Synthesis and Luminescence Properties of (N-Doped) ZnO Nanostructures from aDimethylformamide Aqueous Solution

Brigitte Sieber,*,† Hongqin Liu,‡,§ Gaelle Piret,‡ Jacky Laureyns,| Pascal Roussel,⊥

Bernard Gelloz,# Sabine Szunerits,‡ and Rabah Boukherroub*,‡

Laboratoire de Structure et Proprietes de l’Etat Solide, UMR CNRS 8008, UniVersite Lille 1, Batiment C6,59655 VilleneuVe d’Ascq, France, Institut de Recherche Interdisciplinaire (IRI), USR CNRS 3078, Parc de laHaute Borne, 50 aVenue de Halley, BP 70478, 59658 VilleneuVe d’Ascq, France, Institut d’Electronique, deMicroelectronique et de Nanotechnologie (IEMN), UMR CNRS 8520, Cite Scientifique, AVenue Poincare,BP 60069, 59652 VilleneuVe d’Ascq, France, The Institute for Chemical Physics, School of Science, BeijingInstitute of Technology, Beijing, 100081, People’s Republic of China, Laboratoire de Spectrochimie Infrarougeet Raman (LASIR), Batiment C5 - UMR CNRS 8516, UniVersite Lille 1, 59655 VilleneuVe d’Ascq, France,UCCS, Equipe de Chimie du Solide, UMR CNRS 8181, ENSCL et UniVersite Lille 1, BP 90108,59652 VilleneuVe d’Ascq, France, and DiVision of Electrical and Electronic Engineering, Faculty ofTechnology, Tokyo UniVersity of Agriculture and Technology, Koganei-shi, Tokyo 184 8588, Japan

ReceiVed: April 16, 2009; ReVised Manuscript ReceiVed: June 10, 2009

The paper reports on the optical properties of ZnO nanostructures elaborated on a zinc foil substrate by asimple chemical approach. The doping type and density of the ZnO nanostructures were evaluated usingelectrochemical impedance spectroscopy. XRD diffraction patterns and Raman spectroscopy were used tostudy the structural properties evolution upon thermal annealing at 300 °C for 1 h in air. Their optical properties,probed by low temperature photoluminescence and room temperature cathodoluminescence (CL), are correlatedto their electronic and structural properties. The luminescence of the nanorods is dominated by a broad nearband edge emission located in the blue-violet region of the optical spectrum. Analysis of the CL spectra andmonochromatic CL images show that the main luminescence has an extrinsic origin, which is tentativelyassigned to nitrogen impurities.

I. Introduction

Zinc oxide (ZnO) is one of the most studied semiconductorsdue to its outstanding properties. It is a direct, wide bandgapsemiconductor (Egap ) 3.36 eV) with a free exciton bindingenergy large enough (60 meV) to allow excitons to be stable atroom temperature. ZnO is used in blue/UV optoelectronics,1

transparent electronics,2 piezoelectric transducers,3 photovoltaicapplications,4 varistors,5 spintronic devices,6 and gas sensors.7

The preparation of ZnO nanostructures such as nanorods andnanowires has already been described in several reports. Theyhave been elaborated by many different techniques such asreactive sputtering,8 thermal evaporation,9 spray pyrolysis,10

oxidation of Zn,11 pulsed laser deposition,12 chemical vaportransport and condensation,13 and metal organic chemical vapordeposition.14

Among these various techniques, wet chemical approacheshave been increasingly used in the last years15-20 because theyrequire neither sophisticated equipment nor vigorous experi-mental conditions. Thus, they became more widely used nextto vapor-phase deposition techniques. Their optical propertieshave been explored by means of photoluminescence (PL),21-23

spatially and spectrally resolved cathodoluminescence (CL),

which offers a higher spatial resolution,24-28 or by the combina-tion of both PL and CL.29

In this paper, we report on the synthesis, electronic, mor-phological, and structural characterizations, and optical proper-ties of ZnO nanostructures elaborated by a simple chemicalapproach. The synthesis was performed following a slightlymodified approach of the simple and mild strategy for large-area fabrication of high-quality ZnO nanorod arrays proposedby Zhang et al.30 Different positions of the blue-violet band havebeen observed for different nanostructures.31 Because not allof them correspond to the expected excitonic transition, theyare likely related to defects.31 In this study, we put a specialfocus on the analysis of the shape and the origin of the dominantblue-violet emission band. We show that the intense blue-violetband is not always related to an excitonic transition. Theinfluence of postannealing at 300 °C in air has also been studied.

II. Experimental Section

A. Materials. Zinc foils (99.9%, 0.25 mm thick), dimeth-ylformamide (DMF), carbonate propylene, and lithium perchlo-rate (LiClO4) were obtained from Aldrich and used withoutfurther purification.

B. Preparation of ZnO Nanostructures. Zinc substrates(zinc foil cut into 1 × 1 cm2) were ultrasonically degreased inethanol, propanol, and water before use. The clean zinc foilwas immersed in a 5% dimethylformamide (DMF) aqueoussolution and heated up to 95 °C (oil bath) for 24 h, washedwith water, and finally dried in an oven at 130 °C for 1 h. Someof the samples were annealed afterward at 300 °C in air during1 h.32

* To whom correspondence should be addressed. E-mail: brigitte.sieber@univ-lille1.fr (B.S.); rabah.boukherroub@iemn.univ-lille1.fr (R.B.).

† UMR CNRS 8008.‡ USR CNRS 3078 and UMR CNRS 8520.§ Beijing Institute of Technology.| Batiment C5 - UMR CNRS 8516.⊥ UMR CNRS 8181.# Tokyo University of Agriculture and Technology.

