Imaging Luminescent Traps on Single Anatase ... - ACS Publications

Article pubs.acs.org/JPCC

Imaging Luminescent Traps on Single Anatase TiO2 Crystals: The Influence of Surface Capping on Photoluminescence and Charge Transport Riley E. Rex, Fritz J. Knorr, and Jeanne L. McHale* Materials Science and Engineering Program and Department of Chemistry, Washington State University, Pullman, Washington 99164-4630, United States S Supporting Information *

ABSTRACT: The spatial distribution of intra band gap traps in micrometer-sized single crystals of anatase TiO2 was explored using single-particle photoluminescence (PL) spectroscopy and imaging. The PL from microcrystals with well-defined {001} and {101} facets was imaged for the same particle before and after annealing to explore the influence of fluorine, used as a capping agent in the synthesis of the microparticles, on luminescent traps. Unannealed particles reveal weaker photoluminescence and distinctly different spatial distribution of PL emission compared to annealed particles. The results show that the capped particles have fewer surface defects as a result of passivation of surface electron traps associated with undercoordinated titanium. PL images suggest that the remaining surface defects in the fluorine-capped particles concentrate at edges and corners of the microcrystal, and that anisotropic carrier transport takes place via hopping between adjacent Ti atoms. Annealed particles reveal greater surface defect density and more isotropic, diffusional carrier transport than the as-synthesized particles. The results are interpreted in terms of the influence of fluorine-capping groups on the spatial distribution and occupancies of traps and different pathways for carrier transport. The results have implications for applications of TiO2 nanosheets in solar energy conversion and photocatalysis.

■

increasing percentage of fluorine-free surface {001} facets.9 In the case of perovskite/TiO2 heterojunction solar cells, cells containing TiO2 particles with dominant fluorine-free {001} surfaces achieved power conversion efficiencies nearly double that of cells using more conventional {101} dominant particles.7 The improvements were attributed to stronger interfacial connection between {001}-TiO2 and perovskite7 and impedance spectroscopy measurements suggest a lower chargetransfer resistance between {001}-TiO2 and perovskite.8 The above examples show that differences in the atomic environment on different particle surfaces can significantly influence device performance. Important factors include the density and identity of surface trap states as well as the presence of capping agents. Efforts to understand the spatial and energetic distribution and the chemistry of these trap states have included theoretical calculations,12 electron paramagnetic resonance studies,13 electrochemistry,14 and various spectroscopic methods.15−17 In our previous work, we used photoluminescence (PL) spectroscopy to probe luminescent traps on anatase TiO2 as a function of particle morphology18 and contacting environment19 and with Fermi-level control.20,21 From these results the broad visible PL of anatase TiO2 was

INTRODUCTION Nanostructured titanium dioxide is widely researched as a promising material for photocatalytic and photovoltaic applications.1,2 Under typical synthesis conditions, anatase nanoparticles grow with over 90% {101} surface facets to minimize surface energy.3 However, capping agents can be used in particle synthesis to expose higher energy surfaces which may be favorable for many applications.4 For example, using fluorine as a capping agent can reduce the surface energy of {001} from its clean surface value of 0.90 J/m2 to below that of {101}, which has an energy of 0.44 J/m2 for a clean surface.5,6 Fluorine works as a capping agent by binding to the undercoordinated (fivefold) surface Ti atoms, which are twice as prevalent on stoichiometric {001} surfaces compared to {101}.6 After synthesis, a heating process can remove fluorine to expose the reactive undercoordinated Ti sites.6 Ever since the development of titanium dioxide crystals with a high percentage of reactive {001} surfaces,6 much work has been done applying these novel crystals in photovoltaics,7,8 solar water splitting,9 and degradation of environmental pollutants.10,11 In catalyzing the breakdown of methyl orange, nanoparticles with predominately {001} surface facets showed degradation rates up to 9 times faster than a P25 standard, with the degradation rate increasing with increasing percentage of {001}.10 That study also showed an increase in catalytic efficiency after removing the surface-capping fluorine.10 For water splitting, the rate of hydrogen production increased with © 2015 American Chemical Society

Received: September 15, 2015 Revised: October 23, 2015 Published: October 27, 2015 26212

