Jaspreet
Walia
*ab,
Jean-Michel
Guay
ac,
Oleksiy
Krupin
ad,
Fabio
Variola
b,
Pierre
Berini
acd and
Arnaud
Weck
abc
aCentre for Research in Photonics, University of Ottawa, Ontario, Canada. E-mail: jwalia@uottawa.ca
bDepartment of Mechanical Engineering, University of Ottawa, Ontario, Canada
cDepartment of Physics, University of Ottawa, Ottawa, Ontario, Canada
dSchool of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, Ontario, Canada
First published on 28th November 2017
Plasmon assisted generation of silver sulfate from dodecanethiol is demonstrated on a nano-textured silver substrate with a strong surface plasmon resonance in the visible range. The observed photo-physical processes are attributed to hot charge carriers that are generated from the excitation of surface plasmon resonances using 532 nm laser light. Excited charge carriers are responsible for cleaving the alkane chain, and for generating reactive oxygen species which rapidly photooxidize the exposed sulfur atoms. The ability to drive photochemical reactions with photon energies in the visible range rather than in the UV, on nano-textured silver surfaces, will enable researchers to study photochemical transformations for a wide variety of applications. The strong optical absorbance across the visible range, combined with the fact that the substrates can be fabricated over large areas, naturally makes them candidates for solar driven photochemical applications, and for large scale plasmonic reactors.
More recently, plasmon assisted photochemistry on nano-textured metal surfaces has received considerable attention.6–12 On resonance, very strong electric fields are generated near nano-textured features, and even stronger fields have been reported in gaps between nanostructures.13–16 The metal regions where significant electric field enhancement occurs also contain an elevated density of excited charge carriers, which in turn can be used to drive photo-physical processes in adsorbed molecules. Charge transfer can occur via tunneling processes, where charge carriers are transferred to/from frontier orbitals in the adsorbate. When considering charge transfer dynamics between metallic surfaces and molecules (e.g., a metal–organic interface), self-assembled monolayers (SAM) of alkanethiols on highly conductive substrates represents a good model system. Scanning tunneling microscopy,17–19 and conducting probe force microscopy20 experiments are usually performed to determine the position of the Fermi level in a metal relative to the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO).21 Although these measurements provide a consistent value for the HOMO–LUMO gap in alkanethiols, the position of the Fermi level remains under debate, and therefore it is unknown if tunneling processes are dominated by holes or electrons.
In this study, the formation of silver sulfate (Ag2SO4) via CW visible laser irradiation of a dodecanethiol (HS[CH2]10CH3) layer on nano-textured Ag in air is reported. The surfaces are nano-textured using picosecond laser pulses, which generate spherical nanoparticles of varying sizes and SPR throughout the visible range. Previously, it was shown that by varying the fabrication parameters, many different colours could be produced on silver surfaces.22 In the past, photooxidation of dodecanethiol was observed in samples irradiated by UV photon energies,23–28 and significant progress has been made in connection with UV photo-patterning of alkanethiol monolayers for various biological and chemical applications.29–32 Here, we demonstrate that a 532 nm (2.33 eV) source producing an output power of 0.5 mW can drive the formation of sulfate species from an alkanethiol monolayer formed onto a nano-textured Ag surface that generates strong SPR in the visible. During laser irradiation, acquired Raman spectra clearly reveal the formation of R-SO42− units, purported to be Ag2SO4, consistent with other views.23,33,34 We conclude that hot carriers, efficiently generated within silver nanostructures near the surface due to strong coupling of the incident light to SPR modes thereon, are responsible for the observed photochemical transformations.
The demonstration that chemical reactions on large-area nano-textured silver surfaces can be driven by 0.5 mW of visible laser radiation represents a significant step towards realizing new photochemical reactors for potential industrial use. In this connection, catalytic oxidation reactions are of interest. Namely the partial oxidation of ethylene to form ethylene oxide. Furthermore, UV radiation can be hazardous to work with, and requires special optics and materials to perform experiments, driving up costs. The ability to photooxidize alkanethiols using visible light rather than UV light will now enable researchers to examine photooxidation and photo-patterning of thiols for a wide variety of biochemical applications.
To modify the surfaces of the nano-textured silver, samples were immersed in a 2 mM solution of 1-dodecanethiol, HS[CH2]11CH3 dissolved in IPA for 3 hours. To avoid the formation of multilayer alkanethiols at the surface, the sample was thoroughly washed by placing it in three separate IPA baths for 20 minutes each. After each immersion, the sample was agitated for approximately 2 minutes to allow un-reacted alkanethiol molecules to diffuse out into the IPA solution. All the experiments were carried out in glass (Pyrex) Petri dishes.
