Water security is critical for global health, which, however, is continuously challenged by emerging contaminants and pollutants, such as pesticides and plasticizers. Given the ever-increasing water-quality standards, conventional treatments including coagulation/precipitation and media filtration followed by disinfection are becoming less effective and less economical. As a promising alternative, photocatalytic degradation of micropollutant based on metal-promoted photocatalysts, e.g., Au/TiO₂, can achieve more advanced water treatment objectives and has a high socio-economic capacity. Currently, photocatalytic degradation of micropollutants is still limited by the low conversion efficiency from solar energy to chemical reaction energy, wherein the effective adsorption of micropollutants at low concentrations is key to the photodegradation performance. However, little is known on the quantitative adsorption behaviors of micropollutants, especially at the nanoscale and on metal-promoted photocatalysts.

Figure 1. (a) Illustration of adCOMPEITS and experimental design. DP: diethyl phthalate. PM: pirimiphos methyl. (b) SEM image of a representative Au/TiO₂ nanostructure. Vertical lines dissect the TiO₂ nanorods into 300 nm segments lengthwise. (c,d) 2D histograms of single-molecule localizations of reporter fluorophore Cy3.5 adsorption on a single Au/TiO₂ nanostructure in the absence (c) and presence (d) of PM. Magenta-dashed line: structural contour from SEM in b. (e) AdCOMPEITS image derived from delta_n/n = (n₀−n)/n between c and d. (e) 1D projection of the adCOMPEITS image of the TiO₂ nanorod (upper) and the fitted K of individual TiO₂ segments (lower) as dissected in e. Solid red lines: bi-directional fitting with . K_e, K_i, and x₀ reflect the adsorption enhancement amplitude near the Au-TiO₂ interface, the intrinsic adsorption equilibrium constant on the side facets of TiO₂, and the decay distance constant of the adsorption enhancement, respectively.

To bridge the knowledge gap, we developed a new imaging technique adCOMPEITS, namely adsorption-based competition-enabled imaging technique with super resolution, that can quantitatively map micropollutant adsorption on a single metal-promoted photocatalyst under non-reaction and photo(electro)catalytic reaction conditions. As a proof of concept, we implemented adCOMPEITS into studying a pesticide pirimiphos methyl (PM) and a plasticizer dimethyl phthalate (DP) on single Au/TiO₂ heterostructured photocatalysts (Figure 1a,b), in which a cyanine 3.5 based dye (Cy3.5) serves as the fluorescence reporter. The first step of adCOMPEITS was to conduct single-molecule imaging and super localization of individual adsorption events for Cy3.5 molecules on Au/TiO₂ (Figure 1c). Afterward, the micropollutant molecules were introduced into the imaging system to compete with Cy3.5 for adsorbing onto the surface sites of Au/TiO₂, which led to the suppression of detected Cy3.5 adsorption events (Figure 1d). By quantifying the suppression and building the competitive adsorption model, the adsorption affinity K_L of micropollutant was found to scale with the number of fluorescent molecule adsorption events n, i.e.,  µ K_L, where Δn is the difference without and with micropollutants. Based on this equation, we directly obtained an image, termed adCOMPEITS image, that immediately shows that the micropollutant PM adsorption on the Au nanoparticle is stronger than on TiO₂ within a single Au/TiO₂ nanostructure (Figure 1e).

Figure 2. (a) SEM image of a representative Au/TiO₂. (b,c) Sub-particle photoanodic current i_photo at an applied potential of 0 V vs. Ag/AgCl (b) and  sub-particle V_FB of the Au/TiO₂ nanostructure Solid lines: bi-directional exponential fitting.

Strikingly, we found that PM adsorption at the Au-TiO₂ interface show remarkable enhancement than on the distal TiO₂ segment, which is long-range and could reach >1 μm in distance based on the exponential decay constant x₀ (Figure 1f). In contrast, no adsorption enhancement was observed on isolated TiO₂ nanorods, suggesting the key role of Au cocatalyst in triggering the long-range adsorption enhancement. Given that the energy band in TiO₂ will bend when contacting a metal cocatalyst with a lower fermi level E_F, we speculated that the long-range effect might be attributed to the Au-induced surface band bending in TiO₂. To verify the assumption, we measured the sub-particle photoelectrochemical current i_photo of Au/TiO₂ heterostructures because the surface energy change will cause changes in the band bending at the solid-electrolyte interface, which is associated with the charge carrier separation and thus the photoelectrocatalytic activity in water oxidation (Figure 2a). Excitingly, the sub-particle photoelectrochemical current also shows a peak current at Au-TiO₂ interface and then exponentially decays along the TiO₂ nanorods (Figure 2b). By measuring sub-particle photocurrents at various applied potentials, we calculated the flat-band potential V_FB that directly reflects the surface energy band bending along the length of TiO₂ nanorod (Figure 2c). Consistently, V_FB displays a similar trend but in an opposite direction as that of i_photo, with an exponential decay constant of ~1 μm similar to that of the long-range adsorption enhancement distance on the TiO₂ surface. Building on these discoveries, one could be easily convinced that the long-range adsorption enhancement should stem from the long-range surface band bending effect induced by the contact of TiO₂ with Au.

This exciting discovery further inspired us to manipulate the long-range effect through material engineering. On one hand, we replaced Au with Pt as the cocatalyst to enhance the surface energy band bending in TiO₂; on the other hand, we doped TiO₂ with nitrogen to weaken the surface energy band bending in TiO₂ (Figure 3a). As expected, both the amplitude of adsorption enhancement K_e at the metal-semiconductor interface and the long-range enhancement distance x₀ were tunable when the degree of band bending in TiO₂ was changed accordingly (Figure 3b,c). These results from surface band bending engineering provides another demonstration of a cocatalyst’s role in exerting long-range effects on molecular adsorption on photocatalysts.

Figure 3. (a) Schematic showing the band bending of semiconductor induced by metal-semiconductor contact. eV_BB =  E_F^semi - E_F^metal , where  and  are the Fermi levels of semiconductor and metal, respectively. (b,c) Violin plots showing the effect of N-doping of TiO₂ and Pt as the cocatalyst on the long-range adsorption enhancement parameters: x₀ (b) and K_e (c). Magenta lines: mean values. ***p < 0.001; ****p < 0.0001; paired student’s t test.

In the context of photocatalysis, to take advantage of the long-range adsorption enhancement, there should exist some optimal physical size of photocatalysts or optimal spatial spacing (i.e., loading) of metal cocatalysts to be comparable to the width of surface band bending zone. One could also envision a variety of other potential applications, such as in surface-mediated sensing and dye-sensitized solar cells, where molecule adsorption on semiconductors is crucial. The generalizable imaging technique, the obtainable molecular insights, and the general underlying mechanism presented here should open new avenues for scientific discoveries. 

For more detailed information, see our article “Long-range adsorption enhancement of micropollutant on metal-promoted photocatalysts” in Nature Catalysis (https://www.nature.com/articles/s41929-024-01199-0).