J. Phys. Chem. C 2009, 113, 13643–13650 13643

10.1021/jp903504w CCC: $40.75 2009 American Chemical SocietyPublished on Web 07/01/2009

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

C. Characterizations of the Nanostructured ZnO Sub-strates. Scanning Electron Microscopy (SEM). SEM imageswere obtained using an electron microscope ULTRA 55 (Zeiss)equipped with a thermal field emission emitter and a highefficiency In-lens SE detector.

Raman Spectroscopy. Raman measurements were carried outat room temperature using a microspectrometer LABRAMJobin-Yvon. The 1 µm spot diameter on the sample surface wasproduced by the 514.5 nm line of a 10 mW Ar+ ion laser. Ramanspectra are recorded in backscattering geometry with the incidentand scattered light (not polarized) propagating parallel to thec-axis.

Electrochemical Impedance Spectroscopy (EIS). EIS experi-ments were performed using an Autolab potentiostat 30 (EcoChemie, Utrecht, The Netherlands). The Zn/ZnO nanowiresinterface was sealed against the bottom of a single compartmentelectrochemical cell (V ) 5 mL) by means of a rubber O-ring(the electrical contact was made to a copper plate through theZn). A platinum sheet and an AgCl-modified Ag wire were usedas counter and reference electrodes, respectively. EIS wasperformed using the following parameters: amplitude of 20 mV;frequency range of 10 kHz-1 Hz, potential range: -0.8-0.8V. The electrolyte was carbonate propylene/LiClO4 (0.1 M) toavoid ZnO decomposition.32,33

X-ray Diffraction. The XRD patterns were obtained fromθ-θ scans in a Bruker D8 XRD operating at 50 kV and 30mA with a Cu anticathode (λ ) 1.5418 Å).

Photoluminescence. PL spectra were measured while thesamples were in a cryostat under vacuum. The temperature wasvaried from 10 to 300 K. An optical multichannel analyzer(resolution: 1 nm) and the fourth harmonic line (266 nm) of aYAG laser (pulse duration: 12 ps; repetition rate: 10 Hz; power:4 mW; spot diameter: 6 mm) were used for detection andexcitation, respectively. Each spectrum was acquired during 10 sin order to average over many laser pulses. The area of thesample probed by the detector was about 200 µm in diameter.

Cathodoluminescence. The CL experiments were performedat 300 K in a Hitachi 4700 FESEM equipped with a Gatanparabolic mirror. The accelerating voltage of the electron beamwas 8 kV which corresponds to an electron penetration depthof 0.3 µm.34 The beam current is in the range 100-200 pA andthe working distance is equal to 12.4 mm. This corresponds toa focused beam spot size close to 30-50 nm. The spectralresolution of the CL system is equal to 10 meV.

III. Results and Discussion

A. Morphology of the ZnO Nanostructures. ZnO nano-structures investigated in this work were prepared by chemicaloxidation of Zn foils in a 5% dimethylformamide (DMF)aqueous solution at 95 °C as reported recently.30,32

The oxidation of metallic Zn by naturally dissolved oxygenin water is slow due to the formation of a passive oxide layer.The presence of DMF in the aqueous solution acceleratessignificantly the oxidation process of metallic Zn. The underly-ing mechanism for the ZnO nanostructures formation can berationalized in eqs 1-2.

Zn2+ ions produced under these conditions are continuouslyreleased in the DMF aqueous solution, leading to zinc hydroxideZn(OH)2 precipitation on the Zn surface. Working at an elevatedtemperature (95 °C in the present work) ensures that ahomogeneous nucleation process in solution and Zn(OH)2 doesnot block the further dissolution of zinc and allows theconversion of the hydroxide into ZnO.

Figure 1A shows a SEM image of a freshly grown ZnOsample. ZnO nanostructures appear as hexagonal prisms with afaceted hexagonal end face or faceted pyramids with a diameterranging from 60-600 nm. Annealing the ZnO nanostructuresup to 300 °C in air for 1 h does not induce any significant changein their morphology (Figure 1B).

B. Determination of the Charge Carrier Concentration.The charge carrier concentration was determined using elec-trochemical impedance spectroscopy (EIS) in a carbonatepropylene electrolyte (0.1 M LiClO4) to avoid ZnO decomposi-tion as described recently.32 An equivalent circuit, Rs(RZnCZn)-(RZnOCZnO), comprising the electrolyte resistance Rs in serieswith two parallel association of a resistance R with a capacitanceC describes the electrochemical system under investigation. TheRZnCZn couple presents the non-ZnO coated Zn interface, whileRZnOCZnO rationalizes the electrode structure comprising ZnOnanostructures and non-ZnO coated Zn. Figure 2 shows thechange of C-2 values versus to the interface applied potential.From the orientation of the slope of the Mott-Schottky plotsthe doping type of the ZnO nanostructures can be determined.

Zn + 12

O2 + H2O98DMF

Zn2+ + 2OH- (1)

Zn2+ + 2OH- f Zn(OH)298∆

ZnO + H2O (2)

Figure 1. SEM images of as-grown ZnO nanostructured film before(A) and after postannealing at 300 °C for 1 h (B).

