Open Access
Add to BibSonomy
Abstract
We present sequentially timed all-optical mapping photography (STAMP) with a slicing mirror in a branched 4f system for an increased number of frames without sacrificing pixel resolution. The branched 4f system spectrally separates the laser light path into multiple paths by the slicing mirror placed in the Fourier plane. Fabricated by an ultra-precision end milling process, the slicing mirror has 18 mirror facets of differing mirror angles. We used the boosted STAMP to observe dynamics of laser ablation with two image sensors which captured 18 subsequent frames at a frame rate of 126 billion frames per second, demonstrating this technique’s potential for imaging unexplored ultrafast non-repetitive phenomena.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
High-speed optical imaging is an essential tool for studying ultrafast dynamics in a broad range of scientific and industrial fields. Digital image sensors based on charge-coupled devices (CCDs) and complementary metal-oxide-semiconductor (CMOS) have been utilized for sequential movie acquisition of dynamic phenomena at micro- to milli-second timescales [1–3]. Although these image sensors allow us to obtain the motion pictures of phenomena in a single shot, the imaging speed is currently insufficient to resolve ultrafast dynamics at femto- to nano-second timescales. On the other hand, time-resolved imaging based on the pump-probe method is widely used for observing such ultrafast events which cannot be captured with these image sensors [4,5]. However, the target events are limited to phenomena that can be repetitively produced by the pumping. Therefore, single-shot imaging with femto- to nano-second temporal resolution is needed to measure non-reproducible ultrafast phenomena such as plasma evolution [6], chemical reactions, [7] and phase transitions [8].
Recently, remarkable advancements have been made in the single-shot ultrafast imaging technology [9–20], including compressed ultrafast photography [11–13], holographic recording [15,16], frequency recognition algorithm for multiple exposures [17] and others. Although these techniques have achieved movie acquisition with femto- to nano-second temporal resolution, their ultrafast imaging capabilities rely on complicated optical arrangements or computational algorithms which degrade quality of the reconstructed image frames. Sequentially timed all-optical mapping photography (STAMP) [18–20] is one of the single-shot imaging methods that can reach Tfps frame rates. Owing to the principle of STAMP based on the spatiotemporal mapping via spectral dispersion, the image quality of the acquired movie is not sacrificed in its imaging operation. In STAMP, a 4f system plays an important role as a spatial mapper of the laser spectrum, in which spectrally time- and image-encoded pulses are spatially separated toward different areas of an image sensor that captures them in a single exposure. However, this operation results in a trade-off between the number of frames and pixel resolution. Since the pixel resolution denotes the number of the sensor’s pixels allocated to each time- and image-encoded pulse, it is significantly reduced as the number of frames is increased. This trade-off hampers temporally-detailed analysis of ultrafast dynamics.
Here, we propose and experimentally demonstrate a branched 4f system with a slicing mirror [21–24] to increase the number of frames of STAMP while preserving its pixel resolution. The 4f system includes the slicing mirror, lenses and diffraction gratings and is branched into multiple paths by the slicing mirror with multiple facets tilted at distinct angles. The branched path allows for capturing dynamic events by multiple image sensors and thereby overcomes the trade-off between the number of frames and the pixel resolution in STAMP. To demonstrate the boosted STAMP performance with the branched 4f system, we designed and developed a dual-path branched 4f system. A slicing mirror comprised of 18 mirror facets was fabricated by an ultra-precision end milling process to satisfy the requirements on the tilt angles and surface roughness of the mirror facets. With the developed STAMP system with the branched 4f system, the dynamics of femto-second laser ablation was imaged with two image sensors that individually captured 9 frames in 3 × 3 arrangement (i.e., 18 frames in total), doubling the number of frames without scarifying the pixel resolution. The presented ultrafast single-shot imaging technique has the potential to enable detailed analysis of non-repetitive transient phenomena at femto- to nano-second timescales, e.g. the monitoring and optimizing laser machining and understanding shockwave physics in microscale.
