1. Introduction

Three-dimensional inscription of structure inside transparent material using femtosecond (fs) laser writing has proved to be a simple and extremely advantageous technique for fabricating photonic devices. This technique consists of focusing a femtosecond laser beam inside the material to locally change the refractive index at the focal spot by a multi photon non-linear process. The versatility of fs laser writing has been clearly demonstrated from the early study on the optical breakdown of dielectric material [1] to inscription of photonics structures such as waveguides [2], diffractive optic elements [3], Bragg gratings [4], etc. This technology allows the possibility of creating cheap optical sensors important for the implementation of rapid test diagnostics at points-of-care intended for the telemedicine market.

Most of these applications have been realized in stiff materials, such as glasses, that do not allow the material to be stretched or compressed for mechanical tunability of the photonic devices and sensing applications. Poly-di-methyl-siloxane (PDMS) is a highly stretchable glassy material which has been widely used in the optics industry and found acceptance as a material of choice in lab-on-a-chip applications [5] due to its biocompatibility, inertness, durability, elasticity, transparency, low cost and wide availability. However, PDMS, is not known to be suitable for femtosecond laser inscription.

UV photosensitivity and the potential tunability of this material has already been demonstrated for phase mask replication by casting, intended for fiber Bragg grating fabrication [6]. These were either stuck to the optical fiber or used as stretchable UV transparent phase masks for grating inscription using 266 nm UV radiation, and demonstrated to be cheap replaceable phase-masks. However, these phase-masks degraded with long-term UV exposure and it was reported that the degradation process altered the refractive index and weakened the material considerably. Ultraviolet (UV) exposure leads to photoinduced degradation of the material which compromises the optical and mechanical properties. For long term use, it is necessary to avoid degradation [7]. Nevertheless, photoinduced refractive index change in the order ∼ -0.14 [8] has been reported in PDMS using UV radiation through bleaching reaction which also degrades the material properties.

Enhancing the fs photo-inscription response of the material without significantly changing the mechanical properties is the key to obtaining highly tunable integrated optical devices. We recently reported the photosensitization of PDMS to improve its functionalization [9]. Photosensitivity enhancement has also been reported by Panusa et al. [10] using phenylacetylene as a photo-sensitizing agent. They achieved an index change in the order of Δn≈0.06 and demonstrated waveguiding after inscription. Photosensitization is achieved by permeation until saturation into an already cured PDMS sample with phenylacetylene. The laser exposure is performed, followed by ethanol washing and finally heating. This procedure takes several steps which are time-dependent on the sample size and on the chemical agent. Furthermore, although the authors refer to the induction of optical damage under certain processing parameters, no comments on the mechanical properties of fs exposed PDMS as well as the origin of the refractive index change were presented, which still raises the question on the integrity of the material after inscription. In this article, following our previous work [11], we report in detail the material properties of our novel approach to achieve enhancement of photosensitivity of PDMS by incorporating germanium derivatives before polymerization. This allow the controlled functionalization, and homogenized concentration of photosensitive polymer composite. Using this material, low-loss multimode waveguides were obtained using fs laser writing in Ge-PDMS. In addition, highly tunable diffraction gratings were embedded inside the bulk photosensitized PDMS for the first time to our knowledge. The use of this method allows the ease of material fabrication and processability, necessary for the optics industry and potential LOC applications.

2. Photosensitisation of PDMS

2.1 Incorporation of Ge-derivatives into PDMS

Germanium (Ge) doped glasses are well known to be photosensitive [12]. Since PDMS is a easily functionalized glassy material, we incorporate Ge moieties into PDMS (inset of Fig. 1), in the form of germanes derivative such as Allyltriethylgermane (ATEG), Tetraallylgermane (TAG), Methacryloxymethyltrimethylgermane (MACMTG) and Dimethylgermanium dichloride (DMGCl). The goal is to enhance the photosensitivity and study the optical properties of the new blend of PDMS, while avoiding damage to molecular bonds and hence maintaining the strength of the highly elastomeric polymeric glass. Our best results, so far, were obtained using the ATEG derivative which is the one used in this study.

