Enhancing singlet oxygen photocatalysis with plasmonic nanoparticles

Photocatalysts able to trigger the production of singlet oxygen species are the topic of intense research efforts in organic synthesis. Yet challenges still exist in improving their activity and optimizing their use. Herein, we exploited the benefits of plasmonic nanoparticles to boost the activity of such photocatalysts via an antenna effect in the visible range. We synthesized silica-coated silver nanoparticles (Ag@SiO 2 NPs), with silica shells which thicknesses ranged from 7 to 45 nm. We showed that they served as plasmonically active supports for tris(bipyridine)ruthenium(II), [Ru(bpy) 3 ] 2+ , and demonstrated an enhanced catalytic activity under white LEDs irradiation for citronellol oxidation, a key step in the commercial production of the fragrance rose oxide. A maximum enhancement of the plasmon-mediated reactivity of approximatively 3-fold was observed with a 28 nm silica layer along with a 4-fold enhancement in the emission intensity of the photocatalyst. Using electron energy loss spectroscopy (EELS) and boundary elements method simulations, we mapped the decay of the plasmonic signal around the Ag core, and provided a rationale for the observed catalytic enhancement. This work provides a systematic analysis of the promising properties of plasmonic NPs used as catalysis-enhancing supports for common homogenous photocatalysts, and a framework for the successful design of such systems in the context of organic transformations.


Introduction
Visible light photocatalysis is currently at the heart of intense research developments, 1 in particular in the fields of energy, 2 pollution remediation 3 and organic transformations. 4hotocatalysis is often praised as a method enabling high energy-efficiency and chemical selectivity, complying with the principles of green chemistry. 1,5 any visible light photocatalysts have been developed, based on nanomaterials, 6 2D structures, 7 metal organic frameworks, 8 organometallic complexes 9 or organic molecules. 10Among the most popular molecular photocatalysts used in organic synthesis, tris(bipyridine)ruthenium(II), [Ru(bpy) 3 ] 2+ , stands out for a number of reasons.It has the ability to catalyse a large variety of organic [11][12][13] and inorganic 14 reactions, and it absorbs in the visible range.It has shown superior performances compared to other organometallic and organic photocatalysts in a variety of organic transformations. 15,16,17 u(bpy) 3 ] 2+ also features excellent phosphorescence properties, thanks to its ability to facilitate intersystem crossing in the excited state, which is key to enable energy transfer reactions. 18inally [Ru(bpy) 3 ] 2+ features good resistance to photobleaching and is thus stable under catalytic conditions, especially as compared to organo-photocatalysts. 18,19 u(bpy) 3 ] 2+ has been particularly efficient towards the generation of singlet oxygen ( 1 O 2 ), a powerful manifold for oxidative chemistry, even on complex substrates. 20This reactivity has been particularly useful in organic transformations, especially the Schenck-ene reaction, which allows the selective oxidation of citronellol, a key step in the manufacturing of the commercial fragrance rose oxide performed on a 100 T/year 21 scale, and the production of artemisin, a major antimalarial drug. 22Yet, as a member of the platinum series, Ru remains expensive and fairly toxic. 23There is therefore a need to develop new methodologies to enhance the performance of such molecular photocatalysts, so as to limit the quantity used in the process. 19,24 0][31][32][33][34] PNPs are, among other effects, able to act as antennae modulating and concentrating light in their immediate vicinity. 35,36 T8][39][40][41] The antenna effect enabled by PNPs has been exploited to tweak the selectivity and activity of metal based catalysts. 