Plasmon-driven chemical transformation of a secondary amide probed by surface enhanced Raman scattering

We study the molecular transformation of NMSB adsorbed on self-aggregated structures of Ag and Au nanoparticles under plasmonic excitation with 633 nm and 785 nm laser sources. The SEM images in Fig. 1 show the morphological features of the NMSB-coated Ag (A and B) and Au (C) aggregates. The cartoon depiction in Fig. 1D (D1-D2) shows the chemical transformations occurring at the plasmonic nanocavities of Au and Ag. It has been reported before that plasmon-induced processes, and in particular hot-electron induced processes are strongly localized to the plasmonic nanocavities12,30,31. The reason is that the electric field enhancement is significantly higher in these nanocavities than at other parts of the nanoparticle surfaces. These are the locations where the transfer of hot electrons is the most probable. Hence we state that the reactions observed here are taking place in the hot spots formed by the nanoparticles.Fig. 1: SEM images of Ag and Au aggregates and cartoon depictions of the chemical transformations occurring at the plasmonic nanocavities of the aggregates.A, B SEM images of Ag and C that of Au aggregates coated with NMSB molecule deposited on Si-wafer; D cartoon depicting the honey-comb arrangement of the Ag and Au aggregates with NMSB molecules, I (panel D1 – enlarged view) and the subsequent products, II – MBAm and III – MBN (panel D2 – enlarged view) formed in the plasmonic nanocavities upon visible light (633 nm and 785 nm) irradiation.SERS is used as a spectroscopic tool for the in-situ investigation to resolve the chemical transformation in terms of vibrational fingerprint during the reaction course. The overall spectral time evolution of the reaction is demonstrated in the SERS spectra shown in Fig. 2A–C recorded for Ag NP aggregates adsorbed with NMSB under 633 nm laser excitation. SERS spectra at different time intervals and time trajectories for the reaction of NMSB on Au aggregates under 633 nm laser excitations are shown in Supplementary Fig. 1.Fig. 2: Overall time evolution of the transformation of a secondary amide tracked by SERS.A Representative SERS spectra obtained at different time intervals of the molecular transformation processes of NMSB on Ag aggregates (stack plots show the absolute intensities in each case); B Zoom-in SERS spectra to show the spectral change in A for the region 1100–1600 cm−1; C Time-series SERS map recorded for 100 s under 633 nm laser excitation showing the transformation of NMSB molecule; D Reaction time traces of the peaks at 1320 cm−1, 1189 cm−1, and 2230 cm−1 (extracted from time series map C) assigned to the loss of the amide III band, the appearance of the primary amide and the aromatic nitrile, respectively. E Reaction time traces of peaks at 315 cm−1 and 345 cm−1 assigned as NH2 rocking vibration and C-S-Ag angle bending vibration, respectively. The evolution of these bands is attributed to the formation of two products viz., MBAm and MBN. The spectra have been recorded at 633 nm with a laser power of 2.7 mW and an acquisition time of 0.5 s.We considered the formation of different possible reaction products (molecular structures shown in Supplementary Fig. 2), namely Ag-S(C6H4)CONH2, Ag-S(C6H4)CN, Ag-S(C6H4)CHO and Ag-S(C6H4)CO+, which would show characteristic SERS peaks at 1189 cm−1, 2230 cm−1, 1659 cm−1, 1756 cm−1, respectively14,32,33,34.A clear gradual disappearance of the band at 1320 cm−1 with time indicates the loss of the amide III band (secondary amide) during the reaction course33. This is accompanied by the appearance of the consistently rising peak at 1189 cm−1, which is assigned to the NH2 rocking vibration of a primary amide (Fig. 2B), indicating the transformation of NMSB to MBAm14. The additional peak arising at 1412 cm−1 that corresponds to the amide III band of a primary amide further confirms the reaction product32. The reaction is accompanied by the formation of MBN indicated by the peak rise due to nitrile stretching vibration at 2230 cm−1 34, (Fig. 2C, D) reported for the first time in the study.The Raman spectra calculated for different possible reaction products are shown in Supplementary Figs. 3 and 4 considering M9 clusters, M = Ag, Au, and M-S surface binding mode. The experimental SERS data shows peak characteristics of MBAm and MBN in a close match with that obtained from the calculated Raman spectral assignment. Other potential products are Ag-S(C6H4)CHO and Ag-S(C6H4)CO, respectively, which show characteristic peaks at ~1670 cm−1 and ~1770 cm−1 (Supplementary Fig. 3F, G) in the calculated Raman spectra. However, these were not observed in the SERS spectra indicating that such products have not been formed.The overall transformation becomes also apparent by a peak shift from 315 cm−1 to 345 cm−1 during the reaction (Fig. 2E). To