Femtosecond Laser Processing of Nitinol

STR/3/28/MT Femtosecond Laser Processing of Nitinol H. Y. Zheng, A. R. Zareena and H. Huang Abstract - The effects of femtosecond laser machining on surface characteristics and subsurface microstructure of a Nitinol alloy were studied. The femtosecond laser machining produced an average surface roughness of.2 µm. A re-deposited layer of 7 µm and a hardened layer of 7 µm were also resulted from the machining. Keywords: Ultrashort, Femtosecond, Nitinol 1 BACKGROUND Ni-Ti shape memory alloys (nitinol) exhibit excellent shape memory effect, unusual pseudoelasticity, excellent bio-compatibility and very high corrosion resistance, and as such are widely used in the medical applications [1-2]. Self-expanding nitinol stents are one of the examples. Stents are small wire mesh tubes designed to keep arteries from re-clogging after a balloon dilation procedure. However, the mechanical properties of nitinol are sensitive to thermal influence and mechanical tension. This makes it difficult to accurately machine parts with high geometrical complexity. Nonconventional machining techniques, such as ultrshort pulse laser machining and electrodischarge machining (EDM) are believed to produce more acceptable results. Ultrashort pulse lasers such as femtosecond lasers are potentially useful for processing nitinol due to the unique beam-material interaction mechanism and beam characteristics. The extremely short pulses have extremely high power density leading to instant vaporisation of materials at the focus. There is little time for the heat to diffuse in such a short period of time and therefore there is minimal heat affected zone (HAZ) of the machined regions [3-4]. It has been demonstrated that ultra-short laser pulses can reduce the thermal influence on the nitinol material to a minimum [5]. Alternatively, the other hand, longer laser pulses in the µs-range have led to a relatively large HAZ that reduces the shape memory effect. In the near future, the application of ultra-short pulse lasers for the fabrication of micro-parts and micro-structures will become an established industrial technique. This is due to the very promising results obtained in the field of laser material processing and the rapid progress in the development of compact and reliable solidstate femtosecond lasers. 2 OBJECTIVE The study is to understand the machining characteristics of nitinol using the femtosecond laser and to explore its potential applications. 3 METHODOLOGY A chirped pulse amplification (CPA) based Ti:Sapphire femtosecond laser was used. The laser operates at the wavelength of 775 nm, repetition rate of 1 khz, maximum average power of 8 mw, and pulse width of 15 fs. A ¼ waveplate was used to change the linearly polarized beam into circularly polarized beam. A focal lens of 5 mm focal length was used to process the nitinol sheets and strips respectively. A stream of N 2 gas was blown along the sides of the cutting direction to reduce debris deposition. The as received Nitinol (5.6Ni- 49.4Ti) sheets and were polished to remove any surface coatings, oxides and residues prior to machining. The kerf width, depth, and grain structures of etched specimen were analyzed using scanning electron microscope. Further examination was carried out using EDX and XRD to assess microstructural changes. 4 RESULTS AND DISCUSSIONS 4.1 Process studies Grooves were produced by varying the target power on the workpiece as well as the distance between the focusing lens and the workpiece. The cutting length, scanning speed and exposure time were set constant throughout the experiments. It was found that the focal position, corresponding to the narrowest kerf followed almost the same pattern (Fig. 1) irrespective of laser powers. Position (zero) refers to the beams focal position on the surface of the sample,.5, -1, -1.5 are below the surface and 1 to 6 are above the sample surface. 1

