Single and coupled degenerate defect modes in two-dimensional photonic crystal band gaps S. Mahmoodian, * R. C. McPhedran, and C. Martijn de Sterke CUDOS, School of Physics, University of Sydney, NSW 2006 Australia K. B. Dossou, C. G. Poulton, and L. C. Botten CUDOS, School of Mathematical Sciences, University of Technology Sydney, NSW 2007 Australia Received 22 October 2008; published 15 January 2009 We investigate the formation and coupling of defect modes in two-dimensional photonic crystal PCband gaps associated with degenerate edges. Using a method based on Green’s functions and perturbation theory, we derive a condition for the degeneracy of a defect mode in a PC band gap. We show that the interaction between multiple degenerate defects splits this degeneracy and provide a semi-analytic model for the splitting using a 4 4 tight-binding matrix. We observe that the structure of this tight-binding matrix is related to the overlap and corresponding symmetry of the defect modes. We confirm our analysis by comparing with numerical results. DOI: 10.1103/PhysRevA.79.013814 PACS numbers: 42.70.Qs, 42.25.Fx, 42.25.Bs I. INTRODUCTION Photonic crystals PCsare constructed by arranging di- electric material in a periodic fashion 1. This arrangement often consists of circular inclusions forming hexagonal or square lattices, thereby having planes of rotational and re- flection symmetry. Highly symmetric two-dimensional 2D structures often have modes that exist in degenerate pairs. Degeneracy results from the frequency of modes being un- changed after particular symmetry operations 1; that is, the rotated or reflected mode “sees” the same lattice as the origi- nal mode. The occurrence of these degeneracies in perfectly periodic PCs is well understood 1,2. Since the construction of the first PCs with band gaps, much focus has been directed towards the study of localized defect modes. Defects can be introduced into PCs by break- ing the periodicity in some fashion, leading to the formation of levels in band gaps associated with localized modes. De- fect modes created at degenerate band edges come in pairs. Whether these modes are degenerate depends on the nature of the defect and of the band-edge mode. Band-edge degen- eracies can occur in two forms: different bands having the same frequency on the band edge or two points in a band having the same frequency, while not separated by reciprocal lattice vectors. In this paper we discuss the latter type as it is much more common in 2D PCs. Currently, there are many fully numerical studies on de- fect modes that also deal implicitly with defect-mode degen- eracy. For example, Sakoda and Shiroma 3calculated the frequency of defect modes in the first three band gaps of a square lattice, classifying degenerate and nondegenerate de- fect modes by their symmetry. Kuzmiak and Maradudin 4 used finite-difference time-domain simulations to model de- fect frequencies in the first three band gaps of a hexagonal lattice and were able to identify all degenerate and nonde- generate modes belonging to the C 6v mode class. There are, however, no rigorous analytic studies determining under what conditions defect modes are degenerate. The first part of our paper accomplishes precisely this task; that is, we derive a condition to determine whether a perturbation in a periodic PC lattice leads to degenerate or nondegenerate de- fect modes. The PC lattices we are considering are assumed to be made from nonmagnetic, lossless dielectric material forming 2D hexagonal or square lattices with circular inclusions. De- fects are constructed by altering an inclusion such that its symmetry and that of the lattice is preserved. This can be achieved, for example, by changing the refractive index of an inclusion or altering its radius. We have previously shown that in all band gaps the cre- ation of a shallow defect in a PC leads to the formation of defect modes possessing the symmetry of the band-edge Bloch mode 5. Hence, for band gaps with degenerate band- edge modes, the creation of a defect leads to the formation of two localized defect modes. If the defect is created in an arbitrary fashion, in general, the degeneracy is broken and the two defect modes have different frequencies. Painter et al. 6illustrated such a case, where creating a defect which had a lower symmetry than that of the lattice split a degen- eracy. In this paper we demonstrate that the converse is not necessarily true. We show that defects created by preserving the circular symmetry of an inclusion do not necessarily lead to defects modes that are degenerate. In fact, we show that the condition for degeneracy of a defect mode created at a degenerate band edge is twofold: iThe two degenerate band-edge Bloch modes must have the same electric energy inside the defect region. iiThe electric energy overlap of the two band-edge Bloch modes must vanish over the defect region. Photonic crystal components such as channel drop filters 7and coupled cavity waveguides 8rely on the coupling of defect modes. A well-known method for treating modes of any system, not necessarily optical, with weak coupling is the tight-binding TBmethod 9. More recently, the TB method has been applied to nondegenerate PC defects where the defect modes perturb each other weakly 8,10. In these cases the treatment assumed that each defect had only a single nondegenerate defect mode. As mentioned, defect * sahand@physics.usyd.edu.au PHYSICAL REVIEW A 79, 013814 2009 1050-2947/2009/791/01381412©2009 The American Physical Society 013814-1