1 Introduction

In recent years, group III-nitride nanowires (NWs) have gained much interest as a platform for photocatalysis because of the favorable energy position of their band edges with respect to many redox levels in liquid electrolytes and their high surface-to-volume ratio which leads to an improved, highly stable, and controllable photocatalytic activity [1,2,3,4,5,6]. Photocatalytic reactions are usually optimized for activation by sunlight. However, some standard systems for photocatalytic \(\hbox {CO}_{2}\) reduction suffer from low photocatalytic activity caused by light-induced deactivation processes, only use a narrow part of the sunlight spectrum, and are very slow compared to electronic processes in the photocatalyst [7]. These conditions can be best matched by pulsed irradiation using a light-emitting diode (LED) as illumination source [7]. Thanks to their high crystal quality and the possibility to grow on various substrates [8,9,10,11], group III-nitride NWs can be integrated directly on an LED substrate [12]. The optical waveguide character of the NWs given by their high aspect ratio and refractive index enables an efficient outcoupling of the light from the LED over the NW side facets and, thus, an efficient coupling to photocatalytic centers, e.g., molecules or nanoclusters, immobilized on the surface of the NWs or present in the vicinity of the NWs [12,13,14].

In a previous work (published as part I in J. Appl. Phys. 123, 203104), we have investigated by numerical simulations and experimental transmission measurements how the interaction of GaN NWs with incident light is influenced by the NW geometry, i.e., diameter, length and density, and the illumination wavelength [12]. We found that, for large NW periods, the transmission characteristics are dominated by the waveguide effects of the single NWs, whereas, for dense NW arrays, the coupling of the light into the NW arrays is additionally affected by the inter-wire coupling and the excitation of in-plane waveguide modes. For NW-based photocatalytic applications, an outcoupling of the guided light over the NW side facets is desired for an enhanced photocatalytic activity. In addition, the efficiency of photocatalytic reactions can be improved by molecular or nanoparticle co-catalysts attached to the surface of the NWs [1, 15, 16] or by protective coatings [6, 17, 18]. Earlier publications showed that the geometrical design of NW arrays [19] and their surroundings [20,21,22,23] can have a strong impact on the light absorption. In particular, an enhancement of the light absorption was observed for core–shell NWs [22] and NWs embedded in a transparent polymer [21, 23]. These studies focus on the maximization of the absorption of NW arrays related to photovoltaic applications. However, to our knowledge, there are fewer studies on the influence of dielectric shells and liquid environments on the light outcoupling behavior of NWs with regard to photocatalysis applications. In our previous publication [12], we showed, by numerical studies, that a dielectric shell around the GaN NWs has an impact on the waveguide properties of the NWs and, thus, on the transmission characteristics of the NW arrays [12]. In this work (which is part II), we experimentally investigate the influence of the thickness and the refractive index of a dielectric shell on the transmission of GaN NW arrays. Furthermore, the effect of liquid environments on the transmission characteristics of GaN NW arrays is studied. The experimental results are compared with numerical simulations.

