Mass Transport Limitations in Plasmonic Photocatalysis

Abstract

The interpretation of mechanisms governing hot carrier reactivity on metallic nanostructures is critical, yet elusive, for advancing plasmonic photocatalysis. In this work, we explored the influence of the diffusion of molecules on the hot carrier extraction rate at the solid–liquid interface, which is of fundamental interest for increasing the efficiency of photodevices. Through a spatially defined scanning photoelectrochemical microscopy investigation, we identified a diffusion-controlled regime hindering the plasmon-driven photochemical activity of metallic nanostructures. Using low-power monochromatic illumination (<2 W cm^–2), we unveiled the hidden influence of mass transport on the quantum efficiency of plasmonic photocatalysts. The availability of molecules at the solid–liquid interface directly limits the extraction of hot holes, according to their nature and energy, at the reactive spots in Au nanoislands on an ultrathin TiO₂ substrate. An intriguing question arises: does the mass transport enhancement caused by thermal effects unlock the reactivity of nonthermal carriers under steady state?

Hot-carrier-driven photocatalysis revealed unforeseen reaction pathways enabling innovative and selective photochemical processes, opening exciting avenues for sustainable and renewable energy conversion.¹⁻³ The generation of hot carriers is a nonradiative dissipation pathway of surface plasmons, which for nanometals consists in the excitation of the collective oscillation of surface free electrons upon an applied electromagnetic field, i.e., due to light irradiation, at the localized surface plasmon resonance (LSPR). These highly energetic carriers (electron–hole pairs) with sufficient energy, once reaching the solid–liquid interface, will interact with molecules in solution.²⁻⁵ Otherwise, they will convert their energy into heat.^6,7 The former can induce photochemical redox reactions, while the latter affects the diffusion of the species and reaction rate.⁸ Generally, plasmon-driven photochemistry exploits hybrid nanomaterials combining a metallic nanostructure (e.g., Au) contacted with a semiconductor (e.g., TiO₂) to enable the separation of hot carriers, thus increasing their lifetime and the probability that photocatalytic events may occur (i.e., interacting with molecules in solution).^3,9 The presence of the interface was shown to be critical for the enhancement of hot carrier extraction during photochemical processes.¹⁰⁻¹³ As such, the design of efficient hot carrier photodevices demands comprehension of the underlying mechanisms and limitations controlling the photochemical processes occurring at the solid–liquid interface. In particular, the parameters affecting the transport and extraction of hot carriers remain unclear. In the commonly reported imaging approaches, an irreversible photochemical reaction was followed in time capturing the morphological evolution on the nanostructure surrounding and thus revealing the photochemical “hot spots”, an equivalent of “reactive sites” in catalysis.^10,14−17

Studying hot carrier role during plasmon-driven photochemistry is at the center of a heated debate due to the intricacy for distinguishing the contribution from thermal (heat) and nonthermal (hot carriers) plasmonic effects.^18,19 Recently, scanning electrochemical microscopy (SECM) opened exciting opportunities for unraveling fundamental aspects of the different effects at play during plasmonic photocatalysis in liquid media, such as identifying the impact of photothermal effects,²⁰⁻²² quantifying hot carrier energy distribution,²³⁻²⁵ visualizing their nanoscale localization,²⁴ and even deciphering the interfacial state influence on the photocatalytic efficiency.²⁶

When considering photoinduced processes, SPECM (scanning photoelectrochemical microscopy) consists of an electrochemical probe (i.e., ultramicroelectrode; UME) scanning the irradiated nanostructure in its vicinity (e.g., nanometers to micrometers above the surface) that detects the concentration evolution of the reacting molecules in solution via an electrochemical current signal.²⁷ The probe current provides quantitative information on the amount of photogenerated products, which correspond to the hot charge carriers interacting effectively with the molecules in solution.^23,24 Based on its remarkable resolution, sensitivity, and speed to detect a very small amount of molecules in solution,^28,29 SPECM unlocks the investigation of photocatalytic reactions at relatively low power intensities, an illumination regime usually omitted in optical-based microscopies and spectroscopies.

The influence of the probe on electrochemical measurements was thoroughly investigated. These studies revealed meaningful insights regarding the hindering effect of the probe on the mass transport as well as local phenomena (e.g., local pH modification) due to the confinement of species between the probe and the substrate.³⁰⁻³³ However, this aspect is not explored for plasmon-driven photochemical reactions, during which the hot carrier extraction and the enhanced diffusion due to the thermal effect are mixed together. This increases the complexity of the underlying mechanisms driving the catalytic event.