13644 J. Phys. Chem. C, Vol. 113, No. 31, 2009 Sieber et al.

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

The positive slope in Figure 2 is characteristic of an n-typesemiconductor. Extrapolating the linear part of C2--E curvesallows determining the flat band potential EFB, while from theslope the apparent donor density ND can be determined usingeq 3.

where q is the electron charge (1.6 × 10-19 C), εZnO is thedielectric constant of ZnO (εZnO ) 10), εo is the permittivity offree space (8.85 × 10-14 F cm-1), A is the active surface, ESC

is the potential difference across the ZnO space-charge region,and EFB is the flat band potential.32 A linear behavior observedat potentials more positive than ∼0.0 V and a flat band potentialof EFB ∼ -0.13 ( 0.01 V/Ag/AgCl was determined for as-grown samples. Annealing at 300 °C for 1 h induces ananodically shift of the flat band potential and a EFB ∼ 0.17 (0.01 V/Ag/AgCl was determined. Assuming a total activesurface area of A ) 1.00 cm2,32 a donor concentration of ND )3.90 ( 0.88 × 1019 cm-3 was calculated for as grown ZnOnanostructured interfaces, which decreased to ND ) 1.36 ( 0.1× 1019 cm-3 after annealing the substrate at 300 °C for 1 h inair.

C. Structural Properties: XRD Patterns and Micro-Raman Spectra. XRD diffraction patterns (Figure 3) show thatthe nanostructures have a wurtzite structure and that they are

mostly oriented along the c-axis. This indicates that this axiscorresponds to the highest growth rate.

Figure 4 shows the Raman spectra of the ZnO nanostructuresbefore and after annealing at 300 °C. They consist of severalbands and most of them correspond to Raman-active phononmodes of wurtzite ZnO with a C6V symmetry. The Raman activezone-center optical phonons predicted by the group theory are2A1 + 2E2 + 2E1 + 2B1, where E1 and E2 are double degeneratemodes.35 Phonon modes of A1 and E1 symmetry are both Ramanand IR active, the E2 mode is only Raman active, and the B1

mode is silent (forbidden for both Raman and infrared excita-tions). Nonpolar phonon modes with symmetry E2 have twofrequencies: E2 (high) associated with oxygen atoms and E2

(low) associated with Zn sublattice. The phonons of A1 and E1

symmetry are polar phonons and, hence, exhibit differentfrequencies for the transverse-optical (TO) and longitudinal-optical (LO) phonons. All these vibration modes have beenreported in Raman spectra of bulk ZnO.35 In the backscatteringconfiguration, only the E2 (high), E2 (low), and A1 (LO) modesare allowed, following Raman selection rules. The E2 (low)mode, located at about 100 cm-1, is out of range of our spectra,which start at 200 cm-1. In both specimens, the dominant Ramanline is located at 437-438 cm-1 (Figure 4) and corresponds tothe E2 (high) vibration mode. This is a second indication of thewurtzite structure of the nanorods with a main orientation alongthe c-axis, as previously observed by XRD. Because no

Figure 2. Mott-Schottky plots of as-grown ZnO nanowires before(b) and after thermal postannealing for 1 h at 300 °C (9). Solution:LiClO4 (0.1 M)/carbonate propylene, ∆E ) 20 mV.

Figure 3. X-ray diffraction patterns of as prepared ZnO nanostructures(black line) and after postannealing in air at 300 °C for 1 h (gray line).*Peak due to Zn substrate.

1

C2) 2

qεZnOεoA2ND

(EZnO + EFB) (3)

Figure 4. Raman spectra of freshly grown ZnO nanorods (black lines)and after postannealing in air at 300 °C for 1 h (gray lines): (a) fullspectra between 250 and 750 cm-1 and (b) enlargement of the spectrabetween 420 and 690 cm-1.

Properties of (N-Doped) ZnO Nanostructures J. Phys. Chem. C, Vol. 113, No. 31, 2009 13645

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

significant variation of the position of E2 (high) mode has beenobserved, we can exclude the presence of strain in the nanorods.Figure 4 shows that many forbidden vibration modes are presentin the spectra: the 333 cm-1 mode that corresponds to a secondorder (E2 (high) - E2 (low)),35 the 380 and 417 cm-1 peaks,which can be assigned to the A1 (TO) and the E1 (TO) modes,respectively. The observation of such unexpected peaks is thefirst evidence of the presence of a structural and/or dopinginduced disorder within the nanostructured ZnO substrate.

The broadband visible in Figure 4 which ranges from 450 to630 cm-1 examined more carefully can be divided into two parts,one from 450 to 520 cm-1 and the other one from 520 to 630cm-1. First, the intensity of the shoulder between 520 and 630cm-1 remains unchanged after annealing since it can be eitherincreased or decreased. In fact, it seems to be dependent on thesize of the nanorod from which the micro-Raman spectrum isissued, but we do not have enough results at the present timeto give a definitive conclusion. Its shape changes from one placeto another in such a way that it looks like being composed ofdifferent modes. A combination of the forbidden E1 (LO) modeat 580 cm-1 and the allowed A1 (LO) mode expected at 574cm-1, respectively, could compose the high Raman shift side.These values of LO modes are indicative since they can beshifted when the crystal is disordered.36 The B1 (high) silentmode could also be involved in the shoulder since it is locatedclose to the A1 (LO) and the E1 (LO) modes of ZnO.37 But,because it usually comes along with the B1 (low) silent modeat 275 cm-1, which is not observed in our spectra, it may notbe present.37

The lower part of the broadband (Figure 4), which rangesfrom 450 to 520 cm-1 and whose intensity is very important infreshly grown specimens, is observed to decrease after post-thermal annealing at 300 °C. Also its intensity fluctuates fromplace to place on each specimen. Its precise origin is not soeasily determined since it is not composed of any apparentRaman peak.