DOI: 10.1021/acs.jpcc.5b09005 J. Phys. Chem. C 2015, 119, 26212−26218

Article

The Journal of Physical Chemistry C

correlation between scanning electron microscopy (SEM) and optical microscopy. The grid pattern enabled the exact same microcrystal to be interrogated before and after annealing to remove fluorine. Spectroscopic experiments were conducted on the as-grown particles and again after annealing at 600 °C. The annealing procedure was previously shown6,18 using X-ray photoelectron spectroscopy to remove fluorine. The as-grown and annealed samples were confirmed anatase phase using Raman spectroscopy (see Figure S1 of the Supporting Information). Since the micrometer-size particles prepared here cannot be made in sufficient quantities for the determination of optical spectra, absorption spectra of similar TiO2 nanosheets (see ref 20) were measured before and after annealing. These are presented in Figure S2 and confirm that there is negligible difference in the optical spectra. Microscopy and Spectroscopy. Excitation at 350.7 nm was provided by a BeamLok 2060 Spectra-Physics krypton-ion laser. The beam was passed through a 350 nm interference band-pass filter to remove plasma lines and directed into an Olympus IX70 inverted microscope equipped with a 100× objective. The laser power was approximately 0.1 mW before entering the microscope. The emitted light was passed through a 385 nm long-pass filter before being directed toward the spectrometer or the imaging camera. Exposure time for images and spectra was 4 s. To separate the green and red components of the spectrum, some spectra were obtained with an additional filter between the sample and detectors. For green and red components, a green band-pass filter (450−600 nm, Schott VG-6) and a 610 nm long-pass filter were used, respectively. Spectra were obtained using an Acton SpectroPro 2300i with an attached thermoelectrically cooled CCD from Princeton Instruments (SPEC 10:256E). Images were obtained using an Andor Clara CCD and Andor Solis software. Spatial calibration of optical images was done using a reticle with 10 μm divisions.

concluded to be a superposition of emission from three types of traps: e−CB + h +trapped → hνgreen

(1)

e−trapped + h+VB → hνyellow/red

(2)

PL peaking in the green is attributed to recombination of mobile electrons with trapped holes, while two distributions of trapped electrons recombine with valence band holes to give yellow and red PL. We have hypothesized that the trapped electrons and holes primarily reside on {001} facets at fivecoordinated Ti sites and on {101} at oxygen vacancy sites, respectively. This hypothesis is tested in the present work by using microscopy and photoluminescence spectroscopy to directly image the location of PL on anatase single crystals. The micrometer-sized anatase crystals used in the present study, designated “microsheets”, are intended to serve as a proxy for nanoparticles and nanosheets while being large enough to image using optical microscopy. Previous work on the spectroscopy of microscopic single particles has revealed useful information about charge carrier dynamics in ZnO rods,22 charge transport in TiO2 nanotubes23 and nanowires,24 and facet-dependent redox behavior on anatase microsheets.25 The studies of facet-dependent redox behavior suggest that photoreduction occurs preferentially on anatase {101} and photo-oxidation on {001}-exposed surfaces.25−27 However, the role of spatially isolated trap sites in this effect is uncertain. Understanding the spatial distribution of trap states is important to optimizing particle morphology for a particular application. At the same time, the effect of annealing to remove surface modifiers is an important aspect of understanding the link between particle morphology and performance in photocatalysis and solar energy conversion. In the present work, in addition to testing our working model of anatase PL by imaging photoluminescence of single micrometer-sized anatase crystals with well-defined surface facets and microsheet morphology, we explore the influence of fluorine-capping agents on surface defects. The following work clarifies the character of {101} or {001} facets as possible electron or hole sinks for redox reactions and the role of undercoordinated surface Ti sites in electron trapping. Considering the general practice of removing the surface caps to expose the active {001} surface, it is important to understand the effect of annealing on defects by comparing the PL of fluorine-capped and clean anatase microsheet crystals.

■

RESULTS AND DISCUSSION SEM Images. Several different particles and groups of particles in various orientations were examined throughout the study. Examples of such particles and orientations are shown in Figure 1. Typical microsheets are 3.2−3.7 μm across and 2 μm thick. We estimate the percentage of surface {001} facets at 37%. Particles positioned such that the {001} facet is parallel to the coverslip are labeled as being in a “horizontal” orientation (Figure 1a) and particles with {001} approximately perpendicular to the coverslip are labeled as “vertical” (Figure 1b). Raman spectroscopy (Figure S1) confirms the particles are in the anatase phase. Photoluminescence Imaging. The gridded coverslip allowed for correlation between the SEM and optical microscope to find particles for PL imaging and spectroscopy.Figure 2 shows images of PL from the exact same particle illuminated at various locations before and after annealing at 600 °C to remove the surface fluorine attached to undercoordinated Ti sites.6 In Figure 2a we see that illuminating in the center of an unannealed sample results in PL at corners and edges as well as PL from the laser spot, which spreads directionally toward the nearest edge. Illumination near the corner of an unannealed sample, as seen in Figure 2b, results in PL bright spots at four corners, approximately 2.4 μm apart, which is similar to the distance between the corners and edges where {101} and {001} surfaces meet. Illumination along a {101} side, as in Figure 2c, results in PL directly across the microsheet from the illumination spot as well as at the corners.

