ZnO−Latex Hybrids Obtained by Polymer-Controlled Crystallization: A

J. Phys. Chem. B 2007, 111, 697-707

697

ZnO-Latex Hybrids Obtained by Polymer-Controlled Crystallization: A Spectroscopic Investigation Rafael Mun˜ oz-Espı´,*,† Gunnar Jeschke,†,‡ Ingo Lieberwirth,† Clara M. Go´ mez,§ and Gerhard Wegner*,† Max-Planck-Institut fu¨r Polymerforschung, Postfach 3148, 55128 Mainz, Germany, and Institut de Cie` ncia dels Materials, UniVersitat de Vale` ncia, 46100 Burjassot, Vale` ncia, Spain ReceiVed: September 28, 2006; In Final Form: NoVember 17, 2006

Micro- and submicrosized ZnO-polymer hybrid materials were synthesized by precipitating zinc oxide from an aqueous medium in the presence of poly(styrene-acrylic acid) latex nanoparticles, prepared by miniemulsion polymerization. Up to 10 wt % of the latex becomes incorporated into the crystals. Although the long-range order of the inorganic material is essentially not altered by the polymer, studies by photoluminescence (PL) spectroscopy and electron paramagnetic resonance (EPR) show that the latex particles influence the optical and paramagnetic properties of the hybrids, which can be correlated with changes in the defect structure. Typical PL emission spectra showed a narrow UV peak and a defect-related broad band in the green-yellow spectral region. The former emission is attributed to exciton recombination, whereas the latter seems to be related with deep-level donors. Latex acts as a quencher of the visible emission, and compared to pure ZnO, ZnO-latex hybrids show a significantly lower PL intensity in the visible range. A noticeable dynamic behavior of the PL, clearly more remarkable in the presence of latex, was observed, and it is explained in terms of photodesorption of oxygen adsorbed at surface positions. EPR provided additional information about crystal defects and species with unpaired electrons. All EPR spectra showed a single signal at g ≈ 1.96, whose intensity and temperature dependence did not correlate with those of the PL visible band. These findings indicate that the green-yellow emission and the EPR signal of our samples have a different physical origins.

Introduction Although the label “organic-inorganic hybrid” became popular at the end of the past decade, composite materials have always been present in nature and play a crucial role in biological systems, such as vertebrate skeletons, teeth, egg skins, and mollusk shells, in which nucleation and growth of inorganic structures are controlled by the presence of biomacromolecules. Driven by biomimetic strategies, the combination of inorganics with polymers and the study of the resulting products attracts the interest of an increasing number of material scientists. The crystallization of inorganic structures in the presence of hydrophilic polymers has been shown to be an efficient method to obtain hybrid materials with homogeneous and specific morphologies. Two different strategies, sometimes combined with each other, can be distinguished. First, polymers can act as templates or structure-directing agents of inorganic materials, especially in sol-gel processes.1 A second approach is given by the so-called “polymer-controlled crystallization”.2 In this case, soluble polymers are not used as templates but as specific controlling agents of crystal formation. Macromolecules may play a role in the nucleation process or may also adsorb selectively at certain crystal faces and control the growth, leading to changes in the habit and determining the final morphology. * Corresponding authors. Telephone: +49-6131 379 130. Fax: +496131 379 100. E-mail: [email protected] (R.M.-E.), wegner@ mpip-mainz.mpg.de (G.W.). † Max-Planck-Institut fu ¨ r Polymerforschung. ‡ Current address: Universita ¨ t Konstanz, Fachbereich Chemie, 78457 Konstanz, Germany. § Universitat de Vale ` ncia.

In some cases, polymers can also stabilize phases that otherwise are metastable. Many types of polymers, including both natural and synthetic types, have been applied to control crystallization processes. Among them, double-hydrophilic block copolymers are by far the most broadly used.2 Miniemulsion latexes have been much less investigated but provide a novel system with many potential applications.3,4 The developments during the last years in miniemulsion polymerization allow the synthesis of particles with a defined and tunable size as well as chemical surface composition. Carbonates (mainly CaCO3 and BaCO3)5-8 and barium sulfate9 are probably the systems that have been subjects of most of the research in the field of polymer-controlled crystallization. Many other materials, such as barium chromate,10 calcium phosphates,11 and zinc oxide,4,12-14 have also been studied. However, most of the published work up to now has been based on morphological effects and, sporadically, on some studies about growth mechanisms and kinetics. So far, very little effort has been devoted to investigate how polymers influence the crystal structure and defects and how this affects the physical features, especially optical, magnetic, and electric properties. In the present work, we contribute to fill this lack for a model hybrid system formed by zinc oxide embedding poly(styreneacrylic acid) latex particles. The materials obtained in the course of the crystallization of zinc oxide are particularly suited for systematic studies regarding polymer-mediated crystallization, because ZnO can be easily precipitated from aqueous media as only one polymorph (zincite). Parallel to the interest as a model system, the material

10.1021/jp066380d CCC: $37.00 © 2007 American Chemical Society Published on Web 01/09/2007

698 J. Phys. Chem. B, Vol. 111, No. 4, 2007 science of this II-VI semiconductor offers a wide variety of industrial applications, including well-established ones (e.g., pigments, catalysis, and gas sensors) and new potential uses in optoelectronic devices, such as light emitters, solar cells, or transparent conducting films.15-17 In semiconductors, intrinsic or extrinsic defects can act as electron donors or acceptors and can lead to localized energy states within the band gap. Thus, when optoelectronic applications are concerned, the defect structure stays in the foreground. Photoluminescence (PL) spectroscopy and electron paramagnetic resonance (EPR) are tools commonly used to obtain indirect information about defects in zinc oxide. The presence of defects in a solid determines the features of excitation and emission. Consequently, the analysis of PL spectra may help to elucidate the microstructure. EPR allows the detection of defects and species in a paramagnetic state. The g value of a signal should be, in principle, characteristic for the type of defect. Unfortunately, in both PL and EPR spectroscopies, the unambiguous assignment of an observed spectral feature with a particular defect/species is not a trivial task, because different defects can cause overlapping bands at similar positions. In spite of the large amount of publications on the topic, there is no consensus in the ascription of the observed signals to concrete entities in the ZnO structure. Depending on sample preparation, PL emission peaks in the whole visible range (blue, green, yellow, orange, and red emissions) have been reported and attributed without unanimity to all possible defects (i.e., interstitial zinc and oxygen, zinc and oxygen vacancies, antisites, and substitutional impurities). The discrepancy is even larger when the mechanisms of the absorption and emission processes are concerned. A systematic overview of the optical transitions in zinc oxide and the related literature is given by O ¨ zgu¨r et al.18 in an extensive review published recently. Also in EPR spectra, peaks at different values of g are found by different authors, and a large number of contradictory results have been reported (see, for instance, the discussions in refs 19 and 20). Recently, we have shown that hydrophilic surface-functionalized latex particles are able to control the morphology of zinc oxide precipitated from aqueous media.4 Depending on the surface chemistry of the latex particles, the morphology of the zinc oxide crystals changes dramatically, ranging from needlelike to plate- or flowerlike crystals. Among the different types of polymer compositions screened, poly(styrene-acrylic acid) latex particles evolved as a convenient model system for quantitive analysis. A mechanistic approach was proposed, assuming a preferential adsorption of the latex particles onto the fast-growing {001} faces of ZnO, which reduces the growth rate in the c-axis direction. Whether or not the incorporation of latex is the origin of specific distortions and changes in the crystal microstructure and whether or not this affects the physical properties of the materials remained open questions. Here, we study how the presence of embedded poly(styreneacrylic acid) latex particles influences the physical properties of zinc oxide precipitated from an aqueous medium. In particular, we present the effect of the polymer on the optical and paramagnetic properties of the resulting hybrid materials, correlating the experimental observations with the defect microstructure. Experimental Section Synthesis of ZnO-Latex Hybrid Materials. The hybrids were prepared by precipitating zinc oxide in the presence of different amounts of poly(styrene-acrylic acid) (P(S-AA))

