The inherent ability of plasmonic bowtie nanoapertures (NAs) to localize the electromagnetic field at a subwavelength scale was exploited to engineer the H removal process in dilute nitrides at the nanometer level. Dilute nitride semiconductor alloys (e.g. GaAsN with a small percentage of nitrogen) are characterized by peculiar optoelectronic properties and, most importantly, by an even more peculiar response to hydrogen incorporation. In this class of materials, it is indeed possible to tune post-growth the alloy bandgap energy by a controlled incorporation of hydrogen atoms. The formation of N-H complexes neutralizes all the effects N has on the host matrix, among which is the strong narrowing of bandgap energy. In the present work, bowtie NAs resonant to the N-H complex dissociation energy were numerically modeled by finite element method simulations, realized by a lithographic approach, and characterized by scanning probe microscopy and resonant scattering spectroscopies. The conditions to get the maximum field enhancement at a specific position below the metal/semiconductor interface, namely at the dilute nitride quantum well position, were identified, demonstrating the ability to achieve a plasmon-assisted spatially selective hydrogen removal in a GaAsN/GaAs quantum well sample. Hydrogen removal through bowtie NAs turns out to be way more efficient (approximately two orders of magnitude) than through the plain surface, thus indicating that bandgap engineering through plasmonic nanostructures can be optimized for future efficient realization of site-controlled single-photon emitters and for their deterministic integration in plasmonic devices.
The study of plasmonic nanostructures continues to attract increasing interest, due to their potential application both in fundamental research and in device fabrication , , . Among plasmonic nanostructures, nanoapertures (NAs) show the remarkable property to transmit and enhance the electromagnetic field over a subwavelength scale (i.e. well below the diffraction limit) , , . This ability arises from collective excitations of the conduction electrons referred to as surface plasmon modes , , , . Having the ability to produce highly localized light spots, NAs can efficiently link propagating and localized optical fields, enhancing the interaction of light with matter at subwavelength resolution , , . NAs could thus find potential applications in a number of fields where high optical resolution and field enhancement are critical, such as nanolithography , , , data storage , optical tweezers , or biochemical sensing , , . In particular, given their geometry that merges both a high field enhancement due to plasmon coupling (feed-gap antenna effect) and a strong field localization due to the lighting rod effect , bowtie-shaped NAs are able to generate a very large field enhancement at their apexes , up to 15,000 times the illumination field .
In this work, we exploit the large field transmission and localization properties of bowtie NAs to directly write the bandgap energy of dilute nitrides, i.e. III-V semiconductor alloys containing small amounts of substitutional N atoms , , . In previous works on the GaAsN alloy, to cite the most investigated dilute nitride material, we have shown that it is possible to tune post-growth the alloy properties – e.g. the bandgap energy  or the effective mass ,  and gyromagnetic factor  of conduction electrons – by a controlled incorporation of hydrogen atoms. This possibility stems from the formation of N-H complexes within the material lattice, which neutralize all the effects N has on the host matrix , . More recently, we also demonstrated the possibility to modulate spatially, at a nanometer scale, the material properties, particularly the bandgap energy, by making use of lithographic  or laser-assisted ,  approaches to control the incorporation or removal of hydrogen atoms, respectively. This ability has been successfully employed to create site-controlled quantum emitters and to deterministically integrate them in photonic crystal nanocavities , . Here, by making use of plasmonic NAs, we engineer the plasmon-induced hot spot to control the N-H complex dissociation at the nanometer scale, opening a novel route to the hydrogen-induced in-plane bandgap engineering of dilute nitride semiconductors.
In particular, we perform finite element method (FEM) calculations (Section 2.1) to optimize the bowtie NA geometry in order to obtain a strong field localization (<100 nm) and enhancement (a factor of >10) 33 nm below the metal/semiconductor interface – namely, at the GaAsN quantum well (QW) center – for an incoming radiation wavelength of 700 nm. At that wavelength, indeed, the N-H complex dissociation by light has been reported to have the maximum efficiency . Arrays of isolated bowtie NAs in Al thin films are then fabricated and characterized (Section 2.2) by scanning probe microscopies [i.e. scanning electron microscopy (SEM) and atomic force microscopy (AFM)] as a function of their geometrical properties. An electron beam lithography (EBL) fabrication approach is used, rather than the most widely used focused ion beam (FIB) milling approach, to keep the sample damage negligible during the fabrication of NAs and to allow a scalable processing of many devices in parallel over a relatively large area. The realized NAs are also experimentally characterized (Section 2.3) in terms of their optical response to the incoming radiation by cross-polarized resonant scattering spectroscopy, testing their resonant response to the target 700-nm radiation. Finally, the ability of the NAs to remove H in a spatially selective way from a fully hydrogenated GaAsN/GaAs QW sample is demonstrated (Section 3). The hydrogen removal through the NAs is way more efficient (up to ~100 times) than on the plain sample surface, thus demonstrating the potential of plasmon-induced in-plane bandgap engineering for the future efficient realization of site-controlled single-photon emitters in dilute nitrides and for their deterministic integration in plasmonic devices.