Width of cut (mm).5.4.3.2.1 3 35 4 45 5 Vs Kerf width -1.5-1 -.5 1 2 3 4 5 6 Z - Value(mm) Fig. 1. vs. kerf width. Fig. 2 shows relation between the kerf width and the focal position at the power of 3 mw and the scanning speed of.1 mm/s. It is clear from this graph that the width of the kerf reduced when the focal position was closer to the surface and the narrowest when focused exactly on the surface of the sample. When the beam was focused above the sample surface, air plasma was observed which could change the beam propagation path by functioning as a thin lens. The slight non-symmetry of the kerf width shown in Fig. 2 may be the result of the air plasma effect. The depth of the slots measured at the cross-section of the kerfs at different focal position is shown in Fig. 3. It was also observed that the cut kerf was deeper when the laser beam was focused on the sample surface. The fact that the narrowest cut kerf corresponds to the deepest cut is significant, and that has been experimentally confirmed that the principle of the technique is widely adopted for determining the focal position. Width of Kerf (mm) Vs Width of Kerf.6.5.4.3.2.1 -.6 -.4 -.2.2.4.6 Fig. 2. vs. kerf width (3 mw). The SEM photoshots (Fig. 4) shows the slots machined at different focal positions. The relation between the taper angle and the focal position was studied and plotted in Fig. 5. The taper angle was calculated by measuring the difference in the cut width at the beam entrance and the exit. When the beam was focused on the top surface, the taper angle was the smallest and its measured value is about 84.9. This agrees with the results shown in Fig. 2. Depth of cut (mm) Focal Position Vs Depth of cut.5.4.3.2.1 -.6 -.4 -.2.2.4.6 Fig. 3. vs. kerf depth. Fig. 4. SEM photoshots of fs laser machined slots at different focal positions. Kerf Angle ( ) Focal Position Vs Kerf Angle 9 75 6 45 3 15 -.6 -.4 -.2.2.4.6 Fig. 5. vs. wall angle (P=3 mw). Redeposition was observed to be in the powder form along the cut, which can be wiped off using alcohol. Assist gas (e.g. N 2 ) helped to reduce the amount of the re-deposition. It was observed that the width of the slots varied with power levels. The kerf width was lesser with lower powers (Fig. 6). This indicates that the femtosecond laser processing is a threshold-based ablation process. The material removal rate was calculated for different power levels at a fixed scanning speed and plotted in Fig. 7. A high power resulted in a higher removal rate as expected. However, better machining quality was achieved with lower power levels. 2

WOC ( m) 25 2 15 1 5 Power Density Vs WOC (Nitinol) 1-1 -1.5 Fig. 6. Power density Vs Kerf width. 2 25 3 35 4 45 5 The obtained results are significant, as it was reported earlier that the cut surface started to taper when the aspect ratio was greater than 1. Assuming a Gaussian beam mode, the beam diameter (1/e²) at the focusing lens (D) was estimated about 4 mm. The depth of focus (DOF) is then calculated to be 178 µm according to the following equation DOF = 1.48 (f / D)² λ MRR (x1-5 mm 3 /sec) 1 8 6 4 2 Pow er Vs MRR where f is the focal length (5 mm) and λ the wavelength (771 nm). The deep and near parallel cut beyond the depth of focus has indicated that there are other mechanism for the beam to propagate through the substrate, one of which may be the beam self-focusing effect due to multiple reflections off the side walls. 1 2 3 4 5 6 Pow er (m W) Fig. 7. Power density Vs MRR. SEM pictures of top and cross-sectional views of the kerfs at different power levels are shown in Fig. 8. The slot machined at lower power was straighter as the effect of power instability was minimised due to the threshold-based ablation process. An aspect ratio as high as 3 was obtained (depth: 75 µm, width: 28 µm) at 15 mw output power. The wall was nearly vertical. The laser processed nitinol as well as the unmachined specimen were etched to expose the micro-grain structure. The etchant used is a solution of HF, HNO 3 and H 2 O. SEM pictures of the etched surface of the unmachined nitinol and that of the kerf edges are shown in Fig. 9. Fig. 9. Microstructure of cut edge (left) and unmachined nitinol (right). 