Facet-Dependent Photoreduction on Single ZnO Crystals.

Photocatalytic reactions occur at the crystal-solution interface, and hence specific crystal facet expression and surface defects can play an important role. Here we investigate the structure-related photoreduction at zinc oxide (ZnO) microparticles via integrated light and electron microscopy in combination with silver metal photodeposition. This enables a direct visualization of the photoreduction activity at specific crystallographic features. It is found that silver nanoparticle photodeposition on dumbbell-shaped crystals mainly takes place at the edges of O-terminated (0001̅) polar facets. In contrast, on ZnO microrods photodeposition is more homogeneously distributed with an increased activity at {101̅1̅} facets. Additional time-resolved measurements reveal a direct spatial link between the enhanced photoactivity and increased charge carrier lifetimes. These findings contradict previous observations based on indirect, bulk-scale experiments, assigning the highest photocatalytic activity to polar facets. The presented research demonstrates the need for advanced microscopy techniques to directly probe the location of photocatalytic activity.

Zinc oxide (ZnO) photocatalysts have shown great potential for applications in environmental remediation as they can, for example, be used in wastewater treatment, converting contaminants into harmless substances.[1,2] It is generally accepted that structural and morphological features influence the performance of semiconductor photocatalytic nanomaterials, including ZnO.[3−6] Literature reports synthesis procedures of a wide range of ZnO crystals with different morphologies and sizes, enabling an indirect investigation of structure-dependent photocatalytic activity by specifically expressing certain crystallographic facets.[7−12] In particular, for rod-like ZnO crystals, a lot of effort has been put in changing the ratio of oxygen- (0001̅) or zinc- (0001) terminated polar facets to nonpolar crystal facets. The different crystallographic facets with the corresponding indices are shown for a representative dumbbell-shaped particle in Figure a. In general, it has been found that a higher ratio of polar to nonpolar crystal facets greatly enhances the overall performance of the ZnO photocatalyst.[13−18] Typically, a relative and absolute quantification of polar versus nonpolar crystal facets obtained by electron microscopy is linked to the overall photocatalytic performance determined by monitoring the bulk photodegradation of colored or fluorescent compounds.[10,14−17] Clearly, the knowledge of the facet-dependent photocatalytic performance of ZnO is mostly based on indirect evidence obtained from ensemble-averaged experiments. However, the observed photocatalytic performance does not only depend on the facet abundance; also, the crystal size and the nature and density of crystal defects is of importance. All of these factors are highly intertwined and are readily altered by slight changes to the synthesis conditions. An approach that enables direct visualization of the structure–activity relationship is thus highly needed.

Figure 1

(a,c) SEM images of dumbbell-shaped ZnO crystals obtained before and (b,d) after UV photodeposition of silver nanostructures (20 s of illumination). The principle facets of hexagonally shaped ZnO crystals are indicated with their Miller–Bravais notation in panel a.[13] As the nonpolar side facets of hexagonal ZnO crystals belong to the same equivalent family of lattice planes, they can also be denoted as {101̅0}. The images are obtained with a back-scattered electron detector (BSED). The insets offer a more detailed view on the exact location of the silver photodeposition. Scale bars: 2 μm (a,b), 1 μm (b2), 5 μm (c,d), and 2 μm (d1).

The accessibility of photogenerated charge carriers at the photocatalyst surface can be directly observed by means of the photoreduction of silver ions to silver nanoparticles.[19] The UV photodeposition of silver on ZnO crystals is often used to improve the photocatalytic performance as the silver acts as an efficient trap for the photogenerated electrons, reducing the probability of electron–hole pair recombination.[20−23] However, so far, the facet-dependent photodeposition of silver nanoparticles has not been studied in great detail.[24−26] In this work, the photoinduced silver deposition is explored at the single-particle level for two commonly encountered ZnO structures with different crystallographic facets using an integrated light and electron microscope (iLEM).[27,28] For this purpose, an optical microscope was integrated into a FEI Quanta 250 FEG environmental scanning electron microscope (SEM) using the SECOM platform of Delmic (Scheme S1).[29,30] This iLEM configuration allows the nanoscale observation of ZnO crystals by SEM, before and after UV-induced silver cocatalyst deposition.[28] As such, silver metal deposition acts as a probe reaction to monitor the influence of catalyst structural features on the photoinduced activity.[21,31−33] The combination of SEM with optical microscopy was indispensable in this study, as the required resolution for a proper structural investigation could not be obtained with the optical microscope used to provide the UV irradiation. On the contrary, a correlative approach, in which structural SEM imaging and UV-mediated silver photodeposition is achieved by means of two dedicated setups, is, in principle, possible. However, this requires the sample to be repeatedly transferred between setups, which is time-consuming and increases the risk of sample contamination and losing the regions of interest (ROI).

