1. Introduction

With tightly focused, ultrashort laser pulses, femtosecond laser micromachining can locally modify the refractive index (RI) of bulk materials via multiphoton absorption. Through careful control of the micromachining parameters, such as laser repetition rate, laser power and scanning speeds, three-dimensional optical devices can be created in different types of materials. In 2006, our group first reported that RI change up to 0.06 ± 0.005 could be induced in silicone-based and non-silicone-based hydrogels using a high repetition rate femtosecond laser [1]. Intra-tissue Refractive Index Shaping (IRIS), a new paradigm for refractive vision correction in which corneal refractive properties can be modified without flap cutting or tissue ablation, was then proposed and demonstrated in excised cornea and live cats in vivo [2]. Following the success of creating phase structures in various ophthalmic hydrogels and corneal tissue at laser repetition rate of 80 MHz; we now find that operating at lower repetition rates enables us to obtain much higher RI changes at much lower average powers [3]. This enables fast customization of commercial refractive devices; therefore, it is critical to characterize the post-operation effects and to fully understand the mechanisms underlying the RI changes.

It has been verified that negative RI changes, calculated through phase change, can be induced in hydrogel materials. Change in wavefront phase (in units of waves) is expressed as $\Delta \phi = \frac{{({\Delta n} )\,d}}{\lambda }$, where Δn is the index difference, d is the length of index change region parallel to optical axis and λ is the radiation wavelength. RI changes of different magnitude are achieved in ophthalmic materials such as acrylic hydrophobic materials used for intraocular lenses, Poly(methyl methacrylate)-based (PMMA) copolymers and Hydroxyethyl methacrylate (HEMA)-based hydrogels [4–6]. However, the dependence of RI changes on laser micromachining conditions and the underlying mechanisms are still not completely understood. So far, researchers have different hypotheses for the interaction between femtosecond laser and materials. Baumthat et al. reported in PMMA that high density free-electrons, generated under high repetition-rate laser pulses, can directly cleave the polymer backbone and locally reorganize the polymer, leading to photochemically-induced RI changes [7]. Pradhan et al. argued that under high energy irradiation, water in hydrogels can undergo plasma formation via electron generation, leading to a rapid bubble formation. Photocavitation of the vapor bubbles can indirectly cause rupture of chemical bonds in the polymer chains, resulting in physical degradation of hydrogels in the focal volume [8]. Other mechanisms, such as photo-induced hydrolysis of polymer, color-center formation, stress-induced structural changes, densification via multiphoton ionization have also been proposed for the laser-induced changes in bulk materials under different processing systems [9–14]. In particular, the role of water in different laser operation regimes has been noticed in several studies. For example, water-confined ablation was reported to yield higher ablation threshold for nanosecond IR laser machining of glass-ceramics [15]. The heat affected zone, as well as machining defects like cracks and debris deposition, can also be significantly reduced [16–18]. Our previous study has observed a dramatic increase of water content in various hydrogel materials after micromachining [19], but further investigations need to be carried out to explain this phenomenon, and how this water content change is related to the material properties and laser exposure parameters.

With inherently narrow vibrational bands, Raman scattering allows the observation of vibrational spectrum of a specific chromophore in a complex system, thus it is an ideal tool for studying the origins, dynamics, and mechanisms of chemical changes in polymers [20,21]. Some researchers have used Raman spectroscopy to characterize the biocompatibility of contact lenses, to understand the chemical basis for altering polymers with femtosecond lasers, as well as the interaction between material surfaces and biological environment [14,22,23]. The development of quantitative Raman spectroscopy has demonstrated its accurate and repeated spectrometric analysis of different components such as water-dissolved hydrogen peroxide, catalyzed Heck reaction in liquid phase and silicic volcanic glasses [24–26]. The feasibility of using micro-Raman spectroscopy to resolve the laser-induced local chemical changes in ophthalmic hydrogels has therefore been demonstrated. Compared with conventional Raman spectroscopy, confocal Raman detection enables rejection of the Raman-scattered light from out-of-focus regions, therefore providing much better spatial resolution and signal-to-noise ratio [27,28]. Unlike Raman inelastic scattering which relies on the changes in polarizability of molecules, Fourier-transform infrared (FT-IR) spectroscopy measures the light absorption of materials caused by the changes in molecular dipole moment [29]. It is known that symmetric vibration and ring breathing modes normally exhibit strong Raman response, but are weaker or even nonactive in FT-IR spectra. However, several other asymmetric vibrations along bonds with a strong dipole moment are more easily detected by FT-IR [30]. As a result, Raman spectroscopy, together with FT-IR, could provide synergistic and complete vibrational characterization on the composition and microstructure of materials.

