A. Isotropic growth of Ag NPs

Figure 2 shows the typical morphology of isotropic Ag NPs, which were grown in the 180 mg/L (∼1 mM) AgNO₃ solution. The SEM images show that the Ag NPs at the growth time of 2, 4, 8, 16, and 120 min exhibited irregularly spherical shapes. At a 2 min growth time, the Ag NPs are ∼27 nm of average size and ∼4 × 10⁹/cm² of density. When increasing the growth time to 4 min, the Ag NP density dramatically reduced to ∼2.3 × 10⁶/cm², and the size increased to ∼97 nm, indicating that the coalescence process took place between Ag primary particles at a growth time in between 2–4 min. By fitting the time dependence of average size (d) after the coalescence phenomenon, we found d ³ ∝ t, as shown in Fig. 2(f). Such a relationship is an indication of diffusion-limited growth. In other words, Ostwald ripening (OR) ^{Reference Viswanatha, Santra, Dasgupta and Sarma15} dominates the coarsening process of Ag NPs.

FIG. 2.

(a–e) SEM images of Ag–TiO₂ films grown in 180 mg/L AgNO₃ solution at the growth time of 2, 4, 8, 16, and 120 min, respectively. The insets in (a) and (b) are the size distributions of Ag NPs. (f) Growth-time dependence of Ag NP size.

Ex situ SEM observation is incapable of detecting the dynamic process of Ag NP growth and cannot provide the required resolution to recognize the Ag primary NPs in the nucleation stage. Therefore, in situ extinction spectroscopy was used to study the photocatalytic nucleation of Ag NPs. Figure 3(a) shows the in situ extinction spectra from the photocatalytic beginning to 80 s. The extinction was defined by ln(I ₀/I), where I ₀ is the transmitted intensity before photocatalytic reduction and I is the intensity during photocatalytic reduction. The extinction spectra were recorded in 13 ms with intervals of 10 s. As shown in the figure, two extinction peaks, which are associated with Ag primary particle density and size, appeared at ∼400 and ∼510 nm. By plotting the extinction intensity at ∼400 nm as a function of growth time, as shown in Fig. 3(b), the formation of Ag primary particles was extrapolated to a time as early as <1 s after the photocatalytic beginning. As a matter of fact, the extinction of Ag primary particles was observed in the 0-s spectrum, as shown in the inset of the figure, indicating that the nucleation took place truly in a very short time after the photocatalytic reduction started. The extinction spectra show that the number of Ag NPs is increasing with the irradiation time, whereas the change in sizes is not obvious. The formation of Ag primary particles originates from supersaturated Ag atoms, which are related to the photocatalytic reduction ability of TiO₂ films. Based on the statistical analysis of Ag NPs, the Ag reduction rate was estimated to be ∼2 × 10¹³ cm⁻² s⁻¹, which is reasonable in comparison with the irradiation photons of ∼10¹⁶ cm⁻² s⁻¹.

FIG. 3.

(a) In situ extinction spectra of Ag NPs grown in 180 mg/L AgNO₃ solution from the photocatalytic beginning to 80 s, in which the spectra were recorded in 13 ms with intervals of 10 s. (b) Growth-time dependence of extinction intensity at ∼380 nm and the inset is 0-s extinction spectrum.

B. Anisotropic growth of Ag NPs

The Ag reduction rate is determined by the parameters related to UV irradiation, reduction activity of TiO₂ film, AgNO₃ concentration of solution, etc. By increasing the AgNO₃ concentration, we found that the increased Ag reduction rate gave rise to a change in Ag NP growth, resulting in the formation of Ag nanoplates in the product. Figure 4 shows an example of Ag NP growth in the 2800 mg/L AgNO₃ solution. As shown in the figure, a large amount of thin Ag nanoplates were produced at the growth time as early as 2 min. The Ag nanoplates were ∼10 nm thick and 0.1–1 μm of edge length. With an increase in the growth time, the sizes of Ag nanoplates were not varied obviously, but the thickness increased. TEM analysis shows that the nanoplates are faced with Ag (111) planes, as shown in Fig. 4(f). At a 120 min growth time, the Ag nanoplates were rarely observed, instead evolved into three-dimensional Ag NPs with the well-defined shapes, which are a little different from the irregular Ag NPs grown in the 180 mg/L AgNO₃ solution. The SEM images reveal that the increased AgNO₃ concentration promoted an early coalescence resulting in the formation of Ag nanoplates. In other words, the Ag nanoplates were formed in the early period of coalescence, rather than gradually grew up with the increase in the growth time.

