- Email: [email protected]
Surface-enhanced femtosecond CARS spectroscopy (SE-CARS) on pyridine
Vibrational Spectroscopy 56 (2011) 9–12
Contents lists available at ScienceDirect
Vibrational Spectroscopy journal homepage: www.elsevier.com/locate/vibspec
Surface-enhanced femtosecond CARS spectroscopy (SE-CARS) on pyridine V. Namboodiri, M. Namboodiri, G.I. Cava Diaz, M. Oppermann, G. Flachenecker, A. Materny ∗ Center of Functional Materials and Nanomolecular Science, Jacobs University Bremen, Campus Ring 1, 28759 Bremen, Germany
a r t i c l e
i n f o
Article history: Received 3 June 2010 Received in revised form 12 August 2010 Accepted 13 August 2010 Available online 27 August 2010 Keywords: Surface enhancement Femtosecond spectroscopy Surface-enhanced Raman spectroscopy (SERS) Coherent anti-Stokes Raman scattering (CARS) Surface-enhanced CARS (SE-CARS)
a b s t r a c t Surface-enhanced Raman scattering (SERS) has become an integral part of spectroscopy. The inelastic scattering process is enhanced by several orders of magnitude when molecules are in close contact to nano-structured coin metals. However, the use of surface enhancement in combination with nonlinear spectroscopy is by far not as common as in linear spectroscopy even though a more drastic effect could be expected. In our work, we report the observations we made from the preliminary studies on surface enhancement mechanisms in combination with coherent anti-Stokes Raman scattering (CARS) using femtosecond laser pulses. Silver colloids were used as enhancement medium. Molecules, which show conventional SERS were selected for the experiments. Femtosecond CARS was performed on these molecular systems in the presence and absence of silver colloids. The scattered CARS signal was collected both in the forward and sideward directions. From the analysis of the results general observations were made about the factors affecting the performance of SE-CARS. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Surface-enhanced Raman Scattering (SERS) has become one of the most widely used techniques since its first observation by Fleischmann et al. in 1974 from the molecule pyridine adsorbed on to roughened silver electrodes. Thereafter, the effect has been demonstrated with a variety of molecules and with a number of metals [1,2]. Most commonly used metals are coin metals, such as silver, gold or copper. Metal surfaces containing several coupled microscopic domains are seen to give the highest enhancement for the Raman spectra. Such surfaces are termed as SERS active systems. Besides the metal surfaces, metal colloids consisting of isolated metal particles of nanometer size in aqueous media are also identified to be good candidates for enhancing the Raman spectra of adsorbed molecules. However, the most intense SERS effect is observed from aggregated metal colloids consisting of large groups of aggregated metal nanoparticles. Even though the use of SERS effect is widespread, the mechanism behind the effect is still not clearly understood. There are two theories describing the enhancement effect. One is the electromagnetic theory (EM theory) of enhancement and the other is the chemical enhancement theory [3]. Results from experiments show that the enhancement factor due to the chemical effect is less than that due to the EM effect. This difference in the enhance-
∗ Corresponding author. Tel.: +49 421 200 3231; fax: +49 421 200 49 3231. E-mail address: [email protected] (A. Materny). 0924-2031/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.vibspec.2010.08.005
ment factors due to EM effect and chemical effect is still a matter of debate. Nonlinear optical processes are also seen to be enhanced when using rough surfaces [4,5]. Since the induced nonlinear polarization depends on higher powers of the electric field, huge enhancement of nonlinear optical effects can also be expected. This is because nonlinear optical effects involve the interaction of multiple laser fields with the matter and if any of the incident fields is within the plasmon resonance, the enhancement of the field results in a huge enhancement of the nonlinear optical effect. This principle can be applied to coherent anti-Stokes Raman scattering (CARS), which is a third order nonlinear optical process involving three incident fields, pump (ωp ), Stokes (ωS ), and probe (ωpr ). Chew et al. theoretically predicted the possibility of surface-enhanced coherent anti-Stokes Raman scattering (SE-CARS) from molecules located near colloidal spheres [6]. The calculated results for benzene based on the electromagnetic theory of enhancement gave maximum enhancement factors of about 1010 . In their calculation, the enhancement factor was shown to depend critically on the excitation profile and fall off sharply with increase in the distance between the metal particle and molecule. Surface enhancement of CARS was observed experimentally by Liang et al. using a nanosecond laser [7]. In their experiment, samples used were organic solvents, such as benzene and toluene adsorbed onto colloidal silver particles prepared in N,Ndimethylformamide. The CARS signal scattered at right angles to the input lasers was detected and it showed enhancement of up to two orders of magnitude. Scanning the pump wavelength (ωp ) over the absorption spectrum of the mixture of silver colloid with
10
V. Namboodiri et al. / Vibrational Spectroscopy 56 (2011) 9–12
Fig. 1. Absorption spectra of the silver colloids prepared using the Lee Meisel Method (dashed) and the Hiramatsu Method (solid).