45 mw 35 mw 25 mw 15 mw Fig. 8. Cross-sectional view of the slots machined at different power levels. No obvious change in the microstructure was observed at the kerf edges, which is in contrast to the HAZ observed for long-pulse laser machining. The minimization of the HAZ is due to the negligible thermal diffusion, which is proportional to the square root of the pulse width. The threshold-based ablation is able to produce features smaller than the spot size. Therefore, the efficiency and precision of the process of the ultrashort pulses are high as compared to long pulse laser machining approaches, and thermal damage to the surrounding material is minimised. Also, experiments were carried out on nitinol strips of about 1.2 mm in thickness. Slots were machined at different output power levels at a constant scanning speed and the number of passes was increased to make cut-through slots. The slots machined at different number of passes were examined in SEM and the shots are shown in the Fig. 1. The machined slot 3

edges were smooth with almost no heat affected zone but there was more redeposit along the edges. SEM analysis has shown no significant change in the microstructure of the kerf edges both at the surface and at the cross-section. From these graphs it is clear that both average roughness (Ra) and total roughness (Rt) increased with the increase in output power and was low with the increase in the overlapping space. 6 5 4 Power Vs Ra 4 3 2 1 (a) (b) Ra ( m) 3 2 1 Fig. 1. Slots machined (a) after 6 passes, (b) after 3 passes..2.4.6.8.1.12.14.16 Overlap (mm) Fig. 12. Output power vs. Ra. In another set of experiments, nitinol strip of 2.5 mm in thickness was processed using a longer focal lens (1 mm). 25 2 Power Vs Rt Rt( m) 15 1 5.2.4.6.8.1.12.14.16 Overlap (mm) 4 3 2 1 Fig. 11. SEM shots of kerfs machined in nitinol strips of 2.5mm thick (P=35mW, n=.1mm/s). The aspect ratio as high as 6 was obtained when laser beam was focused on the sample surface as it produced a kerf width of 42 µm (Fig. 11). Similar to our earlier results, the SEM analysis of etched surface of this nitinol strips also did not show any change in the grain structure. This is probably because the energy is deposited in an optical skin depth before thermal diffusion may occur. The vaporized material hydrodynamically expands from the surface with little energy deposited in the bulk. Thus, there is little heat affected zone in the remaining bulk material or recast layer. Also, the lasers induced breakdown at high laser field and extremely short interaction time lead to high accuracy ablation threshold and effectively minimize the heat diffusion into the material. In order to measure the roughness of the processed nitinol surface, overlap cutting was carried out. The scanned lines were overlapped and the overlapping space was varied from.5 to.15 mm at different output power. The following graph shows the measured Ra (Fig. 12) and Rt (Fig. 13) for different output power and overlapping space. Fig. 13. Output power vs. Rt. This is because at higher output power (4 mw) the ablation rate is high thus resulting in a rougher surface (Ra: 5 µm, Rt: 25 µm). Also, when the overlapping space is more than.1 mm, the area exposed to laser processing becomes less and so the surface roughness. At 1 mw, with overlapping space of.15 mm, Ra is 1.8 µm and Rt is 1µm. 4.2 Stent cutting an application After optimizing the cutting parameters for the laser processing of nitinol, stent patterns were cut on nitinol sheets.12 mm thick. The good shape memory effect and unique volume memory effect of nitinol can simplify the implanting process of medical implants and make patients suffer less pain. The compressibility and porosity of nitinol are beneficial for the ingrowth of bone mineralization tissue and fibrous tissue, which makes the fixation of implant much more natural and reliable. Stents can be made either by cutting finer lines on thin walled nitinol tubes or on thin nitinol sheets. The edges of the processed nitinol sheet will be welded together to form the stents. In this study stent pattern was machined on thin 4