■

EXPERIMENTAL SECTION Materials. The anatase microsheets were synthesized following the method of Yang et al.6 Deionized (DI) water was adjusted to pH 2.08 with 1.5 M HCl and used to prepare a stock solution of 0.051 M TiF4. Then the stock solution was diluted to 5.33 mM using pH 2.08 water. Finally, 20 mL of 5.33 mM TiF4 solution was mixed with 245 μL of 10% (w/w) hydrofluoric acid in a Teflon-lined stainless steel autoclave and held at 175 °C for 17 h. Caution: Hydrof luoric acid is extremely corrosive and a contact poison; it should be handled with extreme care. The resulting white precipitate was then washed with DI water in repeated cycles of centrifugation and decanting. The as-grown particles were then drop-cast onto ultrathin quartz coverslips from Chemglass (CGQ-0660-01). The coverslips had been previously engraved in a grid pattern using a diamond-tipped cutter to aid in particle identification and 26213


Article


observed for all particles and groups of particles observed in this study; several other examples can be seen in Figure S3. The difference in spatial PL between unannealed and annealed microsheets is also seen using wide-field excitation to illuminate the entire particle. As shown in Figure 3, wide-field illumination

Figure 3. Unannealed crystal (a) and annealed crystal (b) illuminated by wide-field 350 nm excitation. The black boxes in (a) and (b) represent the approximate location of the intersection of {001} and {101} surfaces as determined by SEM and spatial calibration of optical images. The scale bar indicates 3 μm.

results in brighter PL from the edges than the center in unannealed microsheets and brighter PL from the center than from the edges in annealed microsheets where fluorine has been removed. With use of the spatial calibration from the optical images and comparison to the known particle dimensions from SEM, the square of brightest PL seen in Figure 3a is determined to be approximately 2.2 ± 0.2 μm across, corresponding closely to the size of the {001} surface of this particle. Figure 4 shows spectra of the PL imaged in Figure 2a,d. The observed broad visible emission, with a peak in the red, is

Figure 1. SEM micrographs of an example of anatase crystals and groups of particles examined in this study. The particle in (a) is designated as horizontal and (b) is designated vertical. All scale bars indicate 5 μm.

Figure 2. Images of PL from the same particle illuminated at various spots before (a−c) and after (d−f) annealing at 600 °C. The false color is representing the most intense emission as white and the lowest intensity as black with the gradient as shown in the scale bar at the right. All intensities have been normalized. The most intense spot indicates the location of the incident 350 nm laser. The scale bar is 4 μm.

Figure 4. PL spectra of an unannealed and annealed particle corresponding to Figure 2a and Figure 2d, respectively. The PL intensity of annealed particles is typically about 5 times stronger than that of unannealed particles.

The brightest regions here, from the illumination side to the remote side, are approximately 2.4 μm apart, suggesting the PL from unannealed samples is concentrated at the intersection of {101} and {001} facets. The remote PL observed in Figure 2a− c reveals preferential carrier transport in directions parallel to the crystallographic a and b directions. After annealing, the spatial distribution of PL changes from being concentrated at the edges to simply diminishing around the excitation spot in a Gaussian-like manner, as expected for diffusion-driven charge carriers. This is seen in Figure 2d with center illumination as well as Figure 2e and 2f with corner and edge illumination. The marked difference in the spatial PL distribution between unannealed and annealed particles was

typical of anatase PL.18−21 For comparable illumination spots, the intensity of PL increases by a factor of about 5 after annealing. This evidence suggests the previously fluorinecapped Ti surface sites on {001} play a significant role in the broad red anatase emission, in agreement with previous work.18−21 This result is consistent with the passivation of electron traps by fluorine.28,29 To test the influence of the incident beam location and to get various perspectives, we imaged the PL from particles in 26214