2. Materials and methods
2.1 Design of optical system and slicing mirror
The optical setup of the branched 4f system for boosted STAMP is shown in Fig. 1. To perform STAMP, an ultrashort laser pulse is temporally stretched so that its duration matches to the timescale of the target event. At the target, each spectral component of the chirped pulse arrives at a different time determined by its wavelength and the temporal dispersion. The pulse then enters the branched 4f system consisting of two pairs of a diffraction grating and a lens with a slicing mirror placed in the Fourier plane. The slicing mirror contains two adjacent sets of 9 mirror facets (i.e., 18 in-line mirror facets in total) that realizes the dual-path design. The mirror facets are tilted around a shared axis that coincides with the spatially-dispersed spectrum of the chirped pulse with intra- and inter-set angular differences of ∼0.3 degrees/facet and >10 degrees/set, respectively. Thus, each mirror facet reflects a certain spectral bandwidth of the incoming pulse with a distinct angle, creating a spatially-separated daughter pulse. With one of the two sets of the mirror facets, the corresponding 9 daughter pulses are reflected back to the same grating-lens pair, while with the other set, the remaining 9 daughter pulses are directed to the other grating-lens pair. Note that the reflective 4f arrangement of the grating-lens pairs allows us to preserve the spatial information carried by the incoming pulse in the output daughter pulses, as shown in the inset of the figure. After passing through an imaging lens placed in the downstream of each of the two branched paths, the time- and image-encoded daughter pulses form images of the target event on different areas of the two image sensors.
Fig. 1. Schematic of the branched 4f system for boosted STAMP. One of the two branched paths of the 4f system is comprised of diffraction grating 1, lens 1 and the slicing mirror (SM), offering round-trip trajectories of the daughter pulses. The other path consists of diffraction grating 1, lens 1, SM, lens 2 and diffraction grating 2.
2.2 Fabrication of slicing mirror
An aluminum alloy (A5052) coated with electroless NiP was selected as the substrate material for the slicing mirror to ensure high reflectivity at the near-infrared wavelengths. Monolithic processing by a five-axis ultra-precision processing machine (ULG-100D (5A), Shibaura Machine, JP) was employed for fabrication to improve the relative angular accuracy in the tilt of each mirror facet. To fabricate each facet, the substrate was cut by milling in a plane perpendicular to an end mill equipped on a spindle that runs at 20,000 rpm. A single crystal diamond square end mill (φ = 0.5 mm) was utilized to reduce the surface roughness (<10 nm Ra) and figure error (<80 nm P-V) of the processed facet (Fig. 2(a)). During the cutting process, the substrate was mounted on a holder of the five-axis machine such that the axis of the slicing mirror was positioned to coincide with the rotation axis ($\gamma $) of the machine (Fig. 2(b)). By rotating the machine axis ($\gamma $) with respect to the spindle, the tilt angle of each facet of the slicing mirror was precisely controlled. The cutting process was performed by translating the five-axis machine with a feed speed of 25 mm/min, cutting depth of 2.0 µm and tool pitch of 0.1 mm. The width of each mirror facet was designed to be 2 mm, which was found wide enough to avoid the lack of focused imaging pulse on the facets. The tilt angles of the 18 facets were designed to be -0.9166°, -0.6302°, -0.3437°, 0.0000°, 0.2863°, 0.5724°, 0.9153°, 1.2008°, 1.4859°, 12.2538°, 12.5389°, 12.8244°, 13.1673°, 13.4534°, 13.7397°, 14.0834°, 14.3698° and 14.6563°, respectively. In this design, two sets of the array of 1 × 9 pulses exit the 4f system. The 9 pulses are divided into 3 groups of 3 pulses, which has a gap of 0.5 mm and 2.5 mm between the groups and the pulses, respectively. Additionally, the 4th facet was set as a reference (0.0000°) to relay the pulses with 2-inch diameter lens.
Fig. 2. Fabrication of the slicing mirror. (a) Single crystal diamond square end mill tool (φ = 0.5 mm). (b) Fabrication setup with a five-axis ultra-precision processing machine.
After the fabrication, the mirror surface condition was measured using a coherence scanning interferometer (NewView7200, Zygo, US) and analyzed based on Ra and P-V values. The errors in the width and tilt angle of each facet of the slicing mirror were measured by a non-contact 3D shape measurement instrument (PFU-3, Mitaka Kohki, JP). The width was measured at a resolution of 20 µm. The angle of each facet was defined with respect to the 4th facet and estimated by the least-square linear fit of a 1D profile of the facet taken in the direction perpendicular to the axis of the slicing mirror.
