Open Access
Add to BibSonomyAbstract
A behavior is reported where the index change process used for UV writing of integrated optical waveguides in deuterium loaded Ge:SiO2 glass can become unstable and suddenly switch off or on. It is shown that such discontinuities are associated with abrupt changes in the amount of absorbed UV power. We suggest that these events are controlled by a coupling between UV absorption, local heating and the D2-GeO2 reaction rate. From our findings we predict, and confirm experimentally, that strong waveguides can not be fabricated under normal UV writing conditions in thin core layers with a low initial UV absorption. Our findings show that an improved understanding of the waveguide formation process and future process development requires that thermal effects are taken into account.
©2005 Optical Society of America
1. Introduction
Integrated optical waveguides may be fabricated in glass by a process of direct writing with ultraviolet (UV) light, where a photosensitive planar sample is scanned under a focused laser beam to induce a permanent refractive index change [1]. The technique does not require photolithography or etching and is becoming more widely adopted in the research community for increasingly advanced applications [2–5]. To date the best performance has been obtained in Ge-SiO2 glass which has been photosensitized by hydrogen (H2) or deuterium (D2) loading [6].
It is widely believed that direct UV writing of waveguides in silica-based glass involves the same physical processes as fiber Bragg grating fabrication techniques, which have been much more widely implemented and investigated. However, the waveguide formation process can exhibit a behavior where the index change process can become unstable and switch off or on repeatedly, even though care is generally taken to stabilize the processing parameters to the highest degree possible. From a research point of view such a behavior is highly interesting since it seems at odds with proposed models of photosensitivity, while in a fabrication situation it is highly unwanted since it increases losses and degrades reproducibility.
In this paper we describe experimental and numerical investigations of this behavior. Our results show that discontinuities are associated with large changes in the UV absorption and that a local heating of up to several hundred K occurs. The observed discontinuity behavior can be explained as a result of an interplay between changes in UV absorption, local heating and the D2-GeO2 reaction rate. The amount of induced heating, in addition to the presence of Ge and D2, is a critical factor required to initiate the waveguide formation process.
2. Sample and UV writing
The samples used here consist of a silica-on-silicon buffer/core/cladding structure with a photosensitive core layer. The core contains germanium and boron in a relative concentration so that the refractive index is matched (within 5×10-4) to that of the surrounding layers [7]. Core layer index matching enables the UV-written waveguides to exhibit a circular mode profile and low coupling loss to standard telecom fiber [8]. The thickness of the buffer/core/cladding layer is 16/5.4/12 μm. After deposition the sample was annealed at 1100 °C in an oxygen-rich atmosphere. As a consequence hereof the intrinsic concentration of germanium related oxygen defect centers (GODC’s) [9], and therefore also the intrinsic photosensitivity, is quite low.
Prior to UV writing the photosensitivity is enhanced by loading the sample with D2 at 190 bar [6]. UV writing is carried out with a focused 257 nm continuous wave laser beam (measured 1/e2 spot size = 3.1 μm, Rayleigh length = 55 μm) and computer controlled scanning stages [4]. The sample is housed in a small vacuum chamber and cooled to -30 °C to slow down the rate of D2 outdiffusion [10]. To investigate the photosensitivity a series of straight waveguides is written with scan velocities ranging from 20-1000 μm/s. This covers reasonably well the entire range of velocities that can be applied in a production situation, due to limitations on the maximum processing time and the accuracy of trajectory scanning. After UV writing the sample is annealed at 80 °C for 12 hours to outdiffuse residual D2.
Fig 1. Waveguide width (red squares) and index step (blue circles) versus scan velocity. A sudden ‘shut-down’ behavior occurs at 220 μm/s, after which the waveguide becomes very weak.
Fig 2. Close-up view (40×30 μm2) of a Δn discontinuity. Image a) is a brightfield micrograph sensitive to index contrast while image b) shows blue luminescence from GODC’s upon excitation with a weak UV field. The transition occurs over a length of just 2-3 μm, which is comparable in size to the UV spot. In these images it is evident that an oscillation in the scanning stages was present, this has since been removed and we have verified that it is unrelated to the topics discussed in this paper.
