Femtosecond Laser Direct‐Writing Ablation of Transparent Fluoropolymer Toward Super‐Resolution Imaging of Cell Movements

Live imaging of living cells in 3D micro/nano‐environment is important for advanced biological studies. An issue for super‐resolution imaging is refractive index mismatch between the materials used for imaging platform and medium containing cells (water). Fluoropolymer CYTOP is the ideal platform material due to its refractive index close to that of water. Herein, femtosecond laser (220 fs) is exploited to ablate CYTOP at three laser wavelengths of 257, 515, and 1030 nm. Linear power dependences of the ablation rate obtained for all wavelengths indicate that ablation mechanisms of CYTOP are attributed to the production of fragmented CYTOPs (frag‐CYTOPs) inside the film to induce efficient light absorption for subsequent laser pulses. The frag‐CYTOPs can effectively absorb UV–Vis light, whereas those left behind generate green fluorescence. Importantly, the shortest wavelength of 257 nm offers the best ablation quality in terms of surface smoothness and low fluorescence with the highest ablation rate. We elucidate that the refractive‐index matching of CYTOP/water is effective for clear imaging of shapes/fluorescence of cells on 3D structures, e.g., HeLa cells on V‐shaped trenches. The fabrication of a microscopic 3D fluorescence code tag embedded in the CYTOP films is further demonstrated.


Introduction
Live imaging of cell movements in 3D nano/micro environment is important in investigations of the mechanisms of immune systems or disease progression, and has been attracting much attention, especially in efforts to understand cancer cell invasion into other tissues, [1,2] neutrophil migration at the sites of infection, [3] and axon extension of neurons in brain development. [4] The invasion of prostate or other cancer cells with a greatly deformed shapes in nanochannels has been observed by high-resolution imaging using an inverted microscope with a 3D nanofluidic biochip. [1,2] The forward and reverse migration of neutrophils through a collagen hydrogel was tracked with a 3D microfluidic device using time-lapse microscopy. [3] Axon extension of neurons through microchannels with a chemical gradient or a unidirectional configuration has been investigated using various 3D microfluidic chambers; [4,5] in addition, and further, organelle movements in individual cells such as mitochondria or the trans-Golgi apparatus have been observed using high-resolution confocal microscopes. [6,7] New approaches of nanoparticle-based bioprobes have also been reported to provide high spatial resolution at subcellular levels. [8] Further details of cell/organelle movements are expected to be obtained using super-resolution imaging techniques such as stochastic optical reconstruction microscopy (STORM), [9] single-molecule localization microscopy (SMLM) or photo-activated localization microscopy (PALM), [9] and spinning-disk super-resolution confocal live imaging microscopy (SCLIM). [10] A remaining problem impeding super-resolution imaging of cell/organelle movements in 3D microfluidic devices is the refractive-index mismatch between the materials used for devices and liquid media containing cells (typically water). Almost all the microfluidic devices used in the aforementioned studies were constructed from glass or polydimethylsiloxane (PDMS), both of which have a larger refractive index (1.517 and 1.412 for glass and PDMS, respectively) than water (1.333). Such a refractive-index mismatch distorts and degrades high-magnification images through reflection and refraction, especially when light from a fluorescent marker passes through an irregular interface in the 3D device structure. A highly transparent material having a refractive index similar to that of water is therefore desirable for super-resolution imaging of cell/organelle movements. The ideal material satisfying those requirements will be amorphous fluoropolymer CYTOP (AGC, Japan), [11,12] which exhibits excellent transparency and a refractive index (1.334-1.340) close to that of water.