Mun˜oz-Espı´ et al. latex. The latex was synthesized by miniemulsion polymerization as follows. First, the aqueous phase was prepared by dissolving 3.0 g of the nonionic surfactant Lutensol AT50 (BASF) with 48 g of water. Second, the oil phase was prepared by mixing 11.52 g of styrene (Fluka, puriss, g99.5%), 0.48 g of acrylic acid (Aldrich, purum, g99.0%), 500 g of hexadecane (Aldrich, anhydrous, g99.0%), and 240 mg of the oil-soluble initiator 2,2′-azobis(2-methylbutyronitrile) (AMBN, Wako Pure Chemical Industries). Both the oil and aqueous phases were stirred for 30 min separately and then mixed and stirred for another 45 min. The resulting mixture was ultrasonicated (Branson Digital Sonifier 250-D) to achieve the miniemulsion state, cooling simultaneously in an ice-water bath to avoid polymerization due to heating. The reaction was carried out during 8 h at 72 °C and under an argon atmosphere. The obtained latex had an average diameter of 71 ( 1 nm, determined by dynamic light scattering, and a surface charge density of 1.3 nm-2, determined by titration with a 0.001 N solution of poly(diallyl dimethylammonium chloride) (polyDADMAC, Mu¨tek Analytic, molecular weight in the range 40 000-100 000 g‚mol-1). A more detailed description of the synthesis, purification, and characterization of analogously prepared latex particles can be found elsewhere.4 In a typical crystallization setup, a solution of Zn(NO3)2‚6H2O (1.784 g, Fluka, g99%) in water (195 - x mL, Milli-Q) was placed in a jacketed reactor connected to a thermostated water circulator (95.0 ( 0.1 °C), under reflux and continuous magnetic stirring. The pertinent quantity of latex emulsion (x mL, x ) 0 in the case of the reference sample) was added to the reactor. After achieving thermal equilibrium, the reaction was started by adding hexamethylenetetramine (HMTA, 0.840 g, Aldrich, ACS reagent, g99%) dissolved in water (5 mL). After 90 min, the reaction mixture was cooled in an ice-water bath and the precipitate was separated by centrifugation, washed several times with water, and dried under vacuum at 40 °C for at least 12 h. To check the influence of copper traces on the photoluminescence, a high-purity precursor with no detectable level of copper, Zn(NO3)2‚xH2O (Aldrich, 99.999% based on trace metals analysis), was used. For the thermally treated samples, the as-grown materials were placed in a porcelain combustion boat and calcinated in an oven Linn High Term EVA-1700 under an O2 atmosphere. The heating-annealing program was as follows: (1) 5-min stabilization plateau at 25 °C (room temperature, RT); (2) RT f 600 °C, 10 °C‚min-1; (3) 1-h plateau at 600 °C; (4) 600 °C f RT, 10 °C‚min-1. In some cases, samples turned yellowish or brown in color after calcination. The brown color is an indication of carbon residues. Photoluminescence Spectra. Photoluminescence (PL) spectra were registered under atmospheric pressure using a Spex Fluorolog 212 spectrometer equipped with a Xe lamp. Measurements were carried out in a specially designed cell, composed of two quartz plates, placing the powder sample in between. A 345-nm high-pass filter was used for the emission spectra registered at λexc ) 310 nm, as well as for the excitation spectra registered at λem ) 560 nm. Similarly, 295- and 400-nm highpass filters were used for emission spectra registered at excitation wavelengths of 260 and 350 nm, respectively. The scanning velocity was always 2 nm‚s-1. For the temperature-dependent spectra, a cryostat was introduced into the sample chamber. In the dynamic studies, a time of not less than 3 min without irradiation elapsed between the different measurements to allow complete relaxation of the samples.

ZnO-Latex Hybrids: A Spectroscopic Investigation Electron Paramagnetic Resonance. A certain quantity of 4-oxo-2,2′,6,6′-tetramethyl(piperidin-1-oxyl) radical (9.44 mg, 4-oxo-TEMPO, Aldrich) was mixed with o-terphenyl (3.000 g, OTP, Fluka, g99%) and homogenized by melting the OTP at 80 °C. After cooling down, a portion of this mixture (0.301 g) was homogenized with OTP (2.995 g) and heated again to 80 °C. Known quantities of this final mixture (5-12 mg) were homogenized accurately with known quantities of the ZnO samples (50-90 mg). A part of the mixture was introduced in the EPR tubes for the quantitative measurements. All EPR measurements were carried out on a Bruker Elexsys 580 spectrometer at frequencies of approximately 9.8 GHz, using a 4103TM cavity at room temperature and an MD4EN Bruker Flexline ENDOR resonator at low temperature. A microwave power of 2 mW was applied at room temperature. To avoid saturation of the nitroxide reference signal at lower temperature, power was reduced to 20 µW at 80 K and 0.2 µW at 20 K. A modulation amplitude of 0.2 mT was used throughout. Spin concentration in the ZnO samples was determined from the ratio of the double integrals of the ZnO and nitroxide signals and the known spin concentration of the nitroxide. Further Characterization Techniques. Scanning electron microscopy (SEM) images were recorded by using a fieldemission microscope LEO EM1530 Gemini working at an accelerating voltage of 1-1.5 kV. X-ray diffraction (XRD) patterns were registered by a Seifert XRD 3000 TT diffractometer with Cu KR radiation (λ ) 1.54 Å, voltage ) 40 kV, current ) 30 mA). Thermogravimetric analysis (TGA) was carried out with a thermobalance Mettler Toledo ThermoSTAR TGA/SDTA 851 under an oxygen atmosphere (heating rate of 10 °C‚min-1, from room temperature to 600 °C). Before TGA, the excess latex in the hybrid materials (i.e., the rest of the nonincorporated latex and particles merely adsorbed at the crystal surface) was removed by repetitive washing. Results and Discussion Zinc oxide crystalline powders were precipitated from an aqueous solution in the presence of different concentrations of poly(styrene-acrylic acid) (P(S-AA)) latex particles (with an average diameter of 71 ( 1 nm and a surface charge density of 1.3 nm-2). The core of these polymer particles consisted essentially of polystyrene and was surrounded by a carboxylicgroup-containing corona. The latex corona contains short segments of the hydrophilic component of the nonionic surfactant used in the synthesis (a commercial poly(ethylene oxide) derivative, Lutensol AT50, with structure (C16H33)(EO)50) and hydrophilic segments of the copolymer of styrene and acrylic acid. During the crystallization of the zinc oxide, the hydrophilic functionalized polymer particles adsorb at the growing surfaces and become incorporated into the growing structure, resulting in polymer-inorganic hybrid materials. In the presence of the latex particles, the samples become more monodisperse. It was observed that by increasing the latex concentration, the resulting crystals become shorter and wider and the length-to-width ratio decreases systematically, as shown in the scanning electron microscopy (SEM) images of Figure 1. Here, a series of materials prepared with an increasing concentration of latex and a pure ZnO reference sample are compared. X-ray diffraction (XRD) showed that zincite is the only crystalline phase present in all cases. Increasing amounts of latex become occluded into the final materials as its overall concentration in the crystallizing system is increased. Table 1 lists the latex contents as estimated from the weight loss from thermogravimetrical analysis (TGA).

