Electron energy loss spectroscopy of ZnO nanocrystals with differentoxygen vacancy concentrations

K. Dileep,1 L. S. Panchakarla,2 K. Balasubramanian,1 U. V. Waghmare,3 and R. Datta1,a)

1International Center for Materials Science, Jawaharlal Nehru Center for Advanced Scientific Research,Bangalore-560064, India2Chemistry and Physics of Materials Unit, Jawaharlal Nehru Center for Advanced Scientific Research,Bangalore-560064, India3Theoretical Sciences Unit, Jawaharlal Nehru Center for Advanced Scientific Research, Bangalore-560064,India

(Received 22 November 2010; accepted 13 January 2011; published online 22 March 2011)

ZnO nanocrystals with different oxygen vacancy concentrations were characterized by

high-resolution electron energy loss spectroscopy (HREELS). ZnO nanocrystals show a decrease in

green emission with increasing annealing temperature in an oxygen environment and which

significantly quenches in a sample annealed at 400 �C. O K and Zn L3 pre-edge absorption

structures of ZnO nanocrystals were studied by HREELS as a function of annealing temperature.

Absorption edge peak broadening and variation in pre-edge absorption edge structure of Zn L3

were observed in experimental electron energy loss spectroscopy spectra with different oxygen

defect concentrations. All electron density functional theory (DFT) based (WIEN2k) calculation of

electronic density of states and electron energy loss near edge structure were carried out with

different oxygen vacancy concentrations and compared with experimental observations.

Appearance of additional peaks in pre-edge electron energy loss structure with increasing oxygen

vacancy is shown to be due to the oxygen defect states in the bandgap just below the conduction

band. VC 2011 American Institute of Physics. [doi:10.1063/1.3555604]

I. INTRODUCTION

ZnO is an important semiconductor with a wide range of

device applications, such as light emitting diodes (LEDs),

laser diodes (LDs), transparent conductors, photovoltaics,

etc.1–4 ZnO has a wide bandgap of �3.3 eV at room temper-

ature and a large exciton binding energy (�60 meV) com-

pared to GaN (�21 meV). ZnO-based LEDs and LDs have

already been demonstrated.5,6 Transition metal-doped ZnO

is a promising system as a dilute magnetic semiconductor for

probable application in emerging areas of spintronics.7 ZnO

doped with transitions metal ions such as Mn2þ and Co2þ

was first reported to be ferromagnetic,8,9 but later studies

have cast serious doubt on this aspect.10–12 In fact, ZnO

nanoparticles themselves exhibit weak ferromagnetism at

room temperature.13 The ferromagnetism is attributed to sur-

face defects, probably oxygen vacancies.14 The bandgap of

ZnO (3.3 eV) can be tuned, for example, by alloying with Cd

(redshift) and Mg (blueshift).15 Undoped ZnO shows n-type

conductivity, whereas p-type doping remains to be an open

issue in ZnO and compensation of p-type carriers by n-types

has been reported to be responsible for this difficulty.16

Besides the usual UV emission from ZnO, defect related

green emission has also been reported.17–20 Chemically syn-

thesized ZnO nanocrystals exhibited broad green emission

in addition to the expected UV emission band.14 Green

emission was reduced with increasing annealing temperature

in oxygen atmosphere and was explained in terms of reduc-

tion in oxygen point defects in ZnO lattice. Here we report

Zn L3 and O K pre-edge electron energy loss structure of the

same ZnO nanocrystals annealed at different temperatures

where green emission is observed to be significantly

quenched after annealing at 400 �C, in oxygen atmosphere.

These experimental results were discussed with the help of

WIEN2k-based DFT calculations of electronic density of

states (DOS) and electron energy loss near edge spectra

(ELNES). Zn K absorption edge appears at �9670 eV, there-

fore the Zn L3 edge was chosen (�1020 eV) for both core

edge electron energy loss spectroscopy (EELS) experimenta-

tion and theoretical calculations to study the change in spec-

tral feature considering oxygen point defects in ZnO.