 

Fig. 1. DSC heating thermograms of pristine PDMS and Ge-doped PDMS in comparison to Soda-lime and Gorilla glasses. To erase thermal history a 2^nd heating ramp (10 K/min) was undertaken after a cooling ramp.

Download Full Size | PDF

Differential Scanning Calorimetry (DSC) measurements of pristine and Ge-doped Sylgard PDMS are shown in Fig. 1 in order to evaluate if the germanium derivative is well blended inside the PDMS host matrix. Any inhomogeneity of the phase such as crystallization, can jeopardize the mechanical strength and cause local variation of the photo-inscription response of the material.

In Fig. 1, DSC measurements show no significant difference (∼ 0.1 ^οC) for the glass transition occurring around -140 °C between pristine and Ge-doped PDMS. One can observe that the heat flow of Ge-doped-PDMS is roughly 3% times higher than the pristine sample which is roughly twice higher than for the two other typical glasses. The heat flow is directly linked with the thermal resistivity by Fourier’s law of heat [Eq (1)]

Thermal property of the system is also an important factor to monitor. Fs laser writing consist of depositing energy in a small volume which will lead to an increase in the local temperature. To maintain the optical properties of the material, the degradation limit should not be reached. Therefore, thermogravimetric analysis (TGA) of pristine and Ge-doped PDMS under nitrogen heated up to 600 °C using a constant heating ramp of 10 ^οC are shown at Fig. 2.

 

Fig. 2. Thermogravimetric analysis of pristine PDMS and Ge-doped PDMS realize under nitrogen at a temperature ramp of 10^οC per min. The 1% weight loss limit which is the temperature where material degradation starts to occur, is reached at a temperature of 265 ^οC.

Download Full Size | PDF

As observed in Fig. 2 the decomposition losses of pristine and Ge-doped PDMS were mostly at about 500 °C. The 1% weight loss point (Tc) was found to be around 265 °C in both cases which gives a safe limit to not cross in order to preserve optical quality. For the pristine Sylgard PDMS, decomposition occurred in two steps meaning that there is a chemical transition that can be linked to depolymerization of the PDMS chain [13]. However, for the Ge-doped PDMS, those steps are not observed, which means that the material suffers direct decomposition. This observation would mean that the pristine Sylgard PDMS polymer chains were modified by cross-linking with doped Ge-derivative.

Besides DSC and TGA graphs providing insights on the mechanical and thermal property of the material, the optical properties are also of concern for the intended applications. Therefore, the transmission spectra of pristine PDMS and unexposed Ge-doped PDMS are shown in Fig. 3. The transmission spectra were measured using a simple Ocean Optics USB-4000 spectrometer and a conventional 100 Watts incandescent light bulb. The traces shown in Fig. 3 are the mean values of 50 spectra and each sample is compared relative to an original spectrum taken without any sample. Consequently, the transmission values on these spectra include the absorption of the sample, scattering loss and the Fresnel loss from both surfaces. The samples were 1 mm thick.

 

Fig. 3. Transmission spectrum of pristine and unexposed Ge-doped PDMS including both writing wavelengths (515 and 1030 nm) used for fs laser writing.

Download Full Size | PDF

Figure 3 shows the transmission spectra of pristine and Ge-doped PDMS. A slight difference between the two spectra is measured and can be attributed in part to the increase in the general refractive index by the inclusion of Germanium, the purity (presence of dust or microbubbles which scatter light) of the area under investigation or a real increase in absorption due to the presence of Germanium. However, the difference between the two compounds is not significant enough and no new absorption peak due to the introduction of Germanium is observed. Therefore, the optical quality is conserved or at least high enough for our applications.

2.2 Photosensitivity

The effect of fs laser irradiation on pristine and Ge loaded PDMS was then investigated. A Pharos laser system providing 250 fs pulses centered at a wavelength of 1030 nm was frequency doubled to obtain 515 nm using an Orpheus OPA. 1030 nm and 515 nm wavelengths were tested using a repetition rate varying from 101 kHz to 606 kHz, a pulse energy from 1.65 nJ to 165 nJ, a translation speed from 1 to 50 mm/s and a writing lens with numerical aperture (NA) of 0.1 and 0.25. The repetition rate was changed with a pulse picker in order conserve the pulse energy.