Article


precedence for observing spatial modulation in PL intensity due to optical resonance in single-particle PL imaging.22 House et al.22 observed a spatial pattern of PL on ZnO rods for experiments with scanning excitation but none was observed for single-point excitation as we are doing here. Regardless, we made several efforts to check whether physical differences between annealed and unannealed particles could account for only the latter acting as an optical resonator. First, Raman spectroscopy was used to check the phase and degree of crystallization of particles before and after annealing (see Figure S1). While annealed particles are shown to be more crystalline than unannealed ones, the Raman results indicate that all particles are well-crystallized anatase and any possible waveguiding or optical cavity behavior cannot be attributed to a phase difference or a large change in crystallinity. Second, many annealed and unannealed particles were carefully examined with SEM to look for differences in surface roughness. No significant increase in roughness as a result of the annealing process was observed (see Figure S4). In addition, though scattering and internal reflection of the emitted light might have resulted in the observation of bright PL remote from the laser focus, there is no reason to expect such phenomena to be different for the annealed and unannealed particles, which have the same shape and optical spectrum. Lastly, we considered the possibility that a difference in the interface between quartz substrate and TiO2 particle caused the optical cavity behavior to disappear after annealing. To test this, we altered the experiment setup so that the incident and emitted light beams were no longer passing through the grid substrate between the objective and sample particle. Instead, the sample grid was flipped over and placed sample-side down on a clean quartz coverslip to ensure the path of illumination and collection between objective to particle remained the same for annealed and unannealed samples. After this rearrangement, the results remained as they were before, with an obvious spatial modulation in PL intensity from unnannealed particles and a typical, apparently diffusiondetermined, spatial distribution of PL from annealed particles. With no clear evidence to support the existence of optical cavity modes, we should consider the possibility that the spatial distribution of PL in unannealed particles is due to directional charge carrier transport and trap distributions which favor edges and corners. With this view, we interpret the above results as follows. The broad PL, centered in the red, is primarily the result of valence band holes recombining with electrons trapped by five-coordinated Ti as Ti3+. The passivation of these traps by fluorine on the as-prepared particles effectively prevents these sites from trapping electrons. When the unannealed particles are illuminated in the center of the {001} facet, PL images with fourfold symmetry are observed, resembling four-leaf clovers with leaves aligned in the [100] and [010] directions. The photoholes and photoelectrons created in the vicinity of the laser focus have a driving force to diffuse outward. Electrons transport via polaron hopping through neighboring Ti4+, and at the same time holes can undergo free diffusion in the valence band. We speculate that the hopping transport of electrons is somewhat more favorable in the directions along the equivalent crystallographic a and b directions than in a diagonal direction owing to shorter near-neighbor distances. Additionally, there may be a driving force for electrons to move from {001} toward {101} due to the latter having a lower conduction band energy.30 The superposition of this directional hopping with more isotropic diffusional pathways of holes could reasonably explain the

horizontal and vertical orientations. The PL from a vertically oriented particle can be seen in Figure 5 where the excitation is

Figure 5. Image of PL (a) from an unannealed particle in vertical orientation. The same particle is seen in SEM in (b). The curved shape of the PL emission shown in the PL image suggests that radiative recombination is occurring primarily at the corners and edges. The scale bar in (b) measures 3.7 μm.

incident on the upper right side of the particle and emission is concentrated along the right edge and remotely along the left edge. This view supports the contention that PL of unannealed particles is primarily occurring near corners and less on the {001} or {101} faces. The appearance of remote PL at the edge opposite rather than adjacent to the edge where the laser was focused again indicates preferential carrier transport in the crystallographic a and b directions. In addition to imaging and measuring spectra of single particles, groups of particles were examined to look for evidence of interparticle charge carrier transport. The groups of agglomerated particles are likely the result of particles growing together during synthesis, as can be seen by SEM in Figure S4; thus, electrical connection between particles is expected. Figure 6 shows examples of remote PL occurring on

Figure 6. Image of PL emission from a group of unannealed particles (a). Interparticle charge carrier transport is clearly seen with radiative recombination occurring along the edges of particles several micrometers away from the excitation spot. The image of PL emission from annealed particles (b) shows remote PL emitted from neighboring particles with no preference for corners or edges. The scale bar indicates 3 μm.

neighboring particles in unannealed and annealed samples. In both cases, PL is observed several micrometers and several particles away from the illumination spot. The remote PL from the unannealed cluster shows the same phenomenon observed in single particles, namely, that PL occurs most strongly from edges and corners. The fact that this happens on particles not directly excited by the incident beam supports the idea that the spatial distribution of PL is influenced by charge carrier transport and the spatial distribution of radiative recombination sites rather than optical resonator modes. However, there is 26215


Article


The fivefold increase in PL from annealed microsheets compared to unannealed indicates that photogenerated charge carriers are not as readily able to access surface traps in Fcapped particles. This is consistent with as-grown particles having much lower photocatalytic efficiency than F-free particles as demonstrated in ref 10. The evidence here suggests that the surface traps responsible for red PL in anatase are the same traps participating in photocatalysis. Now we turn to a discussion of our hypothesis that the broad anatase PL is composed of two distinct types of transitions, namely, trapped electrons recombining with valence band holes to produce yellow/red PL and trapped holes recombining with conduction band electrons to produce green PL. On the basis of this model, and from the assumptions that the PL spectrum from anatase microsheets (Figure 4) has three components20 and that the diffusion lengths of holes and electrons are different, we expect to observe a difference in the spatial distribution between the PL at the green edge of the spectrum and that on the red side. To test this, we compared singleparticle PL imaged through a 610 nm long-pass filter to that imaged through a green band-pass filter (450−600 nm). The results in Figure 7 show that, for excitation in the center of an