sample showed that the enhancement of CARS signal occurs for a very narrow range of wavelengths (≈6 nm) around the absorption peak of the mixture. The CARS signal detected in the forward direction was weaker, which was explained to be due to the scattering of the signal by the silver particles. The work described in this paper is an extension of this experiment using femtosecond pulses and is aimed at understanding the mechanism of enhancement of the CARS signal from molecules adsorbed onto colloidal nanoparticles of coin metals, such as silver and gold. The use of femtosecond laser pulses in combination with SE-CARS adds an interesting aspect. The possibility to perform surface-enhanced time-resolved experiments would help to also access molecules present at very low concentrations. Additionally, an investigation of the ultrafast dynamics could help to better understand the SE effect itself. 2. Experimental The sample used for testing the SE-CARS was pyridine due to its well known SERS properties. The chemical and electromagnetic effects responsible for SERS in pyridine has been well investigated and it has been shown that the chemical effect due to the formation of a pyridine/silver charge transfer complex has a strong effect on the SERS intensities shown by pyridine [8–10]. Thus, by choosing pyridine as the test sample, the effect and properties of additional resonances created by the charge transfer states, as predicted by the chemical effect theory, can be investigated. Two previously known methods were used for the preparation of silver colloids.
Fig. 2. Raman spectra of 10−2 M pyridine recorded without (dashed) and with (solid) the addition of silver colloid.
2.1.2. Procedure 2 In this procedure, a quick and more reproducible method reported by Hiramatsu and Osterloh [12] was used for the silver sol preparation. In this method, 21 mg of hydroxylamine hydrochloride was dissolved in 5 ml of distilled water and the solution was then quickly and dropwise added to the solution of 17 mg of silver nitrate dissolved in 90 ml of water, which resulted in Ag-sols in a few seconds. The absorption spectra of the colloids prepared in this way showed a maximum around 409 nm (full line in Fig. 1). 2.2. Raman experiment The silver sols prepared were tested for their SERS activity with pyridine using a Raman microscope setup (backscattering geometry). For the excitation of the Raman scattering the 514 nm output of an Ar-ion laser (Coherent, Inova 308 Series) was used with a power of ≈20 mW. The Raman light was dispersed using a triple spectrometer (T64000, Jobin Yvon, France) in its single mode option using the 1200 grooves/mm diffraction grating and an entrance slit width of 100 m. The spectrometer was equipped with a N2 cooled CCD detector with optimal sensitivity in the visible (green/red) and a chip size of 1024 × 512 pixels. The Raman spectra of pyridine recorded with and without adding silver colloid are shown in Fig. 2. The dashed curve shows the Raman spectra obtained for 10−2 M pyridine without adding any silver sol. Large enhancement (4 orders of magnitude) of the Raman spectra can be seen for the same sample with the addition of silver colloid (solid line in Fig. 2). 2.3. CARS experiment
2.1. Preparation of silver colloids 2.1.1. Procedure 1 This procedure for the preparation of silver sol is based on the well known Lee Meisel Method [11]. In this method, 9 mg of AgNO3 was dissolved in 100 ml of distilled water and the solution was brought to boiling under vigorous stirring. A solution of 1% (w/v) sodium citrate (1 ml) was added drop wise very slowly under dark conditions. The mixture was then kept boiling for 1 h. The Ag-sol prepared by this procedure had a greenish yellow color with an absorption maximum around 437 nm (dashed curve in Fig. 1) and the sol was stable for a few weeks if stored under dark conditions. This method is limited to the production of aqueous solution of silver sols and another major limitation of this method is the low reproducibility of silver sols.
The experimental setup used to realize the CARS scheme was as follows. Femtosecond pulses at a repetition rate of 1 kHz with a center wavelength of around 775 nm were produced by a commercial femtosecond laser system (Clark-MXR Inc., CPA-2010). The pulses had an average energy of 1 mJ per pulse. These pulses were equally split and used to pump two optical parametric amplifiers (OPAs) (TOPAS, Light Conversion). The output pulses of the OPAs were then compressed in double-pass two-prism arrangements, resulting in pulse lengths of approximately 85–95 fs. The output of one of the OPAs was equally split, giving rise to the pump and the probe pulses of the CARS process. The other OPA was used to generate the Stokes pulses. In order to achieve phase matching necessary for the CARS process, a three-dimensional forward geometry of beams (folded BoxCARS) was chosen. The folded BoxCARS arrangement of beams
V. Namboodiri et al. / Vibrational Spectroscopy 56 (2011) 9–12
11
3. Results and discussion
Fig. 3. Detection schemes employed for the SE-CARS experiments. Both the forward directed signal and 90◦ scattered signal are detected. In the femtosecond experiments only the forward directed (phase matched) signal could be observed.
spatially separates the signal for background-free detection. The CARS signal was collected both in the phase matched forward direction and non-phase matched 90◦ sideward direction. The presence of enhanced CARS signal in the sideward direction is ascribed to the relaxation of the CARS phase matching condition due to the optical inhomogeneity of the sample media. This effect is different from the partially coherent anti-Stokes Raman scattering (PCARS) described by Hamaguchi and co-workers [13,14]. The detection part of the experiment is shown in Fig. 3. The lowest concentration of pyridine that gave a detectable CARS signal was 2.5 M. The pump and probe pulses for the CARS process were derived from the same laser and their wavelengths varied from 490 to 550 nm. In each case, the Stokes wavelength was chosen such that the frequency difference between the pump and Stokes lasers was 1000 cm−1 , so as to excite the strongest Raman modes of pyridine at 1002 cm−1 and 1035 cm−1 . All Raman modes within about 1000 ± 300 cm−1 can be easily excited due to the broad spectrum of the femtosecond lasers (FWHM ≈4 nm). For the measurement of the resonant CARS spectra of pyridine, the pump and Stokes lasers were kept overlapped in time while the probe laser was delayed by 1 ps. This was done to avoid the non-resonant contribution to the CARS spectra at time-zero. The CARS spectra obtained from 2.5 M pyridine for different concentrations of silver colloid are shown in Fig. 4.