Hexagonally shaped ZnO crystals (4 μm in diameter × 7 μm in length) were prepared according to the synthesis procedure reported by Wen et al.[34] Representative SEM images of the dumbbell-like ZnO microcrystals prepared via this method are presented in Figure . The electron micrographs in Figure a,c were obtained after the sample was drop-casted onto a cover slide for optical microscopy and prior to the UV photodeposition of silver nanoparticles from an aqueous silver nitrate solution. The crystallographic facets and the presence of structural imperfections on the bare ZnO can be readily observed from these micrographs. Note that the (0001) and (0001̅) facets in these dumbbells were identified through an additional experiment by selectively dissolving the (0001̅) facet in acetic acid (Figure S2).[35]Figure b,d shows the corresponding electron micrographs obtained after UV photodeposition of silver nanoparticles. The photodeposition procedure consisted of venting the sample chamber to atmospheric pressure, adding a 1 mM aqueous silver nitrate solution to the ZnO particles deposited on the glass slide, followed by a 20 s illumination from the bottom side with 365 nm UV light. Afterward, the liquid was carefully removed using tissue paper before pumping down the EM chamber and recording the SEM images of the same photocatalyst particles. Because the surface of the pristine ZnO crystals can be studied from the scanning electron micrograph acquired before silver photodeposition, the actual locations of the photodeposited silver nanoparticles can be related to specific crystal facets or structural imperfections at the ZnO surface, prior to being covered by silver.

Zooming in on Figure b, the deposited silver nanoparticles are readily visible in the back-scattered electron (BSE) micrographs as bright features at the ZnO surface. Clearly, structural imperfections and the crystal edges between the (0001̅) and the lateral {101̅0} facets display a strongly increased tendency to photoreduce silver ions compared with the main (0001̅) and {101̅0} facets themselves. The effect of these structural imperfections is highlighted in the insets that show enlarged images of selected ROI. Similar observations can be made for the ensemble of ZnO crystals shown in Figure c. However, after silver deposition (Figure d), it is revealed that structural imperfections are not homogeneously active. The photodeposition seems to primarily take place at the edges assigned to local O-terminated (0001̅) crystal facets. Nonetheless, some so-called nonpolar planes of the crystals in Figure d show silver deposition that cannot be directly correlated to any structural defect; this photodeposition might be attributed to crystal imperfections that remain unresolved in the recorded electron micrographs. Furthermore, even within individual particles, not all structural defects show equal photoreduction activity.

To minimize the risk for electron-beam-induced structural damage, the beam currents were kept as low as possible. This was also necessary to keep sample charging to a minimum as imaging was performed on nonconducting cover glasses. Still, electrons could remain accumulated on the ZnO particles. Before performing the photocatalytic experiments, the sample chamber was vented, bringing in gas molecules that are known to efficiently neutralize surface charge. As such, the influence of the mild SEM conditions is assumed to have a negligible influence on the outcome of this work.

Another point of attention was the used imaging geometry (Scheme S1) with SEM structural imaging from the top of the ZnO crystal and UV illumination via the optical microscope from the bottom side. Hence, exciton formation preferentially takes place in the bottom 100 nm inside the ZnO crystals. These excitons diffuse through the bulk of the particle toward the crystal surface where reactions can take place. The charge carrier diffusion inside these ZnO crystals (vide infra) exceeds the typical particle thickness used in this study. This reasoning is also supported by our experimental observations. Besides clear photoreduction activity at the top crystal facets, the edge of the (0001̅) in the slightly tilted crystal shown in Figure d does not show a gradient in activity from the bottom to the top.