In this paper, we explored the effect of the diffusion of the species on plasmonic photocatalysis by tuning the hindering effect of the electrochemical probe during local SPECM investigations. We compared the photocatalytic efficiency of plasmonic Au nanoparticles having two different size regimes and supported on TiO₂ films. Exploring different light intensities and energies as well as various electrochemical probe sizes and tuning the substrate–probe distance, we highlighted the influence of diffusion of the species (i.e., mass transport) in solution on the obtained signals related to the plasmon-driven photoactivity. This study revealed the critical, yet intricate, role of hot carrier population on their extraction at the solid–liquid interface, which is regulated by the mass transport and closely dependent on the incident light wavelength, photon flux, and nanoparticle size.

Samples with big (B–Au–TiO₂) and small (S–Au–TiO₂) Au nanoislands (NIs) on a TiO₂–ITO support are probed by SPECM. (1) Under light excitation, the Au/TiO₂ interface enhances the hot carrier separation due to the collection from TiO₂ of the excited electrons with sufficient energy to pass the Schottky barrier (ϕ_B), which then relax into the TiO₂/ITO underlayers.^23,24 (2) The remaining holes with sufficient energy oxidize the species surrounding the NIs. Here, we used the ferrocenedimethanol (Fc(MeOH)₂, FcDM), which is a fast and reversible one-electron-transfer redox probe with an outer-sphere mechanism and a highest occupied molecular orbital (HOMO) energy level matching the Au–TiO₂ Fermi energy level (i.e., −4.87 and −4.7 eV vs vacuum for E_HOMO,FcDM and E_F,Au–TiO2, respectively).^23,24 The oxidation rate of FcDM was only limited by its diffusion in the solution and the available hot carriers reaching the NI surface. (3) The electrochemical probe detects the concentration evolution of the redox mediator from either the depletion of reduced species, i.e., FcDM (redox competition mode, RC),^34,35 or the production of oxidized species, i.e., FcDM⁺ (substrate generation–tip collection mode, SG-TC)^36,37 in the vicinity (Figure 1a).

Figure 1.

Method and physical characterization of B- and S-Au-TiO₂. (a) Illustration depicting the SPECM measurements at an Au/TiO₂ interface for SG-TC and RC modes. (b) Absorptance spectra of the Au NIs (blue curve, S–Au–TiO₂; red curve, B–Au–TiO₂). (c, d) SEM images overlaid with the population distribution of the Au NIs for (c) B–Au–TiO₂ and (d) S–Au–TiO₂ samples.

The Au NI fabrication is detailed in the Supporting Information. The optical properties of B–Au–TiO₂ and S–Au–TiO₂ (Figure S1) were obtained by diffuse reflectance spectroscopy (DRS). B–Au–TiO₂ presented a broadband absorption at wavelengths >550 nm because of the large NI distribution in size and shape, while S–Au–TiO₂ showed a LSPR peak at ∼580 nm mirroring the formation of hemispherical nanoparticles with relatively homogeneous size (Figure 1b).^24,38 The Au NI populations were 211.3 ± 103.1 and 14.6 ± 8.7 nm in diameter (Figure 1c,d), while the heights were in the ranges of ∼20–40 and ∼10 nm (as previously reported for similar Au NIs)^24,39 for B–Au–TiO₂ and S–Au–TiO₂, respectively.

In order to evaluate the significance of the mass transport from the bulk solution to the Au NI photocatalytic hot spots, we employed different probe sizes with different active (r_T) and inactive parts (r_ins) (Figure 2a), as well as varying distances between the probe and plasmonic photocatalysts. Three probes were tested for both SG-TC and RC modes: Probe 1 (r_T = 5 μm, r_ins = 25 μm, and RG = r_ins/r_T = 5), Probe 2 (r_T = 0.5 μm, r_ins = 25 μm, and RG = 50), and Probe 3 (r_T = 100 nm, r_ins = 2.5 μm, and RG = 25). The probe–substrate distance (d) was determined thanks to an approach curve. The probe was positioned above the Au NIs for measuring the concentration of the photogenerated products at a fixed position. The potential applied to the probe (E_T) was set at the diffusion-limited plateau (Figure S2) to reach a steady-state regime. Notably, we performed our experiments under light intensities (<10 W cm^–2) producing negligible temperature increase (i.e., <1 K) to avoid misjudgment associated with plasmonic heating.²¹ This was confirmed by dark measurement with heated solutions, where the measured probe current (I_T) drifted by +44.6 and −0.7 pA K^–1 for the RC and SG-TC modes, respectively (Figure S3). This current drift revealed the mass transport enhancement caused by a temperature increase (Figure S4), which was not observed during our photocatalytic experiments.