To study the material effects of femtosecond laser micromachining in ophthalmic hydrogels, single layer dense line patterns were inscribed into Contaflex hydrogels using a 405 nm femtosecond laser operating at 8.3 MHz repetition rate. Below the material saturation threshold, a phase change profile was built, upon which the nonlinear absorption coefficient could be calculated. With an edge-detection method, change in the local water content was able to be studied quantitatively via micro-Raman spectroscopy. Thermogravimetric analysis (TGA) has been proved to be useful for evaluating the decomposition kinetics and stability of various polymers. Therefore, this technique was applied to study the thermal stability, as well as to verify the observed water content change in written regions. Beside phase patterns, structural alterations in the damage regime were characterized by both Raman and FT-IR to understand the depolymerization mechanisms.

2. Methods

2.1 Femtosecond laser micromachining instrumentation

The femtosecond laser system used a custom Y-Fi fiber laser (KMLABS, Boulder, CO, USA) running at 405 nm, 8.3 MHz. After passing through a half-wave plate and beam expander, the pulse width was compressed to ∼190 fs by a Gires-Tournois Interferometer. The laser beam slightly overfilled a 0.45 numerical aperture (NA), air-immersion microscope objective (Olympus, Center Valley, PA, USA) and was focused into the central layer of samples (Fig. 1(A)). With hydrogel sandwiched between a microscope slide and coverslip, large-scale linear motion was realized by a two-axis translation stages (PRO115LM, Aerotech Inc., Pittsburgh, PA, USA), while vertical axis motion was achieved by another single-axis stage (GTS30 V, Newport Corporation, Irvine, CA, USA). Eighteen single-layer dense line patterns were successively inscribed into Contaflex GM Advance 58 hydrogels (Contamac, Grand Junction, CO, USA) ∼150 µm below the top surface, with 0.5 µm line spacing and beam intensity ranging from 20 mW to 230 mW. The travel length of each pattern was 20 mm to guarantee a constant writing speed of 200 mm/s inside samples. For hydrogels used for Raman measurements, each pattern was designed to be 200 µm wide with 120 µm separation; while for phase change quantification, the patterns were 40 µm wide with 80 µm separation. Contaflex GM Advance 58, also known as “Acofilcon A”, is a non-ionic hydrophilic polymer used for manufacturing soft lenses [31]. It is composed of high-purity Glycerol Methacrylate, Methyl Methacrylate (MMA), N-vinyl-2-pyrrolidone, 2-HEMA, 4-Methacryloxy-2-hydroxybenzophenone (MOBP), Reactive Blue 4 and crosslinked with Diallyl Maleate. With 58%-wt water content, it is transparent within 380 nm - 780 nm and has a RI of 1.402 at 20 °C. After writing, hydrogels were immersed in saline solution for two days before subsequent experiments.

 

Fig. 1. (A) Mounting scheme of laser micromachining in hydrogels. (B) Custom-built MZI used for phase change quantification.

Download Full Size | PDF

2.2 Phase change quantification

As mentioned in our earlier study, the induced phase changes were measured on a custom-built Mach-Zehnder interferometer (MZI) using a 543 nm He-Ne laser (Fig. 1(B)) [32]. With two identical microscope objectives (NA = 0.60, LUCPlanFLN40X, Olympus, Center Valley, PA, USA) on both the reference and testing arm, sample was placed on a three-axis translation stage and the interferogram from each pattern can be collected. The phase map corresponding to each interferogram was calculated using the carrier fringe method and the induced phase change was unwrapped with the Goldstein’s branch cut unwrapping algorithm [33,34]. To remove phase variations caused by aberrations in the incoming wavefront and interferometer, the phase map of the local background was measured from the non-written region adjacent to each pattern and was subtracted from calculation. All these calculations were conducted automatically in MATLAB (MathWorks, Natick, MA, USA). Laser micromachining and phase change quantification were repeated in three different hydrogels using the same procedure.