2 Bowtie NA optimization
Plasmonic NAs are characterized by a strong electromagnetic field localization and by a high transmission, which make them suitable for a variety of subwavelength applications. However, if not properly designed, the waveguide cut-off effect  can drastically reduce the field transmission through the plasmonic NA. For this reason, before fabrication, we have performed a series of FEM simulations, in order to define a range of geometric parameters suitable for the target application: namely, the site-controlled removal of H from a dilute nitride QW buried in a GaAs sample. In particular, the target bowtie NAs should be able to localize the field well below (33 nm) the metal/substrate interface, namely at the center of the QW, and they should sustain a plasmonic resonance at the near-infrared wavelength (λ~700 nm) where the N-H complex dissociation and H removal from dilute nitrides are more efficient . These are two stringent conditions specific to our target application and scarcely investigated in the literature.
As the performance of a bowtie NA is strongly dependent on its geometry, in order to obtain reliable results when simulating the NA properties, it is necessary to adopt a realistic geometry, also taking into account the limits of the fabrication technique being used. Figure 1A shows the geometry of the bowtie apertures studied in this work. It consists of a square waveguide defined by a length L, loaded with two conducting triangular ridges whose apexes meet at the center of the aperture (initially no gap, g=0). A square gap of size g (≥10 nm, to meet the limit of the fabrication technique) is then opened at the center of the aperture. All the corners of the obtained geometry are rounded, with a minimum curvature radius of r=20 nm, thus mimicking the finite resolution of the fabrication technique in the xy plane while also avoiding field singularities in the simulations. The obtained bowtie geometry is used to define an aperture in a metal film of constant thickness (t) deposited on top of a GaAs substrate. The FEM simulation is then conducted by exciting such a geometry with a monochromatic plane wave of wavelength λ, linearly polarized with the electric field at a fixed angle (α) with respect to the y-axis, originating from the air side and propagating at normal incidence with respect to the substrate. The simulations were performed for a range of wavelengths between 300 and 1100 nm. The refractive indices of the substrate (GaAs) and of the metal layer used for the NA were wavelength dependent and were interpolated based on empirical data , , while a wavelength-independent refractive index of 1.0 was used for the environment (air). The computational domain box had a size of 1500 nm with a perfectly matched layer of 150 nm around it . A finer mesh (5–7 nm) was used around the NA, while a coarser mesh (50–80 nm) was employed in the outer regions. A perfect electric conductor was used in the y-z and x-z edge planes. The simulations were performed by calculating first a background scattered field (E0), defined as the solution of the interaction between the plane wave and an infinite metallic slab without any aperture (Floquet periodic boundary conditions were used in this case). Such an E0 field was then used as the excitation field for a second calculation run, where a single bowtie NA was opened in the metallic slab. The total electric field (Etot) obtained in such a way can be seen as the superposition of the background field produced by the metallic slab and the scattered field (ENA) produced by the NA: Etot=E0+ENA. Finally, the field intensity enhancement (I) produced by the single NA was obtained for each point of the simulated three-dimensional (3D) space as the ratio between the square norm of the total field and the square norm of the background field: I=||Etot||2/||E0||2. Such a simulation strategy yields accurate results even in the case ENA≪E0 and does not require a strong calculation power, as it can be performed on a desktop (32 Gb RAM) computer; however, it is a bit more time consuming, as any simulation has to be performed twice (with and without the NA) for each geometry and incident radiation condition.
As widely reported in the literature , , , the spectral response of metallic bowtie NAs is dominated by a surface plasmon resonance at visible/far-infrared wavelengths and by a Fabry-Pérot-like resonance at visible/ultraviolet wavelengths. By varying the NA geometry, mainly the film thickness, the relative intensity of the two resonances can be varied and one resonance can be enhanced with respect to the other. Figure 1 summarizes the results of the simulation studies done by varying all the geometrical parameters of the NA (i.e. outline dimension L, gap size g, metal thickness t, and the type of metal employed). The final goal of this study was to maximize the field intensity and localization (i.e. hot-spot lateral size) at the QW position (33 nm below the metal/substrate interface) for an incoming radiation of 700-nm wavelength. As a general trend, and in agreement with the literature , , , , , we found that an efficient plasmon resonance excitation can be achieved only for incoming light linearly polarized orthogonal to the bowtie main axis (i.e. along the y-axis), while surface plasmons are not excited for polarization along the bowtie main axis (i.e. x-axis). Such a polarization dependence is retained by the radiation transmitted through the NA toward the substrate (see Figure 1B). The field enhancement and resonant wavelength values are then analyzed adopting the convenient polarization for the incoming light (i.e. along the y-axis). We found that the field enhancement increases for decreasing the gap size from 100 to 10 nm, exhibiting a rather flat maximum around g~20 nm (Figure 1D), and that the intensity increases and the resonance frequency redshifts for increasing the outline dimensions from 200 to 600 nm, saturating for L≥400 nm (Figure 1C). Finally, a strong dependence of the plasmonic peak, both in intensity and resonant wavelength, is found by varying the metal thickness (see Figure 1E). Fixing all the geometrical parameters to the values that produce the strongest resonance peak for an incoming radiation of 700-nm wavelength, we investigated the different metals usually employed in plasmonics – such as gold, silver, and aluminum – looking for the minimum hot-spot size that can be produced at the QW level (z=−33 nm). Although Au and Ag are able to produce a stronger field intensity enhancement (about three times that produced by Al), Al guarantees the most localized hot spot among the three metals, as well as the most symmetric one (see Figure 1F).