To confirm the presence of dodecanethiol on the surface, a grazing angle IR absorbance spectrum was recorded and is shown in Fig. 2. A background spectrum was recorded from a coloured silver surface that was maintained under the same conditions. Emphasis was given to the 3050–2750 cm−1 spectral region, where the strongest IR absorbance bands in dodecanethiol occur and are associated with C–H stretching vibrations. The assignment of specific vibrational modes of various alkanethiols in this spectral region has been reported previously.37–40 There are nine vibrational modes that can potentially be observed, all corresponding to CH vibrations in the methylene chain and methyl headgroup. Spectral deconvolution of the data was performed to accurately determine the center frequency positions of the various peaks, and is also shown in Fig. 2. Center frequencies for the symmetric and asymmetric CH stretching vibrations in the methylene groups were assigned values of νs(CH2) = 2848 cm−1 and νa(CH2) = 2917 cm−1. Frequencies related to asymmetric CH vibrations of the methyl terminal headgroup were assigned values of νa(CH3,ip) = 2963 cm−1 and νa(CH3,op) = 2955 cm−1, where ip and op denote in-plane and out-of-plane stretching vibrations.38 The band related the symmetric CH stretching mode in the methyl group is actually split into two modes resulting from Fermi resonance interactions, commonly denoted as νs(CH3,FR), and have previously are observed at 2938 cm−1 and 2879 cm−1 in alkanethiol monolayers.37,38
It is generally accepted that a well-defined SAM of n-alkanethiol chains on a flat surface will form a closely packed structure in the all-trans conformation with all of their alkyl chains forming a well-defined tilt angle, α, away from the surface normal as shown in inset of Fig. 2. The molecule can also be twisted along its molecular axis r′ defined by the angle β. Previous works have shown that RAIRS can be used to estimate α on flat surfaces.40 In RAIRS, a grazing angle of incidence is used, so the generated electric field is mostly perpendicular to the surface, and vibrational modes that have transition dipole moments parallel to the electric field are observed in IR spectra.40 This forms a set of selection rules for molecules which have a preferential orientation. For the case of the n-alkyl thiols in the all-trans conformation, the transition dipole moment of the νs(CH2) mode lies in the molecular plane, and is orthogonal to the molecular plane for an asymmetric stretch. The transition dipole moment for symmetric stretch in the methyl group, νs(CH3), also lies in the molecular plane, oriented along the long axis of the molecule. Correspondingly, for α, β = 0°, only the symmetric CH vibrations in the methyl group would be seen, whereas for α = 90° and β = 0°, only the symmetric CH vibrations in the methylene groups would be observed. In general, the chain tilt and twist angles are determined by fitting experimental results to a model capable of simulating experimental spectra for various values of α and β. However, on nano-textured metal surfaces that enable resonant coupling into plasmonic modes, the electric field may be oriented along several directions depending on the local nanostructure. The high topography of the surface will also influence the local orientation of the molecule. In this connection, there is a breakdown of the selection rules described above and the spectrum plotted in Fig. 2 can be deconvolved into all possible observable modes. The excellent agreement between the measured IR absorbance frequencies and the frequencies reported in the literature for closely packed n-alkyl thiols not only confirms that dodecanethiol is present on the Ag surfaces, but also suggests that the alkyl chains favour a trans conformation.37
Photooxidation of the dodecanethiol layer was initiated using an irradiance of 15 kW cm−2 (0.5 mW power) in a beam focussed through a 20×, 0.4 N.A. objective at a wavelength of 532 nm. The products of the reaction were monitored using Raman spectroscopy using the same laser. In contrast to previous works,23,41,42 Raman scattering signals from C–C stretching modes in the thiol layer were not resolvable. A coloured silver sample that did not undergo the thiolation procedure, was also measured for comparison. Unlike the coloured silver surfaces, smooth bulk silver does not support surface plasmonic resonances, and therefore surface enhanced Raman scattering (SERS) effects cannot be observed, making it very difficult to obtain a measurable signal from any type of species which may be on such a surface. In contrast, the plasmonic nature of the silver surface used in this study generates enhanced local electric fields22,36 which amplify Raman scattering intensities. In keeping with this, previous studies in which Raman spectroscopy was used to study UV induced photooxidation of alkanethiol monolayers required the use of roughened silver substrates.23 It is also instructive to note that quantitative analysis of any products or reactants can be very difficult to achieve due to highly localized and variable field intensities, commonly referred to as hotspots throughout the literature. From a qualitative perspective, Raman scattering signals provide strong evidence for the presence for molecular species of interest.
