SEARCH

SEARCH BY CITATION

Keywords:

  • polychlorinated biphenyls;
  • silver substrates;
  • SERS;
  • DFT;
  • TD-DFT

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

To better understand experimentally observed surface-enhanced Raman Scattering (SERS) of polychlorinated biphenyls (PCBs) adsorbed on nanoscaled silver substrates, a systematic theoretical study was performed by carrying out density functional theory and time-dependent density functional theory calculations. 2,2′,5,5′-tetrachlorobiphenyl (PCB52) was chosen as a model molecule of PCBs, and Agn (n = 2, 4, 6, and 10) clusters were used to mimic active sites of substrates. Calculated normal Raman spectra of PCB52–Agn (n = 2, 4, 6, and 10) complexes are analogical in profile to that of isolated PCB52 with only slightly enhanced intensity. In contrast, the corresponding SERS spectra calculated at adopted incident light are strongly enhanced, and the calculated enhancement factors are 104 ~ 105. Thus, the experimentally observed SERS phenomenon of PCBs supported on Ag substrates should correspond to the SERS spectra rather than the normal Raman spectra. The dominant enhancement in Raman intensities origins from the charge transfer resonance enhancement between the molecule and clusters. Copyright © 2014 John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

Polychlorinated biphenyls (PCBs) were once widely used in a variety of industrial applications such as in transformers and capacitors as dielectrics, and in pesticides, paints, carbonless copy paper as additives, because of their nonflammability, chemical stability, high boiling point, and electrical insulating properties.[1, 2] Since the rice oil contamination incident in Japan caused by PCBs exposures, the production and use of PCBs have been banned in the 1970s.[3] Afterwards, extensive researches on PCBs found that these chemicals are dangerous to humans and wildlife as they may cause cancer in humans and animals. They damage people's immune, reproductive, nervous, and endocrine systems.[4, 5] Although PCBs have been banned from production and use for many years, they are globally distributed all over the world and threaten human health and the environment.[6, 7] Thus, it is urgent to develop effective PCB detection methods, which is the primary step in sound management and abatement of PCBs.

The commonly used methods for determination of PCBs, such as the most widely used combination technology of high-resolution gas chromatography with mass spectrometry, generally involve complicated and time-consuming sample preparation as well as expensive analysis process, so they cannot meet the need of ecological risk assessment for PCBs.[8-10] Recently, the newly developed surface-enhanced Raman scattering (SERS) method, using noble metal nanostructures as substrates, has attracted much attention because of its good performance on detection of trace amounts of organic pollutants.[11-14] This method has been applied successfully to the trace analysis and detection of PCBs by using nanorod arrays[15, 16] and dendritic nanostructures[17] of Ag as SERS substrates.[15-19]

Although the facts that Ag nanostructures are indicated to be ideal substrates for PCB detections using SERS method, the interaction mechanism between PCBs and silver substrates and the SERS enhancement mechanism are still not well understood, which are of substantial importance for preparing sensitive substrates for the detection of PCBs.

Up to now, two mechanisms, electromagnetic enhancement mechanism (EM) and chemical enhancement mechanism (CM), have been used to explain the huge SERS enhancement for molecules adsorbed on Ag substrates.[20-23] The EM is caused by the strong surface plasmon resonance of the rough metal surface coupled to the incident light, and Raman signals can be enhanced by several orders of magnitude and even up to 12–14 orders of magnitude in single molecular SERS cases.[20, 24] The CM can be attributed to the nonresonant interactions between molecules and metal surfaces and to the charge transfer (CT) resonance enhancement with the excitation wavelength being resonant with metal–molecule CT transitions.[22, 23] Much effort has been spent to better understand the individual enhancement mechanisms and their relative contributions to the overall enhancement. Zhao et al.[25] put forward a model system (Pyridine–Ag20 cluster) for studying SERS mechanisms and studied the chemical enhancement, CT resonance enhancement, and EM in details. And Sun et al. described the EM and CM of SERS for pyridine adsorbed on silver clusters through charge difference density (CDD) method.[22] Recently, the CM has attracted great attention.[25] Part of the reason is that the chemical enhancement can obtain the factor in the order of 105.[26-28] On the other hand, the chemical enhancement reflects the nature of molecules adsorbed on metal substrates, which is essential for selective detection of SERS.

