Near infra-red (NIR) self-guided photo-polymerization is investigated in the context of micro-optics photo-fabrication on VCSELs (Vertical-Cavity Surface Emitting Lasers). We present the optimized process we have developed to allow for a collective fabrication on III-V devices wafers under real-time optical monitoring. The influence of photo-chemical parameters on final micro-elements dimensions is studied for two types of single mode 760nm VCSELs. The difference of the resulting tip shape between the two lasers is due to the strong differences of their emissions, as they are nicely reproduced by the computed near-field profiles. The tip shapes are also compared to those produced by the light emitted by an optical fiber and differences with VCSEL tips are discussed. Also the VCSEL characteristics with fabricated tips are discussed and found in good agreement with optical modeling.
©2012 Optical Society of America
Self-guided photo-polymerization is now a well-established method for micro-optical elements fabrication at the end of optical fibers. Using this method, polymer microtips were built in the visible range  and in the near infra-red region . More recently, this approach was successfully adapted for integrating self-aligned microtips at the surface of 760nm single-mode VCSELs (Vertical-Cavity Surface-Emitting Lasers) . To reach this goal, two key issues were solved: (i) transparency after fabrication to allow for device operation at the creation wavelength, (ii) low roughness to achieve high optical quality micro-objects. VCSEL beam focusing at short distance was thus demonstrated, opening a new way for micro-optics integration on such laser diodes. The main advantages of this technique lie on a single fabrication step and on a full compatibility with an application after VCSEL processing or after VCSEL packaging. In this work, we depict the full fabrication process we have optimized and we study how chemical and electrical parameters influence the tip geometry, by comparing in particular VCSEL case to optical fiber case. This analysis constitutes an essential step to identify available fabrication ranges and further applications in the field of instrumentation and optical sensing (compact probes for optical scanning probe microscopy).
2. Fabrication method
The NIR photo-polymerizable formulation used in this study is a mixture of three components: a NIR sensitizer (HITC 0.18 wt. %), associated with a co-initiator (MDEA 4 wt. %), is used to trigger the polymerization of a triacrylic monomer (Ebecryl 605). Such a monomer presents good performances in terms of spatial resolution and it leads to a crosslinked polymer with suitable mechanical and optical properties for applications in micro-optics . Moreover, the absorption coefficient spectrum of the photopolymer presents a maximum value in the NIR region, and is totally bleached after fabrication, as required.. Due to the absorption of the cyanine dye, the NIR photopolymer shows a strong absorption band comprised between 700 and 900 nm. After irradiation at 760 nm, the photopolymer is getting transparent in this range because of dye photobleaching. Spectra are given in . Basic principle of NIR photo-fabrication on VCSELs devices is the following: the NIR photo-polymerizable formulation is spin-coated on the VCSEL wafer (Fig. 1 ). Spin-coating was here developed preferentially to local dispending, as it ensures a high uniformity of resist thickness for devices arrays. Consequently a parallel fabrication is possible. Resist viscosity and spin coating conditions were optimized to obtain a good wetting on the sample surface and a thickness control in the range [20-50] µm. Using this method, the film roughness is lower than 1nm and the thickness uniformity on a 2 inches wafer is better than 2µm, with a 2.5mm-large edge bead. After formulation deposition, photofabrication is performed on a simple electrical probe station, commonly used for VCSEL L-I-V test. A metallic microprobe is placed on the VCSEL contact pad to allow for an electrical injection. Precise probe positioning is possible thanks to formulation transparency in the visible range. The specific probe we used is thinned at its extremity (radius = 5µm) to prevent from any local thickness variation during its positioning on the VCSEL pad by capillarity effects. An electrical current is applied to the VCSEL device leading to a laser emission above threshold that triggers a local photo-polymerization of the NIR-sensitive formulation. Optical refractive index and absorption modifications occurring in the resist under exposure lead to a self-guiding effect of light propagating across the layer. After rinsing the wafer in Isopropanol, a solid and transparent microtip remains at the centre, perfectly aligned with the VCSEL emission beam.
