A controlled Ag–Au bimetallic nanoshelled microsphere array and its improved surface-enhanced Raman scattering effect

Abstract

We report the synthesis of polystyrene core, and Ag and Au bimetallic shell (PS@Ag_nAu_50−n) composite microsphere arrays with different shell-thickness combinations by a two-step ion-sputtering deposition method based on monolayer colloidal crystal, and studied their surface-enhanced Raman scattering (SERS) properties experimentally and theoretically. The SERS properties of PS@Ag_nAu_50−n composite microsphere arrays were compared with those of single silver or gold shell array structure, and the influence of the thickness of the gold outer shell was studied. The results demonstrated that partially replacing the outer silver shell with a gold layer can improve the SERS activity in addition to good stability. This work can give an optimized method for some complex Ag structures as highly active SERS-substrates, and provide a good solution to the interference caused by substrate impurity. As a demonstration, an experiment on hierarchically rough Ag-substrate has performed.

---

1. Introduction

Since the surface-enhanced Raman scattering (SERS) effect was first discovered in 1974,¹ due to its unique properties including high sensitivity, no aqueous interference, non-destructive and non-preparative samples, SERS spectroscopy has been used as a powerful analytical tool in a wide range of applications, such as ultra-sensitive detection,^2,3 biological sensing,^4–8 explosive^9–11 and environmental contaminant detection.^12–14

At present, the SERS phenomenon is observed primarily on the surface of gold, silver, copper and other noble metals.^15,16 And generally believed that, among them, Ag-based substrates may provide higher electromagnetic enhancement leading to larger enhancement in Raman intensity,¹⁷ but they have intrinsically low chemical stability towards surface oxidation. In contrast, Au-based substrates are long-term stable, though they provide moderate enhancement in SERS experiments. Nowadays, single silver- and gold- based SERS active substrates were studied much, but the composite structures were relatively less.

In order to meet high sensitivity and stability of SERS substrate, the uniform Raman signal and reproducible preparation are also important. Thus, a variety of synthesis approaches have been developed for the fabrication of ordered structured substrate, especially template synthesis method including anodic aluminum oxide template method^18–20 and colloidal crystal template method.^21–26 For example, the vacuum sputtering deposition method based on a colloidal crystal template is a simple and flexible method for preparing SERS active substrates. The morphologies of the as-prepared micro/nanostructures can be controlled by the sizes of the colloidal spheres and the deposition parameters, so that the localized surface plasmon resonance (LSPR) position can be adjusted to achieve an optimized SERS enhancement. Many groups have used this method to fabricate the highly ordered SERS active substrates. For instance, Van Duyne's group prepared nanoparticle array and metal film over colloidal sphere by using nanosphere lithography (NSL) on self-assembly monolayer polystyrene or silica sphere colloidal monolayer template.^24,27,28 Xu et al. fabricated silver nanobowl arrays via thermal evaporation deposition on a self-assembled polystyrene monolayer.²⁹ Zhao's group prepared Ag semishell arrays with controlled size and tunable interparticle distance by combining reactive ion etching and sputtering deposition.³⁰ Such ordered arrays used as SERS substrates show large-area uniform Raman activity.^31,32

In this paper, we put forward a simple and effective strategy to synthesize silver and gold bimetallic composite structure, hoping the composite structure can get good stability, while also keeping high SERS activity. Therefore, a polystyrene-sphere core Ag and Au bimetallic shell (PS@Ag_nAu_50−n) (the subscript indicates the thickness in nanometer scale) composite microsphere array, with Ag as inner shell and Au as outer shell, was obtained based on PS monolayer colloidal crystal template via a two-step ion-sputtering deposition method (the strategy shown in Fig. 1). The activity of these PS@Ag_nAu_50−n composite structures as SERS substrates was evaluated by using rhodamine 6 G (R6G) as probe molecules. The experimental results showed a thin Au layer (no more than 10 nm) leaded to a substantial increase in SERS intensity for adsorbed R6G molecules. This work provides an effective strategy to prepare highly sensitive and reproducible SERS substrate, which may give an optimized method for other complex structures as highly active SERS substrate.