FIG. 4.

(a–e) SEM images of Ag–TiO₂ films grown in 2800 mg/L AgNO₃ solution at the growth time of 2, 4, 8, 16, and 120 min, respectively. (f) TEM image of a triangular Ag nanoplate and diffraction pattern (inset).

Figure 5 shows the in situ extinction spectra from the beginning to 80 s. It is different from the growth of Ag NPs in the 180 mg/L AgNO₃ solution that the extinction in the visible region is slowly increased and the extinction peaks associated with individual Ag NPs are not clearly distinguishable. On the other hand, the extinction in the NIR region is rapidly increased with the increase in the growth time. According to the effective medium theory, ^{Reference Moores and Goettmann16,Reference Ung, Liz-Marza and Mulvaney17} the enhanced extinction in the NIR region can be attributed to the extreme high-density primary particles and/or the formation of Ag nanoplates. In fact, the Ag nanoplates were formed at a quite early time, because a small extinction peak at ∼340 nm, which has been assigned to out-of-plane quadrupole resonance of Ag nanoplates in the AgNO₃ solution, ^{Reference Jin, Cao, Hao, Me, Schatz and Mirkin18} emerged in the 30-s spectrum and gradually changed to be clear with the increase in the growth time, as indicated by the arrow in the figure. SEM images show that Ag nanoplates and irregularly spherical Ag NPs were co-existed in the photocatalytic growth using 2800 mg/L AgNO₃ solution, whereas the isotropic growth of Ag NPs was restrained. The Ag reduction rate in 2800 mg/L AgNO₃ solution was estimated to be ∼2 × 10¹⁴ cm⁻² s⁻¹, which is ∼10 times of that in 180 mg/L AgNO₃ solution. Therefore, we conclude that the increased reduction rate promotes the formation of Ag nanoplates. Such nanoplates are of interest in the fields such as stereoscopic visions, storing multiple information, certification, and anticounterfeit technologies and polarizers, ^{Reference Sakai, Tanabe and Tatsuma19,Reference Tanabe, Matsubara, Sakai and Tatsuma20} because they have three major LSPR modes, namely, out-of-plane mode, in-plane longitudinal and transverse modes, and respond selectively to light polarized in a specific angle.

FIG. 5.

In situ extinction spectra of Ag nanoplates grown in 2800 mg/L AgNO₃ solution from the photocatalytic beginning to 80 s, in which the spectra were recorded in 13 ms with intervals of 10 s.

C. Growth mechanism of Ag NPs

Using 50–3200 mg/L AgNO₃ solution, we systematically investigated the photocatalytic growth dynamics of Ag NPs using in situ extinction spectra. As shown in Fig. 6, the extinction spectra reveal that the Ag NPs are of typical isotropic growth in the solutions of ≤300 mg/L and transit into anisotropic growth in the solutions of ≥400 mg/L, indicating that a critical AgNO₃ concentration in between 300 and 400 mg/L is needed for the formation of Ag nanoplates. Besides AgNO₃ concentration, the surface properties of TiO₂ films were also found to be responsible for the formation of Ag nanoplates. All the films used for the above study were kept in moisture environment for weeks. Using a fresh film, which was immediately used for Ag NP growth after calcination, or using films kept in a dry atmosphere, Ag nanoplates were never obtained by photocatalytic growth in 50–3200 mg/L AgNO₃ solutions, suggesting that H₂O absorption modified the surface properties of TiO₂ films and changed the photocatalyic reaction. We believe that the re-oxidation of reduced Ag atoms and the electron–hole recombination are the likely reasons for decrease in Ag reduction rate. To prevent Ag atoms from re-oxidation and/or to effectively separate charges, a layer of vacuum pump oil was coated on the surface of TiO₂ film, and then the anisotropic Ag NP growth was successfully induced, ^{Reference Li, Tao, Li and Zhang21} evidencing our assumption.