Figure 2.

Fixed position investigation of B–Au–TiO₂. (a) Scheme representing the investigation with Probe 2 (r_T = 0.5 μm and RG = 50) at 2 μm from the substrate. (b) I_T versus time, with pulsed light (10 s, 595 nm) and increased light power at every pulse, for RC (red curve) and SG-TC (blue curve) modes corresponding to the scenario depicted in (a). (c, d) ΔI versus light power at 530 nm (green symbols), 595 nm (orange symbols), and 660 nm (red symbols) for (c) RC and (d) SG-TC modes obtained from values in (b) for 595 nm. The different regimes are represented in (c): (1) nonthermal and (2) diffusion-controlled regimes. The error bars represent the standard deviation. The solid and dashed lines correspond to the linear fit for measurements using light power <2 W cm^–2 and >2 W cm^–2, respectively. The power densities of the LEDs are reported in Table S1.

To explore the influence of the probe hindrance on molecular diffusion, we performed an experiment, depicted in Figure 2a, consisting of a chronoamperometry (CA) with the light turned on and off every 10 s at increasing light intensity setting at a fixed distance from the investigated sample (Figure 2b). The diffusion profiles corresponding to each scenario were simulated using a COMSOL model for the SG-TC mode.²⁴ The measurements were performed for each probe at d = 2, 5, 10, and 20 μm (Figures S5–S7). Different photoactivities (differential current between light_on and light_off, |ΔI|) were observed depending on the applied SPECM mode (ΔI_SG-TC ≤ ΔI_RC). For instance, for Probe 2, d = 2 μm and λ_595 nm = 7.64 W cm^–2; the ΔI values were −97.6 ± 0.3 and −62.3 ± 0.4 pA for SG-TC and RC modes, respectively (Figure 2c,d). The |ΔI_SG-TC| was 1.57 times higher than |ΔI_RC| for Probe 2, d = 2 μm (Figure S8). This difference was directly related to the influence of r_T on the FcDM/FcDM⁺ concentration in solution.^34,40 During RC mode, the probe consumes the FcDM decreasing the number of available species at the Au NIs. By contrast, the photo-oxidized species (FcDM⁺) at the Au NIs are reduced back by the probe regenerating FcDM during SG-TC measurements. This phenomenon was not observed (ΔI_SG-TC ≈ ΔI_RC; Figure S8) for higher d and lower r_T because of the negligible influence of the probe active part on the measurement, as the concentration of the species at the Au NIs was not impacted by it under these conditions (Figure S6c,d and Figure S7b–d).³⁰