3. Index change discontinuity
The waveguide width is measured with brightfield optical microscopy while the peak index step is determined by detecting interference fringes produced by the waveguide and its surrounding area upon illumination with partially coherent light [11]. The width and index step are plotted versus the applied scan velocity in Fig. 1 for an incident UV power of 31 mW. For the lowest scan velocity of 20 μm/s a width of 5.8 μm and an index step of 0.0073 is achieved. For increasing scan velocity both values decrease smoothly until a discontinuity is reached at ~220 μm/s. Here the width suddenly drops from 4.2 μm to ~1 μm and the index step drops from 0.0054 to below the detection limit (<0.001). For higher UV power the ‘critical velocity’ increases, reaching ~540 μm/s for 47 mW and ~870 μm/s for 51 mW. This behavior was also reported recently in luminescence microscopy investigations [12].
Near the critical velocity the Δn process can shut down and turn on several times during one scan, most likely triggered by small irregularities in UV power, scan velocity or material properties. A close-up view through the top cladding of such a discontinuity is shown in Fig. 2, using optical brightfield microscopy and luminescence microscopy [12]. The brightfield image is sensitive to index contrast. The luminescence image is obtained by illuminating the sample with a weak UV field through the cladding and detecting the blue luminescence band which is associated with GODC’s [9,13]. Both images show that the transition occurs over a length of just 2-3 μm, which is comparable in size to the UV spot. In addition, it is seen that the discontinuity occurs in terms of both index contrast and GODC concentration.
4. UV absorption
Since GODC’s are known to contribute strongly to UV absorption at the wavelength used for UV writing [9,13] it is likely that substantial changes in absorption are associated with an index change discontinuity. To investigate this further the amount of UV power that is absorbed in the core layer was measured. The setup for doing so is shown in Fig. 3; it is just the usual UV-writing setup supplemented with a UV mirror and two silicon PIN detectors connected via operational amplifiers to a digital oscilloscope [14]. Due to reflections one detector had to be blocked while using the other. The incident beam passes through the core layer down to the substrate (refractive index 1.59+i3.86 at 257 nm [15]) where 60% is reflected up through the core layer again and finally to detector D1. Detector D1 therefore provides information on the amount of UV light remaining after the core layer interaction. A signal proportional to the amount of incident UV power is provided by detector D2. Ignoring multiple reflections we can write the signal on D1 as:
where c1 is a constant dependent on the glass/silicon reflections, absorption outside the core layer and the detector response for D1. The core layer thickness is d, while β is the core layer absorption coefficient and Pin is the incident UV signal. Measurements on D1 were performed during UV writing close to the critical scan velocity on a normal sample and on a sample that contained only buffer/cladding layers; i.e. without a core layer. The core layer absorption coefficient has then been obtained from the ratio between these signals.
Fig. 4. The UV power absorbed in the core vs. time during a discontinuity event. The transition between ‘low’ and ‘high’ absorption occurs over a time interval of just 6×10-4 sec. An illustration of the corresponding index structure is shown in the inset.
A measurement of the total power absorbed in the core layer is shown in Fig. 4 for a scan velocity of 350 μm/s, just at the time when a discontinuity occurs. For this measurement the incident UV power is 39 mW, measured at the location indicated by ‘PIN’ in Fig. 3. This corresponds to 35 mW entering the core layer when mirror reflectivity, surface reflections and window absorption is taken into account. Before ‘turn-on’ 6.2 mW is absorbed in the core layer, while 27 mW is absorbed after ‘turn-on’, corresponding to an absorption coefficient β=1.8×104 m-1 and β=1.3×105 m-1, respectively. The transition between these two states take place in just ~6×10-4 s. During this time the sample moves only 0.2 μm, which is consistent with Fig. 2 which shows that the length over which the discontinuity occurs is similar to the UV spot size.
Fig. 5. Calculated 2D steady state temperature change profile for the ‘high’ UV absorption state (β=1.3×105 m-1). The top/bottom of the plot corresponds to the glass boundaries, while the layer boundaries and incident/reflected UV beam are indicted with white lines. The peak temperature increase is ~600 K.
