We demonstrate machining of precision slots in silica with nanoscale roughness for applications in photonics. Using our in-house developed milling system we have achieved machined slots with surface roughness of 3.0 nm (Sa) and 17 µm depth of cut. This result represents eight times improvement in surface roughness and forty times increase in depth of cut than previously reported. We also demonstrate integration of these milled slots with UV-written waveguides and Bragg gratings to create optical refractometers, based on monitoring Fabry-Pérot spectral fringe changes.
© 2015 Optical Society of America
Silica is an important material for a host of photonic applications, including: fiber optics, arrayed waveguide devices , planar lightwave circuits , integrated quantum optics [3,4], delay lines , MOEMS/MOMS [6,7] and bulk optics [8,9]. However, silica photonics presently suffers a lack of machining techniques suited to producing three dimensional features on the tens of micron scale for photonic applications. Machined features that optically interact with photonic waveguides typically need smooth features in the nanometer regime [10,11] and form requirements of hundreds of nanometers . Here we demonstrate for the first time Flame Hydrolysis Deposition (FHD) silica physically micromachined using a precision milling technique where a single pass achieves a smooth, chip-free slot with a tens of micron depth of cut and form control to fractions of a micron.
In photonics, structural features such as a milled slot allow the evanescent optical mode of a waveguide to interact with nearby media, allowing the guided light to be perturbed. This type of device lends itself to applications in biological sensing , plasmonic devices , refractometers  and chemical sensing . Figure 1(a) and Fig. 1(b) shows a schematic and a micrograph, of the micromilled evanescent device integrated with waveguides and Bragg gratings presented in this work. The effectiveness of the interaction with the evanescent field will depend on propagation loss and proximity of the waveguide to the slot, as illustrated in Fig. 1(a). Figure 1(b) shows an array of milled slots with an integrated waveguide and Bragg gratings that are being illuminated by 633 nm light via a pigtailed optical fiber launch. Analysis of the fabrication conditions required for low-loss micromilled structures and their subsequent use as an optical evanescent field refractive index sensor are discussed in this paper.
Our precision milling technique [Fig. 2(a)] machines silica with smooth, chip free surfaces by utilizing the ductile cutting regime, which removes material via plastic deformation, the result of which is shown in Fig. 2(c). Figure 2(b) depicts the stress/strain characteristics of silica , where the curve represents the material’s reaction to structural change and tells us about its strength, failure mechanics and machining characteristics. Ductile machining occurs if the strain applied allows for the material to behave plastically i.e. between the two red arrows in Fig. 2(b). To achieve this, the applied strain rate must be controlled by applying the correct machining parameters: tool choice, translation speed, rotational speed, and depth of cut . Typically, reducing the translation/rotational speed and depth of cut subsequently decreases the strain rate and thus can initiate ductile mode machining. The tool’s material and geometry are also important factors in controlling strain rate; sharper tools with fewer flutes and less aggressive rakes can also help reduce the applied strain and again instigate ductile mode machining. The transition between brittle and ductile regime machining in FHD silica and its dependence on strain rates is complex  and highly material dependent and we would expect it to depend strongly on both the FHD silica composition and thermal treatment. Arif et al. suggests that for ductile regime micromilling in soda lime glass a feed rate of <45 nm per cutting flute is required . Outside the narrow regime of plastic deformation a more conventional brittle machining process occurs in glass. This brittle type machining is caused by applying a shear stress at the cutting face that exceeds the material’s plastic limit thus creating cracks that propagate and chip away material; producing an unsmooth, pitted and cracked surface . Figure 2(c) is a micrograph of a slot machined in the ductile cutting regime, the reader’s attention is drawn to the similarity in machined and unmachined glass. Figure 2(d) is a micrograph of a slot milled in the brittle regime where large chips and deep pits are evident. Figure 2(c) and Fig. 2(d) depict the marked difference between the two machining regimes in terms of the surface generated.