42,43 s early as 1968, Drexhage and coworkers foresaw the potential of associating PNPs to fluorophores, when witnessing the altered optical properties of a Eu complex in the proximity of a plasmonic Ag film. 44][47] The Heyne group has recently studied silica covered Au and Ag PNPs decorated with Rose Bengal (RB) towards the generation of 1 O 2 for antibacterial activity. 48They have studied in depth the role of the PNP metal by comparing Ag and Au performances, 48,49 as well as the influence of the PNP shape on photocatalysis enhancement. 50Moreover, they noticed that in all systems, there was an optimal shell thickness at which the LSPR effect was maximal.Interestingly, the thickness at which the maximum RB luminescence was measured slightly differed from the one at which 1 O 2 production peaked.Mendoza et al. have further studied similar silica covered Au PNPs and analyzed 1 O 2 generation by three independent methods to conclude that the SPR promoted of the absorption, intersystem crossing and triplet state generation in the RB photosynthesizer. 513][54] Non rigid spacers such as human serum albumin 55 , phospholipid membrane 56 and DNA 57 were also used to immobilized dyes at the surface of PNPs.Despite these developments, the application of this strategy for catalysis has been scarce.Mori et al. reported in 2010 the use of Ag@SiO 2 NPs decorated with [Ru(bpy) 3 ] 2+ for the oxidation of styrene. 58More recently, Reinhard and coworkers used a phospholipid membrane-covered Ag NP to deposit [Ru(bpy) 3 ] 2+ and demonstrated its efficiently for light-enhanced urea degradation. 56Yet, considering the importance of 1 O 2 methodologies in the pharmaceutical industry, more work is needed into establishing a structure/activity relationship, and studying the mechanism of this enhancement in the specific context of these organic transformations. 20As highlighted in a recent review by Linic and coworkers, such hybrid plasmonic materials relying on a plasmonic antenna to boost catalysts are a promising venue for the development of new plasmon-mediated catalysts. 59 thus synthesized silica coated Ag PNPs decorated with [Ru(bpy) 3 ] 2+ , and studied their luminescence and catalytic activity towards citronellol oxidation.In particular, we carefully tuned the silica layer thickness as a mean to control the distance between the supporting PNP and the catalytically active species.Here, immobilisation was performed by simple absorption onto the silica surface, following a strategy developed in our group (Amara) 60,61 and simplifying past approaches relying on covalent bonding. 58Photocatalysis for this reaction under optimized conditions was enhanced by the presence of plasmonic core, affording an almost 3-fold enhancement in reaction yield.The shell thickness affording the maximum for luminescence emission for [Ru(bpy) 3 ] 2+ was found to be 33 nm while the maximum catalytic enhancement was observed at 28 nm.In order to shed light onto the mechanism of plasmonic enhancement, we conducted a thorough LSPR mapping study using scanning transmission electron microscopyelectron energy loss spectroscopy (STEM-EELS).

Synthesis of plasmon-active supports for photocatalysis
Figure 1A) Synthesis of the silver plasmonic nanoparticles core with a silica shell and immobilization of the photocatalyst [Ru(bpy) 3 ] 2+ B-E) TEM images of Ag NPs (1B) and Ag@SiO 2 NPs with various thicknesses 1C-E (7 nm, 30 nm and 37 nm) F) High-angle annular dark-field image (HAADF), overlaid with EDS mapping for Ag (blue), Si (pink), and O (green) of an individual Ag@SiO 2 NP with a 28 nm shell thickness; G-I) EDS mapping of the same particle for Ag, Si and O; J) Intensity for HAADF and EDS of Ag, O and Si across the linescan shown in G. Scale bars for F), G), H), and I) is 50 nm.