The structure–activity relationship was quantified in more detail by inspecting 25 individual dumbbell-shaped ZnO crystals from the same batch, before and after UV-induced silver photodeposition. (Five are displayed in Figure ; the others can be consulted in Figure S3.) The correlation of silver photodepositions to specific structural features of the ZnO photocatalyst reveals that ∼50% of the silver nanoparticles are deposited at the outlines of structural imperfections at the nonpolar facets, 45% at the edges of the O-terminated polar facets, and the remaining 5% are distributed randomly at what appear to be defect-free locations. The latter is concluded based on the available structural resolution offered by the recorded scanning electron micrographs. This observation is remarkable because literature mostly suggests that the polar (0001) or (0001̅) crystal facets are the most photocatalytically active.[10,14−17]

To further rationalize this observation, we turned to literature where the structural characteristics of ZnO crystals have been described in detail.[36−38] Instead of the simplified representation of hexagonally shaped ZnO crystals that is often used in photocatalytic studies (Figure a), extensive investigations have been performed to determine the intrinsic crystal facets of pyramidal-shaped ZnO crystals. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) revealed that the {0001} basal planes are almost atomically flat, containing a very small amount of defects, while the pyramidal planes consist of alternating {101̅0} and {101̅1} crystal facets (Figure a); the semipolar {101̅1} facets are oxygen-terminated.[36,37] From the average angle of 85° between the pyramidal facets and the polar (0001) planes (Figure S3), it is estimated that ∼15% of the total area of the pyramidal plane displays a semipolar {101̅1} character because pure {101̅1} facets should make a 62° angle, while perfect {101̅0} facets should be at exactly 90°. Furthermore, scanning tunneling microscopy has revealed an additional roughness at the pyramidal ZnO planes due to the presence of horizontally oriented grooves.[38] Besides the semipolar and nonpolar facets, these surface imperfections feature additional polar Zn- and O-terminated facets. Figure b displays a schematic representation of these different crystal facets at a structural imperfection on the pyramidal plane. The enhanced photoactivity of semipolar {101̅1} pyramidal crystal facets has previously been proposed by Chang et al. based on the bulk performance of different ZnO crystal morphologies.[13] Hydroxylation of the under-coordinated oxygen facets would enhance electron diffusion toward these facets.[35,39,40]

Figure 2

(a) Miller–Bravais indices of the principle facets of the schematic hexagonally shaped ZnO crystal are shown.[13] The enlarged view gives a detailed picture of the pyramidal facets in dumbbell-shaped crystals consisting of alternating semipolar {101̅1} and nonpolar {101̅0} crystal facets. Their relative contribution was calculated based on the distribution of the pyramidal α-value measured for 30 dumbbell-shaped ZnO crystals (Figure S3). (b) Besides semipolar {101̅1} and nonpolar {101̅0} crystal facets, pits and grooves feature polar Zn- and O-terminated facets. The blue circles highlight the most active silver photodeposition sites as found in this study. Scale bar: 2 μm. (c) Yellow shade highlights the zones with elevated photoreduction activity at the outline of the O-terminated (0001̅) facets at the crystal extremes as well as in the structural imperfections.

Differences in charge-carrier mobility were assessed via spatially and temporally resolved photoluminescence (PL) measurements on individual ZnO crystals. It has previously been reported that exciton lifetimes in ZnO vary with crystal morphology.[41,42] Moreover, the results presented here clearly show spatial variations in charge carrier lifetimes within individual ZnO crystals. Figure a shows that the PL decay strongly depends on the position along the c axis of the dumbbell-shaped crystal with the longest lifetimes recorded near the oxygen-terminated (0001̅) facet. As such, this effect seems to be directly correlated to the crystal length, as confirmed by plotting the relative increase in PL decay times near the O-terminated (0001̅) plane versus the PL decay time at the Zn-terminated (0001) plane as a function of the distance between the O- and Zn-terminated polar planes (Figure c). The longer charge-carrier lifetimes detected at the crystallographic edges of the O-terminated (0001̅) crystal faces and at the outlines of structural defects correspond to longer free carrier diffusion lengths (LD) according to LD ≈ (D τ)1/2, in which τ represents the typical PL lifetime of a few nanoseconds for ZnO and D represents the diffusion coefficient. The charge carrier diffusion length in ZnO can be estimated to be a few micrometers; the effective diffusion coefficient for ZnO rods is 0.5 cm2 s–1, as obtained from literature.[42] This additionally confirms the validity of our previous assumptions that electrons formed near the bottom surface of the ZnO crystals can indeed diffuse through the bulk of the particles to reach the top surface and result in silver nanoparticle photodeposition.

