defects in ZnO nanorods prepared by a hydrothermal method

Defects in ZnO Nanorods Prepared by a Hydrothermal Method

K.H.Tam,?C.K.Cheung,?Y.H.Leung,?A.B.Djurisˇic′,*,?C.C.Ling,?C.D.Beling,?

S.Fung,?W.M.Kwok,?W.K.Chan,?D.L.Phillips,?L.Ding,§and W.K.Ge§

Department of Physics,The Uni V ersity of Hong Kong,Pokfulam Road,Hong Kong,Department of Chemistry,

The Uni V ersity of Hong Kong,Pokfulam Road,Hong Kong,and Department of Physics,The Hong Kong

Uni V ersity of Science and Technology,Clear Water Bay,Hong Kong

Recei V ed:May26,2006;In Final Form:July18,2006

ZnO nanorod arrays were fabricated using a hydrothermal method.The nanorods were studied by scanning

electron microscopy,photoluminescence(PL),time-resolved PL,X-ray photoelectron spectroscopy,and positron

annihilation spectroscopy before and after annealing in different environments and at different temperatures.

Annealing atmosphere and temperature had significant effects on the PL spectrum,while in all cases the

positron diffusion length and PL decay times were increased.We found that,while the defect emission can

be significantly reduced by annealing at200°C,the rods still have large defect concentrations as confirmed

by their low positron diffusion length and short PL decay time constants.

1.Introduction

Fabrication of ZnO nanostructures1-4and the optical proper-ties of ZnO3-15have been extensively studied.Among the various methods for fabrication of ZnO nanostructures,the hydrothermal method1-4is of particular interest since it is a low cost,environmentally friendly method with growth tem-peratures as low as90°C.As a natural consequence of solution-phase growth at low temperature,the as-grown nanorods have large defect concentrations and typically exhibit weak UV but prominent defect emission at room temperature.3,4While the morphology and orientation of the nanorods can be significantly improved by using the ZnO seeds prepared from zinc acetate solution instead of ZnO nanoparticles,1improvements of their optical quality have not been investigated in detail.While it has been shown that the room-temperature photoluminescence (PL)spectra of nanorods can be changed by annealing under different conditions,3,15other techniques have not been used to conduct a comprehensive investigation of the effects of anneal-ing on their properties.It is generally assumed that a large UV to visible emission ratio implies excellent quality of the ZnO nanostructures,but typically,no other techniques are used to verify whether the defect density is indeed low.In addition, the origin of the visible emission is highly controversial.16ZnO can exhibit different emissions in the visible range(violet,blue, green,yellow,and orange-red)which are associated with defects in the material.Green emission is the most commonly observed defect emission in ZnO.However,there is still no consensus on the origin of this emission,and a number of different hypotheses have been proposed.16The possibility that green emissions in different samples can have different origins despite similar position and width has also been proposed.16If this is indeed the case,then the origin of the defect emission cannot be postulated simply by noting the similarities with previously published spectra for ZnO prepared by other fabrication methods.Therefore,considerable interest exists in studying the defect emissions in ZnO in general and ZnO nanorod arrays in particular due to their great potential for practical applications. Since applications of nanorod arrays in optoelectronic devices require control of their emission properties,it is necessary to perform a comprehensive study of their optical properties. Here,we report an investigation of the properties of ZnO nanorods using field emission scanning electron microscopy (SEM),variable temperature PL,time-resolved PL,X-ray photoelectron spectroscopy(XPS),and positron annihilation spectroscopy(PAS).We show that PL spectra do not provide sufficient evidence of defect density in the material.Contrary to common assumption,it is possible to have strong UV emission and weak defect emission and still have considerable defect concentrations in the sample.Thus,to fully characterize sample properties,other measurements in addition to PL spectroscopy are necessary.

2.Experimental Section

The nanorod arrays were prepared from solutions of zinc nitrate hydrate and hexamethylene tetramine on Si substrates with a seed layer prepared from zinc acetate solution.1Poly-ethyleneimine was added to the solution to increase the nanorod aspect ratio.2The morphology of the nanorods before and after annealing was examined using a Leo1530field emission SEM. The rods were annealed in a tube furnace at temperatures200, 400and600°C.The annealing atmospheres were air(atmo-spheric pressure,relative humidity～55-60%),argon,nitrogen, and forming gas(90%N2/10%H2).For annealing in a gas flow, the gas flow rate was0.1Lpm and the pressure was～1Torr. For variable temperature PL measurements,samples were mounted at the coldfinger and placed in a closed-cycle He cryostat(APD Cryogenics,Inc.model HC-2).The excitation source was a HeCd laser(325nm).The spectra were dispersed by a spectrometer SPEX500M and recorded by a photo-multiplier tube R943and PDA-512-USB(Control Development Inc.)fiberoptic spectrometer.Time-resolved PL was measured by using the Kerr-gated fluorescence technique17with1ps

*To whom correspondence should be addressed.Tel:+852********. E-mail:dalek@hkusua.hku.hk.