Fig. 1 Schematic illustration of preparation of PS@Ag_nAu_50−n composite microsphere array through using a two-step ion-sputtering deposition method. (a) A PS colloidal crystal monolayer on a glass slide. (b) PS@Ag_n microsphere array after ion-sputtering deposition of Ag on the surface of PS spheres. (c) PS@Ag_nAu_50−n composite microsphere array after further ion-sputtering deposition of Au. The total thickness of silver and gold bimetallic shell is 50 nm, and the single thickness of silver and gold layer is adjusted by sputtering deposition time under the same sputtering deposition current (25 mA).

2. Experimental section

2.1. Preparation of PS@Ag_nAu_50−n composite microsphere array

The composite microsphere arrays were prepared through a two-step ion-sputtering method, and the fabrication process is shown in Fig. 1. First, a large area of uniform PS (500 nm in diameter) monolayer colloidal crystal was prepared on well cleaned glass slide by a gas–liquid–solid interface self-assembly method, as illustrated in detail previously.^33,34 After dried naturally, a silver layer and gold layer were successively coated on it by using ion-sputtering deposition method. Through adjusting the sputtering time, a series of PS@Ag_nAu_50−n (subscript n is a multiple of 5, from 0 to 50 nm) composite microsphere arrays, with the same total thickness (i.e. 50 nm) but different silver and gold layer thickness combinations were thus obtained. The deposition thickness was estimated according to the sputtering deposition rate curve provided by the instruction manual of the sputter coater (Emitech K550X). In all the experiments, the sputter current is 25 mA, accordingly the deposition rate is 7.5 nm min⁻¹ in thickness.

2.2. Preparation of hierarchically rough PS@Ag–Au array

Standing Ag nanoplate-built microsphere array (PS@Ag array) was firstly fabricated by an electrodeposition onto PS colloidal monolayer template, detailed as reported in ref. 26. A certain amount of gold layer was then deposited by ion-sputtering on the PS@Ag array, the hierarchically rough PS@Ag–Au array was thus obtained.

2.3. Characterization

The morphologies of PS@Ag_nAu_50−n composite microsphere array and hierarchically rough Ag-substrate were observed on a field-emission scanning electron microscope (FESEM, Sirion 200 FEG). For transmission electron microscopic (TEM, JEM-2010) examination, the PS@Ag_nAu_50−n arrays were scraped from the substrate and dispersed in ethanol. Optical absorption spectra were recorded on a spectrophotometer (Cary 5E UV/vis-NIR) in the wavelength range from 200 to 800 nm. For Raman spectral examination, the as-prepared samples were immersed into 10⁻⁶ mol L⁻¹ R6G aqueous solution for 30 min, and then picked up, rinsed with deionized water for several times and dried at room temperature. The Raman spectra were recorded on a confocal microprobe Raman spectrometer (Renishaw inVia Reflex), using a laser beam of 633 nm wavelength, 5% of 17 mW total power, 10 s integral time. And five different points on each substrate were selected to detect the R6G molecules.

3. Results and discussion

Based on the strategy as shown in Fig. 1, the PS@Ag_nAu_50−n composite microsphere array was thus obtained via successively ion-sputtering deposition of silver and gold on PS colloidal monolayer. The FESEM and TEM images in Fig. 2 show a typical surface morphology of PS@Ag_nAu_50−n composite microsphere array. From the TEM image, we can see the deposited metal-film is not uniformly distributed on the whole PS sphere surface, where the thickness in upper part (facing regions) is larger than the bottom part (undersides). Since the Ag and Au layer were directly deposited onto the self-assembled PS colloidal monolayer, without any other treatments, it exhibits large-area uniform hexagonally closed packed pattern. In addition, compared with general hierarchically rough SERS substrate, such as nanoflowers, dendrites, and multi-pods, and so on, the surface of the composite microsphere arrays fabricated by ion-sputtering deposition is relatively smooth, (see Fig. S1†) that is advantageous to study the gold and silver combination effect in the contribution of the Raman enhancement. The thickness of silver and gold layer can be controlled by changing the sputtering deposition time under the same deposition rate. In our experiment, the total shell-thickness of silver and gold layer was 50 nm, and the single thickness value of inner silver layer (or outer gold layer) was adjusted in multiple of 5, from 0 to 50 nm. Considering the inhomogeneity of shell thickness, the actual shell thickness was hard to conform, thus the thickness values were estimated by instruction manual.

Fig. 2 (a) FESEM image of PS@Ag₀Au₅₀ microsphere array with 50 nm gold shell on the well-cleaned glass slide; (b) corresponding TEM image.