For this study, Ag was selected as the plasmonic metal, since it had been shown to exhibit better field enhancement performances than Au, which was proven to be key for metal-enhanced luminescence. 33,62,63 W designed Ag NPs covered with a silica layer, Ag@SiO 2 NPs, to serve as supports for photocatalyst [Ru(bpy) 3 ] 2+ thus affording Ag@SiO 2 -[Ru] (Figure 1A).Briefly, Ag NPs were first synthesized using AgNO 3 as a precursor and polyvinylpyrrolidone (PVP) as a capping agent following previously reported methods (see SI for details). 64The Ag NPs were then coated with SiO 2 via the Stöber process, which consists of repeated addition and hydrolysis of tetraethyl orthosilicate (TEOS) until reaching the desired silica shell thickness. 65Following this step, Ag@SiO 2 were decorated with [Ru(bpy) 3 ] 2+ taking advantage of electrostatic interactions, and provided access to Ag@SiO 2 -[Ru]. 60,66,67 Ithis design, Ag was selected as the plasmonic metal, since it had been shown to exhibit better luminescence enhancement performances than Au. 62,63 e initially optimized the Ag core diameter and the thickness of the silica shell to ensure that the SPR signal effectively overlapped with [Ru(bpy) 3 ] 2+ , which has a maximum absorption at 450 nm. 68[71][72] In particular, we made sure that this overlap remained good across a range of silica shell thicknesses.This initial study allowed to identify an Ag core of 80 nm in diameter, with a shell of 25-35 nm, as a good fit.Afterwards, we optimized the amount of photocatalyst immobilized on the surface to maximise the luminescence of the photosynthesizer (See complete discussion in section 2 ESI).

Ag@SiO 2 and Ag@SiO 2 -[Ru] NPs characterization
Transmission electron microscopy (TEM) analysis of the resulting Ag@SiO 2 NPs confirmed the formation of Ag cores with diameters of 80.4 ± 8.1 nm (Figure 1B).Their silica shells featured thicknesses of 7 (Figure 1C), 10, 15, 19, 25, 28, 30 (Figure 1D), 33, 37 (Figure 1E), 40 or 45 nm, a parameter we could control via the number of TEOS additions.In all cases, the particles core and shells were very uniform across samples, with no core-free particles or irregular shells observed.High-angle annular dark-field image (HAADF) (Figure 1G) and mapping by Energy Dispersive X-Ray Spectroscopy (EDS) (Figures 1H-J) confirmed the presence of the silver core surrounded by a shell made of Si and O of approximatively 28 nm as measured on the TEM images.We then evaluated the specific surface of Ag@SiO 2 NPs by Brunauer−Emmett−Teller (BET) measurements.It ranged from 4 m²/g to 21 m²/g for shells of 7 nm to 45 nm, respectively (Table S1 ESI).While the variation in specific surface area was small, it did increase as the particle size grow, which is counter intuitive.We attributed it to surface roughening as layers of silica were added, as we found in an earlier report.The total pore volume obtained from the BJH method is in the order of E-05 cm 3 /g, indicating the absence of a porous network in the material.During subsequent [Ru(bpy) 3 ] 2+ immobilization, care was taken to control their density per surface unit, as a way to maintain two consecutive photocatalysts at a similar distance, on average.Upon optimization (see ESI section 2), the value of 6.5 10 -5 mmol of [Ru(bpy) 3 ] 2+ /m 2 of support was kept constant for all Ag@SiO 2 NP samples.Indeed, it is well known that photoactive molecules may undergo self-quenching when they are in close proximity, thus this parameter was considered the most important one to lock for comparison purposes. 73The subsequent immobilization of [Ru(bpy) 3 ] 2+ onto the silica shell was evaluated by ζ -potential measurements (Table S2 ESI).The naked Ag@SiO 2 NP surface featured a negatively charged surface of -29.0 mV, a value which dropped to -17.2 mV upon functionalization with [Ru(bpy) 3 ] 2+ .The amount of photocatalyst immobilized at the surface of Ag@SiO 2 NPs was determined by UV-Vis absorption spectroscopy, by subtracting the concentrations of [Ru(bpy) 3 ] 2+ before and after exposure to the support.Optimal immobilization was achieved when collected supernatants were colorless.Prior to silica coating, the core Ag NPs featured a plasmonic band centered at 420 nm Figure 2black).5][76] As the photocatalyst was immobilized at the surface of Ag@SiO 2 NPs, the extinction profile of the ensemble enveloped both the SPR band and the absorption of [Ru(bpy) 3 ] 2+ (Figure 2 -red and orange).][79] In order to evaluate the ability of the synthesized plasmon-active nanosupports to enhance the optical properties of [Ru(bpy) 3 ] 2+ , we turned to fluorescence measurements.First, we irradiated Ag@SiO 2 -[Ru] samples, at the excitation wavelength of [Ru(bpy) 3 ] 2+ , namely 450 nm, and the intensity of the luminescence was measured at its emission wavelength, 615 nm.The photoluminescence properties were studied for silica layers ranging from 7 to 45 nm with an identical concentration of photocatalysts immobilized on commercial SiO 2 NPs (figure S5).For all samples with medium and large silica shell, we observed an emission peak, which confirms the presence of [Ru(bpy) 3 ] 2+ onto the Ag@SiO 2 NPs and the preservation of its luminescence (figure S5).