2.3 Apparatus for observing laser ablation
To demonstrate the boosted imaging performance of our STAMP setup with the branched 4f system, we monitored ultrafast laser ablation dynamics of a glass plate (Fig. 3). To excite laser ablation, we focused an ultrashort laser pulse (with a center wavelength of 803 nm, bandwidth of 35 nm, pulse width of 35 fs and pulse energy of 260 µJ) from a Ti:sapphire laser with a chirped pulsed amplifier (CPA) system (Astrella-USP-1K, Coherent, US) with a lens (f = 30 mm) onto the surface of a 50 µm-thick glass plate. To monitor the ablation dynamics, an uncompressed chirped pulse (with a center wavelength of 803 nm, bandwidth of 35 nm and pulse width of ∼200 ps) from the CPA system was used. The STAMP setup was equipped with two sCMOS image sensors (ORCA-Flash4.0 V3, Hamamatsu Photonics, JP), a 10x objective lens (M Plan Apo, Mitsutoyo, JP), the branched 4f system, imaging lenses (f = 300 mm) and rearranging mirrors in a shadow-graph configuration. The theoretical spatial resolution according to Rayleigh's criterion was 1.74 µm. The rearranging mirror was incorporated to image daughter pulses on an image sensor with the 3 × 3 array. The mirrors were composed of a round mirror and two pickoff mirrors as shown in Fig. 3. Each mirror had a different tilt angle and enabled the rearranging of daughter pulses from the 1 × 9 to 3 × 3 array. The rearrangement of daughter pulses allows us to fully use the pixels in an image sensor. A band-pass filter (FF02-809/81-25, Semrock, US) was mounted on each camera to block the light emitted at the ablation point. To reduce the noise in imaging resulting from the scattering of the excitation pulse at the edge of the glass plate, we used a half waveplate (WPH10M-808, Thorlabs, US) and polarizer (LPNIRE100-B, Thorlabs, US). The polarization state of the imaging pulse was rotated 90 degrees with respect to the polarization direction of the excitation pulse. The polarizer was positioned after the objective lens to allow the imaging pulse to pass through it while partially blocking the excitation pulse scattered at the edge of the glass plate. Two pairs of a diffraction grating (1800 lines/mm) and a lens (f = 250 mm) and the fabricated slicing mirror were used for the branched 4f system. The time delay between the excitation and imaging pulses was adjusted with an optical delay line. The field of view and magnification obtained with the setup we constructed was approximately 180 µm and 16x magnification, respectively.
Fig. 3. Experimental setup of STAMP with the branched 4f system. A 50 µm-thick glass piece was ablated by a high-intensity femto-second laser pulse. The dynamics was captured by two image sensors.
3. Results
3.1 Evaluation of slicing mirror
A photograph of the fabricated slicing mirror and the surface roughness of its 9th facet are shown in Figs. 4(a) and 4(b), respectively. The field of view in Fig. 4(b) is 500 µm × 600 µm (horizontal × vertical) with the boundary between the 9th and 8th facets appearing at the right edge of the color map. The Ra value was less than 4.5 nm for all the mirror facets. The surface profile of each mirror facet was measured across lines perpendicular to and parallel to the tool advancement direction as represented by Figs. 4(c) and 4(d), respectively. The P-V values were measured for each facet in a field of view of 500 µm × 2500 µm and were determined to be less than 43 nm. The maximum error from the designed mirror width (2 mm) was 60 µm. The maximum angular error from the designed tilts of the facets was 0.0036°.
Fig. 4. Evaluation of the fabricated slicing mirror. (a) Photograph of the fabricated slicing mirror. (b) Color map of the surface roughness on the 9th facet. (c)-(d) Surface roughness profiles of the dashed lines indicated as (c) A-A’ and (d) B-B’ in (a). Other facets also show similar profiles.
The processing accuracy of the slicing mirror, including the surface roughness and angular error in the tilts of the facets, was improved compared to previous slicing mirrors [21–24]. Although we have not investigated the effects of the processing accuracy on imaging performance such as spatial resolution, we assume from the values of Ra and P-V (∼$\lambda $/20 in this experiment) that the slicing mirror did not deteriorate the imaging performance.
