Morphology Effects in Photoactive ZnO Nanostructures: Photooxidative Activity of Polar Surfaces

A series of ZnO nanostructures with variable morphology were prepared by a microemulsion method and their structural, morphological, and electronic properties were investigated by a combined experimental and theoretical approach using microscopy (high resolution transmission electron microscopy) and spectroscopic (X-ray diffraction, Raman, and UVvisible) tools, together with density functional theory calculations. The present experimental and computational study provides a detailed insight into the relationship between surfacerelated physicochemical properties and the photochemical response of ZnO nanostructures. Specifically, the present results provide evidence that light-triggered photochemical activity of ZnO nanostructures is related to the predominance of highly-active (polar) surfaces, in particular, the amount of Zn-terminated (0001) surfaces, rather than bandgap sizes, carrier mobilities, and other variables usually mentioned in the literature. Computational results highlight the oxidative capability of polar surfaces, independently of the degree of hydration.

3 structural factors by sampling a plethora of ZnO morphologies; from single crystal surfaces, 42 to thin films, 35 powders, 32 nanostructures featuring a great diversity of sizes and shapes (e.g. rods, disks, rings, screw caps, burgers, tubes, hourglass, trigonal prisms, toilet brushes, spheres, abalone shells, and so on), 37,41,[43][44][45][46][47][48] and, eventually, nanoparticles with well-defined facets. 31,33,44,45 From such studies many aspects affecting the photochemical process can be invoked. For instance, concerning the acid brown 14 dye degradation, it has been shown that media acidity seems to affect the reaction, but, whereas a basic media enhances the yield for ZnO it hinders the yield for TiO 2 photocatalyst. 31 Surprisingly, recent studies did not reveal any correlation between photochemical activity with neither nanoparticle size 33 nor surface area. 41 Furthermore, relatively few of these studies analyse in detail the direct relation among ZnO morphology and the enhanced photocatalytic activity. The catalyst nanostructured morphology (primary/secondary particle size and shape) can be controlled depending on the synthesis method, the employed solvent, the addition of capping agents, and/or eventually the growing conditions. 37,41,43,49 The synthesized oxide may feature distinct polar/nonpolar facet ratio depending on the experimental procedure, and this is often argued to be a key aspect on the photochemical catalytic activity. However, the discussion remains open to a large extent.
On one hand, Mclaren and coworkers 33 have shown that a higher proportion of polar surfaces, i.e. hexagonal disk-like nanoparticles, show a higher photocatalytic activity towards the methylene blue dye decomposition, in line with the results of Li et al. for the N-formylation of aniline. 43 In the former work 33 it is posed that the Zn-terminated (0001) polar surface is the responsible for the high catalytic activity, backed by the study of Xu and colleagues on phenol degradation. 41 The idea behind such hypothesis is that such a surface together with the Oterminated (0001 ത ) is the most unstable one, i.e. exhibits the highest surface energy, and so 4 is prone to attach the reactants, although this explanation neglects the photochemical aspect.
However, these and other authors 33,41,43 also speculate on the possible critical role of surface oxygen vacancies on ZnO surfaces. Nonetheless, pristine polar surfaces suffer from an electric dipole instability issue, and in fact, do feature a variety of reconstructions which generally involve surface vacancies. 50,51 Last but not least, although in some cases the role of nonpolar surfaces is ruled out, 33,41 Kislov et al. have recently shown that the nonpolar surfaces of ZnO present the highest photochemical activity for the methyl orange dye degradation. 42 As long as the alcohol photodegradation is concerned, little is known on the key aspects governing the photocatalytic activity. Typically phenol is used as a probe molecule for liquid-phase photodegradation studies whereas isopropanol is the test case in the gas-phase counterpart. 52 Regardless of that, isopropanol is commonly used as a liquid-phase hydroxyl scavenger additive. 53,54 Previous studies on phenol photodegradation on ZnO nanostructures point that the Zn-terminated (0001) polar surface is the most catalytically active, 41 and that ZnO nanoparticles exhibits a superior activity than TiO 2 ones. 55 This enhanced activity is also observed in the pioneering work of Kulkarni and Wachs on the photooxidation of isopropanol, in which it was found that ZnO selectively converts isopropanol into acetone via dehydrogenation. 32 In this latter work the pivotal role of acid/basic sites on the catalytical activity and specificity was posed.