Optical properties
For a given [Ru(bpy) 3 ] 2+ surface coverage, its emission intensity was strongly dependant on the thickness of the silica shell.In order to establish a clear comparison framework, we calculated an enhancement factor for each support (Equation S4).It is defined as the ratio between the intensity measured for each system and the one for commercial 80 nm SiO 2 NPs, without metallic core, decorated with the photocatalyst (SiO 2 -[Ru]).
The trend in emission intensity observed when varying the shell thickness (figure 3 orange dotted line) is in agreement with previous studies by the Heyne group, on the effect of plasmonic enhancement on the emission of Rose Bengal immobilized on Ag@SiO 2 NPs. 48,50 or very thin SiO 2 layers, the photocatalyst experienced a significant non radiative decay 38 and the emission was strongly quenched, presumably due to charge transfer between the photocatalyst and the metal.As we increased the shell thickness, the emission intensity started to increase significantly, until reaching a maximum for NPs decorated with [Ru(bpy) 3 ]Cl 2 with a layer of a 33 nm.Under these conditions, the catalysts were no longer quenched by the metal surface, thus allowing radiative pathways to take place.The increase in emission intensity was attributed to the enhanced electric field caused by the plasmonic Ag core, which boosts the excitation of the fluorophore and its emission.The maximum luminescence enhancement corresponded to a 4fold increase, as compared to [Ru(bpy) 3 ] 2+ immobilized on commercial SiO 2 .For thicker shells, the emission intensity lowered and reached a plateau which correlates to the emission intensity of plasmonic free system (SiO 2 -[Ru]), suggesting that the influence of the plasmonic core becomes negligible.

Photocatalytic study
We then proceeded to the study of the photocatalytic activity of Ag@SiO 2 -[Ru] NPs for the Schenck-ene reaction, specifically, the photooxidation of citronellol to the corresponding allylic hydroperoxydes (Scheme 1).The oxidation of citronellol was performed in a photoreactor illuminated with white LEDs using Ag@SiO 2 -[Ru] with shell thicknesses ranging from 7 to 45 nm, under aerobic conditions in acetonitrile for 3 hours (Figure 3 blue and green).The formation of the desired alcohol regioisomers was monitored using GC-MS and 1 H NMR, following quenching of the peroxides by triphenylphosphine.The molar catalytic loading was selected to be 3.4 mmol% in [Ru(bpy) 3 ] 2+ in all cases discussed onwards, while maintaining the surface coverage equal for the Ag@SiO 2 -[Ru] series.The effect of the plasmonic core on the photocatalyst is illustrated in Figure 4. Due to the SPR-induced field enhancement, the photocatalyst locally a higher electromagnetic field and the photoexcitation of [Ru(bpy) 3 ] 2+ increases.Relaxation (radiative or non-radiative) thus also increase.A proposed full reaction mechanism is provided in Figure S8.This mechanism is consistent with the fact the conversion was observed to grow linear as a function of light intensity (Figure S14).S3).Such a drop is commonly observed for immobilized homogeneous catalysts. 80We then turned to the reaction catalyzed by the Ag@SiO 2 -[Ru] series.With thin SiO 2 shells, of 7 and 10 nm, conversions of 17 and 18 % were observed.They were lower than the conversion observed with SiO 2 -[Ru] NPs, our model for non-plasmonic supported photocatalysts.As mentioned above, this result can be rationalized by the presence of the nearby Ag core which can quench the catalyst and negatively impact its activity.A similar drop in activity was observed when adding Ag NPs to a reaction solution containing homogenous [Ru(bpy) 3 ]Cl 2 .In this case the conversion dropped from 41 % to 12 % (Table S3).We hypothesized that this could be due to either photo quenching or competitive absorption of light by the material, as observed by our group with an Fe(0) support. 