2 Experimental

The GaN NWs investigated in this work were grown by plasma-assisted molecular beam epitaxy (MBE) in the selective area growth (SAG) mode directly on double-side polished c-plane sapphire substrates. Due to the transparency of these substrates, transmission measurements are possible. Structured titanium nitride masks are used to achieve SAG of GaN NWs. For this purpose, a \(10\,\hbox {nm}\) thin titanium film was thermally evaporated onto the substrate. Arrays of holes with different diameters as nucleation sites for the NW growth were defined by e-beam lithography and wet chemical etching. The GaN NWs were grown for 90 min under highly nitrogen-rich conditions at a nominal substrate temperature of \(1000\,^{\circ }\hbox {C}\) (real substrate surface temperature \(\approx 850\,^{\circ }\hbox {C}\)) yielding an NW length L of \(\approx 510\,\hbox {nm}\) and NW diameters D ranging from 90 to \(190\,\hbox {nm}\). A detailed description of the GaN NW SAG can be found in previous publications [12, 24]. In Fig. 1a, a scanning electron microscopy (SEM) image of as-grown SAG GaN NWs for an NW period \(p = 500\,\hbox {nm}\) and a diameter of \(D = 120\,\hbox {nm}\) is shown in tilted-view (\(45\,^{\circ }\)). The NWs are homogeneous in height and diameter with almost no parasitic nucleation on the mask surface. Prior to the transmission measurements, the metallization of the substrate backside, which is needed for the heat transfer during the MBE growth, and the titanium nitride mask are removed by wet chemical etching in a concentrated HF solution. To analyze the influence of a dielectric shell on the transmission characteristics of GaN NWs, \(\hbox {SiO}_{2}\) and ZnO shells with different shell thicknesses were deposited on the GaN NWs. The \(\hbox {SiO}_{2}\) shells were sputtered onto the GaN NWs using a reactive sputter coater, whereas the ZnO shells were grown by plasma-assisted MBE in a separate system. The ZnO shell growth was realized at a substrate temperature of \(380\,^{\circ }\hbox {C}\), a Zn beam equivalent pressure of \(\hbox {BEP}_{Zn} = 2.5 \times 10^{-7}\,\hbox {mbar}\), an RF power of \(400\,\hbox {W}\), and an oxygen flow rate of 0.6 sccm. Different \(\hbox {SiO}_{2}\) and ZnO shell thicknesses were achieved by varying the sputter and growth time, respectively. Exemplary tilted-view SEM images of \(\hbox {GaN}-\hbox {SiO}_{2}\) and GaN–ZnO core–shell NWs with a shell thickness \(d_{\text {shell}}\) of \(36 \pm 1\,\hbox {nm}\) are illustrated in Fig. 1b, c, respectively. The \(\hbox {SiO}_{2}\) shell is homogeneous and smooth along the NW sidewalls, whereas the ZnO shell shows an inhomogeneous shell thickness. The NW top facets and the substrate surface are also covered independent of the deposition process. Note that all measurements were performed on the same sample. To this end, the \(\hbox {SiO}_{2}\) shells were etched in an ammonium-fluoride 95/5 solution, whereas a concentrated hydrochloric acid solution was used for the ZnO shell etching. After each etching step, the NW morphology was investigated by SEM (not shown). A homogeneous shape was observed for all NWs, and the diameter and length of the NWs were unchanged after each etching step.

Transmission measurements were performed by means of UV/VIS spectroscopy using a Perkin Elmer Lambda 900 spectrophotometer. A detailed description of the setup used can be found elsewhere [12]. To investigate the influence of liquid environments on the transmission of GaN NW arrays, the sample was mounted on an aperture slit and placed within a cuvette that is filled with a liquid, i.e., water or toluol. A schematic illustration of the measurement configuration is shown in Fig. 1d.

Fig. 1
figure 1

Tilted-view (\(45\,^{\circ }\)) SEM images of SAG GaN NW arrays with a fixed period of \(p = 500\,\hbox {nm}\) and a diameter of \(D = 120\,\hbox {nm}\) for a bare GaN NWs, GaN NWs with b a \(37\,\hbox {nm}\)-thick \(\hbox {SiO}_{2}\) shell, and c a \(36\,\hbox {nm}\)-thick ZnO shell. d Schematic illustration of the measurement configuration used for the transmission measurements in liquid environment

Full size image

The simulated data presented in this study are obtained by numerically solving Maxwell‘s equations using the lumerical FDTD solutions package [25]. Three-dimensional finite-difference time-domain simulations were performed for the transmission calculations. We considered a hexagonally ordered array of GaN NWs on sapphire for the simulations. A plane wave is injected along the NW axis from the substrate and the light transmitted through the NW array is detected. A detailed description of the simulation setup can be found in a previous publication [12]. The lumerical MODE solutions package was used to calculate the dispersion relations shown in this work [26].

3 Results and discussion

In Fig. 2a, the measured optical transmission is shown as a function of the illumination wavelength and the NW diameter of a GaN NW array with a fixed \(p = 500\,\)nm and \(L \approx 510\,\)nm. The color scale denotes the value of the light transmission. Blue and red color corresponds to low and highlight transmission, respectively. Two branches of lowlight transmission can be observed. Transmission profiles for three selected NW diameters are plotted in Fig. 2b, highlighting the occurring transmission minima. For small NW diameters, only a single transmission minimum (marked with 1 in Fig. 2b) can be observed, whereas, for larger NW diameters, an enhanced number of transmission minima appear (additional minimum marked with 2 in Fig. 2b). These minima can be attributed to the coupling of light to specific waveguide modes which are supported within the NW and longitudinal Fabry–P\(\acute{\text {e}}\)rot modes which are formed when the guided mode is reflected at the NW top facet due to the refractive index mismatch between the NW and the surrounding medium and interferes with itself [27,28,29,30]. The coupling efficiency of these waveguide modes is determined by the confinement of the corresponding mode, which, in turn, is defined by the effective refractive index \(n_{\text {eff}}\) depending on the wavelength and the NW diameter [12, 31, 32]. The confinement of the mode increases with increasing NW diameter for a fixed wavelength and decreases for increasing wavelength for a fixed NW diameter [31]. The inset of Fig. 2b shows the wavelength where the value of the light transmission T equals 60% (indicated by black arrows in Fig. 2b), i.e., the transition between the region of high and lowlight transmission in the transmission map (Fig. 2a) as a function of the NW diameter. A similar value of the light transmission implies a similar confinement of the mode, i.e., effective refractive index \(n_{\text {eff}}\) of the mode. With increasing NW diameter, an increase of the wavelength can be observed which can be explained by an enhanced confinement of the mode for an increasing NW diameter and a fixed wavelength.