There exist a plethora of techniques for three dimensional micro-structuring of silica on the tens of micron scale. These will be discussed with attention to achievable surface roughness and form. Wet etching of silica often uses hydrofluoric acid and using this Nagarah et al. have performed 600 μm deep etching of fused silica, with a surface roughness of 10 nm (Ra) . Even though smooth features were produced, poor sidewall verticality was achieved thus showing that form is difficult to control to the hundreds of nanometer scale required for waveguide interactions. Silica waveguide structures have been constructed using dry etching on the tens of micron scale by Sheng et al. . Using an inductively coupled plasma with CHF3 and CF4 as reactive ions, the silica was etched to a depth of 17 μm. The etched waveguide sidewall surface roughness was measured to be ~100-200 nm from qualitative analysis, which is too rough for low loss light interaction. Laser machining of silica has recently been shown to produce smooth, micron scale features with the ability to manipulate features in three dimensions. Hunger et al. have shown that micro lenses can be produced with pulsed 10.6 µm laser light, with a typical surface roughness of 0.2 nm (Sq) being achieved and depths of cut ranging from 0.01 to 4 µm . Evaporation of the silica is the main mechanism of material removal, while a low viscosity melt layer creates low surface roughness. However, it is difficult for this type of laser machining to create deep structures or surfaces without a rounded shape because of the thermal nature of surface formation .
Precision micromilling has previously been investigated as a route to machining silica in smooth, chip-free fashion, albeit with variable results. Arif et al. demonstrated ductile regime slot milling with a surface roughness (Sa) of 23.8 nm , but limited to a 400 nm depth of cut per pass. Foy et al. utilized a ball nosed slot mill at 45° to the silica, producing smooth, chip free surfaces but with a relatively high surface roughness of ~60 nm (Sa) at a 17.0 μm depth of cut . Morgan et al. produced 50 μm diameter grinding tools with micro electro-discharge machining in polycrystalline diamond. These tools ground 1.5 μm deep slots in Ultra Low Expansion glass (Corning Code 7972) and achieved a surface roughness (Sa) of 5.7 nm but was limited to a 100 nm depth of cut per pass . Neither Arif or Foy reported on form measurements within these works and neither demonstrated interaction with an optical waveguide. Morgan did include a slot profile and deviations in slot form were on the hundreds on nanometer scale. To this end, in this paper we will demonstrate that precision micromilling can be optimized to achieve slots with nanoscale surface roughness and a micron scale depth of cut, providing an attractive fabrication technique for three dimensional micromachining for silica photonic applications. Table 1 summarizes the various techniques for three dimensional micro-structuring of silica on the tens of micron scale.
The optical system performance of our milled-slot refractometers stems from the interaction of the evanescent field with a liquid analyte deposited into the slot. The fidelity of this interaction will depend on both propagation loss and proximity of the waveguide to the slot. The importance of surface roughness to propagation loss can be explained by the Payne and Lacey model . Structures physically machined to interact with waveguides will modify the surface roughness. The Payne and Lacey model tells us that this will affect the propagation loss via altering the scattering characteristics of the waveguide. The propagation loss is affected by both short and long range surface roughness, causing stochastic scattering and Bragg type scattering, respectively. The Payne and Lacey model shows that by reducing surface roughness, and thus light scatter, the associated propagation loss is reduced; hence smooth features are desired as verified by [26,27] and . Typically, in etched silica waveguides sidewall surface roughnesses of ~10 nm have been achieved [28,29] and propagation losses as low as 0.01 dB/cm for these types of waveguides have been shown in low refractive index contrast systems . However, this ultra low loss occurs because of the close match of refractive index between the core and cladding glass layers, leading to extremely low scatter. Therefore, a micromilled slot with surface roughness of <10 nm should enable acceptably low-loss optical interactions, assuming a reasonably close refractive index match to the liquid analyte, i.e. oil or water. Within this study a micromilled slot with a surface roughness of 3.0 nm (Sa) has been achieved.
Evanescent field interactions can also be severely affected if the offset between waveguide and mechanical feature (slot) is not optimized for both proximity and roughness, as varying form will affect both penetration depth and scatter. Thus minimizing slot form variation is critical to the performance of evanescent field devices. In this paper, the waveguide offset is defined as the mean distance of the machined surface with respect to the waveguide. Figure 1(a) illustrates this offset with respect to waveguide and slot. For the direct UV written waveguides we have used within this work the effect of varying waveguide offset can be modeled to discover how this affects the optical power of the evanescent field. An example waveguide structure and refractive indices are shown in Fig. 3. The upper cladding height was varied from 0 to 15 µm and the ratio of the total optical power in an 80 nm layer of silica above the cladding (denoted by the red line) was calculated using the commercially available mode solving software (Fimmwave, Photon Design).
