*3.4. Laser Powder Bed Fusion-LPBF (Moog)*

At Moog, flexures made from metal powders, AlSi10MG and 17-4 stainless steel, were fabricated using laser powder bed fusion (LPBF). Table 2 lists the process parameters. The flexures were then cut to desire thickness using wire EDM. Flexures with and without the tabs were built. Figure 9 shows the two 0.51 mm thick aluminum flexures (9a with tabs and 9b without tab) and a 0.64 mm thick stainless steel (9c. without tab). The surface pattern observed in the figure was left behind by the wire EDM trimming. From the high-resolution micrographs, the width of the elements is consistent. There is however minor distortion of the elements in terms of bending are observed, resulting in small variations in the gap width between the elements. The degree of distortion is less for the ones with tabs. Without the tabs, the distortion is less for the 17-4 build than for the aluminum counterpart. The density of metals produced via LPBF is typically 99.7% that of (fully dense) wrought material [10]. The reduction in density is due to various forms of material defect, which are typically small i.e. less than 0.10 mm. These defects in concert with a rough as built surface finish can result in reduced fatigue performance when compared with smooth surfaced wrought material. For example, the S-N (stress versus number of cycles to failure) curves for LPBF-built AlSi10MG using the ALB1 process show that the average fatigue life of specimens printed at 0, 30, 60, and 90 degrees is about 65% of that of the aluminum wrought [11]. The fatigue life of the SLM (selective laser melting)-built steel 630 is about 57% that of its wrought [9]. It has been demonstrated that the fatigue life of SLM AlSi10Mg parts can been extended to about 8% by machining and heat treatment [12]. Tests will be needed to determine whether 3D printing would be suitable to fabricate flexure for the intended application. It should be pointed out that the fatigue lives of AWJ-cut aircraft aluminum and titanium were able to extend considerably through dry-grit blasting on the part edges [13,14].



§ Used wire EDM to trim to desired part thickness. ¥ For stress release only.

(**c**) 17-4 Stainless steel (0.64 mm THK)

**Figure 9.** Metal flexures—LBPF.

#### *3.5. 3D Printing (NanoArch Micro Scale—BMF InP140*/*InS140)*

Flexures were printed at BMF Material Technology, Inc. in Shenzhen City, Guangdong Province, China. Flexures were printed using two materials, GR and HTL, that were acrylic based photosensitive resin developed by BMF. Refer to its material properties in Table 3 (https://bmf3d.com/materials/). Two GR flexures (0.51 mm and 1.02 mm THK) are shown in Figure 10a,b. The surfaces of the flexures were quite smooth and flat. Both showed noticeable distortion along the direction of the *Y*-axis on the X-Y plane. A portion of the element segments were bent slightly as indicated by the nonuniform gap width between several straight sections of the element. The distortion is more severe on the thick flexure than on the thin one. Figure 10c shows a third flexure built from the HTL material in black color. Distortion of the flexure element was also observed.



(**c**) 0.51 mm THK HTL

**Figure 10.** 3D-printed flexures (BMF).

#### *3.6. 3D Printing (Formlabs—Form 2)*

Stereolithography, an additive manufacturing process that polymerizes a liquid resin with light, was used at Formlabs to print two different sets of flexure with different supporting structures [15]. A Model Form 2 printer, a galvanometer system to steer a laser on a cure plane for this purpose, was used in this case. A model is sliced into layers as thin as 0.025 μm and created layer-by-layer on this cure plane (https://formlabs.com/blog/ultimate-guide-to-stereolithography-sla-3d-printing/). Wherever the laser hits the resin, the material hardens into the final part. An inverse stereolithography process, parts are formed "upside down", and are drawn up from a tank full of rigid resin that was reinforced with glass to offer very high stiffness and polished finish [16].

Figure 11 illustrates two flexures built with the Form 2. The horizontal and radial elements of both flexures display considerable distortion. According to Formlabs, the peel and squish forces of the print process were most likely responsible for the distortions. After each layer, the part separates ("peels") from the tank. This motion is a combination of the tank moving laterally and the *Z*-axis moving upwards. After separation, the part then returns to its original position, though one-layer-thickness higher. For small fragile parts with long thin features, this separation and return can generate forces that cause the part to return slightly off of position. Since there are lots of thin features next to each other, over time this displacement added up enough to cause them to get close enough such that the liquid resin around them caused them to stick together through viscous forces such as surface tension, enough to hold them in place.

#### *3.7. Micromachining (Datron)*

High-speed CNC milling was used to machine two flexures (full and 1/2 scales) at Datron. They were cut on a double flute end mill using ethanol coolant on a NEO CNC Machine equipped with the Autodesk Fusion 360 software. The material was 2024 aluminum 0.51 mm thick. The aluminum sheet was secured with masking tape and super glue. The cutting parameters were given in Table 4.