3.2 Observation of laser ablation
The frame interval and time window of the constructed STAMP system were calibrated by finding the position of the delay line between the excitation and imaging pulses that made the early stage of air breakdown (i.e., refractive-index change of air) generated by the excitation pulse to be observed in each time frame. They were respectively calculated as 7.9 ps/frame (corresponding to 126 Gfps) and 134.3 ps/movie from the change of the optical path length in the delay line. In this experimental condition, the exposure time is theoretically calculated as 7.9 ps [18]. Then, the delay time between the excitation and imaging pulses was adjusted and fixed such that the laser ablation was initiated in the second frame. After this calibration, single-shot imaging of the laser ablation dynamics was performed.
Figure 5 shows the obtained sequential frames of the glass ablated by the excitation laser pulse. Owing to the branched 4f system, we achieved the large pixel resolution of 450 × 450 pixels/frame with the large frame number of 18. Each frame was trimmed and adjusted in its brightness and contrast to compensate for the intensity-variation between the daughter pulses. In the second to 18th frames, generation and evolution of plasma were observed at the center of the field of view where the ablation pulse was incident at the air-glass interface. These effects appeared black in the obtained images, since dense electrons and ions of the plasma partially absorbed the imaging pulse. For example, in the 11th frame (i.e., 71.1 ps after the excitation pulse was focused onto the glass surface), the plasma-specific black area was composed of a semicircle and a protrusion above it. The former represents the photoelectric effect in which the free electrons were generated on and emitted from the glass surface, while the latter ionization of air. In this experiment, we monitored both plasma generation and evolution seamlessly. Here we focused on the imaging of generation and initial evolution of plasma with constant frame rate and magnification. As we reported previously [18], we can tune the time window and field of view in imaging by changing the STAMP’s optical setup. In the laser ablation process, the development of plasma, shockwave propagation and resolidification would be observed with a slower frame rate.
Fig. 5. Single-shot movie of ultrafast plasma dynamics obtained by STAMP with the dual-path 4f system. Each frame occupied 450 × 450 pixels on the image sensor with the frame interval of 7.9 ps. In a single-shot time window of 134.3 ps, plasma progress generated by laser-induced ablation was monitored.
4. Discussion
We quantitatively characterize the number of frames boosted without sacrificing the pixel resolution in each frame in the presented STAMP with the branched 4f system. Since our interest is in the trade-off between the number of frames and the pixel resolution, we consider a product of the number of frames N and the normalized pixel resolution ${P_n} = {P_f}/{P_i}$, where ${P_f}$ is the number of pixels allocated to each frame and ${P_i}$ the total number of pixels in one image sensor. In the previously reported STAMP systems where only one image sensor was installed, the product $N \times {P_n}$ was 0.290 [18], 0.423 [19] and 0.190 [20], respectively. In contrast, the product in this work is 0.869 which is more than two times larger than those of the previous works. This clearly validates the reduced trade-off barrier in STAMP performance with the branched 4f system.
The imaging quality of the presented STAMP system is yet limited and needs to be improved to perform a more quantitative analysis. Specifically, the striped patterns and blurring in the images shown in Fig. 5 need to be solved. The former were caused by the polarizer which was used to reduce the scattered excitation pulse while allowing the imaging pulse to pass through it. They can be removed by utilizing an optimized polarizer such as a nanoparticle film polarizer. The latter issue was mainly due to misalignment of the dual-path 4f system. The light reflected on the slicing mirror which passed through the region just above the incident light was used as a reference light for alignment of the daughter pulses corresponding to the 10th to 18th frames. On the other hand, the alignment accuracy was lower for the other daughter pulses corresponding to the first to 9th frames because there was no analogous reference light. Therefore, a method to introduce this reference light for alignment is needed.
For further development of STAMP performance with the presented branched 4f system, we expect that a multi-path 4f system with multiple image sensors could allow us to further increase the number of frames without decreasing the pixel resolution. Although a multi-path 4f system would have some technical challenges such as difficulty of alignment or potentially large size of the system compared to the sizes of systems in previous techniques, it could be realized by changing the tilt angles of each facet of the slicing mirror. One of the possible improvements of the optical system is the integration of the slicing mirror and the rearranging optics. A slicing mirror which has mirror facets tilted not only around the shared axis but also around the longer midline on each facet can be substituted for rearranging mirrors. Although its fabrication process would become more complicated, the slicing mirror with 2-axis tilted facets reduces the optical components and the alignment process of the rearranging optics. In general, increasing the number of frames sacrifices signal-to-noise ratio (SNR) of frames because the optical energy of imaging pulse is distributed to each frame. However, in ultrafast imaging based on STAMP, the pulse energy required for imaging is small because high-sensitivity image sensors can be used for image acquisition. The SNR of the frames can be preserved by adjusting the power of imaging pulse as increasing the number of frames.