The results of this model predict an exponential relationship between waveguide offset and the optical power located at the interface above the waveguide. If 100% of the optical power is present when the waveguide offset is zero, a 0.1 µm increase in waveguide offset reduces the optical power at the interface by 7.4%, a 0.5 µm offset reduces the optical power by 33.0% and a 2 µm offset by 88.9%. The tolerance of the waveguide offset is dependent on the application and in general the waveguide offset should be reduced to <0.1 µm to achieve efficient evanescent coupling and consistent modal interaction.
The following sections describe our high precision micromill and examples of silica machined using this system. Metrology data is provided for silica machined in both the brittle and ductile milling regimes and the achieved nanometer scale surface roughness. We further combine slots milled with optimised machining parameters with photonic waveguides and Bragg gratings to create an evanescent refractometer. The device is then used to sense refractive index change, which is achieved through monitoring Fabry-Pérot spectral fringe changes.
2. Precision micromill
Cutting experiments were performed using our in-house developed precision milling machine . The doped silica substrates used in this research were fabricated by Flame Hydrolysis Deposition (FHD) on a silicon wafer (silica-on-silicon); primarily because of its use within the telecommunications sector in the form of arrayed waveguide grating devices and planar lightwave circuits . A large machine parameter space was chosen by varying spindle rotational speeds from 10 to 60 krpm and translational speeds from 0.25 to 1.5 mm/min, these feed rates being similar to the ones used by Arif et al.  who demonstrated ductile regime micromilling. Batches of slots were prepared at fourteen different machining parameters, each 1.5 mm in length. The slots were milled into the top surface of the silica and photographic examples of slots are provided in Fig. 1(b) and Fig. 2(c). In each case the workpiece was mounted to the workbed via low-viscosity wax. For cutting, a new tungsten carbide two-flute slot mill with a Chemical Vapor Deposited (CVD) diamond coating (diamond size 2-4 µm) and a diameter of 254 µm was used. A deionized water jet was used to provide tool and workpiece coolant at a rate of ~175 ml/min in every stage of machining and dressing, which is the maximum flow rate of our system. We note that dry-mode machining results in catastrophic sample failure so continuous coolant flow is necessary. Key differences in our study compared with earlier works, is the machine tool path, and pre-machining tool dressing. Our machine tool path has been developed such that the entrance and exit of each slot is approached via a linear gradient, to a final depth of 17.0 µm, the entrance and exit both being 0.5 mm long. This was found to be a successful way of suppressing brittle machining and preventing premature mill wear and was far more successful than standard plunge milling. To ensure mill concentricity to the spindle and reduce static runout, a pre-machining tool dressing routine was utilized. The dressing routine consisted of drilling into the silicon carbide, to a depth of ~20 μm, with a translational speed of ~2.5 mm/min in the z axis and was repeated five times.
3. Milled silica metrology
After the slots were machined, surface metrology was carried out using a white light interferometer (Zemetrics ZeScope) to measure surface roughness at the bottom of each slot, this being the part of the feature that will interact with the evanescent wave of local waveguides in later refractometer experiments. The white light interferometer has a vertical spatial resolution of 1 nm and a lateral spatial resolution of ~500 nm while using a 50x objective, which was used to collect all surface roughness data within this paper. Each surface roughness value was calculated by selecting twenty 10 x 10 μm square samples from each 90 x 70 μm white light interferometer scan, in five columns and four rows. Rather than using the entire rectangular white light interferometer scan area, smaller samples were used to stop the bias of surface texture effecting average surface roughness calculated from the difference in lengths of the white light interferometer scan areas. Polynomial leveling was applied before the three dimensional average surface roughness (Sa) was calculated. Both leveling and the surface roughness calculations were performed using SPIP, Image Metrology A/S software. A mean was then taken over the twenty square samples and the Sa plotted in Fig. 4 against feed rate. Where the feed rate is defined as the distance travelled along the sample for one mill revolution.
