Silver island films (SiFs), were annealed at various temperatures from 75 to 250°C, Fig. . The silver plasmon absorption (extinction) spectra typically show a decrease in the ≈420 nm plasmon absorption band, as well as a shift to shorter wavelengths. This effect is more pronounced with increased temperature, where the spectral changes at 75°C, are less dramatic than those at much higher annealing temperatures, c.f. Fig. , panels A and D. Subsequently by ratiometrically plotting the 420 nm absorption band as a function of annealing temperature and time, we can clearly see that the more significant changes to the silver islands are indeed occurring at higher temperatures, Fig. .
Absorption spectra of SiFs after annealing at different temperatures. A − 75°C, B − 121°C, C − 190°C, and D − 250°C.
Time-dependent change in absorption at 420 nm of SiFs after annealing at different temperatures. A − 75°C, B − 121°C, C − 190°C, and D — 250°C.
As well as measuring the plasmon absorption spectra we were able to get additional insights into the size and shape changes of the SiFs using Atomic Force Microscopy (AFM). Figure shows a typical AFM image of a SiF both before and after annealing at 75°C for 10 hr, top and bottom, respectively. The right hand-side inserts show surface height histograms. Interestingly, while only slight changes in plasmon absorption were observed, Fig. , Panel A, the AFM image shows that much larger particles are present after annealing. In addition, by considering many spectra, taken with many samples under different conditions, we also observed the general trend that the spacing between the particles increased, and that the particles became spherical as a function of both time and annealing temperature. A height analysis of the surfaces revealed the structures to increase in size as a function of both annealing temperature and time, Fig. , where the changes in film height when annealed at 75°C are thought to be within the experimental error, i.e., no change in height is likely for a 75°C annealing temperature.
Time-dependent change in height of SiFs after annealing at different temperatures. A − 75°C, B − 121°C, C − 190°C, and D−250°C.
These changes in SiF morphology on surfaces have also been observed by others, [26
] while developing platforms for Surface-Enhanced Raman Scattering [31
]. Similar to previous interpretations, our changes in plasmon extinction, coupled with an insight into the size and shape changes revealed by AFM, can be explained by net surface plasmon changes. While an increased particle size would be expected to result in a red shifted plasmon absorption maximum [34
], the dominant influence here is the decreased interaction of neighboring SiFs, as a result of an increased particle separation, noting that one would expect mass to be conserved on the surface during annealing. The interparticle dipole-dipole interactions are known to determine the width of the extinction bands of SiFs [37
]. Hence, as the SiFs become spherical, larger, and more spaced, the net change in silver plasmon absorption is a blue shifted and decreased absorption maximum. Interestingly for films annealed at 250°C, then a slightly different behavior is observed, whereby the silver plasmon absorption eventually red shifts after initially shifting blue. This is thought to be due to the particles being sufficiently spaced that the dipole-dipole interactions become very weak, the plasmon absorption now dominated by the size and shape of the isolated particles.
The effects of thermally annealing silver colloid films were, however, markedly different. The plasmon absorption spectra both red shifted and increased for temperatures over 190°C, Fig. . Interestingly, the size of the colloids did not significantly change from their initial 30–80 nm diameter size distribution, (data not shown) suggesting that the changes observed in the absorption spectra were dominated again by particle-particle dipole-dipole interactions [37
], and not by size and shape.
Absorption spectra of silver colloids deposited on APS-coated glass slides after thermal annealing at different temperatures.
To investigate the use of both annealed SiFs and colloidal films in metal-enhanced fluorescence, films were coated with fluorescein-labeled Human Serum Albumin (HSA), which passively absorbs onto the surface and is known to form a monolayer coverage [1
]. Figure top and middle shows that enhanced fluorescence can be observed from annealed SiFs and colloidal films, as compared to a control sample, which is a monolayer of labeled-protein-coated glass. By considering the enhancement factor, Fig. bottom, which is the ratio of the fluorescein emission from the annealed silver surfaces as compared to glass, we can see notable differences between annealed SiFs and colloids. Firstly, the enhanced fluorescence from colloidal films is generally larger than for SiFs. This finding is consistent with previous reports from our laboratory [2
] and is in part due to the enhanced electric fields around the colloids, which also can enhance fluorescence due to modifications in a fluorophores’ excitation rate [4
]. Secondly, the enhancements from colloidal films are approximately constant, irrespective of surface thermal annealing, which is in contrast to SiFs, which show an increase of up to 50% in the enhancement. This result suggests the use of preannealed silver island or colloid films as substrates in MEF, which involve modest temperature increases, such as in substrates for hybridization assays [7
], or for use as reusable autoclavable clinical sensing platforms. In addition, Fig. bottom shows similar enhancements for SiFs and colloids after annealing at high temperature (>190°C) for 3 hr. Given that both the plasmon absorption spectra and AFM images show a rearrangement from SiFs to colloids as a function of both temperature and annealing time, coupled with the fact that colloidal films undergo little shape and size change, then the similar fluorescence enhancements observed are considered to be due to there being similar nanostructures present on both the surfaces after annealing. While beyond the scope of our laboratory annealing equipment at this time, we question whether very similar surface structures would eventually be observed after annealing at higher temperatures than were employed in our study, and for longer periods of time? i.e., would the curves in Fig. bottom eventually converge?