Fig. 3 is the measured absorption spectra of 500 nm PS monolayer colloidal crystal (the bottom curve) and PS@Ag_nAu_50−n composite microsphere arrays with different Ag and Au thickness combinations (from Ag₅₀Au₀, Ag₄₅Au₅, to Ag₅Au₄₅, Ag₀Au₅₀). For PS monolayer, there exist two bands at ∼510 and ∼622 nm, as reported in the literatures,^35,36 which may be related to the optical diffraction effect on monolayer PS colloidal crystal array. As can be seen from the top curve in Fig. 3, three strong bands of PS@Ag₅₀Au₀ array appear at ∼445, ∼513 and ∼611 nm. The absorption peak at ∼445 nm is coming from the silver resonance absorption. With decreasing the thickness of inner silver layer (or increasing the thickness of outer gold layer) but keeping the total thickness unchanged, the peak at ∼445 nm is gradually broadened and even disappeared while the silver layer thickness is lower than 25 nm.

Fig. 3 The measured absorption spectra of PS@Ag_nAu_50−n composite microsphere arrays (total in 50 nm thickness) with different Au and Ag thickness combinations (correspondingly shown in detail beside the curves) and pure PS colloidal crystal monolayer.

The PS@Ag_nAu_50−n composite microsphere arrays with different Au and Ag thickness combinations was first immersed in 10⁻⁶ mol L⁻¹ R6G solution for 30 min, subsequently rinsed with deionized water for several times to remove the free or physically adsorbed molecules.³⁷ After drying in air, SERS characterization was conducted. Fig. 4 compares the SERS signal intensities of R6G absorbed on PS@Ag_nAu_50−n composite microsphere arrays. As can be seen, though the PS@Ag₀Au₅₀ array and PS@Ag₅₀Au₀ array produce relatively weak SERS signals, the spectra still exhibit well-defined peaks characteristic of the R6G molecule at 611, 773, 1363, 1512, 1653 cm⁻¹,³⁸ exhibiting clean substrate-surface obtained by ion-sputtering deposition. From Fig. 4, it can also be seen the PS@Ag₅₀Au₀ array showed relatively stronger Raman signal than that of PS@Ag₀Au₅₀ array, and all the composite microsphere arrays exhibited higher signal than the PS@Ag₀Au₅₀ array. The Raman intensities of PS@Ag_nAu_50−n composite microsphere arrays with different Au and Ag thickness combinations, at 610 cm⁻¹, 1182 cm⁻¹, 1362 cm⁻¹, which are assigned to C–C–C ring in plane bending, C–H out of plane bending and aromatic C–C stretching^38,39 are listed in Fig. 5, clearly showing the changing tendency. From the intensity tendency at 1362 cm⁻¹, all the composite arrays were enhanced and higher than the PS@Ag₅₀Au₀ array. When the out part of silver layer was partially replaced by gold layer, the Raman intensity of composite substrate is first enhanced and then weakened, with the maximal intensity being 4 times stronger than that of single PS@Ag₅₀Au₀ while the gold thickness is reach to 10 nm. Generally, the SERS effect is related to the substrate surface roughness, molecule absorption quantity and orientation etc.^37,40,41 Here it is hard to know the specific absorption conditions, but we carried out the experiments under the same experimental parameters including immersing duration time, sample size, and so on, making sure the molecule–substrate interaction on this series of PS@Ag_nAu_50−n composite microsphere arrays' surfaces were nearly similar. In addition, because the changing tendency of SERS signals is different from the tendency of Au–Ag composition, the small differences of the R6G density on substrates should not be the main reason in our cases. In a word, the SERS signal intensity is dependent on the Ag and Au thickness ratio of these composite arrays, and there is an optimal combination for the maximal enhancement. Thus, the silver and gold composite array obtained in the present study appears highly promising as remarkably sensitive, simple, reproducible, and low-cost SERS-substrate. And the changing tendency can account for the combination effect of Ag and Au bimetallic shell.

Fig. 4 Raman spectra of 10⁻⁶ mol L⁻¹ R6G on the PS@Ag_nAu_50−n composite microsphere arrays (total in 50 nm thickness) with different Au and Ag thickness combinations (correspondingly shown in detail beside the curves).

Fig. 5 Comparison of Raman intensities at 610 cm⁻¹, 1182 cm⁻¹, 1362 cm⁻¹ on PS@Ag_nAu_50−n composite microsphere arrays (total in 50 nm thickness) with different Au and Ag thickness combinations.