J. Phys. Chem. B, Vol. 111, No. 4, 2007 699

Figure 1. SEM micrographs of ZnO crystals obtained in the absence and in the presence of P(S-AA) latex: (a) without latex, (b) 1, (c) 3, and (d) 9 g‚L-1.

TABLE 1: Latex Content and Spectroscopic (PL and EPR) Data for a Pure ZnO Sample and for ZnO-Latex Hybrids Obtained in the Presence of Different Concentrations of P(S-AA) Latex latex latex green/UV spin densityc EPR relaxation conc [g·L-1] contenta [%] PL ratiob [cm-3]/1018 time, T2 [ns] 0 1 3 5 9

0.0 4.3 7.0 9.1 9.5

47.8 12.3 10.8 8.4 6.8

3.70 3.15 3.11 2.4 1.38

38 75 61 52 48

a Estimated by thermogravimetrical analysis. b Ratio of the integrated intensities of the green and UV peaks. c From EPR measurements.

The volume and mass of one hybrid crystal can be roughly calculated from the dimensions (measured statistically from the electron micrographs) and the density of the material, taking 5.606 g‚cm-3 for pure ZnO and 1.00 g‚cm-1 for the latex. As an example, for the material shown in Figure 1d, with a latex content of ca. 9.5 wt %, the volume of a crystal is in the order of 0.30 µm3. The volume of a spherical latex particle is 3 orders of magnitude smaller (1.8 × 10-4 µm3 for 70-nm particles). Consequently, hundreds of latex particles (ca. 1000 for the highest polymer content) can be present in the volume of one crystal, which impliessassuming an ideal homogeneous distributionsinterparticle distances of only a few nanometers. Hybrids containing up to approximately 10 wt %, corresponding to volume percentages of as much as up to 50%, are formed without causing deterioration in the long-range order of the crystalline material, as proved by XRD.4 I. Room-Temperature Photoluminescence. Photoluminescence (PL) excitation and emission spectra of pure ZnO and ZnO-latex hybrid samples were measured in the solid state, placing the crystalline powders between two quartz plates of a specially designed cell. Slight differences in sample preparation are unavoidable in our procedure: since the samples contained changing quantities of polymer and had, in consequence, different morphologies, they might be differently packed. However, considering that the preparation was similar in all cases and taking into account that the latex introduces only small differences in the density, we are confident that the spectra are qualitatively comparable, as it has been proven by independent measurements.21 PL Spectra of Pure ZnO Samples. The excitation spectrum of a pure ZnO control sample (see Figure 2), registered at the emission wavelength of 560 nm (2.21 eV), shows a pronounced

700 J. Phys. Chem. B, Vol. 111, No. 4, 2007

Figure 2. (a) PL emission spectra (λexc ) 310 nm) and (b) PL excitation (PLE) spectra (λem ) 560 nm) of ZnO-latex hybrid samples crystallized in the presence of different concentrations of P(S-AA) latex, compared with a pure ZnO sample.

peak at 384 nm (3.23 eV). In the emission spectra of the same sample, registered at an excitation wavelength of 310 nm (4.00 eV), two bands can be distinguished: a narrow and weak UV peak at 385 nm that appears at almost the same position as the excitation peak, with no significant Stokes shift, and a very broad and more intense green-yellow visible band centered at around 560 nm. The change of the excitation wavelength from 260 to 350 nm leads only to a change in the intensity of the peaks, but not in the position. The UV peak can be attributed to near-band-edge exciton annihilation,22 whereas the origin of the visible emission is less clear. In literature, a green band, ranging from 2.2 to 2.5 eV (typically 2.45-2.5 eV), is the visible emission most commonly found in ZnO. Although some authors have attributed this band to residual copper impurities,23,24 it is more typically ascribed to transitions related to intrinsic defects. A large number of possibilities have been proposed: different processes involving oxygen vacancies (VO),25-29 zinc vacancies (VZn) acting as acceptor levels,30-33 zinc interstitial (Zni),25,34 donor-acceptor pair recombination of a Zni+ donor with a VZn- acceptor,30 and zinc and oxygen antisites.35,36 The possible origin of the green emission through copper impurities, was ruled out after comparing the spectra of a conventional ZnO control sample (prepared with reagent-grade Zn(NO3)2‚6H2O) and an analogous sample synthesized by employing a high-quality copper-free zinc nitrate. While the conventional precursor had a purity of 99.4% and contained

Mun˜oz-Espı´ et al. copper on the level of e50 ppm, the high-quality zinc nitrate had a purity of 99.999% and no presence of copper at detectable levels. However, the emission spectra did not show any appreciable difference, and the visible band was similar in form and intensity in both cases. Therefore, the ascription of the visible emission to interstitial or vacancy defects in the crystal structure appears to be more reasonable than substitutional impurities. Hybrid Materials: Photoluminescence Versus Latex Concentration. Although both UV and visible emission are present in all samples (control and hybrid samples), their intensities and ratios change. Figure 2 shows that the visible band decreases when increasing the concentration of P(S-AA) latex during the crystallization. This goes parallel to an increase of the UV emission in the presence of incorporated latex. Ratios of the integrated intensities of the visible band to the integrated intensities of the UV peak are compiled in Table 1. The strong quenching of the visible emission in the presence of latex leads to a significant decrease of the ratio of green-to-UV emission intensity. Photoluminescence excitation (PLE) spectra of the samples (Figure 2b) show the onset and the peak at approximately the same positions, but the intensities change for the different latex concentrations. When the polymer is removed by calcination under an oxygen atmosphere, the PL spectra of the samples, presented in Figure 3, change with respect to the original samples. The visible emission is red-shifted from 560 nm (2.21 eV) in the as-grown pure ZnO sample to 603 nm (2.06 eV) in the O2-annealed ones. Samples crystallized in the presence of intermediate latex concentrations (3 and 6 g‚L-1) show two components in the visible region of their PL spectra: a yellow-orange peak at ∼2.0-2.1 eV and a green peak at ∼2.5 eV. The emission spectra of the sample obtained at the highest latex concentration (9 g‚L-1) is dominated by the emission at ∼2.5 eV. The UVemission peak is slightly blue-shifted for all annealed samples, and analogously, the peak in the excitation spectra shifts also to higher energies (from 2.22 to 2.28 eV). The observation of yellow emission in ZnO samples annealed in an oxidative atmosphere is not unique and has been often reported. For instance, Studenikin et al.37 observed orange PL emission in ZnO films prepared by oxidative annealing, while samples annealed in a reductive atmosphere presented a green emission. Yellow-orange emission appears to be related to an excess of oxygen and has been ascribed to interstitial oxygen.38-40 The almost disappearance of the excitonic emission and the parallel enhancement of the visible emission for samples thermally treated under oxygen is also a frequent observation.40 Our results suggest that the visible emission is composed of more than one contribution, probably of different physical origins. Actually, the visible emission of all the samples can be perfectly fitted to the sum of two Gaussian functions, with centers in the green and in the yellow-orange regions. But, why, if the oxidative atmosphere was similar for the different samples, was strong yellow emission observed in the control sample after annealing, whereas the green band dominated the emission in the sample prepared with high amount of P(SAA) latex? Let us assume that both emissions are related to defects, but the concentration of defects responsible for the yellow peak increases in the presence of oxygen, while the concentration of those responsible for the green increases under oxygen-deficient conditions. In the case of pure ZnO, an excess of oxygen is available to create the interstitial oxygen required for the orange emission (assuming that this ascription is correct). For hybrid samples, the oxygen will be first consumed in the

ZnO-Latex Hybrids: A Spectroscopic Investigation

Figure 3. (a) PL emission spectra (λexc ) 310 nm) and (b) PL excitation (PLE) spectra (λem ) 560 nm) of different ZnO-derived samples, after calcination under an O2 atmosphere at 600 °C.