?Department of Physics,The University of Hong Kong.

?Department of Chemistry,The Univeristy of Hong Kong.

§Department of Physics,The Hong Kong University of Science and Technology.20865

J.Phys.Chem.B2006,110,20865-20871

10.1021/jp063239w CCC:$33.50?2006American Chemical Society

Published on Web09/20/2006

excitation pulses at 267nm.The decay times have been determined by fitting the decay curves monitored at the peak position of the UV emission (～380nm).The decay times of the visible emission were considerably longer,of the order of nanoseconds,and thus could not be reliably determined,but it should be noted that annealing may affect both UV and visible emission decay times.It should also be noted that the ratios of emission peaks measured in different PL systems may be different due to the dependence of UV to defect emission ratio on excitation intensity.However,excitation intensity does not affect the peak positions.XPS spectra were obtained using PAL 102XPS,and the position of the carbon C 1s peak (285.0eV)was taken as a reference.

In the positron annihilation experiments,low-energy positrons were focused to a ～1mm diameter spot on the sample under magnetic guidance and the implantation energy of the positrons was varied between 0and 25keV.The principles of PAS have been reviewed in detail in refs 18and 19.The positrons implanted into the sample bulk would be rapidly thermalized and undergo diffusion.The diffusing positrons can be trapped by the neutral or negatively charged open volume defects or by the positron shallow traps (i.e.,ionized acceptors or disloca-tions).The positron would thus finally annihilate at the delocalized state or the localized defect state with a surrounding electron and emit a pair of gamma photons (511keV),which contain information about the electronic environment in which the positron annihilates.The Doppler broadening technique used in the present PAS study monitored the Doppler broadening of the 511keV annihilation radiation to obtain the so-called S parameter,which was defined as the ratio of the fixed central region of the 511keV annihilation peak to the total area of the peak.The measured S parameter represents a linear combination of the corresponding annihilation states,that is,S )Σf i S i ,where f i and S i are the fraction and the characteristic S parameter of the positrons annihilating at a state i .Thus,the measured S parameter is related to the fraction of positrons annihilating with the valence electrons.Because a positron annihilating from the trapped vacancy state usually has a higher overlap with the valance electrons which possess higher momentum with respect to the core electrons,the S parameter corresponding to the positron annihilating in the defect state is higher than that in the delocalized bulk state.On the other hand,the positron annihilating from the shallow trap state experiences a very similar electronic environment to the bulk and thus the corre-sponding S parameter is indistinguishable from the bulk value.Effective positron diffusion length L eff +,which is related to the mobility of the positron,was another parameter that could offer information of the positron trapping process,since trapping in a vacancy state or a shallow trap state reduces the L eff +value.The experimental data for “low-momentum”S parameter S (E )were fitted by the computer program VEPFIT,20taking into account positron implantation,positron diffusion,and positron annihilating at different possible sites.Relevant parameters

obtained by positron annihilation experiments for ZnO nanorods (as-grown and annealed at different conditions)are summarized in Table 1.

3.Results and Discussions

3.1.Electron Microscopy Characterization.Figure 1shows the morphology of ZnO nanorods.It can be observed that the rods are ～800nm long with diameters in the range from 55to 70nm,while the rod density is of the order of several hundred rods per μm 2.SEM images do not indicate significant changes of the nanorod morphology after annealing.To obtain better information on the nanorod structure and morphology,TEM measurements have been performed and the obtained results are shown in Figure 2.All of the rods grow along the [0001]direction.It can be observed that annealing in an oxidative environment (air,oxygen)does not affect the morphology of the nanorods but annealing in a reducing environment (forming gas)causes some surface damage and thinning of the rods.3.2.Photoluminescence and Positron Annihilation Spec-troscopy.While electron microscopy can provide information on overall morphology and structure of the nanorods,other experimental techniques are needed to study point defects.Thus,PL,time-resolved PL,PAS,and XPS have been used to examine the defects in ZnO nanorod arrays before and after annealing.It can be observed that annealing significantly affects the optical properties of the nanorods.Photoluminescence spectra have been measured at different temperatures to obtain more information about the nature of point defects in the rods.At low temperature,ZnO typically exhibits sharp bound exciton lines,commonly labeled I 0-I 11,6which correspond to the excitons bound to different donor and/or acceptor defects.The low-temperature (12K)PL spectra of the as-grown nanorods and nanorods

TABLE 1:Comparison of Properties of ZnO Nanorods Annealed under Different Conditions a

annealing conditions L eff +(nm)S PL decay times O 1/Zn O 1/O 2(O 1+O 2)/Zn

none

2.80.54726ps,30ps 0.440.69 1.07600°C,air 140.54626ps,49ps 0.410.64 1.04600°C,oxygen 150.55421ps,47ps 0.390.58 1.05600°C,nitrogen 390.55626ps,61ps 0.420.46 1.35600°C,Ar

280.55431ps,63ps 0.460.55 1.31600°C,forming gas 200.55726ps,120ps 0.320.29 1.41200°C,forming gas 50.56824ps,74ps 0.410.58 1.11400°C,forming gas

24

0.554

23ps,57ps

0.44

0.41

1.50

a

The areas of XPS peaks have been adjusted by the sensitivity factors,0.733for oxygen and 2.768for

Zn.

