The fabrication process is schematically presented in . The surface of a Si(100) wafer was patterned into a periodic array of pyramidal pits (Figure S1, Supporting Information File 1
) by using SCIL, reactive ion etching (RIE), and KOH etching. The spatial period of these pits is 520 nm. A 200 nm layer of SiO2
was thermally grown on the Si wafer, and then an array of holes was defined by SCIL. SCIL was developed as a new nanoimprint lithography technique that combines the advantages of both UV nanoimprint lithography techniques, with a rigid stamp for best resolution and with a soft stamp for large-area patterning [23
]. The soft stamp was carried on glass in SCIL, using a sequential imprinting principle instead of backside pressure to minimize air inclusions on large areas and to ensure high uniformity. The imprinted structure was then transferred through the SiO2
layer with RIE. The SiO2
pattern with the array of holes acted as a mask during the anisotropic etching of Si in a KOH solution, and the Si surface was patterned into a periodic array of pyramidal pits. After removal of the SiO2
mask, about 20 nm of SiO2
was then again thermally grown on the structured Si surface to avoid a reaction between the subsequently deposited metal films and the Si substrate. Au/Ag bilayers with different layer thicknesses (10 nm/20 nm, 10 nm/25 nm, 10 nm/30 nm, 15 nm/25 nm, and 15 nm/30 nm) were deposited onto the prepatterned substrates by e-beam evaporation, and then annealed at 700 °C in Ar for 15 min to induce dewetting. This temperature is well below the solidus temperature of the Au–Ag system, i.e., the dewetting is solid-state dewetting. Hence, interdiffusion of Au and Ag occurred, and perfectly ordered arrays of Au–Ag alloy nanoparticles were formed on the prepatterned substrates. Subsequently, dealloying by submerging the samples in a HNO3
solution resulted in the transformation of the Au–Ag alloy nanoparticles into the nanoporous gold nanoparticles, due to the dissolution of Ag out of the alloy. As the final result, a well-defined ordered array of nanoporous gold nanoparticles was obtained. The dewetting of the bilayers on the prepatterned substrate with 20 nm thermal SiO2
was also performed by annealing at 800 and 900 °C. However, growth of some Si or silicide particles was observed and the particles remained after dealloying; this will be investigated in more detail in the future. In addition, reference samples with 15 nm Au/20 nm Ag bilayers deposited onto a flat Si substrate with 100 nm thick, thermally grown SiO2
layer was annealed at 900 °C in Ar for 15 min and then dealloyed for comparison. Dewetting on a prepatterned substrate takes place at a lower annealing temperature [24
], and a higher temperature (900 °C) is required for the dewetting on a flat substrate to form particles.
Schematics of the fabrication process for an ordered array of nanoporous gold nanoparticles.
Excess local chemical potential can be introduced by the prepatterned structure according to the Gibbs–Thomson relation, Δμ = κ·γ·Ω, where Δμ is the local excess chemical potential, κ the local curvature, γ the surface energy, and Ω the atomic volume. There is an excess positive chemical potential at peaks or ridges due to the positive local curvature and an excess negative chemical potential at pit valleys due to the negative local curvature. Consequently, there is an additional driving force for the diffusion of the metal atoms from the peaks to the valleys during dewetting on the prepatterned substrate, leading to the formation of the ordered array of nanoparticles. However, in addition to the curvature-driven diffusion, the capillary driven diffusion (the dominating process for the dewetting on a flat substrate [24
]) and grain growth are additionally two important processes during dewetting, making the formation of the ordered nanoparticle arrays thickness-dependent [19
]. For example, the Au–Ag alloy nanoparticles are irregularly distributed after dewetting 10 nm Au/20 nm Ag bilayers on the prepatterned substrate (Figure S2, Supporting Information File 1
). This is probably due to insufficient total layer thickness. As the total layer thickness is increased adequately (Au/Ag: 10 nm/25 nm, 10 nm/30 nm, 15 nm/25 nm, and 15 nm/30 nm), the dewetting on the prepatterned substrate can lead to the formation of ordered arrays of the nanoparticles. It is interesting to note that the optimized thickness for the formation of ordered arrays of particles by dewetting of a thin film in this study is twice as much as that in the previous work [19
], although the pit arrays used in this work and the previous work have the same spatial period. However, the pits were fabricated by KOH etching in this study and have a depth of about 360 nm, whereas the pits in the previous work were fabricated by reactive ion etching and have a depth of 150 nm. This means that a larger optimized thickness is required for the formation of an ordered array in the deeper pits with the same spatial period. Additionally, it is possible to control the particle size and spacing by varying the structural parameters (period and depth) of the prepatterned structure, and the layer thickness [19
Subsequently, dealloying of the ordered arrays of the Au–Ag alloy nanoparticles can result in the formation of ordered arrays of nanoporous gold nanoparticles. shows the SEM images of the ordered array of Au–Ag alloy nanoparticles dewetted from 15 nm Au/30 nm Ag bilayers and the obtained ordered array of nanoporous gold particles after dealloying. The arrays of the nanoporous gold nanoparticles induced from the 10 nm Au/25 nm Ag, 10 nm Au/30 nm Ag, and 15 nm Au/25 nm Ag bilayers are still not perfect, and even instances of two particles in a single pit can be observed (Figure S3, Supporting Information File 1
). A perfectly ordered array of nanoporous gold nanoparticles was obtained from the 15 nm Au/30 nm Ag bilayers and there is only one nanoporous gold nanoparticle in every pit. displays the perfectly ordered array of nanoporous gold nanoparticles in a large area, and the porosity of the particles can be seen more clearly in the corresponding magnified SEM image (). In and , the white circular lines outside of the particles can be identified as the previous contours of the dewetted Au–Ag alloy nanoparticles before dealloying. This clearly hints to shrinkage of the particles by dealloying. In a previous work, 29% volume shrinkage of the nanoparticles was reported after dealloying [13
Figure 2 SEM micrographs of samples before and after dealloying: (a) ordered array of Au–Ag alloy nanoparticles dewetted from the 15 nm Au/30 nm Ag bilayers, and (b) ordered array of nanoporous gold nanoparticles formed after the subsequent dealloying. (more ...)