J. Phys. Chem. B, Vol. 111, No. 4, 2007 701 phenomenon is shown in Figure 4a and b, where the PLemission intensities for detection windows at 385 ( 1 and 560 ( 1 nm (under continuous excitation at 310 nm) are plotted versus the irradiation time. The dynamic effect appears to be clearly less important for the pure ZnO sample, whose intensity changes less than 10%, and is stronger for the hybrid materials, whose intensities change up to more than 40% (see Figure 4c and d). After ca. 20 s, the emission intensities remain approximately constant. The process was found to be reversible and the curves are reproducible, if the measurements are repeated with a time lag of few minutes to allow the complete relaxation of the samples. The emission spectra previously presented (Figures 2 and 3) are assumed to show the situation at equilibrium of the mentioned dynamic process, because after several seconds the intensity become negligibly small and the total time to register the spectrum is much longer (around 3 min at an scanning velocity of 2 nm‚s-1). The shape of the curves shown in Figure 4 and the reversibility of the process suggest a photoinduced desorption of molecules adsorbed onto the surface of the materials, presumably oxygen from the medium, a known phenomenon in zinc oxide surfaces.41 A fact that attracted our attention was that the curves fit well to adsorption-isotherm-like expressions. Somehow, the dynamic behavior of the luminescence appears to mirror the assumed photodesorption process. Thus, the PL curves of Figure 4a and b might be rescaled and fitted to such equations. For this purpose we assume that (i) the PL intensity is proportional to the coverage of the ZnO surfaces by oxygen species (I ∝ θ) and (ii) the oxygen concentration is proportional to the time of photoirradiation (c ∝ t). Although this empirical treatment of plotting luminescence data with an adsorption-like expression could be questioned from a strict formal point of view, it offers clear evidence of the correlation of the PL with adsorption processes. In Figure 4c, the dependence of the UV emission intensity with the time of irradiation, t, is perfectly fitted with a Langmuir-Freundlich-like expression

I′UV ) combustion of the embedded polymer. Once the polymer has been burned out, the creation of interstitial oxygen is possible. Thus, in samples containing relatively low amounts of latex, both bands are present, the orange being larger than the green. When the amount of latex is high, the oxygen will be essentially consumed in its combustion. The latex material can work as a reducing agent toward ZnO in the heating/pyrolysis process; this is highly probable as ZnO can be reduced by carbon at high temperatures according to the reaction 2ZnO + C f CO2 + 2Zn. Although the process takes place under continuous oxygen flux, the partial oxygen pressure can diminish if the amount of consumed oxygen is larger than the entrant. An involvement of oxygen vacancies in the green emission could be therefore assumed, since the concentration of this defect type should increase when the oxygen pressure decreases. However, although this agrees with one of the “classical” assignments of the green emission, we could not find any unambiguous proof supporting this assumption. II. Dynamic Behavior of the Photoluminescence. Under continuous irradiation with monochromatic light from the xenon white lamp of the spectrometer, a dynamic behavior of the PL emission was observed (i.e., the intensity changed with the time of excitation). A time-dependent increase of the UV emission occurred simultaneously with a decrease of the visible emission when the samples were studied under ambient conditions. This

a′(b′t)1/m′ 1 + (b′t)1/m′

(1)

in which a′, b′, and m′ are the fitting parameters and I′UV is the rescaled PL intensity, calculated according to

I′UV )

It - I0 Imax

(2)

where It is the PL intensity at a certain time and I0 and Imax are the initial and maximum intensities, respectively. In the case of the visible emission, the intensity decreases with time, but the representation 1 - I′vis versus time (being I′vis the visible emission normalized to 1) can also be adjusted to an analogous equation, as illustrated by Figure 4d:

1 - I′vis )

a′(b′t)1/m′ 1 + (b′t)1/m′

(3)

It is known that oxygen has a strong tendency to adsorb to metal oxide surfaces, acting as an electron acceptor. Adsorbed O2 can capture electrons from the conduction band (eq 4), forming a negatively charged layer with low conductivity, a so-called depletion layer. When the material is irradiated with UV light of higher energy than the band gap (around 3.3 eV for ZnO), the photogenerated holes (h+), with a positive effective

702 J. Phys. Chem. B, Vol. 111, No. 4, 2007

Mun˜oz-Espı´ et al.

Figure 4. Dynamic of the PL emission under irradiation at 310 nm, detected at (a) λem ) 385 ( 1 nm and (b) λem ) 560 ( 1 nm. Graphs c and d show the data of a and b, respectively, rescaled and adjusted to Langmuir-Freundlich-type equations (eqs 1 and 3).

charge, migrate to the negatively charged surface, discharging the charged oxygen species (eq 5).42

O2(g) + e- h O2-(ads)

(4)

O2-(ads) + h+ h O2(ads)

(5)

Discharged O2 species can be more easily desorbed than O2-. The photodesorption of the oxygen leads to the destruction of the depletion layer. The photogenerated electrons can move freely and the conductivity increases. The O2 desorption continues until the equilibrium O2(ads) h O2(g) is achieved. In reality, the adsorption of oxygen at ZnO surfaces is the origin of a much more complex scenario than that shown by eqs 4 and 5. O2- is not the only charged chemisorbed oxygen species; but, the oxidation state can change, and different species, such as O- and O2-, can be present (see, e.g., ref 43). Furthermore, oxygen can react with crystal defects in surface positions and create more complex species. Keeping in mind the described adsorption-desorption processes supposed to take place at the surface of the particles, we suggest an explanation for the experimental observations based on a model proposed by Bahnemann et al.,44 who also observed a bleaching of the visible emission parallel to an increase of the UV peak. This model assumes that “electron traps” are initially present in the system. After photoexcitation, three competing processes are possible: (i) nonradiative relaxation,

Figure 5. Schematic representation of possible processes after photoexcitation in ZnO-derived products.