Fabricating 3D nano/microstructures in CYTOP has been challenging because the conventional fabrication techniques for glass or PDMS are not well suited to chemically stable and fragile CYTOP. Short-pulsed laser ablation should be a suitable technique because it offers features of high speed, high flexibility, and high fabrication resolution. The laser with a wavelength shorter than 170 nm (absorption edge of CYTOP) is required to ablate CYTOP with a single photon process. Obata et al. fabricated microfluidic trenches in CYTOP using 157-nm nanosecond F 2 laser ablation, where smooth half-pipe trenches with a width and depth of ≈200 μm were formed. [13] They demonstrated DNA electrophoresis by enclosing the half-pipe trench via CYTOP cap bonding. A disadvantage of their approach is that, to avoid the absorption of laser light by the lens and/or O 2 molecules in air, it requires a hygroscopic CaF 2 lens and ambient N 2 gas for the ablation system. More importantly, F 2 laser providing enough energy for material processing is currently not commercially available. Alternatively, ultrafast laser provides capability of processing CYTOP through nonlinear multiphoton absorption. [14][15][16] Hanada et al. fabricated 3D microfluidic channels in CYTOP using 775-nm femtosecond laser ablation and observed the swimming movements of a Dinoflagellate cell in the channel. [17,18] However, the as-ablated surfaces of CYTOP were rough, with a thick damaged layer remaining: this layer must be removed via a lengthy (> 30 h) solution process, followed by thermal annealing.
In this paper, we report single-step femtosecond laser ablation of CYTOP, where smooth as-ablated surfaces can be obtained with a conventional lens under ambient air without post treatments such as solution etching and thermal annealing. We also investigated how ablation of CYTOP with a femtosecond pulse laser proceeds by carrying out spot irradiation using a fixed number of laser pulses. Significantly, we demonstrated the effect of refractive index matching for CYTOP/water by observing GFPlabeled HeLa cells on V-shaped trenches with a width and depth of 12 and 35 μm, respectively. A clear differential interference contrast (DIC) and fluorescence image of the cells was obtained with a confocal fluorescence microscope, showing the ability of developed technique to fabricate a platform for super-resolution imaging of living cells. As another potential application of femtosecond laser processing, we demonstrated the fabrication of a microscopic 3D fluorescence code tag embedded in the CYTOP films.

Wavelength Dependence of CYTOP Ablation
The surface morphology, fluorescence image, and depth profile for 300 × 100 μm 2 ablated areas processed with a wavelength and laser power of 1030 nm and 200 mW, 515 nm and 80 mW, and 257 nm and 7 mW, respectively, are shown in Figure 1a-c; all of the samples were prepared with a focusing position of z = 0.0 μm, i.e., the visible-light focal point was at the surface. Each laser power was selected to produce a similar ablation depth of ≈40 μm. A rough surface and strong fluorescence were observed with 1030-and 515-nm laser ablation, whereas a smooth surface with negligible fluorescence was obtained with 257-nm laser ablation. The depth profiles in Figure 1a-c show a sloped sidewall where laser scanning began, a moderately inclined bottom, and a steep edge where laser scanning ended. As we show later (Figure 2a), the ablated debris was deposited onto the sidewalls, corners, and the bottom as ablation proceeded with scanning the laser beam, causing a slope and a moderate incline at these faces.
A broad fluorescence spectrum with a peak wavelength of ≈510 nm was observed for the ablation areas subjected to 1030and 515-nm processing, as shown in Figure 1d. Similar but weaker fluorescence was also observed for 257-nm processing ( Figure 1c and Figure 2c). Because the original CYTOP shows no fluorescence, the observed bluish-green fluorescence was attributed to laser-induced photochemically and/or photothermally modified CYTOP, such as fragmented CYTOP debris or damaged CYTOP with dangling bonds, which we refer to as fragmented CYTOP (frag-CYTOP, hereafter) for simplicity. The fluorescence spectrum could be deconvoluted into four Gaussian peaks (Figure 1d), indicating that various frag-CYTOPs were generated by the laser irradiation. Notably, the fluorescence was observed under 473-nm excitation, which implies that the frag-CYTOPs effectively absorb UV-Vis light, unlike the original CYTOP. The lower fluorescence intensity and smoother surface in the 257-nm case (inset in Figure 1e) indicate that this wavelength is preferable to 1030-and 515-nm laser light for high resolution bioimaging using the 3D structured CYTOP.
In Figure 1e, a linear dependence of the ablation depth on the laser power is observed for all three wavelengths. The ablation efficiency at 1030, 515, and 257 nm was evaluated to be 0.23, 0.62, and 5.5 μm mW −1 , respectively, revealing higher efficiency at shorter wavelength, as expected. The difference in numerical aperture (NA) and the wavelength of the laser in the experiments can significantly impact the size of the laser beam spot on the sample. However, this difference proves to be insignificant in our experiment and analysis, as evidenced by the plateau observed in Figure 2d for ablation depth. This observation indicates that the defocusing of the laser spot has a minimal effect on the ablation results, as compared to the total power supplied per unit area. We also evaluated the laser fluence for each wavelength later in discussion section ( Figure S1, Supporting Information).