To understand the CM mechanism of PCB molecules adsorbed on silver substrates, we here present a systematic theoretical study by performing density functional theory (DFT) calculations. 2,2′,5,5′-tetrachlorobiphenyl (PCB52) was chosen as a model molecule of PCBs as performed in previous studies,[29, 30] and several small silver clusters Agn (n = 2, 4, 6, and 10) are used to mimic active sites on Ag substrates. Cluster model does not support strong plasmon resonances because of its small size and thus is expected to show only small electromagnetic enhancement,[31, 32] and to not cover the Raman signals resulted from the CM. It is known that active sites on nanostructures consist of only one or few atoms (referred to as adatoms or adclusters).[33] In particular, Peyser-Capadona et al. have shown that neutral and small (two to eight atoms) silver clusters enclosed in a dendrimer or peptide scaffold can lead to huge enhancement in Raman scattering intensities.[34] So in recent years, there are numerous studies to elucidate the mechanism of SERS using neutral and small silver clusters to model the Ag substrate used in experiment. For example, Jensen et al. investigated size-dependence of the enhanced Raman scattering of pyridine using small silver clusters Agn (n = 2–8, 20) by performing DFT and time-dependent (TD)-DFT calculations. They found that the results from small silver clusters are similar to those from the larger particles.[35] So the present calculations used small Ag clusters to be feasible for time-consuming SERS calculations.

Theoretical methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

All calculations in this work were carried out using gaussian 09 program package.[36] The geometries of isolated PCB52 molecule, silver clusters Agn (n = 2, 4, 6, and 10), and the complexes between them,PCB52–Agn (n = 2, 4, 6, and 10), were fully optimized using BP86 functional[37, 38] with the 6-311G(d,p) basis set for C, H, and Cl atoms and LANL2DZ basis set[39] for Ag atom. Optical spectra of PCB52, Agn, and PCB52–Agn complexes were calculated by the TD-DFT method[40] using LC-BP86 functional, which was always used in the optical absorption calculations,[26, 41] with the same basis set adopted in the ground state calculations. Previous investigations indicated that the BP86 functional can reliably describe the frequency information of silver clusters with organic compound molecules.[25, 40, 42] Additionally, to support our choice of pseudopotential and basis set used in the present work, we performed additional calculations using a larger combined basis set, 6-311 + G(d,p)-LANL2TZ(f) [the 6-311 + G(d,p) for C, H, and Cl and the LANL2TZ(f) for Ag]. Taking PCB52-Ag6 as an example, its optimized geometry parameters and calculated normal Raman spectra (NRS) obtained using the 6-311G(d,p)-LanL2DZ basis set and the 6-311 + G(d,p)-LANL2TZ(f) basis set are compared in Figs S1 and S2. It is found that these two basis sets almost lead to the same results, confirming the enough accuracy of the current basis set we used.

The relative Raman intensities (Ii) are obtained from the calculated Raman activities (Si) using the following relationship derived from the intensity theory of Raman scattering.[43]

  • display math(1)

In this equation, ν0 is the exciting frequency in cm−1, νi is the vibrational wavenumber of the ith normal mode, c is the speed of light in vacuum, h and k are Plank and Boltzmann's constants, respectively, and f is a suitably chosen common normalization factor for all peak intensities. The directly calculated Raman activities Si (in unit Å4/amu) are determined by the following formula:

  • display math(2)

where αi and γi are the isotropic and anisotropic polarizabilities, respectively, and Qi is the normal coordinate. The normal modes were assigned by means of visual inspection using the gaussview 5.0.8 program.[44]

Results and discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

In this section, we first discuss the geometries and NRS of PCB52 and PCB52–Agn (n = 2, 4, 6, and 10) complexes, then the optical absorption spectra of PCB52–Agn to find the absorption wavelength with the strongest transition, and finally the SERS of complexes.