B–Au–TiO₂ was excited at light energies (2.34, 2.04, and 1.88 eV) lower than the TiO₂ bandgap (i.e., E_G,TiO2 = 3.2 eV),⁴¹ avoiding possible contributions from the TiO₂ underlayer. The excitation energies were selected according to the contribution of interband and intraband transitions: mainly interband transitions (2.34 eV, 530 nm), a mix of intraband and interband transitions (2.04 eV, 595 nm), and mainly intraband transitions (1.88 eV, 660 nm).^24,42 The photoactivity showed a wavelength dependency (Table S2) with higher |ΔI| measured at longer wavelengths (|ΔI₅₃₀| < |ΔI₅₉₅| ≤ |ΔI₆₆₀|; Figure 2c,d). The measured ΔI function of the light power density (P) showed two trends (Figure 2c). At P < 2 W cm^–2, (1) ΔI evolved quasi-linearly with P. This linear increase was observed for every measurement, and it is considered as revelatory of a nonthermal photoactivity regime due to hot holes. By contrast, for P > 2 W cm^–2, (2) the slope decreased drastically reaching, in the extreme case, a steady state, which corresponds to a diffusion-controlled regime (Table S3). Indeed, the measured ΔI was close to the diffusion limiting current for 1 mM FcDM/FcDM⁺; dark I_T,RC = ∼108 pA (Figure S6d) and |ΔI_SG-TC| = 97.6 ± 0.3 pA at λ_595 nm and P = 7.64 W cm^–2 (Figure 2d). These results suggest that the oxidation rate is directly related to the incident light power P and wavelength (λ). When the slope equals 0 pA W^–1 cm², the increase in P decreases the quantum yield. Considering the hindrance of the probe on the diffusion of the species (e.g., RC vs SG-TC), it emphasizes that the observed phenomenon was related to mass transport evolution, dismissing other possible mechanisms (e.g., optical saturation and Au/TiO₂ charge transfer limitation). As such, the plasmonic activity was limited by the FcDM available at the NI active sites. Since the NI active sites depend on the NI geometry, the mass transport evolution differs according to P and λ. Furthermore, a wavelength dependency was also observed for the slope (S) relating the evolution of ΔI with P. From the S values (Table S3), we observed that |S| increased from 530 to 660 nm for P < 2 W cm^–2 (|S₅₃₀| < |S₅₉₅| ≤ |S₆₆₀|). However, the opposite trend was observed for |S| at P > 2 W cm^–2 (|S₅₃₀| > |S₅₉₅| ≥ |S₆₆₀|), while the photoactivity trend remained the same (|ΔI₅₃₀| < |ΔI₅₉₅| ≤ |ΔI₆₆₀|). This suggests that the reactive site availability is linked to the nature of the transitions. At 530 nm, the interband transition mainly occurs while plasmonic hot carriers can also be generated.²⁴ The former produces highly energetic holes with shorter lifetime and generated in the whole nanostructure volume (NIs of ∼200 nm diameter and ∼30 nm height for B–Au–TiO₂), which will mostly recombine before reaching the solid–liquid interface and reacting with the redox mediator.⁴³ In contrast, plasmonic hot carriers are generated through a surface-assisted mechanism and can therefore more easily participate to the photocatalytic events.⁴⁴ Additionally, interband transitions in Au are optical losses that usually diminish the efficiency of plasmonic hot carrier generation. Overall, these effects explain the lower |S| and |ΔI| observed at 530 nm than when the photocatalyst was excited at 595 nm (lower generation from interband transition) and 660 nm (mainly intraband transition).²⁴

Further, the external quantum efficiency (EQE) was calculated from the ΔI_SG-TC, based on a previously introduced diffusion model,²⁴ for different probes and λ_exc at d = 20 (Figure 3a) and 2 μm (Figure 3b). First, we note that the EQE decreased with the increase in P, particularly for d = 2 μm and P < 2 W cm^–2 (∼5 × 10¹² photons s^–1). This EQE evolution was observed for all the probes and was significantly faster for 595 and 660 nm compared to 530 nm: a photon flux increase of 1 order of magnitude reduced the obtained EQE to 1/3 and 1/2 for 660 and 530 nm, respectively, using Probe 2 at d = 2 μm. Moreover, the EQE at P < 2 W cm^–2 showed a dependency with the probe size and distance from the investigated surface: the smaller the probe and the distance from the investigated surface, the bigger the EQE. At P > 2 W cm^–2, all EQE stabilized ∼0.032 ± 0.013%.

Figure 3.

Quantum yields of B–Au–TiO₂. (a, b) External quantum efficiency (EQE) function of the incident photons for B–Au–TiO₂ excited under LED lights (530 nm, green symbols; 595 nm, orange symbols; 660 nm, red symbols) with Probe 1 (filled squares) and Probe 2 (empty circles) at d = 20 μm (a) and with Probe 2 (empty circles) and Probe 3 (filled diamonds) at d = 2 μm (b). (c–f) Schemes depicting the influence of the system on the FcDM recycling and the probe hindering effects on the diffusion of the species for Probe 1 at d = 20 μm (c), Probe 2 at d = 2 μm (d), Probe 1 at d = 2 μm (e), and Probe 3 at d = 5 μm (f). Insets in (d) and (e) represent the half-cell (as highlighted by dashed white and black rectangles in parts d and e, respectively) probing area and probe influence for Probe 2 and Probe 1, respectively. The concentration profiles of the species were simulated with the COMSOL diffusion model from ref (24). Similarly, the EQEs were calculated from ΔI_SG-TC with the same diffusion model. The error bars represent the standard deviation. The photon fluxes are reported in Table S1.