interpret the peak shift in the low wavenumber region, we compared the SERS spectra with the calculated Raman spectra obtained by DFT calculations shown in Fig. 3 and Supplementary Fig. 3. The stretching vibration at 311 cm−1 (Fig. 3B) in the calculated Raman spectra considering Ag9 clusters (details in the Methods section) has been assigned to NH2 rocking vibration of MBAm whereas the one at 336 cm−1 (Fig. 3C) corresponds to the C-S-Ag bending vibration in MBN.Fig. 3: Comparison of calculated Raman spectra and experimental SERS spectra.Calculated Raman spectra (250–600 cm−1) (A) NMSB, (B) MBAm and (C) MBN, compared with (D) experimental SERS spectra to support the peak assignment at 315 cm−1 and 345 cm−1 to NH2 rocking vibration and C-S-Ag angle bending vibration of MBAm and MBN, respectively. Calculations were performed at the B3LYP/aug-cc-pVDZ level calculating the molecule adsorbed on an Ag9 cluster (Supplementary Fig. 3).This is in line with the observation made in SERS measurements and we infer the temporal blue shift observed from wavenumbers 315 cm−1 to 345 cm−1 to support the subsequent conversion of the primary product MBAm to MBN during the reaction course35,36. A relative red-shift (of the relevant bands in this study) in the SERS spectra for both Ag and Au could be seen when compared to the calculated Raman spectra in most of the cases. Although the spectral feature in the range 310–345 cm−1 is quite broad, we refer for simplicity to the peak position at 315 cm−1 and 345 cm−1 to highlight the relative shift in the peak position during the reaction. The reaction under study is assumed to be irreversible because the reaction products are likely to diffuse away or even desorb, and hence a returnable reaction is not favorable in this case.As introduced before, plasmon-induced chemical transformations are typically assigned to electron transfer and/or heat-induced processes. To further gain insights into the chemical transformation of NMSB at the plasmonic nanocavities, we also carried out a DEA study on the NMSB molecule in the gas phase using mass spectrometry (see Methods for details). Previous studies on the reactivity of peptides reveal an electron attachment-mediated dissociation pathway via Cα-N cleavage in the presence of a proton source studied by mass spectrometry14. In a similar context, low-energy electron attachment-induced bond dissociation of small peptides has also been reported in the gas phase37. Here, we experimentally observed four anionic fragments formed upon electron attachment to NMSB in the electron energy range from ~0 eV up to 12 eV (Fig. 4 and Supplementary Fig. 5, Supplementary Table 1). At low energies, hydrogen atom dissociation is observed, giving rise to [M–H]– at m/z 166. The peak at low energies is considerably asymmetric, with a maximum at 1.1 eV, another peak is observed at 4.4 eV. The ion yield of [M–H]– is by about two orders of magnitude higher than all other yields. Another fragment anion is observed at m/z 109, [M–C2H4NO]–. The corresponding anion efficiency curve shows ion yield at energies above 4.3 eV, with two peak maxima at 5.7 eV and 8.6 eV. Finally, SH– and S– ions are observed at m/z 33 and 32 at energies above 2.5 and 3.6 eV, respectively. The latter fragments are formed after passing over a [M–SH]…SH– pre-dissociation state; a detailed discussion about these fragment anions can be found in Supplementary Discussion 1. No ion yield indicating the relevant Cα–N bond cleavage was observed within the detection limit of the used apparatus. This would correspond to the [M–CH3]– fragment anion as well as the CH3– fragment anion, for which it should be also noted that the electron affinity of CH3 is very low (0.03 eV at the CCSD(T)/aug-cc-pVTZ//CCSD/aug-cc-pVDZ level), and thus the formation probability of CH3– could be low compared to other products. For the complementary reaction channel leading to [M–CH3]–, we computationally predict a threshold of 1.01 eV.Fig. 4: Dissociative electron attachment to the NMSB molecule.Left: Anion efficiency curves for dissociative electron attachment to the NMSB molecule in the gas phase, producing [M–H]–, [M–C2H4NO]–, SH–, and S– fragments. Right: Proposed ion structure and reaction energies (in eV) of forming the fragments from M + e– as calculated at the CCSD(T)/aug-cc-pVDZ//ωB97XD/aug-cc-pVDZ level (see the SI for further details).Therefore, the DEA studies in the gas phase do not provide direct evidence of Cα–N cleavage of the NMSB molecule to prove the experimentally observed MBAm formation reflected in the SERS spectra, though the gas phase studies probe the action of low-energy electrons with similar kinetic energies like in the condensed phase experiment. It is known that plasmon excitation results in the generation of hot electrons in the range of 0–3 eV. Thus, we ascribe the discrepancy, when going from single molecules in the