fourfold symmetry of the PL images when as-prepared particles are illuminated in the center of the {001} surface. Clearly, the PL of the as-prepared particles, though dimmer overall than annealed particles, is brighter at corners and edges, reflecting a population of Ti4+/3+ that remains undercoordinated at the intersection of facets. Furthermore, when the laser is aimed at a defect-rich corner or edge, we observe PL from the illumination spot as well as directional transport to opposite corners and edges where remote PL maps the locations of the highest concentration of pre-existing electron traps. The weak emission observed from {001} while illuminated at an edge (e.g., see Figure 2c) possibly indicates a small population of residual, uncapped Ti sites that are more readily observed by exciting on {001} directly. Unannealed particles can be concluded to have fewer surface defects and the unfluorinated sites that exhibit PL are concentrated on the edges and corners. The resulting images, averaged over a time (4 s) which is long compared to that of carrier transport, result in a pattern which reflects both the directionality of the transport and the distribution of trap states. The annealed particles show a different PL distribution as a result of increased density of defects at surface Ti atoms compared to capped particles. In addition, surface reconstruction may occur when the capping agents are removed, leading to microfacets and surface roughening.31,32 Luminescent defects are obviously concentrated at corners and edges in the as-prepared microsheets. Thus, the additional faceting that occurs on 1 × 4 surface reconstruction, predicted to occur on removal of fluorine,31,32 would introduce many more trap states and lead to the observed Gaussian distribution of PL centered about the laser focus. In the absence of fluorine, the Fermi level is high enough that many of the surface Ti are already present as Ti3+ in the absence of UV illumination,28,31,33 and emission occurs when these previously trapped electrons recombine with photoholes in the valence band. Diffusion of the valence band holes away from the spot where they are created during the exposure time of the image thus gives the Gaussian distribution of PL. To support these concepts, we looked for values of the diffusion length of electrons and holes in bulk anatase crystal. The diffusion length L is a function of the diffusion coefficient D and the lifetime τ of carriers: Ln = (Dnτn)1/2 for electrons and Lp = (Dpτp)1/2 for holes. The diffusion coefficient D is in turn related to the mobility μ through μ = qD/kBT. The electron diffusion length in nanocrystalline anatase has been estimated to be in the range 7−32 μm,34 and though it is assumed that diffusion is faster in the bulk crystal,35 the carrier lifetime may be very different as well. Reference 36 used time-resolved PL, transient absorption, and photoconductivity to determine the lifetime of electrons and holes in bulk anatase, from which they estimated the diffusion length Ln to be 10 μm. Significantly, this estimate was based on pulsed laser excitation with average powers several orders of magnitude larger than those used in this work. Such high powers result in enhanced recombination and shorten the carrier lifetimes. Thus, the diffusion length Ln in the present study might be even longer than 10 μm. The diffusion length of holes in bulk anatase is, to our knowledge, not available in the literature. If effective masses of holes and electrons are similar,36,37 then barring a large difference in the lifetimes τn and τp the diffusion lengths of holes would also be on the order of several micrometers. Thus, the extent of the remote PL observed in the images is reasonable in view of the expected mobilities and lifetimes of the carriers.

Figure 7. PL images of an annealed particle imaged through a 610 nm long-pass filter (a) and a green band-pass filter (b). Image (a) corresponds to the red trace spectrum in (c) and image (b) corresponds to the blue trace spectrum in (c). The white lines in (a) and (b) indicate the sample location for plotting PL intensity versus distance in (d). The ratio of the remote PL 1.5 μm away from the maximum to that at the maximum intensity is approximately 0.16 for red PL and 0.11 for green PL. Note that 19 pixels is approximately 2 μm.

annealed microsheet, the contribution of the green component to remote PL near the particle edge is approximately 30% less than the contribution of the red component. Similar results were observed for green and red components of PL from unannealed particles (see Figure S5). The results here are suggestive of red and green PL components resulting from distinct mechanisms. It should be noted that there is possible overlap between the components of anatase PL20 and thus the separation into green and red shown here is only an approximate sampling. Furthermore, the 1 s integration time used to obtain images of Figure 7a,b is very long compared to transport times, making it difficult to observe transport 26216


Article


states on anatase microsheets with well-defined {001} and {101} surface facets. For as-grown particles, where the majority of undercoordinated surface Ti atoms are capped by fluorine, the PL is concentrated along edges and corners where {101} and {001} facets intersect. Additionally, the spatial pattern of PL from F-capped particles suggests anisotropic transport of photogenerated electrons, possibly via polaron hopping across nearest neighbor Ti atoms. After heat treatment to remove surface fluorine from surface Ti atoms, the microsheet PL increases intensity fivefold, is no longer concentrated at the edges and corners, and charge transport, as evidenced by remote PL, appears to be diffusional. The removal of fluorine significantly increases the population of trapped electrons at surface Ti3+ sites which act as radiative recombination centers in the presence of valence band holes. It is possible these surface electron traps, which are more prevalent on {001} than on {101}, are critical for the surface reactivity needed for effective sensitization and photocatalysis.

