by: Timothy Kinney and Professor Bruce Johnson
Abstract
One of the most significant developments in the field of Raman spectroscopy was the discovery of Surface-Enhanced Raman scattering (SERS). Raman signals from the early days of SERS were known to be enhanced by a factor up to a million-fold, and more recently by extraordinary factors of up to 1012-14. To understand what this means we should understand how Raman scattering works, and then apply this understanding to the Surface-Enhanced case. SERS can be thought of as at least two kinds of enhancement: an electromagnetic enhancement which is well understood and a chemical enhancement which is not completely understood. Charge-Transfer has been shown to contribute some chemical enhancement, but more mysteries remain. We are working to characterize the wavelength dependence of the Anti-Stokes to Stokes ratio, thus deriving what contributions are made by metal-adsorbate-complex transitions approximately resonant with both the laser and surface plasmon frequencies in the absence of Charge-Transfer effects. Work by this group and by others indicates that anti-Stokes to Stokes ratios can be sensitive indicators of this chemical enhancement mechanism in SERS when it occurs.
Background: Raman Scattering
A new type of light-matter interaction was experimentally verified by C.V. Raman and K.S. Krishnan in 1928, now called Raman scattering.1 In contrast with Rayleigh scattering, it is the scattering of photons with a change in frequency. Though the effect holds for all wavelengths of light, it is easier to consider a monochromatic case. During a Raman event, an incident photon of a particular wavelength and energy interacts with the vibronic modes of a matter system. The term vibronic indicates a combination of electronic states and vibrational states. The energy of the system increases by an amount allowed by vibronic transitions and a photon having lost this exact energy is re-emitted. The molecule thus moves from a lower energy state to a higher energy state, while photons of a higher energy are absorbed and photons of a lower energy are emitted. This is called a Stokes shift, but each vibronic transition also has an Anti-Stokes shift. (See Figure 1.) In the Anti-Stokes case, the matter is already in an excited vibrational state. An incident photon interacts with the excited system, resulting in the emission of a higher energy photon and a loss of energy in the system as it returns to a lower energetic state. This happens less often and is dependent on how many molecules in the system are in an excited state, generally due to thermal interactions. The probability of any Raman event occurring is on the order of one for every million incident photons, but it is highly dependent on the polarizability of the matter, and the frequency of the incident radiation. The ratio of Anti-Stokes events to Stokes events is a useful quantity for observing resonance between vibronic transitions and incident radiation.
Since the advent of laser technology in the mid-20th-century, scientists have had access to collimated beams of nearly monochromatic radiation. From this, Raman spectroscopy was developed to probe the vibronic transitions of matter systems by examining the emitted Raman light with a spectrometer. If the energy of the incident radiation is known, the energy of the emitted radiation can be measured, and the energy gained by the system can be calculated exactly. This energy then corresponds to vibronic modes in the system, offering experimental verification of theoretical calculations. Like Infrared or Fluorescent Spectroscopy, Raman Spectroscopy identifies groups of molecules by their chemical bonds, but also provides additional information. The vibrational modes of matter are easily characterized and studied, and a large body of work dedicated to this field exists. In Raman spectroscopy, vibrational modes are called Raman modes. To display Raman spectra, the wavenumber of the incident radiation (usually a laser) is marked as zero. In what is known as the wavelength shift, the wavenumber of each Raman mode is shifted from zero, corresponding to the energy of that mode. For example, a carbon-carbon stretch occurs at approximately 1600 wavenumber shift. It will always occur at 1600 wavenumber shift, regardless of the frequency of the incident radiation. Changing the incident radiation changes the absolute wavenumber position, but not the relative shift. Note that varying the surrounding environment or the physical state of the matter may shift the Raman modes slightly. Figure 4 shows two Surface-Enhanced Raman spectra of 1-dodecanethiol. Several carbon-carbon modes are visible between 1050 and 1200 wavenumber shift.
Background: Surface-Enhanced Raman Scattering
Due to the rarity of Raman events, the Raman signal is often very weak relative to the intensity of incident radiation. The experimentalist desires very intense incident radiation to be ensured of a strong signal. However, over-heating the sample can lead to unwanted side effects, so a balance must be sought between strength and authenticity of the signal. One method to mitigate this is to place the matter in close proximity with a noble metal, such as gold, silver, or copper. Because the intensity of Raman events is dependent on the polarizability of the matter, the strength of the signal is closely dependent on the number of free electrons in the system. Noble metals have an abundance of free electrons that become polarized in resonance with the incident radiation and the molecules being studied, resulting in an increased intensity of Raman photons. This is called Surface-Enhanced Raman Scattering, or SERS. When light interacts with a noble metal, the free electrons on the surface tend to polarize in resonance with the light, and this collective motion is called a surface plasmon. The peak resonance of the surface plasmon depends on the type of noble metal and its geometry. Understanding surface plasmons and their wavelength dependence is critical to understand surface enhancement.
