J. Phys. Chem. B 2006, 110, 7743-7749
7743
Nanostructure of Silver Metal Produced Photocatalytically in TiO2 Films and the Mechanism of the Resulting Photochromic Behavior K. Lance Kelly* and Koichi Yamashita Department of Chemical System Engineering, Graduate School of Engineering, The UniVersity of Tokyo, 7-3-1 Hongo Bunkyo-ku, Tokyo 113-8656, Japan ReceiVed: September 9, 2005; In Final Form: February 14, 2006
The optical activity of composite films created by the photocatalytic reduction of silver or gold ions in TiO2 upon irradiation by UV light has up to now been discussed in terms of the formation and light-induced destruction of distinct nanoparticles molded inside the porous nanocrystalline film. We present results from classical light scattering calculations and a logical analysis of experimental observations to add detail to the mechanism. As opposed to large, solid metal nanoparticles, coatings and small particles in heterogeneous external dielectric environments account for observations such as the broad optical spectrum and multiwavelength photochromic responses. For some steps of the photochromic process, we propose that visible light permits an equilibrium promoting the growth of small metal features or suspended particles. We use a new expression for the restricted path length in our size-dependent broadening corrections of metal shells and discuss this briefly. We conclude by discussing the consequence of plasmon absorption in the proximity of the electronically active TiO2 surrounding matrix, leading to mass transfer and shape change of the metal and photochromic properties of the film.
I. Introduction Advances in metal nanometer-scale optics hold recognized potential in applications as diverse as optical data storage, biodiagnostics, advanced waveguides, photovoltaic cells, and single-molecule spectroscopy.1-5 One powerful theme in this experimental research is the manipulation of the metal nanostructure utilizing far-field light illumination. Particle size in solution is affected by high-energy laser pulses6,7 which is explained well by considering thermal energy dissipation, but nanostructure is also strongly directed by illumination by lowpower ambient room light.8-10 Careful investigation11 has shown that the growth of large nanoparticles is directed by the surface plasmon resonance, i.e., the collective oscillation of conduction electrons whose resonance energy depends on the particle’s size, shape, and local environment. However, the solution phase is not conducive to the practical use of metal nanoparticles in many envisioned applications. For metal suspended in solid-phase materials, nanometer-scale optical metal has been manipulated again by high-energy methods such as pulsed lasers,12,13 applied DC fields,14 and heating.15 And interestingly, despite the immobility of a solid substrate support, there is also evidence of nanoparticle shape change due to illumination by low-power visible light. For silver and gold metal embedded in nanocrystalline TiO2 films, reversible photochromic1,16,17 and photovoltaic fuel cell18,19 behaviors have been observed. For the efficient use of these composite materials in practical devices, a detailed knowledge of the natural processes leading to the advanced optical properties is desirable. However, study of these systems is often hindered by the embedded nature and low concentration of the metal structures. Experimentally, standard surface microscopy methods are inapplicable, and * To whom correspondence
[email protected].
should
be
addressed.
E-mail:
electron methods yield little. In addition, these systems present new challenges to theory and analysis by computation. Unfortunately, the size of optical metal structures prohibits atomistic electronic structure calculations, and in addition, there is no clear way to incorporate the surface plasmon in an ab initio method. In this work, we apply methods of classical light scattering to better understand the optical properties of optically active silver metal embedded in nanocrystalline TiO2. Inner details of the chemical mechanism of shape change are not directly afforded by this route; however, via analysis and a careful account of experimental observations, some insight is afforded. The remainder of this paper is organized as follows. Section II briefly discusses the experimental and theoretical background utilized in this work, directing readers to previous publications for full information. Section III follows a linear investigation, reviewing results and the motivation for new calculations along the way. Section IV is a discussion of the impact of these results on the details of metal nanometer-scale shape change in the films, and section V summarizes the main findings. II. Background and Method This work was motivated by recent experiments.1,16-19 Figure 1 presents one example of the measured optical spectra obtained for a photochromic silver/TiO2 composite film. Silver metal is reduced from AgNO3 in solution by means of electrons promoted to the conduction band of TiO2 by UV light. The amount of silver reduced is small; the film is t for all ratios. In the limit of zero core radius, a ) r, the radius of a homogeneous particle. Equation 1, though simple, is sufficient for the present application. A more rigorous geometrical analysis for the average path length inside a concentric spherical shell is of interest, but such future work should be matched with a precise theory for the coefficient A so that it does not act as a fitting parameter. In Figure 5, A equals 1 and a is 3.93, 7.44, and 33.1 nm for increasing t values in panel A and 7.44, 9.12, and 13.7 nm for increasing rc values in panel B. In panel A, the corrections strongly broaden thinner shells with less broadening for thicker shells. However, the influence of the
corrections on the peak height is more complicated, especially for core/shell geometries large enough to support multiple resonances, for example, t ) 30 nm in panel A. In this case, both blue and red resonances are widened very slightly, but the blue peak is shortened and the red peak heightened by the corrections. It is clear that a broadening correction is critical for very small particles35 (diameter of 2 nm) and that they are unnecessary for particles with path lengths similar to the mean free path length. In question is the intermediate path length below which corrections become significant. This question is complicated by the assortment of expressions used for the path length for different particle shapes and vague treatment of the factor A. To our knowledge, the smallest silver particle for which singleparticle spectra are directly compared with (unbroadened) Mie theory has a diameter of 20 nm;36 however, because the scattering cross section magnitude is small for small particles, the measured spectra have an unfortunate signal-to-noise ratio. Previous surveys of metal particle spectra suggest that, for silver, corrections are significant with a radius of