Figure 4 shows that for feed rates <20.8 nm/rev the surface roughness (Sa) reduces below 10 nm with a small standard error, showing good consistency in surface roughness over the twenty metrology samples. Conversely, for feed rates >20.8 nm/rev a rougher surface finish is generated with Sa larger than 10 nm and larger variance in standard errors. Figure 4(b) shows that the smallest surface roughness of 3.0 nm (Sa) was achieved at a machining rate of 6.3 nm/rev. A feed rate of 6.3 nm/rev corresponds to a rotational speed of 40 krpm and a translation speed of 0.25 mm/min. A large increase in surface roughness for samples machined at 10 krpm is evident. This is caused by higher feed rates leading to increased strain rates on the sample, such that the machining becomes dominated by brittle mode fracture causing high surface roughness, chipping and cracking. Of the four machining feed rates that achieved surface roughnesses <4.3 nm (Sa), the feed rate at 18.8 nm/rev demonstrated both 4.3 nm roughness and the smallest number of sidewall defects and chips and thus was chosen for further investigation. The feed rate of 18.8 nm/rev corresponds to a rotational speed of 40 krpm and a translation speed of 0.75 mm/min. Figure 5 shows examples of the white light interferometer plots that were used to calculate the surface roughnesses for Fig. 4.
Figure 5(a) shows a white light interferometer scan of brittle mode machining, with micron scale chips and rough surface indicated by the large range (~-800 nm to ~600 nm) of the color bar scale. Figure 5(a) was machined at a feed rate of 150 nm/rev, which corresponds to a rotational speed of 10 krpm and a translation speed of 1.5 mm/min. Figure 5(b) depicts ductile mode milling with no evidence of pitting or chipping and shows a smooth surface and has a color bar (roughness scale) range ~46 times smaller than Fig. 5(a). Figure 5(b) was milled with a 18.8 nm/rev feed rate, corresponding to the optimal machining parameters of 40 krpm rotational speed and 0.75 mm/min translation speed. It should be emphasized that we believe the low amount of chipping and a smooth slot bottom results from a combination of optimized machining parameters, which include: tool choice, translation speed, rotational speed, depth of cut, coolant type and coolant flow rate.
Our demonstrated ability to physically micromill silica with a 17 µm depth of cut and average surface roughness (Sa) of 3.0 nm is unprecedented. Previous work by Arif et al. demonstrated precision milled slots in silica that were smooth and chip free, using a feed rate of 80 nm/rev to achieve surface roughness (Sa) of 23.8 nm but with a cut depth limited to just 400 nm . Our technique enables forty times deeper cutting and eight times smoother features than this result, albeit at four times slower feed rate. Silica has also been machined with ball nosed cutters at 45° to the substrate by Foy et al.. In this case smooth, chip free surfaces were created with surface roughness (Sa) of <60 nm at 17 µm depth of cut using a feed rate of <16 nm/rev . We have created surfaces nearly twenty times smoother at a similar feed rate (18.8 nm/rev) and are not limited by the ball nosed cutter geometry.
4. Integration of slot with waveguides and Bragg gratings
A number of slots were fabricated in silica-on-silicon samples at the optimal machining parameters, which corresponds to a rotational speed of 40 krpm and a translation speed of 0.75 mm/min. Out of eight milled slots a mean surface roughness (Sa) was calculated to be 5.1 nm with a standard deviation of 1.2 nm and a mean depth of cut calculated at 16.6 µm with a standard deviation of 229 nm. In order to demonstrate photonics capability in our machined structures, a slot with surface roughness of 4.9 nm and 17.0 µm depth of cut was inscribed with a waveguide and Bragg gratings. This micromilled slot meets both our target requirements of a slot bottom surface roughness <10 nm and a form error <0.1 µm [see waveguide offset in Fig. 1(a)]. The waveguide and Bragg gratings were created by using Direct UV Writing as described by Sima et al. . The waveguide consisted of a set of spectrally matched Bragg gratings to form a Fabry-Pérot cavity, its design is shown in Fig. 6 and the resultant spectra is shown in Fig. 7(a).
The spectra in Fig. 7(a) were measured while the slot was exposed to air (blue line) and 1.45 refractive index oil (green line) respectively. Spectra collection in both Fig. 7(a) and Fig. 7(b) utilized optical backscatter refractometry (Luna Technologies OBR Model 4400). As refractive index oil alters the effective refractive index of the waveguide between the Bragg gratings, and thus the cavity length, the spectral Fabry-Pérot fringe position changes. The change in the fringe position at ~1550.6 nm [blue curve Fig. 7(a)] was monitored for varying refractive index oils and is shown in Fig. 7(b). The total fringe shift in Fig. 7(b) was ~160 pm for a refractive index change of 1.4 to 1.45. Thus, Fig. 7 shows successful refractive index sensing from a waveguide integrated into a micromilled slot.