Fig. 2
figure 2

a Transmission as a function of the illumination wavelength and the NW diameter and b selected transmission spectra for different NW diameters of bare GaN NWs. Transmission minima arising from the coupling of light into the same waveguide mode are marked with numbers. c Transmission as a function of the illumination wavelength and the NW diameter and d selected transmission spectra for different NW diameters of \(\hbox {GaN}-\hbox {SiO}_{2}\) core–shell NWs with \(d_{\text {shell}} = 60\,\hbox {nm}\). The insets in b and d show the wavelength as a function of the NW diameter at the positions where \(T = 60\%\), indicated by black arrows in the corresponding transmission spectra

Full size image

To investigate the influence of a dielectric shell on the transmission characteristics of a GaN NW array, a \(60\,\hbox {nm}\)-thick \(\hbox {SiO}_{2}\) shell was sputtered onto the GaN NWs. In Fig. 2c, the measured transmission is again illustrated as a function of the NW diameter and the illumination wavelength. Transmission profiles for the same NW diameters as shown for the bare GaN NWs (Fig. 2b) are plotted in Fig. 2d. The NW diameters quoted are the GaN core diameters. The core–shell NWs show similar transmission characteristics as the bare GaN NWs. However, the observed minima in the transmission data are shifted to larger wavelengths. As for the bare GaN NWs, the wavelength where the value of the light transmission equals 60% (indicated by black arrows in Fig. 2d) is plotted as a function of the NW diameter in the inset of Fig. 2d. The NW diameter-dependent wavelength shows a similar behavior as for bare GaN NWs, but it is shifted to larger wavelengths. This indicates that the waveguide modes and Fabry–P\(\acute{\text {e}}\)rot modes are more strongly confined within the core–shell NWs compared to the bare GaN NWs with identical NW diameter. These observations are in agreement with our previous theoretical studies that showed a shift of the transmission characteristics of GaN NWs with a dielectric shell to smaller NW diameters/larger wavelengths, respectively [12].

In Fig. 3a, transmission spectra of \(\hbox {GaN}-\hbox {SiO}_{2}\) core–shell NWs with varying shell thickness are plotted for a fixed \(D = 105\,\hbox {nm}\). The observed transmission minimum is shifted to larger wavelengths with increasing shell thickness compared to bare GaN NWs, indicating an enhanced confinement of the waveguide mode. In addition, the transmission minimum is less pronounced with increasing shell thickness. A similar behavior can be observed for bare GaN NWs with increasing NW diameter (Fig. 2a, b). For larger NW diameters, the transmission minimum is shifted to larger wavelengths due to a stronger confinement of the waveguide mode and several, less distinct transmission minima can be observed, which can be ascribed to the efficient coupling of light into higher order waveguide modes and Fabry–P\(\acute{\text {e}}\)rot modes [12]. The less pronounced transmission minimum observed for increasing shell thickness, thus, can be explained by the coupling of light into several modes. The same behavior was measured for GaN–ZnO core–shell NWs with varying shell thickness (not shown). The transmission as a function of the wavelength of core–shell NWs with varying shell materials and a fixed \(d_{\text {shell}} =36 \pm 1\,\hbox {nm}\) is illustrated in Fig. 3b for a fixed \(D = 105\,\hbox {nm}\). For both shell materials, i.e., \(\hbox {SiO}_{2}\) and ZnO, the measured transmission minimum appears at larger wavelengths compared to bare GaN NWs. For GaN–ZnO core–shell NWs, a larger shift can be observed. The refractive index of \(\hbox {SiO}_{2}\) has a value between 1.57 and 1.55 [33], and the refractive index of ZnO ranges between 2.10 and 1.96 in the investigated wavelength regime [34]. The more pronounced shift of the transmission minimum for GaN NWs with a ZnO shell, thus, can be explained by a stronger confinement of the mode due to the higher refractive index of ZnO. The enhanced confinement of the modes for an increasing shell thickness and an increasing refractive index of the shell is in agreement with the shift of the dispersion relations of the modes to smaller NW diameters for NWs with a dielectric shell [32]. As mentioned above, all measurements were performed on the same sample. For this purpose, the \(\hbox {SiO}_{2}\) and ZnO shells were wet chemically etched. To analyze the influence of the etching steps on the transmission characteristics of the bare GaN NWs, transmission measurements of etched GaN NWs were performed (not shown). Similar transmission characteristics were observed. Only slight deviations in the absolute values were measured. However, the transmission minima occur at the same spectral positions. The slight deviations between the measured transmission spectra of bare and etched GaN NWs can be attributed to measurements on different NW array sites. Thus, we conclude that the etching steps have no impact on the transmission properties of the bare GaN NWs.