The performed simulation study allows us to identify the parameters a bowtie NA should have to produce an efficient H removal from a dilute nitride QW sample, which turn out to be L=400–600 nm, g=20–50 nm, t=100–120 nm in an Al film. This defines relatively lax conditions on the bowtie geometry, such that even relatively large variations in the geometric properties of the fabricated NAs (up to 20% around the optimal value of L and t, and up to 40% for g) are not expected to sizably worsen (<20%) the hot-spot intensity generated by the NA. Moreover, the use of Al (instead of Au or Ag) as the metal employed for making the NAs contributes to keeping the process cheap, which, paired to an inherently scalable fabrication technique (see below), greatly enhances the potential for applications of the results presented here.
The possibility to realize plasmonic nanostructures with gap size <10 nm has been demonstrated for different fabrication approaches: angled deposition , self-assembly of nanoparticles , chromium expansion with a second lithography , and template stripping from a silicon substrate , . Nevertheless, FIB milling is still the most used technique to achieve narrow-gap plasmonic aperture devices , having shown the capability to produce – with an He+-ion lithography configuration – apertures with gap size down to 8nm in coaxial shapes  and 5nm in dimers , , . Although FIB-based fabrication techniques allow ultimate gap resolution, they are often very challenging, not scalable, and not very versatile. Moreover, they are serial in nature, and thus unsuited for the fabrication of large arrays of devices. On the contrary, EBL – possibly combined with a single lift-off step to produce the final structure – is a much more flexible technique, naturally suited to be scalable and capable of producing high-resolution features at a much higher production rate. In this study, we adopted one such EBL-based fabrication technique by optimizing a negative-tone lift-off process with a hydrogen silsesquioxane/poly(methylmethacrylate) (HSQ/PMMA) bilayer resist for the realization of plasmonic NAs. A lithographic approach followed by a modified lift-off process has recently shown the ability to fabricate 10-nm-gap bowtie NAs . However, in that case, a single-layer HSQ resist was employed paired with a lift-off performed in hydrofluoric acid, which is not suitable for all substrates nor is it compatible with Al films. Here, to avoid hydrofluoric acid as the remover, an HSQ/PMMA bilayer resist was employed, thus allowing to remove the resist in an organic solvent (i.e. acetone). The negative-tone lift-off process with the HSQ/PMMA bilayer resists has already been proposed in the literature , with a reported sub-10-nm resolution ; however, to the best of our knowledge, it has never been applied to the fabrication of plasmonic nanostructures.
The fabrication procedure used is outlined in Figure 2. Our devices are fabricated on epitaxial GaAs samples (with and without the presence of a buried GaAsN QW; see Figure 2A for a sketch of the whole processing chain). First, a positive-tone PMMA electron resist is spin-coated on the sample, forming a 300-nm-thick resist film on top of which a second layer of negative-tone HSQ electron resist is deposited with a thickness of 70 nm. An EBL system (VISTEC EBPG 5-HR working at 100 kV) is used to pattern bowtie shapes with an e-beam dosage ranging from 1300 to 1900 μC/cm2, depending on the feature size. After the exposure, HSQ is developed at 20°C in an aqueous solution of tetramethylammonium hydroxide at 2.4% for 60 s, and then backed on a hot plate at 130°C for 5 min to increase the chemical and physical stability of the HSQ to the subsequent processing steps (Figure 2B). The bowtie shapes are then transferred to the PMMA layer by reactive ion etching in an O2:Ar (3:1) gas mixture plasma, thus leaving behind tall (~370 nm) bowtie posts (see Figure 2C). The presence of Ar+ ions in the plasma helps to reach near perpendicular sidewalls in the HSQ layer and a pronounced undercut in the PMMA layer, both conditions extremely important to prevent sidewall coating during the metal deposition (Figure 2D) and, thereby, to allow for a high-quality lift-off even with a rather small resist-to-metal ratio (3:1 in our case). Any possible oxidation of the sample surface and/or PMMA residuals is removed just before the metal deposition by a further mild reactive ion etching step in CHF3 plasma. Next, a 120-nm-thick layer of aluminum is evaporated onto the sample, and the final lift-off is performed in hot acetone (56°C) in a sonication bath cleaner for 2 min.