Fig. 4. Surface-enhanced CARS spectra obtained from pyridine. The CARS spectrum from pyridine alone is shown as solid curve. The other curves are for different concentrations of silver colloids (dashed curves).
The largest enhancement factor observed for SE-CARS was 10 when using a pump wavelength of 550 nm. No enhancement of the CARS spectra was observed for the other wavelengths used. Also, no enhancement effect was observed in the 90◦ detection scheme in contrast to the earlier observations made when using nanosecond laser pulses [7]. The observed enhancement of CARS spectra from pyridine is seen to have a sensitive dependence on a variety of parameters. In the nanosecond experiments a strong dependence of the signal intensity on the excitation wavelength had been observed. The excitation profile of SE-CARS was found to be much narrower than the observed absorption of the metal sol. Additionally, the peak of the excitation profile was red-shifted to the absorption maximum. Therefore, firstly, we have investigated the dependence on the excitation wavelength. In the experiments performed by varying the pump laser from 490 to 550 nm, no enhancement effect was observed except when the pump laser was tuned to approximately 550 nm. The observation of this extremely narrow enhancement profile is in accordance with the earlier findings. The wavelength of the Stokes laser in this case was set to 580 nm corresponding to the strongest Raman modes of pyridine, resulting in a wavelength of the CARS signal centered around 522 nm. These wavelengths are on the red side of the resonance frequency of the surface plasmons observed as maximum in the absorption spectrum (Fig. 1) at approximately 420 nm. In the nanosecond SE-CARS experiment this separation was not so obvious. The conclusion there was that only very specific aggregates of silver or gold with the adsorbed molecules contributed to the enhancement effect. Since in those experiments only the sidewards scattered signal (not phase matched) was considered, the mechanism of the process was still not absolutely clear. The main question remained for the nanosecond results was, whether the signal detected under 90◦ resulted from non-phase matched SE-CARS or just from the regular CARS signal scattered by the metal particles. In the experiments presented here, the phase matched forward CARS signal was investigated. Without surface enhancement, a decrease of forward CARS signal intensity can be expected due to diffuse scattering processes to all sides. Thus the observed increase of forward CARS signal intensity definitely is due to the SE effect. The fact that the enhancement is found only for a very narrow wavelengths range of the exciting lasers supports the assumption that only a minority of the metal aggregates within the distribution of different sizes and shapes gives rise to the enhancement. Since the observation of the SE-CARS strongly depends also on the correct phase matching arrangement of the three lasers pulses (folded BoxCARS geometry), we believe that the macroscopic phase relation between the enhanced fields plays an important role. This is in accordance to the observations reported by Liang et al. The sidewards scattered signal, which itself was off the phase matched direction, only could be produced by using pump, Stokes, and probe lasers in an geometric arrangement predicted by the momentum conservation condition. The obvious enhancement of the CARS signal should depend on the concentration of the silver sol. This could be verified experimentally. Here, a critical dependence on the concentration of the silver colloids was observed. Fig. 4 shows the enhancement obtained for different concentrations of the silver colloid added to pyridine. For a low concentration of silver colloid (e.g. 5%) a small enhancement results. The effect is maximum for a concentration of colloid of approximately 10%. Further addition of colloids resulted in a decrease of the enhancement factor. Starting from a very small concentration of silver sol the increase of the enhancement with addition of silver sol is according to the expectations. The decrease in the enhancement factor for higher colloid concentration can be explained by the increase of scattering of both the lasers and the
12
V. Namboodiri et al. / Vibrational Spectroscopy 56 (2011) 9–12
CARS signal by the large number of colloidal silver particles away from the phase matched forward direction of the CARS process. As was already mentioned in the discussion above, our experiments investigation the wavelengths dependence strongly point to the fact that only specific metal aggregates contribute to the SE-CARS effect. Two further experimental findings support this assumption. The properties of the silver sol (cluster size, shape) can be influenced by the preparation. Also in SERS, the effect of surface enhancement is seen to have very sensitive dependence on the method of synthesis. The principle preparation technique (see Section 2), temperature and light conditions, speed of addition of reagents for the preparation of colloids, etc. affects the SERS activity of the colloids. Unfortunately, even colloids prepared using the same method and parameters not always yield the same efficiency of enhancement. In our experiments a correlation between SERS activity and SE-CARS efficiency could be observed. Therefore, for each silver sol used, firstly Raman spectra were taken, in order to check the quality of the silver sol. It is obvious that not all sols, i.e. cluster sizes and shapes, result in the enhancement of CARS similar to the observations made for SERS. However, at the moment we do not know to what extent the SERS and CARS enhancement capabilities of a certain sol are correlated. Another indication of the strong dependence of the SE-CARS effect on the cluster properties can be found when the dependence on the laser intensity is investigated. The stability of the colloids was seen to be affected by the intensity of the laser pulses. In general, large