In the present work we explore the relationship between ZnO morphology and chemical activity in a series of nanometric samples and analyse the structural/electronic and chemical differences encountered for different nanostructures as a function of the specific surface planes exposed to the media, either naked or hydrated, but also, for polar terminations, including surface defects. To this end, ZnO wurtzite nanoparticles were synthesized by means of a microemulsion method and have been characterized by X-Ray Diffraction (XRD) patterns and High-Resolution Electron Microscopy (HR-TEM) images together with Density 6 the structural characteristics of the ZnO nanostructures. Raman spectra were acquired using a Horiba iH320 spectrometer and He:Ne laser excitation (632.8 nm). UV-visible diffuse reflectance spectroscopy experiments were performed with a Shimadzu UV2100 apparatus with a nominal resolution of ca. 1 nm using BaSO 4 as reference. Band gap analysis was carried out following standard procedures (for a direct band gap semiconductor) by plotting (hνA) 2 (where hν = excitation energy and A = absorption coefficient) vs. energy.

Computational Details
Different slab models have been constructed to simulate the ZnO nanoparticle facets.
From present (see below) and previous XRD studies 16,33,37,41,45 it is clear that wurtzite ZnO preferentially displays the nonpolar (101 ത 0) and (112 ത 0) surfaces, and the Zn-and O-terminated (0001) and (0001 ത ) polar surfaces, respectively. Nonpolar surfaces can perfectly feature an unreconstructed termination whereas polar surfaces suffer from a well-known surface energy instability issue. Considering ZnO as an ionic material (with the caveat that is in the fringe region of ionic compounds), the different charged terminations of the polar surfaces exhibit a net dipole moment, which de facto increases with the separation between them. This introduces an electrostatic component to the surface energy that diverges with the surface separation. 57,58 There are various ways to nullify this dipole moment. The bulk cut termination may exist involving a charge transfer from electron-rich O-terminated (0001 ത ) surface to electronpoor Zn-terminated (0001) surface. However, despite such charge transfer is feasible for systems with few layers, it becomes unrealistic for the systems here contemplated. So, for nanometer-sized ZnO (0001)/(0001 ത ) systems it has been found that pristine terminations must feature 1/4 ML of surface vacants to compensate the under/overcharge. 59 Many previous studies reveal a variety of surface reconstructions as a function of the O 2 and H 2 pressure. 50,51,60 Aside from this, it has been found that when fully hydrogenated/hydrated both 7 polar surfaces show a perfect (1×1) arrangement, i.e. apparently no surface defects. [61][62][63] Because of this, we decided to consider the ideal pristine polar surface models only as a conceptual reference for dehydrogenation/dehydration processes, thus allowing for decoupling the desorption from surface reconstruction (vacancy formation) process.
To ascertain how the electronic structure affects the photochemical activity we investigated pristine nonpolar surfaces and polar surfaces with surface vacants. Moreover, following previous investigations, 64 we considered also fully hydrated ZnO surfaces, e. g.   The H 2 O mixed dissociated situation on nonpolar (101 ത 0) surface, as profusely studied both experimentally and theoretically, has been modeled using a (2×1) unit cell, 64,76,77 whereas a (1×1) unit cell is adequate for the full dissociated situation on nonpolar (112 ത 0) surface. 64 A minimum vacuum of 1 nm was applied in the surface direction to avoid interaction between repeated slabs. A counterdipole was set in the middle of the vacuum gap to compensate long-range dipole-dipole interactions among repeated slabs for pristine polar surfaces. An optimal Monkhorst-Pack k-points grid of 17×17×1 has been used for the ZnO surface calculations, guaranteeing an energy convergence below 0.01 kJ mol -1 as tested using denser grids.
Present DF calculations were performed with the VASP code, 78 carrying out periodic Kohn-Sham calculations for the above-described surface slab models. The projector augmented wave method has been used to represent atomic cores effect on the valence electron density. 79 This representation of the core states allows one to obtain converged results -variations in energy below 0.01 kJ mol -1 -with a cut-off kinetic energy of 415 eV for the plane-wave basis set. Geometry optimizations were performed using a conjugated gradient algorithm and applying a tetrahedron smearing method with Blöchl corrections. The structural optimization was finalized when forces acting on atoms were below 0.01 kJ mol -1 pm -1 . Unless stated otherwise, all calculations were carried out in a non spin-polarized fashion. All DF calculations have been carried out using the Perdew-Burke-Ernzerhof (PBE) exchange correlation (xc) functional. 80 This xc functional has proven to deliver a realistic description of bulk ZnO and low-index Miller surface structures. 61,65,66,69 Note that PBE, among other xc functionals, severly understimates materials bandgap.
Indeed, the bandgap for ZnO is here computed to be 0. 83 where fix slab E is the total energy of the slab unit cell containing two equal surfaces, and bulk E is the total energy of the slab atoms in bulk environment. A is the area of each surface created within the slab unit cell. Given that γ fix is known, one can extract the surface energy of a relaxed surface, rel γ , from the following equation; where ‫ܧ‬ ௦ is the total energy of the relaxed slab -here with only one its surfaces being relaxed-. Alternatively, one could calculate the relaxation energy per unit area, γ relax according to; and then adding the relaxation energy to γ fix .