When laser scanning was repeated in the same ablation area, the ablation depth increased with increasing number of scans (Figure 2a,b). As the number of scans increased, the depth profile tended to deviate from a rectangular shape to a V-shape (Figure 2a), whereas the ablation depth increased from 50 μm for single scan to ≈130 μm for five scans (Figure 2b). We examined the removal of redeposited debris by rinsing the sample in ethanol with ultrasonic agitation for 15 min. The debris removal changed the depth profile to more closely resemble a rectangular shape, although the profiles corresponding to four or five scans were still inclined trapezoids. Figure 1. a-c) Surface morphology, fluorescence image, and depth profile for a 300 × 100 μm 2 ablated area processed with a wavelength and laser power of a) 1030 nm and 200 mW, b) 515 nm and 80 mW, and c) 257 nm and 7 mW, respectively. The surface morphologies are the reflection images of the surfaces taken with a 405-nm laser confocal microscope. Fluorescence images were obtained with a fluorescence microscope equipped with a WBS (Olympus) mirror unit. The depth profile was derived as a cross-sectional line profile along the 100-μm edge of the ablated area. The horizontal and vertical scale bars in the depth profiles indicate 100 and 20 μm, respectively. Ablation proceeded with repetition of a 300-μm line scan with a scan speed of 200 μm s −1 , followed by 2-μm spacing in the direction of the arrow in the depth profiles. d) The fluorescence spectrum observed at a 300 × 100 μm 2 ablated area processed with a laser with wavelength and laser power of 515 nm and 80 mW, respectively. Note that the original CYTOP shows no fluorescence; therefore, the fluorescence was attributed to photochemically and/or photothermally modified products resulting from laser ablation of CYTOP, such as fragmented CYTOP debris or damaged areas with dangling bonds. Four deconvoluted Gaussian peaks and a restored curve are shown for reference. e) Dependence of the ablation depth on the laser power for beams with a wavelength of 257, 515, and 1030 nm and a focusing position z = 0.0 μm. The inset shows the difference in surface smoothness (arb. units), the inverse of the brightness intensity of surface image taken with a 405-nm laser confocal microscope, and fluorescence intensity (arb. units) among the three wavelengths, corresponding to (a-c). The noise level of the fluorescence intensity was subtracted. Higher (lower) bar in the smoothness histogram corresponds to a smoother (rougher) surface. See also Figure S1   Figure S2 (Supporting Information). When placing the focusing position above the surface (z > 0), the ablation depth decreased, the surface became smoother (although some irregular patterns appeared at z > 60 μm), and fluorescence remained low. When the focusing position was moved beneath the surface (z < 0), the ablation depth decreased rapidly, becoming negative when z < −50 μm. Negative ablation depth values indicate that the surface is swelling in the irradiated area, suggesting that the material beneath the surface is expanded by laser irradiation. The expansion was accompanied by surface roughening and a remarkable increase in fluorescence. Transmission video image recorded in situ during laser ablation at z < −50 μm shows that unstructured round regions were formed and moved inside the film under laser irradiation (not shown here). This observation strongly suggests that the frag-CYTOPs were in the liquid phase during laser irradiation. A plateau of the ablation depth (33-38 μm) was observed in the focusing-position range of −30 < z < 20 μm. When the focusing position z was 100 μm above the surface, the ablation depth becomes less than 5 μm in Figure 2d. On the other hand, the ablation depth increased from 105 μm at fourth scan to 127 μm at fifth scan in Figure 2b, suggesting that scan repetition enhances the ablation efficiency, as discussed in the following section.
It is natural that the frag-CYTOP produced by laser-induced chemical reaction has a refractive index slightly different from that of original CYTOP, although we have not yet evaluated the refractive index of a thin frag-CYTOP layer due to technical difficulty. Nonetheless, we consider that mismatch between frag-CYTOP and water could be remarkably smaller than those between glass and water or PDMS and water, since the chemical composition of frag-CYTOP should be close to that of the origi-nal CYTOP and frag-CYTOP should be amorphous as well as the original CYTOP.