fluctuations are observed in the CARS signal due to the random motion of the colloidal particles within the laser focus. Therefore, it was required to accumulate the signal over a long time to average out the fluctuation effects. However, exposing the silver colloid to the intense femtosecond laser pulses affected the stability of the colloid, which could be seen from a steady decrease of the signal with time. Increasing the laser intensity was increasing both signal intensity and fluctuation (noise) and at the same time resulted in a steeper decrease of the signal life time. These two counter acting effects pose a major difficulty in the reliable performance of the experiment. After a short interruption of the laser interaction (few seconds) the signal is found to start at the high level again before decaying depending on the laser intensity. Additionally, we have observed that vigorous stirring of the sample first reduced the signal intensity. After few minutes without stirring the signal completely recovers. The destruction of the cluster sizes and shapes needed for the SE-CARS process mechanically or due to laser ablation result in a signal decrease. In the case of the interaction with intense lasers only few seconds are sufficient to replace the damaged metal clusters in the focal range. The stirring obviously breaks up existing aggregates, which are formed again after a short while. Thus, the results of the experiments on the surface enhancement effect of coherent anti-Stokes Raman scattering (CARS) showed that the CARS signal can be enhanced in the presence of colloidal metal particles. However, the enhancement factor observed is much less than that observed for the conventional Raman scattering. While the effect is seen to be sensitively dependent on the method of preparation of the colloids, concentration of the colloids, the wavelengths of the laser pulses involved, and the laser intensity, a considerable improvement of the enhancement factor seems to be unlikely. The enhancement effect was observed only when the phase matching condition of the lasers (folded BoxCARS arrangement) was fulfilled. The exact mechanism of the observed effect is still not understood. Since the surface enhancement will occur for those molecules adsorbed to the statistically distributed silver clusters, which in each case produce locally enhanced electro-
magnetic fields with different geometry and intensity, one would expect at least a partial destruction of the phase relation. This might be the main reason for the fact that the enhancement factor observed is much smaller than that expected making simplified model assumptions. A systematic study of the effect under the described experimental conditions is difficult. In order to overcome the problems due to the random motion of colloidal particles within the laser focus, alternative techniques such as the use of rough metal surfaces has to be taken into consideration. We have also found that the laser intensity is a critical parameter. In order to reduce the laser intensity which affects the stability of the colloids, the use of a high repetition rate laser system would be advantageous. 4. Summary Based on earlier experiments performed with nanosecond laser pulses, we have performed new experiments investigating the surface enhancement of coherent anti-Stokes Raman scattering (CARS). For these experiments we have used femtosecond laser pulses, enabling us to also study dynamical aspects, not discussed here. The experiments confirmed that surface enhancement of CARS is possible. While the conventional SERS experiments involving a single laser beam gave enhancement factors of up to four orders of magnitude, the enhancement factor observed for the SECARS process, which involves the interaction of three laser beams, was just a factor of 10. This is mainly due to the increase of diffuse scattering of laser and signal light for increasing silver sol concentration. Interestingly, the SE-CARS effect requires a geometric arrangement of the lasers according to the phase matching condition. A very narrow excitation profile, which was found to be red-shifted relative to the absorption peak of the silver sol, points to the fact that only few silver clusters have the right shape and size to fulfill the requirements of the nonlinear coherent Raman scattering. A destruction of these clusters by the interaction with intense laser pulses as well as due to e.g. strong stirring therefore also results in the decrease or vanishing of the surface enhancement. In future experiments, we plan to use a micro-CARS arrangement, where due to the strong focusing the phase matching condition is relaxed. The use of structured metal surfaces as SE substrates would then contribute to a better characterization of the SE-CARS process. Acknowledgments The authors thank Patrice Donfack, Dr. Animesh Kumar Ojha and Rasha Hassanein for their helpful discussions. Financial support from Jacobs University Bremen gGmbH is gratefully acknowledged. References [1] K. Kneipp, M. Moskovits, H. Kneipp, Surface-Enhanced Raman Scattering, Springer, Berlin, 2006. [2] W.E. Smith, Chem. Soc. Rev. 37 (2008) 955. [3] M. Moskovits, Rev. Mod. Phys. 57 (1985) 783. [4] V.M. Shalaev, Nonlinear Optiics of Random Media, Springer, Berlin, 2000. [5] T. Ichimura, N. Hayazawa, M. Hashimoto, Y. Inouye, S. Kawata, Phys. Rev. Lett. 92 (2004) 22. [6] H. Chew, D.-S. Wang, M. Kerker, J. Opt. Soc. Am. B 1 (1984) 56. [7] A.J. Liang, A. Weippert, J.M. Funk, A. Materny, W. Kiefer, Chem. Phys. Lett. 227 (1994) 115. [8] H. Seki, J. Vac. Sci. Technol. 18 (1981) 633. [9] M. Muniz-Miranda, G. Cardini, V. Schettino, Theor. Chem. Acc. 111 (2004) 264. [10] G. Cardini, M. Muniz-Miranda, M. Pagliai, V. Schettino, Theor. Chem. Acc. 117 (2007) 451. [11] P.C. Lee, D. Meisel, J. Phys. Chem. 86 (1982) 3391. [12] H. Hiramatsu, F.E. Osterloh, Chem. Mater. 16 (2004) 2509. [13] T. Ishibashi, H. Hamaguchi, Chem. Phys. Lett. 175 (1990) 543. [14] T. Ishibashi, H. Hamaguchi, J. Chem. Phys. 103 (1995) 1.