In N-doped ZnO samples grown by CVD on GaN substrate38

as well as in Ga-N codoped ZnO layers,39 an extra Ramanpeak has been observed at 510 cm-1 and related to the presenceof nitrogen.38,39 But following the analysis by Manjon et al.37

and the results of ab initio calculations,35 this peak could beassigned to the 2B1 (low) second-order. The 2B2 (low) modeclose to 520 cm-1 could also be present in few spectra inFigure 4.37

The activation of all the forbidden vibration modes in oursamples as well as the existence of the broadband from 450 to630 cm-1 should be due to the breakdown of the translationalcrystal symmetry by defects and impurities. Considering thehigh doping level of the specimens, a doping-induced disordercan be thus proposed. Its main effect is obviously the occurrenceof a broad Raman band in our samples in the spectral range of450-520 cm-1. The decrease of its intensity after annealing aswell as that of the intensity of the forbidden modes A1 (TO)and E1 (TO) modes (Figure 4) could be related to the measureddecrease of the doping level, and probably to the reduction ofthe doping induced disorder. This suggestion is confirmed bythe observation of a broadened E2 (high) peak, whose line widthdecreases after annealing from 9.2 to 8.2 cm-1,40 but remainslarger than the 6-7 cm-1 usually observed in undoped ZnOsubstrates.38

As proposed previously,32 the reduction of the disordertogether with the reduction of the doping level after postan-nealing at 300 °C is most likely due to the out diffusion ofshallow donors such as interstitial zinc atoms Zni and hydrogen

atoms. But Zni are fast diffusers and, thus, unstable in n-typeZnO.41 Hydrogen atoms have already been proposed to inducedisorder;42 no change in our Raman spectra and impedancemeasurements has been observed when the annealing temper-ature is lower than 300 °C. Thus, hydrogen atoms are the mostprobable impurities responsible for the doping induced disorderand its reduction. Indeed, it was reported that for temperaturesabove 125 °C, all interstitial hydrogen atoms (Hi) diffuse outwithout the dissociation of hydrogen atoms on oxygen site(HO).43 The HO atoms start to evolve into Hi at around 225 °Cand completely diffuse out at temperatures above 475 °C.43

Specifically, hydrogen atoms should be responsible for the broadRaman band observed between 450 and 520 cm-1. We alsopropose to associate the spatial variation of the intensity of theRaman band between 450 and 520 cm-1 to doping levelfluctuations related to the strength of the doping induceddisorder.

D. Optical Properties of ZnO Nanostructures. The lumi-nescence emitted by the specimens is mainly located in the UVpart of the spectrum, as shown in Figure 5. In the following,we will focus on the origin of the UV band. Since its energypeak is located close to the stoichiometric band gap valueexpected at 3.36 eV at RT, it will be named NBE (near bandedge). The NBE band shown in Figure 6 is rather broad: itsaverage full width at half-maximum (fwhm) is 150 and 158meV in freshly grown and postannealed samples, respectively.These values are much larger than the thermal broadening equalto 1.8kBT, that is, to 45 meV at 300 K,44 and that thehomogeneous broadening of 40 meV expected at RT in highquality ZnO bulk samples.45 Several mechanisms can contributeto the broadening of the NBE band: (i) homogeneous and/orinhomogeneous strain that arises from stoichiometric fluctuationsand thermal expansion differences between the substrate andthe nanowires, (ii) disorder in the nanowires, (iii) band tailing,indirect band-to-band and band-to-impurity transitions. We havealready seen (§ IIIB) that the presence of strain can be excluded(mechanism i). Disorder in the nanowires (mechanism ii) hasbeen evidenced in freshly grown samples by Raman spectros-copy, which also showed that it was strongly reduced afterannealing at 300 °C. On one hand the average broadening ofthe NBE CL band is nearly constant before and after annealing.On the other hand it is as large at 4 K than at 300 K (105 meVat 10 K and 150 meV at 300 K, as detected by PL; Figure 7).This implies that disorder is certainly important in the ZnO

Figure 5. Typical CL spectrum recorded on freshly grown ZnOnanostructures. The spectrum has been obtained on a (10 × 10) µm2

area.

13646 J. Phys. Chem. C, Vol. 113, No. 31, 2009 Sieber et al.

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

nanorods before and after annealing, and that the broadeningof the whole NBE peak could have another origin that theH-related disorder as previously proposed from Ramanexperiments.

The third mechanism which could also explain the wideningof the NBE band holds usually in highly doped semiconductorsand therefore should be examined here more carefully. Whenthe ionized donor density (in n-type semiconductor) is largerthan the effective density of states neff of the conduction band(CB; degeneracy limit), excessive electrons fill the low statesof the CB. The Fermi level being above the bottom of the CB,indirect transitions become possible because the k selection rulesare violated as a result of an enhanced carrier scattering by theionized impurities. At low temperature, the band broadensasymmetrically with a low energy tail following a functionaldependence �(E - Egap) and a steeper high energy tail, but atroom temperature (RT) the band tends to a more Gaussianshape.46,47 This band-filling effect is known as the Burstein-Moss(BM) effect and leads to a blue shift of the energy position ofthe optical bands.48,49 A close examination of all the NBE CLbands such as that shown in Figure 6 shows that the spectralpeak positions are always shifted toward low energies by about150 meV from the stoichiometric band gap value expected at3.36 eV at RT.50 Additionally, the shape of the NBE bands isnot Gaussian at RT and their low energy tail does not followthe �(E - Egap) dependence at low temperature. Therefore wecan conclude that the broadening of the whole NBE band is

not connected with the BM effect, implying that the electronconcentration in the areas from which are issued the NBE bandsis lower than that measured by impedance spectroscopy, thatis, 3.9 1019 cm-3. This seems quite reasonable because it hasalready been shown that nonradiative recombinations aredominant at very high doping level.51 The depth underneath thenanostructures surface probed with the two techniques is verydifferent. Impedance spectroscopy probes the surface of thenanostructures. In CL, the information depth is evaluated to 300nm for an accelerating voltage of 8 kV.34 Also the monochro-matic (and also polychromatic) CL images recorded on all thesamples reveal a spatial heterogeneity of the luminescenceintensity (Figure 8). The broad shape of the NBE band rendersits origin difficult to determine, but it is worth trying because itseems to be a general feature of ZnO nanorods prepared at lowtemperature.52,53