gas phase to molecules adsorbed on surfaces, to significantly change the DEA characteristics in terms of resonance energies, intensity of fragment ion formation, and/or specific dissociation pathways38. In detail, the main fragmentation reaction observed in the gas phase corresponds to S–H bond dissociation near electron energies of about 1.1 eV (Fig. 4 and Supplementary Fig. 5, Supplementary Table 1). Following a well-known mechanism in DEA39,40, we can propose the capture of the electron into a π* orbital of the aromatic moiety, followed by coupling with the dissociative σ*(S–H) orbital.The second reaction step is not possible to occur in the condensed phase since the sulfur atom is bound to the metal-nanoparticle. Instead, the electronic coupling of the π* state to the dissociative σ*(N–Cα) state seems to become prevalent for condensed NMSB and thus would explain the observed N–Cα bond cleavage, absent in the gas phase experiment40. Other bond cleavages observed in the gas phase (like for example C–S bond) occur only at higher electron energies (Fig. 4) and will not play a considerable role in the plasmon experiment. Following the dissociation of the N–Cα bond, addition of a proton led to the formation of MBAm. Plausibly, water molecules trapped within the plasmonic substrate acted as a proton source albeit its energetically sluggish reaction kinetics (high water oxidative potential). This explains the reaction channels active in inducing the observed chemical transformations (although the detailed stepwise reactivity of the NMSB molecule is out of the scope of this study).Additional mechanistic details can be obtained from SERS data recorded under various conditions. The reaction was also monitored under 785 nm excitation wavelength for Au and Ag NPs which showed a similar reaction trend (SERS spectra at different time intervals shown in Supplementary Figs. 6 and 7. That the reaction is not purely light-induced (i.e. without the involvement of plasmonic nanoparticles) is clear from our experimental observation where no reaction of the NMSB molecule deposited on Si-substrate (without nanoparticles) was observed at 633 nm and 785 nm laser excitation (Fig. 5A, B), indicating the role of plasmon excitation. Further, to understand the role of power and external temperature on the reactivity of NMSB molecule and to comment on the thermal or non-thermal nature of the reaction, we deduce the reaction rate constant for the dissociative transformation of NMSB molecule to MBAm. Considering the fast excitation and relaxation of hot-charge carriers under continuous wave (CW) illumination, the time-average concentration of the hot electrons is considered constant which allows us to consider the process to follow a pseudo-first-order rate law (see ref. 12 and Supplementary Discussion 2 for details) using the equation below:$${{\mathrm{ln}}}\frac{{\left[{{{\rm{NMSB}}}}\right]}_{0}}{\left[{{{\rm{NMSB}}}}\right]}=\frac{{k}_{f}}{1-h}{{\cdot }}{t}^{(1-h)}$$
(1)
where \(\left[{{{\rm{NMSB}}}}\right]\) and \({\left[{{{\rm{NMSB}}}}\right]}_{0}\) represent the concentration of NMSB at time t and initial concentration, respectively. For kinetic calculation, the SERS intensity at a given wavenumber is considered proportional to the molecular concentration at a time “t”. \({k}_{f}\) is the kinetic rate constant and h is a fractal term41. Relation (1) represents a power law function represented by y = a·tb and therefore kf = (1–h)·a = b·a. The time-series spectra obtained for loss of amide III peak at 1320 cm−1 in all measurements are fitted to Eq. (1) to deduce the reaction rate constants for all the kinetic measurements (at different laser power, wavelength, and substrate; Fig. 5C–F). A characteristic plot with the kinetic fit is shown in Supplementary Fig. 8. To further validate the contribution of plasmon in the molecular transformation of NMSB, we investigated the influence of laser power and external temperature on the reaction rate. A photochemical reaction is often marked by a linear increase of the reaction rate with increasing laser power, on the other hand, a thermal pathway could be predicted based on a super-linear dependence of the same42. We observed a near-linear dependence of the reaction rate constant on the laser power (calculated above) in the case of both Ag and Au under 633 nm, however, neither linear nor super-linear dependence of the reaction rate constant was observed in the case of 785 nm laser excitation (Fig. 5C, D). Additionally, hot-charge-driven reactions are often associated with thermal heating due to localized plasmon heating43. Therefore, a direct conclusion on the non-thermal nature of the dissociation reaction could not be deduced from Fig. 5C, D. This will be further substantiated in a later discussion of the study.Fig. 5: Reactivity of NMSB molecule without nanoparticle and reaction rate study of the NMSB