Contents lists available at ScienceDirect
Vibrational Spectroscopy journal homepage: www.elsevier.com/locate/vibspec
Surface-enhanced femtosecond CARS spectroscopy (SE-CARS) on pyridine V. Namboodiri, M. Namboodiri, G.I. Cava Diaz, M. Oppermann, G. Flachenecker, A. Materny ∗ Center of Functional Materials and Nanomolecular Science, Jacobs University Bremen, Campus Ring 1, 28759 Bremen, Germany
a r t i c l e
i n f o
Article history: Received 3 June 2010 Received in revised form 12 August 2010 Accepted 13 August 2010 Available online 27 August 2010 Keywords: Surface enhancement Femtosecond spectroscopy Surface-enhanced Raman spectroscopy (SERS) Coherent anti-Stokes Raman scattering (CARS) Surface-enhanced CARS (SE-CARS)
a b s t r a c t Surface-enhanced Raman scattering (SERS) has become an integral part of spectroscopy. The inelastic scattering process is enhanced by several orders of magnitude when molecules are in close contact to nano-structured coin metals. However, the use of surface enhancement in combination with nonlinear spectroscopy is by far not as common as in linear spectroscopy even though a more drastic effect could be expected. In our work, we report the observations we made from the preliminary studies on surface enhancement mechanisms in combination with coherent anti-Stokes Raman scattering (CARS) using femtosecond laser pulses. Silver colloids were used as enhancement medium. Molecules, which show conventional SERS were selected for the experiments. Femtosecond CARS was performed on these molecular systems in the presence and absence of silver colloids. The scattered CARS signal was collected both in the forward and sideward directions. From the analysis of the results general observations were made about the factors affecting the performance of SE-CARS. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Surface-enhanced Raman Scattering (SERS) has become one of the most widely used techniques since its first observation by Fleischmann et al. in 1974 from the molecule pyridine adsorbed on to roughened silver electrodes. Thereafter, the effect has been demonstrated with a variety of molecules and with a number of metals [1,2]. Most commonly used metals are coin metals, such as silver, gold or copper. Metal surfaces containing several coupled microscopic domains are seen to give the highest enhancement for the Raman spectra. Such surfaces are termed as SERS active systems. Besides the metal surfaces, metal colloids consisting of isolated metal particles of nanometer size in aqueous media are also identified to be good candidates for enhancing the Raman spectra of adsorbed molecules. However, the most intense SERS effect is observed from aggregated metal colloids consisting of large groups of aggregated metal nanoparticles. Even though the use of SERS effect is widespread, the mechanism behind the effect is still not clearly understood. There are two theories describing the enhancement effect. One is the electromagnetic theory (EM theory) of enhancement and the other is the chemical enhancement theory [3]. Results from experiments show that the enhancement factor due to the chemical effect is less than that due to the EM effect. This difference in the enhance-
∗ Corresponding author. Tel.: +49 421 200 3231; fax: +49 421 200 49 3231. E-mail address: [email protected] (A. Materny). 0924-2031/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.vibspec.2010.08.005
ment factors due to EM effect and chemical effect is still a matter of debate. Nonlinear optical processes are also seen to be enhanced when using rough surfaces [4,5]. Since the induced nonlinear polarization depends on higher powers of the electric field, huge enhancement of nonlinear optical effects can also be expected. This is because nonlinear optical effects involve the interaction of multiple laser fields with the matter and if any of the incident fields is within the plasmon resonance, the enhancement of the field results in a huge enhancement of the nonlinear optical effect. This principle can be applied to coherent anti-Stokes Raman scattering (CARS), which is a third order nonlinear optical process involving three incident fields, pump (ωp ), Stokes (ωS ), and probe (ωpr ). Chew et al. theoretically predicted the possibility of surface-enhanced coherent anti-Stokes Raman scattering (SE-CARS) from molecules located near colloidal spheres [6]. The calculated results for benzene based on the electromagnetic theory of enhancement gave maximum enhancement factors of about 1010 . In their calculation, the enhancement factor was shown to depend critically on the excitation profile and fall off sharply with increase in the distance between the metal particle and molecule. Surface enhancement of CARS was observed experimentally by Liang et al. using a nanosecond laser [7]. In their experiment, samples used were organic solvents, such as benzene and toluene adsorbed onto colloidal silver particles prepared in N,Ndimethylformamide. The CARS signal scattered at right angles to the input lasers was detected and it showed enhancement of up to two orders of magnitude. Scanning the pump wavelength (ωp ) over the absorption spectrum of the mixture of silver colloid with
10
V. Namboodiri et al. / Vibrational Spectroscopy 56 (2011) 9–12
Fig. 1. Absorption spectra of the silver colloids prepared using the Lee Meisel Method (dashed) and the Hiramatsu Method (solid).
sample showed that the enhancement of CARS signal occurs for a very narrow range of wavelengths (≈6 nm) around the absorption peak of the mixture. The CARS signal detected in the forward direction was weaker, which was explained to be due to the scattering of the signal by the silver particles. The work described in this paper is an extension of this experiment using femtosecond pulses and is aimed at understanding the mechanism of enhancement of the CARS signal from molecules adsorbed onto colloidal nanoparticles of coin metals, such as silver and gold. The use of femtosecond laser pulses in combination with SE-CARS adds an interesting aspect. The possibility to perform surface-enhanced time-resolved experiments would help to also access molecules present at very low concentrations. Additionally, an investigation of the ultrafast dynamics could help to better understand the SE effect itself. 2. Experimental The sample used for testing the SE-CARS was pyridine due to its well known SERS properties. The chemical and electromagnetic effects responsible for SERS in pyridine has been well investigated and it has been shown that the chemical effect due to the formation of a pyridine/silver charge transfer complex has a strong effect on the SERS intensities shown by pyridine [8–10]. Thus, by choosing pyridine as the test sample, the effect and properties of additional resonances created by the charge transfer states, as predicted by the chemical effect theory, can be investigated. Two previously known methods were used for the preparation of silver colloids.