A custom precision micromill was developed to perform milling experiments in Flame Hydrolysis Deposition (FHD) silica-on-silicon to achieve surface roughness and depths of cut relevant for photonic devices. By milling in the ductile regime we have achieved surface roughnesses of 3.0 nm (Sa) at 17 µm cut depth using a feed rate of 6.3 nm/rev, which corresponds to a rotational speed of 40 krpm and a translation speed of 0.25 mm/min. Our result pushes the state of the art of silica micro slot milling by increasing the achievable depth of cut by forty times and achieving a surface roughness that is eight times smoother than previously published results. To demonstrate photonics capability, waveguides and Bragg gratings were integrated with a micromilled slot to create a refractometer. Fabry-Pérot spectral fringe shifts were monitored for varying refractive indices, applied to the slot, and a shift of 160 pm was monitored for a refractive index change of 1.4 to 1.45. Therefore micromilling, in terms of both surface roughness and form, has been demonstrated to be a viable and attractive route to fabricate slots in silica for photonic devices and applications.
References and links
1. C. R. Doerr and K. Okamoto, “Advances in silica planar lightwave circuits,” J. Lightwave Technol. 24(12), 4763–4789 (2006). [CrossRef]
2. M. R. Poulsen, P. I. Borel, J. Fage-Pedersen, J. Hübner, M. Kristensen, J. H. Povlsen, K. Rottwitt, M. Svalgaard, and W. Svendsen, “Advances in silica-based integrated optics,” Opt. Eng. 42(10), 2821–2834 (2003). [CrossRef]
4. J. B. Spring, B. J. Metcalf, P. C. Humphreys, W. S. Kolthammer, X.-M. Jin, M. Barbieri, A. Datta, N. Thomas-Peter, N. K. Langford, D. Kundys, J. C. Gates, B. J. Smith, P. G. R. Smith, and I. A. Walmsley, “Boson sampling on a photonic chip,” Science 339(6121), 798–801 (2013). [CrossRef] [PubMed]
6. D. Iannuzzi, K. Heeck, M. Slaman, S. de Man, J. H. Rector, H. Schreuders, J. W. Berenschot, V. J. Gadgil, R. G. P. Sanders, M. C. Elwenspoek, and S. Deladi, “Fibre-top cantilevers: design, fabrication and applications,” Meas. Sci. Technol. 18(10), 3247–3252 (2007). [CrossRef]
7. L. G. Carpenter, C. Holmes, B. D. Snow, J. C. Gates, and P. G. R. Smith, “Photonic microcantilevers with interferometric Bragg grating interrogation,” IEEE Photon. J. 4(5), 1387–1395 (2012). [CrossRef]
8. P. Comley, P. Morantz, P. Shore, and X. Tonnellier, “Grinding metre scale mirror segments for the E-ELT ground based telescope,” CIRP Ann.Manuf. Technol. 60(1), 379–382 (2011). [CrossRef]
9. D. Ravindra and J. Patten, “Duc tile regime single point diamond turning of quartz resulting in an improved and damage-free surface,” Mach. Sci. Technol. 15(4), 357–375 (2011). [CrossRef]
10. F. P. Payne and J. P. R. Lacey, “A theoretical analysis of scattering loss from planar optical waveguides,” Opt. Quantum Electron. 26(10), 977–986 (1994). [CrossRef]
11. F. Grillot, L. Vivien, S. Laval, D. Pascal, and E. Cassan, “Size influence on the propagation loss induced by sidewall roughness in ultrasmall SOI waveguides,” IEEE Photon. Technol. Lett. 16(7), 1661–1663 (2004). [CrossRef]
12. D. Kumar, V. K. Sharma, and K. N. Tripathi, “Design and fabrication of multilayer metal-clad dielectric surface plasmon waveguide polarizers,” Opt. Eng. 45(5), 054601 (2006). [CrossRef]
13. S. Watts, “Bragg gratings: optical microchip sensors,” Nat. Photonics 4(7), 433–434 (2010). [CrossRef]
14. C. Holmes, K. R. Daly, I. J. G. Sparrow, J. C. Gates, G. D’Alessandro, and P. G. R. Smith, “Excitation of surface plasmons using tilted planar-waveguide Bragg gratings,” IEEE Photon. J. 3(5), 777–788 (2011). [CrossRef]
15. C. Holmes, L. G. Carpenter, H. L. Rogers, I. J. G. Sparrow, J. C. Gates, and P. G. R. Smith, “Planar waveguide tilted Bragg grating refractometer fabricated through physical micromachining and direct UV writing,” Opt. Express 19(13), 12462–12468 (2011). [CrossRef] [PubMed]
16. R. M. Parker, J. C. Gates, D. J. Wales, P. G. R. Smith, and M. C. Grossel, “An investigation into dispersion upon switching between solvents within a microfluidic system using a chemically resistant integrated optical refractive index sensor,” Lab Chip 13(3), 377–385 (2013). [CrossRef] [PubMed]
17. H. Eda, “Ductile Grinding of Ceramics: Machine Tool and Process,” in Handbook of Advanced Ceramic Machining, I. Marinescu, ed. (CRC Press, Taylor & Francis Group, 2007), pp. 1–4.
