Characterization of ZnON films prepared by annealing sputtered zinc oxynitride films
Characterization of ZnO:N?lms prepared by annealing sputtered zinc oxynitride?lms at different temperatures
J.P.Zhang,a?L.D.Zhang,L.Q.Zhu,Y.Zhang,M.Liu,and X.J.Wang
Key Laboratory of Materials Physics,Anhui Key Laboratory of Nanomaterials and Nanostructure,
Institute of Solid State Physics,Chinese Academy of Sciences,Hefei230031,China
G.He
Department of Applied Chemistry,School of Engineering,The University of Tokyo,7-3-1Hongo,
Bunkyo-ku,Tokyo113-8656,Japan
?Received21July2007;accepted25September2007;published online4December2007?
N-doped ZnO?lms were prepared by annealing zinc oxynitride?lms deposited by rf reactive
sputtering.Two Raman peaks were observed at274and580cm?1.According to the variation of the
integral intensity of these two peaks,the nitrogen activation at500°C?the activation temperature
?AT??has been obtained.Below the AT,the integral intensities of them show a different variation
trend.X-ray photoelectron spectroscopy?XPS?indicates the N chemical state variation for them and
?nds the activated Zn-N bond.Further analyses by photoluminescence?PL?spectra and
spectroscopic ellipsometry?SE?have been carried out.The activated sample exhibits a symmetric
emission peak at3.22eV assigned to be the A0X emission at room temperature.SE investigation
takes account of samples within the different temperature span divided by the AT.Different factors,
such as nitrogen dopant?N?O and the nanocrystal growth,which affect the redshift of the absorption
edges,have been discussed.?2007American Institute of Physics.?DOI:10.1063/1.2817255?
I.INTRODUCTION
Zinc oxide?ZnO?has been considered as an excellent
material for ultraviolet light emitting diodes and lasers due to
its band gap?3.37eV?and large exciton binding energy?60
meV?,which favors the high ef?cient room temperature?RT?
excitonic emission.However,synthesizing p-type ZnO re-
mains to be a serious challenge because of the self-
compensation effect induced by the presence of native donor
defects such as oxygen vacancies?V O?and zinc interstitials ?Zn i?.Additionally,low dopant solubility1and the deep ac-ceptor levels of the dopants may yield low carrier concentra-
tion,which makes the realization of p-type ZnO more dif?-
cult.Nitrogen is widely considered to be one of the most
promising p-type dopants due to its similar size to oxygen
and its resistance to forming AX centers.2Nitrogen could be
incorporated either in atomic or molecular form.It is be-
lieved that the atomic one is preferred for achieving p-type
doping.However,the high energy of the N?N bond ??9eV?makes it dif?cult for its dissociation to achieve atomic form incorporation3by introducing N2into the tradi-tional radio-frequency?rf?sputtering ambience.It results in the formation of a N2?molecule at an O site and therefore leads to the compensation rather than p-type doping as?N?O.
To overcome the crucial problem concerning with the
low solubility of the N acceptors in ZnO,some works have
been reported to investigate the thermal oxidation of Zn3N2.
Li et al.4prepared p-type ZnO thin?lms by thermal oxidiza-
tion of Zn3N2?lms and observed the structural transforma-
tion from Zn3N2to ZnO.Nakano et al.1synthesized ZnO:N
?lms in the same way and systematically investigated the in?uence of annealing temperature on the electrical proper-
ties.In this study,we tried preparing Zn3N2?lm by introduc-
ing N2and Ar?ows into the reactive ambience of rf sputter-
ing.Due to the residual oxygen in the chamber and its high
reaction activity with metal Zn,the oxygen is unintentionally
incorporated into the?lm.As a result,the as-grown sample
is estimated to be zinc oxynitride?lm.However,the high N2
concentration in the chamber could reduce the formation en-
ergy and increase the solubility of the nitrogen dopant.5The
ex situ oxidation annealing of zinc oxynitride?lms has been carried out to form polycrystalline ZnO?lm and adjust the
nitrogen chemical states.Based on the investigations of zinc
oxynitride?lms annealed at different temperatures,we found
an activation temperature,which is consistent with other
reports.1,6,7The Raman spectra and x-ray photoelectron spec-
tra have been employed to identify the nitrogen activation
temperature?AT?.The optical properties have been investi-
gated by photoluminescence?PL?spectra and spectroscopic
ellipsometry?SE?.This study could provide valuable optical
information for the development of ZnO doped with nitrogen
in the applications of optoelectronic and other optical de-
vices.