FIG. 6.

In situ extinction spectra of Ag–TiO₂ films grown in 50–3200 mg/L AgNO₃ solutions, in which the extinction spectra were recorded in 13 ms with intervals of 6 min.

By carefully observing the SEM images [Figs. 4(a)–4(c)], we found that the nanoplates generally lie on the irregularly spherical NPs, and the small NPs beneath thin nanoplates are visible, indicating that the nanoplates were produced in the solution, rather than on the surface of TiO₂ film. In other words, Ag nanoplates were produced primarily in the area adjacent to the film surface, and adhered to the surface when they were taken out from the solution. As a matter of fact, a large amount of low-dimensional Ag nanostructures, such as nanowires, nanoribbons, and nanoplates, had been taken from the solution when the photocatalytic growth of Ag NPs was conducted for a longer time using TiO₂ or ZnAlO films as the reduction media, as shown in the supporting material (Fig. S1). To understand the formation mechanism of Ag NPs and Ag nanoplates, we proposed a growth model by taking Ag atom and Ag⁺ ion diffusion into consideration. Because the solution density is close to that of a solid, the force applied to Ag atoms by the substrate atoms is approximately equal to that by the atoms in the solution. Thus, it is different from the film growth in a vacuum that Ag atoms on the surface of TiO₂ film have a large possibility to diffuse into a solution. Considering that the Ag atoms were produced on the surface of TiO₂ film by the reduction of Ag⁺ ions, the one-dimensional diffusion equations and the boundary conditions for Ag atom and Ag⁺ ion diffusion can be written as

where u(x, t) and v(x, t) are the Ag and Ag⁺ densities at position x and at time t, respectively; D ₀ and D ₊ are the corresponding diffusion coefficients; k _R and k _O are the reduction and oxidation rates. Numerical calculations showed that the reduction on TiO₂ film reaches equilibrium with the oxidation in a time scale of <<1 ms. Thus, u(x, t) is something like the diffusion function with u(0, t) = constant. Taking the film roughness and the transverse diffusion into account, a specific u(x, t) in Fig. 7(a) is suggested. Once u(x, t) is supersaturated, nucleation is possible. The critical free energy (ΔG _c), critical radius (r _c), and critical density (n _c) of primary particles are related to supersaturated concentration (C) by ^{Reference Ohring22}

where C ₀ is the saturated concentration of Ag atoms, σ is the surface free energy density of primary particles, v _a is the atomic volume, and n _s is the number of nucleation sites. Equations (5) and (7) tell us that the distribution u(x, t) of Ag atoms produces nonuniformly distributed primary particles, which might be decisive for the formation of Ag nanoplates or not. According to Eqs. (6) and (7), the large value of u(x, t) produces the high-density and small-size primary particles. Therefore, if the supersaturated u(x, t) is lower than a critical level in the nucleation gestation, the density of primary particles is low, and then the OR mechanism dominates the coalescence process leading to the isotropic growth of Ag NPs. If u(x, t) within a large enough area is higher than the critical level, the density of primary particles is so high that a new coalescence mechanism, probably the oriented attachment (OA) mechanism, promotes the formation of Ag nanoplates, as shown in Figs. 7(b) and 7(c). At the same time, the OR growth still dominates the area where u(x, t) is lower than the critical level, and then the isotropic Ag NPs are present. The restrained isotropic growth of Ag NPs can be attributed to the larger surface free energy of irregular Ag NPs in comparison with that of Ag nanoplates because the atom transfer is always from where surface free energy is high to where surface free energy is low.