5. Thermal modeling
By combining measurements of the sample glass layer thicknesses, the UV beam profile and the absorbed power it is now possible to numerically calculate the temperature, T, induced by the UV beam. Using a cylindrical coordinate system defined by the UV beam axis and assuming rotational symmetry we have in the core layer:
where L1 and L2 define the lower and upper core boundary, respectively. Outside the core layer β=0 so that only the heat diffusion term remains. Rotational symmetry corresponds to the sample being stationary under the UV beam, however this is not a severe approximation for regions close to the optical axis since it will be shown that the thermal rise time is much smaller than the effective UV exposure time. For the specific heat capacity cp, density ρ and thermal conductivity k we used values for silica (cp=800 J kg-1K-1, ρ=2203 kg m-3, k=1.4 W m-1K-1), i.e. the various dopants were not taken into account. The downwards and upwards propagating UV intensity profiles, Idown, Iup are modeled as Gaussian profiles using measured profile data. The substrate is considered as an infinite heat reservoir (T(substrate)=const) and heat conduction at the glass surface is neglected (dT/dz=0). Especially the former boundary condition might seem unreasonable, since 50% of the incident UV is absorbed over a very short distance in the substrate. However, since the thermal conductivity of silicon is ~100 times larger than that of silica the heating effect here is much smaller and it does not affect the temperature in the core layer. To verify this we have numerically simulated the substrate heating and obtain a peak temperature increase of just ~30 K which is well isolated from the core.
Calculations were performed for an incident UV power of 39 mW, i.e. the same as was used for the absorption measurements. The steady state temperature profile is shown as a 2D plot in Fig. 5 along with the layer boundaries and UV beam profile used for the calculation. The vertical profile along the optical axis is shown in Fig. 6. The peak temperature change for the low and high absorption states is ~130 K and ~600 K, respectively. Our calculations show that the thermal response time is quite short, typically a few times 10-5 s. Consequently, during a transition between low and high absorption, which lasts an order of magnitude longer, the peak temperature should nearly be in equilibrium with the instantaneous absorption value.
Fig. 6. Calculated vertical temperature profile corresponding to the measured ‘low’ and ‘high’ UV absorption state (β=1.8×104 m-1, β=1.3×105 m-1). The peak temperature increase is ~130 K for the low absorption state and ~600 K for the high absorption state. The temperature profile due to substrate absorption is also plotted, showing only a modest heating due to the large thermal conductivity of silicon.
Fig. 7. Calculated ‘low’ UV absorption state vertical temperature profile for a thick (5.5 μm) and a thin (2.8 μm) core layer. The peak temperature increase is ~130 K for the thick core and ~80 K for the thin core. Experiments show that strong waveguides can not be written in the thin core.
6. Discussion
The fact that the UV absorption (Fig. 4) changes at the same time as the luminescence intensity (Fig. 2) is a strong indicator that a rapid change in the concentration of GODC’s does occur. The transition has been measured to occur over a time scale of ~6×10-4 s, during which the local heating by the UV beam changes from roughly one hundred K to several hundred K. It is well known that molecular deuterium can be thermally activated to react with GeO2 sites with GODC formation as a byproduct [16,17]. The reaction rate is strongly dependent on temperature, with an activation energy of 29 kcal/mol [16]. From the temperatures calculated above it follows that the initial D2 activation rate changes by roughly 8 orders of magnitude during a discontinuity event. This makes the following scenario quite robust, even if our calculated temperature changes should be incorrect by, say, a factor of two: As a core volume is approached by the UV beam the GODC concentration, and thus also the UV absorption, rises slowly in a positive feedback loop controlled by the D2 reaction rate. If the temperature / effective UV exposure time reaches a certain threshold value the positive feedback loop leads to a fast, run-away process where all the available D2 suddenly becomes thermally activated. It is widely agreed that GODC’s serve as catalyst sites for the usual array of UV induced index change mechanisms [18] and thus the waveguide index step will switch from a low to a high value in this process as well. This scenario accounts for the observed discontinuities in index step, luminescence and UV absorption. It also accounts for the fact that the discontinuity occurs over a length similar to the diameter of the UV beam.