Fig. 2. Raman spectra of 10−2 M pyridine recorded without (dashed) and with (solid) the addition of silver colloid.
2.1.2. Procedure 2 In this procedure, a quick and more reproducible method reported by Hiramatsu and Osterloh [12] was used for the silver sol preparation. In this method, 21 mg of hydroxylamine hydrochloride was dissolved in 5 ml of distilled water and the solution was then quickly and dropwise added to the solution of 17 mg of silver nitrate dissolved in 90 ml of water, which resulted in Ag-sols in a few seconds. The absorption spectra of the colloids prepared in this way showed a maximum around 409 nm (full line in Fig. 1). 2.2. Raman experiment The silver sols prepared were tested for their SERS activity with pyridine using a Raman microscope setup (backscattering geometry). For the excitation of the Raman scattering the 514 nm output of an Ar-ion laser (Coherent, Inova 308 Series) was used with a power of ≈20 mW. The Raman light was dispersed using a triple spectrometer (T64000, Jobin Yvon, France) in its single mode option using the 1200 grooves/mm diffraction grating and an entrance slit width of 100 m. The spectrometer was equipped with a N2 cooled CCD detector with optimal sensitivity in the visible (green/red) and a chip size of 1024 × 512 pixels. The Raman spectra of pyridine recorded with and without adding silver colloid are shown in Fig. 2. The dashed curve shows the Raman spectra obtained for 10−2 M pyridine without adding any silver sol. Large enhancement (4 orders of magnitude) of the Raman spectra can be seen for the same sample with the addition of silver colloid (solid line in Fig. 2). 2.3. CARS experiment
2.1. Preparation of silver colloids 2.1.1. Procedure 1 This procedure for the preparation of silver sol is based on the well known Lee Meisel Method [11]. In this method, 9 mg of AgNO3 was dissolved in 100 ml of distilled water and the solution was brought to boiling under vigorous stirring. A solution of 1% (w/v) sodium citrate (1 ml) was added drop wise very slowly under dark conditions. The mixture was then kept boiling for 1 h. The Ag-sol prepared by this procedure had a greenish yellow color with an absorption maximum around 437 nm (dashed curve in Fig. 1) and the sol was stable for a few weeks if stored under dark conditions. This method is limited to the production of aqueous solution of silver sols and another major limitation of this method is the low reproducibility of silver sols.
The experimental setup used to realize the CARS scheme was as follows. Femtosecond pulses at a repetition rate of 1 kHz with a center wavelength of around 775 nm were produced by a commercial femtosecond laser system (Clark-MXR Inc., CPA-2010). The pulses had an average energy of 1 mJ per pulse. These pulses were equally split and used to pump two optical parametric amplifiers (OPAs) (TOPAS, Light Conversion). The output pulses of the OPAs were then compressed in double-pass two-prism arrangements, resulting in pulse lengths of approximately 85–95 fs. The output of one of the OPAs was equally split, giving rise to the pump and the probe pulses of the CARS process. The other OPA was used to generate the Stokes pulses. In order to achieve phase matching necessary for the CARS process, a three-dimensional forward geometry of beams (folded BoxCARS) was chosen. The folded BoxCARS arrangement of beams
V. Namboodiri et al. / Vibrational Spectroscopy 56 (2011) 9–12
11
3. Results and discussion
Fig. 3. Detection schemes employed for the SE-CARS experiments. Both the forward directed signal and 90◦ scattered signal are detected. In the femtosecond experiments only the forward directed (phase matched) signal could be observed.
spatially separates the signal for background-free detection. The CARS signal was collected both in the phase matched forward direction and non-phase matched 90◦ sideward direction. The presence of enhanced CARS signal in the sideward direction is ascribed to the relaxation of the CARS phase matching condition due to the optical inhomogeneity of the sample media. This effect is different from the partially coherent anti-Stokes Raman scattering (PCARS) described by Hamaguchi and co-workers [13,14]. The detection part of the experiment is shown in Fig. 3. The lowest concentration of pyridine that gave a detectable CARS signal was 2.5 M. The pump and probe pulses for the CARS process were derived from the same laser and their wavelengths varied from 490 to 550 nm. In each case, the Stokes wavelength was chosen such that the frequency difference between the pump and Stokes lasers was 1000 cm−1 , so as to excite the strongest Raman modes of pyridine at 1002 cm−1 and 1035 cm−1 . All Raman modes within about 1000 ± 300 cm−1 can be easily excited due to the broad spectrum of the femtosecond lasers (FWHM ≈4 nm). For the measurement of the resonant CARS spectra of pyridine, the pump and Stokes lasers were kept overlapped in time while the probe laser was delayed by 1 ps. This was done to avoid the non-resonant contribution to the CARS spectra at time-zero. The CARS spectra obtained from 2.5 M pyridine for different concentrations of silver colloid are shown in Fig. 4.