FIG. 7.

Photocatalytic growth model of Ag NPs in AgNO₃ solution: (a) reduction of Ag atoms, (b) nucleation of Ag primary particles, and (c) coalescence of Ag NPs and/or Ag nanoplates. The top panel is for high AgNO₃ concentration and the bottom panel is for low AgNO₃ concentration.

The growth model can explain the phenomenon of Ag nanoplates lying on the small Ag NPs. On the other hand, the growth model is consistent with the conclusion of Yin et al. that the OA growth is facilitated in high-density primary particles. ^{Reference Yin, Huang, Zhang, Zheng and Lin23} Similarly, the formation of Ag nanoplates using AgNO₃ ethanol solution ^{Reference Tanabe, Matsubara, Stridge, Kazuma and Kelly12} or using oil-decorated TiO₂ films ^{Reference Li, Tao, Li and Zhang21} can also be attributed to the increase in the Ag reduction rate due to the photodegradation of organic molecules. It should be emphasized that the growth model suggested the importance of surface roughness in the formation of Ag nanoplates, which is consistent with the observation of Tanabe et al. that the growth of Ag nanoplates was related to the surface roughness and random features of substrate. ^{Reference Tanabe, Matsubara, Stridge, Kazuma and Kelly12} According to our growth model, a smooth TiO₂ film facilitates the formation of large-scale Ag nanoplates. As a matter of fact, a Ag nanoplate larger than ∼10 μm × 10 μm was obtained by using a smooth TiO₂ film, as shown in the supporting material (Fig. S2). On the other hand, the growth model is proposed on the basis of our experimental setup with strong UV irradiation. Under strong UV irradiation, the nucleation on TiO₂ films is not important in the photocatalytic growth of Ag NPs because the reduction–oxidation environment does not facilitate the formation of stable Ag primary particles. To realize the epitaxial growth of Ag NPs, we recommend using a low UV photon flux and a single-crystal TiO₂ substrate.

D. Extinction spectra of Ag–TiO₂ films

Figure 8 shows the extinction spectra of Ag–TiO₂ films grown at the growth time of 2, 4, 8, 16, and 120 min. For the films grown in the 180 mg/L AgNO₃ solution, the Ag–TiO₂ film at 2 min growth time exhibits a UV extinction peak at ∼380 nm and a blue–green extinction peak at ∼450 nm, as shown in Fig. 8(a). With an increase in the growth time, the UV peak was blue-shifted and the blue–green peak was red-shifted. The two extinction peaks are very like the quadrupole and dipole resonances of Ag spheroids, ^{Reference Kelly, Coronado, Zhao and Schatz24} however, they are more likely due to a combination of factors because visible quadrupoles require extreme size, shape, and external medium regularity in a typical macro geometry which averages over thousands of particles. In other words, the extinction spectra are the hybridized plasmon response of the complex nanostructures ^{Reference Prodan, Radloff, Halas and Nordlander25} or due to an asymmetric dielectric environment. ^{Reference Beck, Verhagen, Mokkapati, Polman and Catchpole26} When increasing the growth time, the extinction peak at ∼380 nm was rapidly enhanced and the extinction peak at ∼450 nm was weakened. The extinction peak at ∼450 nm was no more observed for the Ag–TiO₂ film at a 120 min growth time.

FIG. 8.

Extinction spectra of Ag–TiO₂ films at the given growth times: (a) Ag–TiO₂ films grown in 180 mg/L AgNO₃ solution and (b) Ag–TiO₂ films grown in 2800 mg/L AgNO₃ solution.

