Experimental investigation of new optical antenna designs found by a genetic algorithm

Most currently studied optical antenna geometries are based on radio wave antenna designs. However, material properties at optical frequency largely differ from the perfect metal case at radio frequencies. Recently, evolutionary algorithms (EA) [1] were used in the field of Plasmonics to find novel geometries however with strongly limited configuration space [2]. We developed an EA capable of handling complex multiparticle geometries. In a first application we optimized near-field intensity enhancement produced by antenna structures in a single point [3]. We found a novel geometry, i.e. a nano antenna/split-ring hybrid antenna for which the near-field intensity enhancement is strongly increased. Yet these structures are not suitable for experimental realization. Here we show the results of an adapted EA which uses primitive elements that are compatible with an experimental fabrication step using FIB. The performance of the realized structures is characterized by means of confocal two-photon-photoluminescence-(2PPL)-microscopy. We find that the hierarchy of performances found in individuals taken from sequential generations of the EA can be reproduced in fabricated structures.