Figure 1.Representative SEM image of the ZnO nanorods:(a)top view and (b)side view.

20866J.Phys.Chem.B,Vol.110,No.42,2006

Tam et al.

annealed in different atmospheres exhibit considerable differ-ences,as shown in Figure 3.The spectra have been normalized for easier comparison of peak positions and peak widths.It can be observed that the as-grown nanorods exhibit a broad,asymmetric,featureless emission with a maximum at 3.379eV and a full-width at half-maximum (fwhm)of 25meV.Broad,featureless emission is expected in the case of a large defect concentration since energy band instead of discrete defect levels would be formed.On the other hand,the annealed samples exhibit sharp and narrow donor-bound exciton peaks.For all annealing conditions,the spectra are dominated by donor-bound exciton peaks,but the peak positions and consequently the identities of the donors are different.The PL intensity of the UV emission is also considerably enhanced after annealing,in agreement with the increase in the decay time (see Table 1).For annealing in air,the main peak is located at 3.364eV (fwhm )6meV)with a small shoulder at 3.369eV.The forming gas-annealed sample exhibits a peak at 3.366eV (fwhm )7meV)and a shoulder at 3.371eV,while the sample annealed in Ar shows only a peak at 3.366eV (fwhm )6meV).The sample

annealed in nitrogen exhibits a slightly broader peak with a fwhm of 9meV,which is centered at 3.368eV.All of the observed peaks are in good agreement with previously reported donor-bound exciton lines.6The weak peak at ～3.337eV is likely a two-electron satellite of the donor-bound exciton peak or possibly related to excitons bound to structural defects.5To obtain more information about the nature of defects in the nanorods,we performed detailed investigation on the influence of annealing in an oxidative (air)and a reducing (forming gas)environment.Figure 4shows the variable tem-perature PL spectra of as-grown nanorods (parts a and b of Figure 4),nanorods annealed in air (parts c and d of Figure 4),and nanorods annealed in forming gas (parts e and f of Figure 4).It can be observed that the UV emission intensity is the lowest for as-grown nanorods and the highest for

nanorods

Figure 2.Representative TEM images of ZnO nanorods:(a)as-grown,(b)annealed in forming gas at 600°C,(c)annealed in air at 600°C,and (d)annealed in oxygen at 600°C.The insets show corresponding HRTEM

images.

Figure 3.12K PL spectra of ZnO nanorods annealed in different atmospheres.The spectra have been normalized for easier comparison of peak

positions.

Figure 4.Temperature dependence of the UV and defect emission for ZnO nanorods (a and b)no annealing,(c and d)annealed in air,and (e and f)annealed in forming gas.

ZnO Nanorods Prepared by a Hydrothermal Method J.Phys.Chem.B,Vol.110,No.42,200620867

annealed in forming gas.The PL intensity in the UV spectral range at12K in the case of as-grown nanorods is～67times smaller than that of nanorods annealed in forming gas(600°C) and～41times smaller than that of nanorods annealed in air (600°C).

We can also clearly observe different behavior of defect emissions with temperature.As-grown nanorods exhibit a peak shift and decreasing intensity of the defect emission with increasing temperature.After annealing in air at600°C,red and green emission peaks can be observed.The red emission decreases significantly with increasing temperature,while green emission intensities remain similar at all temperatures.After annealing in forming gas,there is no change in green emission intensity up to～150K and then there is a small increase at 200and300K.Both green and red emissions show a negligible peak shift with increasing temperature.A different dependence on measurement temperature of the defect emissions observed indicates different origins of these emissions.It should also be

noted that although emergence of the red band has been previously reported for ZnO single crystals annealed in air,16 the temperature dependence of the green and red bands was different from the one observed in nanorods.While the green emission in air-annealed ZnO single crystal quenched rapidly above200K,16this was not the case for ZnO nanorods.In addition,the dependence of green emission intensity on tem-perature cannot be described by the commonly used expressions since it does not decrease with increasing temperature.On the other hand,red and yellow defect emission intensities decrease with increasing temperature as expected.The PL emission intensity temperature dependence is usually described with21,22

where R is the process rate parameter,E T is the activation energy,k is the Boltzmann constant,and T is temperature.The expression with two different activation energies E T1and E T221-23