73As the shell thickness increased in the Ag@SiO 2 -[Ru] series, conversion went up steadily, until a maximum conversion of 68 % was obtained for a SiO 2 thickness of 28 nm.This value is higher than the one observed with both homogeneous [Ru(bpy) 3 ]Cl 2 and SiO 2 -[Ru] NPs.Specifically, this means that, at the maximum of plasmonic enhancement (Equation S5), the yield is improved by a factor of 2.6, as compared to the supporting on the simple SiO 2 NPs.Above a shell thickness of 28 nm the yields dropped rapidly and, beyond 35 nm, conversion plateaued at around 30 %, a value close to the performance of the simple SiO 2 -[Ru] system.In this range, we reasoned that the effect of the SPR is very limited, thus explaining an activity comparable to photocatalysts immobilized on a non-plasmonic support.Control experiments show that almost no conversion occur under light irradiation in the absence of catalyst, or with Ag@SiO 2 -[Ru] NPs without light irradiation (Table S3).When naked Ag NPs were used alone as catalyst, the conversion remained negligible.
Previous study showed that Ag PNPs could directly catalyze the formation of 1 O 2 , [81][82][83] but in our hands, this did not prove to be efficient for the Schenck-ene process.GC-MS studies of the conversion over time suggest that the hybrid nanomaterial was stable during the length of the reaction (Figure S12).
Interestingly, the most intense luminescence of Ag@SiO 2 -[Ru] was measured for a shell of 33 nm, while 28 nm of SiO 2 performed the best for the photocatalytic oxidation of citronellol.We have also checked the effect of the surface loading on catalysis at various shell thickness and confirmed that the most active system was at 0.53 μmol/g (Figure S13).A similar discrepancy in optimal shell thicknesses was observed by Heyne and coworkers when comparing the enhancement of the emission of a PNP enhanced photocatalyst (Rose Bengal) with its ability to produce 1 O 2 .While a 20 nm SiO 2 layer offered the highest luminescence for Ag@SiO 2 -RB NPs, the maximum 1 O 2 production was observed for a 11 nm silica shell. 48We note that similarly to our system, the maximum for catalysis occurs for thinner shells than the maximum of the luminescence enhancement.Also, similar to this system, Heyne and coworkers reported that, at the peak of 1 O 2 emission enhancement, the amount of 1 O 2 available in solution is enhanced by a 3-fold factor.This values are in agreement with the 2.6 enhancement factor we observed.Other antenna-photocatalyst systems employing a variety of metallic core, molecular dyes and spacing component have been developed.The characteristic of each system and the optimal spacing between the plasmon-active surface and the photocatalyst are detailed in section 5 of the ESI.A number of systems were developed with silica shells and they feature similar thicknesses for the maximum effect (most emission) in the range of 20-30 nm distances, similar to what we are reporting.A couple of systems do report much shorter distances, in the order of a few nm, one with silica 58 and one of a phospholipid layer. 56At such short distances though, we did observe significant quenching, incompatible with high enhancement.The quenching of dye fluorescence close to PNPs has been extensively studied and confirms that with shell thicknesses of the order of 10 nm and lower significant quenching should take place. 84This effect should be maximized when there is overlap between the LSPR and the dye excitation wavelength, which is the case here. 84Beyond these values, plasmonic enhancement has been shown before to remain strong even into the long distances, all the way to 200 nm. 72Our system is the sole one presenting such an optimization of the shell thickness in the context of catalysis (Table S4). 85inally, it is well known that dyes coupled with PNPs excited at their LSPR wavelength do feature a reduction in their lifetime.This effect is associated as well with the broadening of the LSPR observed upon deposition of [Ru(bpy) 3 ] 2+ onto Ag@SiO 2 . 84We did not observe that this had any significant effect on catalytic results.