Figure 3

Spatially and temporally resolved PL measurements on individual ZnO crystals after excitation with a picosecond-pulsed 405 nm laser using a confocal microscope. Distribution map of free charge carrier lifetimes for (a) three dumbbell-shaped ZnO crystals and (b) two ZnO microrods. (c) Correlation of the relative increase in PL decay times near the O-terminated polar plane versus the Zn-terminated plane in function of the distance between both crystallographic facets.

Note that because of practical limitations, PL lifetime experiments were performed using 405 nm laser excitation and visible PL (430–850 nm) detection. Hence these results are related to crystal imperfections rather than exciton emission. The significant correlation between the structural defect PL lifetime and the zones with enhanced silver photodeposition are undeniable. Figure S6 shows the related emission spectra recorded at the different positions along the ZnO crystals. The positions with enhanced lifetime are related to red-shifted emission, which is typically assigned to zones with increased concentrations of crystallographic imperfections. Because the role of defects on exciton lifetime and photocatalytic performance has already been actively discussed in literature, this observation can further be investigated by UV excitation. The blue circles in Figure b highlight the most active silver photodeposition sites at the edges of O-terminated polar faces occurring at the top and bottom planes of the crystal as well as in structural imperfections.

Interestingly, rod-like ZnO crystals are sometimes encountered in a batch of dumbbell-shaped crystals (see right side of Figure S4a,b). These particles display a markedly different photoreduction performance compared with the dumbbell-shaped ZnO crystals, as a more homogeneous silver nanoparticle photodeposition (inset Figure S4a.1) is observed on the seemingly defect-free crystal side faces (inset Figure S4b.1). To further investigate the activity of such ZnO microrods with respect to photoinduced silver deposition, a batch of ZnO microrods was synthesized as described by Shoja Razavi et al.[43] In Figure , electron micrographs of microrods before and after silver nanoparticle photodeposition are shown; the experimental procedure was identical to that for the dumbbell-shaped particles described before. Clearly, at the current resolution of the scanning electron micrographs, structural imperfections are less commonly encountered compared with the dumbbell-shaped sample. However, the observation that silver photodeposition occurs more homogeneously at crystal edges and {101̅0} facets of the rod-like ZnO crystals is confirmed in this batch (Figure b,c). Importantly, an additional crystal facet seems to have developed in these crystals. At the boundary between the (0001̅) and the {101̅0} side facet, {101̅1̅} facets can be recognized. This semipolar {101̅1̅} Zn-terminated facet has previously been observed as a stepwise transition of the polar (0001̅) O-terminated face into the pyramidal {101̅0} facets in a stack of ZnO nanopyramids.[37] Ensemble photocatalytic activity measurements of these complex nanopyramids indicated that these facets lead to an increased photocatalytic activity. Our photodeposition measurements support this hypothesis, as the largest silver nanoparticle deposits are observed at these facets. Similar to the ZnO dumbbells, time-resolved PL studies on individual ZnO microrods (Figure b) demonstrate increased free charge-carrier lifetimes near the O-terminated polar facets, indicating an enhanced charge-carrier diffusion toward these crystal facets.

Figure 4

(a) SEM image of a ZnO microrod and schematic representation displaying the crystallographic Zn-terminated {101̅1̅} semipolar facets adjacent to the O-terminated polar face. (b) SEM image of a ZnO microrod obtained before and (c) after 20 s of UV-induced photodeposition of silver nanostructures. The insets c1 and c2 reveal that the boundary areas of the polar (0001̅) crystal faces toward the {101̅1̅} facets display an increased photoinduced silver deposition. Scale bars: 1 μm.