Light emerging from the polymer layer is measured during fabrication by placing a silicon photodiode above the device surface (Fig. 2 ). Thanks to this real-time optical monitoring, the resist absorbance at the lasing wavelength can be inferred from the optical signals measured at the beginning of the reaction and in the saturation zone applying Beer-Lambert law. We found an absorption coefficient of 800cm−1, in good agreement with preliminary measurements of formulation absorbance. Moreover, control of the reaction time can be achieved using this set-up. Bleaching time was defined as the time needed to reach 99% of the final light power, which corresponds to a bleaching of 99% of the initial NIR dye. As shown in Fig. 2, bleaching time is found to vary from 4 to 16 s, depending on the initial applied power. Interestingly, we have also checked that these curves superpose well when they are plotted as a function of optical dose instead of reaction time and that saturation occurs for the same value of optical dose (40s.kW/cm2). This means that the fabrication mechanism is not dependant of optical power level in this range of incident power. However, the shape of the tips was found to vary significantly depending on the photochemical conditions. This discussion is presented in the next part.
3. Photo-chemical study
Several series of microtips were fabricated on the surface of VCSEL devices as well as at the extremity of optical silica fibers (SMF28/core diameter 8.2µm, ON = 0.06) with the same initial polymer thickness (40µm) to make possible a direct correlation with well-known systems (Fig. 3 ). Two kinds of 760nm single mode GaAs-based VCSELs used in this study were provided by Oclaro Inc. (Zürich, Switzerland) and Philips Technology GmbH U-L-M Photonics (Ulm, Germany). They differ from each other by the optical confinement method used for their fabrication. The first ones comprise a 5µm-diameter metallic contact aperture (laser threshold~2.5mA), as the seconds include a 3µm-diameter buried oxide aperture near the optical active zone (laser threshold~0,4mA). This technological distinction allows us exploring two numerical apertures (NA) for the optical laser source. NA is indeed equal to 0.14 for the contact aperture case and to 0.19 for the oxide one. Applied optical powers used for experiments were calculated in each case so that a direct comparison between different types of sources or fiber is possible, by taking into account initial power densities difference.
Effect of initial resist thickness
Because free-radical polymerization is highly sensitive to O2 quenching , we observed that polymerization of the NIR photopolymer could not be led efficiently within the upper layer of the material (around 5µm). Consequently, the NIR formulation thickness plays a key role on tip fabrication. If the layer is too thin, oxygen inhibition at the surface, favored by continuous replenishment of O2 by diffusion from surface, is predominant within the whole photoresist thickness and tip creation is not possible . Figure 4(a) illustrates this point: from a film thickness evaluated to 5 μm, a very short tip could be generated (around 1μm). On the other hand, thick polymer films are either not favorable. Indeed, as shown in the next section, the length of the tip is usually of the order of magnitude of the thickness of the deposited film (the difference corresponds to the inhibition volume). Consequently, as shown in Fig. 4(c), the high aspect ratio of the polymer tip is usually too high to provide straight tips. For these reasons, the initial thickness of the NIR-photoresist film was chosen in the [20-50μm] range (Fig. 4(b)).
Effect of optical dose
Whatever the considered type of VCSEL, final tip height is found to increase monotonously with exposure time, i.e. with applied optical dose (see example for the case of a metal aperture in Fig. 5 ). The tip appears after only 2 s of irradiation at 2kW/cm2. This time, also called inhibition time, corresponds to the time needed to consume the O2 dissolved in the photopolymer film. Interestingly, this time is 5 times lower than bleaching time (cf. Figure 2). Therefore, measuring the optical power during fabrication could constitute a way to monitor the bleaching ratio, and thus the final tip height. For larger times, length increases with irradiation time but remains far shorter than the deposited film thickness. This means that the photo-polymerization is not efficient up to the free photopolymer film surface. Same trends were observed for the fiber case. Main difference is concerned with the shape of the tip. The maximum tip section is greater than the emission surface in the case of the VCSEL whereas the tip section is always decreasing from the fiber surface to the end of the tip. Such behavior has to be confronted to the higher NA of the VCSEL which results in a higher divergent actinic beam. The self-focusing effect described previously allows refocusing the beam and ensure a sharp tip end. Such effect explains also why the tip length is lower in the case of the VCSEL case compared to the fiber.
Tip dimensions were measured as a function of the exposure dose (Fig. 6 ). In the same time, Radius of Curvature (RoC) initially equal to ~0.5µm increases up to 2µm in the saturation regime. The increase of the tip ROC is coupled to tip height increase and is due to the propagation of the polymerization volume towards the surface of the film. As for tip lateral extent, it is mainly fixed by the emission zone size and by self-guiding effect. Considering these results, it appears that it is possible to tune the radius of curvature of the tip and thus its focal length by playing with simple parameters such as the irradiation time or the laser power.