is also commonly used.Equation2can be used to describe two competitive nonradiative channels21or to describe different behaviors in low-temperature and high-temperature regions.22,23 The value of the activation energy dominant at low temperatures, E T1,is commonly very low.22,23One possible interpretation of this is that such a low value would be expected to occur as a result of donor ionization in the case of a donor-acceptor pair emission.22However,it has been shown that the small values of E T1are a consequence of T-2temperature dependence of the capture section for carriers at recombination centers.22 Figure5shows the temperature dependence of the emission intensity of the yellow and red bands(yellow for as-grown rods, red for rods annealed in air and forming gas).The temperature dependences can be fitted with eq2.For E T1,small values(～6 meV for as-grown and rods annealed in air and16meV for rods annealed in forming gas)are obtained,in agreement with the literature.22Thermal activation energies E T2were253meV for yellow emission,47meV for red emission of samples annealed in air,and209meV for red emission of samples annealed in forming gas.For both annealing atmospheres,the emission in the visible spectral range could be decomposed into two Gaussian peaks,one in the red and one in the green spectral region.However,different activation energies indicate that despite the similarities of peak positions,the samples annealed in different atmospheres have very different nonradiative recombination rates and consequently different types and concentrations of point defects.The same conclusion can be reached from the comparison of the decay times of the UV emission(Table1).It should also be noted that,for the sample annealed in forming gas,there is an increase in green emission after the red emission is quenched(opposite to the air-annealed ZnO single crystals,where quenching of green emission leads to the rise in red emission16)which indicates possible competi-tion for holes between two different acceptor levels.16A similar phenomenon is also observed for blue and yellow defect emission bands in GaN.24The difference in temperature behavior of two different GaN samples(increase vs no increase of emission from one channel after the decrease of emission from the other channel),similar to that observed in air-and forming gas-annelaed ZnO nanorods,was explained by the differences in nonradiative recombination rates between the two samples.24 We have also investigated the influence of annealing tem-perature on the PL spectra.The room-temperature PL spectra for nanorods annealed in air and forming gas flow are shown in parts a and b of Figure6,while the comparisons between different annealing atmospheres at200and600°C are shown in parts c and d of Figure6.As-grown nanorods exhibit the UV emission peak and the broad yellow defect emission.They also have short PL decay times and a very short effective positron diffusion length.Since the effective positron diffusion length represents a measure of how much positrons trap into all of the defects,including both trapping and scattering defects, the short diffusion length for the as-grown sample is in good agreement with its poor luminescent properties.On the other hand,the S parameter is sensitive only to certain types of trapping defects,which in ZnO are Zn vacancy and vacancy clusters or complexes because O vacancy represents a weakly bound state similar to the ionized acceptor and the dislocation.25 Thus,the corresponding S parameter would not differ signifi-cantly from the bulk state.25This implies that a large measured S parameter would be associated with more positron events originating from the Zn vacancy and/or the vacancy cluster.On another hand,a low value of L eff+implies that more positrons are trapped in the various kinds of defects in ZnO,namely,O vacancy,Zn vacancy,vacancy cluster,and ionized acceptors. The high value of the S parameter and the short diffusion length in ZnO nanorods as compared with that of the ZnO epilayer grown on sapphire(measured to be0.502(0.002and76(1

I(T))

I

1+R exp(-E

T

/kT)

(1)

I(T))

I

1+R

1

exp(-E

T1

/kT)+R

2

exp(-E

T2

/kT)

(2)

Figure5.Temperature dependence of the intensities of different visible

luminescence lines in ZnO.

20868J.Phys.Chem.B,Vol.110,No.42,2006Tam et al.

nm)indicate that there is a large concentration of the vacancy type defects in the samples.

With the increase of annealing temperature,the positron diffusion length L eff +increases and so do the decay times obtained from time-resolved measurements.All of the samples exhibit biexponential decay,and the slow component exhibited a substantial increase after annealing,as shown in Table 1.After annealing at 200°C,the ratio of UV to visible emission is improved,with samples annealed in forming gas showing negligible defect emission,which is commonly considered as an indication of excellent crystal quality and low defect concentration.However,although L eff +has been increased after annealing at 200°C,it is still considerably lower than that of samples annealed at higher temperatures which exhibit larger defect emission.Also,annealing the samples in N 2,Ar,oxygen,or forming gas results in a higher value of the S parameter implying more Zn vacancy or vacancy cluster formation.However,the increase of the S parameter upon annealing in these samples was accompanied by the increase of L eff +,which could only be explained by the decrease in concentrations of weakly binding positron states such as O vacancy,ionized acceptor,or dislocations.It was plausible to conclude that annealing the samples in N 2,Ar,oxygen,and forming gas would have the effect of increasing the concentrations of Zn vacancy and/or vacancy cluster and decreasing the concentrations of O-vacancy,ionized acceptor,and/or dislocation.For the case of annealing in air,the S parameter had no significant change and L eff +increased with respect to the as-grown one,which implied no change in Zn vacancy/vacancy cluster concentration,