Based on the results on catalytic tests, we turned to electron energy loss spectroscopy (EELS) in order to better rationalise the optimal enhancement effect observed for a certain shell thickness.

Scanning transmission electron microscopy (STEM)-electron energy loss spectroscopy (EELS) study
Electron energy loss spectroscopy (EELS), used in conjunction with scanning transmission electron microscopy (STEM), is a powerful technique to map both bulk or surface plasmons intensity and energy.One of the advantages of STEM-EELS is to simultaneously provide both geospatial and energetic information of the surface plasmons. 86,87 sing STEM-EELS, the surface plasmon energy of the naked Ag cores was measured to be 3.4 eV (365 nm), at the surface of the NPs (Figure S7). 88This value is different from the SPR value obtained for the same systems by UV-Vis spectroscopy, because the STEM-EELS experiment takes place in ultra-high vacuum, and not in solution.[91][92] Coating Ag NPs with silica affected the surface plasmons, as it altered the dielectric medium around the PNP.Indeed, the energy of the surface plasmon measured at the surface of the NP decreased as a function of silica shell thickness(Figure S7). 88For naked AgNPs, and Ag@SiO 2 with 7, 19 and 28 nm shells, we studied the intensity of the surface plasmon as a function of the distance to the Ag core (Figure 4).As expected, the intensity of the surface plasmon decreased exponentially in all cases.For no shell or a thin shell (7 nm), intensity was high and the decrease constant (Figure 4 grey and blue).For thicker shells (19 and 26 nm), there was a threshold effect, whereby almost no surface plasmon intensity was observed within the shell, while at the surface of the silica, intensity was restored and the signal decayed as a function of the distance like the naked system (Figure 4 green and orange).Interestingly in the latter systems, the intensity at the SiO 2 /vacuum interface was greater than at a similar distance from the plasmonic core for naked NPs and with a thin shell of 7 nm.Moreover, the LSPR peak shifted in energy as the silica shell became thicker, from 3.4 eV for naked Ag NPs down to 3.0 eV for a shell of 28 nm, which parallels the shift in the maximum of UV-Vis absorption shown in Figure 2. As STEM-EELS analysis are performed on individual particle, we selected NPs characteristic of the samples they belonged to.Analysis of shells larger than 30 nm could not be performed because the specimen was then too thick, preventing electrons from reaching the detector. 89gure 4 -Intensity of the LSPR peak measured by STEM-EELS at 3,4, 3.2, 3.0 and 3.0 eV, for naked Ag, Ag@SiO 2 _7nm, Ag@SiO 2 _19nm and Ag@SiO 2 _28nm NP respectively, measured as a function of the distance from the Ag surface.
We then performed boundary element method (BEM) calculations with Matlab for metallic nanoparticles using MNPBEM toolbox 93,94 on 80 nm Ag spheres with various thicknesses of SiO 2 shells, in order to simulate electron energy loss spectra and study the influence of the solvent on the surface plasmon energy position.
The energies determined by the simulations using a dielectric medium equals to 0 are similar to the ones obtained in the STEM experiment (I.e.under vacuum).Both simulation in vacuum and experimental data showed a peak centered at 3.4 eV for naked Ag NPs and a decreasing energy as a function of SiO 2 shell thickness (Figure 5A).By changing the value of the dielectric medium around the core shell NPs to mimic materials in a solvent, the general trend obtained is similar but the energies are altered by the environment surrounding the material.When switching from vacuum to CH 3 CN, the trend is similar but there is a small shift in energy associated to the change in the dielectric environment (Figure 5B).In both media, when the thickness of the shell increases, the energy and the intensity decrease.Yet for thin silica layers, between 0 and 15 nm, a peak with a lower energy but more intense than for naked NPs is observed, similarly to data collected using STEM-EELS.