In order to understand the inner physical process, we simulated the absorption spectra and field enhancement of the PS@Ag_nAu_50−n composite microsphere arrays by finite-difference time-domain (FDTD) method. Because it is quite difficult to simulate the sample model by using the real structure in Fig. 2b, in our simulations, the coated metal-film was assumed to be uniformly distributed on the whole PS sphere surface and the shell-thickness is 50 nm, the diameter of PS sphere and the periodicity of array are 500 nm, as used in the experiment. We just want to know the changing tendency or whether an optimal maximum exists by using such simple model. The refractive indexes of Ag and Au are from ref. 42. Firstly, we changed the silver and gold layer thickness combination, and calculated their absorption spectra. The simulation results are plotted in Fig. 6. There are obvious resonant peaks at each spectrum. For example, when the coating layer is entirely silver, we can see three peaks in the spectrum, which are located at the wavelengths of 452 nm, 516 nm and 599 nm, respectively, as obtained in the experiment though with some deviation. Here we primarily discuss the peaks in the right side for each line, which are nearest to the wavelength of the stimulating light (633 nm). As the thickness of Au layer increasing, this peak red-shifts from 599 nm to 609 nm, and the absorption coefficients at the peaks become larger.

Fig. 6 The simulated absorption spectra of PS@Ag_nAu_50−n composite microsphere arrays with different Au and Ag thickness combinations by simulation through finite-difference time-domain (FDTD) method.

It is well known that the SERS enhancement ratio varies when the polarization of the incident light changes. This effect may cause different SERS signal intensities in multiple measurements since the orientation of the array arrangement direction of the microspheres is unknown in the Raman measurement process. This will bring uncertainty to the measurement and descend its stability. Here we more concern the extent of the uncertainty brought by this effect. A 500 nm PS sphere coated by 50 nm Au shell is considered, as an example. And the wavelength of incidence is 633 nm. We calculate the SERS enhancement factor for incidence polarization angle (the angle between electric field E and x-axis) from −30° to 30° using FDTD method. The result is shown in Fig. 7. The SERS enhancement factor is obtained by averaging the biquadrate of the electric field (E⁴/E₀⁴) in a thin layer which is within 1 nm from the metal coating layer. From the Fig. 7, we can find that the enhancement is symmetric about the x-axis, and the minimum and maximum enhancement factor is 289 and 352, respectively. So the enhancement factor is within the range 320.5 ± 31.5, which indicates the error from the polarization of light is about ±9.8%. In fact, monolayer colloidal crystal is of multi-domain structure. In our case, average effect will bring much lower error.

Fig. 7 The SERS enhancement factors versus the polarization angle of incident light.

The simulation result of the SERS enhancement factors for different Ag and Au layer thickness combinations is plotted in Fig. 8. We can see that the field enhancement is 241 when the shell is composed entirely of silver. When gradually increasing the thickness of the gold from 10 nm to 40 nm until the shell is composed entirely of gold, we can see the SERS enhancement factor first increases, and decreases later. The maximum of the SERS enhancement, which is 701, appears when the thickness of the silver and gold layer is 30 nm and 20 nm, respectively. Comparing with the case of pure silver shell, the SERS enhancement factor is 1.9 times larger. We think the main reason for this is that the gold layer can change the dielectric environment which the PS@Ag nanoshell experiences. And with an optimized thickness of the gold layer, the nanoshell can get a maximum SERS enhancement.

Fig. 8 Comparison of SERS enhancement of PS@Ag_nAu_50−n with different thickness combinations obtained by FDTD simulation.

By comparing the theoretical simulation and experimental results of such simple Ag–Au bimetallic shell structure, we can find the changing tendency of SERS intensities is similar though there are some differences, since the theoretical simulation model and the actual situation are not completely consistent. Based on above results, it is clear that, if replacing part of silver outer layer by gold layer, it will not reduce but increase the substrate's SERS activity, and there exists an optimal Au and Ag thickness ratio for the maximal Raman enhancement. This phenomenon should be important for the application of SERS effect.