**Figure 11.** Flexures built with rigid resin—3D printing (Formlabs).

**Table 4.** Cutting parameters for micromachining with Datron Neo CNC Machine.


**¥** Depth of cut; **§** Width of cu.

Figure 12 shows the two flexures. There is no apparent distortion on the horizontal and radial segments of the full-scale flexure element. However, there is observable distortion on the top three horizontal segments with non-uniform gap spacing. It was noted that workpiece holding issues prevented optimization of speeds and feeds. A very shallow depth of cut and slow feed rate was required to prevent tool breakage. An uneven application of the superglue underneath the masking tape could cause tools to break. This allowed for slight movement of the workpiece during machining. It was also very difficult and time consuming to remove the tape and super glue on the flexure without damaging the delicate part.

**Figure 12.** Aluminum flexures—Neo at Datron.

Micrographs of the half-scale flexures machined with the Zund, MicroMAX, and Neo (at Datron) are replotted side-by-side in Figure 13. Figure 13a1,b1,c1,a2,b2,c2,a3,b3,c3 show the as-cut full flexures, cores of the flexure elements overlaid with the tool paths, and the magnified views of the mid-span and two end loops of the flexures, respectively. Visual comparison of the micrograph of the waterjet-cut flexure and the tool path shows no degradation (column b) resulted from the downsizing. The maximum deviation is still about 0.1 mm. Since there is a slight edge rounding on the jet entry surface

of the flexure, the micrograph shown in column b corresponds to the jet-exit surface of the flexure. In the presence of the edge taper, the width of the flexure element is slightly but consistently wider than the tool path in the presence of the edge taper. For the Zund-cut counterpart, however, the downsizing has led to certain degradation in the match between the flexure and the tool path. Figure 13a3 displays noticeable rotational distortion. The maximum mismatch was measured to be 0.27 mm, nearly three times that for the full-scale flexure. Considerable rotational distortion is observed for the Neo-cut half-scaled flexure, as shown in Figure 13c3. The maximum mismatched was measured to be 0.66 mm.

**Figure 13.** 0.5 scale fluxures cut with Zund, MicroMAX, and Neo at Datron (from left to right): As-cut flexures with frames (**a1**, **b1**, and **c1**); tool paths superimposed onto flexure elements (**a2**, **b2**, and **c2**); and tool paths superimposed onto flexure elements—zoomed in (**a3**, **b3**, and **c3**).

## *3.8. Micgraphs of Flexure Elements*

The 0.5-scale flexures cut with the Zund, MicroMAX, and Datron (Neo) were further inspected under a microscope to compare the performance of the three machines for micromachining. Figure 14 illustrates typical micrographs of top and bottom views of a single end loop cut with the three tools. Note that the designed width of the flexure element and the gap between the horizontal segments of the element are 0.25 and 0.38 mm, respectively. The sum of the two is 0.634 mm. From the micrographs, we measured the average values of these dimensions. The corresponding values for the flexures machined with the Zund, MicroMAX, and Datron are (0.240, 0.387 mm), (0.262, 0.366 mm), and (0.156, 0.462 mm), respectively. Comparison of the dimensions of the flexures machined with the three tools is summarized in Table 5. Note that the gap is governed by the diameter of the cutting tool while the sum of the measured element width and gap is the same as that of the designed dimensions.



\* Average value derived from micrographs based on 10 to 20 measurements.

**Figure 14.** Visual comparison of micrographs of top and bottom surfaces of half-scale flexures cut with Zund, MicroMAX, and Neo at Datron.

In Figure 14 each part displays certain anomalies. For example, there is a consistent flat spot on the outside loop of the Zund-machined flexure. For the MicroMAX-machined flexure, a minute rounding on the edge of the jet entry surface and an edge taper can be observed. The average edge taper was measured to be 33 μm. The corresponding edge taper for the Zund- and Neo-machined flexures are considerable smaller, that is 8 μm. Note that waterjet and the end mill of CNC milling are a flexible and a hard tool, respectively. During machining, waterjet bends and spreads and its cutting power reduces as it cuts into the workpiece. As a result, a natural taper forms on the waterjet–cut edges. On the other hand, the end milling is in direct contact with the cut edge of the workpiece. The minute edge taper is likely caused by the deflection of the miniature end mill. The JetMachining Center is equipped with a 5-axis accessory, Tilt-A-Jet (TAJ), capable of compensating edge taper (https://www.omax.com/accessories/tilt-a-jet). The TAJ is however not effective in machining thin materials; it will be applied to stack cutting in Section 3.9.

Comparison of these values with the designed dimensions indicates that the flexures machined with the Zund and the MicroMAX match well with the designed dimensions. The large deviations of the element width and the gap on the flexure cut with the neo at Datron is attributed to the large diameter of the end mill (0.48 mm). Figure 14f shows that the Neo did not cut through the materials at several spots (below the green dashed line). The poor performance of the Neo is partly attributed to the imperfect fixturing to secure the workpiece, according to Datron. As a result, the workpiece might have been moved during machining. Using better tape and more even application of super glue, similar to the process used on the setup of the Zund, would likely allow an increase to depth of cut and feed rate, reducing the cycle and handling times of this operation.