Ablation Process with 257-nm Femtosecond Laser
We further investigated ablation mechanisms with 257-nm wavelength laser, which is the best choice to fabricate a platform using CYTOP for super-resolution imaging of living cells, as revealed by the results shown above. To this end, the progress of CYTOP ablation with a 257-nm femtosecond laser was visualized by spot irradiation using a fixed number of laser pulses with a power of 8 mW, corresponding to a laser fluence of 3.2 J cm −2 , with a 100 kHz repetition rate (assuming a laser spot size of 2.0 μm). Figure 3a,b shows the ablation traces obtained by spot irradiation with pulse numbers between 100 and 800k. No traces were observed in a height image (Figure 3a) for pulse numbers of 10k or less. However, when a DIC image was obtained with a confocal microscope with an observation depth z of −42 μm (Figure 3b), a small spot with clear contrast was observed for pulse numbers as low as 100. Such a laser-induced small spot trace was observed even with a single shot (not shown here). These results indicate that 257-nm laser light was focused 42 μm beneath the surface when the focusing position z was set to zero for visible light. A laser-induced reaction (fragmentation of CYTOP) occurs The fluorescence intensity was estimated by summing the image intensity in a cylindrical volume around each spot (i.e., a cylinder with a horizontal radius of 21 μm and the whole depth range in (c) (from 23.9 to −130.2 μm), whereas the ablation was calculated by summing the black/white pixels in the 8-bit grayscale range of <130 or >200 (background grayscale was 165 on average) in the same cylindrical volume at each spot.
by multiphoton absorption at the laser-beam focusing point, resulting in a clear spot at this point; by contrast, no trace was formed on the surface for pulse numbers as high as 20k.
The spatial distribution of a fluorescence/transmission DIC image observed when the observation depth z was varied is presented in Movie S3 (Supporting Information), where underlying fluorescent spots with fluorescence are clearly observed for a certain depth range. The transmission image contrast represents the change in refractive index caused by the frag-CYTOPs as well as the formation of voids for a large number of pulses. We further visualized the fluorescence distribution together with the transmission DIC image as cross-sectional depth profiles in Figure 3c.
The fluorescence intensity and the volume of frag-CYTOPs and voids ("ablation" for simplicity in Figure 3d) were estimated from the depth profile and are plotted in Figure 3d. Fluorescence generated by frag-CYTOPs was observed for the spot generated by 100 pulses; it grew larger, mostly in the upward direction, as the pulse number was increased to 10k. When the pulse number was less than 20k, the volume of the frag-CYTOPs was limited inside the film and no trace appeared on the sample surface. At 20k pulses, the volume reached the surface and outward eruption of ablated frag-CYTOPs began. The dark areas in Figure 3c correspond to voids formed in the CYTOP film. When the number of pulse shots was greater than 20k pulses, the dark areas increased both in diameter and depth, and the fluorescence intensity decreased simultaneously. These results reveal that fluorescent frag-CYTOPs were removed by outward eruption. The change proceeded for at least 800k pulses.

Cell Observations Using Microfabricated CYTOP
We characterized the effect of refractive index matching for CY-TOP/water by observing cells placed on V-shaped trenches fabricated in CYTOP by the 257-nm femtosecond laser. Figure 4 shows a confocal fluorescence microscopy image, where HeLa cells GFP-tagged for cytoplasm were observed when we focused on a flat surface of CYTOP (Figure 4a,b) or on a V-shaped trench (Figure 4c,d). The trench width and depth were 12 and 35 μm, respectively. A similar observation was carried out for a glass substrate with trenches with a width and depth of 18 and 16 μm, respectively. (Figure 4f-i).
The shape of the cell on the flat glass surface is clearly observed in Figure 4f (cell C), and strong fluorescence is seen (Figure 4g). However, the shape of the cell on a glass trench (the width and depth were 18 and 16 μm, respectively) was not identifiable (Figure 4h) and the fluorescence was weak (Figure 4i) (cell D). These results are attributed to a rough glass/water interface with a large refractive index mismatch distorting the cell shape in DIC imaging, and to random reflection of fluorescence light at the glass/water interface weakening fluorescence intensity. By contrast, the shape of the cell on the CYTOP trench is clearly observed (Figure 4c) (cell B), together with strong fluorescence (Figure 4d), because of refractive index matching for CYTOP/water. The dependence of the image sharpness and fluorescence intensity on the focusing position is shown in Figure 4e. The sharpness represents the depth distribution of a small organelle in the cell, whereas fluorescence indicates the depth profile of the cell volume. The depth profiles show that cell A was viewed from its top, whereas cell B was viewed nearly from its side.