SEM images (false color) at 25° tilt of the perfectly ordered array of the nanoporous gold nanoparticles formed from the 15 nm Au/30 nm Ag bilayers.
The surface morphology of the nanoporous gold particles on the prepatterned substrates changes with the thickness ratio of the as-deposited bilayers. Au and Ag are fully miscible, thus the Au concentration in the formed Au–Ag alloy nanoparticles can be roughly calculated based on the layer thickness ratio. It is 29 atom % for the 10 nm Au/25 nm Ag bilayers, 25 atom % for the 10 nm Au/30 nm Ag bilayers, 38 atom % for the 15 nm Au/25 nm Ag bilayers, and 34 atom % for the 15 nm Au/30 nm Ag bilayers. The average ligament size <θ> is plotted as a function of the Au concentration, as shown in . <θ> decreases from 25 nm (for the particles induced from the 10 nm Au/30 nm Ag bilayers) to 9 nm (for the particles induced from the 15 nm Au/25 nm Ag bilayers) with increasing Au concentration. As Au concentration approaches above 34 atom %, the ligament size seems to approach a lower saturation value of <θ> = 10 nm. Therefore, the ligament size (or pore size) of the nanoporous nanoparticles can be controlled by varying the layer-thickness ratio of the as-deposited Au/Ag bilayers. The size shrinkage decreases with increasing Au concentration, as observed from the contours (white circular lines) of the original Au–Ag alloy nanoparticles and those of the nanoporous gold nanoparticles (inset SEM images in ).
Plot of the average ligament size as a function of the Au concentration of alloy nanoparticles. The error bars represent the standard deviation. Insets show the corresponding SEM images. Scale bars in the insets are 100 nm.
shows a comparison of the data from both the array of nanoporous gold nanoparticles induced from the 15 nm Au/30 nm Ag bilayers on the prepatterned substrate and the irregularly distributed nanoporous gold nanoparticles induced from the 15 nm Au/20 nm Ag bilayers on the flat substrate. and are the SEM images, and are histograms of the particle size distribution, and the mean particle size <m
> is determined. and are the plots of the radially averaged autocorrelation. The radially averaged autocorrelation is calculated from the autocorrelation or pair correlation as a function of radial distance. The first minimum of this function shows information about the mean particle size <m
′>, and the subsequent first maximum denotes the characteristic particle spacing s
. There is a small difference between <m
> and <m
′> due to the different determination methods. Normally, the mean particle size or diameter <m
>, width of the particle size distribution <λp
>, and the characteristic particle spacing s
increase with increasing film thickness for the dewetted nanoparticles on flat substrates [19
]. However, the prepatterned substrates with nanostructures lead to an obvious reduction of the particle size and spacing [19
]. Although the total bilayer thickness of the as-deposited bilayers for the ordered array of nanoporous gold nanoparticles is clearly larger than that for the nanoporous gold nanoparticles on the flat substrate, it can be seen that the mean particle diameter (323 nm) and characteristic particle spacing (538 nm) of the ordered array of nanoporous gold nanoparticles are much smaller than those (639 nm and 1377 nm) of the nanoporous gold nanoparticles induced on the flat substrate. The particle size distribution for the array of nanoporous gold nanoparticles on the prepatterned substrate possesses a much smaller width comparing to that for the nanoporous gold nanoparticles induced on the flat substrate, i.e., the particles on the prepatterned substrate are much more uniform and almost perfectly ordered. The characteristic particle spacing (538 nm) of the array of the nanoporous gold nanoparticles corresponds nearly to the spatial period (520 nm) of the pits of the prepatterned substrate, and the small deviation is probably due to the uncertainty of the radially averaged autocorrelation. Comparing the radially averaged autocorrelation of the nanoporous gold nanoparticles on the flat substrate, the plot of the array of the nanoporous gold nanoparticles on the prepatterned substrate shows a periodic wave shape, denoting the high regularity of the nanoparticle array, which is well confirmed by the autocorrelation image (inset in ).
Figure 5 Data for the ordered array of nanoporous gold nanoparticles formed from the 15 nm Au/30 nm Ag multilayers on the prepatterned substrate (a, c, and e) and the irregularly distributed nanoporous gold nanoparticles formed from the 15 nm Au/20 nm Ag bilayers (more ...)