The x-ray absorption spectroscopy technique has been

widely used to study the defect structure of ZnO. Hsu et al.reported an x-ray absorption near-edge structure (XANES)

(Zn K absorption edge) of Co-doped ZnO annealed under

different conditions.21 ZnO film annealed in oxygen shows

an increase in p orbital DOS (given by the area under the

XANES absorption curve) compared with the sample

annealed under the Ar/H2 condition. The appearance of an

extra feature at the pre-edge structure of the Zn K absorption

edge was attributed to the increase in oxygen vacancy for

samples annealed under Ar/H2 gas and associated with the

highest saturation magnetization among samples annealed

under different environmental conditions. Theoretical calcu-

lation in real space based on the FEFF 8.2 code also supported

the appearance of an extra peak in the Zn K pre-edge struc-

ture, corresponding to an increasing number of O atom

a)Author to whom correspondence should be addressed. Electronic mail:

ranjan.bapi@gmail.com.

0021-8979/2011/109(6)/063523/7/$30.00 VC 2011 American Institute of Physics109, 063523-1

JOURNAL OF APPLIED PHYSICS 109, 063523 (2011)

removed from the structure. Muller et al. reported that the O

K edge absorption peak and associated fine structures wash

out with an increasing amount of oxygen vacancy in SrTiO3

thin film.22 The O K absorption edge of bulk ZnO can be

explained by the transition from O 1s to O 2p unoccupied

levels. O 2p states hybridize with Zn 4s states to give

ELNES in the energy range �530�539 eV. From 539 to

�550 eV, the main contribution is from O 2p states hybri-

dized with Zn p states. Above 550 eV is the transition to O

2p�Zn 4d hybridized levels.23 The Zn L3 corresponds to

electronic transitions from Zn 2p3/2 to Zn 3d and 4s states

hybridized with O 2p states.24 The appearance of four peaks

(named A, T, E, TT) in Zn L3 ELNES of w-ZnO was

explained by point-group symmetry in the first coordination

unit. (ZnO4)6� was taken to be in tetrahedral symmetry (Td)

and under this condition Zn 4s gave rise to a1, Zn 4d gave

rise to t2, and e molecular orbitals, respectively. 4p became

t2. Zn 4d and 4p hybridizes to form two t2 molecular orbi-

tals. The four peaks in Zn L3 EELS spectrum were assigned

to electron transitions to these four molecular orbitals of

w-ZnO.25 EELS has the advantage over XANES for its sim-

plicity, experimentation can be easily performed at localized

regions (approximately a few-nanometer region to a single

atomic column) in a transmission electron microscope.

However, quantitative EELS simulation have been very dif-

ficult due to computational difficulty to generate spectra

closely resembling its experimental counterpart. Fingerprint-

ing in EELS is widely used to compare absorption edge

spectra for an element in various compounds to derive any

inferences qualitatively. We have used the WIEN2k-based

TELNES2 program (which is fully relativistic, in a TEM at

300 kV, the speed of the electron is 0.88c, where c is the

speed of light in a vacuum) to calculate EELS spectra of

various oxygen deficient w-ZnO lattice. Relativistic calcula-

tion in EELS is essential for materials that are anisotropic or

have a lower symmetry. WIEN2k describes periodic system

in reciprocal space and orbitals are solved using DFT. These

orbitals are used to calculate an EELS cross section for the

final derivation of ELNES spectra. Core-hole consideration

is very important to quantitatively study ELNES. However,

change in the pre-edge structure in ELNES spectra with

varying oxygen vacancy does not require accurate quantita-

tive estimation of leading peak energy positions and hence

omitted from the present calculations. The calculation using

a large super cell in order to reduce the interaction between

neighboring core-holes renders the calculation computation-

ally highly expensive. However, neglecting core-hole effects

may introduce incorrect relative peak height at the pre-edge

feature, which may not correspond with the experimental

EELS spectra well. The TELNES2 program calculates EELS

spectra nearly 50 eV from the onset of the core absorption

edge fairly well and for higher energy losses up to 100 eV,

the FEFF 9 code is the appropriate choice.26 In the present

report we have studied primarily the pre-edge feature at the

core absorption edge of both experimental and theoretical

spectra.