Figure 5 -Simulations of the intensity of the LSPR peaks at energies ranging between 2.25 and 3.75 eV for Ag@SiO 2 with shell ranging from 0 to 45 nm obtained with Matlab MNPBEM toolbox in vacuum (A) and in CH 3 CN (B) Despite some discrepancies in the energy values obtained by STEM-EELS, MNPBEM toolbox and UV-Vis spectroscopy, which can be attributed to the polydispersity of the NPs and the different dielectric media, similar trends in the energy decrease with increasing silica thickness were observed for the three methods.Even though, the intensity is dumped when going away from the metallic core, the plasmonic NP still has a significant impact in its surrounding environment as far as 40 nm away from its surface (Figure 6).Moreover, the intensity at the surface of the silica shell is particularly compelling for thicker coating.Thus higher conversions were observed for Ag@SiO 2 -[Ru] NPs for which the separation between the photocatalyst and the plasmonic core was enough to avoid surface quenching but still offers an appealing enhanced electromagnetic field.These results suggest that each component in the catalyst design is essential to efficient plasmonic enhancement.First, the Ag core allows the needed electro-magnetic field enhancement by SPR.Second, [Ru(bpy) 3 ] 2+ acted as the catalyst for the reaction.Third, both the core and fluorophores needed to be properly sheltered from one another by a thick enough silica shell, while the shell could not be too thick, otherwise the plasmon effect was too weak.Fourth, light was needed to activate both the SPR and the fluorophore itself.If any of these conditions was not met, catalysis was sub optimal.Like for the fluorescence study, we observed a maximum in catalysis activity for a medium shell thickness, namely at 28 nm.Thinner shells are less efficient, likely because of quenching.For larger ones, the SPR of the core is likely to be less effective.

Conclusion
In conclusion, we have synthesized a series of plasmonic nanoparticles coated with silica at various thicknesses and decorated with a photocatalyst (Ag@SiO 2 -[Ru] NPs).This system allowed for the first time a systematic study of the phosphorescence and the photocatalytic activity of these materials.The light-induced field enhancement in the vicinity of the plasmonic Ag NPs is responsible for the increased production of 1 O 2 thus the catalytic activity of the [Ru(bpy) 3 ]Cl 2 towards the oxidation of alkenes.We have demonstrated an up to 4-fold activity increased compared to [Ru(bpy) 3 ]Cl 2 immobilized on SiO 2 NPs for nanomaterials with a 28 nm silica shell.Careful engineering of the hybrid plasmonic nanomaterial and the given photocatalyst is crucial as quenching of the photocatalyst was observed for thinner silica shells, while the photocatalyst was not affected by the LSPR for materials with thicker shells.Interestingly, EELS measurements and Matlab MNPBEM toolbox simulations revealed that even though there is an isolating layer, the intensity of the surface plasmons is strong at the interface between the silica layer and the photocatalyst, thus separation between the photocatalyst and the plasmonic core avoid surface quenching but still offers an appealing enhanced electromagnetic field in the photocatalyst's environment.The LSPR-induced field enhancement is the strongest at the surface of naked Ag NPs yet photocatalysts undergo quenching and for our Ag@SiO 2 -[Ru] NPs, maximum luminescence emission and catalytic activity were observed for thicker silica coatings, for which the intensity of the surface plasmon is still significant.

Experimental section
All the chemicals were purchased from Sigma Aldrich and used as received.
Silver nanospheres were prepared using the protocol reported by Zhao et al. 95 with slight modifications.18,0 mL of ethylene glycol (EG) was used to dissolved 0,90 g of polyvinylpyrrolidone (PVP) 30K MW 40K in a round bottom flask previously washed with aqua regia.200 mg of AgNO 3 dissolved in 2,0 mL of EG were added to the PVP solution at room temperature.After stirring at 145 °C for 4 h, 5 mL of acetone and 5 mL of ethanol were added, and the reaction mixture was let to cool down.The solution turned from colorless to green-brown when reaching the set temperature and the final solution was beige.The Ag NPs were washed three times using 30 mL of 50:50 acetone and ethanol mixture and centrifuged at 12 000 rpm at ambient conditions for 15 min.The resulting pellets were easily to suspend in ethanol and were dried in a vacuum oven at room temperature.