while the concentration of weakly binding positron traps was slightly decreased.However,there is no obvious,direct relation-ship between the L d +and PL decay times,in agreement with a previous study on ZnO epitaxial layers.26Positron annihilation experiments also exhibit a lack of obvious correlation with visible emission,in agreement with a previous study which assigned visible emission to interstitial defects (zinc for green and oxygen for yellow emissions).27Imperfect correlation between L d +and radiative lifetime may indicate that crystal imperfections play a significant role in obtained results,similar to GaN where imperfect correlation between L d +and dislocation density,as well as the radiative lifetime and dislocation density,was found.28The lack of obvious relationship between L d +and the PL decay times is likely due to the large number of possible defects in ZnO.While L d +is a measure of total defect concentration,UV emission PL decay times are sensitive to defects which participate in carrier relaxation following optical excitation.

While for all annealing atmospheres defect emission is reduced by annealing at 200°C,the UV to visible emission ratio worsens for annealing at 400°C and then improves again for 600°C.The defect peak shape is strongly dependent on the annealing atmosphere.After annealing in air at 400°C and 600°C,green and orange-red components can be observed.On the other hand,the samples annealed in forming gas show almost no visible emission for annealing at 200°C,while green emission appears for annealing at 400and 600°C.The samples annealed at 400°C exhibit longer L d +but shorter PL decay times and stronger defect emissions compared with samples annealed at 200°C.One possible explanation for this is that the nonradiative defect concentration is lower,but the inferior UV emission is caused by an increased career transfer to deep defect levels followed by their radiative relaxation and conse-quent decrease in the intensity ratio of UV and defect emissions.3.3.XPS Characterization.To clarify the issue of the possible origin of the observed defect emissions,XPS measure-ments were performed.The results obtained are shown in Figure 7and summarized in Table 1.The curves for samples annealed in air at 200°C (similar result as forming gas annealing at 200°C)and forming gas at 400°C (similar result as forming gas annealing at 600°C)have been omitted to improve the image clarity.The lower energy peak in the O 1s spectrum can be attributed to oxygen ions in the wurtzite ZnO structure,and the position of this peak in different reports was from 529to 530.7eV.29-31The higher energy peak,at 531-532eV,29-31is typically assigned to loosely bound oxygen on the surface,such as OH groups,30,31or O 2-ions in oxygen deficient regions.29Two peaks (O 1and O 2)are sufficient to achieve a good fit for O 1s spectra in all cases,although the shift of the higher

energy

Figure 6.Room-temperature PL spectra of ZnO nanorods for:different annealing temperatures for samples annealed in the following:(a)air and (b)forming gas,(c)different annealing atmospheres for annealing at 200°C,and (d)different annealing atmospheres for annealing at 600°

C.

Figure 7.XPS spectra of ZnO nanorods annealed under different conditions:(a)O 1s (b)Zn 2p 2/3.

ZnO Nanorods Prepared by a Hydrothermal Method J.Phys.Chem.B,Vol.110,No.42,200620869

O2peak position for samples annealed in forming gas indicates a possible small contribution of an additional peak at higher energy(peak at～533eV,attributed to water30,31).The peak shift of the O1s peak can also occur due to band bending.31As for the Zn2p3/2peak,small shifts to higher energy can be observed for all annealed samples.The O1/Zn ratio in all samples is lower than0.5,indicating oxygen deficiency in the bulk of the rods.

3.4.Origins of Defect Emissions.The origin of the visible emission in ZnO is highly controversial.It has been shown that different defect centers are responsible for green,yellow,and red emissions.9,32Yellow and red emissions also exhibited different decay properties,which were attributed to different initial states(conduction band and shallow donors).8For the origin of green emission,a number of hypotheses have been proposed,such as singly ionized oxygen vacancy V o+,10V o2+ center,11oxygen antisite,12and zinc vacancy V Zn.13It has been suggested that the green emission is associated with oxygen

deficiency,while the orange-red emission is associated with excess oxygen.5On the other hand,the red emission was also attributed to the interstitial zinc Zn i.14Yellow emission is commonly attributed to interstitial oxygen defects O i,3,4although some impurities such as Li could also contribute to this emission.4Unlike the green emission,the yellow emission was unaffected by surface modifications,4and assignment of the emission to O i has been confirmed by the reduction of this emission after annealing in a reducing environment.3,15Visible (green and yellow)emission in ZnO has also been attributed to the presence of Zn(OH)2on the surface.33,34Enhancement of the UV emission due to water desorption has also been demonstrated.35Hydrogen impurities also play a significant role in ZnO,36-38and the desorption temperatures of hydrogen and OH groups are different.35