Raman spectra from the two samples (with and without the thiol layer), after 3 s and 1 h of exposure to the 532 nm laser are plotted in Fig. 3(a) and (b), respectively. During the first acquisition, neither sample generated any detectable Raman scattering signal. However, after 1 hour, clear spectral differences are seen between the samples. On surfaces where no dodecanethiol was present, a Raman spectrum that is relatively featureless is observed, whereas the irradiation of dodecanethiol on the coloured silver surface generates a spectrum containing high intensity narrowband lineshapes, and confirms that spectral differences result from the presence of the alkanethiol layer. Specifically, five new peaks centred at 470, 635, 711, 973, and 1190 cm−1 appear and are easily resolved in the spectrum, and represent the products of alkanethiol photooxidation that has taken place on the surface.23–28 Time evolution Raman spectra over the course of one hour were recorded every minute and are plotted in Fig. 3(c). For the purpose of clarity, twelve spectra are plotted over five-minute intervals. One can see that the photooxidation process begins within the first five minutes of exposure. The most intense feature appearing in the 975 cm−1 region is assigned to the symmetric stretch of a sulfate (SO42−) ion.23,42 The peaks at 470, 635 and 1190 cm−1 are also assigned to Raman scattering from SO42−, and follow the previous assignments in the literature.23
Locally normalized Raman intensities as a function of time for the five peaks associated with scattering from sulphate species is shown in Fig. 4(a) for an incident power of 0.5 mW. All of the peaks essentially grow at the same rate, consistent with the assignment that they are all from the same surface species. The data were then fit to a single exponential, from which the rate was determined for each of the five peaks. The rates were then averaged to calculate a rate for a given incident power. This calculation was repeated for incident laser powers of 0.75, 1, 1.5, 2 and 2.5 mW and is plotted on a logarithmic scale in Fig. 4(b). The slope of this plots is ∼1.5 and suggests a strong non-linear dependence on light intensity. Generally, for a chemical reaction whose rate depends on temperature, a strong non-linear dependence of this type is expected.43 Both thermal and hot carrier induced processes can generate various power laws.
Previous studies have attributed photooxidation of the S–C bond in alkanethiols on Ag to hot electron excitation.23 In our study, the same chemical transformation is observed, but using a photon energy of only 2.33 eV. However, in contrast to previous literature on the subject, our surfaces consist of nano-textured silver supporting strongly localized plasmons in the visible, so care must be taken to understand their role in a photochemical reaction. Plasmonic excitations can decay non-radiatively by giving up their energy to carriers in the metal. These carriers can then be emitted into an adsorbate, thereby inducing a chemical reaction, or they can thermalize (heating of the lattice) with the resulting heat inducing a chemical reaction.
The silver surface used in this work exhibits a very strong optical absorptance in the visible which may lead to significant local heating of the sample surface. If indeed the observed chemical changes are driven by thermal effects, then identical chemical transformations should be observed when thermal energy only is supplied to the sample via resistive heating.6 The results of such an experiment indicate that this is not the case. A sample was placed on a heating stage and raised to various temperatures ranging from 40 to 160 °C in 30 °C increments. Raman spectra were acquired at each temperature after 40 minutes of heating. Up to temperatures of 130 °C, all recorded Raman spectra were featureless, showing no evidence of sulfur compounds. Subjecting the sample to elevated temperatures did however significantly alter its colour. No colour changes were observed when samples were irradiated with the laser. It is instructive to note however, that the onset of thermal damage was found to occur when the irradiance was increased to 45–60 kW cm−2. Based on these observations, the energy supplied for the observed S–C bond cleavage and subsequent photooxidation of the exposed sulfur atom is not thermal. This is somewhat expected since the nano-textured Ag surface is effectively in direct contact with a relatively large high purity (99.99%) Ag substrate which has a very high thermal conductivity of 429 W m−1 K−1.44
Ruling out thermally-driven reactions, we now consider photochemistry involving charge carriers, specifically, what their role might be in cleaving the S–C bond in adsorbed dodecanethiol molecules. In understanding photochemistry on metal surfaces, the energies of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of the adsorbate molecule relative to the Fermi level of metallic surface is of central importance. In addition, if the metallic surface is nano-textured or consists of nanoparticles, then knowledge of LSPR frequencies and the corresponding enhancements in local electric field intensities are also required. Generally, when molecules adsorb onto metal surfaces, transition energies between the HUMO and LUMO level in the adsorbed molecule can be either blue- or red-shifted from their gas phase energies by as much as 1 eV. Lastly, these orbitals experience broadening when the metal–adsorbate complex forms and no longer consist of discrete energies. The HOMO–LUMO gap for adsorbed molecules on metal surfaces can be determined using scanning tunnelling microscopy (STM). In particular, the HOMO–LUMO gap for dodecanethiol has been reported to lie between ∼7 to 8 eV.21 The Fermi level of Ag is 4.3 eV and lies between the HOMO and LUMO levels of the adsorbed molecule.
