The used tools are shown in Fig. 2.

There are several steps to execute before the milling process action. The spindle motor must be warmed-up before the first machining. Then, tool and material are manually attached to the CNC station. The preparation of the material is the most delicate step before milling. Flatness and tilting errors may appear during the attachment of the laminate if it is not carefully executed. This results in geometrical and finishing defects on the final machined part. After the laminate cutting, the workpiece material is flattened in a hydraulic press. The laminate is glued to a PVC substrate; this contributes to the rigidity of the laminate during machining. This attachment permits the milling of the ulterior release of the machined microparts. The complete attachment system is shown in Fig. 3. It includes two stainless steel case parts that press and fixes the laminate by using four screws tightening.

The design is programmed on CAD-CAM software, where machining processes can be translated to a G-CODE program which is read by the software of the machine. Spindle rotation velocity, feed rate, tool diameter, bulk material thickness, depth of each step and geometry are specified on the CAD-CAM software. After all the preparation steps are achieved, the machining process can start. The machining is done under a big drop of lubricant coolant to improve heat dissipation and the process of chip cutting. Finally, the workpiece material is removed and immersed in a soft solvent like acetone to eliminate the adhesive layer, separating the machined metallic plate or the micropart from de PVC glued substrate.

The micromilling systematic study includes different machining parameter values. The variation of the different parameters is listed in Table 1. The study is done by milling simple 1-mm straight grooves with single step and depth values to determine the optimal combination of machining parameters. The range of parameters such as feed or depth step have been selected after selecting the best trial-and-error results based on visual inspection and burr measurements. The reasons of sweeping these parameters are, mainly, to achieve values that allow minimum burr formation and minimum geometrical deviation (due to plasticization effects in work material). The selected speed of 30,000 rpm corresponds to the minimum amplitude of vibration measured in the XY plane of the spindle structure through accelerometers, although the speed of 60,000 rpm is also studied to test the effect of those vibrations.

The analysis of the milling results is done using an Olympus DSX1000 digital microscope that allows two-dimensional and three-dimensional (2D and 3D) measurements. This microscope is connected to a computer located next to the cabin to check the results. The microscope enables fast macro to microlevel view, and it also provides accurate measurements due to its telecentric optical system. This microscope provides information about the geometric and surface roughness values of the manufacturing results.

Once the 1-mm grooves are done at each different combination, measurements of resulting depth and width were done by profilometer, and geometric measurements were done by the confocal digital microscope. Several measurements were taken along each groove obtaining mean values and typical deviations for depth and width. For width measurements, a typical deviation is less than 2 μm for each groove, while it is less than 0.6 μm for depth of cut measurements. The excess of expected dimension is plotted in Fig. 4, for width, and in Fig. 5, for depth. The excess in width is calculated in respect to the tool diameter and the excess in depth of cut is the difference in μm with the programmed depth of cut in the G code program. Feed rate (f), tool diameter (D) and depth of cut (p) are considered. The rotation spindle speed is set up as n = 30,000 rpm for all the geometrical characterization experiments. The excess of width is assumed to come from small eccentricity of the spindle and from non-controlled vibrations. Excess of width is reduced; thus, we consider that feed speed and microcutting forces are adequate.

Regarding cut width, the feed rate of 0.48 mm/min presents the smaller deviations for all diameters and depth of cuts, it is particularly critical for the 50-μm-diameter tool at 4-μm depth of cut. It is visible that the excess in width is not much dependent on the feed for the 250-μm diameter tool in the range observed. It seems to be a depth threshold from where the width excess increases and becomes more dependent on the feed velocity. The set of parameters that more accuracy offers in 2D geometrical features is the 50-μm diameter tool at depth of cuts as small as 2 μm, and machining at feed velocity of 0.48 mm/min, with width excess less than 1% of the diameter.

Regarding depth of cut, small deviations on the depth of grooves have been obtained, and the measured values vary between 0.15 and 0.45 μm. The minimum value achieved for the 50-μm diameter tool has been found at 2 μm and 0.24, while for the 250 μm diameter too, it is found that at 16-μm depth and 0.48. There is not a clear sign on influence neither tool, depth of cut, nor feed rate. They all seem to be adequately selected. We assume this variation comes from variation in the spindle tool position and from the zero-tooling process.

Figures 6 and 7 show pictures of the machining achieved by the two tools on their optimal parameter configurations, respectively. Straight walls, flat machined surface, fine width accuracy, and a small amount of burr have been achieved in this milling process. This is the starting point for more complex 2D features in multiple depth of cuts milling processes.

Roughness is calculated by the microscope Olympus software; several parameters of roughness can be measured being the most representative Ra which is defined as Average or arithmetic average of profile height deviations from the mean line. Ra measurements of the studied grooves have been taken. In addition, the effect of runout at not-optimal spindle velocities for the machine on the machined surface roughness has been studied.