dynamics. To further understand the differences between charge carrier transport in capped and clean anatase particles, it would be useful to apply time-resolved photoluminescence imaging. A dynamics study may also elucidate the recombination mechanisms responsible for the different components of anatase PL. In past work, we have observed a transition from red/yellow PL to green PL of anatase nanoparticles in the presence of hole scavengers such as ethanol,19 in conventional nanoparticles and nanosheets of TiO2 at negative applied potential,20,21 on vacuum annealing,38 and on increase of the power of incident UV illumination.23 We have also previously reported that in mixed phase (anatase−rutile) P25 nanoparticles only green PL is observed, suggesting that the red- and yellow-emitting traps are passivated.39 In the present study, neither increased laser power nor exposure to ethanol succeeded in affecting a significant transition from red/yellow to green PL. In addition, we observed only a small difference in PL spectra upon using illumination mostly focused on {101} versus {001} facets (see Figure S6). Illumination on {001} results in PL only slightly blue-shifted compared to illumination of {101}, which is consistent with observations by Tachikawa and co-workers.25 The present work does not support the hypothesis that there is a clear separation of luminescent hole- and electron-trapping sites on different facets. However, the enhanced redox properties of TiO2 nanosheets are often attributed to the difference in the positions of the conduction and valence bands of the {001} and {101} surfaces.40 In this picture, slightly higher energy conduction and valence band edges for {001} than for {101} surfaces favor the sequestration of holes on the former and electrons on the latter. When a microsheet is illuminated in the center of the {001} surface, this creates a concentration gradient for both electrons and holes to diffuse outward. Electrons may have longer diffusion lengths than holes and additionally there is a driving force for them to undergo transport from the {001} to the {101} surface. In the as-prepared, F-capped microsheets, there are few trapping sites and thus photoelectrons created within the focal volume do not all recombine there and can undergo hopping mediated by shallow traps until finding vacant trap sites where they are captured and imaged whenever they recombine with the diffusing photoholes. Annealed microsheets, on the other hand, have a high density of pre-existing trapped electrons, resulting in PL images which are concentrated around the focal region rather than spread out toward regions of higher trap density. In either case, the observation that a slightly larger spread in the PL is seen at red and yellow wavelengths than at green wavelengths appears to be in accord with our assignments of the PL to electron and hole traps, respectively, and the greater extent of electron as compared to hole transport. The difference is small in accord with only a small contribution of green PL, suggesting that this component is manifested more in samples with higher surface-area to volume ratio and/or more interparticle connections. It may be that the difficulty to observe green PL in the present study, as compared to previous work using nanosheets20 and nanoparticles,18−21 is the result of fewer hole traps on the microsheets. We have associated these hole traps with oxygen vacancies, which may be less prevalent in the microsheets studied here.

■

ASSOCIATED CONTENT

S Supporting Information *

Raman spectra of anatase crystals before and after annealing (Figure S1); absorbance spectra of anatase nanosheet films before and after annealing (Figure S2); photoluminescence (PL) images of anatase crystals before and after annealing (Figure S3); SEM images of anatase crystals before and after annealing (Figure S4); PL images and spectra through 610 nm long-pass and green band-pass filters (Figure S5); PL images and spectra of {101}- and {001}-illuminated particles (Figure S6). The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ acs.jpcc.5b09005. (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to acknowledge the use of the Franceschi Microscopy & Imaging Center at Washington State University for electron microscopy work. We also thank Dr. Matthew McCluskey for a useful discussion on optical imaging. The support of the National Science Foundation (DMR1305592) is gratefully acknowledged.

■

REFERENCES

(1) Fujishima, A.; Honda, K. Electrochemical Photolysis of Water at a Semiconductor Electrode. Nature 1972, 238, 37−38. (2) O’Regan, B.; Grätzel, M. A Low-Cost, High Efficiency Solar Cell Based on Dye-Sensitized Colloidal TiO2 Films. Nature 1991, 353, 737−740. (3) Lazzeri, M.; Vittadini, A.; Selloni, A. Structure and Energetics of Stoichiometric TiO2 Anatase Surfaces. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 155409/1−155409/9. (4) Liu, S.; Yu, J.; Jaroniec, M. Anatase TiO2 with Dominant HighEnergy {001} Facets: Synthesis, Properties, and Applications. Chem. Mater. 2011, 23, 4085−4093. (5) Diebold, U. The Surface Science of Titanium Dioxide. Surf. Sci. Rep. 2003, 48, 53−229.