Effect of dye concentration
NIR sensitizer concentration is a parameter of importance since the production of reactive species under irradiation is directly related to this species. As shown in , dye concentration can have two main antagonist influences in self-guiding polymerization: on the one hand, an increase of the dye concentration will allow reaching a higher concentration of reactive species able to start the polymerization. Thus, this effect corresponds to an increase of the polymerization efficiency and thus, for given photonic conditions, the tip will be longer and will present a higher section. On the other hand, high dye concentration increases the absorption of the material and may result in high light attenuation along the optical pathway. Two NIR sensitizer concentrations were tested here (0.09% and 0.18%) with two different applied optical powers (40µW and 80µW). Figure 7 illustrates the effect of intensity and dye concentration on the tip length. In our case, the increase of the intensity for a given concentration favors the extent of the polymerization and thus, long tips are obtained at high intensity. This observation demonstrates that the main effect of the dye concentration is to improve the efficiency of polymerization and thus, the internal filter effect is not significant. The high bleaching rate of the HITC from excited states accounts for this behavior.
Effect of initial beam profile
As seen in Fig. 8 , for the same applied optical density and the same initial polymer thickness (40µm), tips fabricated on VCSELs are shorter than those self-written on optical fibers. Tip height is also found to be lower for oxide-confined VCSELs than for metallic- aperture ones. These discrepancies can be attributed to the different initial beam diameters, as a better consumption of inhibitor is expected to occur in case of a larger beam, leading to a better reaction efficiency. Moreover, tip shapes are also significantly different. This can be explained by the different initial beam transverse profiles.
To model the optical field propagating in the polymer medium at the beginning of the photochemical reaction, we have used a fully vectorial three dimensional optical model VELM . This tool is able to predict output beam profiles emitted by VCSELs devices, as it does take into account lasing mechanisms as well as full vertical multilayer resonator and transverse geometries. Initial values of refractive index (1.48) and optical absorption (800cm−1) of the formulation were also taken into account. Resulting 2D color contour maps for oxide-confined devices are plotted in Fig. 9 . Each contour can be related to a different photo-polymerization threshold depending of the initial applied optical dose. Doing so, we found a good agreement with experimental shapes achieved for increasing VCSEL powers with a constant exposure time. In particular, the narrowing of the tip basis for the highest exposure dose is well reproduced.
It is worth noting that this modeling gives only an indication on the final tip shape. In fact RoC and height are not only a function of the initial optical field but they also depend on dynamic parameters of the system such as oxygen inhibition mechanisms at the polymer surface as well as optical indices modification during polymerization process. Nevertheless, VELM was also proven to reproduce tip shapes differences between oxide-confined and metallic-aperture VCSELs devices. The presence of a small metal aperture at the VCSEL surface produces indeed strong field perturbations. Those result in different near-field profiles which lead to the different experimentally observed tip shapes (Fig. 10 ). Cross sections for three increasing distances from VCSEL surface (z = 0,10and 20µm) are also plotted for both cases. They also confirm that tip height is higher for the metallic contact aperture case, as more light is expected to propagate at z = 20µm.
4. Optical properties after tip fabrication
VCSEL light-voltage-current (L-I-V) characteristics measured with a standard set-up before and after tip fabrication are presented for the oxide case in Fig. 11(a) . Electrical properties are found unchanged, while the laser threshold of the device is slightly increased (0.45mA instead of 0.35mA). Furthermore, the measured output power seems also reduced by a factor two, but this effect is due to the very short focal length of the tip (see next paragraph) that prevents from collecting the whole emitted beam using our detector located at 3 millimeters from the tip surface. We found similar results for metallic-aperture VCSELs. In the latter case, laser threshold is equal to 4.5mA instead of 3mA, so the increase is larger for the metallic-aperture devices, which can be well understood by being the metal in contact with the emitted light. As for the emission spectrum, it stays nearly unchanged in both cases; as seen in Fig. 11(c), the oxide-confined laser stays single mode (SMSR>34dB) and still emits a single mode beam at the same wavelength. The result is similar for the metal-aperture case (see Fig. 11(d)).