To understand the obtained EQE, it is critical to consider the influence of the experimental conditions (e.g., r_T, r_ins, and d) (Figure 3c–f).^27,30,36,40 Based on that, we believe that the almost constant and lower EQE for Probe 1 (Figure 3a) originated from the size of the probing area, which corresponded to the irradiated area of Au NIs (r_T = r_beam), as shown in the insets of Figure 3d,e. Thus, the obtained EQE included the effect of the counter reaction (FcDM⁺ + e^– → FcDM) occurring outside of the illumination beam spot due to the relaxation of electrons (e.g., bipolar electrode effect).⁴⁵ The influence of the probe on the FcDM regeneration and the hindered diffusion of FcDM species was low due to the large d (Figure 3c). A similar scenario with a r_T 10 times smaller (i.e., probed area 100 times smaller) showed higher EQE at lower light intensity, since the species from the counter reaction reaching the probe detection area were negligible for lower light intensities (Figure 3a). By hindering the diffusion of the species (Figure 3d,e), this effect was enhanced, leading to a greater number of oxidized species diffusing to the probe compared to the reduced species from the counter reaction. Consequently, the EQE for the lowest power intensity at d = 2 μm was 1.58 times higher than that at d = 20 μm. Interestingly, the EQE decreased drastically according to the light power intensity for Probe 3 (Figure 3b), despite the negligible influence of the probe on the measurement (Figure 3f). This suggests the effect of another phenomenon related to the hot carrier extraction at the solid–liquid interface. Further, the EQE stabilization at higher P suggests that the mass transport limitation observed in Figure 2c,d is related to the most reactive sites of Au NIs, while the less reactive sites remain available for FcDM species. The reactive sites correspond to the sites where the molecules are oxidized, meaning that the charge carriers are extracted at the solid–liquid interface. Since the hot charge carrier has a limited mean free path, it has a higher probability to separate at the liquid–metal–semiconductor interface (e.g., FcDM/Au/TiO₂ interface), which is considered as the most reactive site. By increasing the light power, the population of generated charge carriers increase. If a reactive site is already highly populated (limited by the extraction of hot holes by the molecules in solution), a mass transport limitation is observed (e.g., a slope decrease). The observed stabilization at higher P is related to the extraction probability for the charge carriers generated “far” from the FcDM/Au/TiO₂ interface (e.g., at the FcDM/Au interface related to the ∼30 nm NI height). We believe these sites as less reactive (e.g., lower population of charge carriers reaching the interface). Consequently, lower to no mass transport limitation occurs for those reactive sites, as the FcDM oxidation rate is lower than the FcDM diffusion.

Next, we performed similar measurements on S–Au–TiO₂ (diameter ∼15 times smaller than that of B–Au–TiO₂). We investigated S–Au–TiO₂ at 455 nm (mainly interband transitions) and 530 and 660 nm. The obtained EQEs at 530 and 660 nm showed similar decay at low power density (Figure 4a), suggesting that the diffusion of the species influenced the extraction of hot charge carriers even for smaller Au NI size (14.6 ± 8.7 nm) within the same order of magnitude as the hot charge carrier mean free path. Interestingly, the EQE at 455 nm showed a quasi-steady state regardless of the quantity of incident photons. This suggests that the mass transport limitation is directly related to the photogenerated hole population reaching the interface.^23,44

Figure 4.

Quantum yields of S–Au–TiO₂ and comparison with those of B–Au–TiO₂. (a) EQE function of the incident photons for S–Au–TiO₂ excited under LED lights (455 nm, blue symbols; 530 nm, green symbols; 660 nm, red symbols) with Probe 2 at d = 2 (filled squares) and 20 μm (empty circles). (b) IQE function of the incident photons for S–Au–TiO₂ (filled squares) and B–Au–TiO₂ (stars) excited under LED lights (455 nm, blue curve; 530 nm, green curve; 660 nm, red curve) with Probe 2 at d = 2 μm. The error bars represent the standard deviation. The photon fluxes are reported in Table S1.