molecule with nanoparticle under different laser power and temperature conditions.A Reaction time traces extracted from normal Raman spectra of NMSB (2.5 mM ethanolic solution) deposited on Si-substrate at 1320 cm−1 and 1189 cm−1 recorded under laser excitation 633 nm (7 mW) and (B) 785 nm (39 mW); Plot of reaction rate constant versus increasing laser power on the surface of Ag and Au under (C) 633 nm and (D) 785 nm laser excitation. Temperature dependence of the reaction rate constant on the surface of (E) Au (under 1.5 mW laser power) and (F) Ag (under 0.5 mW laser power) aggregates monitored at 633 nm laser illumination. The error bars represent the standard deviation of the rate constant values from three independent measurements.To be noted, Fig. 5D shows the comparison of three rate constant values calculated for Au aggregates under 785 nm excitation as no reaction could be observed at lower laser powers. Additionally, the reaction rate calculated under 633 nm laser excitation for both Ag and Au (Fig. 5C) is higher than that observed under 785 nm excitation (Fig. 5D). This can be attributed to the usually stronger plasmon absorbance band at 633 nm than at 785 nm for Ag and Au aggregate structures as shown in our previous micro absorbance study on 40 nm particle aggregates12. The UV-vis spectra recorded for AgNP dispersion coated with NMSB molecule (Supplementary Fig. 9) after centrifugal wash also substantiate the fact, although the UV-vis spectral feature reflects the situation for NMSB-treated particles in solution. The UV-vis spectra for NMSB-treated AuNPs dispersion is also shown in Supplementary Fig. 9B. Deconvolution of the UV-vis spectra (Supplementary Fig. 9A) shows a secondary plasmon band centered at 640 nm (approx.) which creates a near resonance with a 633 nm laser excitation than with a 785 nm excitation. This further supports the fact that the reaction rate depends on the availability of hot electrons (which is more favorable in the case of 633 nm than for 785 nm), Additionally, it was shown before that the hot electron generation rate is higher for Ag than Au, which can account for the respectively higher reaction rates44.A schematic depicting the hot-electron attachment-mediated dissociative decomposition of NMSB at the plasmonic interface of Ag is shown in Scheme 1. The LUMO + 8 (Ag9) and LUMO + 7 (Au9) (Supplementary Fig. 10) represent the accessible ligand state molecular orbital for electron attachment respectively. The metal-molecule hybrid state representing either the metal or, ligand state molecular orbital when anchored on the surface of Au and Ag is shown in Supplementary Fig. 10.Scheme 1Schematic depicting the accessible HOMO-LUMO orbitals of the NMSB molecule when complexed with the metal surface (considering Ag9 atoms) and the overall depiction of the hot-electron mediated chemical transformation of NMSB adsorbed on Ag nanostructure.Further, we conclude that the MBAm formed as the main product in this study further undergoes a reaction to generate MBN (although with a low yield, Fig. 2A, D) plausibly through a dehydration reaction channel. (Scheme 2) This is an important observation as it opens up a secondary reaction route for nitrile formation on the plasmonic surface under visible light irradiation.Scheme 2Schematic showing the proposed reaction steps for the formation of the reaction products under laser excitation at 633 nm and 785 nm on the surface of Au and Ag nanostructures.A recent study on the dehydration reactions of a primary amide to nitrile at plasmonic junctions revealed the active and cooperative role of gold surface adatoms and hot electrons at the nanocavities34. With the help of the SERS study, Zhou et al. demonstrated that the gold adatoms at the plasmonic junction can form various metal-molecule complexes, stabilizing the reaction intermediates and facilitating the dehydration reaction. One may also speculate that the plasmonic heat could facilitate the low-yield MBN side product through a dehydration channel. In this context, the effect of external heat on the reactivity and reaction rate of NMSB was studied by monitoring the reaction at seven different temperatures between 25 °C and 85 °C (Fig. 5E, F). A slow rise in the reaction rate with increasing temperature was noted, most significant at 55 °C and 85 °C, for Au and Ag, respectively. However, the error bars in the case of both Au and Ag reflect a significant variation which does not allow us to conclude on a clear effect of temperature on the reaction rate. No significant effect of external heat on the dehydration reaction yield of MBN was observed. This further supports the less-dominant role of heat in the conversion of NMSB to MBAm and MBN. Therefore, we propose the predominance of hot-electron mediated DEA pathways in the chemical transformation of NMSB to MBAm and MBN on the surface of Ag and Au nanoparticles.