Heating may therefore be an important factor in controlling the waveguide formation process, especially in samples with a low intrinsic concentration of GODC’s. Even though progress has been made in understanding the physical mechanisms of photosensitivity, accurate modeling of the waveguide formation process still remains a daunting task. Our work shows that besides the electronic interactions between various defect species involved in microscopic photosensitivity models [18], the coupling between UV irradiation, UV absorption and heating must also be taken into account. In addition, at the elevated temperatures reported here the diffusion of D2 from non UV exposed regions to the illuminated region becomes important on time scales comparable to the ~0.01 second UV irradiation time [19].
Local heating has previously been suggested to affect the waveguide UV writing process, first as the explanation for difficulties in writing symmetrical splitters and couplers due to redistribution of D2 around illuminated regions [20]. More recently, the coupling of electronic reactions, heating and diffusion has been suggested as a means for explaining why the index profiles of UV written waveguides have a higher degree of morphological complexity than can arise from a monotonical relationship depending only on UV irradiation [12].
The proposed scenario also predicts that there should exist a minimum core thickness for which the induced heating cannot in practice reach the threshold value required to initiate a run-away D2 activation. As the core becomes thinner the surface-to-volume ratio of the illuminated cylinder becomes greater and the importance of heat loss due to diffusion increases. Consequently, the peak temperature will become lower and eventually the temperature rise will not be large enough to facilitate a run-away activation of D2 for practically applicable UV writing parameters. We have confirmed this behavior in samples with a core thickness of 2.8 μm, where only very weak waveguides could be formed. Using secondary ion mass spectroscopy it was verified that the chemical composition of these samples was the same as for those used earlier in this paper. The D2 loading parameters, UV power and the applied range of scan velocities were also the same. Using the measured ‘low’ absorption value (β=1.8×104 m-1) the peak heating in these samples was calculated to be ~80 K as opposed to the ~130 K of the thicker core layer samples. Hence the required critical temperature for run-away thermal activation of D2 and UV writing of strong waveguides is in the 80-130 K range for our experimental conditions. Our claim that heating plays a crucial role in the waveguide formation process is further strengthened by the fact we have successfully written strong waveguides in samples with thin core layers that had not been subjected to a high temperature annealing in oxygen during the fabrication process. In these samples the concentration of GODC’s, and thus also the UV absorption, is intrinsically high so that the requirement for UV induced photosensitivity activation is relaxed.
7. Conclusion
We have described a behavior which occurs during UV writing of waveguides where the index change process can become unstable and switch off or on repeatedly. It has been shown that such discontinuities are associated with abrupt changes in the UV absorption coefficient of the photosensitive core layer. Numerical calculations of core layer heating by the UV beam and published values for the thermally induced D2-GeO2 reaction rate suggest that discontinuity events are controlled by the amount of UV induced heating of the core layer. As a core volume is approached by the UV beam the D2-GeO2 reaction rate, and thus also the UV absorption, rises slowly in a positive feedback loop controlled by the D2 reaction rate. If the temperature / effective UV exposure time reaches a certain threshold value the positive feedback loop leads to a fast, run-away process where all the available D2 suddenly becomes thermally activated. Since GODC’s serve as catalyst sites for UV induced index change mechanisms the induced index step will therefore switch from a low to a high value in this process as well. This scenario accounts for the observed discontinuities in index step, luminescence and UV absorption. The proposed mechanism predicts that thin core layers with a low initial GODC concentration should not be able to sustain a heating sufficient for the formation of strong waveguides. This behavior was confirmed experimentally. To understand the waveguide formation process and to optimize future process development thermal effects due to the UV beam must therefore be taken into account.