5. Conclusion
In conclusion, we proposed and experimentally demonstrated STAMP via a branched 4f system with a slicing mirror. We fabricated the special optical component called a slicing mirror to incorporate a dual-path 4f system into STAMP and used the constructed system to monitor the evolution of plasma generated from laser ablation. The slicing mirror which has two sets of 9 facets (with the tilt angles in the first set ranging between -0.9166° and 1.4859° and in the second set ranging between 12.2538° and 14.6563°) was fabricated with a surface roughness of less than 10 nm Ra and 80 nm P-V. In the demonstration of STAMP using the dual-path 4f system, ultrafast imaging of 18 frames with a frame interval of 7.9 ps was achieved. The proposed imaging technique is expected to be a powerful tool for capturing ultrafast, non-repetitive phenomena with high image quality and a large number of frames.
Funding
Ministry of Education, Culture, Sports, Science and Technology (JPMXS0118067246); Precursory Research for Embryonic Science and Technology (JPMJPR17P9).
Disclosures
The authors declare no conflicts of interest.
References
1. T. G. Etoh, D. Poggemann, G. Kreider, H. Mutoh, A. J. P. Theuwissen, A. Ruckelshausen, Y. Kondo, H. Maruno, K. Takubo, H. Soya, K. Takehara, T. Okinaka, and Y. Takano, “An image sensor which captures 100 consecutive frames at 1000 000 frames/s,” IEEE Trans. Electron Devices 50(1), 144–151 (2003). [CrossRef]
2. S. Kleinfelder, S. Lim, X. Liu, and A. El Gamal, “A 10000 frames/s CMOS digital pixel sensor,” IEEE J. Solid-State Circuits 36(12), 2049–2059 (2001). [CrossRef]
3. A. Krymski, D. Van Blerkom, A. Anderson, N. Bock, B. Mansoorian, and E. R. Fossum, “A high speed, 500 frames/s, 1024/spl times/1024 CMOS active pixel sensor,” 1999 Symp. VLSI Circuits. Dig. Pap. (IEEE Cat. No.99CH36326), 137–138 (1999).
4. M. C. Downer, R. L. Fork, and C. V. Shank, “Femtosecond imaging of melting and evaporation at a photoexcited silicon surface,” J. Opt. Soc. Am. B 2(4), 595–599 (1985). [CrossRef]
5. M. Domke, S. Rapp, M. Schmidt, and H. P. Huber, “Ultrafast pump-probe microscopy with high temporal dynamic range,” Opt. Express 20(9), 10330–10338 (2012). [CrossRef]
6. R. Kodama, P. A. Norreys, K. Mima, A. E. Dangor, R. G. Evans, H. Fujita, Y. Kitagawa, K. Krushelnick, T. Miyakoshi, N. Miyanaga, T. Norimatsu, S. J. Rose, T. Shozaki, K. Shigemori, A. Sunahara, M. Tampo, K. A. Tanaka, Y. Toyama, T. Yamanaka, and M. Zepf, “Fast heating of ultrahigh-density plasma as a step towards laser fusion ignition,” Nature 412(6849), 798–802 (2001). [CrossRef]
7. P. Hockett, C. Z. Bisgaard, O. J. Clarkin, and A. Stolow, “Time-resolved imaging of purely valence-electron dynamics during a chemical reaction,” Nat. Phys. 7(8), 612–615 (2011). [CrossRef]
8. S. Wall, D. Wegkamp, L. Foglia, K. Appavoo, J. Nag, R. F. Haglund, J. StäCurrency Signhler, and M. Wolf, “Ultrafast changes in lattice symmetry probed by coherent phonons,” Nat. Commun. 3(1), 721 (2012). [CrossRef]
9. J. Liang and L. V. Wang, “Single-shot ultrafast optical imaging,” Optica 5(9), 1113 (2018). [CrossRef]