No fluorescence of residual frag-CYTOP was observed at the trench interfaces in Figure 4d, showing that the weak florescence observed in Figure 1c is weak and does not disturb the fluorescent imaging of cells. Spectroscopic separation of GFP or RFP fluorescence also contributes to the elimination of frag-CYTOP fluorescence.
We provided the 3D images for cell-A and -B on CYTOP as Figure 5, along with 360-degree rotation 3D-image Movie S4 (Supporting Information). The cell-B in the trench showed a larger spreading area and shorter height compared to cell-A on the flat surface of CYTOP, because the cell-B deformed its body according to the smaller width of CYTOP V-trench (only 12 μm in width as shown in Figure 5c). The observation reveals that HeLa cells have a capability to slip into gap spaces narrower than its original body height to invade into other tissues.

Microscopic 3D fluorescence Code Tag
The strong fluorescence of frag-CYTOPs inside the film shown in Figure 3c and Movie S3 (Supporting Information) opens a new application of laser-processing of CYTOP for microscopic 3D fluorescence code tags and 3D data storage. The fluorescence spots generated by 257-nm laser irradiation with 2k pulses (8 mW) in Figure 3c exhibit a typical diameter and height of 3.4 and 15 μm, respectively. Multilayer irradiation of the spot-pattern (dot matrix) created using a 257-nm laser beam with a spacing greater than the typical size of fluorescence spots should result in a multiply stacked fluorescent spot array. We demonstrated this effect by designing nine characters (RIKEN and four expanding star marks) in a 9 × 9 dot matrix, and performed irradiation using laser-beam spots with a horizontal (x and y) and vertical (z) spacing of 16 and 10 μm, respectively. The irradiated area exhibited no surface traces as in the case of Figure 3a. However, the embedded characters (codes) were observed clearly with a confocal fluorescence microscope (Movie S5, Supporting Information). This demonstration indicates that microscopic 3D fluorescence tags embedded in CYTOP films can be realized by 257-nm femtosecond laser processing. The size of the pattern and vertical spacing can be reduced further by tuning the laser wavefront and using a larger NA objective lens. Reading the fluorescence code requires a confocal fluorescence microscope, thereby restricting who can read the code tag.

Discussion
In many previous reports on short-pulse multiphoton laser ablation of polymer films, the power dependence was reported to be nonlinear, with a threshold and saturation at higher fluences: Kumagai et al. reported the fluence threshold in 798-nm femtosecond laser ablation of Kapton polymer. [19] Küper et al. reported a nonlinear dependence of the ablation rate on fluence for 193-351-nm nanosecond laser ablation of Kapton polymer. [20] Ibrahim et al. showed a nonlinear dependence of ablation depth on the incident laser power for poly(methylmethacrylate), polystyrene, poly(butyl methacrylate), www.advancedsciencenews.com www.advmattechnol.de and poly[2-(3-thienyl)ethoxy-4-butylsulfonate]. [21] Given the reports on other polymer materials, [22][23][24][25][26][27][28][29][30][31][32][33] our result of a linear power dependence suggests that a unique mechanism is involved during ablation of CYTOP by femtosecond lasers. In the ablation experiments of Figure 1, we used two objective lenses for the three wavelengths, leading to three different diffraction-limited spot sizes. The diameter of focused laser spot was estimated theoretically [34] as 3.93, 1.97, and 0.98 μm for the wavelength of 1030, 515, and 257 nm, respectively. In the estimation, we used the effective focal length of 10 mm for the two lenses and assumed 4.0 mm and 1.2 for the laser beam diameter and laser beam quality (M 2 ). Using the estimated spot sizes, the data in Figure 1e was replotted in Figure S1 (Supporting Information) as the dependence of ablation depth per pulse on laser fluence.