II. METHODS

A. Experimental details

ZnO nanocrystals were synthesized by a procedure al-

ready described in the literature.14 As synthesized particles

were annealed at 200 and 400 �C in oxygen flow [�500

SCCM (where SCCM denotes cubic centimeters per minute

at standard temperature and pressure)], for 10 h. The photolu-

minescence spectra of the nanocrystals were recorded with a

Perkin–Elmer model LS55 luminescence spectrometer. Elec-

tron energy loss near edge structures were acquired using a

gun monochromator (both in image and diffraction mode) in

a FEI-TITAN3TM 80–300 kV transmission electron micro-

scope operating at 300 kV with an energy resolution better

than 0.18 eV. In order to improve the signal over noise, cam-

era length and energy dispersion scale were reduced to 73

mm and 0.03 eV/pixel, respectively, which reduces the energy

resolution from 0.18 to 0.28 eV, as measured from full-width

at half-maxima of zero loss peak. A Gatan imaging filter en-

trance aperture size of 1 mm was chosen, which ensured high

energy resolution during spectra acquisition. A near edge

structure up to 50 eV from the absorption edge for both O K

FIG. 1. ZnO nanocrystals: (a) as-synthe-

sized at room temperature, (b) annealed

at 200 �C, and (c) annealed at 400 �Cunder oxygen.

FIG. 2. Electron diffraction pattern from three different samples showing no

change in hexagonal wurtzitic structure.

063523-2 Dileep et al. J. Appl. Phys. 109, 063523 (2011)

and Zn L3 edge were acquired for samples with different oxy-

gen vacancy concentrations and emphasis was given to

changes in the features at the onset of the absorption edge.

B. Calculation details

We calculated the DOS and ELNES (with TELNES2 pro-

gram in WIEN2k) with a full potential linear augmented plane

wave basis to study the change in density of states with oxygen

vacancy concentrations and respective ELNES spectra both

for O K and Zn L3 edges. The generalized gradient approxima-

tion was employed with the Perdew-Burke-Ernzenhof (PBE)

functional. Four cases were studied; (i) ZnO with no vacancies,

(ii) 25% O vacancies simulated with a super cell of ZnO (four

Zn, four O, 2� 1� 1 unit cell) with one O atom removed, (iii)

12.5% O vacancies simulated with a super cell of ZnO (eight

Zn, eight O, 2� 2� 1 unit cell) with one O atom removed,

(iv) 6.25% O vacancies simulated with a super cell of ZnO (16

Zn, 16 O, 2� 2� 2 unit cell) with one O atom removed. We

used wurtzite structure with space group P63mc generated

with lattice parameters of a¼ 3.2973 A and c¼ 5.2824 A and

atomic positions given by Zn (2/3, 1/3, 0) and O (2/3, 1/3, u),

where u¼ 0.3775. Lattice parameters of all the structures were

relaxed internally (using the minimization program min_lapw).

The following values were used for various simulation param-

eters: Rmt¼ 2.00 and 1.77 a.u. for Zn and O atoms, respec-

tively, the valence and core states were separated by �6.0 Ry

of energy, and Rk(max)[(Rmt(min)� kmax)]¼ 7.00. We sampled

integration over Brillouin zone with uniform 15� 15� 9 mesh

of k points. We used instrumental broadening of 0.2 eV, col-

lection angle �2.5 mrad, and convergence angle of �0 (paral-

lel illumination) for a meaningful comparison between the

theoretical and experimental spectra.

III. RESULTS AND DISCUSSION

Before we start discussing ELNES spectra of ZnO

with different oxygen vacancy concentrations, we provide

some information of ZnO nanocrystals. Figure 1 shows

HRTEM images from three different ZnO nanocrystals.