As-grown nanorods exhibit yellow defect emission,while after annealing typically red and green components are present in the emission spectra,and for some annealing conditions, defect emission becomes negligible.The yellow emission was reduced with annealing in all cases after annealing at temper-atures as low as200°C,although the decrease was lower in air.The desorption rate of hydroxyl groups peaks at～150°C,35 and this desorption is likely responsible for the reduction of yellow defect emission after annealing at200°C.The assign-ment of this emission to the presence of hydroxyl groups is in agreement with previous reports on visible luminescence in ZnO nanocrystals.33,34Studies of the O-H local vibrational modes in ZnO also confirm that annealing at150°C results in the removal of OH groups,while H does not leave the sample for annealing at this temperature for30min.36However,O-H is only one type of hydrogen-related defect in ZnO,and the other defect has been shown to be thermally stable up to～500°C.36 The study of hydrogen treatment of ZnO nanorods followed by annealing at400°C demonstrated that the presence of hydrogen improves the UV to visible emission ratio,while annealing at 400°C restores the original UV to visible emission ratio.37Thus, we can conclude that a decrease in the UV to visible emission ratio after annealing at400°C is most likely due to the removal of hydrogen.It should be noted that an increase in the defect (green)emission after hydrogen plasma treatment was also reported in ZnO,which was attributed to surface damage of the samples,similar to our results for forming gas annealing.38 Therefore,the yellow emission of the as-grown sample is likely due to the presence of OH groups,while worsening of the UV to visible emission ratio can be attributed to the removal of hydrogen.

The green component of the emission(centered at～2.47eV) is present in all annealed samples which exhibit defect emission. It is prominent in the samples annealed in forming gas,which exhibit evidence of surface damage(Figure2b).A similar observation has been reported for hydrogen plasma treatment of ZnO.38Thus,the green emission likely originates from surface defects,in agreement with previous reports.4,32The appearance of a green component of the emission is in most cases observed together with an increased positron diffusion length and S parameter,as well as lower O1/O2ratios.The increase in the S parameter with the appearance of green emission,as well as a decrease of O1/O2ratio,are in agreement with the involvement of V Zn14or V Zn complexes,as well as surface trapping by chemisorbed oxygen.11It should be noted that samples annealed in oxygen and argon at600°C also have similar values of the S parameter and thus similar concentrations of zinc vacancy related defects,but they exhibit a higher O1/O2ratio.Also, calculations of the energy levels of different defects predict that a transition involving V Zn would result in a blue emission.12,39 The summary of calculated energy levels from different studies12,39-41are shown in Figure8.It should be noted that the majority of calculations only predict energies of single point defects.While the energy of the V o Zn i complex has been calculated to be2.16eV below the conduction band minimum,39 energy levels of other defect complexes are unknown.Calcula-tions also predict that V Zn and V o have low formation energies,42 while Zn i is the most likely shallow donor candidate but it has a higher formation energy and it is a fast diffuser and thus not likely to be stable.42Electron irradiation experiments have shown that V Zn are a part of two different defect complexes and that one of those complexes anneals out simultaneously with neutral V O after annealing at～550K.43Since all samples which exhibit green emission also exhibit a large S parameter value,while the opposite is not the case(i.e.,a large S parameter does not imply the existence of strong green emission),the likely origins of this emission are defect complexes related to V Zn and these defect complexes are mainly located at the surface.The assignment of this emission to a defect complex is also supported by the fact that single point defects,such as V Zn43and Zn i,42 are expected to be mobile at relatively low temperatures and thus would anneal out below600°C.

Concerning red emission,it should be noted that there is a difference in the peak positions of the red components for samples annealed in forming gas(～2.0eV)and samples annealed in air(～1.8eV).A different peak position,together with different behaviors with increasing temperature,

indicates Figure8.Energy levels of different defects in ZnO from different studies in the literature:(a)ref12,(b)ref40,(c)ref39,and(d)ref 41.V Zn,V Zn-,and V Zn2-denote neutral,singly charged,and doubly charged zinc vacancy,respectively.Zn i°and Zn i indicate neutral zinc interstitial while Zn i+denotes singly charged zinc interstial.V O°and V O denote neutral oxygen vacancy while V O+denotes singly charged oxygen vacancy.O i and H i represent oxygen and hydrogen interstitial, respectively.V O Zn i denotes a complex of oxygen vacancy and zinc interstitial.