(ii) recombination of photogenerated electrons and holes (responsible for the sharp excitonic UV emission), and (iii) trapping of photogenerated electrons. The visible emission occurs after tunneling of trapped electrons from the conduction band to preexisting holes. Adsorbed oxygen species are supposed to play a crucial role in the electron tunneling from the conduction band to the preexisting trap sites. Figure 5 depicts schematically the possible processes after photoirradiation. Immediately after excitation, oxygen is adsorbed at the surface and can capture electrons, diminishing the probability of excitonic electron-hole recombination. These electrons are tunneled to the traps and visible light is emitted. Under

ZnO-Latex Hybrids: A Spectroscopic Investigation

J. Phys. Chem. B, Vol. 111, No. 4, 2007 703

Figure 6. PL emission spectra (λexc ) 310 nm) at decreasing temperatures (RT f 77 K) for a ZnO-latex hybrid sample crystallized in the presence of 3 g‚L-1 of P(S-AA) latex.

irradiation, the photogenerated holes begin to migrate to the surface and O2 is photodesorbed. The disappearance of “tunneling agents” from the surface implies the decrease in the visible emission and a simultaneous growth of the excitonic recombination. The evolution of the photoluminescence goes, therefore, parallel to the photodesorption process. Two possible defects in zinc oxide could act as traps: oxygen vacancies and interstitial zinc. The proposed model allows the explaination of the visible emission and its behavior without involving other less probable defects in n-type ZnO. It is worth mentioning that van Dijken et al.29 proposed a similar model to explain the emission of nanocrystalline ZnO, but in which O2- sites at the surface capture a hole (O2- + h+ f O-), which is transferred to a VO+ level (VO+ + h+ f VO2+). They assumed that the visible emission arises from the transition of photogenerated electrons of the conduction band to the deep trapped hole of VO2+. Independently of the exact mechanism behind the process, photoluminescence appears to depend on adsorption-desorption processes at surface sites. In the presence of P(S-AA) latex, the adsorption of O2 is hindered and the visible emission decreases accordingly. Additionally, latex particles accelerate the photodesorption process, as judged from the slopes of the curves at initial times in Figure 4c and d. III. Temperature Dependence of the Photoluminescence. Investigations on the temperature dependence of the photoluminescence were carried out for further understanding of the optical properties of the prepared ZnO-latex hybrid materials. PL emission spectra were registered at different temperatures in the range 77-305 K. For all samples, including a pure ZnO reference and the hybrids, the following features were observed as temperature decreased: (i) both UV- and visible-emission intensities increased, (ii) the maximum of the UV peak shifted to higher energies (blue shift), (iii) the visible band shifted to lower energies (red shift), and (iv) both peaks became narrower, that is, the full width at half-maximum decreased. An example of this behavior is presented in Figure 6 for a hybrid sample crystallized in the presence of 3 g‚L-1 of P(S-AA) latex. The two recognizable bands of the spectra can be relatively well fitted with single Gaussian curves, but in the case of the visible emission, the fitting improves if two Gaussians are used. This suggests once more the possibility of two contributions in the visible emission: a green and a yellow-orange one. The

Figure 7. Temperature dependence of the PL emission bands (area of the peaks normalized to unity) for pure ZnO and ZnO-latex hybrid samples: (a) UV emission and (b) visible emission. Solid lines are fits according to eq 6.

evolution of these two visible peaks seems to follow the same tendency as the envelope curve. However, since the exact position and the contribution of each component at different temperatures is not obvious and only mathematical speculations can be done, fits of the whole visible emission to simple Gaussians, which are nonetheless good fits, are considered further on, keeping in mind that more than one energetic transition might be behind the visible emission. The growth of the integrated PL intensity when the samples are cooled down can be considered as the result of the quenching of nonradiative recombination processes and the freeze-out of phonons.45 Figure 7 presents the evolution of the integrated PL intensity (normalized to 1) of the UV- and visible-emission bands for a set of hybrid samples, comparing them with a pure ZnO reference. It can be seen that the quenching of the UV peak is more important for the ZnO-latex materials than for the control sample: the thermal quenching from 77 K to room temperature is of ∼90% for the hybrids, against a ∼20% for the reference. The behavior appears to be opposite for the visible emission: quenching is here less significant for samples prepared with increasing concentration of latex. The temperature dependence of the PL intensity, I(T), can be expressed by a simple thermal activation model according

704 J. Phys. Chem. B, Vol. 111, No. 4, 2007

Mun˜oz-Espı´ et al.

to the equation45,46

I(T) )

I0

( )

-Ea 1 + A exp kBT

(6)

where I0 is the emission intensity at 0 K, A is a constant, and Ea is the activation energy of the thermal quenching process. For pure zinc oxide, activation energies of 3 and 41 meV were calculated for the UV and the visible band, respectively. For the hybrids, energies in the range of 62-19 meV (UV emission) and 84-35 meV (visible emission) were obtained, the lower values corresponding to the highest additive concentration. The temperature dependence of energy bands at constant pressure originates from both thermal expansion of the lattice and renormalization of band energies by electron-phonon interactions.47 The blue shift of the UV excitonic emission when decreasing temperature arises essentially from a band gap shift, assuming the exciton binding energy to be independent of the temperature. Thermal expansion and electron-phonon interaction have a similar temperature dependence, and all the phonons, in principle, contribute to the shift. The overall contribution to the shift can be reproduced by a Bose-Einstein distribution, using the effective phonon energy pω.48 Thus, the temperature dependence of the excitonic emission can be expressed by

E(T) ) E0 -

λ pω exp -1 kBT

( )

(7)

where E0 is the exciton energy at 0 K and λ is a proportionality coefficient. Figure 8a shows that the evolution with the temperature of the UV-peak energy, taken at the maximum of emission, can be fitted to eq 7. The parameter pω has a value of 12 meV for ZnO and ranges from 17 to 29 meV for the hybrids. A different tendency was observed for the visible emission: the band shifts systematically to lower energies as the temperature decreases, as can be seen in Figure 8b. This behavior, although not fully understood, has been sometimes reported in semiconductor crystalline materials. For instance, Leiter et al.28 reported for ZnO a continuous blue shift of the excitonic emission as temperature decreased, parallel to a shift to higher energies of the oxygen-vacancy-related (according to their assignment) green emission (∼2.45 eV). However, other authors have also found a different temperature dependence: van Dijken et al.29 reported a shift to higher energies with decreasing temperature in the range 75 < T < 180 K and a temperature independence of the energy position for T < 75 K and T > 180 K. In our case, the energy shift was found to fit well with an empirical Arrhenius-type equation

E(T) ) E0 + A exp

( ) - kBT

(8)

where E0 is the energy at 0 K, A is a constant, and  is an energetic parameter. The energetic parameter  calculated from the fits is 33 meV for the control samples; in the hybrid samples, it ranges between 37 and 40 meV. The opposite shift of the visible emission with respect to the shift of the exciton-related peak could indicate, as proposed by Leiter et al.,28 that the visible PL involves intradefect transitions whose energies behave differently from the temperature dependence of the band gap.

Figure 8. Temperature dependence of the energy of the maximum emission of pure ZnO and ZnO-latex hybrid samples: (a) UV emission and (b) visible emission. Solid lines are fits of the experimental data according to eqs 7 and 8.