Geometries and normal Raman spectra

Fig. 1 shows the calculated geometrical parameters of the molecule and clusters as well as their complexes, where the available experimental and theoretical values are also given for comparison. Experimental and calculated NRS of isolated PCB52 molecule and the vibration modes of several main peaks are illustrated in Fig. 2. Calculated NRS of PCB52–Agn (n = 2, 4, 6, and 10) complexes are given in Fig. 3, and those for silver clusters are shown in Fig. S3.For the silver clusters, the most sable structures proposed in previous researches[45] were used as the initial geometries and then were optimized at the selected level of theory. As indicated in Fig. 1, Agn (n = 2, 4, and 6) clusters prefer planar geometries, while Ag10 cluster resembles two square pyramidal Ag5 structures with their four-membered square-planar rings rotated by 45°. It is found that the calculated Ag-Ag bond bond lengths are in good agreement with previous results. PCB52 is a noncoplanar molecule, where two of the chlorine atoms are in the ortho-positions of the biphenyl moiety, and the dihedral angle between two benzene rings is calculated to be 101.2°. The calculated NRS of PCB52 and the corresponding normal vibration modes for several main peaks are shown in Fig. 2. For comparison, the experimental Raman spectrum of PCB52 obtained using Ocean Optics miniaturized Raman system with semiconductor laser in 785 nm wavelength is also given in Fig. 2. It is found that the calculated result is in good agreement with the experimental one. For example, the calculation gives four strongest Raman peaks at 1576, 1242, 1033, and 676 cm−1, while the corresponding experimental values are at 1587, 1233, 1045, and 678 cm−1. It should be noted that the characteristic peaks of PCB52 occur at high-frequency regions (1000 ~ 1600 cm−1), in contrast to the positions of characteristic Raman peaks of silver clusters (50 ~ 200 cm−1, Fig. S1). This fact implies that Ag substrates used in SERS experiments do not mask characteristic PCB spectrum bands and thus are suitable for serving as the substrates carrying PCBs.

image

Figure 1. Optimized geometries of PCB52 molecule, silver clusters Agn (n = 2, 4, 6, and 10), and the complexes between them. The bond lengths are in Å. The distances given in italics are the experimental or the theoretical results in the study of Tian et al.[46] and Harb et al.[47] Values in parentheses and square brackets are calculated interaction energies (in eV) and amounts of electric charge transferred from PCB52 to the clusters (in e), respectively. This figure is available in colour online at wileyonlinelibrary.com/journal/jrs.

Download figure to PowerPoint

image

Figure 2. Experimental (the blue line in panel) and calculated (the black line) normal Raman spectra of PCB52 and the corresponding normal vibration modes for main peaks. This figure is available in colour online at wileyonlinelibrary.com/journal/jrs.

Download figure to PowerPoint

image

Figure 3. The calculated normal Raman spectra of PCB52–Agn (n = 2, 4, 6, and 10) complexes.

Download figure to PowerPoint

All the most stable complexes between silver clusters and PCB52 are located to be top-coordinate structures with one Ag atom binding to two Cl atoms of PCB52. Calculated Ag–Cl distances are in the range of 2.909–3.792 Å. Compared to bare silver clusters, geometries of silver clusters in the complexes are only slightly changed upon binding to PCB52. For example, the Ag–Ag distance is 2.577 Å in naked Ag2, while it is 2.581 Å in PCB52–Ag2. The small structural change indicates weak interaction between Ag clusters and PCB52 molecule. To quantitatively evaluate the stability of these complexes, we calculated the amount of electric charge transferred (Q) from PCB52 to silver clusters and the interaction energies (ΔE) between them using the following relationship:

  • display math(3)

where EAgn, EPCB52, and EPCB52–Agn are energies of Agn, PCB52, and the complexes, respectively, and the results are listed at the bottom of Fig. 1. Calculated Q and ΔE values are 0.088, 0.119, 0.057, and 0.019 e, and 0.133, 0.266, 0.101, and 0.016 eV for PCB52–Ag2, PCB52–Ag4, PCB52–Ag6, and PCB52–Ag10, respectively. These small Q and ΔE values also indicate a weak interaction between the molecule and clusters. Calculated NRS of the complexes are shown in Fig. 3. The shifts in harmonic frequency and NRS and SERS enhancement factors for selected normal modes of PCB52 and PCB52–Agn are given in Table 1.

Table 1. Shifts (υ) in harmonic frequency (in cm−1), NRS, and SERS enhancement factor for PCB52 and PCB52–Agn (n = 2, 4, 6, and 10) complexes
 PCB52PCB52–Ag2PCB52–Ag4PCB52–Ag6PCB52–Ag10Assignment
  NRSSERSNRSSERSNRSSERSNRSSERS 
  1. EF, enhancement factor.

υ676675620674674621621676621Ring deformation
EF 1.79.4 × 10342.4 × 1041.91.9 × 1042.41.3 × 105
υ103310279941024992102899410311032Ring deformation
EF 1.15.0 × 1030.3921.54.4 × 1030.55.9 × 103
υ107510751073107610761075107510751075Ring breathing
EF 1.61.2 × 1042.03.8 × 1042.35.9 × 1042.14.2 × 103
υ121812181125121912191219112112181218C–H bending
EF 2.34.3 × 1031.52.5 × 1013.51672.51.0 × 105
υ124212431233124412441243123212431244C–H bending
EF 1.22.4 × 1031.43.2 × 1031.24321.46.9 × 103
υ15761575157515751575157615761575Ring stretching
EF 1.02.7 × 1031.18.6 × 1031.23.3 × 1030.7