20870J.Phys.Chem.B,Vol.110,No.42,2006Tam et al.

that these emissions may have different origins.Depth resolved cathodoluminescence measurements have shown that green, yellow,and red emissions originate from different types of defects.9However,it has been proposed that yellow and red near-IR emissions may involve the same final acceptor state but different initial states(conduction band or shallow donor).8 Both yellow3and red7emissions are commonly assigned to defects related to excess oxygen.The samples annealed in air which exhibit prominent red emission also exhibit higher O1/ O2and O1/Zn ratios,as well as lower S parameter values.This indicates that Zn vacancies or vacancy complexes are likely not involved in this emission.It should be noted that as-grown samples,samples annealed in air,and samples annealed in oxygen have a very similar composition based on XPS results. The positron diffusion lengths and PL decay times are similar for samples annealed in air and oxygen,while S parameter values are similar for as-grown and air-annealed samples.Yet, the PL spectra of these three samples are very different.This fact,together with the persistence of the emission for annealing up to600°C,indicates the possible involvement of defect complexes.The chemical nature of this defect complex requires further study,but comparatively lower S parameter values indicate that it does not involve V Zn-related complexes.

4.Conclusions

To summarize,the as-grown ZnO nanorods have large defect concentrations,but this can be substantially improved by annealing.Annealing at200°C in all cases results in consider-able improvement in the UV to visible emission ratio,but positron annihilation measurements indicate that a considerable number of defects are still present in the material.Thus,PL spectroscopy measurements are not a sufficient indication of the sample quality and defect density.The defect emission shape and intensity are strongly dependent on the annealing atmo-sphere,and there is a complicated relationship between data obtained from different measurements.The yellow emission in hydrothermally grown nanorods is likely due to the presence of OH groups,while green emission likely originates from surface defects.Both green and red emission possibly involve defect complexes,which contain V Zn in the case of green emission,while the chemical nature of defects responsible for red emission requires further study.

Acknowledgment.This work is partly supported by the Research Grant Council of the Hong Kong Special Administra-tive Region,China(HKU7019/04P,HKU7032/04P,and HKU 1/01C).Financial support from the Strategic Research Theme, University Development Fund,and Seed Funding Grant(ad-ministrated by The University of Hong Kong)are also acknowl-edged.

References and Notes

(1)Greene,L.E.;Law,M.;Tan,D.H.;Montano,M.;Goldberger,J.; Somorjai,G.;Yang,P.Nano Lett.2005,5,1231.

(2)Law,M.;Greene,L.E.;Johnson,J.C.;Saykally,R.;Yang,P. Nat.Mater.2005,4,455.

(3)Greene,L.E.;Law,M.;Goldberger,J.;Kim,F.;Johnson,J.C.; Zhang,Y.;Saykally,R.J.;Yang,P.Angew.Chem.,Int.Ed.2003,42,3031.

(4)Li,D.;Leung,Y.H.;Djurisˇic′,A.B.;Liu,Z.T.;Xie,M.H.;Shi, S.L.;Xu,S.J.;Chan,W.K.Appl.Phys.Lett.2004,85,1601.

(5)Teke,A.;O¨zgu¨r,U¨.;Dogˇan,S.;Gu,X.;Morkoc?,H.;Nemeth,B.; Nause,J.;Everitt,H.O.Phys.Re V.B2004,70,195207.

(6)Meyer,B.K.;Alves,H.;Hofmann,D.M.;Kreigseis,W.;Forster,

D.;Bertram,F.;Christen,J.;Hoffmann,A.;Strassburg,M.;Dworzak,M.; Haboeck,U.;Rodina,A.V.Phys.Status Solidi B2004,241,231.

(7)Studenikin,S.A.;Golego,N.;Cocivera,M.J.Appl.Phys.1998, 84,2287.

(8)Lauer,R.B.J.Phys.Chem.Solids1973,34,249.

(9)Ong,H.C.;Du,G.T.J.Cryst.Growth2004,265,471.

(10)Vanheusden,K.;Seager,C.H.;Warren,W.L.;Tallant,D.R.; Voigt,J.A.Appl.Phys.Lett.1996,68,403.

(11)Dijken,A.van;Meulenkamp,E.A.;Vanmaekelbergh,D.;Meijerink,

A.J.Phys.Chem.B2000,104,1715.

(12)Lin,B.;Fu,Z.;Jia,Y.Appl.Phys.Lett.2001,79,943.

(13)Zhao,Q.X.;Klason,P.;Willander,M.;Zhong,H.M.;Lu,W.; Yang,J.H.Appl.Phys.Lett.2005,87,211912.

(14)Gomi,M.;Oohira,N.;Ozaki,K.;Koyano,M.Jpn.J.Appl.Phys. 2003,42,481.

(15)Cross,R.B.M.;De Souza,M.M.;Sankara Narayanan,E.M. Nanotechnology2005,16,2188.

(16)O¨zgu¨r,U¨.;Alivov,Ya.I.;Liu,C.;Teke,A.;Reshchikov,M.A.; Do’’an,S.;Avrutin,V.;Cho,S.-J.;Morkoc?,H.J.Appl.Phys.2005,98, 041301.