2.3 Quantitative spectroscopic study

The micro-Raman system used a laser diode module (LQC660-110C, Newport Corporation, Irvine, CA, USA) to produce a 660 nm excitation beam (Fig. 2(A)). After passing through a beamsplitter and a metallic variable neutral density filter, the beam was focused into the sample with an air-immersion objective (LUCPlanFL60X, Olympus, Center Valley, PA, USA), and the maximum output power was 75 ± 1 mW. The laser beam profile was measured using a knife-edge method and the effective excitation NA was calculated to be 0.25 [35,36]. A vertical translation stage was mounted on top of two single-axis linear translators (GTS30 V and GTS70, Newport Corporation, Irvine, CA, USA) to enable precise, three-axis motion control of samples. Illuminated with a white light-emitting diode, the sample position was monitored on the imaging arm using a CCD camera. The excitation beam and fluorescence background were blocked by a notch filter and two long-pass filters (NF658-26 and FELH0700, Thorlabs, Newton, NJ, USA). Raman scattering signals were finally collected by a customized, high-sensitivity spectrometer (QEPro, OceanOptics, FL, USA). The axial resolution was calibrated from the beam depth profile of a 2 µm-thick nitrocellulose film using signals of the 866 cm^-1 peak. With a 75 µm pinhole, the resolution was estimated by the full width at half maximum (FWHM) to be 12 µm (Fig. 2(B)). Hydrogels used for measurement were cut into ∼240 µm thick cross sections and Raman spectra were collected by laterally scanning across the width of each section with 1 µm step size. Each spectrum was averaged from 2 scans with 1 s integration time. To avoid dehydration, sections were immersed in saline solution and sealed between two quartz windows during the entire measurement. Motion control, spectra collection and baseline correction were all conducted in MATLAB.

 

Fig. 2. (A) Confocal micro-Raman system sketch. (B) Beam depth profile used for calibrating the axial resolution, obtained by scanning a 2 µm-thick film and extracting the 866 cm^-1 peak intensity at each axial position.

Download Full Size | PDF

FT-IR was additionally used to characterize the local chemical changes in damaged patterns. Patterns of 6 mm wide were written on the top surface of hydrogels to guarantee direct probe of the written layer. Transmittance spectra were recorded using a Shimadzu spectrometer (IRPrestige-21, Shimadzu Scientific Instruments, Columbia, MD, USA) in the range of 700 - 4000 cm⁻¹ with ± 4 cm⁻¹ resolution, and signals were averaged from 32 scans.

2.4 Thermal analysis

To investigate the thermal stability of the hydrogels after micromachining, 3 mm by 2 mm wide large phase change patterns were inscribed across the whole sample thickness. Each bulk phase pattern contained 25 layers with a 20 µm layer separation, and the phase change of a single layer was ∼0.51 waves at 543 nm. One of the bulk phase patterns and another pristine hydrogel piece of similar size were dehydrated for one week and stored in a desiccator before measurement. Each hydrated sample was blotted lightly with Kimwipes and immediately sealed in an aluminum hermetic pan. To allow water removal during heating, a pinhole was punched on the lid of the aluminum pan right before each measurement. To confirm bulk changes in water content, TGA was performed and repeated three times on both hydrated and dried samples between 40 °C and 600 °C in nitrogen atmosphere using a Q500 TGA analyzer (TA Instruments, New Castle, DE, USA).