A further consideration may be that the deep-level (visible) emission could be affected by the intensity of the absorbed light either by saturation effects for the reason that the incoming light has a varying penetration depth. However, in our experiments, we varied the intensity of the exciting light by a factor of 50 coming from the lowest intensity at which an emission was detectable, and no change in the position and form of the emission band was observed. IV. Electron Paramagnetic Resonance. Electron paramagnetic resonance (EPR) spectra of all our samples showed a signal at g ≈ 1.96, which is very strong compared to many published spectra. According to the literature, this signal seems to be attributable to shallow donors and its position appears to be independent of the shallow-donor identity.49-51 Many authors assigned also the EPR peak at g ≈ 1.96 to the singly ionized oxygen vacancy, VO+,27,52,53 but another signal at g ≈ 1.99 seems to be responsible for this type of defect.49,50 In our samples, the g ≈ 1.96 signal was the only one observed for any condition and temperature. A quantification of the spin concentration was performed by using the free radical 4-oxo-2,2′,6,6′-tetramethyl(piperidin-1oxyl) (4-oxo-TEMPO) as an internal reference. Figure 9 shows the EPR spectra for the same batch of the hybrid samples presented in previous sections, compared with the spectrum of a ZnO reference sample. Here, the signals of the organic radical (that appear below 351 mT and do not interfere with the ZnO signals) have been cut off. Spin densities, compiled in Table 1,

ZnO-Latex Hybrids: A Spectroscopic Investigation

J. Phys. Chem. B, Vol. 111, No. 4, 2007 705

Figure 9. Room-temperature EPR spectra for ZnO-latex hybrid samples, compared with pure ZnO. Measurements were carried out in the presence of the free radical 4-oxo-TEMPO, used as an internal reference (the peaks of this radical, all bellow 351, have been cut off in the graph).

Figure 10. Temperature dependence of the EPR signal for pure ZnO and a ZnO-latex hybrid sample prepared in the presence of 6 g‚L-1 of P(S-AA) latex. Solid lines are fits according to eq 10, and the dashed lines are fits to the same equation fixing I0 ) 0.

were calculated taking into account the ratio between the mixed quantities of free radical and ZnO sample and the ratio between the intensities of the EPR signals. The EPR signal corresponds to the spin concentration in the total material. To obtain the concentration of paramagnetic defects in ZnO particles, the density of the samples was calculated taking into account the latex content, obtained from the weight loss in the thermogravimetrical analysis (TGA), and approximating the density of the latex to 1.00 g‚cm-3. A density of 5.606 g‚cm-3 was considered for the pure ZnO sample. A spin density of 3.7 × 1018 cm-3 was calculated for the reference sample, and values between 1.4 × 1018 and 3.2 × 1018 cm-3 were obtained for the hybrid samples. The spectra of Figure 9 also show that the EPR signal becomes systematically narrower when the concentration of latex increases. Assuming that the transverse electron spin relaxation time, T2, determines the peak width, Γ, these two magnitudes can be related with the expression Γ ) 2‚(3T2)-1/2, being Γ in megahertz and T2 in nanoseconds. Accordingly, values of T2 were estimated for the different samples. The calculated values, listed in the last column of Table 1, range from 38 ns for the pure ZnO to 75 ns for one of the hybrid samples. The dynamic behavior of the photoluminescence has been explained through the photodesorption of adsorbed oxygen species in section II. Paramagnetic species adsorbed at surface positions (e.g., O2-) should also be detectable in EPR spectra. O2- adsorbed at surface positions in ZnO has been reported to give a triplet signal placed at g ≈ 2.00-2.05.19,54,55 In our case, such signal was not detected at this position. However, this does not necessary contradict the presence of oxygen species. Adsorbed oxygen can react at the surface and lead to species without unpaired electrons; for example, it might react with zinc interstitial according to56

vacancies, VO+, and correlated it with the PL green emission band. However, this ascription has been refuted by other authors20,57 and is also contradictory to our results. In the PL visible emission spectra of ZnO-latex hybrids, a difference of up to more than a factor of 5 is observed in the integrated intensity (cf. Figure 2) with respect to pure ZnO, while the concentrations of paramagnetic defects determined by EPR differs by less than a factor of 3. Both intensities tend to decrease as the latex concentration increases, but the quenching of the visible emission in the presence of latex is much more significant than the reduction of the EPR intensity. The evolution of both signals with the latex concentration is also different. In addition, further measurements of samples not presented here indicated that some materials showing lower PL emission than others could have, however, higher spin concentration calculated from the EPR signal. These facts induce us to believe that, in our samples, the centers observed in EPR and PL spectra are different. The definite evidence that the EPR signal and the visible emission in our samples have different origins came from comparing the temperature dependence of both types of spectra. Figure 10 presents the temperature dependence of the EPR signal for the reference and a hybrid sample, which is taken as a representative example, prepared in the presence of 6 g‚L-1 of P(S-AA) latex. In stark contrast to the intensity of the PL visible emission (cf. Figure 7), the intensity of the EPR signal increases with increasing temperature. Analogous thermal behavior for the g ≈ 1.96 signal is found in the literature,53,58 but the contrary tendency (i.e., signal decreases with increasing temperature) has been also reported.51 Considering the increase of the EPR signal with the temperature, we assume that, at low temperature, the shallow-donor centers have two paired electrons and are EPR-silent. With increasing temperature, more thermal energy is available to excite one of the two electrons from the shallow-donor level to the nearby conduction band. Assuming that the EPR signal at g ≈ 1.96 is coming from unpaired electrons placed in this shallow-donor level, the increase of the signal with temperature would correspond to this excitation process. Accordingly, it should be possible to fit the temperature dependence to an Arrhenius-type activation expression with the form

Zni+ + 1/2O2(ads) + e- h Zni2+ - O2-

(9)

We suggest that, in the absence of light, the adsorbed oxygen species are essentially in a nonparamagnetic state and therefore cannot be detected by EPR (at least not in the form O2-). Many authors have studied the correlation between the PL emission and the EPR signals in ZnO, in an attempt to clarify the defect structure and to identify possible paramagnetic species adsorbed at surface positions. In a very often cited work, Vanheusden et al.27 assigned this signal to singly ionized oxygen

I(T) ) I0 + A exp

( ) -E kBT

(10)

706 J. Phys. Chem. B, Vol. 111, No. 4, 2007

Mun˜oz-Espı´ et al.

where I0 is the intensity at 0 K (that should be 0, if only one thermally activated paramagnetic center is present) and E is the activation energy. The fitting of the data appears to be clearly better if I0 * 0 is considered (compare the solid and dashed lines in Figure 10). This suggests that two different paramagnetic centers are responsible for the observed signal. The first one, center type I, is initially paramagnetic and will become diamagnetic by thermal activation. The second center type, type II, is diamagnetic at 0 K and paramagnetic in the thermal activated state; this center is predominant and determines the general tendency of the EPR signal with temperature. Thus, the general temperature dependence might be expressed by a twoterm equation of the form

I(T) )

I0

( )

-EI 1 + AI exp 2kBT

+ AII exp

( ) -EII 2kBT

(11)

where AI and AII are constants, I0 is the intensity at 0 K due to centers of type I, and EI and EII are the thermal activation energies for the two different paramagnetic centers. The first term corresponds to the contribution of the centers of type I and the second term corresponds to the centers of type II. Equation 11 has five fitting parameters, which makes the simultaneous determination of the activation energy for both center types a complex task. As a first approach, we assume that at high temperature (arbitrarily T > 220 K) the first term can be neglected. The activation energies for the centers of type II can be thereby calculated: 33 ( 2 meV for the pure ZnO sample and 38 ( 3 meV for the hybrid. The values of EII and AII calculated by this approximation can be now fixed in eq 11, and only three parameters remain to be adjusted: AI, I0, and EI. In this way, we can calculate the activation energies for the centers of type I. Values of 111 ( 54 and 126 ( 28 meV are obtained for the reference sample and for the hybrid, respectively, although in this case the error in the determination is large. In principle, the two different center types might correspond to two different shallow donors with energy levels placed close to the conduction band. The thermal evolution (decrease with temperature) of the centers of type I agrees with the typical behavior for shallow donors in ZnO.51 Although the centers of type II could correspond to a second shallow-donor level, we speculate that they can be also conduction electrons trapped in oxygen species adsorbed at the surface. In future work, optically detected electron paramagnetic resonance could provide further information to understand the relationship between the defect centers and adsorbed species in our materials and the optical and the paramagnetic properties. Conclusions Embedded latex nanoparticles have been shown to influence the optical and paramagnetic properties of ZnO precipitated from aqueous media. The materials are composed of an inorganic and largely undisturbed crystalline matrix in which the organic latex particles are embedded. In spite of the morphology changes and a content of incorporated latex up to 10 wt %, X-ray diffraction proved that the long-range order is conserved. However, photoluminescence (PL) spectroscopy and electron paramagnetic resonance (ERP) showed an effect of the polymer in the defect structure. A weak UV peak and a more intense very broad visible emission appear in all PL emission spectra, but the ratio and intensity of the peaks changes considerably, depending on the