The fully hydrated cleavage energy, ‫ܧ‬ ௩ ு మ ை , of polar (0001) Electron/hole pair conductivity is explained based on the their effective masses, estimated in a first approximation assuming a parabolic dispersion of Conduction Band (CB) maximum and Valence Band (VB) minimum, in the form; where E(k) is the band energy around its stationary point, here arbitrarily defined as k = 0, i.e.
at Γ Γ Γ Γ point; ħ is the reduced Planck constant, k is the k-vector distance, and m * is the effective mass. In this sense, the effective masses are obtained from the curvature of the band minima or maxima adjusted to a second degree polynom, as done in previous related studies. 88 For that, the bandstructure within the irreducible Brillouin zone has been sampled from Γ Γ Γ Γ pointat which direct band gap is found-towards A, M, and K points, which correspond in real

12
The electronic structure has been also studied based on the Density Of States (DOS) or, more specifically, the projected DOS. In this case, an energy resolution of of 0.05 eV was employed, and the projection was carried out over the surface and near surface atoms, including the adsorbates upon when found. For a proper comparison of VB and CB of the different surfaces and situations, the bands were aligned using the vacuum energy level as zero energy reference.

Results and Discussion
The XRD patterns of ZnO samples obtained with the microemulsion method by varying the surfactant to water ratio are shown in Figure 2. Table 1  To support this, DF based calculations of the surface energies have been carried out, although, for the present purposes, a more important quantity is the estimated cleave energies, following the procedure above-described. Thus, DF cleavage energies for the vacant-free This is a reasonable explanation for the morphologies obtained in sample C, which otherwise, due to the low surfactant concentration -water saturation conditions-should feature completely water saturated surfaces, and a morphology similar to sample B. However, in B the growth appears to be governed by the thermodynamics, whereas in C kinetics seems to play a key role. Take for instance a ZnO nanoparticle with all surfaces fully hydrated due to the capping effect of H 2 O, OH, or H moieties. In order to grow in a particular surface direction, a moiety must desorb of the surface to allow a ZnO growing unit to be inserted.
Desorption can be difficult in water rich conditions, i.e. the desorbed water molecule can easily be replaced by another water molecule. However, the limiting step seems to be still the desorption energy.
In the case of (0001)/(0001 ത ) polar surfaces the desorption is more problematic; on one hand, it would imply desorption of hydroxyls from the (0001) surface, or H atoms in the (0001 ത ) case. Here we conceptually considered the simultaneous desorption of both moieties from both surface terminations to form a water molecule in the media, and thus, leaving both surfaces free to further grow, thus allowing a more fair comparison with the other surfaces.
On the other hand, moiety desorption is considered leaving the polar surfaces naked and pristine, i.e. free of vacancies. This is done because when fully hydrated, polar surfaces are found to feature a regular vacant-free (1×1) pattern. 61-63 Therefore, we decided to separate desorption from a possible posterior vacancy formation process, which is out of the present discussion. One must keep in mind, however, that when water would be removed from the polar surfaces, they would unavoidably reconstruct forming vacancies. Given said that, present calculations reveal desorption energy to be 2.55, 2.04, and 2.09 J m -2 , for vacant-free Having explained the morphology based on the synthesis media and growing conditions, we now focus on the photocatalytic aspect. For isopropanol photooxidation it is well known that the initial radical attack is carried out by hole-related species. 92 The alcohol degradation assisted by UV light irradiation is based on the principle that UV light creates electron/hole (e -/h + ) pairs within the semiconductor. Holes generated in the VB are capable of oxidizing the alcohol substrate and the CB electrons are capable of reducing oxygen present in air. The generated e -/h + pairs suffer competitive fates concerning their recombination or involvement in chemical steps. The efficient separation of photogenerated electron/hole is one of the important factors in the reaction.
As well-known, the relative energy of VB and CB levels determines the (potential) transfer path(s) of photoexcited holes and electrons, which can directly influence their fate as charge recombination and/or chemically related species. Primary size and shape are important morphological parameters affecting photoactivity by influencing VB and CB levels. Size is rather important but the crystalline dimension range here explored ( Table 2; Table 2. Results indicate that charge mobility has no significant differences (maximum of ca. 10%) as Indirectly the VB plots in Figure 5 show that work function values for different surface terminations differ within a range of 1.2 eV, which is in line with variations of up to 0.8 eV as found in the literature. 96 Note that the ZnO work function is generally accepted to be around 4.3 eV, as found for nanocrystalline thin films. 97 Indeed, variations between different surface terminations are well below this value, and also smaller than the changes in the work function produced by adsorption of polar molecules upon, found to vary the work function by up to 2.9 eV. [98][99][100] Coming back to Figure 5, it indicates that the (0001)  Raman active. Figure 6 shows the Raman spectra for the three samples discussed in this work.