From the DRS measurements (Figure 1b), we calculated the internal quantum efficiency (IQE). The influence of the probe on the mass transport was identical for the IQE (Figure S9) and the EQE (Figure 3a,b). S–Au–TiO₂ showed the highest IQE (Figure 4b) compared to B–Au–TiO₂, which is in agreement with the literature considering the dependency of the Au NI geometry on the charge carrier generation and recombination mechanisms.^12,38,42,46 These results were expected since and (1) the absorption of nanostructures increases with their size and (2) the generation of plasmonic hot charge carriers is a surface-assisted mechanism, therefore significantly favored for smaller nanostructures.⁴⁷ Furthermore, the IQE showed an enhancement from B–Au–TiO₂ to S–Au–TiO₂ of 2.4–3.1-fold at 530 nm and 4.7–10-fold at 660 nm, depending on the incident photon flux. This enhancement was more evident for 660 nm, as the charge carriers were generated mainly by intraband transitions, meaning plasmonic hot carriers. Thus, the generated charge carriers surviving recombination, reaching the interface, and interacting with the molecules in solution were more abundant for S–Au–TiO₂ (smaller nanostructures) than B–Au–TiO₂. Notably, the ratio between the IQE at low (∼0.6 × 10¹² photons s^–1) and high (∼14 × 10¹² photons s^–1) photon fluxes (IQE_L/H) provided further insights into the hot carrier availability at the solid–liquid interface: IQE_L/H,530 nm = 2.1 and 2.7; IQE_L/H,660 nm = 2.3 and 4.7 for S–Au–TiO₂ and B–Au–TiO₂, respectively. The IQE_L/H displayed that the mass transport limitation of hot carriers for plasmon-driven photochemistry decreased as follows: B–Au–TiO₂@660 > B–Au–TiO₂@530 > S–Au–TiO₂@660 > S–Au–TiO₂@530. Further, the IQE ratio obtained for S–Au–TiO₂@455 (IQE_L/H,455 nm = 1.4) emphasizes the significance of the carrier population available at the interface. Since the different interfaces were responsible for the extraction of the hot carriers (i.e., hole extraction through FcDM/Au interface and electron extraction through Au/TiO₂ interface), the extraction of hot charge carriers was most likely occurring first at the FcDM/Au/TiO₂ interface because of their mean free path.^2,3,44 The presence of the FcDM/Au/TiO₂ interfaces is directly related to the size of the Au NIs: smaller NIs = a higher amount of FcDM/Au/TiO₂ interfaces relative to the volume of the NIs. This explains further the difference in IQE between S–Au–TiO₂ and B–Au–TiO₂, as S–Au–TiO₂ presents a higher quantity of FcDM/Au/TiO₂ reactive sites compared to B–Au–TiO₂. Considering these interfaces and the hot carrier population function of the λ_exc, we attribute the significant drop of quantum yields depending on the P increase as a consequence of the most reactive spots (i.e., most populated in hot carriers) of the Au NIs being limited by the mass transport, which limits the extraction of the holes from the system. In other words, the diffusion-controlled regime was reached when the extraction of hot holes through FcDM was limited by the FcDM availability at the Au reactive spots. Moreover, due to the wide variety of photocatalytic hot spots at the Au NI surface, the drop in quantum yields caused by the diffusion limitation occurred at lower photon flux than expected from the ΔI measurements reported in Figure 2c,d, showing the relevance of investigating plasmon-driven photocatalytic systems at low P (<1 W cm^–2).

For P generating plasmonic heating (>10 W cm^–2), we believe that the thermalization of the unextracted hot carriers will break the mass transport limitation. This effect will enhance the apparent efficiency from the plasmonic nanostructure while shadowing the limitation related to the diffusion of the species. However, this increase in activity is not related to the reactive sites of the investigated materials but to the availability of molecules reaching them. This phenomenon can mislead in the attribution of the mechanism contributing to the observed activity.

In conclusion, we showed the critical role of mass transport during plasmon-driven photochemistry investigated locally with scanning photoelectrochemical microscopy. The measured quantum yield was dependent on the diffusion of the species in solution, which was intimately related to the charge carrier extraction rate at the solid–liquid interface. These results showed the significance of the nanostructure size on the availability of the hot charge carriers to react with the molecules in solution and the relevance of multiscale investigations to reliably decipher the related mechanism contributing to photochemical reactions. Our findings highlighted that mass transport is an essential factor in why photocatalytic processes are far from reaching theoretical limits. Moreover, it explains why the results reported in literature present a wide heterogeneity in assessing if photocatalytic efficiency in solution is due to either hot carrier chemistry or photothermal effects. For the former, an incorrect evaluation of local temperature at the nanoscale may shadow the mass transport enhancement caused by photothermal effects, thus favoring the availability of nonthermal charge carrier reactive spots. For the latter, the obvious increase in temperature measured at the macroscale can lead to hasty conclusions, while the real actors “behind the scene” are the hot carriers activated thanks to this photothermal effect. Finally, these measurements pave the way toward the identification and quantification of reactive hot spots from plasmonic nanostructures and the rationalization of their photocatalytic reactivity.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.nanolett.4c01386.

• Experimental section, additional SPECM data, and DRS measurements (PDF)

Author Contributions

O.H. and A.N. conceived and designed the experiments. A.N. coordinated and supervised the project. O.H. performed the preparation of the samples and the measurements and analyzed the data. O.H. prepared the figures and wrote the first draft. All authors discussed and edited the manuscript.

The authors declare no competing financial interest.