3. Results

3.1 Phase change as a function of beam intensity

Uniform phase patterns were obtained when writing below 60 mW at 200 mm/s. With the power increasing, damage started to appear and expanded to the whole written region after 70 mW (Fig. 3(A)). However, the gross damage was highly confined within the written zone for power as high as 230 mW, and no heat-induced melting features such as micro-cavities, nano-holes or cracks were observed [37]. The rectangular patterns can be distinguished from substrate by the phase shift on the interferogram, with the retrieved phase maps showing negative phase changes (Fig. 3(B)). Below 60 mW, phase change increased nonlinearly with laser power increasing, and maximum phase change up to -1.26 waves at 543 nm, could be obtained (Fig. 3(C)). Fitting of phase change as a function of laser power obtained a line gradient of 1.741, indicating that the micromachining process under 405 nm was mainly dominated by two-photon absorption.

 

Fig. 3. (A) DIC images of patterns written at different power. Scale bar: 50 µm for all. (B) Interferograms and retrieved phase maps of the 40 mW (top) and 60 mW (bottom) pattern written on sample 1. (C) Phase change measured at 543 nm as a function of laser power, with a nonlinear fit and R² values included in the graph.

Download Full Size | PDF

3.2 Change in the local water content

The voxel of each pattern, appeared as black rectangles, were little bit blurry under differential interference contrast (DIC), possibly due to the anti-waveguide effect of the written volume (Fig. 4(A)). Both phase patterns and damaged ones were highly confined to the written volume, but patterns edges became rough when power was over 150 mW. Due to the materials saturation effect confirmed in our two-photon photochemical model [38], the thickness of the written layer tended to increase with the applied average power. The characteristic Raman peaks can be assigned to different vibrational modes based on the components of Contaflex GM 58 (Fig. 4(B)). The bands at 1156 cm^-1, 1236 cm^-1 and 1663 cm^-1 corresponded to ν(C-C), a combination vibration of ν(C-O) and ν(C-COO), and ν(C = C) respectively [39,40]. The band at 1506 cm^-1 was attributed to δ(-CH₂) and ν_ar(C-C), while the 1757 cm^-1 band arose from ν(C = O) [41–43]. In particular, the band at 3000 cm^-1 corresponded to ν(C-H) from the -CH₂ group [40,44], while the broad band between 3150 cm^-1 - 3700 cm^-1 was due to ν(-OH) in water molecules. This cumulative band included contribution from both the symmetric ν(-OH) at ∼3300 cm^-1 and the anti-symmetric ν(-OH) at ∼3400 cm^-1 [22,45].

 

Fig. 4. (A) DIC images of the cross sections. Scale bar: 50 µm for all. (B) Spectrum of Contaflex GM 58 showing all characteristic Raman bands. (C) Lateral scan profile of the cross section written at 40 mW. (D) Normalized spectra from the 40 mW cross section, from which a dramatic increase of the Int₃₄₅₆/Int₃₀₀₀ ratio was observed. (E) Change in local water content as a function of laser power, expressed as a percentage. (F) Representative spectra from damaged patterns demonstrating different chemical changes.

Download Full Size | PDF

Quantifying water content with the “internal calibration” method, in which Raman intensity of the water band is normalized to the intensity of lower wavenumber signature bands, has been reported to be less affected by the instrumental and analytical conditions [46,47]. Therefore, the relative water content can be quantified using the integral intensity of the 3000 cm^-1 and ∼3456 cm^-1 bands, and “Difference in Int₃₄₅₆/Int₃₀₀₀ ratio” was used to represent the local water content change, expressed as a percentage. A significant increase of the Int₃₄₅₆/Int₃₀₀₀ ratio was observed from the lateral scan profile collected from the cross section written at 40 mW, and this change was only confined within the written region (Fig. 4(C)). Estimated from the FWHM of the scan profile, the written layer was ∼15 µm thick, offering a more accurate evaluation than that from DIC microscopy. Two representative spectra extracted from both written region and the adjacent non-written region illustrated this significant local hydration (Fig. 4(D)). Change in the local water content could be divided into three distinctive stages based on their exhibited behaviors (Fig. 4(E)). The Int₃₄₅₆/Int₃₀₀₀ ratio of phase patterns increased nonlinearly with increasing laser power, reaching a ∼17.5% average increase of local water content at 60 mW. In stage II, the local water content reached a plateau and did not vary significantly between 70 mW ∼ 150 mW. Due to the non-uniformity of damaged regions, the Int₃₄₅₆/Int₃₀₀₀ ratio was more discrete in stage III, but it kept increasing and achieved a maximum of ∼30% increase at 230 mW. Although patterns in both stage II and III exhibited gross damage, different laser-induced chemical changes were observed. While all Raman bands remained the same for 150 mW damaged pattern, the 1156 cm^-1, 1236 cm^-1and 1757 cm^-1 bands disappeared in the 230 mW pattern, and the 3000 cm^-1 band tended to split into two sub-bands (Fig. 4(F)).