As depicted in Fig. 3, NRS appearances of complexes are quite similar to the NRS of isolated PCB52 molecule. As listed in Table 1, six important vibrational modes of PCB52 located at 1576, 1242, 1218, 1075, 1033, and 676 cm−1 were compared with the corresponding vibrational modes in NRS of PCB52–Agn. It is found that in complexes, all six modes present slight enhancement, which should be attributed to static chemical mechanism due to lack of any excitation for systems studied. This can be understood according to Formula (2), which indicates that the polarizability is one of decisive factors of Raman intensities. From Table 2, it is clear that all the complexes have larger polarizabilities than PCB52, resulting in NRS enhancement of complexes compared to the isolated PCB52. However, the calculated enhancement factors are smaller than 5 because of the weak interaction of PCB52 with silver clusters. Thus, the calculated NRS cannot explain the experimentally observed remarkable enhancement of SERS with a factor up to 105.

Table 2. Calculated static polarizabilities (in au) of isolated PCB52 and PCB52–Agn (n = 2, 4, 6, and 10) complexes at ground states, respectively
 xxyyzz
  1. PCB, polychlorinated biphenyl.

PCB52237.197174.830128.810
PCB–Ag2398.140292.866190.307
PCB–Ag4435.512489.795246.312
PCB–Ag6656.626543.047271.084
PCB-Ag10855.793527.295451.021

Absorption spectra of PCB52–Agn (n = 2, 4, 6, and 10) complexes

The calculated absorption spectra of PCB52 and the complexes are shown in Figs 4 and 5, and those for silver clusters are given in Fig. S4. For Ag clusters, calculated results agree well with the experimental and theoretical data in the study of Neugebauer et al., Dennington et al., and Idrobo[43-45]. For instance, the calculated energies of the lowest transitions for Ag2, Ag4, Ag6, and Ag10 are 3.12, 3.09, 3.22, and 3.51 eV, which are consistent with the available experimental values of 2.80, 2.97, 3.23, and 3.79 eV, respectively.[45-47] For the isolated PCB52 molecule, all the excitations are below 250 nm, which are in the ultraviolet region. Upon binding to silver clusters, the absorption spectrum present remarkable red shift, and the maximum absorption peaks of the complexes occur in the range of 300 ~ 400 nm. The excitations with the largest oscillator strengths (f) are found to be at 379.64 nm (f = 0.6706), 408.44 nm (f = 0.9267), 331.98 nm (f = 1.4659), and 353.03 nm (f = 1.5698) for complexes of PCB52 with Ag2, Ag4, Ag6, and Ag10 clusters, which correspond to the S1, S3, S8, and S14 transitions, respectively. These wavelength numbers are used as the incident lights for the following SERS calculations.

image

Figure 4. The calculated absorption spectra of PCB52 molecule. The blue lines correspond to several strong absorption peaks. This figure is available in colour online at wileyonlinelibrary.com/journal/jrs.

Download figure to PowerPoint

image

Figure 5. The calculated absorption spectra of PCB52–Agn (n = 2, 4, 6, and 10) complexes. The blue lines correspond to several strong absorption peaks. This figure is available in colour online at wileyonlinelibrary.com/journal/jrs.

Download figure to PowerPoint

The newly developed CDD method was extensively used to visualize CT between target molecules and the metal clusters at different electronic excited states,[22, 23, 26, 48] which provides us a powerful tool to distinguish the CT enhanced mechanism from the EM. Thus, the CDDs of the three complexes at their strongest transition states were calculated and listed in Fig. 6. Clearly, the S1, S3, S8, and S14 states for PCB52–Agn (n = 2, 4, 6, and 10) are CT-excited states (electron transfer from silver clusters to PCB52) in absorption, which are mainly resulted from intermolecular CT, and the holes and electrons are localized on silver clusters and PCB52, respectively.

image

Figure 6. Charge difference densities for different singlet excitation states of PCB52–Agn (n = 2, 4, 6, and 10) complexes (the green and red stand for holes and electrons, respectively, and the isovalue is 4 × 10 − 4 au). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.). This figure is available in colour online at wileyonlinelibrary.com/journal/jrs.

Download figure to PowerPoint

SERS spectra of PCB52–Agn (n = 2, 4, 6, and 10) complexes.