In the late seventies, electrochemists took spectra of pyridine adsorbed on silver, copper or gold electrodes electrolyte solutions. 3-5 Scanning the potential of the electrode (by comparison with Saturated Calomel Electrode), they demonstrated variations in the Raman signal. It was discovered that oxidation/reduction cycling would dissolve and redeposit monolayers of metal, augmenting the SERS enhancement by two orders of magnitude. This additional enhancement was attributed to molecular roughening of the surface of the electrodes.6 The SERS intensity was found to increase with the number of deposited monolayers for small numbers of layers.7 Allen et al. were able to show that multiple mechanisms contribute specified magnitudes to overall enhancement. They suggested that approximately 3.5 x 101 enhancement could be contributed from a surface-roughness mechanism, and 2 x 103 enhancement could come from roughness-independent mechanisms for pyridine adsorbed on copper electrodes.8
Wavelength Dependence of SERS: Multiple Mechanisms
Pockrand et al. determined that the multiple mechanisms were not limited to substrate activity. They collected Raman spectra of pyridine on evaporated silver films and on silver electrodes, and compared SERS-active films with SERS-inactive films investigating wavelength dependence of Raman intensities. 9 They explicitly noted that excitation profiles could be plotted showing SERS intensities dependent on the laser excitation wavelength, with a single maximum in the visible region for each vibrational mode. (See Figure 2.) Their paper demonstrated that all the excitation profiles they took have similar resonance behavior, but that the curves shift to greater energy of the incident light with the increasing energy of the vibration. The surface plasmon enhancement of silver was qualitatively the same with evaporated silver films and with silver electrodes, ruling out mechanisms solely dependent on the substrate. Regarding the adsorbed molecule, they found that doublet signals were obtained from thick layers of pyridine on SERS-active films; one of the pair correlated to the bulk pyridine and the other to a red-shifted version of the same vibrational mode in the surface pyridine. This corroborates the conclusion that the SERS response for pyridine is dependent on both surface plasmon activity attributed to the silver substrate and excitation energy of the laser. Furthermore, the large enhancement factors from SERS-active films could not be described by local field effects alone, and Pockrand concluded that a chemical contribution must play a role.
The multiple mechanism picture has prevailed over the years, leading to the proposal of various models, more or less dependent on specific data obtained from SERS experiments. In general, it is now well-accepted that a local-field electromagnetic enhancement occurs, which resonates with vibronic modes in the molecule and thus substantially increases the intensity of Raman scattered radiation. This mechanism requires that the molecule’s Raman-shifted vibronic frequencies fall within the spectral width of the substrate surface plasmons, although direct proximity between metal and molecule are not required. 10, 11 It is also well-accepted that this mechanism does not completely describe SERS. For physisorbs/chemisorbs, molecule-specific chemical effects are found in enhancement patterns and intensities. 10 A Charge-Transfer mechanism, requiring direct proximity between metal and molecule, has been suggested to explain these.11-13 Charge-Transfer involves the overlapping of molecular orbitals to provide pathways for the migration of electrons from the metal to the adsorbate, or vice versa. This is thought to provide enhancement by modifying the Raman polarizability tensor, causing increased intensity in Raman scattered radiation, though some scientists argue that the Charge-Transfer mechanism should be regarded as an altered adsorbate complex rather than an enhancement mechanism.11
Early work with electrodes, colloids and island films was insufficient to fully understand the complexity of SERS substrate plasmon resonance activity. Researchers called for experiments with increased control of surface roughness, but technology was lacking. 8 A solution to this problem has been developed at Rice University and involves the fabrication of nanoscale particles, like nanoshells. The study of nanoparticles has led to major advancements in the understanding of SERS, such as Jackson et al. recently investigating the relationship between nanoshell geometry and SERS enhancement. 14 Using nanoshells, a new monodispersed substrate, Jackson et al. showed that, in particular for those modes which the laser is resonant with, the Raman Effect is enhanced proportionately with the density of the nanoshells. 14 In other work by the same authors, it was discovered that SERS intensity was linearly dependent on the density of the nanoshell substrate, proving that the SERS response in their experiments was due to single nanoshell resonance response, not dimer resonances. 15 It is important to note that tuning the laser wavelength to excite single nanoshell resonances or dimer resonances results in data corroborating one theory or the other. This reinforces the notion that wavelength dependence is critical to understanding SERS mechanisms.
Maher et al. specifically investigated wavelength dependence of SERS on multiple substrates by plotting the anti-Stokes to Stokes ratio for p-mercaptoaniline (pMA) at different excitation energies and discovered that asymmetries arise. 16 At a fixed temperature a calculable amount of anti-Stokes emission is usually expected based on the thermal activity of the sample. They used the ratio of anti-Stokes to Stokes intensities as a sensitive indicator of resonant conditions involving both the metal and the adsorbate molecule. They argued that this asymmetry is systematically structured according to underlying resonances that need more than a single wavelength to be seen. One series of data they took for laser line 633 nm produced a higher anti-Stokes to Stokes ratio than expected, and another series using a 514 nm excitation produced a lower ratio than expected. (See Figure 3.) This may correspond to an anti-Stokes emission intensity peak around the 585 nm to 610 nm region. Their data lacked the excitation energy density necessary to determine this with certainty, but they did show that the anti-Stokes to Stokes ratio is a potentially valuable means of mapping the resonant interaction between the probes and the metal.