Our study revealed that ablation of CYTOP by a femtosecond laser involves multiple processes. At the beginning, the laser shots modify the CYTOP photochemically and/or photothermally via multiphoton processes, producing frag-CYTOPs, which absorb more UV-Vis light, generate fluorescence, and exhibit a different refractive index compared with the original CYTOP. Then, the frag-CYTOPs absorb laser light more efficiently and promote the fragmentation of surrounding CYTOP by forming a temporary liquid phase inside the CYTOP film. When the volume of melted frag-CYTOPs reaches the surface, the frag-CYTOPs are eruptively ablated outward, leaving nonfluorescent voids on the surface. We expect that some amount of frag-CYTOPs remain at the bottom of the voids, as suggested by the residual fluorescence intensity in Figure 3c when the pulse number is greater than 100k. The residual frag-CYTOPs promote further laser ablation of CYTOP at depths at least as large as z < −100 μm (Figure 3c), because the frag-CYTOPs are evaporated instantly at the bottom of the voids. In total, high-efficiency laser ablation of CYTOP is mediated by frag-CYTOPs, resulting in a linear dependence of the ablation depth on laser power in Figure 1e and in the multiple scan effects shown in Figure 2a,b. The fragmentation of CYTOP depends on the laser wavelength, i.e., shorter wavelength lasers more effectively produce frag-CYTOPs and evaporate them.
Three critical factors influence this ablation mechanism. First, CYTOP at the initial stage is highly transparent to radiation with wavelengths as short as 170 nm. Fragmentation of CYTOP therefore occurs by multiphoton absorption only at the focusing point of the laser beam. In our experiments, the focusing point was 42 μm beneath the substrate when the focusing position was set to zero for visible light, which led to frag-CYTOPs remaining inside the film instead of being ejected. This behavior is contrary to that in cases discussed in previous reports related to multiphoton laser ablation of polymers. [22][23][24][25][26][27][28][29][30][31][32][33] Second, the frag-CYTOPs have a far larger absorption coefficient than the original CYTOP; the volume of the frag-CYTOPs therefore increases more effectively as a result of further laser pulses. This volume change functions as positive feedback for frag-CYTOP production. Third, after the outward eruption of the frag-CYTOPs, the frag-CYTOPs produced at the bottom region of the void evaporate instantly, leading to the nonsaturated power dependence in Figure 1e and Figure 2b.
Although the frag-CYTOPs play a key role in femtosecond laser ablation of CYTOP, their details (e.g., molecular weight, chemical structure, or absorption spectrum) were not elucidated in the present study. The fact that the frag-CYTOPs are excited by 405-nm light and emit fluorescence reveals that they absorb visible light better than the original CYTOP. In addition, the fluidic behavior observed inside the film during laser irradiation suggests that the frag-CYTOPs have lower molecular weights than the original CYTOP. Furthermore, the frag-CYTOPs continued to emit fluorescence (no bleaching) for more than a couple of months, indicating that they are chemically stable under ambient conditions.
The effectiveness of refractive-index matching for CY-TOP/water was clearly demonstrated by the observation of cells on V-shaped trenches. Fluorescence generated in CYTOP by the 257-nm femtosecond laser is sufficiently low to capture clear and sharp images of tagged cells. By contrast, the demonstration on glass suggests that the cell shape cannot be clearly observed when the refractive index of microstructure around the cell is mismatched to that of water. The microstructures of CYTOP are useful for obtaining clear image of small organelles in cells from various angles ( Figure 5, cell B). This study verifies that CYTOP microstructured by the 257-nm femtosecond laser is highly promising as a platform for super-resolution imaging of living cells. The primary objective of the cellular experiments here is to demonstrate the superiority of CYTOP over glass for observing cells on 3D structures. We focused on refractive index matching rather than the size of the 3D structures in this study, although smaller 3D structures are of interest in the investigation of complex cell behaviors, as reported with some laser structured polymer surfaces of PLLA or Kapton. [35,36] We also demonstrated a microscopic 3D code tag, where the generation of fluorescent frag-CYTOPs was used for fluorescence spot writing inside a CYTOP film, which might have new industrial applications (e.g., as a depth scale for microscope observations, or a 3D fluorescence standard sample). Further, the visualization of laser beam profile along the incident direction z, which cannot be achieved with conventional x-y laser beam profilers, can be realized through the generation of fluorescent frag-CYTOPs around the focusing point in a CYTOP film, providing an effective method of estimating the quality of the laser beam.

