370
J. Eur. Opt. Society-Rapid Publ. 22, 37( 2026)
modulation frequency from 27 Hz to 225 Hz, the resonant leaky-waveguide modes appeared in the normalized signal. However, due to the photo-acoustic cell design, the amplitude of the PAS signal decreases with the increase of f [ 9 ], and the signal to noise ratio of the normalized signal deteriorates. Hence, to eliminate the influence of the substrate, modern PAS set-ups require the power tunability of the excitation light beam. Moreover, polarization control of the absorption in asymmetric NW in a wide wavelength range can be assessed if the lamp source is replaced by a widely tunable laser.
In this work, we apply PAS tunability in terms of light wavelength, and beam modulation frequency, to push the limits of the mode detection in vertically standing NW ensembles. We investigate two samples: GaAs-based NWs, and the same sample asymmetrically covered by Au( 2 out of 6 NW sidewalls). The semiconductor-only sample supports a leaky waveguide resonance at 780 nm. We apply PAS with a laser, widely tunable in the( 680 – 1000) nm range and in power, which allows us to access higher modulation frequency. The normalized absorption signal sharpens when f is changed from 54 Hz to 1225 Hz, as the influence of the substrate decreases. We further show that Au-covered sample exhibits extrinsic chirality in the whole wavelength range.
2 Materials and methods
Figure 1a shows schematic of the photo-acoustic set-up, with sketches of the two investigated samples. The photoacoustic cell is positioned on a rotation / translation stage. The laser wavelength is tunable in the( 680 – 1000) nm range( including tunable power), and the modulation frequency of the chopper frequency is tunable up to 2000 Hz. We enable the precise laser position on the sample by a microscope. In this work, the light is focused to around 300 lm diameter.
The investigated samples are ensembles of vertically standing, coaxial GaAs-AlGaAs-GaAs core-shell-supershell NWs, grown by molecular beam epitaxy on p-Si( 1 1 1) substrates. This process starts with lithography-free Si / SiOx patterns defining the nucleation sites for self-catalyzed growth of GaAs core [ 10 ]. Since the patterns were obtained by means of droplet epitaxy of GaAs nanocrystals, the final ensemble has randomly positioned NWs. The GaAs core growth results in predominantly defect free zincblende structure. Next, AlGaAs shell and GaAs supershell are grown using axial growth promoting conditions. As a result, the NW ensemble has density of 1 NW / lm 2, andagood uniformity of geometric parameters, as shown by detailed statistical analysis of scanning electron microscope images in Ref. [ 7 ].
In the investigated Sample C, a single NW is 4.65 lm tall, and has the overall diameter of 165 nm, which contains 11.6 nm AlGaAs shell, and 5.9 nm GaAs supershell. The shell / supershell growth results in additional parasitic layers on the top of Si substrate( 25 nm and 5 nm of AlGaAs and GaAs, respectively). The investigated Sample C-Au has the same semiconductor base as Sample C, on the top of which a thin Au layer was grown using electron beam evaporation.
In order to break the symmetry in the Au-covered NW sidewalls, the Au flux was tilted for 14 ° angle with respect to the nanowire axis. This resulted in around 17 nm of Au on two out of six NW sidewalls. Samples of thinner NW diameters, fabricated by the same technique, were previously investigated by reflection spectroscopy [ 11 ].
Furthermore, we use finite-difference-time-domain simulations to predict the optical behavior of NWs in the same wavelength range. We employ a commercial 3D solver of Maxwell’ s equations by Lumerical [ 12 ]. Complex refractive indices of GaAs and AlGaAs are taken from Ref. [ 13 ]. We use a broadband total-field-scattered-field source to excite a single NW in the air or on the substrate, under normal incidence. This method separates the power absorbed by the medium from the scattered one, which numerically mimics the photo-acoustic measurements, as shown later in the manuscript. The absorption cross-section is defined by a box of monitors surrounding the NW( 200 nm 200 nm 6 lm); the absorption efficiency g abs is then calculated by dividing the absorption cross-section with the geometric cross-section.
3 Results and discussion
A single GaAs NW with diameters in( 60 – 200) nm range supports leaky resonant absorption modes below the GaAs bandgap [ 14 ]. Figure 1b shows such mode in a single NW of the Sample C, peaking at 780 nm; the inset plots 3D absorption density at that wavelength. When the NW is placed on the substrate, the same absorption cross-section box contains a small volume of the absorbing substrate material; nonetheless, the presence of the substrate leads to the background absorption, slightly modifying the spectral shape of the resonance. Sample C-Au is modelled by taking into account all Au material influences: the Au layer on the two sidewalls, the Au layer present on the substrate, including the NW shadowing effect, and a plasmonic“ cap” on the top of the NW. This strongly reshapes the resonances of the semiconductor NW. In Figure 1c, the absorption is redshifted for the NW with two sidewalls in the air; when such NW is placed on the substrate with Au, this substrate redshifts and strongly increases the absorption. Finally, the hybrid semiconductor-plasmonic nanowire has a prominent absorption for wavelengths greater than the GaAs bandgap, i. e. in the( 870 – 1000) nm range. Dashed lines in Figures 1b – 1c present scattering cross-section spectra of a single NW, without and with Au; at these NW dimensions, they differ in the spectral position of the resonances [ 15 ]. Hence, trying to characterize the resonant absorption by performing extinction measurements would be misleading, especially in terms of ascribing the observed signal peaks to the electromagnetic field enhancement at the same wavelength, e. g. for improved sensing or heat transport at the nanoscale [ 16 ].
We next apply PAS to detect the resonant absorption modes in the two samples; as a reference, we use the substrate of the Sample C( Si, without Au). We excite them at normal incidence, with linear polarization, in the( 680 – 1000) nm range, and tune the chopper frequency f. Figure 2a