Advancing Our Understanding
We anticipate that in some cases unexpected resonances (called “hidden” by Maher et al. 16) for the metal-molecule-complex will be observed, even in the absence of Charge-Transfer processes. We are specifically interested in 1-dodecanethiol, though we also intend to do experiments with p-mercaptoaniline, due to the other work that has been conducted at Rice University. 10,16 Initial observations for the 1-dodecanethiol and gold colloids system using a Renishaw Invia Raman Microscope have been made. (See Figure 4.) We are currently building a Tunable Raman Darkfield Microscope system to restrict the region of measurements to those where the gold colloids have deposited in a smooth and dense manner. Using a dye laser, we can tune continuously from 558 nm to 606 nm, providing good coverage of the visible spectrum in the region where gold colloids enhance strongly. Compiling excitation profiles as a function of excitation energy, we will calculate the theoretical thermally-allowed anti-Stokes to Stokes ratio for the sample and then examine deviations from this result. Building on the work of Gibson et al. (see subsequent section of this paper), 10 we may prove or disprove the theory of hidden resonances for the metal-molecule-complex in the absence of Charge-Transfer for pMA attached to gold nanospheres and nanoshells. This work can later be applied to other substrates and molecules to provide data to support the general case.
Theoretical Support
As theoretical support for our work, we turn to Gibson et al. who independently ran theoretical density-matrix calculations for pMA and obtained results corroborating the notion that unexpected resonant behaviors between molecule-metal-complex may play a significant role in Surface Enhancement. 10 Their model was able to obtain anti-Stokes resonant excitation behavior in the absence of Charge-Transfer, supporting our hypothesis that these unexpected resonances occur when the anti-Stokes emission is in resonance with vibronic transitions of the molecule-metal-complex. Gibson et al. suggest that anti-Stokes emissions which are resonant with a metal-adsorbate molecule may provide anti-Stokes intensity peaks which then fall away in intensity on either side of the peak. Their calculations agree with the work of Maher 16 and may explain why his observed anti-Stokes to Stokes ratio is asymmetrical to thermal expectations; lower or higher, depending on laser excitation energy.
Acknowledgements
The author would like to thank Rice University and the Rice Quantum Institute. He also acknowledges funding provided by the National Science Foundation and hopes that Research Experience for Undergraduates programs will continue to be funded for the long-term development of research sciences.
References
1. Long, D.A. “Raman Spectroscopy”, (McGraw-Hill, Great Britain, 1977).
2. Image from Vrije University Brussels, 2007, Dec 18, . Accessed 2008 Mar10. Axis labels added.
3. M. Fleischmann, P.J. Hendra and A.J. McQuillan, Chem. Phys. Letters 26, (1974) 123.
4. U. Wenning, B. Pettinger, H. Wetzel, Chem. Phys. Lett. 70, (1980) 49.
5. J. Billmann, G. Kovacs and A. Otto, Surface Sci. 92, (1980) 153.
6. D.L. Jeanmarie and R.P. Van Duyne, J. Electroanal. Chem. 84, (1977) 1.
7. B. Pettinger and U. Wenning, Chem. Phys. Letters 56, (1978) 253.
8. Craig S. Allen, George C. Schatz, Richard P. Van Duyne, Chem. Phys. Letters 75, (1980) 201.
9. I. Pockrand, J. Billman, A. Otto, J. Chem. Phys. 78 (11): 6384-6390 1983.
10. J.W. Gibson, B.R. Johnson, J. Chem. Phys. 124, 064701 (2006).
11. M. Moskovits, Rev. Mod. Phys. 57 (3): 783 1985.
12. M. Osawa, N. Matsuda, K. Yoshii, I. Uchida, J. Phys. Chem. 1994, 98 12702-12707.
13. J.R. Lombardi, R.L. Birke, T. Lu, J. Xu, J. Chem. Phys. 84 (8) 4174, 1986
14. J.B. Jackson, S.L. Westcott, L.R. Hirsch, J.L. West, N.J. Halas, Appl. Phys. Lett. 82, 257 (2002).
15. J.B. Jackson, N.J. Halas, Proc. Nat. Acad. Sci. USA 101, 17 930 (2004).
16. R.C. Maher, J. Hou, L.F. Cohen, E.C. Le Ru, J.M. Hadfield, J.E. Harvey, P.G. Etchegoin, F.M. Liu, M. Green, R.J.C. Brown, M.J.T. Milton, J. Chem. Phys. 123, 084702 (2005).