Contrary to glass, PDMS cannot tolerate heat and tends to burn at much lower energy densities. Therefore, thermal accumulation regime must be avoided. At the same time high energy density at the focal spot is essential for processing PDMS. Hence, 1030 nm irradiation using the range of parameters listed above always resulted in burning of the PDMS. However, 515 nm irradiation, below a pulse energy of 3.3 nJ, with a writing speed of 5 mm/s at a repetition rate of 101 kHz, leads to an isotropic refractive index modification of the PDMS sample. Since single photon absorption in PDMS is near 300 nm, moving from 1030 nm to 515 nm leads to a 2-photon absorption regime which is more effective than a 4-photon regime. Therefore, less intensity was used to write at 515 nm than at 1030 nm. Since the background absorption at both writing wavelengths of interest is similar, as shown in Fig. 3, less heat is generated at the shorter wavelength. On the other hand, at 1030 nm, higher intensity is required due to the lower two-photon cross-section, and so more heat is transferred to the host matrix resulting in the characteristic burnt regions. Moreover, heat accumulation regime must be avoided since this material tends to begin decomposing at 265 ^οC. This result imposes hard limits on the energy per pulse, focal volume and repetition rate. The heating resulting from single pulse cannot exceed this temperature as the heat accumulates from the temporal overlap between the remaining energy at the time of the arrival of the next pulse. A limit can be estimated from the literature for the static temporal overlap for glasses and this regime occurs above a repetition rate of 100 kHz [14]. This limit has been found using a 1045 nm laser, 0.65 NA writing lens and a pulse length of 375 fs and pulse energy of 2.5 µJ. If we consider our writing parameters (515 nm, 0.25 NA, 250 fs and 3.3 nJ), one can compare the intensity at the focal areas by taking into consideration the pulse intensity as a function of 1/Δt and the area where the radius is given by the waist (ω₀ = λ/(π x NA)). We found an intensity difference of 9% in favor of Eaton’s set of parameters, which is almost no change and within experimental error. As shown in Fig. 1, the heat flow of two different glasses, soda-lime and Gorilla, is approximately twice lower than in PDMS which, from Eq. (1), means that the thermal resistivity of those two glasses is higher than the PDMS sample. Therefore, strictly from the point of view of the dynamics between heat dissipation and the heat supply, considering a similar photoexcited area at the focal region as well as a similar fluence, the 100 kHz limit on the repetition rate for glass should be appropriate for writing without heat accumulation for the PDMS. This limit, deduced from a comparison with Eaton et al. [14] and supported by our preliminary calculation, is adequate since no burnt areas were observed upon a single pass writing which will occur if the temperature exceeds 265 ^οC. Furthermore, the width of the writing presented at the Fig. 4 is similar to the diameter of our waist which are both around 1.2 µm where thermal effects are well known to induce modification outside of the focal area. Further investigation will be conducted to find this threshold limit, which is beyond the scope of this article.

 

Fig. 4. Evolution of the refractive index depending on the number of passes for pristine and Ge-loaded PDMS. Burnt areas, characteristic by their black appearance indicating decomposition of the polymer chain through oxidation process to carbon-carbon and silicon-oxygen-carbon bonds are observed after 34 passes. Top view [a-1), b-1)] and white light transmission side view [a-2), b-2)] of a waveguide written at 34 and 36 showing the profile of the waveguide.

Download Full Size | PDF

In order to find the maximum refractive index change produced by fs laser inscription, a multi-pass technique was used until the PDMS showed signs of burning. The recipe used was a repetition rate of 101 kHz, 3.3 nJ, 20 mm/s and the number of passes varied from 1 to 40 as shown on Fig. 4.