Fig. 7 Enhanced fluorescein-HSA emission on annealed SiFs (Top) and silver colloids (Middle), as compared to a glass control sample, which is unannealed. The annealing time was 3 hr. (Bottom)—The enhancement factor (mean of three measurements) as a function (more ...)
In many examples of fluorescence-based sensing it is fluorophore detectability that governs the utility and sensitivity of the sensing approach [21
]. In general, the detectability of a fluorophore is determined by two factors: the extent of background emission from the sample and the photostability of the fluorophore. A highly photostable fluorophore, such as tetramethyl rhodamine [4
] can undergo about 106
excitation-relaxation cycles prior to photobleaching. While this can yield as many as 104
detectable photons per fluorophore it should be realized that the vast majority of fluorophores photodegrade after far fewer excitation-emission event cycles [4
]. Subsequently, we tested the photostability of fluorescein-HSA on the annealed SiFs as compared to an identical unannealed SiF sample as well as a glass control sample. Figure top shows fluorescein-HSA emission as a function of time, excited at 470 nm and observed through a 500 nm long pass filter. The relative intensities of the plots reflect the benefits of using silvered surfaces for MEF, in that more detectable photons can be observed per unit time from the annealed SiFs, as compared to both unannealed SiFs and indeed the glass control surface, where the integrated areas under the plots are proportional to the photon flux from the respective samples. By additionally adjusting the laser power to match the same initial steady-state intensities of the samples, Fig. bottom, we can see that the annealed SiF produces more photostable fluorescein-HSA. This finding is consistent with a reduced fluorescein lifetime, the fluorescein in essence spending less time in an excited state, and therefore is less prone to photodestruction [1
], i.e., it is more photostable.
Photostability of FITC on glass slides and SiFs before and after annealing at 190°C. Top—Same laser power, Bottom—Laser power adjusted to give the same steady-state intensity.
We additionally measured the lifetime of the samples used in the photostability study, Fig. . The intensity decay shows that a much shorter lifetime is evident for fluorescein-HSA as compared to the glass control sample, further supporting our photostability observations.
Typical intensity decay of fluorescein-HSA-coated annealed SiFs (190°C) and glass. IRF—Instrumentation Response Function.
Finally, it is interesting to note that the annealed films yield both increased emission from locally positioned fluorescein, as well as a reduced fluorescein lifetime. As briefly mentioned in the Introduction, this combination of an increased quantum yield, coupled with a decreased lifetime is unusual in fluorescence spectroscopy, and is consistent with a modification of the intrinsic radiative decay rate of the fluorophore, see Eqs. (1
). This forms the basis of our earlier interpretations of metal-enhanced fluorescence [1
Metal-Enhanced Fluorescence: Interpretation in Terms of Radiating Plasmons
Until recently the emission of fluorophores in close proximity to metallic nanostructures was thought to originate solely from the fluorophore, the excited plasmons interacting with the fluorophore and changing its free-space spectral characteristics [1
]. However, recently our interpretation of metal-enhanced fluorescence has changed somewhat [22
] as shown in Fig. , to one whereby excited fluorophores can nonradiatively transfer energy to surface plasmons which in turn, radiate the fluorophores’ photophysical characteristics, in essence the system radiates [22
]. While there is very little experimental evidence to date that themetal-enhanced fluorescence phenomenon is due to fluorophore-coupled radiating plasmons, the data presented here for thermally annealed silver island films, certainly goes some ways to support this interpretation, as is described below.