Since a dramatic increase of water content was observed in those phase patterns in stage I, TGA was applied to investigate how this change affected the polymer structure. Compared with pristine Contaflex samples, samples with the bulk phase pattern exhibited similar thermal behavior, indicating that polymer thermal stability did not degrade after laser micromachining (Fig. 5(A)). The first regime between 40 °C and 200 °C was due to the elimination of free and bound water in samples. Because of the high water content, a rapid weight loss was observed in hydrated samples below 100 °C. After an isothermal state in which polymer structure remained stable but all water has been removed out, decomposition of the polymer network initiated from 350 °C until 500 °C, characterized by a significant weight loss [48]. The bulk water content difference between hydrated pristine and written samples can be calculated from the percentage of weight loss at the end of isothermal state. The weight loss of pristine sample and written one at 200 °C was 57.8% and 65.1% respectively, suggesting that written samples contained more water than that of pristine hydrogels. For dehydrated samples, the derivative of weight loss was nearly plateaued below 200 °C (Fig. 5(B)). The hydrated pristine sample displayed a broad zone centered around ∼99 °C and a smaller zone around ∼124 °C, corresponding to the loss of free and bound water, respectively [49]. Although the written hydrogel exhibited similar behavior, the two peaks both shifted to a lower temperature.

 

Fig. 5. (A) TGA thermograms of pristine hydrogels (the black and red curves), and samples containing a bulk phase pattern as in stage I (the blue and magenta curves). (B) Derivative of the weight loss.

Download Full Size | PDF

3.3 Chemical alterations of laser-induced depolymerization

The FT-IR spectra from two damaged patterns and one non-written sample all presented a broad band at ∼3360 cm^-1 assigned to ν(-OH) and a superimposed band at 2953 cm^-1 assigned to ν(C-H) (Fig. 6). The band at 1711 cm^-1 corresponded to ν(C = O) from the ester and ketone groups [50]. The 1647 cm^-1 band included contribution of ν(C = C) from alkene groups as well as δ(O-H-O) from bulk water molecules [44]. The band at 1485 cm^-1 can be assigned to ν(C-N), while the stronger side peak at 1452 cm^-1 was from δ(-CH₂) [37,51–53]. The band at 1277 cm^-1 could be due to in-plane δ(-OH) on the main chain, and the 1252 cm^-1 band was attributed to ν(-CH₃) [44,54]. While the narrow band at 1167 cm^-1 was assigned to ν(C-O), the broad band at 1055 cm^-1 was due to symmetric ν(C-O-C) [55,56]. Similar to that observed in Fig. 4(F), all characteristic bands remained the same for 150 mW damaged pattern, but most disappeared for the 230 mW one, both of which indicated possible depolymerization of polymer backbone.

 

Fig. 6. FT-IR spectra of non-written region (black curve), 150 mW (blue curve) and 230 mW (red curve) damaged patterns.