As is known, many highly active SERS substrates are fabricated based on silver hierarchical structures which usually prepared by chemical methods with adding some surfactants. In the fabrication process of these substrates, some surfactants are easily adsorbed onto the silver's surfaces, which may produce some Raman peaks to interference the identification of target molecules. In addition, the silver-substrate has high chemical activity and weak stability. Based on our above results, if introducing a thin gold layer onto such high active silver-substrate, the SERS activity may be held or even improved, while the interference from impurities could be depressed due to a clean surface from the ion-sputtering deposited gold film. Our further experiments have proven this deduction. A hierarchically rough Ag-substrate was prepared as reported,²⁶ as shown in Fig. 9a, and then surface-coated with different thicknesses of gold layer. Their corresponding SERS patterns are characterized by using R6G as probing molecules. As is shown in Fig. 9b, we can know the pure Ag-substrate not only shows the characteristic Raman peak of R6G molecule, but also some peaks caused by the impurities. However, a surface coating of ∼30 nm gold layer on the Ag-substrate can obviously reduce the interference Raman peaks, and even enhance the SERS activity.

Fig. 9 (a) FESEM image of hierarchically rough Ag-substrate; (b) Raman spectra of 10⁻⁶ mol L⁻¹ R6G on hierarchically rough Ag-substrate, 10 nm-Au and 30 nm-Au surface-coated hierarchically rough Ag-substrate.

4. Conclusion

In summary, the SERS activity of simple PS@Ag_nAu_50−n composite microsphere arrays with different Ag and Au thickness combinations, which fabricated by a two-step ion-sputtering deposition method on PS colloidal monolayer template, was investigated at 633 nm excitation choosing R6G as the probe molecule. The experimental and simulation results demonstrated that, replacing silver outer shell by a certain thickness of gold layer can enhance the stability of SERS substrate and improve the SERS activity, and under a certain ratio of gold and silver thickness can lead to the most enhanced signal. And further experiment on hierarchically rough Ag-substrate, demonstrated that this work can give an optimized method for some complex structures as highly active SERS-substrates, and provide a good solution to the interference caused by substrate impurity.

Acknowledgements

The authors acknowledge the National Basic Research Program of China (973 Program, grant no. 2011CB302103), the financial supports from Natural Science Foundation of China (grant no. 11174286 and 11374303), and provincial Natural Science Foundation of Anhui (grant no. 11040606M62).