As seen in Fig. 11(b) (case of oxide-confined VCSELs), threshold change after fabrication remarkably differs from one device to another and presents a mean value of around 15% for the realized devices. This effect is well reproduced by VELM simulations; in Fig. 12 the variations of threshold gain compared to the device without tip are reported. The variations show a period of λ/(2 npolymer) = 254nm, which proves beyond any doubt the cause of this effect is the in/out of phase interference of the polymer material placed at the output of the device. This effect can be well observed in a much simpler 1D simulation of the structure, where the tip is just a plane layer of the stack, providing very similar quantitative results (not reported here for sake of brevity).
Variations with RoC, shown at right, can then well be seen as an effective variation of the polymer thickness experienced by the mode-rays. In fact, when RoC is varied on the entire range achieved in the experiments, the range of variation of threshold gain and emission wavelength stay pretty stable and average to almost the same values (continuous and dotted red lines in the plots). The computed values for variation of emission wavelength show this parameter stays almost unchanged, as in the experiments (cf. Figure 11(c-d)). The increase of threshold compared to the unprocessed devices is found to vary in the range 0-30%, as in the experimental results. The difference with the average of 15% in the experiments is due to a limited number of samples (seven), two of which show an increase close to the maximum one. The sensitivity to tip RoC and to tip height is mainly observed on the laser threshold, but it is much lower for the output power. Indeed, we estimate its value to ~90% of the initial VCSEL light at the focal plane. Consequently, the modifications due to the presence of the tip are not detrimental for the use of such VCSELs in miniaturized optical probes.
Owing to their small Radius of Curvature, polymer microtips can strongly focus the light at short working distances, as shown for example by FDTD simulations performed in the case of polymer tipped optical fiber [8,9]. To evaluate the general focusing properties of the VCSEL-based microtips, we have conducted similar FDTD simulations with an impinging mode similar to the VCSEL output modes in terms of width and numerical aperture. Calculated spot diameters and working distances are shown in Fig. 13(a-b) . These simulations were done for typical RoC values ranging from 0.6 µm (minimum value to get a spot outside the probe) to 1.55 µm and a height of 10 µm. For this increasing RoC, the tip base diameter was also simultaneously reduced from 5 to 3.5 µm, as for the investigated/characterized structures a high RoC was correlated with a smaller base diameter. The refractive index of the polymer after polymerization is estimated to be 1.52; as for the absorption loss after fabrication, it is assumed to be negligible for the considered dimensions. In these conditions, working distances were found to increase from 0 µm to 1µm with a tight focus spot ranging from 0.4 to 1µm in diameter. On the other hand, the tip height was found to have only a small influence on working distances and spot diameter.
Focusing properties were first roughly observed using a simple LIV prober . Here, to confirm the calculated trends, we have mounted several devices and imaged their focus spot with a high numerical aperture (NA = 0.95) objective having a resolution of about 0.5µm. Two typical measured focus spots images and profiles are shown in Fig. 13(c-d).
As expected, the measured intensity is found to be notably focused compared to the bare VCSEL case, although slightly bigger than calculated. In some case, we observed a slight modification of field symmetry. Our measurements confirm also that more than 80% of the VCSEL output light is transmitted by the tip and that the focusing effect is observed whatever the applied operating current. Furthermore, smaller spot is obtained for the sharpest probe, as predicted by our modeling. One can also point out that some tip geometries may impact significantly the profile of the VCSEL mode itself. The precise study of this effect is currently undergoing for various tip shapes.
Near infra-red self-writing of polymer microtips on VCSELs devices (Vertical-Cavity Surface-Emitting Lasers) has been studied. A spin-coating process has been developed for a uniform deposition of the NIR photopolymer. Thanks to this collective process, a photochemical study has been performed to determine available ranges for tip height and tip RoC. Moreover, the dependence of tip dimensions to VCSEL optical density power, exposure time and dye concentration has been investigated. This study was carried out on two kinds of single mode 760nm-emitting VCSELs exhibiting different active zone sizes related to a different type of transverse confinement (metallic aperture or buried oxide aperture). Photo-fabrication was also performed on a standard optical fiber for reference. Experimental results show a strong dependence of tip shape to the initial beam profile, in good agreement with our model taking into account laser resonator and tip geometry (VELM). VCSEL electro-optic characteristics after microtip fabrication were also studied: a lasing threshold modification is observed. This effect is dependent on tip height and on tip curvature radius. Nevertheless, laser emission wavelength is almost unchanged. This behavior is also well reproduced by VELM simulations. Finally, the focusing properties of fabricated tips were characterized and successfully compared to optical modeling using an FDTD simulation tool. Thanks to this study, we can conclude that using this fabrication method, laser spots focused with sizes ranging from 0.4 to 0.9µm can be obtained, as well as working distances from 0,2 to 1µm, with VCSEL emitted power equal to at least 80% of the initial one. It is worth noting that RoC tip could be further increased by deporting tip fabrication from VCSEL surface on an intermediate layer or by modifying fabrication conditions to obtain larger working distances. As final remarks, we would like to outline few potential applications of fabricated devices. The photo-detection properties of VCSELs diodes relying on optical feedback or on lateral internal detection could be exploited in these micro-tipped devices to achieve a miniaturized multi-functional probe [10,11]. Specific designs including packaging consideration can allow for insertion of these compact probes in SNOM systems. Moreover, such compact self-focusing emitting devices are of great interest for butt coupling to optical waveguides .