Download Full Size | PDF

4. Discussion

The present set of experiments mainly studied the material effects of low-repetition rate femtosecond laser micromachining to gain insights into its underlying mechanisms. RI change is assumed to be induced by the change of polymer molecular density in the excitation volume [57]. It is known that in single-photon absorption, the reaction rate would be proportional to the product of beam intensity and activated moiety [58–61]. However, the energy of a single 405 nm photon (3.1 eV) is smaller than the dissociation energy of most single bonds, such as C-C bond (3.6 eV), C-H bond (4.7 eV) and O-H bond in water (4.8 eV) [62], therefore, single-photon absorption cannot provide sufficient energy to achieve significant structural changes. We have done extensive studies and built quantitative photochemical scaling models for femtosecond laser micromachining in hydrogels in both four-photon and two-photon regime [3,38]. In the two-photon absorption limit, phase change induced by a pattern of lines with spacing t can be calculated by $\Delta \phi = \gamma \cdot \frac{{{P^2}}}{{\nu \cdot \tau \cdot {\lambda ^2} \cdot S \cdot t}}$, in which P is the average power, ν is the repetition rate, τ is the pulse duration, λ is the laser wavelength, S is the scanning speed and γ represents the multiphoton absorption coefficient. Here γ = 1.01e^-23 m⁴·waves/(W²·s) describes the energy conversion efficiency of materials and can be determined by least square fitting of the phase change data. Consistent with previous scaling model, the fitting results in part 3.1 readily confirms that under 405 nm, two-photon absorption is the driving force for inducing localized RI change and other chemical changes. Additionally, compared with early results written at 80 MHz in Contaflex GM 58, over one wave of phase change at 543 nm can be obtained with single layer writing at a much lower average power and faster scan speed [63]. This suggests that the induced RI change depends on the amount of pulse energy absorbed by the material, and operating at a lower repetition rate could largely improve the micromachining efficiency.

The observed morphological changes as well as the spectroscopic features both suggest different modification regimes of micromachining. Below 60 mW, the local water content increased in proportion to the beam intensity, corresponding to an enhancement of the phase change magnitude. The local structural changes are likely initiated from nonlinear absorption by the active molecules in polymer. High photon density in the focal volume could trigger two-photon absorption by initiator molecules along the traces of laser focus [64,65]. Due to the high thermal conductivity of water and its high concentration in hydrogels, at a lower repetition rate, part of the energy from the absorbed photons will dissipate as thermal energy even before the next pulse arrives [66,67]. Some single bonds, e.g., -OH or -H will be broken by the remaining energy, generating free radicals [68]. As the two-photon absorption proceeds, polymer segments will be removed gradually, and water molecules start to permeate in the modified volume. Due to the lower RI of water than that of polymer matrix, the written regions thus exhibit negative phase changes. Besides direct Raman probing of the written layer, the thermal analysis of bulk phase patterns provided another evidence for the increase of local water content. And shift of the two weight loss zones to lower temperatures shown in Fig. 5(B) suggested that the infused water likely presented as free water and interacted through hydrogen bonds with other water molecules in polymer [69]. To enable the two-photon absorption process, certain compounds should be responsible for absorbing and delivering energy to the polymer backbone. Based on the composition of Contaflex GM 58, we hypothesize that MOBP is most likely responsible for absorbing photon energy, with which MMA could share its electron structure. The benzophenone group in MOBP is a strong UV absorber that has been widely used as photo-initiators to prepare photo-reactive polymers [70]. Researchers also noticed that the oxygen atom in the carbonyl group could inter- or intra- molecularly interact with electron acceptors like fullerenes and metal nanoparticles by donating a lone electron pair. The donor and acceptor are able to share the localized charge density to form linear coordinative bonds or charge-transfer complex [71,72]. Drawing from previous studies, MMA and MOBP may be bridged by the free Na⁺ cations in saline solution to form activated moiety, which deliver energy to the polymer backbone and break single bonds. The Na⁺ may interact with both the oxygen atoms from ester group in MMA, with a free electron pair sharing between C-O and C = O equally. In the moiety structure, Na⁺ is likely surrounded by –OH from MOBP and C-O, C = O groups from MMA to form a single bridge. It is also possible that the C = O and –OH group in benzophenone could interact with MMA together to form a double bridge [73–75]. Small bubbles appeared, randomly dispersed, after 60 mW and patterns no longer exhibited uniform dense lines, as shown in Fig. 3(A). Correspondingly, with beam intensity kept increasing, the local water content reached a plateau in stage II, indicating an equilibrium state of the hydrogel network has been reached. This may explain why phase change cannot further increase after the materials saturation, and the layer thickness was roughly the same. Hydrogel network is composed of polymer segments that are connected by universal joints to form an equilibrium geometry with minimum system energy [76]. Upon laser irradiation, the most stretched or compressed segments that beyond the most-favored conformation are likely to be broken first to narrow down the free energy distribution in polymer network [77]. Since random and gradual removal of the activated segments would not significantly influence the conformational states of others, under low pulse energy, generation of free radicals could occur without the collapse of the polymer backbone [78]. As a result, all spectroscopic signatures in Fig. 4(F) and Fig. 6 remained the same for phase patterns and even for the 150 mW damaged ones, and no photoproducts or dissolved monomers have been detected.

