Thermal poling of thin silica glass films: Design rules for optical fibers and waveguides Yves Quiquempois * Université des Sciences et Technologies de Lille, Laboratoire PhLAM, Bâtiment P5, 59655 Villeneuve d’Ascq Cedex, France Nicolas Godbout and Suzanne Lacroix Centre d’optique, photonique et lasers, Laboratoire des fibres optiques, Département de génie physique, École Polytechnique de Montréal, C.P. 6079, Succ. Centre-ville, Montréal (Québec) H3C 3A7 Canada Received 2 January 2005; published 16 June 2005 The charge migration models which account for the creation of a second-order nonlinear susceptibility in poled silica glasses give significant results if the sample thickness is at least 1 order of magnitude larger than the induced nonlinear layer thickness. We present herein a model valid for thin samples applicable to twin-hole optical fibers or waveguides. DOI: 10.1103/PhysRevA.71.063809 PACS numbers: 42.65.An, 42.70.Ce, 66.30.Dn, 72.20.Ht It is now well established that the creation of a second- order nonlinear susceptibility in silica glasses by thermal poling is mainly due to charge migration occurring during the process. The existing models predict that a cation deple- tion region is induced near the anode. The thickness of this region depends mainly on the initial concentration of cations c 0 and on the applied voltage V app 1–3. If we neglect charge injection, the thickness of the cation depletion region is given by the following equation: x a = 2 ec 0 V app - V t , 1 where is the material permittivity and e is the elementary charge. V t is a threshold voltage below which no 2 suscep- tibility is induced 3. The expression of x a in Eq. 1is identical to that of x a in Eqs. 19of Ref. 3. The superscript is added to emphasize the fact that this expression is valid when the total thickness of the sample is at least 1 order of magnitude larger than x a . For example, in high purity silica glasses, x a is typically larger than 20 m. The value given by Eq. 1is then a good approximation for larger than about 200 m, i.e., bulk silica samples. In the opposite case x a , the spatial electric field distribution is expected to strongly depend on the thickness of the sample. This fact raises an important question about the efficiency of poling experiments in thin silica samples such as twin-hole optical fibers or waveguidesin which the distance between elec- trodes could be smaller than 20 m 4. The blocking elec- trode condition assumed in the present paper, i.e., the fact that the charge injection phenomenon is neglected, is a valid assumption in the case of short poling duration. Indeed, there exists experimental observation 5and theoretical demon- stration 6of the existence of an optimum poling duration for maximizing the magnitude of the recorded electric field. Beyond this optimum, the maximal value of the built-in elec- tric field decreases due to charge injection phenomena 5. For poling durations shorter than this optimum, the charge injection can be neglected in first approximation. The purpose of this paper is to give a model for the de- termination of the 2 spatial distribution within thin silica samples after they are thermally poled and to propose some simple rules for the design of the optimal geometry of waveguides and twin-hole optical fibers. The electric field EFExinside the sample during pol- ing is related to ithe chemical composition of the glass through the initial concentration of alkali ions c 0 and the threshold voltage V t , iithe poling conditions through the applied voltage V app and the poling temperature T, and iii the geometric characteristics of the sample via its thickness . According to the charge migration model introduced in Ref. 3, specifically Eqs. A36, A37, and A43therein, the steady-state EF distribution Exduring polingcan be expressed as Ex= 2c 0 e V app - V t + E min 2 - c 0 e x 0 x x a E min x a x , 2 where E min is the value of Exoutside the anodic layer of thickness x a . A sketch of the electric-field distribution during poling external electric field appliedis shown in Fig. 1. E in Fig. 1 corresponds to the difference between the maximal *Electronic address: yves.quiquempois@univ-lille1.fr FIG. 1. General shape of the electric-field distribution during the poling process while the external electric field is applied. The sample thickness can be larger or smaller than x a . PHYSICAL REVIEW A 71, 063809 2005 1050-2947/2005/716/0638094/$23.00 ©2005 The American Physical Society 063809-1