Fig. 3
figure 3

Transmission spectra of a \(\hbox {GaN}-\hbox {SiO}_{2}\) core–shell NWs with varying shell thicknesses and b core–shell NWs with varying shell materials and a fixed \(d_{\text {shell}} = 36 \pm 1 \,\hbox {nm}\) for a fixed \(D = 105\,\hbox {nm}\).

Full size image

Up to now, all transmission measurements have been carried out in air. However, photocatalytic reactions usually take place in a liquid environment, e.g., water or in an electrolyte. To this end, transmission measurements were performed in different liquid environments, i.e., water and toluol. The refractive index of water is between 1.35 and 1.33 for wavelengths between \(330\,\hbox {nm}\) and \(800\,\hbox {nm}\) [35], whereas the refractive index of toluol has a value between 1.56 and 1.48 [36, 37]. Transmission spectra for a bare GaN NW array without dielectric shell with \(p = 300\,\hbox {nm}, L \approx 580\,\hbox {nm}\) and \(D = 155\,\hbox {nm}\) are shown in Fig. 4a for different environments. A transmission minimum can be observed for all environments, i.e., air, water, and toluol, and can be ascribed to the coupling of the incident light into the fundamental waveguide mode. In contrast to the effect of a dielectric shell around the GaN NWs, the measured transmission minimum appears at smaller wavelengths for environments with increasing refractive index. Note that, for \(\lambda < 360\,\hbox {nm}\), a constant low transmission can be observed which is attributed to the onset of absorption of light within the GaN NWs. Corresponding simulated transmission profiles for a GaN NW array with \(p = 300\,\hbox {nm}, L = 580\,\hbox {nm}\) and \(D = 160\,\hbox {nm}\) are illustrated in Fig. 4b. Sharp transmission features superimposed upon a broad minimum can be observed for all environments. In previous work, we have shown that for NW arrays where the NW period is similar to or smaller than the illumination wavelength, the light transmission through these NW arrays is affected by the waveguide character of the single NWs and the NW array serving as photonic crystal slab [12]. Photonic crystal slabs support in-plane waveguide modes that can trap light in the NW array, resulting in a strong suppression of the light transmission [12, 21]. In addition, guided so-called Fano resonances can be excited giving rise to sharp spectral features in the transmission spectra [38,39,40]. Based on these considerations, we attribute the broad transmission minimum to the coupling of light into the fundamental waveguide mode, whereas the sharp transmission features can be explained by the excitation of Fano resonances supported within the NW array. Comparing the transmission spectra for the different surrounding media, two effects can be observed: first, the transmission minimum arising from the coupling of light into the fundamental mode is shifted to smaller wavelengths with increasing refractive index of the environment, indicating a weaker confinement of the fundamental waveguide mode. In the inset of Fig. 4b, the dispersion relations for the fundamental (\(\hbox {HE}_{11}\)) and the \(\hbox {EH}_{11}\) mode are plotted for a bare GaN NW in air and in water for \(\lambda = 400\,\hbox {nm}\). The effective refractive indices approach the refractive index of the respective environment, i.e., 1 for air and 1.34 for water, for NW diameters close to the cut-off diameters. In addition, the dispersion relations shift to slightly larger NW diameters for the higher index environment, indicating a weaker confinement of the waveguide modes for an increasing refractive index of the environment. Therefore, the observed shift of the transmission minimum to smaller NW diameters can be explained by the shift of the dispersion relations. Second, the sharp transmission features originating from the excitation of Fano resonances occur at different wavelengths for the different surrounding media. Due to the varying environments, the effective refractive index of the underlying photonic crystal slab changes and, thus, different Fano resonances are supported within the NW array [40]. Comparing the measured transmission spectra (Fig. 4a) with the corresponding simulations (Fig. 4b), a good agreement of the spectral positions of the transmission minimum can be observed. However, the measured transmission minima are broadened compared to the simulations and the sharp spectral features do not appear in the measured spectra. The broadening can be explained by disorder in the NW geometry, i.e., variations of the diameter and length of individual NWs, and the occurrence of tripods (fraction of tripods \(\approx 10 \%\), not shown) [12]. The coexistence of NWs and tripods was already observed by Schuster et al. for SAG GaN NWs on diamond substrates, and explained by different polarities of the nucleation seeds [24]. Variations of the NW diameter lead to different coupling efficiencies of the fundamental waveguide mode, resulting in broadened transmission minima. Besides, differences in the NW length and the occurrence of tripods cause a different light extraction of individual NWs and tripods and, therefore, also give rise to broadened transmission features. The absence of the sharp spectral features is, thus, attributed to the disorder in the photonic crystal slab induced by the occurring tripods [41].