The fact that the spectral peak position is red-shifted fromthe band-to-band (BB) transition expected in stoichiometric ZnOtogether with the absence of obvious BM effect allow us topropose that it is more related to a extrinsic transition than toan intrinsic BB one. Such a possibility is corroborated by theobservation of spectra such as that shown in Figure 6, whichexhibit a more symmetrical shape. This kind of NBE CLspectrum was only recorded in the freshly grown nanostructured

Figure 6. Two types of CL spectra recorded on ZnO nanostructuresat 300 K: (a) fit of the spectrum with two Gaussian curves (2G-spectra);(b) fit of the spectrum with two Gaussian curves (3G-spectra); (b)experimental data.

Figure 7. Temperature dependence of (a) PL spectra and (b) energyof the main PL peak.

Properties of (N-Doped) ZnO Nanostructures J. Phys. Chem. C, Vol. 113, No. 31, 2009 13647

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

ZnO; it represents about half of the spectra. From the shape ofthe spectrum displayed in Figure 6, it is now obvious that theband corresponds to the convolution of several transitions. Thespectra of the type 1 (Figure 6a) and of type 2 (Figure 6b) areeasily convoluted by 2 and 3 Gaussian bands, respectively. Inthe following they will be named 2G- and 3G-spectra. The firsttwo Gaussian bands with the lowest energies are labeled B-and V-bands and that with the highest energy UV-band. 3Gspectra were observed in CL spectra but not in PL spectra; thisis likely due to the smaller volume probed in the CL experiments.

The V-band is always the predominant band of the NBEemission. The fit of the spectra such as those in Figures 6 isalso justified by the observation of featureless and broad NBEbands even at low temperature (Figure 8) by monochromaticCL images which can appear different when recorded with theB- and the V-bands (Figure 8) and also by a spatial dependenceof the V-band to B-band intensity ratio.

The results of the fits of the 300 K CL-spectra are given inTable 1.

In the following we analyze the NBE band by first studyingthe UV-band in the freshly grown sample, thanks to itsoccurrence in the CL spectra. The UV-band ranges from 3.289to 3.363 eV with a mean value at 3.337 eV. Because it is justbetween the values expected for an excitonic and a BB transitionin stoichiometric ZnO (3.30 and 3.36 eV, respectively), it shouldcorrespond to an intrinsic transition. The excitonic transition isfavored in low doped semiconductors, whereas the BB transition

is favored in highly doped semiconductors. This shows that thedetermination of the UV-band origin is not so straightforwardbecause it requires the knowledge of the local electron concen-tration. Thus, it is further discussed below by also taking intoaccount the influence of the electron density on the optical bandgap.

When the carrier density in the conduction band exceeds theMott critical density for the semiconductor-metal transition, anarrowing of the band gap (BGN) is induced by the band tailingresulting from the merging of the donor band with theconduction band, leading to a red shift of the optical bands.This occurs in parallel to the BM effect. Additionally, thecarriers can be localized by the local variations of the ionizedimpurities density, resulting in a band tailing of both the valenceand conduction bands, and thus an increased BGN effect. Whenthe carrier density becomes even larger, electron-electronexchange and Coulomb interactions as well as the screenedelectron-ion interactions become significant, enhancing theBGN effect and thus the red shift of the optical bands.

Because the BGN effect is expected to compete with the BMeffect, the relation between the carrier density and the positionof the NBE peak is not so straightforward to establish. Theoptical experiments performed have often lead to contradictoryconclusions: by increasing the donor density in the degeneratedrange, a blue shift of the absorption edge and of the PL bandwas always observed, as a result of the BM effect. But thisblue shift could be preceded or followed by a red shift in therange 8 × 1018 to 2 × 1020 cm-3. And a fix experimental valueof the carrier concentration at which the blue shift changes toa red shift (or inverse) has not been found. Calculations haveshown that the optical band gap is constant for electronconcentrations lower than 2 × 1019 cm-3.52 Beyond thatconcentration, the blue shift should be preceded by a very slightred shift of a few meV. From optical absorption measurementson Al-doped thin ZnO films, the optical gap was found toincrease with the electron concentration approximately as n2/3

for ne < 4.2 × 1019 cm-3,53 followed by a sudden decrease at5.4 × 1019 cm-3 and an increase at about 1020 cm-3. Also, thevalue of the Mott density found in the literature varies from 3× 1017 cm-3 to 5.5 × 1019 cm-3 and even to 7 × 1019

cm-3.51,53-56 The discrepancy being very large, we have toassume a reasonable value of the Mott density in the following.We take the value of 3 × 1017 cm-3 calculated by Klingshirnet al.57 When the electron concentration is above the Mottdensity value of 3 × 1017 cm-3, the excitonic transition shouldprogressively be replaced by the BB transition. The value ofthe energy peak of the excitonic transition should remainunchanged since the decrease of its binding energy is compen-sated by the red shift of the band gap,57 whereas that of the BBtransition most probably undergoes a red shift. When theelectron density reaches a value of 5 × 1018 cm-3 thatcorresponds to the onset of a degenerate population,57 the BBtransition should blue shift a little as a result of a small BMeffect.

Thus, a precise identification of the nature of the NBE bandis not so straightforward. In the following we make theassumption that the UV band corresponds to the BB transitionsince a variation of its energy has been experimentally observed.The quenching of the free exciton recombination could alsoresult from the presence of hydrogen as a shallow donor.18 Inthe case of a nondegenerate electron density, the energy of thetransition being located at Eg+kT/2,44 the optical gap probedin our experiments should vary from 3.25 to 3.324 eV. Thiscorresponds to a shrinkage of the gap by the BGN effect. Thus,

Figure 8. SEM images of as-prepared ZnO nanostructures (a)secondaries electrons mode; (b and c) monochromatic CL imagesrecorded at (b) 3.228 eV (spectrum peak), (c) 3.024 eV.