10. F. Mochizuki, K. Kagawa, S. Okihara, M.-W. Seo, B. Zhang, T. Takasawa, K. Yasutomi, and S. Kawahito, “Single-event transient imaging with an ultra-high-speed temporally compressive multi-aperture CMOS image sensor,” Opt. Express 24(4), 4155 (2016). [CrossRef]
11. L. Gao, J. Liang, C. Li, and L. V. Wang, “Single-shot compressed ultrafast photography at one hundred billion frames per second,” Nature 516(7529), 74–77 (2014). [CrossRef]
12. J. Liang, C. Ma, L. Zhu, Y. Chen, L. Gao, and L. V. Wang, “Single-shot real-time video recording of a photonic Mach cone induced by a scattered light pulse,” Sci. Adv. 3(1), e1601814 (2017). [CrossRef]
13. J. Liang, L. Zhu, and L. V. Wang, “Single-shot real-time femtosecond imaging of temporal focusing,” Light: Sci. Appl. 7(1), 42 (2018). [CrossRef]
14. K. Goda, K. K. Tsia, and B. Jalali, “Serial time-encoded amplified imaging for real-time observation of fast dynamic phenomena,” Nature 458(7242), 1145–1149 (2009). [CrossRef]
15. T. Inoue, A. Matsunaka, A. Funahashi, T. Okuda, K. Nishio, and Y. Awatsuji, “Spatiotemporal observations of light propagation in multiple polarization states,” Opt. Lett. 44(8), 2069 (2019). [CrossRef]
16. M. Sasaki, A. Matsunaka, T. Inoue, K. Nishio, and Y. Awatsuji, “Motion-picture recording of ultrafast behavior of polarized light incident at Brewster’s angle,” Sci. Rep. 10(1), 7638 (2020). [CrossRef]
17. A. Ehn, J. Bood, Z. Li, E. Berrocal, M. Aldén, and E. Kristensson, “FRAME: femtosecond videography for atomic and molecular dynamics,” Light: Sci. Appl. 6(9), e17045 (2017). [CrossRef]
18. K. Nakagawa, A. Iwasaki, Y. Oishi, R. Horisaki, A. Tsukamoto, A. Nakamura, K. Hirosawa, H. Liao, T. Ushida, K. Goda, F. Kannari, and I. Sakuma, “Sequentially timed all-optical mapping photography (STAMP),” Nat. Photonics 8(9), 695–700 (2014). [CrossRef]
19. T. Suzuki, F. Isa, L. Fujii, K. Hirosawa, K. Nakagawa, K. Goda, I. Sakuma, and F. Kannari, “Sequentially timed all-optical mapping photography (STAMP) utilizing spectral filtering,” Opt. Express 23(23), 30512–30522 (2015). [CrossRef]
20. T. Suzuki, R. Hida, Y. Yamaguchi, K. Nakagawa, T. Saiki, and F. Kannari, “Single-shot 25-frame burst imaging of ultrafast phase transition of Ge2Sb2Te5 with a sub-picosecond resolution,” Appl. Phys. Express 10(9), 092502 (2017). [CrossRef]
21. D. Magrin, E. Giro, F. Bortoletto, G. Crimi, C. Pernechele, and R. Tomelleri, “Design and construction of a focal plane slicing mirror,” Proc. SPIE 7018, 701837 (2008). [CrossRef]
22. Y. Suematsu, K. Saito, M. Koyama, Y. Enokida, Y. Okura, T. Nakayasu, and T. Sukegawa, “Development of compact metal-mirror image slicer unit for optical telescope of the SOLAR-C mission,” Proc. SPIE 9904, 990411 (2016). [CrossRef]
23. M. Tamamitsu, Y. Kitagawa, K. Nakagawa, R. Horisaki, Y. Oishi, S. Morita, Y. Yamagata, K. Motohara, and K. Goda, “Spectrum slicer for snapshot spectral imaging,” Opt. Eng. 54(12), 1 (2015). [CrossRef]
24. Y. Kitagawa, Y. Yamagata, S. Morita, K. Motohara, S. Ozaki, H. Takahashi, M. Konishi, N. M. Kato, Y. Kobayakawa, Y. Terao, and H. Ohashi, “Fabrication of a wide-field NIR integral field unit for SWIMS using ultra-precision cutting,” Proc. SPIE 9912, 991225 (2016). [CrossRef]