A sharp peak at 433-453 cm −1 can be assigned to the E 2 high mode of non-polar optical phonons, which is the characteristic peak of the hexagonal wurtzite phase. Other E 2 related mode peaks appear at 90 cm -1 (E 2low ), 195 cm -1 (2E 2low ) and 331-327 cm -1 to (E 2high -E 2low ), the last two due to multiple phonon scattering processes. The peak at 383cm −1 corresponds to A 1 transverse optical (TO) mode. A small peak at 580 cm −1 attributed as the E 1 (LO) mode has also been observed. Generally, it is believed that the E1 (LO) mode is related to the structural defects (oxygen vacancies, zinc interstitials, free carriers, etc.) in ZnO. The low intensity of the E 2high peak and the absence of significant differences among the presented spectra suggest that the samples exhibit well crystalline hexagonal wurtzite entities without substantial differences in lattice/defect related structural properties.
Last, Figure 7 presents the normalized photooxidation rate of the samples (e.g. rates eliminating the effect of the different surface area of the samples) as a function of the (101 ത 0)/(0002) XRD intensity ratio. A "natural" order evolves from this graph; needle crystals (sample C) showing much higher activity than flat crystals (sample A), and these two structures displaying larger activity than brick structures (sample B). These results highlight the requirement of the presence of polar surfaces for obtaining high photocatalytic performance. This, in combination with present DF-based photochemical activity trends, and considering, as above-mentioned, the h + key role in the process, indicates that isopropanol photoelimination is likely to occur on polar ZnO nanoparticle facets, independently of the degree of hydration. In particular, Zn-terminated (0001) surface is the one with a priori better photooxidation activity. This rules out the effect of water displacement by isopropanol as a key factor.
Thus, apparently, the exposure of polar surfaces, in particular the Zn-terminated (0001) surface, appears to be a key factor for the photooxidation capability of ZnO nanostructures. This results further supports the previous work of Maclaren and coworkers, 33 showing that a higher proportion of polar surfaces is needed for a good organic compound decomposition, although here needle-like structures show a better activity than previous 33 and present hexagonal disk-like nanoparticles. Aside, the key role of (0001) surfaces is also in line with previous studies on alcohol degradation. 41 This shows why sample C, showing the higher ratio of (0001) surfaces, maximizes the isopropanol photooxidation rate as seen in Figure 7. To obtain such a high ratio of polar surfaces, one needs to synthesize ZnO nanoparticles in a water rich environment. Thus, this shape comparison in equal conditions allows one to figure which polar surface is best to improve the photocatalytic power of a ZnO-based nanoparticle material.

Conclusions
A series of ZnO nanostructures were prepared by microemulsion using a single pot procedure. A multitechnique XRD, Raman, TEM, UV-visible experimental study combined with DF calculations allowed us to provide a complete structural, morphological, and electronic analysis of the materials. The nanostructured ZnO entities displayed disk-, brick-, and needle-like morphologies as the surfactant to water ratio of the microemulsion procedure continuously decreases from ca. 60 to 10 values.
The analysis of the chemical response of such ZnO nanostructures in the photoelimination of isopropanol was rationalized in terms of the structural/electronic properties of the nanosolids. The combined experimental and theoretical approach was able to provide evidence that the process is better carried out on polar surfaces, independently on the degree of hydration, and consequently, ruling out the effect of water displacement by isopropanol adsorption. Furthermore, the key role played by the ratio between polar and nonpolar facets exposed at the external surface of the ZnO nanomaterial is disclosed. It is shown that maximization of activity is not obtained by physicochemical properties (e.g. defects, band gap energy, etc.) previously described in the literature, but by maximizing the exposure to the external solid surface of polar surfaces in general, and ZnO Zn-terminated (0001) surface in particular. This is shown to be grounded in the preferential stability of holes upon light excitation at (0001) surfaces. This feature would be relevant for all photodegradation processes and thus hold general validity for ZnO nanomaterials. Table 1 Water to surfactant molar ratio microemulsion parameter values (w), primary particle size (Size, in nm) calculated from XRD data, and BET areas (A BET , in m 2 g -1 ), (10 1 0)/(0002) diffraction peaks intensity (I) ratio, and band gap energy (eV) measured from UV-visible data.