13648 J. Phys. Chem. C, Vol. 113, No. 31, 2009 Sieber et al.

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

in our samples, the electron density should be larger than theMott density (3 × 1017 cm-3) but smaller than 5 × 1018 cm-3.A spatial variation of the doping level could account for theobserved fluctuation of the BB energy transition, in agreementwith the conclusions drawn from the analysis of the Ramanspectra.

The V-band, in the 3G-spectra, is located at about 120 meVbelow the intrinsic UV-band: its energy peak ranges from 3.178to 3.247 eV, with a mean value of 3.214 eV. Its intrinsic natureseems unlikely since it should give rise to a strong localizationof carriers, a process that can be excluded.

So the extrinsic nature of the V-band should now beconsidered. At room temperature, the donors are ionized thenrendering more probable a free to bound transition than adonor-acceptor pair (DAP) one.18 Also, in a n-type material,the donor-hole type transition (D°h) is very improbable.58 Inthe case of the (eA°) transition with a peak at energy EeA, theenergy level EA of the acceptor is given by EA ) Egap - EeA +kT/2. By assuming that the UV-band corresponds to a BBtransition, one finds that EA varies from 108 to 155 meV (witha mean value of 128 meV) above the valence band maximum.The lowest value of the energy level EA (108 meV) correspondsto the lowest value of the BB transition. The energy level EA isof the order of that usually attributed to nitrogen on asubstitutional oxygen site (NO) in the ZnO lattice,59,60 even if itis a little smaller. A different location of the hydrogen atomsaround the nitrogen atoms61 or a larger doping level in oursamples could explain this result. The close position of theV-band energy peak in 2G and 3G spectra allows us to suggestthat the origin of the V-band in all the spectra and samples isrelated to nitrogen. This is corroborated by the enhancement ofthe V-band intensity observed after exposure to a NH3 plasmatreatment.

The formation of NO and related complexes is very likely inZnO as shown by first-principles calculations.62 Furthermore,SIMS analysis of the nanostructured ZnO substrate indicatesthe presence of hydrogen and nitrogen.

Concerning the B-band, it seems reasonable to assume thatit is of the donor-acceptor pair (DAP) type since it is locatedabout 80 meV below the V-band. The energy of a DAPtransition is given by EDA ) Eg - ED - EA + e2/4πεr, whereε is the dielectric constant and r is the pair separation. Theaverage Coulomb energy ∆E ) e2/4πεr can be very roughlyestimated by assuming <r> ) (3/4πNA)1/3.63

Applied to two V-bands detected in the 3G-spectra, we findthat nitrogen could be also involved in the B-band: for instance,an EDA energy of 3.09 eV, corresponding to that of the V-band,is found for Ed ) 0.05 eV, Ea ) 0.11 eV, Na ) 1014-1015

cm-3, or Ed ) 0.1 eV, Ea ) 0.11 eV, Na ) 5 × 108 cm-3. ForEDA ) 3.14 eV, one finds Ed ) 0.035 eV, Ea ) 0.155 eV, Na) 1015-1016 cm-3, or Ed ) 0.05 eV, Ea ) 0.155 eV, Na ) 5× 1017 cm-3. So, the donor could be either hydrogen located at35 meV below the conduction band and/or a deeper donor likeinterstitial zinc atoms (Zni) for instance. It is also important tonote that the B-band seems to be linked with the V-band in thesense that they always occur together.

Annealing of the freshly grown sample in air at 300 °C leadsto a slight blue shift (11-13 meV) of the two bands (see Table1). This can be easily explained if we remember that the electrondensity n is in a range (3 × 1017-5 × 1018 cm-3) such that theband gap decreases with n. Then, the blue shift of all thetransitions could result from the decrease of n as it has beendetected by impedance spectroscopy.

IV. Conclusion

ZnO nanostructures were synthesized by a chemical dissolu-tion of a zinc foil in a 5% DMF aqueous solution at 95 °C. Theas-grown nanorods crystallize in the wurtzite structure withthe c-axis as the main growth axis. No modification of themorphology could be evidenced after postannealing of thenanostructures in air at 300 °C for 1 h. However, disorder inthe as-grown nanorods could be observed from the Raman andluminescence spectra. Its reduction after annealing at 300 °Csuggests that hydrogen atoms were involved in the disorder.Luminescence spectra exhibit a major near band edge (NBE)emission band tentatively assigned to nitrogen atoms in sub-stitutional oxygen sites. Analysis of the CL spectra andmonochromatic CL images show that the main luminescencehas an extrinsic origin.

Acknowledgment. Claude Vanmansart (LSPES) is greatlyacknowledged for his participation to the CL experiments. H.L.thanks the Chinese government for the China ScholarshipCouncil Award.

References and Notes

(1) Kim, H. S.; Lugo, F.; Pearton, S. J.; Norton, D. P.; YWang, Y.-L.;Ren, F. Appl. Phys. Lett. 2008, 92, 112108.

(2) Song, D. Y.; Aberle, A. G.; Xia, J. Appl. Surf. Sci. 2002, 195, 291.(3) Wang, X.; Zhang, J.; Zhu, Z.; Zhu, J. Appl. Surf. Sci. 2007, 253,

3168–3173.(4) Zhao, Q.; Yu, M.; Xie, T.; Peng, L.; Wang, P.; Wang, D.