As observed in Figs. 4(a)–4(b) on the top view of the waveguide, inscription in Ge-PDMS shows burning only after 34 passes. The white light side illumination view of inscription in Fig. 4 shows the affected region inside the PDMS sample. For 34 passes, the structure supports light all across the inscribed area. For 36 passes, a burnt area can be seen from the top view perspective. One can observe from the side views using white light that the light contrast is less for 36 passes than for 34 passes and that the guiding light intensity is not uniform along the structure, especially for the center region where a huge diminution is observed. The center region is where the material has been subjected to the highest dose of light and therefore is the expected region that should burn first. The structure inscribed is characteristic of a weakly focused Gaussian beam, which creates an elongated elliptical structure. The maximum RI change induced without burning as shown in Fig. 4, was in the order of 4 × 10⁻³ for 40 passes, which is sufficient to write photonics structure in it.

Inscription in pristine PDMS is also demonstrated using the same writing recipe. For this sample, no sign of burning of PDMS was observed even for 40 passes as shown in Fig. 4. However, RI change was, at best, of 7 × 10⁻⁴, 6 times lower than the Ge-doped PDMS demonstrating the photosensitization using our ATEG derivative. The low refractive index resulting from inscription of pristine PDMS is a problem when making photonic structures such as curved waveguide in which loss induced by the radius of curvature increases as the refractive index decreases, whilst maintaining the same waveguide aspect ratio. In order to make compact device, higher refractive index change is preferable. Refractive index profile was measured by the phase-shift caused in the inscribed region using a Mach-Zehnder interferometer. A reconstruction algorithm coupled with the structure profile obtained by direct microscopy, is used to fit the phase profile with the refractive index profile [15].

2.3 Raman spectroscopy of pristine and exposed/unexposed PDMS

To understand where the refractive index change comes from, Raman spectroscopy was performed at 633 nm on pristine, Ge-doped unexposed and exposed samples. As one can observe in Fig. 5, there are four new peaks located around 535 cm⁻¹, 1220 cm⁻¹, 1602 cm⁻¹ and 2734 cm⁻¹. These are associated to the introduction of the germanium compound inside PDMS. The peak around 535 cm⁻¹ has been reported to belong to the carbon-germanium liaison (Ge-C) [16] and its intensity increases by 20% after laser exposure. The 1220 cm⁻¹ peak can be associated to the CH₂ twisting and CH bending [17] which is present in the germanium compound. The 1602 cm⁻¹ peak is attributed to the C-C bond which is found in the original compound [18] and it does not change with fs laser exposure. The 2734 cm⁻¹ peak is an overtone of the 1368 cm⁻¹ peak which is not observable in Raman spectroscopy for the C-CH₃ symmetric deformation. The intensity of this peak decreases with laser exposure.

 

Fig. 5. Raman spectrum at 633 nm of pristine (bleu), unexposed (orange) and exposed PDMS (red), and a table of the intensity ratio change for each of the peaks between the unexposed and exposed PDMS. 4 new peaks are observed in the spectrum by adding the Ge derivative compound.

Download Full Size | PDF

The table shown on Fig. 5 summarizes the peaks measured on each Raman spectra and gives the intensity ratio between the exposed and unexposed PDMS samples. As one can observe in the Raman spectrum shown in Fig. 5 no peak was detected around 300 cm⁻¹ associated to crystalline germanium (Ge-Ge) [19] and crystalline silicon (Si-Si) at 520 cm⁻¹. [20] before and after laser exposure which agrees with the DSC measurement. Also, no peak showing SiO₂ [21] formation is detected on this Raman spectrum. The peaks at 2906 cm⁻¹ and 2967 cm⁻¹ are respectively attributed to the asymmetric and symmetric stretching of the methyl end group (CH₃) [22], which increase with the introduction of the Ge compound. It is to be noted that this methyl group is also present in the germanium derivative. However, laser exposure seems to affect this group and reduces its intensity by 25% compared to the intensity of the unexposed Ge-modified compound. Furthermore, if we observe the intensity of the peak associate to the deformation of the backbone structure (C-Si-C) at 158 cm⁻¹ and 187 cm⁻¹ [23], it seems that the laser inscription increases these two peaks. Therefore, more deformation of the backbone structure is reported. Thus, the refractive index modification seems to come from the fixing of the Ge to the carbon which seems to affect the methyl group and causes a perturbation of the C-Si-C backbone deformation causing a local densification.