In the fabrication protocol, each step has been carefully optimized to control the final feature sizes and metal roughness; in particular, the etching steps are those most critical in this respect. As discussed in the next section, the metal roughness strongly affects the plasmon resonance in a way hard to quantify by simulations, so a fabrication strategy able to control such a parameter is instrumental to get efficient, reproducible plasmonic structures. The developed fabrication protocol allows for the realization of large arrays of (thousands of) identical bowtie NAs with outline dimension (L) down to 200 nm, gap size (g) down to ~30 nm, in an Al film (t=120 nm) having a root mean squared (Rms) surface roughness as low as 1.1 nm (see Figure 2E).
2.3 Optical characterization
The optical properties of the fabricated bowtie NAs, in particular their ability to enhance the intensity of an incoming electromagnetic radiation at the plasmon resonance wavelength, have been tested by resonant scattering spectroscopy performed in a cross-polarization configuration , . In such a configuration, a laser light linearly polarized at an angle α with respect to the bowtie y-axis (see inset in Figure 3A) is sent to the sample through a microscope objective. The light scattered from the sample is then collected by the same objective, in a cross-polarization scheme (i.e. selecting only the light linearly polarized at α+90°). This technique efficiently removes the reflected light coming from the sample surface, thus enhancing the sensitivity to the scattered light. A confocal Raman microscope equipped with a 100× objective (with a numerical aperture of 0.9) was employed to reach a laser spot size of ~0.6 μm at the sample surface, thus allowing us to investigate the response of a single NA (the NA-NA spacing was ≥1 μm in the fabricated arrays; see Figure 3A,E). The incident light was provided by a super-continuum laser source with wavelengths ranging from ~480 to ~2400 nm, then selected and linearly polarized in the range 480–1000 nm by a combination of a 1000-nm short-pass filter, a 435-nm long-pass filter, and a linear polarizer. Meanwhile, the collected light was polarization-rotated by a half-wave plate and filtered by a polarizer set (i) at 90° with respect to the polarization of the incident light and (ii) aligned with the direction of the maximum response of the detection system. Finally, the signal was analyzed by a spectrometer equipped with a 300 groove/mm grating and an Si charge coupled device detector. The spectrum of the scattered light intensity produced by the NA is obtained by normalizing the scattered light measured from the NA to that measured on an unpatterned area of the same metal film, which automatically rids us of the wavelength dependence of the incident radiation intensity and of the response of the optical elements (see Figure 3A–C). A Gaussian profile centered at the target wavelength of 700 nm is found for all the investigated NAs (i.e. Al bowties with L=300–600 nm, g=30–70 nm, and t=120 nm; see also Figure 4D), thus validating the simulation studies and the fabrication protocol.
The ability of EBL to easily create arrays of identical objects on a large area is evident by the optical and surface probe microscopy (SEM and AFM) maps acquired on the realized samples (see Figure 3A,D,E).
A 2D map of the scattered light intensity integrated around 700 nm is reported in Figure 3D. Although the spatial resolution of the map (limited by the laser spot size of ~0.6 μm) is comparable with the size of the investigated NA, and so a map of the plasmonic mode itself is not possible, it is possible to envisage two darker regions around a bright spot centered on the bowtie aperture. We tentatively attribute this feature to the plasmon-induced localization of the scattered field, which is collected from the bowtie arms and concentrated into the bowtie gap. Maps at higher spatial resolution, as those obtainable by near-field spectroscopy (planned for the near future), could validate or reject this hypothesis.