Nanotechnology 2008, 19, 245706.(5) Subasri, R.; Asha, M.; Hembram, K.; Rao, G. V. N.; Rao, T. N.

Mater. Chem. Phys. 2009, 115, 677–684.(6) Cheng, C.; Xu, G.; Zhang, H.; Luo, Y. Mater. Lett. 2008, 62, 1617–

1620.(7) Son, J. Y.; Lim, S. J.; Cho, J. H.; Seong, W. K.; Kim, H. Appl.

Phys. Lett. 2008, 93, 053109.(8) Tuzemen, S.; Gur, E.; Dogan, S. J. Phys. D: Appl. Phys. 2008, 41,

135309.(9) Huang, M. H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.;

Weber, E.; Russo, R.; Yang, P. Science 2001, 292, 1897.(10) Bacaksiz, E.; Yılmaz, S.; Parlak, M.; Varilci, A.; Altunbas, M. J.

Alloys Compd. 2009, 478, 367–370.(11) Chao, L.-C.; Liau, C.-C.; Lin, S.-J.; Lee, J.-W. J. Vac. Sci. Technol.,

B 2008, 26, 2601.(12) Premkumar, T.; Manoravi, P.; Panigrahi, B. K.; Baskar, K. Appl.

Surf. Sci. 2009, 255, 6819–6822.(13) Reiser, A.; Raeesi, V.; Prinz, G. M.; Schirra, M.; Feneberg, M.;

Roder, U.; Sauer, R.; Thonke, K. Microelectron. J. 2009, 40, 306–308.(14) Rosina, M.; Ferret, P.; Jouneau, P.-H.; Robin, I.-C.; Levy, F.;

Feuillet, G.; Lafossas, M. Microelectron. J. 2009, 40, 242–245.(15) Zhang, Z.; Yu, H.; Shao, X.; Han, M. Chem.sEur. J. 2005, 11,

3149.(16) Greene, L. E.; Law, M.; Tan, D. H.; Montano, M.; Goldberger, J.;

Somorjai, G.; Yang, P. Nano Lett. 2005, 5, 1231.(17) Ahsanulhaq, Q.; Umar, A.; Hahn, Y. B. Nanotechnology 2007, 18,

115603.

TABLE 1: Average Values of the Energy Band Positions and their FWHM of the Gaussian Transitions that Compose the TwoKinds of NBE CL Spectra

B-bandECL (eV)

B-bandfwhm (meV)

V-bandECL (eV)

V-bandfwhm meV)

UV-bandECL (eV)

UV-bandfwhm (meV)

ICLV-band/ICLB-band

ICLV-band/ICLUV-band

freshly grown 3.131 220 3.214 129 3.337 87 3.1 23300 °C annealed 3.147 208 3.225 130 2.6

Properties of (N-Doped) ZnO Nanostructures J. Phys. Chem. C, Vol. 113, No. 31, 2009 13649

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w

(18) Ozgur, U.; Alivov, Y. I.; Liu, C.; Teke, A.; Reshchikov, M. A.;Dogan, S.; Avrutin, V.; Cho, S.-J.; Morkoc, H. J. Appl. Phys. 2005, 98,041301.

(19) Feng, X.; Feng, L.; Jin, M.; Zhai, J.; Jiang, L.; Zhu, D. J. Am.Chem. Soc. 2004, 126, 62.

(20) Ku, C.-H.; Wu, J.-J. J. Phys. Chem. B 2006, 110, 12981.(21) Perkins, L. C.; Lee, S.-H.; Li, X.; Asher, S. E.; Coutts, T. J. J. Appl.

Phys. 2005, 97, 034907.(22) Yang, M.; Yin, G.; Huang, Z.; Liao, X.; Kang, Y.; Yao, Y. Appl.

Surf. Sci. 2008, 254, 2917.(23) Djurisic, A. B.; Leung, Y. H.; Tam, K. H.; Ding, L.; Ge, W. K.;

Chen, H. Y.; Gwo, S. Appl. Phys. Lett. 2006, 88, 103107.(24) Yuan, X. L.; Zhang, B. P.; Niitsuma, J.; Sekiguchi, T. Mater. Sci.

Semicond. Process. 2006, 9, 146.(25) Zhou, X.; Kuang, Q.; Jiang, Z.-Y.; Xie, Z.-X.; Xu, T.; Huang, R.-

B.; Zheng, L.-S. J. Phys. Chem. C 2007, 111, 12091.(26) Yang, Y. H.; Chen, X. Y.; Feng, Y.; Yang, G. W. Nano Lett. 2007,

7, 3879.(27) Piechal, B.; Yoo, J.; Elshaer, A.; Mofor, A. C.; Yi, G.-C.; Bakin,

A.; Waag, A.; Donatini, F.; Dang, L. S. Phys. Status Solidi B 2007, 244,1458.

(28) Chen, C.-W.; Chen, K.-H.; Shen, C.-H.; Ganguly, A.; Chen, L.-C.; Wu, J.-J.; Wen, H.-I.; Pong, W.-F. Appl. Phys. Lett. 2006, 88, 241905.

(29) Hsu, J. W. P.; Tallant, D. R.; Simpson, R. L.; Missert, N. A.;Copeland, R. G. Appl. Phys. Lett. 2006, 88, 252103.

(30) Zhang, Z.; Yu, H.; Wang, Y.; Han, M.-Y. Nanotechnology 2006,17, 2994.

(31) Djurisic, A. B.; Leung, Y. H. Small 2006, 2, 944, and referencestherein.

(32) Liu, H.; Piret, G.; Sieber, B.; Laureyns, J.; Roussel, P.; Xu, W.;Boukherroub, R.; Szunerits, S. Electrochem. Commun. 2009, 11, 945–949.