nitinol sheets with the laser output power of 3 mw and at a scanning speed of.5 mm/sec. A focusing lens of 5 mm focal length was used. The following SEM shots (Fig. 14) shows the stent pattern machined on the 12 µm thick nitinol sheets. the presence of B2 phase austenite was confirmed from the obtained data and thus no change in the microstructure of the processed nitinol and the absence of HAZ were reconfirmed in this analysis. XRD Spectra Intensity (arb. unit) 25 5 75 1 125 15 2θ (degree) Fig. 14. SEM shots of nitinol stent pattern. Fig. 15. XRD spectra of laser processed Nitinol stent and the as-received specimen. Fines lines were cut on the nitinol sheets and the smallest cut width machined for the stent pattern was 47 µm. The zig-zag edges at the exit was the debris deposit as no assist gas was used during stent cutting. This zig-zag edges could be overcome by reducing the scanning speed but, the lowest stage feed rate available in the laser equipment was.5 mm/sec. The SEM analysis has proven the absence of HAZ and micro-cracks in the machined stents. This high precision of material removal is possible as a result of the strong intensity dependence of the multiphoton initiation step. The change in material removal mechanism away from a thermal process results in a minimum increase in the temperature of the surrounding material and a significant reduction in the shock of the bulk material. Debris was deposited along the edges of the stent and are collected for further analysis. Also, the debris could be easily wiped off using acetone. The surface roughness (Ra) of the machined stent was measured using Wyko surface profiler was.25 µm. The machined stent was also subjected to x-ray diffraction analysis to examine the effect of the laser machining on the micro structure of the machined nitinol and the HAZ. The stent surface as well as the as-received specimen was scanned using a Philips X-ray Diffractometer and the output XRD spectra is shown in Fig. 15. The XRD spectrum obtained from the laser processed nitinol stents exhibits a similar pattern to that of the as-received nitinol sheet. Also, Fig. 16. SEM shots of laser processed nitinol debris. Debris deposited during the laser processing was also collected and analyzed in SEM as shown in Fig. 16. A typical cluster is about 775 nm in size and these particles were also subjected to EDX analysis to examine the presence of other elements. The EDX analysis confirmed the absence of any foreign particles in the debris collected (Fig. 17). Fig. 17. EDX spectra of laser processed Nitinol debris. 5

5 CONCLUSIONS Femtosecond laser processing of nitinol was studied and the research outcome is summarized as below: Initially the principal technique widely adopted for determining the focal position the narrowest cut kerf corresponding to the deepest cut was experimentally confirmed. Highest aspect ratio of 6 with nearly vertical cut surface was achieved in machining nitinol. Analyses of the laser processed nitinol showed no obvious changes in the micro grain structure and thus resulting in high machining quality with negligible heat affected zone. The machined surfaces are free from recast layer and micro cracks. Stent patterns were machined on thin nitinol and stainless steel sheets using optimised parameters. Debris collected during the laser processing of nitinol showed the presence of nanoparticle clusters (775 nm). Thus a possibility of producing nitinol nano-particles by laser processing was identified and further exploration is required in this aspect. Thus femtosecond laser processing was identified as a more promising method for micro-machining and stent cutting of nitinol alloys with minimum HAZ and thus processing without compromising the unique properties of the material. 6 INDUSTRIAL RELEVENCE Femtosecond laser is a promising tool for processing materials with high precision and minimum HAZ. Applications may be found in microelectronics, biomedical and photonics industries. REFERENCES [1] I. Martynov, V. Skorohod and S. Dolonin, Shape memory and superelasticity behavior of porous Ni-Ti material, J. de physique, Vol. 4(1), pp. 421, (1991). [2] B. Thierry, Y. Merhi, L. Bilodeau, C. Trepanier and M. Tabrizian, Nitinol versus stainless steel: acute thrombogenisity study in an ex vivo porcine model, Biomaterials, Vol. 23, pp. 2997-35, (22). [3] M.D. Feit, et al., Femtosecond laser machining, CLEO (CFD2), pp. 51, (1998). [4] S. Nolte, et al., Micromachining with femtosecond lasers, CLEO (CFD3), pp. 51, (1998). [5] H.Y. Zheng, A.R. Zareena, H. Huang and G.C. Lim, Studies of Femtosecond Laser - processed Nitinol, J. of Materials Science Forum, Trans Tech. Publications Inc., Vol. 437-438, pp. 277 28, (23). 6