The average size of the as-synthesized ZnO nanocrystals at

FIG. 3. Powder x-ray diffraction pattern from three different samples show-

ing no change in hexagonal wurtzitic structure. The fine peaks are not seen

in the case of as-prepared sample due to broadening from small particle size.

FIG. 4. (Color online) Photoluminescence of ZnO nanocrystals of

as-synthesized and annealed at different temperatures.

FIG. 5. (Color online) (a) Experimental

O K edge ELNES and (b) its first deriva-

tive. The pre-edge features are not clear

in this case.

063523-3 Dileep et al. J. Appl. Phys. 109, 063523 (2011)

room temperature (RT), and annealed at 200 and 400 �Care 12 (6 3), 22 (6 5), and 33 (6 5) nm, respectively. Af-

ter annealing, the crystal, have grown in size. ZnO nano-

crystals have wurtzite structure and the crystal structure

remained unaltered upon annealing at various temperatures

as evident from both electron and powder x-ray diffraction

data (Figs. 2 and 3, respectively). Although it is possible

to quantify oxygen occupancy in a given crystal lattice by

HRTEM,27 applying the same methodology to a nanocrys-

talline material form is extremely difficult (as it involves

tilting nanocrystals toward zone axis orientation and they

are not stable for a sufficient time under electron beam ex-

posure). Therefore, we have studied the defect structure in

ZnO through the observation of core absorption edge

ELNES and the emphasis is given to any changes in the

pre-edge feature. From the photoluminescence of three dif-

ferent ZnO nanocrystals (see Fig. 4), one can clearly

observe that green emission is quite intense and broad for

an as-synthesized RT ZnO sample compared to the sample

annealed at 200 �C. Whereas green emission disappeared

significantly for the sample annealed at 400 �C in oxygen,

this suggests that green emission may be related to oxygen

deficiency in ZnO and quenched significantly by annealing

that might have replenished oxygen in the ZnO lattice.

From experimental ELNES spectra and its first derivative

of O K edge (Fig. 5), and Zn L3 edge (Fig. 6) for three dif-

ferent samples, one observes that the fine features in the

spectra are washed out in both O K and Zn L3 with

increasing oxygen vacancy in the lattice structure. Other

appreciable differences in the spectra are labeled. In the O

K spectra, there are no appreciable changes at pre-edge

features as a function of O vacancies, but smearing of the

fine features is observed. The absence of prominent pre-

edge peaks may be because the O atoms come only in the

second coordination shell of any O vacancy. In contrast,

there are clear pre-edge features for the three samples in

the Zn L3 spectra. The four main expected peaks of ZnO

are labeled A, E, T, and TT in Fig. 6. The relative peak

height of b to a is increasing with oxygen vacancy concen-

trations. There are also extra peaks like c, d, etc. that come

up at higher oxygen deficiencies. The change in the pre-

edge structure can be made more prominent from the first

derivative spectra. The ratio of b to a increases marginally

(0.23 – 0.5) with increasing concentrations of oxygen

deficiencies.

We have compared our experimental spectra with theo-

retical ELNES to get better insight into change in pre-edge

structure with increasing oxygen vacancy defects in terms of

change in the electronic density of states. Figures 7(a) and

7(b) show the theoretical ELNES spectra of O K and Zn L3

edges, respectively, of ZnO with different oxygen vacancy

concentrations (i.e., no vacancy, 6.25%, 12.5%, and 25%).

One interesting feature in these spectra is the appearance of

a pre-edge feature with the oxygen vacancy (VO) introduced

FIG. 6. (Color online) (a) Experimental

Zn L3 edge ELNES and (b) its first deriv-

ative. The pre-edge features are labeled.

FIG. 7. (Color online) Theoretical DFT

calculation of ELNES spectra of (a) O Kedge and (b) Zn L3 edge for various oxy-

gen vacancy concentrations.