concentration of latex added during crystallization. Latex acts as a quencher of the visible emission, which is significantly diminished in the presence of the polymer particles. Whereas the PL peak can be attributed to near-band-edge exciton annihilation, the visible emission, centered at ∼2.2 eV, is more complicated to understand. The observed broad band might be a sum of a green and a yellow-orange component. Under oxidative annealing of pure ZnO powders, a strong 2.0-eVcentered emission appears, which could arise from the presence of interstitial oxygen. Under similar conditions, hybrid ZnOlatex samples show a green (at ∼2.5 eV) and a yellow band. The green emission might be ascribed to intrinsic defects in ZnO (i.e., zinc interstitial or oxygen vacancies). Surface-adsorbed species seem to have a determining influence on the optical properties. Under continuous irradiation, a noticeable dynamic behavior of the photoluminescence is observed. This can be explained in terms of the photodesorption of adsorbed oxygen. We assume that O2 plays a role as a tunneling agent of electrons from the conduction band to a deep trap, probably related to oxygen vacancies or interstitial zinc and responsible for the green visible emission. When O2 is photodesorbed, the “electron-shuttle” species disappears and visible emission is bleached, while the UV peak increases due to the higher probability of excitonic recombination. Latex particles seem to block possible adsorption sites, and consequently, the visible emission is quenched. Under ambient conditions, the excitation light may originate the formation of hydroxide radicals able to interact with the surface of the ZnO particles. Future work should try to elucidate this point by a surface-sensitive technique. The study of the time dependence of the luminescence in an inert atmosphere should also help in the understanding of this complex system. Further information on the defect structure of the materials was obtained by electron paramagnetic resonance (EPR). EPR spectra of all samples showed an intense signal at g ≈ 1.96, the intensity of which tended to decrease, as with the PL visible emission, for increasing latex concentrations. However, no quantitative correlation was possible between the intensities of the PL and EPR signals; furthermore, the temperature dependence of both signals followed opposite tendencies. Contrary to results of some authors,27 this indicates that, in our samples, the centers responsible for visible emission and the centers responsible for the EPR signal are different. The analysis of the results showed that two different center types with opposite temperature behavior may contribute to this signal, one being initially paramagnetic at low temperatures and another one that achieves the paramagnetic state by thermal activation. We speculate that the first center type corresponds to shallow donors, whereas the second one could correspond to conduction electrons trapped in oxygen species adsorbed at the surface. Acknowledgment. We gratefully thank Hansjo¨rg Menges for technical assistance in the photoluminescence measurements. References and Notes (1) (a) Simon, P. F. W.; Ulrich, R.; Spiess, H. W.; Wiesner, U. Chem. Mater. 2001, 13, 3464-3486. (b) van Bommel, K. J. C.; Ariannna, F.; Shinkai, S. Angew. Chem., Int. Ed. 2003, 42, 980-998. (2) Yu, S.-H.; Co¨lfen, H. J. Mater. Chem. 2004, 14, 2124-2147. (3) Lu, C. H.; Qi, L. M.; Cong, H. L.; Wang, X. Y.; Yang, J. H.; Yang, L. L.; Zhang, D. Y.; Ma, J. M.; Cao, W. X. Chem. Mater. 2005, 17, 52185224. (4) Mun˜oz-Espı´, R.; Qi, Y.; Lieberwirth, I.; Go´mez, C. M.; Wegner, G. Chem.sEur. J. 2006, 12, 118-129. (5) Marentette, J. M.; Norwig, J.; Stockelmann, E.; Meyer, W. H.; Wegner, G. AdV. Mater. 1997, 9, 647-651. (6) Co¨lfen, H. Curr. Opin. Colloid Interface Sci. 2003, 8, 23-31.

ZnO-Latex Hybrids: A Spectroscopic Investigation (7) Balz, M.; Barriau, E.; Istratov, V.; Frey, H.; Tremel, W. Langmuir 2005, 21, 3987-3991. (8) Guillemet, B.; Faatz, M.; Gro¨hn, F.; Wegner, G.; Gnanou, Y. Langmuir 2006, 22, 1875-1879. (9) (a) Qi, L. M.; Co¨lfen, H.; Antonietti, M.; Li, M.; Hopwood, J. D.; Ashley, A. J.; Mann, S. Chem.sEur. J. 2001, 7, 3526-3532. (b) Li, M.; Co¨lfen, H.; Mann, S. J. Mater. Chem. 2004, 14, 2269-2276. (10) Yu, S.-H.; Co¨lfen, H.; Antonietti, M. Chem.sEur. J. 2002, 8, 29372945. (11) Antonietti, M.; Breulmann, M.; Go¨ltner, C. G.; Co¨lfen, H.; Wong, K. K. W.; Walsh, D.; Mann, S. Chem.sEur. J. 1998, 4, 2493-2500. (12) (a) O ¨ ner, M.; Norwig, J.; Meyer, W. H.; Wegner, G. Chem. Mater. 1998, 10, 460-463. (b) Wegner, G.; Baum, P.; Mu¨ller, M.; Norwig, J.; Landfester, K. Macromol. Symp. 2001, 175, 349-355. (13) (a) Taubert, A.; Palms, D.; Weiss, O.; Piccini, M. T.; Batchelder, D. N. Chem. Mater. 2002, 14, 2594-2601. (b) Taubert, A.; Kubel, C.; Martin, D. C. J. Phys. Chem. B 2003, 107, 2660-2666. (14) Peng, Y.; Xu, A. W.; Deng, B.; Antonietti, M.; Colfen, H. J. Phys. Chem. B 2006, 110, 2988-2993. (15) Look, D. C.; Claflin, B.; Alivov, Y. I.; Park, S. J. Phys. Status Solidi A 2004, 201, 2203-2212. (16) Wang, Z. L. J. Phys.: Condes. Matter 2004, 16, R829-R858. (17) Pearton, S. J.; Norton, D. P.; Ip, K.; Heo, Y. W.; Steiner, T. Prog. Mater. Sci. 2005, 50, 293-340. (18) O ¨ zgu¨r, U.; Alivov, Y. I.; Liu, C.; Teke, A.; Reshchikov, M. A.; Dogan, S.; Avrutin, V.; Cho, S.-J.; Morkoc¸ , H. J. Appl. Phys. 2005, 98, article no. 041301. (19) Neumann, G. In Current Topics in Materials Science; Kaldis, E., Ed.; North-Holland Publishing Company: Amsterdam, 1981; Vol. 7. (20) Djurisˇic´, A. B.; Choy, W. C. H.; Roy, V. A. L.; Leung, Y. H.; Kwong, C. Y.; Cheah, K. W.; Rao, T. K. G.; Chan, W. K.; Lui, H. T.; Surya, C. AdV. Funct. Mater. 2004, 14, 856-864. (21) Demir, M. M.; Mun˜oz-Espı´, R.; Lieberwirth, I.; Wegner, G. J. Mater. Chem. 2006, 16, 2940-2947. (22) Hecht, H.; Mollwo, E. Solid State Commun. 1971, 9, 2167-2171. (23) Dingle, R. Phys. ReV. Lett. 1969, 23, 579-581. (24) Garces, N. Y.; Wang, L.; Bai, L.; Giles, N. C.; Halliburton, L. E.; Cantwell, G. Appl. Phys. Lett. 2002, 81, 622-624. (25) Kro¨ger, F. A.; Vink, H. J. J. Chem. Phys. 1954, 22, 250-252. (26) Riehl, N.; Ortmann, H. Z. Elektrochem. 1956, 60, 149-151. (27) (a) Vanheusden, K.; Seager, C. H.; Warren, W. L.; Tallant, D. R.; Voigt, J. A. Appl. Phys. Lett. 1996, 68, 403-405. (b) Vanheusden, K.; Warren, W. L.; Seager, C. H.; Tallant, D. R.; Voigt, J. A.; Gnade, B. E. J. Appl. Phys. 1996, 79, 7983-7990. (28) Leiter, F.; Alves, H.; Pfisterer, D.; Romanov, N. G.; Hofmann, D. M.; Meyer, B. K. Physica B 2003, 340, 201-204. (29) (a) van Dijken, A.; Meulenkamp, E. A.; Vanmaekelbergh, D.; Meijerink, A. J. Phys. Chem. B 2000, 104, 1715-1723. (b) van Dijken, A.; Meulenkamp, E. A.; Vanmaekelbergh, D.; Meijerink, A. J. Lumin. 2000, 90, 123-128. (30) Bylander, E. G. J. Appl. Phys. 1978, 49, 1188-1195. (31) Egelhaaf, H.-J.; Oelkrug, D. J. Cryst. Growth 1996, 161, 190194. (32) Kohan, A. F.; Ceder, G.; Morgan, D.; Van, de Walle, C. G. Phys. ReV. B: Condens. Matter 2000, 61, 15019-15027. (33) Guo, B.; Qiu, Z. R.; Wong, K. S. Appl. Phys. Lett. 2003, 82, 22902292.