In their pioneering work, Pacholski et al.[24] demonstrated that the photodeposition of silver on very small ZnO nanocrystals occurred at one particular tip, which was assigned to be the O-terminated polar face. Looking closely at these electron micrographs and keeping in mind the observations made in this study, the silver nanoparticles do not seem to be located at the O-terminated polar tip itself but at what appears to be the {101̅1̅} facet.

To correctly relate silver nanoparticle photodeposition to facet-dependent electron accessibility, any contribution of preferential silver cation chemisorption at these facets should be ruled out, as this would locally enhance photodeposition. To minimize the effect of preferential electrostatic attraction of silver cations, the described experiments were performed at pH 10 as the bulk isoelectric point of ZnO was determined to be at pH 8.7 to 10.3.[35] Protonation of the O-terminated polar face was reported to induce a downward band bending, decreasing the recombination probability of the photogenerated charges and inducing a directional electron diffusion toward this polar facet.[35,40] This model has been confirmed by our experimental results, as they reveal longer charge carrier lifetimes at the outlines of O-terminated (0001̅) facets and structural imperfections, as previously discussed (Figure ).

This study further proves that the photocatalytic activity of ZnO particles is the result of several factors. Via the applied iLEM approach, crystal facets and surface imperfections can straightforwardly be related to the photocatalytic activity. As such, differences in silver deposition between the ZnO dumbbells and microrods reveal that the edge of the O-terminated (0001̅) facet shows an enhanced photocatalytic activity. In contrast with the dumbbell-shaped crystals where this edge is very sharp, the activity can directly be linked to a clearly developed {101̅1̅} facet in the microrod sample. Furthermore, the activity of the {101̅0} side facets in the dumbbell-shaped crystals can straightforwardly be associated with structural imperfections and more specifically to the (0001̅) facet at these sites. In the microrods, no surface imperfections could be resolved in the obtained scanning electron micrographs, so this link cannot be directly made. However, we assume that crystallographic imperfections and the related crystal facets are at the origin of the observed photocatalytic deposition. Note that on a bulk scale both ZnO samples showed a similar activity in the photoreduction of resazurin to the fluorescent resorufin. This further supports the idea that many important aspects that determine the overall photocatalytic experiments are masked in a typical bulk activity measurement and that rational optimization relies on the in-depth insights that can be generated via microscopic observations.

By using UV photoreduction of silver ions as probe reaction, correlative light and electron microscopy experiments have revealed a notable intraparticle heterogeneity in photoreduction activity for ZnO crystals. On dumbbell-shaped ZnO crystals, about half of the silver photodeposition occurs at the crystallographic edges of the O-terminated (0001̅) polar planes, whereas the other half is observed at the O-terminated facets of structural imperfections. Identical experiments performed on seemingly defect-free ZnO microrods confirmed that the edges of O-terminated polar facets, outlined with {101̅1̅} facets, display the highest photodeposition activity. Until now, the most active crystal facets have been determined by monitoring the bulk-scale photodecomposition efficiency of ZnO crystals with different sizes and morphologies.[10,14−17] However, such a bulk-scale approach does not provide a direct structure–activity link because only idealized crystals consisting of defect-free crystal facets are considered. This work reveals that the influence of structural imperfections on the photocatalytic performance should not be neglected. The presented approach, that is, detailed structural characterization of individual photocatalysts using SEM, both before and after performing photodeposition, is crucial to correctly rationalize the activity pattern. Besides photoreduction, a lead- or manganese-based approach could be applied to reveal locations with enhanced photo-oxidation activity. Alternatively, fluorogenic reactions could enable an investigation of the structure–activity relationship on the nanoscale. In general, spatially and temporally resolved studies will greatly contribute to a better understanding, further optimization and even rationalization of catalyst design. Furthermore, the observed intraparticle distribution of charge-carrier PL decays is also of interest in the field of composite mesocrystals. Exact knowledge of the surface and internal structures is necessary to achieve efficient charge-transfer processes, which are important for catalysis, optoelectronics, sensing, and energy conversion.[44]