The calculated SERS spectra for PCB52–Agn (n = 2, 4, 6, and 10) complexes are given in Fig. 7. Comparing these SERS spectra with the NRS of the isolated PCB52 shown in Fig. 3, significant enhancements for all complexes are observed, and the total enhancement factor is 103 ~ 105. However, the SERS appearance for these four PCB52-cluster complexes are quite different from each other, implying that the Raman enhancement intensity depends strongly on the size of the cluster; while there is no simple correspondence with cluster size can be found. The similar phenomenon is also found for the pyridine and silver cluster system.[26, 48]

image

Figure 7. Calculated normal Raman spectra and surface-enhanced Raman Scattering of PCB52–Agn (n = 2, 4, 6, and 10) complexes. The wavelengths used to calculate the surface-enhanced Raman Scattering are 379.6, 408.4, 331.9, and 353.5 nm, respectively.

Download figure to PowerPoint

Careful examination of the SERS and NRS of PCB52–Ag2, we can find several differences between them. For instance, the obvious Raman shifts located at 676, 1075, 1218, and 1242 cm−1 in the PCB52–Ag2 NRS are not enhanced enormously in its SERS spectrum. While four vibration modes at 708.5, 1374.5, 1433.1, and 1575.5 cm−1 are found to undergo large enhancement, which are attributed to C–H out-plane deformation, C–H in-plane bending, C–H in-plane deformation, and C–C stretching motions, respectively. The enhanced factors of these vibrational modes exceed 2 × 104, especially with the strongest mode at 1374 cm−1 up to 6.0 × 104. The selective huge enhancement can be attributed to the CM, which is distinctly different from the EM that would lead to the entire vibrational modes enhancing in the same order.

Similarly, for the SERS spectra of PCB52–Ag4 PCB52–Ag6, and PCB52–Ag10 calculated with 408.44, 331.98, and 353.03 nm wavelengths, the vibrational modes are also selectively enhanced relative to their NRS. In the case of SERS spectrum of PCB52–Ag4, three totally symmetric modes at 674.2, 1076.5, and 1575.6 cm−1 are enhanced greatly, which are assigned to the ring deformation, ring breathing vibration, and ring C–C stretching vibration, respectively, and the resonance Raman intensities are increased by factors of 2.0 × 104. The SERS spectrum of PCB52–Ag6 calculated with the incident light of 331.98 nm has the largest enhancement compared with those of PCB52–Ag2, PCB52–Ag4, and PCB52–Ag10, and the largest enhancement occurs at 1075 cm−1, which is ascribed as the ring breathing mode with an enhanced factor up to 1.0 × 105. In the case of PCB52–Ag10, we also observed the selective enhancement of the SERS spectra. It can be seen that there were almost no Raman peaks in the range of 1300 ~ 1800 cm−1 regions in the NRS; however, several peaks can be obviously observed in this region in the SERS spectrum.

As is discussed earlier, the incident light applied to calculate the SERS spectra of all complexes is resonant with the CT excitation states. The CDDs of the PCB52–Agn (n = 2, 4, 6, and 10) complexes at S1, S3, S8, and S14 transition states show that they are all cluster to molecule CT excitations. This means that the wavelengths used are resonance with cluster-molecule CT transitions, which can be ascribed as CT resonance enhancement.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

In this work, the Raman scattering properties of PCB52 adsorbed on small silver clusters are studied by performing DFT calculations at the BP86/6-311g(d,p)-LanL2DZ level. It is found that the appearance of the NRS of complexes PCB52–Agn (n = 2, 4, 6, and 10) is similar to that of the isolated PCB52 molecule, and the slightly enhanced Raman intensity is ascribed to the static chemical enhancement. Calculated electronic excitation spectra using the TD-DFT method at the LC-BP86/6-311g(d,p)-LanL2DZ level show that the largest oscillator strengths are at 379.64, 408.44, 331.98, and 353.03 nm, which were used as incident lights to investigate SERS properties of PCB52-silver complexes. The enhanced factors for the SERS were calculated up to 104 ~ 105 for several vibrational modes. The CDDs methods were applied and provided us a vivid description of the CT transition in the cluster-molecule complexes. The results indicate that the huge enhancement of SERS spectra is mainly caused by the cluster-molecule CT resonance enhancement.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

We gratefully acknowledge the financial support from the National Basic Research Program of China (973 Program 2013CB934301) [Correction made here after initial online publication], the National Natural Science Foundation of China (21273131), and Shandong Provincial Natural Science Foundation, China (Grant No. ZR2013BM019).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Theoretical methods
  5. Results and discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
JRS4421-sup-0001-FigS1-FigS4.docWord document348KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.