The Ag@SiO 2 NPs were obtained via base-catalyzed Stöber process.In a typical synthesis, 25 mg of Ag NPs were added to a round bottom flask containing 40 mL of ethanol, 10 mL of ultra-pure water and 1.25 mL of NH 4 OH 30%.25 μL of TEOS were added to the vigorously stirring solution at room temperature every 12 h until the SiO 2 layer reached the thickness wanted.Ag@SiO 2 NPs were isolated and purified by adding 20 mL of ethanol and centrifuged at 8 000 rpm for 10 min.The nanoparticles were washed two more times using 30 mL of ethanol and 10 mL of water prior to be suspended in ethanol and dried in a vacuum oven.The reaction was scaled up to 125 mg of Ag NPs by increasing the amount of reagents and solvents accordingly.
Immobilization of [Ru(bpy) 3 ]Cl 2 on Ag@SiO 2 NPs was realized following a procedure previously reported by the Amara group. 60Briefly, an aqueous stock solution of [Ru(bpy) 3 ]Cl 2 .6H 2 O 0.4 mM was prepared.A varied volume of this solution was then added to 20 mg of dried Ag@SiO 2 NPs suspended in a total volume of 3 mL of water.The volume added was calculated depending on the BET surface area in order to obtain the desired loading and surface coverage, which was identical of each nanomaterial.The solution was sonicated then stirred in the dark overnight.The final material was washed three times with water and dried by lyophilization.Complete immobilizations resulted in a colorless supernatant, which was confirmed by UV-Vis spectroscopy.
Photooxidation of citronellol was performed in a 125 mL 3-neck jacketed with a condenser, a thermometer and a balloon filed with oxygen were connected to the 3 necks, and the reactor was maintained at 20 °C using a water recirculatory (Figure S9).The reactor was illuminated by white LED strips (Ustellar UT33301) with a daylight white color (6000 K) positioned on a cylindrical support of about 30 cm diameter and 10 cm height and the whole set up was covered with aluminum foil.
In a typical procedure, 0.33 mmol of b-citronellol and 0.20 mmol of ortho-dimethoxybenzene (internal standard) were dissolved in 20 mL of acetonitrile, using part of the solvent to suspend 2.0 mg of Ag@SiO 2 -Ru with a silica shell of 28 nm and surface loading equal to 6.5E-5 mmol of [Ru(bpy) 3 ] 2+ /m 2 .The amount of reagents was adapted to keep the concentration of molecular catalytic loading the same for all experiment while maintaining the concentration of PNPs identical, hence the opacity of the solution the same for all experiments.
For the kinetic studies, aliquots were taken every hour and quenched with 1.2 molar equivalent of triphenylphosphine and analyzed by GC-MS.After 3h, the reaction mixture was quenched with triphenylphosphine, concentrated by rotary evaporation and dissolved in CDCl 3 for 1 H NMR analysis.
Conversions were calculated using the ratio between the internal standard (orthodimethoxybenzene) and the reactant peak areas given by GC-MS at a given time, compared to the one at the beginning of the reaction.The conversion was also calculated using NMR by the ratio of the integration for the product and the internal standard.The reactions were performed in triplicate.

Figure 4 -
Figure 4 -Influence of the Ag NPs on the radiative and non-radiative rates of [Ru(bpy) 3 ] 2+ relaxation

Figure 6 -
Figure 6 -Intensity of the LSPR peak calculated with Matlab MNPBEM toolbox in CH 3 CN as a function of the distance from the Ag surface for naked Ag NPs and Ag@SiO 2 with 7, 19 and 28 nm shells