(17)Ma,C.;Kwok,W.M.;Chan,W.S.;Zuo,P.;Kan,J.T.W.;Toy, P.H.;Phillips,D.L.J.Am.Chem.Soc.2005,127,1463.

(18)Positron Beams and Their Applications;Coleman,P.,Ed.;World Scientific:Singapore,2000.

(19)Schultz,P.J.;Lynn,K.G.Re V.Mod.Phys.1988,60,701.

(20)van Veen,A.;Schut,H.;De Vries,J.;Hakvoort,R.A.;Ijpma,M. R.AIP Conf.Proc.1990,218,171.

(21)Leroux,M.;Grandjean,N.;Beaumont,B.;Nataf,G.;Semond,F.; Massies,J.;Gibart,P.J.Appl.Phys.1999,86,3721.

(22)Kristok,J.;Collan,H.;Hjelt,K.J.Appl.Phys.1997,81,1442.

(23)Garcia,J.A.;Remo′n,A.;Mun?oz,V.;Triboulet,R.J.Cryst.Growth 1998,191,685.

(24)Armitage,R.;Yang,Q.;Weber,R.E.J.Appl.Phys.2005,97, 073524.

(25)Tuomisto,F.;Ranki,V.;Saarinen,K.;Look,D.C.Phys.Re V.Lett. 2003,91,205502.

(26)Koida,T.;Chichibu,S.;Uedono,A.;Tsukazaki,A.;Kawasaki,M.; Sota,T.;Segawa,Y.;Koinuma,H.Appl.Phys.Lett.2003,82,532.

(27)Liu,M.;Kitai,A.H.;Mascher,P.J.Lumin.1992,54,35.

(28)Chichibu,S.F.;Uedono,A.;Onuma,T.;Sota,T.;Haskell,B.A.; DenBaars,S.P.;Speck,J.S.;Nakamura,S.Appl.Phys.Lett.2005,86, 021914.

(29)Ye,J.D.;Gu,S.L.;Qin,F.;Zhu,S.M.;Liu,S.M.;Zhou,X.; Liu,W.;Hu,L.Q.;Zhang,R.;Shi,Y.;Zheng,Y.D.;Ye,Y.D.Appl. Phys.A:Mater.Sci.Process.2005,81,809.

(30)Ogata,K.;Komuro,T.;Hama,K.;Koike,K.;Sasa,S.;Inoue,M.; Yano,M.Phys.Status Solidi B2004,241,616.

(31)Coppa,B.;Davis,R.F.;Nemanich,R.J.Appl.Phys.Lett.2003, 82,400.

(32)Djurisˇic′,A.B.;Leung,Y.H.;Tam,K.H.;Ding,L.;Ge,W.K.; Chen,H.Y.;Gwo,S.Appl.Phys.Lett.2006,88,103107.

(33)Zhou,H.;Alves,H.;Hofmann,D.M.;Kriegseis,W.;Meyer,B. K.;Kaczmarczyk,G.;Hoffmann,A.Appl.Phys.Lett.2002,80,210.

(34)Norberg,N.S.;Gamelin,D.R.J.Phys.Chem.B2005,109,20810.

(35)Xie,R.;Sekiguchi,T.;Ishigaki,T.;Ohashi,N.;Li,D.;Yang,D.; Liu,B.;Bando,Y.Appl.Phys.Lett.2006,88,134103.

(36)Shi,G.A.;Stavola,M.;Pearton,S.J.;Thieme,M.;Lavrov,E.V.; Weber,J.Phys.Re V.B:Solid State2005,72,195211.

(37)Lin,C.-C.;Chen,H.-P.;Liao,H.-V.;Chen,S.-Y.Appl.Phys.Lett. 2005,86,183103.

(38)Strzhemecny,Y.M.;Nemergut,J.;Smith,P.E.;Bae,J.;Look,D.

C.;Brillson,L.J.,J.Appl.Phys.2003,94,4256.

(39)Xu,P.S.;Sun,Y.M.;Shi,C.S.;Xu,F.Q.;Pan,H.B.Nucl. Instrum.Methods B2003,199,286.

(40)Lima,S.A.M.;Sigoli,F.A.;Jafelicci,M.,Jr.;Davolos,M.R. Int.J.Inorg.Mater.2001,3,749.

(41)Nikitenko,V.Optics and Spectroscopy of Point Defects in ZnO. In Zinc Oxide-A Material for Micro-and Optoelectronic Applications; Norberg,N.H.,Terukov,E.,Eds.;Springer:Dodrecht,The Netherlands, 2005;pp69-81.

(42)Janotti,A.;Van de Walle,C.G.J.Cryst.Growth2006,287,58.

(43)Tuomisto,F.;Saarinen,K.;Look,D.C.;Farlow,G.C.Phys.Re V. B:Solid State2005,72,085206.

ZnO Nanorods Prepared by a Hydrothermal Method J.Phys.Chem.B,Vol.110,No.42,200620871