18. T. G. Bifano, T. Dow, and R. O. Scattergood, “Ductile-regime grinding: a new technology for machining brittle materials,” J. Eng. Ind. 113(2), 184–189 (1991). [CrossRef]
20. M. Arif, M. Rahman, and W. Y. San, “Ultraprecision ductile mode machining of glass by micromilling process,” J. Manuf. Process. 13(1), 50–59 (2011). [CrossRef]
21. J. M. Nagarah and D. A. Wagenaar, “Ultradeep fused silica glass etching with an HF-resistant photosensitive resist for optical imaging applications,” J. Micromech. Microeng. 22(3), 035011 (2012). [CrossRef]
22. Z. Sheng, B. Yang, L. Yang, J. Hu, D. Dai, and S. He, “Experimental demnstration of deeply-etched SiO2 ridge optical waveguides and devices,” IEEE J. Quantum Electron. 46(1), 28–34 (2010). [CrossRef]
23. D. Hunger, C. Deutsch, R. J. Barbour, R. J. Warburton, and J. Reichel, “Laser micro-fabrication of concave, low-roughness features in silica,” AIP Adv. 2(1), 012119 (2012). [CrossRef]
24. K. Foy, Z. Wei, T. Matsumura, and Y. Huang, “Effect of tilt angle on cutting regime transition in glass micromilling,” Int. J. Mach. Tools Manuf. 49(3-4), 315–324 (2009). [CrossRef]
25. C. J. Morgan, R. R. Vallance, and E. R. Marsh, “Micro-machining and micro-grinding with tools fabricated by micro electro-discharge machining,” Int. J. Nanomanuf. 1(2), 242–258 (2006). [CrossRef]
26. Y. Hibino, H. Okazaki, Y. Hida, and Y. Ohmori, “Propagation loss characteristics of long silica-based optical waveguides on 5 inch Si wafers,” Electron. Lett. 29(21), 1847–1848 (1993). [CrossRef]
27. M. Ferrera, L. Razzari, D. Duchesne, R. Morandotti, Z. Yang, M. Liscidini, J. E. Sipe, S. Chu, B. E. Little, and D. J. Moss, “Low-power continuous-wave nonlinear optics in doped silica glass integrated waveguide structures,” Nat. Photonics 2(12), 737–740 (2008). [CrossRef]
28. W. T. Li, D. A. P. Bulla, J. Love, B. Luther-Davies, C. Charles, and R. Boswell, “Deep dry-etch of silica in a helicon plasma etcher for optical waveguide fabrication,” J. Vac. Sci. Technol. A 23(1), 146–150 (2005). [CrossRef]
29. W. T. Li, D. A. P. Bulla, and R. Boswell, “Surface oxidation of Al masks for deep dry-etch of silica optical waveguides,” Surf. Coat. Tech. 201(9-11), 4979–4983 (2007). [CrossRef]
30. K. Okamoto, Fundamentals of Optical Waveguides, 2nd ed. (Elsevier Inc, 2006), pp. 342–343.
31. L. G. Carpenter, “Thesis: Precision dicing and micromilling of silica for photonics,” University of Southampton (2013).
32. C. Sima, J. C. Gates, H. L. Rogers, P. L. Mennea, C. Holmes, M. N. Zervas, and P. G. R. Smith, “Ultra-wide detuning planar Bragg grating fabrication technique based on direct UV grating writing with electro-optic phase modulation,” Opt. Express 21(13), 15747–15754 (2013). [CrossRef] [PubMed]