J. Phys. Chem. B, Vol. 111, No. 4, 2007 707 (34) Korsunska, N. O.; Borkovska, L. V.; Bulakh, B. M.; Khomenkova, L. Y.; Kushnirenko, V. I.; Markevich, I. V. J. Lumin. 2003, 102, 733736. (35) Lin, B. X.; Fu, Z. X.; Jia, Y. B. Appl. Phys. Lett. 2001, 79, 943945. (36) van Craeynes, F.; Maenhout, W.; Dekeyser, W. Phys. Status Solidi 1965, 8, 841-846. (37) Studenikin, S. A.; Golego, N.; Cocivera, M. J. Appl. Phys. 1998, 84, 2287-2294. (38) Liu, M.; Kitai, A. H.; Mascher, P. J. Lumin. 1992, 54, 35-42. (39) Wu, X. L.; Siu, G. G.; Fu, C. L.; Ong, H. C. Appl. Phys. Lett. 2001, 78, 2285-2287. (40) Liu, X.; Wu, X. H.; Cao, H.; Chang, R. P. H. J. Appl. Phys. 2004, 95, 3141-3147. (41) (a) Melnick, D. A. J. Chem. Phys. 1957, 26, 1136-1146. (b) Collins, R. J.; Thomas, D. G. Phys. ReV. 1958, 112, 388-395. (c) Medved, D. B. J. Chem. Phys. 1958, 28, 870-873. (d) Tanaka, K.; Blyholder, G. J. Phys. Chem. 1972, 76, 3184-3187. (e) Shapira, Y.; Cox, S. M.; Lichtman, D. Surf. Sci. 1975, 50, 503-514. (f) Zhang, D. H. Mater. Chem. Phys. 1996, 45, 248-252. (g) Li, Q. H.; Gao, T.; Wang, Y. G.; Wang, T. H. Appl. Phys. Lett. 2005, 86, article no. 123117. (42) Takahashi, Y.; Kanamori, M.; Kondoh, A.; Minoura, H.; Ohya, Y. Jpn. J. Appl. Phys. Part 1 1994, 33, 6611-6615. (43) Fujitsu, S.; Koumoto, K.; Yanagida, H.; Watanabe, Y.; Kawazoe, H. Jpn. J. Appl. Phys. Part 1 1999, 38, 1534-1538. (44) Bahnemann, D. W.; Kormann, C.; Hoffmann, M. R. J. Phys. Chem. 1987, 91, 3789-3798. (45) Greene, L. E.; Law, M.; Goldberger, J.; Kim, F.; Johnson, J. C.; Zhang, Y. F.; Saykally, R. J.; Yang, P. D. Angew. Chem., Int. Ed. 2003, 42, 3031-3034. (46) Jiang, D. S.; Jung, H.; Ploog, K. J. Appl. Phys. 1988, 64, 13711377. (47) Allen, P. B.; Cardona, M. Phys. ReV. B: Condens. Matter 1983, 27, 4760-4769. (48) Makino, T.; Chia, C. H.; Tuan, N. T.; Segawa, Y.; Kawasaki, M.; Ohtomo, A.; Tamura, K.; Koinuma, H. Appl. Phys. Lett. 2000, 76, 35493551. (49) Van de Walle, C. G. Physica B 2001, 308-310, 899-903. (50) Garces, N. Y.; Giles, N. C.; Halliburton, L. E.; Cantwell, G.; Eason, D. B.; Reynolds, D. C.; Look, D. C. Appl. Phys. Lett. 2002, 80, 13341336. (51) Hofmann, D. M.; Hofstaetter, A.; Leiter, F.; Zhou, H. J.; Henecker, F.; Meyer, B. K.; Orlinskii, S. B.; Schmidt, J.; Baranov, P. G. Phys. ReV. Lett. 2002, 88, article no. 045504. (52) Kasai, P. H. Phys. ReV. 1963, 130, 989-995. (53) Po¨ppl, A.; Vo¨lkel, G. Phys. Status Solidi A 1989, 115, 247-255. (54) Kakazey, N. G.; Sreckovic, T. V.; Ristic, M. M. J. Mater. Sci. 1997, 32, 4619-4622. (55) Walters, A. B.; Na, B. K.; Liu, C. C.; Vannice, M. A. J. Mol. Catal. A 2000, 162, 287-295. (56) Morazzoni, F.; Scotti, R.; Dinola, P.; Milani, C.; Narducci, D. J. Chem. Soc., Faraday Trans. 1992, 88, 1691-1694. (57) Halliburton, L. E.; Giles, N. C.; Garces, N. Y.; Luo, M.; Xu, C. C.; Bai, L. H.; Boatner, L. A. Appl. Phys. Lett. 2005, 87, article no. 172108. (58) Sancier, K. M. J. Phys. Chem. 1972, 76, 2527-2539.