The French National Research Agency (ANR) is gratefully acknowledged for financial support (ANR-09-BLAN-0168-01) as well as Region Midi-Pyrénées (FIAB SU-8 project). The authors would also like to thank Oclaro Inc. (Zürich, Switzerland) and Philips Technology GmbH U-L-M Photonics (Ulm, Germany) for providing VCSEL devices.
References and links
1. R. Bachelot, P. Royer, G. Wurtz, C. Ecoffet, A. Espanet, and D. J. Lougnot, French Patent N° PCT 9814385, (2001).
3. V. Bardinal, B. Reig, T. Camps, E. Daran, J. B. Doucet, C. Vergnenègre, C. Turck, J. P. Malval, D. J. Lougnot, and O. Soppera, “A microtip self-written on a Vertical-Cavity Surface-Emitting Laser by photopolymerization,” Appl. Phys. Lett. 96(5), 051114 (2010). [CrossRef]
4. C. Croutxe-Barghorn, O. Soppera, L. Simonin, and D. J. Lougnot, “On the unexpected rôle of oxygen in the generation of microlens arrays with self-developing photopolymers,” Adv. Mater. Opt. Electron. 10, 25–38 (2000). [CrossRef]
5. I. Dika, J. P. Malval, O. Soppera, V. Bardinal, D. Barat, C. Turck, A. Spangenberg, and A. Bruyant, “Near-infrared photopolymerization: initiation process assisted by self-quenching and triplet-triplet annihilation of excited cyanine dyes,” Chem. Phys. Lett. 515(1-3), 91–95 (2011). [CrossRef]
6. O. Soppera, S. Jradi, and D. J. Lougnot, “Photopolymerization with microscale resolution: influence of the physico-chemical and photonic parameters,” J. Polym. Sci. Part A: Polym. Chem. 46(11), 3783–3794 (2008). [CrossRef]
8. C. Pang, F. Gesuele, A. Bruyant, S. Blaize, G. Lérondel, and P. Royer, “Enhanced light coupling in sub-wavelength single-mode silicon on insulator waveguides,” Opt. Express 17(9), 6939–6945 (2009). [CrossRef] [PubMed]
9. R. Bachelot, S. Blaize, C. Pang, A. Bruyant, and P. Royer, “Polymer-tipped optical fibers,” Fiber Integrated Opt. 27(6), 542–558 (2008). [CrossRef]
10. T. Camps, C. Bringer, V. Bardinal, G. Almuneau, C. Amat, E. Daran, P. Dubreuil, J. B. Doucet, and C. Fontaine, “High sensitivity integrated lateral detection in VCSELs,” Electron. Lett. 41(3), 129–131 (2005). [CrossRef]
11. D. Heinis, C. Gorecki, C. Bringer, V. Bardinal, T. Camps, J. B. Doucet, P. Dubreuil, and C. Fontaine, “Miniaturized scanning near-field microscope sensor based on optical feedback inside a single-mode oxide-confined Vertical-Cavity Surface-Emitting Laser,” Jpn. J. Appl. Phys. 42(Part2), L1469– L1471 (2003). [CrossRef]
12. T. Yamashita, A. Kawasaki, O. Watanabe, and M. Kagami, “Fabrication of self-written polymeric waveguides using near-infrared light from single-mode devices,” Microoptics Conference (MOC’11), 2011 17th, Sendai, Japan, 1–2, Oct. 30 2011-Nov. 2, 12460175 (2011).