063523-4 Dileep et al. J. Appl. Phys. 109, 063523 (2011)

FIG. 8. (Color online) Superposition of ELNES and DOS

for (a)–(d) O K and (e)–(h) for Zn L3 edge.

063523-5 Dileep et al. J. Appl. Phys. 109, 063523 (2011)

in the ZnO lattice. Peaks A and B in both O K and Zn L3

edges arise from p unoccupied levels of O hybridized with

4s levels of Zn. Peak B and additionally peak C become

more prominent with increasing VO. Also, the ratio between

peaks B and A increases with increasing VO concentrations.

One makes a similar observation in the Zn L3 pre-edge struc-

ture. EELS spectrum provides information about the density

of the unoccupied states, which is related to local atomic

environments. Therefore, we compare ELNES of both O Kand Zn L3 with unoccupied partial-DOS and present details

in Fig. 8, plotting both ELNES and DOS in the same graph

to relate ELNES peaks with corresponding DOS. One can

notice that as oxygen vacancy increases from 0% to 25%, all

the partial density of states redistributes in the energy

bandgap. These defect-related donor levels contribute to the

pre-edge peaks in the spectra and the same effect is observed

on both absorption edges.

Although such high concentrations of oxygen vacancies

are unrealistic (i.e., 25%, 12.5%, etc.) in a practical w-ZnO

lattice, our calculations help in understanding the resulting

trend in change in electronic structure and its reflection in

EEL spectra. Indeed, it was quite a successful attempt to a

very good extent as we could see the similar trends in the Zn

L3 experimental and theoretical spectra as the oxygen va-

cancy concentration is increased (see Fig. 9). The four peaks

of ZnO, A, T, E, and TT are observed in both experimental

and theoretical spectra. The development of pre-edge peaks

and smearing of other fine features of A, T, E, and TT peaks

due to VO have similar trends in both experimental and theo-

retical spectra.

A small change in pre-edge structure in our experimen-

tal spectra indicates that a more realistic estimate of the VO

concentrations would be within a percent. Simulations of

these require the use of a very large super cell to solve for

DOS using WIEN 2K and in such cases the FEFF9 code will

be the appropriate choice (easy to calculate for nonperiodic

structures containing defects in real space). As it is still not

clear that the origin of green emission in ZnO is truly from

oxygen vacancy or some other types of defects, it will be

useful to calculate ELNES spectra for various types of point

defects possible in ZnO lattice (e.g., Frenkel, Schottky type,

etc., considering charge neutrality of the ceramic lattice), as

well as different vacancy complexes to reach a firm conclu-

sion on the assignment of a mechanism responsible for green

emission, and by looking not only at the pre-edge feature but

at near edge structure as well.

IV. CONCLUSIONS

In conclusion, we have compared experimental ELNES

pre-edge structures of three different ZnO nanocrystals with

different oxygen defect concentrations. Broadening of the

main peak in experimental absorption edge, both in O K and

Zn L3, was observed with increasing oxygen deficiency.

From theoretical ELNES spectra, we clearly observe the

appearance of extra peaks with increasing density of VO in

ZnO nanocrystals, similar to the feature observed at the Zn

L3 pre-edge structure experimentally. These features are due

to donor levels in the bandgap arising from oxygen defects.

1W. W. Wenas, A. Yamada, K. Takahashi, M. Yoshino, and M. Konagai, J.

Appl. Phys. 70, 7119 (1991).2H. Ohta, K. Kawamura, M. Orita, M. Hirano, N. Sarukura, and H. Hosono,

Appl. Phys. Lett. 77, 475 (2000).3S. A. Studenikin, M. Cocivera, W. Kellner, and H. Pascher, J. Lumin. 91,

223 (2000).4C. Klingshern, Chem. Phys. Chem. 8, 782 (2007).5J. Bian, W. Liu, J. Sun, and H. Liang, J. Mater. Process. Technol. 184, 451

(2007).6D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama, M. Y. Shen, and

T. Goto, Appl. Phys. Lett. 70, 2230 (1997).7T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science 287,

1019 (2000).8P. Sharma, A. Gupta, K. V. Rao, F. J. Owens, R. Sharma, R. Ahuja,

J. M. O. Guillen, B. Johansson, and G. A. Gehring, Nat. Mater. 2, 673 (2003).9D. Kundaliya, S. B. Ogale, S. E. Lofland, S. Dhar, C. J. Metting, S. R.