Conclusion
The present study has revealed the basic characteristics of femtosecond laser ablation of CYTOP, in terms of wavelength, power, and focusing position dependencies. Among the three wavelengths of 257, 515, and 1030 nm, the shortest wavelength of 257 nm is desirable to fabricate 3D structures of CYTOP, since smoother ablated surfaces are obtained with less byproduct fluorescence. The ablation rate exhibits a linear power dependence for all wavelengths, indicating the unique ablation mechanisms of CYTOP that the production of frag-CYTOPs within the film induces efficient light absorption for subsequent laser pulses. We demonstrated the observation of HeLa cells on V-shaped trenches of CYTOP and glass, and elucidated that the refractive-index matching of CYTOP and water significantly improves the imaging quality of cellular shapes and fluorescence on 3D structures. We further demonstrated the production of a microscopic 3D fluorescence code tag integrated in the CYTOP films. Although the details of frag-CYTOPs such as their molecular weight, chemical structure, and absorption spectrum remain to be elucidated in subsequent investigations, the two demonstrations we presented here indicate that femtosecond laser processing of CYTOP has strong potential for use in cell biology and in industrial fluorescence markers or a laser beam profiler.

Experimental Section
The femtosecond laser source used in the present study was a fully automated industrial design laser (Pharos, Light Conversion) emitting 1030 nm laser light with a pulse width and repetition rate of 220 fs and 100 kHz, respectively. The second (515 nm) and fourth (257 nm) harmonics were generated using a harmonic generator (HIRO, Light Conversion). The laser beam irradiated a CYTOP sample surface through a 20× objective lens with a NA of 0.40 (Mplan Apo NIR, Mitsutoyo) and 0.36 (PFL-20-UV-AG-A, Sigma Optics) for 1030/515 and 257 nm, respectively. The two lenses have the identical effective-focal length of 10 mm. The laser power was measured using an optical power meter (Vega, Ophir Photonics) equipped with a thermal sensor (2A-BB-9, Ophir Photonics) positioned at the sample surface. The focusing position was measured on the basis of a video image of the sample surface observed through the objective lens under visible illumination. In the present report, the focusing position z represents the distance to the focusing point as measured from the surface under visible light, where z was negative (positive) when the focusing point was inside the sample (above the surface).
The CYTOP substrates were films with a thickness of 200-500 μm, which was prepared using a commercially available CYTOP solution (CTX-809SP2, AGC) through thermal solvent evaporation in a flat glass dish at a temperature of 230°C. The glass dish was maintained at this temperature for a duration of 72 h to improve the surface planarity of the CYTOP thin film. Subsequently, the films were detached from the glass dish and sectioned into square substrates (≈15 × 15 mm). A CYTOP substrate was placed on a PC-controlled XYZ stage (KWC06020-LCA, Suruga Seiki) to be scanned under the focused laser light. The resolution of the XYZ stages was 0.2 μm (single axis resolution, repeated positioning accuracy). The sample to ablate was scanned a rectangular area of 300 × 100 μm 2 with a typical scanning speed and line spacing (L/Spacing) of 200 μm s −1 and 2.0 μm, respectively. We also examined the dependence of ablation and fluorescence on the number of pulses in spot-matrix experiments, where 4 × 4 matrix spots with 50 μm spacing were irradiated with 100, 200, 400, 800, 1k, 2k, 4k, 8k, 10k, 20k, 40k, 80k, 100k, 200k, 400k, and 800k laser pulses with a wavelength of 257 nm.
A schematic of the experimental and evaluation methods for the laser scanning experiments was shown in Figure 6. The ablated areas were measured with a laser scanning confocal 3D microscope (OLS5000, Olympus) and a fluorescence microscope (BX51, Olympus). The laser scanning confocal 3D microscope evaluates the surface topography of a 3D structure by detecting the surface through the reflection of laser at the interface between CYTOP and air. The measurement was not affected by weak residual fluorescence. The fluorescence and surface smoothness for a rectangular area of 300 × 100 μm 2 were evaluated by summing RGB values for a centered region of ≈200 × 50 μm 2 in the fluorescence image and the confocal reflection intensity image (Figure 6b). Cross-sectional depth profiles of the ablated area were obtained by averaging depth data obtained with the confocal microscope for a 120 μm-wide band across the 300 × 100 μm 2 area. The surface smoothness was calculated as the reciprocal of the averaged brightness for a centered region of ≈200 × 50 μm 2 in the confocal microscope image. A higher (lower) value of surface smoothness corresponds to a smoother (rougher) surface. For the fluorescence spectrum measurement, the sample was excited with a halogen lamp whose light was dispersed by WUS (Olympus) mirror unit, and the spectrum was measured with a fiber optic spectrometer (USB2000, Ocean Optics).