One of the advantages of the chosen fabrication strategy was to have a direct, easy control on the metal surface roughness. It was possible, indeed, to fabricate identical bowtie NAs with metal surface roughness varying over more than an order of magnitude (i.e. Rms from 1.1 to 18.7 nm) by changing the etch recipe, thus directly testing the metal roughness effect on the plasmon resonance by measuring the optical response of the different NAs. Metal surface roughness is known to play a crucial role in plasmonics, as it is one of the main parameters affecting the plasmon excitation time dephasing and damping mechanics . Its effects are usually detrimental for the properties of the plasmonic nanostructures, particularly in the far-field configuration, even if they can have advantages on the near-field properties, as those reported in surface-enhanced Raman spectroscopy experiments . Nevertheless, surface roughness effects are difficult to model and can often be addressed only through direct measurements. In Figure 4A–C, we report an optical and structural investigation of identical bowtie NAs in a 120-nm-thick Al film (L=600 nm, g=35 nm) with very different metal surface roughness (as measured by AFM). By increasing the surface roughness, the scattered intensity enhancement peak shifts to higher wavelengths, broadens, and loses in intensity, all but vanishing for the sample with the highest roughness (Rms=18.7 nm). Such a behavior agrees with what is reported in the literature for plasmonic gold nanoshells, as a result of the roughness-induced plasmon damping and “hot-spot” formation on the surface corrugations , . For our purposes, being interested in getting the strongest far-field enhancement from the bowtie NAs, in the fabrication process optimization we gave priority to a process leading to smooth metal surfaces, somewhat sacrificing the achievement of ultimately high gap resolution. The actual feature sizes of the fabricated NAs have been determined by surface probe microscopes (SEM and AFM); gap sizes down to 30 nm have been obtained in samples showing very small surface roughness (Rms~1 nm). On such samples characterized by a smooth surface, we investigated the dependence of the scattering intensity enhancement as a function of the bowtie outline dimension and gap. As regards the scattered field enhancement, we found that it increases both by decreasing the gap size (Figure 4E) and by increasing the NA outline dimension (Figure 4F). These behaviors agree well with what is expected on the basis of the performed simulations. Moreover, we investigated the optical response as a function of the polarization angle of the incoming radiation for an NA with L=400 nm and g=30 nm (see Figure 4G). We found an almost vanishing value of the field enhancement for α=0° (polarization of the incoming light along the x-axis), a rather small value (~3) for α=90° , and a marked maximum (~14) for α=45°. Also, this experimental behavior perfectly agrees with the expectations (see Figure 1B) once the cross-polarization configuration employed in the measurements is duly considered. Indeed, when the polarization of the incoming light is along the x-axis, the surface plasmon excitation is negligible. On the contrary, for polarization along the y-axis, the plasmon is maximally excited. However, in this configuration, the scattered light (also polarized along the y-axis) is cut by the collection polarization (fixed along the x-axis)2; thus, the maximum intensity is found for α=45°.
3 Plasmon-assisted H removal
As mentioned in Section 1, the spatially controlled, post-growth incorporation or removal of hydrogen in dilute nitride semiconductors can be applied to the fabrication of site-controlled quantum dots (QDs) , . In particular, as regards H removal, a near-field optical illumination was recently employed to overcome the wavelength diffraction limit, thus reducing the laser-irradiated area down to a size of <100 nm and inducing quantum confinement effects . The ability of plasmonic NAs to localize the electromagnetic field at a subwavelength level may make it possible to obtain similar results simply by illuminating the sample through a standard microscope objective, thus avoiding the need for a complex near-field illumination setup. To this end, we first test and demonstrate the possibility to obtain a controlled H removal, and the ensuing in-plane modulation of the bandgap of dilute nitrides, through plasmonic NAs. Figure 5A shows a sketch of the experiment we performed. We studied a GaAs sample containing a GaAsN QW ([N]=1.1%) buried 30 nm below the surface of the sample, which was hydrogenated (H-dose equals to 1.5×1017 ions/cm2, hydrogenation temperature equals to 190°C) in such a way to get a complete N passivation. The achievement of this condition is confirmed by the photoluminescence (PL) spectrum of the sample after hydrogenation, in which only the GaAs emission can be detected (see black line in Figure 5B). After H irradiation, an array of identical Al bowtie NAs was fabricated on the sample following the procedure described in Section 2.2 (t=120 nm, L=600 nm, g=100 nm, array pitch of 5 μm). Then, laser exposures via a 50× microscope objective (numerical aperture of 0.5, laser spot size ~1.7 μm) were performed on single bowtie NAs by shining a 100-nm band laser centered at λexp=700 nm (as obtained from a super-continuum laser with the aid of optical filters). The laser treatments were performed for a fixed exposure time of 1 s at increasing power densities. The H removal efficiency was tested by measuring the room temperature PL spectrum from the single NAs after each laser treatment. The PL was excited with a 532-nm laser operated at a power (<1 mW) sufficient to generate a sizable PL signal without producing any H removal (in this, we were aided by the fact that in GaAsN, the H-removal efficiency at 532 nm is about four times lower than that at 700 nm ).
For laser treatments performed at a power of >1 mW, it was possible to observe a sizable hydrogen removal, as indicated by the appearance of the GaAsN peak in the PL spectrum, with a relative intensity increasing as a function of the laser power used in the treatment (see Figure 5B, red and blue lines). It is worth mentioning that the laser power necessary for a full dehydrogenation of the GaAsN QW through the bowtie NA (i.e. 6 mW for 1 s, for a GaAsN emission energy matching the one of the hydrogen-free sample) is seemingly unable to remove any H from the unpatterned area of the same sample (see inset in Figure 5B). As extensively detailed in previous studies on similar samples, indeed, a laser treatment at a power higher than ~30 mW for 20 s is necessary in order to produce a full de-hydrogenation on unpatterned samples . The reported results indicate, therefore, that the H removal process through the NA is much more efficient (~100 times, when the amount of energy delivered to the sample by the laser is considered) than on the plain surface, and that the plasmon-induced hot spot created by the bowtie NA in correspondence of the GaAsN QW level is able to produce an in-plane bandgap engineering modulation.