3. Fs written devices

3.1 Fs laser written waveguide

One can use this refractive index change to write photonics structure inside the modified PDMS. Since a single inscription result in a narrow profile as shown in Fig. 4, many side by side inscriptions were made to broaden the cross-section in order to facilitate the coupling of the light from an optical fiber to the waveguiding structure. To do so, 4 inscriptions of 16 passes each, separated by 1 micron written at 515 nm at a speed of 20 mm/s were inscribed in order to create the waveguiding structure presented in Fig. 6. To investigate the waveguiding property, a 633 nm He-Ne laser as well as a 1550 nm laser source were coupled to the waveguide. 633 nm was used to observe the mode-field and the scattering of the light along the waveguide. 1550 nm was used to measure the loss of the waveguide by measuring the Rayleigh scatter with an optical backscatter reflectometer OBR-4600 from Luna, all along the waveguide [24].

 

Fig. 6. Waveguide written inside the PDMS. a) View under the microscope showing scattering from the waveguide. b) The position of the waveguide inside the PDMS layers. c) and d) shows two different modes supported by this waveguide. e) The refractive index profile of the waveguide. f) The light was butt coupled using an SMF-28 fiber into a 4 cm long waveguide sample.

Download Full Size | PDF

No sign of burning was seen with this recipe as shown in Fig. 6(a). A 6 microns wide waveguide is then formed with a Gaussian like refractive index profile as seen in Fig. 6(e). The different mode profiles Figs. 6(c)–6(d) are obtained by using a 20X, 0.4 NA microscope objective lens. Scattering losses from the waveguide can be observed from the top, Figs. 6(a)–6(f) using a 633 nm He-Ne laser. The loss was measured to be of 0.6 dB/cm ± 0.04 dB/cm (with a 95% confidence level) at a wavelength of 1550 nm. A near field view of the layer of PDMS indicates that the waveguide is situated 400 µm below the surface Fig. 6(b).

3.2 Fs laser written stretchable grating

The narrow needle form of the inscription is well suited to the inscription of diffraction gratings in our Ge-doped PDMS. Therefore, gratings of 2 [Fig. 7(a)], 3 and 4 µm [Fig. 7(b)] pitch were inscribed using direct fs laser writing. Grating were written with 18 passes using 515 nm illumination, at a repetition rate of 101 kHz, writing speed of 20 mm/s, pulse power of 1.65 nJ and a focusing lens with a numerical aperture of 0.25. Two layers of writing separated by 20 µm were written in order to obtain a more efficient grating by increasing the phase shift difference.

 

Fig. 7. Fs written grating in the bulk PDMS. a) Scanning electron microscopy (SEM) showing the transverse view of the 2 µm pitch grating. The transverse view is visible in the SEM because of the formation of a relief grating at the broken surface. The reasons for the formation of this relief grating is still under investigation but is presumably due to the conformational changes in the molecular structure induced by the fs laser. b) Diffraction of white light by a 3 µm and 4 µm pitch grating, respectively.

Download Full Size | PDF

One of the advantages of having such a diffraction grating in this material is tunability. In Fig. 8(a), the efficiency of the grating increases with the stretching suggesting that there is overlapping between the writing steps. Therefore, larger steps or better writing resolution defined by the writing parameters, should be used to reduce the pitch. Notwithstanding, an efficiency of 3-4 percent is obtained with an unoptimized recipe showing promising potential. However, when the stretching is too large, a depletion of the diffraction orders is observed. This depletion is probably introduced by the stretching methods which introduces a small curvature in the sample. This problem may be overcome by using a dog bone structure sample, which prevents this behavior. The diffraction in the orthogonal direction [see Fig. 8(c)] under large strains is presently under investigation. Furthermore, stretching to around 50% of the initial sample was realized without mechanical failure or plastic deformation of the material.