For the Ag–TiO₂ film grown in the 2800 mg/L AgNO₃ solution at 2 min growth time, the extinction spectrum is something like that of a smooth and compact Ag film, and no particle-like extinction peak was observed, as shown in Fig. 8(b), indicating that the Ag nanoplates with various sizes dominate the extinction of the Ag–TiO₂ film. With the increase in the growth time, the particle-like extinction emerges and gradually changes to be visible. By observing the SEM images presented in Fig. 4, we suggest that the particle-like extinction results from the transition of Ag nanoplates to the Ag NPs with well-defined shapes. At 120 min growth time, the extinction spectrum changed to be similar to that of the Ag–TiO₂ film grown in 180 mg/L AgNO₃ solution, which is in accordance with SEM observation.

E. SERS of Ag–TiO₂ films

Using R6G as the probe molecule, the SERS activity of the Ag–TiO₂ films was evaluated. Figure 9(a) shows the Raman spectra of R6G molecules on Ag–TiO₂ films grown in 180 mg/L AgNO₃ solution at the growth time of 2, 4, 8, 16, and 120 min. The R6G featured bands are identified to be the in-plane bending vibration of C–C–C ring at 612 cm⁻¹, the out-of-plane and in-plane bending vibrations of C–H at 770 and 1130 cm⁻¹, and aromatic stretching vibrations at 1362, 1510, and 1650 cm⁻¹. ^{Reference Hildebrandt and Stockburger27,Reference Tao, Li, Zhang, Kang, Yang, Qiu and Liu28} The Raman intensities increased when increasing the growth time from 2 to 120 min. Using the Raman intensity of 0.1 M R6G on a pure TiO₂ film as the reference, EF_SERS was determined by carefully calculating the number of R6G molecules within the irradiation area. The Ag–TiO₂ film at 120 min growth time exhibits an enhancement factor of ∼10⁶, indicating that such a cost-effective Ag–TiO₂ film is available for SERS applications.

FIG. 9.

(a) Raman spectra of R6G molecules on Ag–TiO₂ films that were grown in 180 mg/L AgNO₃ solution at the given growth time. (b) SERS enhancement factor as a function of the film surface roughness.

Two kinds of electromagnetic “hot spots”, the sub-10 nm “feed-gaps” between metal NPs and the “lightning rods”, ^{Reference Garcia-Vidal and Pendry29,Reference Yang, Matsubara, Xiong, Hayakawa and Nogami30} have been known making the important contributions to SERS. Based on our SEM observation, the “feed-gaps” were excluded because the adjacent Ag NPs are generally separated by a distance larger than 10 nm. Thus, “lightning rods” are proposed to be the SERS mechanism of Ag–TiO₂ films. To demonstrate our proposal, AFM was used to quantitatively measure the roughness of Ag–TiO₂ films, and the roughness dependence of EF_SERS is plotted in Fig. 9(b). EF_SERS increased with the roughness of Ag–TiO₂ films, indicating that the enhanced Raman signals can be attributed to the “lightning rods” effect at the sharp edges of Ag NPs, similar to the SERS mechanism on a rough Ag metal surface.

Similar to the films grown in 180 mg/L AgNO₃ solution, the Ag–TiO₂ films grown in 2800 mg/L AgNO₃ solution extremely enhanced the Raman signals of R6G molecules, as shown in Fig. 10(a). Figure 10(b) shows the growth-time dependence of SERS enhancement factor, and the EF_SERS values of the Ag–TiO₂ films grown in 180 mg/L AgNO₃ solution are also plotted for comparison. The Ag–TiO₂ films dominated by Ag nanoplates, which are the two samples grown in 2800 mg/L AgNO₃ solution for 2 and 4 min, produced a ∼1 × 10⁵ Raman enhancement, which is larger than that of Ag–TiO₂ films with small Ag NPs, but smaller than that of Ag–TiO₂ films with large Ag NPs. With the increase in the growth time, the EF_SERS values increased. The increase of EF_SERS values can be ascribed to the transition of Ag nanoplates to Ag NPs with well-defined shapes, increasing the density of large Ag NPs. However, the production of large Ag NPs in 2800 mg/L AgNO₃ solution is lower than that in 180 mg/L AgNO₃ solution, as shown in Fig. 4(e). Therefore, the increase in Raman enhancement is limited when the growth time was extended to 120 min. The results suggest that using a high concentration solution cannot bring large SERS values due to the formation of Ag nanoplates.