Fig. 4
figure 4

a Transmission spectra of bare GaN NWs without a dielectric shell with a fixed \(D = 155\,\hbox {nm}\) for different liquid environments. b Simulated transmission spectra of GaN NWs with \(L = 580\,\hbox {nm}\) and \(D = 160\,\hbox {nm}\) for different environments. The inset shows the dispersion relation for the \(\hbox {HE}_{11}\) and the \(\hbox {EH}_{11}\) mode at \(\lambda = 400\,\hbox {nm}\) for a bare GaN NW in air and water

Full size image

In Fig. 5a, the measured transmission is plotted as a function of the illumination wavelength for a bare GaN NW array with \(p = 300\,\hbox {nm}\), \(L \approx 580\,\hbox {nm}\) and \(D = 210\,\hbox {nm}\), for which no tripods were observed in contrast to the NW array with \(D = 155\,\hbox {nm}\), for an air, a water, and a toluol environment (solid lines). The corresponding simulated transmission spectrum for a bare GaN NW array in air is indicated with a gray dashed line. Sharp spectral features superimposed upon a smoothly varying background can be observed. As for the GaN NW array with \(D = 155\,\hbox {nm}\), the constant low transmission for \(\lambda < 360\,\hbox {nm}\) is ascribed to the absorption of light within the GaN NWs. The measured background resembles the transmission minimum observed for the GaN NW array with a smaller \(D = 155\,\hbox {nm}\) (Fig. 4a) and, therefore, is attributed to the coupling of light into the fundamental waveguide mode. The observed shift of the background to smaller wavelengths for an increasing refractive index of the surrounding medium can be explained by a weaker confinement of the mode for a higher index environment. Based on the above considerations on NW arrays serving as photonic crystal slabs, the sharp spectral features are ascribed to the excitation of Fano resonances supported within the NW array [38,39,40]. Comparing the measured transmission spectrum of bare GaN NWs in air (black solid line) with the corresponding simulation (gray dashed line), a good agreement of the spectral positions of the transmission minima can be observed. To analyze the spectral positions of the Fano resonances in more detail, the measured (filled dots) and the corresponding simulated positions (empty dots) of the occurring Fano resonances (indicated by numbers in Fig. 5a) are illustrated in Fig. 5b for the different environments. The measured positions of the Fano resonances agree well with the calculated wavelengths.

Fig. 5
figure 5

a Measured transmission spectra (solid lines) of bare GaN NWs with a fixed \(D = 210\,\hbox {nm}\) for different environments and the corresponding simulated transmission spectrum (dashed line) for an air environment. b Measured and simulated positions of the occurring Fano resonances indicated by numbers in a

Full size image

4 Conclusion

We have investigated the influence of a dielectric shell and dielectric environments on the light transmission through GaN NW arrays by experimental measurements and compared the measured transmission data with numerically simulated results. A dielectric shell around the GaN NWs shifts the transmission minima originating from the coupling of light into fundamental waveguide modes and Fabry–P\(\acute{\text {e}}\)rot modes to larger wavelengths due to a stronger confinement of the modes. An enhanced shift of the transmission minima for an increasing thickness and refractive index of the shell could be explained by the shift of the dispersion relations of the modes to smaller NW diameters, and is in good agreement with our previous numerical simulations. In addition, we have analyzed the impact of different liquid environments on the transmission characteristics of GaN NW arrays. Transmission minima originating from the coupling of light into waveguide modes are shifted to smaller wavelengths with increasing refractive index of the surrounding medium due to a weaker confinement of the modes. The sharp transmission features which are a result of the NW array serving as a photonic crystal slab appear at different wavelengths due to the different effective refractive indices of the underlying photonic crystal slabs. The experimental findings show a good quantitative agreement with the simulations.