Fig. 4. Surface-enhanced CARS spectra obtained from pyridine. The CARS spectrum from pyridine alone is shown as solid curve. The other curves are for different concentrations of silver colloids (dashed curves).
The largest enhancement factor observed for SE-CARS was 10 when using a pump wavelength of 550 nm. No enhancement of the CARS spectra was observed for the other wavelengths used. Also, no enhancement effect was observed in the 90◦ detection scheme in contrast to the earlier observations made when using nanosecond laser pulses [7]. The observed enhancement of CARS spectra from pyridine is seen to have a sensitive dependence on a variety of parameters. In the nanosecond experiments a strong dependence of the signal intensity on the excitation wavelength had been observed. The excitation profile of SE-CARS was found to be much narrower than the observed absorption of the metal sol. Additionally, the peak of the excitation profile was red-shifted to the absorption maximum. Therefore, firstly, we have investigated the dependence on the excitation wavelength. In the experiments performed by varying the pump laser from 490 to 550 nm, no enhancement effect was observed except when the pump laser was tuned to approximately 550 nm. The observation of this extremely narrow enhancement profile is in accordance with the earlier findings. The wavelength of the Stokes laser in this case was set to 580 nm corresponding to the strongest Raman modes of pyridine, resulting in a wavelength of the CARS signal centered around 522 nm. These wavelengths are on the red side of the resonance frequency of the surface plasmons observed as maximum in the absorption spectrum (Fig. 1) at approximately 420 nm. In the nanosecond SE-CARS experiment this separation was not so obvious. The conclusion there was that only very specific aggregates of silver or gold with the adsorbed molecules contributed to the enhancement effect. Since in those experiments only the sidewards scattered signal (not phase matched) was considered, the mechanism of the process was still not absolutely clear. The main question remained for the nanosecond results was, whether the signal detected under 90◦ resulted from non-phase matched SE-CARS or just from the regular CARS signal scattered by the metal particles. In the experiments presented here, the phase matched forward CARS signal was investigated. Without surface enhancement, a decrease of forward CARS signal intensity can be expected due to diffuse scattering processes to all sides. Thus the observed increase of forward CARS signal intensity definitely is due to the SE effect. The fact that the enhancement is found only for a very narrow wavelengths range of the exciting lasers supports the assumption that only a minority of the metal aggregates within the distribution of different sizes and shapes gives rise to the enhancement. Since the observation of the SE-CARS strongly depends also on the correct phase matching arrangement of the three lasers pulses (folded BoxCARS geometry), we believe that the macroscopic phase relation between the enhanced fields plays an important role. This is in accordance to the observations reported by Liang et al. The sidewards scattered signal, which itself was off the phase matched direction, only could be produced by using pump, Stokes, and probe lasers in an geometric arrangement predicted by the momentum conservation condition. The obvious enhancement of the CARS signal should depend on the concentration of the silver sol. This could be verified experimentally. Here, a critical dependence on the concentration of the silver colloids was observed. Fig. 4 shows the enhancement obtained for different concentrations of the silver colloid added to pyridine. For a low concentration of silver colloid (e.g. 5%) a small enhancement results. The effect is maximum for a concentration of colloid of approximately 10%. Further addition of colloids resulted in a decrease of the enhancement factor. Starting from a very small concentration of silver sol the increase of the enhancement with addition of silver sol is according to the expectations. The decrease in the enhancement factor for higher colloid concentration can be explained by the increase of scattering of both the lasers and the
12
V. Namboodiri et al. / Vibrational Spectroscopy 56 (2011) 9–12
CARS signal by the large number of colloidal silver particles away from the phase matched forward direction of the CARS process. As was already mentioned in the discussion above, our experiments investigation the wavelengths dependence strongly point to the fact that only specific metal aggregates contribute to the SE-CARS effect. Two further experimental findings support this assumption. The properties of the silver sol (cluster size, shape) can be influenced by the preparation. Also in SERS, the effect of surface enhancement is seen to have very sensitive dependence on the method of synthesis. The principle preparation technique (see Section 2), temperature and light conditions, speed of addition of reagents for the preparation of colloids, etc. affects the SERS activity of the colloids. Unfortunately, even colloids prepared using the same method and parameters not always yield the same efficiency of enhancement. In our experiments a correlation between SERS activity and SE-CARS efficiency could be observed. Therefore, for each silver sol used, firstly Raman spectra were taken, in order to check the quality of the silver sol. It is obvious that not all sols, i.e. cluster sizes and shapes, result in the enhancement of CARS similar to the observations made for SERS. However, at the moment we do not know to what extent the SERS and CARS enhancement capabilities of a certain sol are correlated. Another indication of the strong dependence of the SE-CARS effect on the cluster properties can be found when the dependence on the laser intensity is investigated. The stability of the colloids was seen to be affected by the intensity of the laser pulses. In general, large