Figure 8 shows that for all cases, smaller depth of cut means a lower roughness level. The effect of feed rate increases drastically when increasing the size of the depth of cut. Ra does not show a relevant dependency on tool diameter in the range of 50–250 μm if the depth of cut is less than 8 μm and the feed rate is between 0.24 and 0.48 mm/min. With both tools, the roughness achievable is small enough to satisfy the mechanical requirements on complex MEMS assemblies. Figure 9 shows a topographical view of the machined surface of the minimum roughness achieved with the 50-µm tool; the Sa value of the marked area is 0.042, which is consistent with the Ra value shown in Fig. 5. Figure 10 shows the surface finish of the machined surface with the minimum roughness achieved of Ra = 0.04 μm. These values of roughness are small enough to achieve the required airgap thickness in microactuators such as micromotors, where airgaps around 10 µm are needed.

The vibration of the tool leads to worse surface finishes and worse geometrical accuracy as shown in Fig. 11. In Fig. 12, it is visible a poorer finish and a worse geometrical shape in comparison to the milling shown in Fig. 6.

For most electric machine applications, individual magnets are mounted in the rotor assembly. Rectangular plates, in radial flux applications, and circular sectors or segments, in axial flux applications, are the typical shapes needed for this kind of machines. One example of these desired shapes is the angular sector performed in FeCo-2 V alloy shown in Fig. 13. Adjusting the machining parameters to the optimized values for the 50-micron endmill, the shape has been performed in several cutting steps to achieve the desired depth of 50 μm. The machined cavity presents high-quality straight walls and a very reduced quantity of burr in the top edges; this is visible in the vertical slice profile plotted in Fig. 13 bottom. This microscope provides information about the geometric and surface roughness values of the manufacturing results.

This kind of geometry can be radially duplicated, as the machined yoke of rotor in Fig. 14, creating ferromagnetic structures where assembling and gluing separated micromagnets is possible. In addition, there is the possibility of cutting out the part by milling, after integrating permanent magnets. The repeatability of this and more complex features is quite stable. In the machined patron of Fig. 14, the grooves were done by a 100-μm milling tool. The width of the machined groove excesses an average of 12.9 μm, with a typical deviation of 2.18 μm. The diameter of tool entrance presents an average excess of 11.5 μm with a typical deviation of 3.83 μm.

Figure 15 shows another example of a micropart made in FeCo-2 V alloy. The circular micropart has been cut out. It can be used as back yoke in stator or rotor assemblies of axial flux micromotors. The micropart is glued to a PVC substrate while it is being machined. Afterwards, the substrate with glued FeCo-2 V alloy laminate and machined part is submerged in acetone. The glue is dissolved, and the part is released without residuals.

Other interesting shapes are solenoid, stator and transformer yokes. This consists of high aspect ratio components with circular or square section shapes. These components are winded with aluminium, silver or copper wires. When an electrical current flows through the conductor, the ferromagnetic yoke becomes an electromagnet.

Some trials have been performed on FeCo-2 V; Fig. 16 shows square yokes of 150 × 150-μm size and 60-μm depth. It is visible the good quality of machined surfaces and the straightness of the walls. It presents a slight amount of burr and chip material on the upper edges. More complex yoke shapes have been achieved such as the one shown in Fig. 17. It is a near square yoke of 100-μm diameter and 200-μm height, over a circular base of 300-μm diameter and 100-μm height. This kind of slender geometries is required as yokes to be winded, winded yokes, are active electromagnetic components needed in devices such as actuators.

Even though machining parameters have been selected to minimize machining imperfections, it is not possible to fully machine without any burr residuals. This defect may affect to the assembly process of micrometric components. An electrochemical process is applied to the machined parts to eliminate the burr imperfection on edges as a complementary process.

To achieve a better surface finish, electropolishing is done in an acid bath. The power supply is set to 0.7–0.8 V, a resistance of 0.4,0.5 Ω appears, the total time needed is between 90 and 135 s to remove impurities and burrs, the positive cathode is hooked to the workpiece material from which the material must be removed (FeCo-2 V alloy) and the negative to the recipient, which is copper. The solution is 50% water and 50% hydrochloric acid of 10% of purity in volume. In Fig. 18 a groove of 250-µm width and 30-µm depth is shown before and after the electropolishing process. A 14-µm bulge of burr and uncut chip gets completely eroded after electropolishing, as Fig. 19. The difference in finishing is key to make a machined part suitable to assembly in MEMS. This process solves one of the most important drawbacks of microcutting processes.

The machined rotor on FeCo-2 V shown in Fig. 14 is used as a practical example of the effectiveness of the electropolishing process. Considering the power applied and the submersed surface of FeCo-2 V workpiece material, the current density applied is about 320 A/m².

Figure 20 shows the micropart after the polishing process, and Fig. 21 shows the comparison between section profiles before and after polishing. It is clearly visible that the process allows eliminating the burr residuals and surface irregularities without damaging the machined shape.