References

1. M. Fleischmann, P. J. Hendra and A. J. McQuilla, Chem. Phys. Lett., 1974, 26, 163–166.
2. S. Nie and S. R. Emery, Science, 1997, 275, 1102–1106.
3. K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. Dasari and M. S. Feld, Phys. Rev. Lett., 1997, 78, 1667–1670.
4. R. A. Tripp, R. A. Dluhy and Y. P. Zhao, Nano Today, 2008, 3, 31–37.
5. K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari and M. S. Feld, J. Phys.: Condens. Matter, 2002, 14, R597–R624.
6. S. P. Xu, X. H. Ji, W. Q. Xu, B. Zhao, X. M. Dou, Y. B. Bai and Y. Ozaki, J. Biomed. Opt., 2005, 10, 031112.
7. Z. Luo, K. Chen, D. Lu, H. Han and M. Zou, Microchim. Acta, 2011, 173, 149–156.
8. C. Y. Song, Z. Y. Wang, R. H. Zhang, J. Yang, X. B. Tan and Y. P. Cui, Biosens. Bioelectron., 2009, 25, 826–831.
9. L. B. Yang, L. A. Ma, G. Y. Chen, J. H. Liu and Z. Q. Tian, Chem. - Eur. J., 2010, 16, 12683–12693.
10. L. B. Yang, G. Y. Chen, J. Wang, T. T. Wang, M. Q. Li and J. H. Liu, J. Mater. Chem., 2009, 19, 6849–6856.
11. H. Zhou, Z. Zhang, C. Jiang, G. Guan, K. Zhang, Q. Mei, R. Liu and S. Wang, Anal. Chem., 2011, 83, 6913–6917.
12. L. Guerrini, S. Sanchez-Cortes, V. L. Cruz, S. Martinez, S. Ristori and A. Feis, J. Raman Spectrosc., 2011, 42, 980–985.
13. K. C. Bantz and C. L. Haynes, Vib. Spectrosc., 2009, 50, 29–35.
14. H. B. Tang, G. W. Meng, Q. Huang, Z. Zhang, Z. L. Huang and C. H. Zhu, Adv. Funct. Mater., 2012, 22, 218–224.
15. X. Gong, Y. Bao, C. Qiu and C. Y. Jiang, Chem. Commun., 2012, 48, 7003–7018.
16. B. Sharma, R. R. Frontiera, A. I. Henry, E. Ringe and R. P. Van Duyne, Mater. Today, 2012, 15, 16–25.
17. D. A. Genov, A. K. Sarychev, V. M. Shalaev and A. Wei, Nano Lett., 2004, 4, 153–158.
18. Q. Liao, C. Mu, D. S. Xu, X. C. Ai, J. N. Yao and J. P. Zhang, Langmuir, 2009, 25, 4708–4714.
19. S. J. Lee, Z. Q. Guan, H. X. Xu and M. Moskovits, J. Phys. Chem. C, 2007, 111, 17985–17988.
20. Z. Huang, G. Meng, Q. Huang, Y. Yang, C. Zhu and C. Tang, Adv. Mater., 2010, 22, 4136–4139.
21. S. Mahajan, M. Abdelsalam, Y. Suguwara, S. Cintra, A. Russell, J. Baumberg and P. Bartlett, Phys. Chem. Chem. Phys., 2007, 9, 104–109.
22. X. M. Lin, Y. Cui, Y. H. Xu, B. Ren and Z. Q. Tian, Anal. Bioanal. Chem., 2009, 394, 1729–1745.
23. T. R. Jensen, M. D. Malinsky, C. L. Haynes and R. P. Van Duyne, J. Phys. Chem. B, 2000, 104, 10549–10556.
24. J. C. Hulteen and R. P. Vanduyne, J. Vac. Sci. Technol. A, 1995, 13, 1553–1558.
25. G. T. Duan, W. P. Cai, Y. Y. Luo, Y. Li and Y. Lei, Appl. Phys. Lett., 2006, 89, 181918.
26. G. Liu, W. Cai, L. Kong, G. Duan, Y. Li, J. Wang, G. Zuo and Z. Cheng, J. Mater. Chem., 2012, 22, 3177–3184.
27. M. Litorja, C. L. Haynes, A. J. Haes, T. R. Jensen and R. P. Van Duyne, J. Phys. Chem. B, 2001, 105, 6907–6915.
28. A. D. McFarland, M. A. Young, J. A. Dieringer and R. P. Van Duyne, J. Phys. Chem. B, 2005, 109, 11279–11285.
29. M. J. Xu, N. Lu, H. B. Xu, D. P. Qi, Y. D. Wang and L. F. Chi, Langmuir, 2009, 25, 11216–11220.
30. C. Wang, W. Ruan, N. Ji, W. Ji, S. Lv, C. Zhao and B. Zhao, J. Phys. Chem. C, 2010, 114, 2886–2890.
31. Y. Wang, N. Lu, W. Wang, L. Liu, L. Feng, Z. Zeng, H. Li, W. Xu, Z. Wu and W. Hu, Nano Res., 2013, 6, 159–166.
32. G. Liu, Y. Li, G. Duan, J. Wang, C. Liang and W. Cai, ACS Appl. Mater. Interfaces, 2012, 4, 1–5.
33. J. Wang, G. Duan, G. Liu, Y. Li, Z. Dai, H. Zhang and W. Cai, J. Mater. Chem., 2011, 21, 8816–8821.
34. Z. Dai, Y. Li, G. Duan, L. Jia and W. Cai, ACS Nano, 2012, 6, 6706–6716.
35. Z. C. Lu and M. Zhou, J. Colloid Interface Sci., 2011, 361, 429–435.
36. D. K. Yi, J. H. Lee, J. A. Rogers and U. Paik, Appl. Phys. Lett., 2009, 94, 084104.
37. Y. Huang and D. H. Kim, Nanoscale, 2012, 4, 6312–6317.
38. P. Hildebrandt and M. Stockburger, J. Phys. Chem., 1984, 88, 5935–5944.
39. A. Weiss and G. Haran, J. Phys. Chem. B, 2001, 105, 12348–12354.
40. L. J. Wan, M. Terashima, H. Noda and M. Osawa, J. Phys. Chem. B, 2000, 104, 3563–3569.
41. X. Zhang, Z. Lu, D. Sim, S. Li, Y. Feng, J. Ma, H. Chen, F. Boey, H. H. Hng and Q. Yan, Chem. - Eur. J., 2011, 17, 13386–13390.
42. E. D. Palik, in Handbook of Optical Constants of Solids, ed. E. D. Palik, Academic Press, 1985.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3ra47882c

‡ These authors should be addressed as co-first authors.

---