Moreover, the simulations indicate that, by reducing the bowtie NA gap to <50 nm, the hot spot created at the GaAsN QW level is localized on a lateral size <100 nm (see Figure 1F, for g=20 nm). As a consequence, we expect to be able to exploit such a plasmon-enhanced H removal to create nanostructures (such as QDs) by illuminating bowtie NAs through a standard microscope objective, in a similar (but much simpler) way to what is done by near-field optical illumination . Low-temperature measurements on NAs treated under specific irradiation conditions indeed show the presence of narrow lines (with full-width-half-maximum down to ~1 nm) peaked at energies higher than that of the pristine GaAsN QW (λ~980 nm at T=5 K), which may indicate the formation of a GaAsN QD within the treated area (see Figure 5C–E). In particular, it is worth noticing that delivering a same amount of energy (6 mJ) through identical NAs (L=400 nm, g=100 nm) but employing different power and treatment times (e.g. 6 mW for 1 s, or 2 mW for 3 s; see blue and red lines in Figure 5C, respectively), the formed nanostructures can exhibit very different emission energies (up to ~100 meV apart each other). This seems to suggest that the actual size of the GaAsN nanostructure created by NAs is not only defined by the NA geometry, but can also be tuned by the laser treatment parameters, similar to what is found in the case of near-field illumination . A detailed investigation on the QD formation by plasmon-assisted H removal is under progress and will be reported elsewhere.
4 Outlook and conclusion
In summary, bowtie-shaped plasmonic NAs have been investigated and fully optimized, opening a new pathway to the efficient in-plane bandgap engineering of hydrogenated dilute nitrides by spatially selective H removal. Bowtie NAs resonant to the N-H complex dissociation energy (λ~700 nm) have been numerically modeled by FEM simulations, realized by a lithographic approach avoiding the use of focused ion milling, and characterized by scanning probe microscopy and resonant scattering spectroscopies. The conditions to get the maximum field enhancement at a specific position below the metal/semiconductor interface (namely, at the center of the dilute nitride QW) have been identified, and we demonstrated the achievement of plasmon-assisted spatially selective hydrogen removal in a fully hydrogenated GaAsN/GaAs QW sample. Hydrogen removal through bowtie NAs turns out to be two orders of magnitude more efficient than through the plain surface of the sample. Ordered arrays of identical Al bowtie NAs have been fabricated by e-beam lithography, followed by a single lift-off step. In spite of its simplicity, this method can be employed to shape a wide range of complex nanostructure designs with good resolution and reproducibility, without involving aggressive etching processes or damaging focused ion milling steps. In addition, the roughness of the metal film, which has been shown to strongly affect the plasmon resonance condition, can be controlled by varying the process parameters and can be kept at a very low level (as low as Rms~1 nm).
The reported findings demonstrate that bandgap engineering through plasmonic nanostructures represents a versatile and powerful tool to control the optoelectronic properties of dilute nitrides in a spatially controlled way. In particular, the reported study represent a first, pivotal step toward the realization of deterministically coupled optically active semiconductor- plasmonic integrated components, of great interest for the practical implementation of quantum information and communication protocols . Indeed, the demonstrated ability to locally manipulate the bandgap energy of hydrogenated dilute nitrides via plasmonic NAs can lead to the realization of spatially controlled QDs, which can be directly coupled with the plasmonic nanostructures used to fabricate them. Ideally, it might even be possible to purposely design doubly resonant plasmonic structures , with one resonance centered at 700 nm – to have efficient H removal to create the QD – and the other resonance peaked at ~900 nm – to boost and control the emission properties of the fabricated nanostructures.
Award Identifier / Grant number: RBFR12RS1W
Funding source: European Union’s Horizon 2020 Research and Innovation Program
Award Identifier / Grant number: 641899 PROMI
Funding statement: This work was supported by the Italian Ministry for Education, University and Research within the Futuro in Ricerca (FIRB) program (project DeLIGHTeD, Prot. n. RBFR12RS1W, Funder Id: http://dx.doi.org/10.13039/501100003407) and by the European Union’s Horizon 2020 Research and Innovation Program under the Marie Skłodowska-Curie Grant Agreement No. 641899 PROMIS (Postgraduate Research on Dilute Metamorphic Nanostructures and Metamaterials in Semiconductor Photonics), Funder Id: http://dx.doi.org/10.13039/100010665. A.P. and M.F. acknowledge funding from the Regione Lazio programme “Progetti di Gruppi di ricerca” legge Regionale n. 13/2008 (SINFONIA project, prot. n. 85-2017-15200) via LazioInnova spa.