References and links
1. M. Svalgaard, C.V. Poulsen, A. Bjarklev, and O. Poulsen, “Direct UV-writing of buried single-mode channel waveguides in Ge-doped silica films,” Electron. Lett. 30, 1401–1402 (1994). [CrossRef]
2. M.Y. Park, W. Yoon, S. Han, and G. H. Song, “Fabrication of low-cost planar wavelength-selective optical add-drop multiplexer by employing UV photosensitivity,” Electron. Lett. 38, 1532–1533 (2002). [CrossRef]
3. G.D. Emmerson, S.P. Watts, C.B.E. Gawith, V. Albanis, M. Ibsen, R.B. Williams, and P.G.R. Smith, “Fabrication of directly UV written channel waveguides with simultaneously defined integral gratings,” Electron. Lett. 38, 1531–1532 (2002). [CrossRef]
4. M. Svalgaard, K. Faerch, and L.-U. Andersen, “Variable optical attenuator fabricated by direct UV writing,” J. Lightwave Technol. 21, 2097–2103 (2003). [CrossRef]
5. C. Peucheret, Y. Geng, M. Svalgaard, B. Zsigri, H.R. Sørensen, N. Chi, H.-J. Deyerl, M. Kristensen, and P. Jeppesen, “Direct UV written Michelson Interferometer for RZ signal generation using phase-to-intensity modulation conversion,” Phot. Tech. Lett. 17, 1674–1676 (2005). [CrossRef]
6. P.J. Lemaire, R.M. Atkins, V. Mizrahi, and W.A. Reed, “High Pressure H2 loading as a technique for achieving ultrahigh UV photosensitivity and thermal sensitivity in GeO2 doped optical fibres,” Electron. Lett. 29, 1191–1193 (1993). [CrossRef]
7. G.D. Maxwell and B.J. Ainslie, “Demonstration of a directly written directional coupler using UV induced photosensitivity in a planar silica waveguide,” Electron. Lett. 31, 95–96 (1995). [CrossRef]
8. D. Zauner, K. Kulstad, J. Rathje, and M. Svalgaard, “Directly UV-written silica-on-silicon planar waveguides with low insertion loss,” Electron. Lett. 34, 1582–1584 (1998). [CrossRef]
9. M.J. Yuen, “Ultraviolet absorption studies of germanium silicate glasses,” App. Opt. 21, 136–140 (1982). [CrossRef]
10. M. Svalgaard, “Effect of D2 outdiffusion on direct UV writing of optical waveguides,” Electron. Lett. 35, 1840–1842 (1999). [CrossRef]
11. T.W. Whitbread, R.A. Betts, and F. Lui, “Non destructive two-dimensional refractive index profiling of integrated optical waveguides by an interferometric method,” Appl. Opt. 30, 4384–4389, (1991). [CrossRef] [PubMed]
12. M. Svalgaard, A. Harpøth, and T. Rosbirk, “Characterization of UV-written waveguides with luminescence microscopy,” Opt. Express 13, 5170–5178 (2005), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-13-13-5170. [CrossRef] [PubMed]
13. L.N. Skuja, A.N. Trukhin, and A.E. Plaudis, “Luminescence in germanium-doped glassy SiO2,” Phys. Status Solidi. A 84, K153–157 (1984). [CrossRef]
14. M. Svalgaard, A. Harpøth, and M. Andersen, “Discontinuities during UV writing of waveguides,” in Proceedings of OSA Topical Meeting on Bragg Gratings, Poling and Photosensitivity, B.J. Eggleton, ed., (Technical Digest Series, Optical Society of America, Washington, D.C., 2005), 121–123.
15. S. Adachi, “Model dielectric constants of Si and Ge,” Phys. Rev. B 38, 12966–12976 (1988). [CrossRef]
16. A. Iino, M. Kuwabara, and K. Kokura, “Mechanisms of hydrogen-induced losses in silica-based optical fibers,” J. Lightwave Technol. 8, 1675–1679 (1990). [CrossRef]
17. P.J. Lemaire, A.M. Vengsarkar, W.A. Reed, and D.J. DiGiovanni, “Thermally enhanced ultraviolet photosensitivity in GeO2 and P2O5 doped optical fibers,” Appl. Phys. Lett. 66, 2034–2036 (1995). [CrossRef]
18. M. Kristensen, “Ultraviolet-light-induced processes in germanium-doped silica,” Phys. Rev. B 64, 144201 (2001). [CrossRef]
19. P.J. Lemaire, “Reliability of optical fibers exposed to hydrogen: Prediction of long-term loss increases,” Opt. Eng. 30, 780–789 (1991). [CrossRef]
20. K. ærch and M. Svalgaard, “Symmetrical waveguide device fabricated by direct UV writing,” Phot.. Tech. Lett. 14, 173–175, (2002). [CrossRef]