Although the experimentally-measured HOMO–LUMO gap energies appearing in the literature for alkanethiol monolayers on gold and silver are consistent, the position of the Fermi energy with respect to these orbitals remains an open question. For example, conducting-probe atomic force microscopy (CP-AFM) measurements have shown that higher work function metals, such as Pt, decrease the contact resistance, and therefore barrier heights to tunnelling, compared to Au and Ag. It was suggested that such trends are indicative that the Fermi level lies closer to the HOMO, and that charge transport is dominated by hole tunnelling.45 On the other hand, the results of ultraviolet photoelectron spectroscopy (UPS) and inverse photoemission spectroscopy (IPES) experiments suggest that the main mechanism is electron tunneling and therefore the Fermi level is positioned closer to the LUMO.
When plasmons are excited on the surface of a metal, hot electrons and hot holes are created and become available to drive chemical reactions within the adsorbate. As shown in Fig. 5, excited carriers will be excited to a maximum energy of hν above the Fermi level. Three mechanisms that lead to chemical transformations in adsorbed molecules are then possible: (i) a hot electron (Fig. 5a) can be excited from the Fermi level to the LUMO of the molecule. (ii) A hot hole (Fig. 5b) enables electron transfer to the metal from the HOMO of the adsorbate. These two scenarios are referred to as indirect adsorbate excitation, and generally require the corresponding frontier orbitals to lie near the Fermi level.6 (iii) The coupling of surface plasmon energy into electronic states in the molecule enabling direct transitions is also possible.7,8,43 In addition to this, an increasing number of reports are appearing in the literature discussing the effect of high intensity electric fields and increased charge-carrier yields.12 Spaces between plasmonic nanostructures can generate higher intensity fields than on isolated nanoparticles and are thought to be areas of increased photochemical activity.
In our study, there are two reactions of interest. First, the cleaving of the S–C bond leading to the formation of silver sulfide, Ag2S, followed by an oxidation reaction producing Ag2SO4. Quantum chemical calculations using time domain density functional theory (TDDFT) provide values for allowed UV/Vis transitions of molecules adsorbed on Ag clusters,46 allowing to determine if charge transfer is from the metal to the molecule or vice versa for a given reaction. For example, theoretical calculations have shown that visible light driven reactions involving ArNO2 on Ag clusters are due to electron transfer from silver to the LUMO of ArNO2. On the other hand, it was also shown that electrons are transferred from the HOMO levels in adsorbed ArNH2 molecules to unoccupied levels in silver.47 Unfortunately, a detailed TDDFT calculation to confirm the direction of charge transfer for each reaction step is beyond the scope of this work. The experimental results demonstrate that the entire reaction can be driven with photon energies of 2.33 eV, and therefore the energy separation between the Fermi energy and either the HOMO or LUMO must lie within this energy range for bond cleaving to proceed, as sketched in Fig. 5. It is unlikely that the HOMO–LUMO energy gap for dodecanethiol would be reduced by ∼5 eV after adsorption, therefore direct excitation is ruled out. To our knowledge, this is the first demonstration of S–C bond scission in thiol molecules by visible light (532 nm radiation). The key to this reaction is a nano-textured Ag surface capable of strong absorption through SPR at visible wavelengths.
In addition to hot carriers, the reaction involving the oxidation of Ag2S to Ag2SO4 requires the presence of oxygen species. From Fig. 4a, the Raman scattering intensity from sulfate species reaches a maximum after 20 minutes. Since an increase in reaction rate is observed as irradiance is increased, the limiting factor must be the concentration of hot carriers to drive the reaction forward and not molecular species in the surrounding environment. In previous works, alkanethiol oxidation on gold and silver surfaces in air, under UV excitation, required the presence of ozone (O3).48,49 It has been suggested that UV excitation leads to decomposition of O3 into O2 molecules in the singlet state. Oxygen molecules in the singlet state are known to be highly reactive. Plasmonic decomposition of O3 has not been reported previously, and to assign the observed plasmonic photooxidation to ozone decomposition is premature; carefully designed experiments where the environment is controlled are required. Furthermore, metal sulfide oxidation will readily proceed in aqueous conditions, and therefore water molecules in the surrounding environment, or hydrogen bonded to the surface may play a role.
From a fabrication standpoint, laser machining represents a high throughput process capable of patterning large areas and requires minimal fabrication steps. When considering industrial scale operations, these attributes are essential if the technology is eventually going to scale-up. On the other hand, nanofabrication tools like electron beam lithography and focused ion beam milling can produce extremely well-defined structures at the nanoscale to support strong optical resonances in the visible range. This in turn would lower the power input requirement to initiate photooxidation. Self-assembled plasmonic silver structures with tunable resonances have also been realized,50 and can be fabricated over large areas as well. Along these lines, the best fabrication tool for photochemical applications involving silver remains an open question and requires further investigation.
This journal is © the Owner Societies 2018 |