Active control and spatial mapping of mid-infrared propagating surface plasmons

Surface waves on metal films with subwavelength features and tunable optical resonances are excited with a quantum cascade laser. The resulting transmission through, and propagation on, the metal/dielectric interface is measured, both spectrally and spatially.


Introduction
It has been known for some time that metallic films with periodic subwavelength apertures or corrugations can exhibit unique optical properties. Ebbesen et al studied such structures and demonstrated the phenomenon of extraordinary optical transmission (EOT) where, at certain wavelengths, more light passes through the sample than predicted by aperture theory [1]. Recently, Liu and Lalanne have developed a microscopic theory for the EOT phenomenon which suggests that the enhanced transmission through EOT gratings is in fact due to a combination of incident light coupling to both surface plasmon polaritons (SPPs) and a rapidly decaying quasi-cylindrical wave (CW) at the metal/dielectric interface [2]. While the work of Liu and Lalanne focuses on the near-IR/visible spectral range, it is noted that relative importance of the two EOT mechanisms (SPPs and CW) changes as the resonant wavelength of the EOT grating is increased. Specifically, coupling to SPPs becomes weaker, and the EOT process is dominated by the CW for longer wavelengths.
In this work we investigate the coupling of incident coherent light to EOT gratings designed for resonance in the mid-infrared (mid-IR). The study of mid-IR plasmonic structures is of significant interest, not only because this spectral range has been, for the most part, overlooked in most of the theoretical and experimental plasmonic work to date, but also because of the interest in novel optical and optoelectronic mid-IR devices for sensing and security applications. In order to measure the interaction of coherent mid-IR light and a tunable mid-IR EOT grating, we have developed a novel spatially-resolved mid-IR spectroscopy experiment. We demonstrate that the propagation of surface waves on the EOT grating can be imaged using such an experimental set-up, and that by controlling the resonance of our plasmonic structure, we can switch the EOT grating so that the incident light is either transmitted or coupled to propagating surface waves. We demonstrate that the spectral position of the grating transmission maximum is distinct from the spectral location corresponding to maximum propagation.

Experiment and Results
An EOT grating fabricated on semi-insulating GaAs, and designed for peak transmission near 10μm is used as the plasmonic device for this experiment. The EOT resonant wavelength is tuned thermally, from room temperature (RT) to 235ºC, resulting in a 20cm -1 peak shift [3]. A dual wavelength, liquid nitrogen-cooled, quantum cascade laser (QCL), operating in pulsed mode and emitting at ~5.6μm and ~9.7 μm, was used as the exciting source. The laser's long wavelength line is nearly resonant with the EOT grating's primary transmission peak, while the 5.6 μm peak is far from any plasmonic resonance (Figure 1(a)).
In order to investigate the transmission through, and propagation on, our gratings, the direction of expected propagation must be determined. At normal incidence, for light polarized in the x-direction, the forward (1,0) and backward (-1,0) propagating SPP modes are degenerate. However, for θ≠0, these two SPP modes split, which is typically evidenced by a splitting of the primary peak in the EOT grating transmission spectra (Figure 1(b)). Thus, photons resonant with the lower frequency (υ -) peak should couple to SPPs propagating in the -x direction, while those resonant with the higher frequency (υ + ) transmission peak would be expected to couple to SPPs propagating in the +x direction. However, if the spectral properties of the grating can be tuned, then a monochromatic light source could couple to plasmons propagating in opposite directions simply by tuning the grating.  In order to image the propagation of the surface modes, a razor blade was attached to a motorized translational stage, and aligned to travel across the metal side of the sample, held at 8º with respect to the incident laser light. By stepping the blade and collecting spectra at each blade position, the spatial dependence of the light transmitted through the grating can be determined. The 9.7 μm laser is focused onto the EOT surface to a spot size of 75 μm and transmitted/scattered light as a function of wavenumber and x-position is measured across the sample surface. Figures 2(a-c) show the results from the above experiment. At RT, with the x-polarized QCL nearly spectrally coincident with the (-1,0) peak of the EOT grating, no propagation is seen on the metal surface. Similarly, no propagation is seen for y-polarized light, which is to be expected as this light should only couple to SPPs propagating in the vertical direction. However, as the EOT grating transmission spectrum is shifted so that the QCL emission now sits on the high-energy falling edge of the (-1,0) EOT peak, a clear propagation tail is seen. A propagation length of 384μm is extracted from this data, in good agreement with the ~900 μm propagation length expected for a smooth GaAs/Au interface.
We demonstrate the ability to selectively couple to propagating surface modes on an actively tunable plasmonic structure. While we have utilized thermal tuning here, optical or electronic tuning mechanisms may offer faster device operation, while more complicated plasmonic device structures offer the possibility of developing multidirectional plasmonic router and switching devices.