 Garcia-Vidal FJ, Moreno E, Porto JA, Martin-Moreno L. Transmission of light through a single rectangular hole. Phys Rev Lett 2005;95:103901.10.1103/PhysRevLett.95.103901Search in Google Scholar PubMed
 Thio T, Pellerin KM, Linke RA, Lezec HJ, Ebbesen TW. Enhanced light transmission through a single subwavelength aperture. Opt Lett 2011;26:1972–4.10.1364/OL.26.001972Search in Google Scholar PubMed
 Wen X, Datta A, Traverso LM, Srisungsitthisunti P, Xu X, Moon EE. Optical nanolithography with λ/15 resolution using bowtie aperture array. Appl Phys A 2014;117:307–11.10.1007/s00339-014-8265-ySearch in Google Scholar
 Wen X, Traverso LM, Pan L, Xu X, Moon EE. High throughput optical lithography by scanning a massive array of bowtie aperture antennas at near-field. Sci Rep 2015;5:16192.10.1038/srep16192Search in Google Scholar PubMed PubMed Central
 Peng C, Jin E, Clinton TW, Seigler MA. Cutoff wavelength of ridge waveguide near field transducer for disk data storage. Opt Express 2008;16:16043–51.10.1364/OE.16.016043Search in Google Scholar PubMed
 Berthelot J, Aćimović SS, Juan ML, Kreuzer MP, Renger J, Quidant R. Three-dimensional manipulation with scanning near-field optical nanotweezers. Nat Nanotechnol 2014;9:295–9.10.1038/nnano.2014.24Search in Google Scholar PubMed
 Fromm DP, Sundaramurthy A, Kinkhabwala A, Schuck PJ, Kino GS, Moerner WE. Exploring the chemical enhancement for surface-enhanced Raman scattering with Au bowtie nanoantennas. J Chem Phys 2006;124:061101.10.1063/1.2167649Search in Google Scholar PubMed PubMed Central
 Henini M. Dilute nitride semiconductors: physics and technology. Oxford, Elsevier, 2005.Search in Google Scholar
 Ayse E. Dilute III-V nitride semiconductors and material systems. Berlin, Germany, Springer, 2008.Search in Google Scholar
 Polimeni A, Baldassarri Hoeger von Hoegersthal G, Bissiri M, et al. Effect of hydrogen on the electronic properties of InxGa1-xAs1-yNy/GaAs quantum wells. Phys Rev B 2001;63:201304.10.1103/PhysRevB.63.201304Search in Google Scholar
 Masia F, Pettinari G, Polimeni A, et al. Interaction between conduction band edge and nitrogen states probed by carrier effective-mass measurements in GaAs1-xNx. Phys Rev B 2006;73:073201.10.1103/PhysRevB.73.073201Search in Google Scholar
 Pettinari G, Polimeni A, Masia F, et al. Electron mass in dilute nitrides and its anomalous dependence on hydrostatic pressure. Phys Rev Lett 2007;98:146402.10.1103/PhysRevLett.98.146402Search in Google Scholar PubMed
 Pettinari G, Masia F, Polimeni A, et al. Influence of nitrogen-cluster states on the gyromagnetic factor of electrons in GaAs1-xNx. Phys Rev B 2006;74:245202.10.1103/PhysRevB.74.245202Search in Google Scholar
 Trotta R, Polimeni A, Capizzi M. Hydrogen incorporation in III-N-V semiconductors: from macroscopic to nanometer control of the materials’ physical properties. Adv Funct Mater 2012;22:1782–801.10.1002/adfm.201102053Search in Google Scholar
 Pettinari G, Felici M, Trotta R, et al. Hydrogen effects in dilute III-N-V alloys: from defect engineering to nanostructuring. J Appl Phys 2014;115:012011.10.1063/1.4838056Search in Google Scholar
 Trotta R, Polimeni A, Martelli F, et al. Fabrication of site-controlled quantum dots by spatially selective incorporation of hydrogen in Ga(AsN)/GaAs heterostructures. Adv Mater 2011;23:2706–10.10.1002/adma.201004703Search in Google Scholar PubMed
 Balakrishnan N, Pettinari G, Makarovsky O, et al. Band-gap profiling by laser writing of hydrogen-containing III-N-Vs. Phys Rev B 2012;86:155307.10.1103/PhysRevB.86.155307Search in Google Scholar
 Biccari F, Boschetti A, Pettinari G, et al. Site-controlled single-photon emitters fabricated by near-field illumination. Adv Mater 2018;30:1705450.10.1002/adma.201705450Search in Google Scholar PubMed
 Pettinari G, Gerardino A, Businaro L, et al. A lithographic approach for quantum dot-photonic crystal nanocavity coupling in dilute nitrides. Microelectron Eng 2017;174:16–9.10.1016/j.mee.2016.12.003Search in Google Scholar
 Pettinari G, Felici M, Biccari F, Capizzi M, Polimeni A. Site-controlled quantum emitters in dilute nitrides and their integration in photonic crystal cavities. Photonics 2018;5:10.10.3390/photonics5020010Search in Google Scholar