If sufficient pulse energy is deposited and local temperature is above the ceiling temperature, the polymer backbone would be depolymerized by free electrons through random main chain scission or unzipping [79]. At this stage, the Int₃₄₅₆/Int₃₀₀₀ ratio fluctuated due to the non-uniformly degradation of polymer network in different samples. Vanishing of the ν(C-C) and ν(C-O) modes at 230 mW, as shown in Fig. 4(F), confirmed this photochemical degradation. The depolymerizaiton reaction likely initiates from the cleavage of the ester side chains, as identified by the disappearance of the ν(C = O) mode in both Raman and FT-IR spectra, and this could trigger the subsequent main chain scission. As dissociation of side chain proceeds, radical electrons generated at the C atoms on polymer main chain may additionally destabilize it [80]. Compared with the other two spectra in Fig. 4(F), a new sub-band assigned to the asymmetric ν(C-H) showed up in the spectrum of 230 mW pattern. Some earlier studies have found that the ν(C-H) mode in alkanes was sensitive to the conformational disorder of polymer chains [81,82], therefore, this spectral change may indicate that besides chain dissociation, the spatial organization of main chain was already disrupted with 27.7 nJ pulse energy applied. Polymer chains were also likely to obtain more rotational freedom, and the functional side groups may rearrange and interact with their polar elements more effectively so that water molecules are less able to interact with functional groups [78]. Depolymerization could decrease the average molecular density and broaden the molecular weight distribution, which significantly reduced the scattering signals [57]. Higher number of endgroups generated during depolymerization may also increase the absorption coefficient of the damaged region, as already observed in UV-ablated PMMA [80,83]. Therefore, although a dramatic increase of Int₃₄₅₆/Int₃₀₀₀ ratio was observed in this regime, it is not clear that permeation of water still occurred in the written region.

5. Conclusion

The present work examined the materials effects of femtosecond laser micromachining by inscribing single layer dense line patterns into ophthalmic hydrogels across a wide range of pulse energy. Operating at 8.3 MHz repetition-rate enabled us to obtain phase change up to ∼1.2 waves at 543 nm with much lower average powers than previous studies. A phase change profile fitted to our photochemical scaling model indicated that the micromachining process was mainly dominated by two-photon absorption. Compared with a qualitative approach generally used in spectroscopic studies, the confocal micro-Raman system, combined with the edge-detection method, allowed us to directly probe the local water content and other chemical changes within the written layer in a quantitative manner. Below material saturation limit, the local water content increased nonlinearly with phase change increasing, suggesting that generation of free radicals followed by water permeation is likely the key contributor for induced RI changes. While micromachining did not degrade the thermal stability of hydrogels, TGA results also provided compelling evidence on the increase of local water, which possibly existed as free molecules in polymer. Besides change in the local water content, the new features presented in the Raman and FTIR spectra offered another perspective on possible photochemical reactions. In the optical breakdown regime, depolymerization may initiate from the cleavage of side chains, followed by main chain scission and even disruption of main chain organization. These findings provide valuable insights into the underlying mechanisms by which femtosecond laser micromachining induces RI changes in hydrogel materials. They may also help gauge the potential of this technique to be used for customization of refractive devices as well as for human vision correction. High quality visual correctors have been demonstrated in hydrogels and related materials with similar laser processing [84].