■

CONCLUSIONS We used single-particle photoluminescence spectroscopy and imaging to reveal the spatial distribution of luminescent trap 26217


Article

The Journal of Physical Chemistry C (6) Yang, H. G.; Sun, C. H.; Qiao, S. Z.; Zou, J.; Liu, G.; Smith, S. C.; Cheng, H. M.; Lu, G. Q. Anatase TiO2 Single Crystals with a Large Percentage of Reactive Facets. Nature 2008, 453, 638−641. (7) Etgar, L.; Gao, P.; Xue, Z.; Peng, Q.; Chandiran, A. K.; Liu, B.; Nazeeruddin, M. K.; Grätzel, M. Mesoscopic CH3NH3PbI3/TiO2 Heterojunction Solar Cells. J. Am. Chem. Soc. 2012, 134, 17396− 17399. (8) Rong, Y.; Ku, Z.; Mei, A.; Liu, T.; Xu, M.; Ko, S.; Li, X.; Han, K. Hole-Conductor-Free Mesoscopic TiO2/CH3NH3PbI3 Heterojunction Solar Cells Based on Anatase Nanosheets and Carbon Counter Electrodes. J. Phys. Chem. Lett. 2014, 5, 2160−2164. (9) Yang, X. H.; Li, Z.; Liu, G.; Xing, J.; Sun, C.; Yang, H. G.; Li, C. Ultra-Thin Anatase TiO2 Nanosheets Dominated with {001} Facets: Thickness-Controlled Synthesis, Growth Mechanism and WaterSplitting Properties. CrystEngComm 2011, 13, 1378−1383. (10) Han, X.; Kuang, Q.; Jin, M.; Xie, Z.; Zheng, L. Synthesis of Titania Nanosheets with a High Percentage of Exposed (001) Facets and Related Photocatalytic Properties. J. Am. Chem. Soc. 2009, 131, 3152−3153. (11) Zhang, D.; Li, G.; Yang, X.; Yu, J. C. A Micrometer-Size TiO2 Single-Crystal Photocatalyst with Remarkable 80% Level of Reactive Facets. Chem. Commun. 2009, 29, 4381−4383. (12) Nunzi, F.; Mosconi, E.; Storchi, L.; Ronca, E.; Selloni, A.; Grätzel, M.; De Angelis, F. Inherent Electronic Trap States in TiO2 Nanocrystals: Effect of Saturation and Sintering. Energy Environ. Sci. 2013, 6, 1221−1229. (13) Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer, M. C. Explaining the Enhanced Photocatalytic Activity of Degussa P25 TiO2 Using EPR. J. Phys. Chem. B 2003, 107, 4545−4549. (14) Berger, T.; Monllor-Satoca, D.; Jankulovska, M.; Lana-Villarreal, T.; Gómez, R. The Electrochemistry of Nanostructured Titanium Dioxide Electrodes. ChemPhysChem 2012, 13, 2824−2875. (15) Szczepankiewicz, S. H.; Colussi, A. J.; Hoffmann, M. R. Infrared Spectra of Photoinduced Species on Hydroxylated Titania Surfaces. J. Phys. Chem. B 2000, 104, 9842−9850. (16) Tamaki, Y.; Furube, A.; Murai, M.; Hara, K.; Katoh, R.; Tachiya, M. Dynamics of Efficient Electron − Hole Separation in TiO2 Nanoparticles Revealed by Femtosecond Transient Absorption Spectroscopy under Weak Excitation Conditions. Phys. Chem. Chem. Phys. 2007, 9, 1453−1460. (17) Boschloo, G.; Fitzmaurice, D. Spectroelectrochemical Investigation of Surface States in Nanostructured TiO2 Electrodes. J. Phys. Chem. B 1999, 103, 2228−2231. (18) Mercado, C. C.; Knorr, F. J.; McHale, J. L.; Usmani, S. M.; Ichimura, A. S.; Saraf, L. V. Location of Hole and Electron Traps on Nanocrystalline Anatase TiO2. J. Phys. Chem. C 2012, 116, 10796− 10804. (19) Knorr, F. J.; Mercado, C. C.; McHale, J. L. Trap-State Distributions and Carrier Transport in Pure and Mixed-Phase TiO2: Influence of Contacting Solvent and Interphasial Electron Transfer. J. Phys. Chem. C 2008, 112, 12786−12794. (20) Rex, R. E.; Knorr, F. J.; McHale, J. L. Surface Traps of TiO2 Nanosheets and Nanoparticles as Illuminated by Spectroelectrochemical Photoluminescence. J. Phys. Chem. C 2014, 118, 16831−16841. (21) Knorr, F. J.; McHale, J. L. Spectroelectrochemical Photoluminescence of Trap States of Nanocrystalline TiO2 in Aqueous Media. J. Phys. Chem. C 2013, 117, 13654−13662. (22) House, R. L.; Mehl, B. P.; Kirschbrown, J. R.; Barnes, S. C.; Papanikolas, J. Characterizing the Ultrafast Charge Carrier Trapping Dynamics in Single ZnO Rods Using Two-Photon Emission Microscopy. J. Phys. Chem. C 2011, 115, 10806−10816. (23) Mercado, C. C.; Knorr, F. J.; McHale, J. L. Observation of Charge Transport in Single Titanium Dioxide Nanotubes by MicroPhotoluminescence Imaging and Spectroscopy. ACS Nano 2012, 6, 7270−7280. (24) Tachikawa, T.; Majima, T. Exploring the Spatial Distribution of Transport Behavior of Charge Carriers in a Single Titania Nanowire. J. Am. Chem. Soc. 2009, 131, 8485−8495.