FIG. 10.

(a) Raman spectra of R6G molecules on Ag–TiO₂ films that were grown in 2800 mg/L AgNO₃ solution at the given growth time. (b) SERS enhancement factor as a function of the growth time.

F. Enhanced fluorescence of Ag–TiO₂ films

Silver island films were studied in fluorescence enhancement, but few reports involved the Ag–TiO₂ film with high-density Ag NPs. Using RhB as the fluorophore, we studied the fluorescence enhancement of the Ag–TiO₂ films grown in 180 mg/L AgNO₃ solution. The fluorophore of 1 mM RhB in TiO₂ sol–gel was coated on the surface of Ag–TiO₂ films to avoid the fluorescence quenching. The average distance between the fluorophores is ∼12 nm, larger than Forster distance (4.2 nm) for the homoresonance energy transfer, and the fluorophore density was estimated to be ∼6 × 10¹⁵ cm⁻² in a 100 nm layer adjacent to the surface of Ag NPs. Considering that the PL spectrum strongly depends on excitation photon energy, we first measured the absorption, PL, and PLE spectra of RhB in solution to determine the excitation wave length. As shown in Fig. 11(a), the PLE spectrum was found essentially superposed with the absorption, and the maximum PLE at 554 nm partially overlapped with PL emission, thus the excitation light for PL measurement was set at the wave length of 500 nm. Similar RhB PL spectra on the pure TiO₂ films were obtained and the PL intensity did not obviously depend on the thickness of fluorophore layer, as shown in the inset of Fig. 11(a), in which the RhB PL spectra of the samples with different thick fluorophore layers are presented. Compared to the RhB fluorescence on the pure TiO₂ films, the PL intensity of RhB on the Ag–TiO₂ films was enhanced and the enhancement strongly depends on the growth time of Ag NPs, as shown in Fig. 11(b). Using the RhB PL spectra on the pure TiO₂ as the reference, the enhancement factor (EF_PL) was determined to be 1.5–6.9, as shown in the inset of Fig. 11(b). The enhancement is consistent with the measurement for the other fluorophore on the silver island films. ^{Reference Lakowicz, Geddes, Gryczynski, Malicka, Gryczynski, Aslan, Lukomska, Matveeva, Zhang, Badugu and Huang31}

FIG. 11.

(a) Absorption, PL, and PLE spectra of RhB solution. The PL spectrum was excited by 330 nm light and the PLE spectrum was detected at 578 nm. The inset shows RhB PL spectra on pure TiO₂ films. (b) RhB PL spectra on Ag–TiO₂ films that were grown in 180 mg/L AgNO₃ solution at the given growth time.

Fluorescence enhancement results primarily from two processes, an increase of the localized electromagnetic fields near the metal NPs and an increase of radiative decay rate. ^{Reference Lakowicz32} Thus, EF_PL = G _exc × G _QY, where G _exc is the enhancement of the localized electromagnetic field, and G _QY is the increase in quantum yield of the fluorophore, which can be estimated with the PL lifetime. For the RhB molecules on the pure TiO₂ films, the determined average lifetime was 0.63 ± 0.14 ns. The RhB PL lifetime on the Ag–TiO₂ films was found to be 0.35–0.58 ns and did not obviously depend on the growth time of Ag NPs. Considering that the quantum yield of a fluorophore in solid cannot exceed a few percents, G _QY was estimated smaller than 2. Thus, the localized electromagnetic field and/or the increased scattering of the Ag–TiO₂ films play an important role in the PL enhancement of RhB besides the increase in quantum yield. In addition, the increased roughness of the Ag–TiO₂ film by deposition time might result in larger surface area allowing more dye molecules located in the near field of the Ag nanostructures, thus leading to more enhancement of PL intensity.