Shinde, Z. Ma, B. Varughese, K. V. Ramanujachary, L. Salamanca-Riba,

and T. Venkatesan, Nat. Mater. 3, 709 (2004).10C. N. R. Rao and F. L. Deepak, J. Mater. Chem. 15, 573 (2005).11G. Lawes, A. S. Risbud, A. P. Ramirez, and R. Seshadri, Phys. Rev. B 71,

045201 (2005).12J. M. D. Coey and S. A. Chambers, MRS Bull. 33, 1053 (2008).13A. Sundaresan, R. Bhargavi, N. Rangarajan, and C. N. R. Rao, Phys. Rev.

B 74, 16136 (2006).14L. S. Panchakarla, Y. Sundarayya, S. Manjunatha, A. Sundaresan, and

C. N. R. Rao, Chem. Phys. Chem. 11, 1673 (2010).15M. K. Yadav, M. Ghosh, R. Biswas, A. K. Raychaudhuri, and A. Mooker-

jee, Phys. Rev. B 76, 195450 (2007).16K. Minegishi, Y. Koiwai, Y. Kikuchi, K. Yano, M. Kasuga, and A.

Shimizzu, Jpn. J. Appl. Phys., Part 2 36, L1453 (1997).17K. Vanheusden, W. L. Warren, C. H. Seager, D. R. Tallant, J. A. Voigt,

and B. E. Gnade, J. Appl. Phys. 79, 7983 (1996).18W. Li, D. Mao, F. Zhang, X. Wang, X. Liu, S. Zou, Y. Zhu, Q. Li, and

J. Xu, Nucl. Instrum. Methods Phys. Res. B 169, 59 (2000).

FIG. 9. (Color online) A comparison

between (a) experimental and (b) com-

puted spectra for Zn L3 edge.

063523-6 Dileep et al. J. Appl. Phys. 109, 063523 (2011)

19K. Vanheusden, C. H. Seager, W. L. Warren, D. R. Tallant, and J. A.

Voigt, Appl. Phys. Lett. 68, 403 (1996).20S. V. Bhat, S. R. C. Vivekchand, A. Govindraj, and C. N. R. Rao, Solid

State Commun. 149, 510 (2009).21H. S. Hsu, J. C. A. Huang, Y. H. Huang, Y. F. Liao, M. Z. Lin, C. H. Lee,

J. F. Lee, S. F. Chen, L. Y. Lai, and C. P. Liu, Appl. Phy. Lett. 88, 242507

(2006).22D. A. Muller, N. Nakagawa, A. Ohtomo, J. L. Grazul, and H. Y. Hwang,

Nature 430, 657 (2004).

23C. L. Dong, C. Persson, L. Vayssieres, A. Augustsson, T. Schmitt, M. Matte-

sini, R. Ahuja, C. L. Chang, and J.-H. Guo, Phys. Rev. B 70, 195325 (2004).24C. Ton-That, M. R. Phillips, M. Foley, S. J. Moody, and A. P. J. Stampfl,

Appl. Phys. Lett. 92, 261916 (2008).25T. Mizoguchia, M. Yoshiyaa, J. Lia, F. Obaa, I. Tanakab, and H. Adachia,

Ultramicroscopy 86, 363 (2001).26K. Jorissen, Ph.D. thesis, University of Antwerp, Antwerpen, Belgium,

2007.27C. L. Jia, M. Lentzen, and K. Urban, Science 299, 870 (2003).

063523-7 Dileep et al. J. Appl. Phys. 109, 063523 (2011)