 

Fig. 8. a) Diffraction angle and efficiency of a grating under stretching condition for the first order. b) Diffraction orders. c) Diffraction efficiency of the orders at higher strains on the RHS axis.

Download Full Size | PDF

Furthermore, to confirm the repeatable tunability behavior of our devices, a 1.5 µm pitch grating was stretched in steps of 10% of its original length. The position of the first order of diffraction of a He-Ne (632.8 nm) laser was monitored before and after each stretch. The stretching and relaxation were performed 20 times before increasing to the next stretching length. For a high strain value, a permanent shift should be registered which means that the sample transitions from an elastic to a plastic deformation regime. The transition between the elastic and the plastic deformation is our limit of interest since there is a permanent deformation of the structure. When this limit is crossed, the starting angle θ_ɛ=0% value changes significantly. Figure 9 shows the evolution of the first order of diffraction for different strains values applied to the grating.

 

Fig. 9. Angular position of the first diffraction order before and after elongation of 20 cycles each in steps of 10% (the dots at the end of the stretch shows the scatter in the data for the different repeated stretch cycles). Mean value (µ), standard deviation (std) and p value of an ANOVA test between the 10% strain population and the others one, are presented on the table for this 1.5 µm pitch grating. One can observe that the angle of diffraction after stretching does not evolve linearly with the strain. This is due to the fact that the grating is inscribed inside the material and the first order will undergo a refraction at the PDMS-air interface resulting in this behavior.

Download Full Size | PDF

One can see from Fig. 9 that the diffraction angle of an unstretched grating (θ_ɛ=0%) is retrieved after stretching to 140% of the original length. The mean values (µ) and the standard deviations (std) of each population are presented in Fig. 9. For the population ɛ=10%, the lowest value around θ=37^ο is an anomaly and therefore can be excluded from the mean and the standard deviation calculations which gives a mean value of 38.71^ο and a standard deviation of 0.07^ο. Therefore, the mean value (µ) and standard deviation (std) are quite similar from the ɛ = 10% to ɛ = 40%. A small shift of the mean value is registered for a strain of 50% and 60%. In order to ascertain if the shift is significant or not, an analyses of variance (ANOVA) is performed relative to the ɛ=10% population. The p values are presented in the table in Fig. 9. From these values, we can conclude that it is only for the ɛ=70% that the mean value is significantly different from all other mean value. Therefore, the mean measured from a 10% up to 60% strain are not significantly different at a confidence level of 95%. Considering when ɛ=70%, the difference could come from the elastic-plastic transition or from the slippage of the PDMS in the holder. The PDMS was held in between two metal pieces squeezed together. Throughout the stretching process, the PDMS sheet suffers deformation of its thickness rendering the holding of the PDMS less effective and permitting the sample to slip. In order to correct this phenomenon, modification of the way the PDMS is held or how the PDMS sample feature is fabricated is necessary to provide a better grip of the sample. Nevertheless, a repeatable strain up to 60% does not seem to cause any plastic deformation or mechanical failure proving the high repeatable tunability of our system. More investigations are being performed to correct the slipping of the PDMS

4. Conclusion

Femtosecond laser inscription in pristine PDMS has been demonstrated and its photosensitivity has been improved by a factor of 6 in our modified PDMS using a germanium based chemical compound. Raman spectroscopy has been performed to understand the enhancement in refractive index change caused by the fs laser writing in germanium modified PDMS samples. Waveguiding structures were successfully written with direct femtosecond laser writing showing losses of 0.6 dB/cm. Highly stretchable diffraction gratings have been imbedded inside a photosensitivity enhanced PDMS sample using a femtosecond laser and their repeatable high tunability have been shown to a strain of 60%. Many other derivates are being investigated to enhance the photosensitivity and the laser induced refractive index change to >0.01 and will be reported on in a future publication. Enhanced photosensitivity will improve the integration of photonics directly into the PDMS used for LOC applications.