3. Results and discussion
After hydrothermal growth, the glass slides were covered by a dense layer of ZnO NR with diameters in the 50–100 nm range. Larger (µm size) ZnO crystals were also present (Fig. 1).
For comparison, we first mention the results of ZnO dip-coating without PLL surface treatment. In this case, hardly any NP could be identified in the SEM images of ZnO NR film and the microcrystals (Fig. 1). The absence of NP was also evident by the lack of plasmonic colouring (Fig. 2(b, d)) in diffuse reflection from the macroscopic samples. It can be reasoned that, although wurtzite-type ZnO NP have positive surface charges [34], due to adsorption of negative ions in the citrate and phosphate buffers, the net force on the NP in stabilized colloids may become repulsive. As a result, the NP would be pushed away from the meniscus zone before they can get trapped by capillary forces in the grooves of the nanostructured ZnO surface. Another factor that may hinder the convective CFA colloid assembly is the hydrophobicity of the ZnO NR films [35], which can prevent formation and sustainability of the wetting film and reduce the evaporation surface area. Reduced evaporation flux leads to reduced particle flux towards the three-phase contact line [36]. As a result, the number of NP that deposit on a surface from the relatively dilute colloids was extremely low.
The effect of PLL surface treatment prior to dip-coating becomes obvious as the samples acquire the characteristic plasmonic colouring of Ag (yellow) and Au (pink) NP (Fig. 2(c, e)). For a uniform sample coverage with NP a steady CFA assembly is required, however, on the samples only few mm in size this was limited by the initial meniscus formation and the colloid accumulation at the bottom of the sample. Nevertheless, the uniform regions are of sufficient size for MExS and PL microscopy.
The metal NP after CFA deposition can attach to the ZnO surface deep in the NR film, which makes them difficult to spot in the SEM images. However, a mechanical micro-scratch can reveal the buried NP (Fig. 3) and enable one to estimate the number of particles per unit area to be of the order 10 µm–2. Such concentration of NP is sufficient to cause the distinct LSPR colouring of the samples. For comparison, consider the ruby red original Au colloid (OD 1, with 1.9·1010 particles/mL according to manufacturers’ data). If all particles from a 1 cm3 volume were deposited on a 1 cm2 surface area, the corresponding coverage density would be 190 µm–2, which is only a single order of magnitude larger than the observed NP density on the ZnO surface with a pink appearance.
A similar NP density can also be found on the facets of the ZnO micro-crystals (Fig. 4). One can observe that the flat micro-crystal surfaces contain larger NP aggregates in comparison to the fine ZnO NR film, which hosts mainly individual NP and small assemblies. This effect is similar to the NP aggregate splitting by NAA films [30]. Finally, we note that for PLL treated samples the number of NP on all ZnO crystal facets is similar without significant preference to crystallographic planes as would be expected for the anisotropic ZnO crystals [22].
The MExS spectra recorded from randomly selected spots on ZnO substrates decorated with Ag and Au NP are shown in Fig. 5. Despite considerable variation, the spectra show a similar wavelength dependence that is clearly different for Ag and Au colloids. According to the manufacturers’ data, for the Ag colloid, the extinction peak is expected at 450 nm (in liquid) and 540 nm for Au NP of the same size (60 nm diameter). The extinction peak wavelength for NP on the ZnO surface may differ slightly from that in the buffer solution due to a different effective refractive index of the surrounding medium. Nevertheless, similar peak wavelength values can be presumed in the measured spectra of dry NP on ZnO (Fig. 5). This is a good indication that a significant number of NPs are isolated and have not aggregated in large clusters. For aggregated Au NP the extinction maximum would shift to a longer wavelength causing a blue/purple appearance of the samples. The difference in spectra can be attributed to the presence of larger ZnO crystals with an uneven distribution across the sample area, which causes different proportions between isolated and aggregated NP.
When illuminated with a UV light from a Hg lamp through the excitation filter 330–385 nm transmission range, the samples emit orange light that can easily be seen by an unaided eye. The PL spectra (Fig. 6(a)) have a maximum at 640 nm or 1.94 eV. The orange emission from hydrothermal ZnO NR has previously been observed [37] and attributed to oxygen interstitials defects on the surface.
The presence of metal NP in our case modified the PL spectra only at shorter wavelengths, below 640 nm. The most notable difference in the PL spectra before and after deposition of NP was found at 560 nm for Au NP on ZnO (Fig. 6(b)). This wavelength corresponds to yellow/green luminescence bands attributed to several defect types in the ZnO structure [9]. This wavelength is also close to the Au NP extinction maximum wavelength (560 nm), which is the likely cause of change of the PL signal. Indeed, the suppression of defect emission at similar wavelengths in composites of ZnO NR and Au NP has been reported before [38] and explained by the resonant excitation of particle plasmons and a nonradiative decay via the excitation of electron–hole pairs. The explanation also agrees with the absence of similar effects in Ag NP, the LSPR resonance of which does not coincide with the defect emission energy. It should be noted that in our case only few ZnO NR are in contact with metal NP. In spite of that the dip-coating in Au NP colloid caused a notable PL suppression. For comparison, sputtered Au NP with much smaller sizes but a larger contact area with ZnO have been shown to almost completely eliminate the visible PL [39].