摘要： Bloch surface waves (BSWs) are surface states excited at the interface between a one-dimensional dielectric photonic crystal (1D-PC) and some ambient material. They are promising alternatives to propagating surface plasmon polaritons thanks to their much longer propagation lengths (up to thousand times the wavelength) that is not limited by absorption and the opportunity to excite them in both polarizations. They are currently considered for multiple applications in integrated optical systems or for sensing devices [1]. To control the propagation of BSWs, a structured functional layer is deposited on top of the 1D-PC. It consists of a thin dielectric layer with a desired shape that controls locally the dispersion relation of the BSW. This allows to implement, in general, optical elements with functions on demand. However, a severe limitation is the finite index contrast to be induced with the functional layer. For typical combinations of material and geometries, an index contrast only in the order of ~ 0.1 is achieved. This is insufficient to tightly focus a BSW with elements perceived by a rational design approach, e.g. with lens-like elements [2]. To solve this problem, we apply here methods developed in the context of computational photonic material design to perceive functional elements that focus BSWs highly efficiently [3]. We basically aim to solve the inverse problem. Computationally, we rely on a finite-difference frequency method as the Maxwell solver. To identify suitable structures, we discretize a finite spatial domain (typically in the order of 40μm x 10μm) where each pixel either is or is not covered by the functional layer. By systematically flipping each pixel, we can optimize structures that can focus the BSW into a predefined spatial domain. For a spatial domain directly behind the element, the BSW can be focused even below half its wavelength (a selected example is shown in Fig. 1a). To verify the findings, we fabricated the respective structures and measured the optical near-fields above them (a selected example is shown in Fig. 1b). The comparison between the simulation and the measurement shows an impressive agreement and allows to verify the ability to tightly focus the BSWs (Fig. 1c). This work introduces a new computational framework to design functional elements that can be used to control the propagation of BSWs and verifies it experimentally. Our approach is suitable for other material platforms where the limited index contrast is an obstacle to control the propagation of some waves. The elements we have studied can find immediate application in lab-on-chip systems where tightly focused BSWs interact with materials carried in fluidic channels to perform spectroscopic measurements.