(25) Tachikawa, T.; Yamashita, S.; Majima, T. Evidence for CrystalFace-Dependant TiO2 Photocatalysis from Single-Molecule Imaging and Kinetic Analysis. J. Am. Chem. Soc. 2011, 133, 7197−7204. (26) Zhang, P.; Tachikawa, T.; Bian, Z.; Majima, T. Selective Photoredox Activity on Specific Facet-Dominated TiO2 Mesocrystal Superstructures Incubated with Directed Nanocrystals. Appl. Catal., B 2015, 176−177, 678−686. (27) Murakami, N.; Kurihara, Y.; Tsubota, T.; Ohno, T. ShapeControlled Anatase Titanium(IV) Oxide Particles Prepared by Hydrothermal Treatment of Peroxo Titanic Acid in the Presence of Polyvinyl Alcohol. J. Phys. Chem. C 2009, 113, 3062−3069. (28) Yu, X.; Jeon, B.; Kim, Y. K. Dominant Influence of the Surface on the Photoactivity of Shape-Controlled Anatase TiO2 Nanocrystals. ACS Catal. 2015, 5, 3316−3322. (29) Kumar, S. G.; Devi, L. G. Review on Modified TiO 2 photocatalysis under UV/Visible light: selected results and related mechanisms on interfacial charge carrier transfer dynamics. J. Phys. Chem. A 2011, 115, 13211−13241. (30) Kavan, L.; Grätzel, M.; Gilbert, S. E.; Klemenz, C.; Scheel, H. J. Electrochemical and photoelectrochemical investigation of singlecrystal anatase. J. Am. Chem. Soc. 1996, 118, 6716−6723. (31) Zhang, J.; Wang, J.; Zhao, Z.; Yu, T.; Feng, J.; Yuan, Y.; Tang, Z.; Liu, Y.; Li, Z.; Zou, Z. Reconstruction of the (001) Surface of TiO2 Nanosheets Inducted by the Fluorine-Surfactant Removal Process Under UV-Irradiation for Dye-Sensitized Solar Cells. Phys. Chem. Chem. Phys. 2012, 14, 4763−4769. (32) Selçuk, S.; Selloni, A. Surface Structure and Reactivity of Anatase TiO2 Crystals with Dominant {001} facets. J. Phys. Chem. C 2013, 117, 6358−6362. (33) Seo, H.; Baker, L. R.; Hervier, A.; Kim, J.; Whitten, J. L.; Somorjai, G. A. Generation of Highly n-Type Titanium Oxide Using Plasma Fluorine Insertion. Nano Lett. 2011, 11, 751−756. (34) Hsiao, P.-T.; Tung, Y.-L.; Teng, H. Electron Transport in TiO2 Nanocrystalline Films of Dye-Sensitized Solar Cells. J. Phys. Chem. C 2010, 114, 6762−6769. (35) Dittrich, Th.; Lebedev, E. A.; Weidmann, J. Electron Drift in Porous TiO2 (Anatase). Phys. Status Solidi A 1998, 165, R5−R6. (36) Yamada, Y.; Kanemitsu, Y. Determination of Electron and Hole Lifetimes of Rutile and Anatase TiO2 Single Crystals. Appl. Phys. Lett. 2012, 101, 133907/1−133907/4. (37) Enright, B.; Fitzmaurice, D. Spectroscopic Determination of Electron and Hole Effective Masses in a Nanocrystalline Semiconductor Film. J. Phys. Chem. 1996, 100, 1027−1035. (38) Rich, C. C.; Knorr, F. J.; McHale, J. L. Trap State Photoluminescence of Nanocrystalline and Bulk TiO2. Mater. Res. Symp. Proc. 2010, 1268, 1268-EE03-08. (39) Knorr, F. J.; Zhang, D.; McHale, J. L. Influence of TiCl4 Treatment on Surface Defect Photoluminescence in Pure and MixedPhase Nanocrystalline TiO2. Langmuir 2007, 23, 8686−8690. (40) Yu, J.; Low, J.; Xiao, W.; Zhou, P.; Jaroniec, M. Enhanced Photocatalytic CO2-reduction Activity of Anatase TiO2 by Co-exposed {001} and {101} Facets. J. Am. Chem. Soc. 2014, 136, 8839−8842.

26218