fluctuations are observed in the CARS signal due to the random motion of the colloidal particles within the laser focus. Therefore, it was required to accumulate the signal over a long time to average out the fluctuation effects. However, exposing the silver colloid to the intense femtosecond laser pulses affected the stability of the colloid, which could be seen from a steady decrease of the signal with time. Increasing the laser intensity was increasing both signal intensity and fluctuation (noise) and at the same time resulted in a steeper decrease of the signal life time. These two counter acting effects pose a major difficulty in the reliable performance of the experiment. After a short interruption of the laser interaction (few seconds) the signal is found to start at the high level again before decaying depending on the laser intensity. Additionally, we have observed that vigorous stirring of the sample first reduced the signal intensity. After few minutes without stirring the signal completely recovers. The destruction of the cluster sizes and shapes needed for the SE-CARS process mechanically or due to laser ablation result in a signal decrease. In the case of the interaction with intense lasers only few seconds are sufficient to replace the damaged metal clusters in the focal range. The stirring obviously breaks up existing aggregates, which are formed again after a short while. Thus, the results of the experiments on the surface enhancement effect of coherent anti-Stokes Raman scattering (CARS) showed that the CARS signal can be enhanced in the presence of colloidal metal particles. However, the enhancement factor observed is much less than that observed for the conventional Raman scattering. While the effect is seen to be sensitively dependent on the method of preparation of the colloids, concentration of the colloids, the wavelengths of the laser pulses involved, and the laser intensity, a considerable improvement of the enhancement factor seems to be unlikely. The enhancement effect was observed only when the phase matching condition of the lasers (folded BoxCARS arrangement) was fulfilled. The exact mechanism of the observed effect is still not understood. Since the surface enhancement will occur for those molecules adsorbed to the statistically distributed silver clusters, which in each case produce locally enhanced electro-
magnetic fields with different geometry and intensity, one would expect at least a partial destruction of the phase relation. This might be the main reason for the fact that the enhancement factor observed is much smaller than that expected making simplified model assumptions. A systematic study of the effect under the described experimental conditions is difficult. In order to overcome the problems due to the random motion of colloidal particles within the laser focus, alternative techniques such as the use of rough metal surfaces has to be taken into consideration. We have also found that the laser intensity is a critical parameter. In order to reduce the laser intensity which affects the stability of the colloids, the use of a high repetition rate laser system would be advantageous. 4. Summary Based on earlier experiments performed with nanosecond laser pulses, we have performed new experiments investigating the surface enhancement of coherent anti-Stokes Raman scattering (CARS). For these experiments we have used femtosecond laser pulses, enabling us to also study dynamical aspects, not discussed here. The experiments confirmed that surface enhancement of CARS is possible. While the conventional SERS experiments involving a single laser beam gave enhancement factors of up to four orders of magnitude, the enhancement factor observed for the SECARS process, which involves the interaction of three laser beams, was just a factor of 10. This is mainly due to the increase of diffuse scattering of laser and signal light for increasing silver sol concentration. Interestingly, the SE-CARS effect requires a geometric arrangement of the lasers according to the phase matching condition. A very narrow excitation profile, which was found to be red-shifted relative to the absorption peak of the silver sol, points to the fact that only few silver clusters have the right shape and size to fulfill the requirements of the nonlinear coherent Raman scattering. A destruction of these clusters by the interaction with intense laser pulses as well as due to e.g. strong stirring therefore also results in the decrease or vanishing of the surface enhancement. In future experiments, we plan to use a micro-CARS arrangement, where due to the strong focusing the phase matching condition is relaxed. The use of structured metal surfaces as SE substrates would then contribute to a better characterization of the SE-CARS process. Acknowledgments The authors thank Patrice Donfack, Dr. Animesh Kumar Ojha and Rasha Hassanein for their helpful discussions. Financial support from Jacobs University Bremen gGmbH is gratefully acknowledged. References [1] K. Kneipp, M. Moskovits, H. Kneipp, Surface-Enhanced Raman Scattering, Springer, Berlin, 2006. [2] W.E. Smith, Chem. Soc. Rev. 37 (2008) 955. [3] M. Moskovits, Rev. Mod. Phys. 57 (1985) 783. [4] V.M. Shalaev, Nonlinear Optiics of Random Media, Springer, Berlin, 2000. [5] T. Ichimura, N. Hayazawa, M. Hashimoto, Y. Inouye, S. Kawata, Phys. Rev. Lett. 92 (2004) 22. [6] H. Chew, D.-S. Wang, M. Kerker, J. Opt. Soc. Am. B 1 (1984) 56. [7] A.J. Liang, A. Weippert, J.M. Funk, A. Materny, W. Kiefer, Chem. Phys. Lett. 227 (1994) 115. [8] H. Seki, J. Vac. Sci. Technol. 18 (1981) 633. [9] M. Muniz-Miranda, G. Cardini, V. Schettino, Theor. Chem. Acc. 111 (2004) 264. [10] G. Cardini, M. Muniz-Miranda, M. Pagliai, V. Schettino, Theor. Chem. Acc. 117 (2007) 451. [11] P.C. Lee, D. Meisel, J. Phys. Chem. 86 (1982) 3391. [12] H. Hiramatsu, F.E. Osterloh, Chem. Mater. 16 (2004) 2509. [13] T. Ishibashi, H. Hamaguchi, Chem. Phys. Lett. 175 (1990) 543. [14] T. Ishibashi, H. Hamaguchi, J. Chem. Phys. 103 (1995) 1.