(33) Mora-Sero, I.; Fabregat-Santiago, F.; Denier, B.; Bisquert, J.; Tena-Zaera, R.; Elias, J.; Levy-Clement, C. Appl. Phys. Lett. 2006, 89, 203117.

(34) Gruen, A. E. Z. Naturforsch. 1957, 12a, 89.(35) Serrano, J.; Romero, A. H.; Manjon, F. J.; Lauck, R.; Cardona,

M.; Rubio, A. Phys. ReV. B 2004, 69, 094306.(36) Falkovsky, A. Phys. ReV. B 2001, 64, 024301.(37) Manjon, F. J.; Mari, B.; Serrano, J.; Romero, A. H. J. Appl. Phys.

2005, 97, 053516.(38) Kaschner, A.; Haboeck, U.; Strassburg, M.; Strassburg, M.;

Kaczmarczyk, G.; Hoffmann, A.; Thomsen, C.; Zeuner, A.; Alves, H. R.;Hofmann, D. M.; Meyer, B. K. Appl. Phys. Lett. 2002, 80, 1909.

(39) Harima, H. J. Phys.: Condens. Matter 2004, 16, S5653.(40) Bergman, L.; Bremser, M. D.; Perry, W. G.; Davis, R. F.; Dutta,

M.; Nemanich, R. J. Appl. Phys. Lett. 1997, 71, 2157.

(41) Janotti, A.; Walle, C. G. V. d. J. Cryst. Growth 2006, 287, 58.(42) Windisch, C. F.; Exarhos, G. J.; Yao, C.; Wang, L.-Q. J. Appl.

Phys. 2007, 101, 123711.(43) Bang, J.; Chang, K. J. Appl. Phys. Lett. 2008, 92, 132109.(44) Bebb, H. B.; Williams, E. W. In Theory in Semiconductors and

Semimetals; Willardson, R. K., Beer, A. C., Eds.; Academic Press: NewYork, 1972; Vol. 8.

(45) Klingshirn, C.; Hauschild, R.; Fallert, J.; Kalt, H. Phys. ReV. B2007, 75, 115203.

(46) De-Sheng, J.; Makita, Y.; Ploog, K.; Queisser, H. J. J. Appl. Phys.1982, 53, 999.

(47) Bugajski, M.; Lewandowski, W. J. Appl. Phys. 1985, 57, 521.(48) Burstein, E. Phys. ReV. 1954, 93, 632.(49) Moss, S. Proc. Phys. Soc. B 1957, 70, 247.(50) Meyer, B. K.; Alves, H.; Hofmann, D. M.; Kriegseis, W.; Forster,

D.; Bertram, F.; Christen, J.; Hoffmann, A.; Stra�burg, M.; Dworzak, M.;Haboeck, U.; Rodina, A. V. Phys. Status Solidi B 2004, 241, 231.

(51) Makino, T.; Segawa, Y.; Yoshida, S.; Tsukazaki, A.; Ohtomo, A.;Kawasaki, M. Appl. Phys. Lett. 2004, 85, 759.

(52) Chen, J.; Ae, L.; Aidele, C.; Lux-Steiner, M. C. Appl. Phys. Lett.2008, 92, 161906.

(53) Tong, Y.; Liu, Y.; Shao, C.; Liu, Y.; Xu, C.; Zhang, J.; Lu, Y.;Shen, D.; Fan, X. J. Phys. Chem. B 2006, 110, 14714–14718.

(54) Sakai, K.; Kakeno, T.; Ikari, T.; Shirakata, S.; Sakemi, T.; Awai,K.; Yamamoto, T. J. Appl. Phys. 2006, 99, 043508.

(55) Lu, J. G.; Fujita, S.; Kawaharamura, T.; Nishinaka, H.; Kamada,Y.; Ohshima, T.; Ye, Z. Z.; Zeng, Y. J.; Zhang, Y. Z.; Zhu, L. P.; He,H. P.; Zhao, B. H. J. Appl. Phys. 2007, 101, 083705.

(56) Zeng, H.; Li, Z.; Cai, W.; Cao, B.; Liu, P.; Yang, S. J. Phys. Chem.B 2007, 111, 14311.

(57) Klingshirn, C.; Hauschild, R.; Fallert, J.; Kalt, H. Phys. ReV. B2007, 75, 115203.

(58) Reshchikov, M. A.; Korotkov, R. Y. Phys. ReV. B 2001, 64, 115205.(59) Look, D. C.; Reynolds, D. C.; Litton, C. W.; Jones, R. L.; Eason,

D. B.; Cantwell, G. Appl. Phys. Lett. 2002, 81, 1830.(60) Xiong, G.; Ucer, K. B.; Williams, R. T.; Lee, J.; Bhattacharyya,

D.; Metson, J.; Evans, P. J. Appl. Phys. 2005, 97, 043528.(61) Hu, J.; He, H. Y.; Pan, B. C. J. Appl. Phys. 2008, 103, 113706.(62) Lee, E.-C.; Kim, Y.-S.; Jin, Y.-G.; Chang, K. J. Phys. ReV. B 2001,

64, 085120.(63) Cao, B. Q.; Lorenz, M.; Rahm, A.; vonWenckstern, H.; Czekalla,

C.; Lenzner, J.; Benndorf, G.; Grundmann, M. Nanotechnology 2007, 18,455707.

JP903504W

13650 J. Phys. Chem. C, Vol. 113, No. 31, 2009 Sieber et al.

Dow

nloa

ded

by U

NIV

LIL

LE

1 o

n A

ugus

t 19,

200

9Pu

blis

hed

on J

uly

1, 2

009

on h

ttp://

pubs

.acs

.org

| do

i: 10

.102

1/jp

9035

04w