It is known that surface plasmons can be created by illumination of thin continuous metal films under very unique optical conditions, such as through a prism, or a medium of high dielectric constant and with p-polarized light [22
]. However, surface plasmons can also be created by direct illumination of metallic solution based colloids or nanostructures, or even by nanostructures bound to surfaces [22
]. Illumination of nanostructures or colloids typically results in the visualization of strong colors, which is due to a combination of both absorption and scattering [35
]. The term “absorption” is generically used for these nanostructures and colloids, but the correct term is “extinction,” as there are both absorption and scattering components to the observed colors. Based on the recent radiating plasmon model postulated by our laboratories [22
], small colloids are expected to quench fluorescence, because the absorption component of the extinction is dominant over scattering, while larger colloids or nanostructures are expected to result in enhanced fluorescence, as the scattering component is now dominant over the absorption component of the extinction spectra. Intuitively, by considering Eqs. (5
) and (6
) we expect the absorption term CA
to cause quenching, and the scattering term CS
to cause fluorescence enhancement due to fluorophore coupling (described in the next section in more detail). Examination of Eq. (5
) shows that CA
increases as the radius of the nanoparticle cubed, whereas CS
increases as the radius to the sixth power. For this reason, larger nanostructures are expected to show greater fluorescence enhancements than smaller nanoparticles. While this understanding underpins Mie theory [35
] for small spherical particles whose radius is less than 0.05 λ
, the general conditions and equations are still valid for much larger nanostructures, but considerably more complex [35
With regard to the fluorescence enhancements observed here for annealed silver island films, it is thought that the fluorescence enhancement increase, Fig. bottom, is due to the increased size of the nanostructures after annealing, the CS
component of the extinction becoming more dominant than the absorption, CA
, component. Similarly, the annealed colloidal film data also bears out this interpretation, as the size of the surface-bound colloids remained approximately the same after annealing, the Fluorescein-HSA enhancement approximately constant as seen in Fig. bottom. In addition, while the electric field effect or sometimes called the “lightening rod effect” [4
] is known to modify fluorophore absorption cross sections, thereby increasing the excitation rate of fluorophores, this effect is dependent on the close locality of the fluorophore to the nanostructures, and would not be expected to increase with an increase in interparticle spacing, as is observed here. Moreover, a modification in the excitation rate of fluorophores would not alter a fluorophores’ fluorescence lifetime [4
]. To the best of our knowledge this is the first experimental observation that supports the hypothesis of fluorophore-plasmon coupled emission from noncontinuous particulate films.
An Increased Quantum Yield, Decreased Lifetime, and Plasmon-Coupled Emission
While we have rationaled the observations of enhanced fluorescence intensities as a function of SiF annealing in terms of the increased CS
component of the extinction spectra, it is informative to comment on the quantum yield and lifetime changes observed, given that current thinking has slightly shifted from a radiative rate modification, as originally depicted by Eqs. (1
The lifetime of surface plasmons are known to be very short, on the order of tens of femtoseconds [41
]. This suggests that in our radiating plasmon interpretation, the energy transfer is essentially one way, from fluorophore to metal [22
]. The increased quantum yield of fluorophores in close proximity to metallic nanostructures can be understood as the result of rapid energy transfer to the plasmons, which then radiate to the far-field [22
]. When discussing excited fluorophores near to metals, we assume that the near field is present while the fluorophore is in an excited state, i.e., a field around an oscillating dipole at distances closer than the wavelength. For far field radiation, we refer to a wave propagating away from its source, whether fluorophore or metal. For a fluorophore, the far-field wave exists after it releases a photon and returns to the ground state.
The concept of donor-acceptor emission was first described by Forster for donor-acceptor pairs [21
]. However, our laboratories have recently shown experimentally for fluorophores that rapid energy transfer from a donor to an acceptor resulted in an overall increase in the quantum yield of the system
, when the quantum yield of the acceptor is greater than the donor [43
]. This effect occurs because the rate of Forster energy transfer is proportional to the radiative decay rate of the donor and is independent of the nonradiative decay rates. If the transfer rate is high, which is thought to be the case for fluorophore-metal combinations, then the energy is transferred before the donor can decay by the nonradiative pathways, which are inherently the same in the absence and presence of the acceptor.
The effect of transfer to the metal can be described well by the following equations, where the rate of energy transfer to plasmons, or groups of plasmons is provided for by Forster’s theory, and is given by:
= 4 or 3 for a plane or filled half-space of acceptors, respectively, R0
is the well-known Forster distance, r
is the distance from the fluorophore to the acceptor (metal), and τ0
is the lifetime of the donor fluorophores in the absence of acceptors. Recalling the quantum yield and lifetime equations in the absence of metal, Eqs. (3
) and (4
), the efficiency of energy transfer into the plasmons is given by:
As the rate of transfer becomes larger than the inverse lifetime, then the transfer efficiency approaches unity.
The total emission of the fluorophore-metal combination can now be given by:
is the fluorophore emission intensity and IP
is the emission due to the radiating plasmons, QS
is the scattering quantum yield, F
is the absorption coefficient of the fluorophore, and k
is an instrumental constant describing collection efficiencies, etc. From Eq. (9
) we can see that as the transfer efficiency, E
, approaches 100% (unity), the total intensity becomes:
This means that for fluorophore-metal combinations, the effective quantum yield of the fluorophore approaches unity and the overall quantum yield of the system becomes the quantum yield of scattering. Remarkably, this occurs irrespective of whether the fluorophore has a low or high quantum yield. This forms the basis of our recent radiating plasmon model [22
In terms of the annealed SiF data presented in this paper, we see an overall increase in the fluorescence intensity as the size of the nanostructures increases, because the quantum yield of scattering is higher for larger structures than for smaller structures, i.e., competition between absorption (quenching) and scattering, the CA
, c.f. Eqs. (5
) and (6
]. The lifetime of fluorescein in the presence of metal also drops, similar to any donor in the presence of an efficient acceptor as depicted by Forster’s theory [21