The spot-matrix ablation samples were observed with a 473-nm laser scanning confocal biological microscope (FV1000, Olympus) equipped with a water-immersion objective lens (LUMPLFLN60XW, NA = 0.80, WD = 2.0 mm, Olympus) or with a 405-nm laser scanning confocal fluorescence microscope (FV3000, Olympus) equipped with a water-immersion objective lens (UPLSAPO60XW, NA = 1.20, WD = 0.28 mm, Olympus). The fluorescence generated around each spot was estimated by summing the image intensity in a cylindrical volume at the spot, i.e., a cylinder with a horizontal radius of 21 μm and a height equal to the depth range measured (388 image slices corresponding to a depth from 23.9 to 130.2 μm). The production of laser-induced photochemically and/or photothermally modified CYTOP was estimated by summing the black/white pixels out from an 8-bit grayscale range of 130-200 (background grayscale was 165 on average) in the same cylindrical volume at each spot.
For cell observations, V-shaped trenches of CYTOP with a width of 12 μm and depth of 35 μm were fabricated by 257-nm laser ablation at a power of 8 mW and a scanning speed of 50 μm s −1 . For comparison, a similar V-shaped trench with a width of 18 μm and depth of 16 μm was fabricated in soda-lime glass by 257-nm laser ablation at a power of 800 mW and a scanning speed of 10 μm s −1 . For the evaluation of depth of the Vtrenches, we filled them with a red fluorescent solution of rhodamine B, and observed the depth profile with a confocal fluorescence microscope (FV1000, Olympus). Only red fluorescence was extracted by filtering out the residual fluorescence of frag-CYTOP, as shown in Figure 5c for CYTOP and Figure 5d for glass sample. The depth of V-trenches was estimated from the images, as mentioned above. The cross-sectional depth profile of CYTOP trenches exhibited a double-tapered and inclined shape, rather than a simple V-shape. The inclination (5−7 degree) is due to a slight misalignment of the beam axis. The observed red fluorescence for the glass V-trench was significantly weaker than that for the CYTOP V-trench, due to the large refraction at the interface of glass and water.
After the substrates were immersed for 2 h in a Laminin solution, two types of cancer cells were seeded to the substrates with V-shaped trenches: HeLa cells with GFP-tagged cytoplasm and PC3 cells with red fluorescent protein (RFP)-tagged histone. A reduced quantity of mixed cells comprising GFP-Hela and RFP-PC3 was seeded, as compared to the confluent condition, to obtain single cells positioned in the trenches, for both CYTOP and glass samples. Cell observations were carried out after 18-24 h incubation in a 5% CO 2 ambient atmosphere. A water-immersion objective lens (UPLSAPO60XW, NA = 1.20, WD = 0.28 mm, Olympus) was used with a 473-nm laser scanning confocal biological microscope (FV1000, Olympus). As evidenced by the large area images in Figure S6 (Supporting Information), the distribution of cells on both CYTOP and glass samples was comparable and non-uniform. Although RFP-PC3 cells were also present on the CYTOP sample, they were not situated unfortunately in the region of Figure 4a-d.
We utilized three distinct combinations of two fluorescence microscopes (FV1000, FV3000) and two objective lenses (LUMPLFLN60XW, NA = 0.80, WD = 2.0 mm, or UPLSAPO60XW, NA = 1.20, WD = 0.28 mm) for the observation of spot-matrix fluorescence and cells, depending on the availability of microscopes and lenses. No essential difference was re-sulted by the selection of microscopes and lenses among these. We employed differential interference contrast (DIC) technique to observe frag-CYTOP in the spot-matrix experiment and cells on CYTOP or glass. The DIC technique converts specimen optical path gradients into amplitude differences and visualizes the difference as improved contrast in the resulting image.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.