 Guo H, Meyrath TP, Zentgraf T, et al. Optical resonances of bowtie slot antennas and their geometry and material dependence. Optics Express 2008;16:7756–66.10.1364/OE.16.007756Search in Google Scholar PubMed
 Park S, Hahn JW, Lee JY. Doubly resonant metallic nanostructure for high conversion efficiency of second harmonic generation. Optics Express 2012;20:4856–70.10.1364/OE.20.004856Search in Google Scholar PubMed
 Guo R, Kinzel EC, Li Y, Uppuluri SM, Raman A, Xu X. Three-dimensional mapping of optical near field of a nanoscale bowtie antenna. Optics Express 2010;18:4961–71.10.1364/OE.18.004961Search in Google Scholar PubMed
 Theiss J, Pavaskar P, Echternach PM, Muller RE, Cronin SB. Plasmonic nanoparticle arrays with nanometer separation for high-performance SERS substrates. Nano Lett 2010;10:2749–54.10.1021/nl904170gSearch in Google Scholar PubMed
 Zehtabi-Oskuie A, Zinck AA, Gelfand RM, Gordon R. Template stripped double nanohole in a gold film for nano-optical tweezers. Nanotechnology 2014;25:495301.10.1088/0957-4484/25/49/495301Search in Google Scholar PubMed
 Chen X, Ciraci C, Smith DR, Oh SH. Nanogap-enhanced infrared spectroscopy with template-stripped wafer-scale arrays of buried plasmonic cavities. Nano Lett 2015;15:107–13.10.1021/nl503126sSearch in Google Scholar PubMed
 Melli M, Polyakov A, Gargas D, et al. Reaching the theoretical resonance quality factor limit in coaxial plasmonic nanoresonators fabricated by helium ion lithography. Nano Lett 2013;13:2687.10.1021/nl400844aSearch in Google Scholar PubMed
 Scholder O, Jefimovs K, Shorubalko I, Hafner C, Sennhauser U, Bona GL. Helium focused ion beam fabricated plasmonic antennas with sub-5 nm gaps. Nanotechnology 2013;24:395301.10.1088/0957-4484/24/39/395301Search in Google Scholar PubMed
 Wang Y, Abb M, Boden SA, Aizpurua J, de Groot CH, Muskens OL. Ultrafast nonlinear control of progressively loaded, single plasmonic nanoantennas fabricated using helium ion milling. Nano Lett 2013;13:5647–53.10.1021/nl403316zSearch in Google Scholar PubMed
 Kollmann H, Piao X, Esmann M, et al. Toward plasmonics with nanometer precision: nonlinear optics of helium-ion milled gold nanoantennas. Nano Lett 2014;14:4778–84.10.1021/nl5019589Search in Google Scholar PubMed
 Huang I-C, Holzgrafe J, Jensen RA, Choy JT, Bawendi MG, Lončar M. 10nm gap bowtie plasmonic apertures fabricated by modified lift-off process. Appl Phys Lett 2016;109:133105.10.1063/1.4963689Search in Google Scholar
 Yang H, Jin A, Luo Q, Li J, Gua C, Cui Z. Electron beam lithography of HSQ/PMMA bilayer resists for negative tone lift-off process. Microelectron Eng 2008;85:814–7.10.1016/j.mee.2008.01.006Search in Google Scholar
 Rommel M, Weis J. Hydrogen silsesquioxane bilayer resists – combining high resolution electron beam lithography and gentle resist removal. J Vac Sci Technol B 2013;31:06F102; and references therein.10.1116/1.4822136Search in Google Scholar
 Englund D, Faraon A, Fushman I, Stoltz N, Petroff P, Vučković J. Controlling cavity reflectivity with a single quantum dot. Nature 2007;450:857–61.10.1038/nature06234Search in Google Scholar PubMed
 Galli M, Portalupi SL, Belotti M, Andreani LC, Faolain LO, Krauss TF. Light scattering and Fano resonances in high-Q photonic crystal nanocavities. Appl Phys Lett 2009;94:071101.10.1063/1.3080683Search in Google Scholar
 Wang H, Goodrich GP, Tam F, Oubre C, Nordlander P, Halas NJ. Controlled texturing modifies the surface topography and plasmonic properties of Au nanoshells. J Phys Chem B 2005;109:11083–7.10.1021/jp051466cSearch in Google Scholar PubMed
 Rodríguez-Fernández J, Funston AM, Perez-Juste J,Álvarez-Puebla RA, Liz-Marzán LM, Mulvaney P. The effect of surface roughness on the plasmonic response of individual sub-micron gold sphere. Phys Chem Chem Phys 2009;11:5909–14.10.1039/b905200nSearch in Google Scholar PubMed
 Lyamkina A, Schraml K, Regler A, et al. Monolithically integrated single quantum dots coupled to bowtie nanoantennas. Optics Express 2016;24:28936–44.10.1364/OE.24.028936Search in Google Scholar PubMed
 Chou Chau Y-F, Chou Chao C-T, Rao J-Y, et al. Tunable optical performances on a periodic array of plasmonic bowtie nanoantennas with hollow cavities. Nanoscale Res Lett 2016;11:411.10.1186/s11671-016-1636-xSearch in Google Scholar PubMed PubMed Central
© 2019 Giorgio Pettinari et al., published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 Public License.