II.EXPERIMENTAL PROCEDURES
The zinc oxynitride?lms were deposited on silicon sub-
strate??100??by rf reactive sputtering.The Si substrate was
cleaned in an ultrasonic bath with a mixed solution ?NH4OH:H2O2:H2O2:1:7,25°C?for15min.Then the Si substrate was etched in a diluted hydro?uoric?HF1%?to
remove native SiO2and then washed using deionized water.
A metallic zinc disk?99.999%?was used as the target.The
sputtering chamber was evacuated to a base pressure below
4?10?4Pa with a turbomolecular pump.Ultrapure
a?Electronic mail:jin.ping.2000@https://www.sodocs.net/doc/042927142.html,.
JOURNAL OF APPLIED PHYSICS102,114903?2007?
0021-8979/2007/102?11?/114903/7/$23.00?2007American Institute of Physics
102,114903-1
?99.999%?Ar and N2gas mixtures were introduced into the chamber through a set of mass?ow controllers with the?ow rates of15and15sccm,respectively.During deposition,the working pressure in chamber was kept at1.0Pa with the substrate temperature kept at room temperature?RT?and the target-substrate distance maintained at40mm.About270 nm thick?lms were produced by a40min deposition with the sputtering power of100W.Due to the oxygen residual in the chamber,the as-grown zinc oxynitride?lms were pro-duced and then annealed in a tube furnace for30min at a temperature ranging from300to800°C in an O2?ow.The O2source is ultrapure?99.999%?oxygen gas with the?ow rate of50sccm.
The structure and surface morphology were studied by x-ray diffraction?XRD?with Cu K?radiation??=0.15406nm?and atomic force microscopy?AFM?,re-spectively.Chemical bonding states and composition of the ?lms were measured by x-ray photoelectron spectroscopy ?XPS?using an Al K?x-ray source.To investigate the local vibration modes affected by nitrogen incorporated into ZnO ?lms,Raman scattering spectra were obtained using the 514.5nm line of an Ar+laser.Photoluminescence?PL?mea-surement was performed at room temperature by the excita-tion from a325nm He-Cd laser to study the in?uence of N doping on the luminescence properties of ZnO?lms.The optical properties were further studied by SE.
III.RESULTS AND DISCUSSION
Figure1?a?presents the???XRD patterns of the?lms annealed at different temperatures.The XRD pattern of the as-grown znc oxynitride?lm is also shown at the bottom in
Fig.1?a?.Only one diffraction peak around34.03°is ob-
served revealing a preferential orientation.Xiao et al.7ob-
served a diffraction peak at34.19°of zinc oxynitride?lms.
The literature assigned this peak to the result of the?002?
peak of ZnO?34.42°?mixing the?222?peak of
Zn3N2?31.70°?.6,7In our case,the residual oxygen in the chamber during reactive sputtering would be advantageous
to the ZnO wurtzite structure growth.The diffraction peak at
36.21°for the samples annealed at700and800°C corre-
sponds to the ZnO wurtzite?101?.The nitrogen dopant as ?N2?O,other than?N?O,will make the lattice constants ex-pand and will lead to a lower angle side shift of the?002?diffraction peak.8Since Zn-N bond length is somewhat smaller than the Zn-O bond length,the?N?O is not respon-sible for the larger lattice constants.9On this viewpoint,it is suggested that the most possible nitrogen state in the wurtzite structure is?N2?O.The strains in the?lms could shift the XRD peak to the lower angle direction as well.As for the sample annealed at300°C,the main peak shows an obvious higher angle side shift to34.30°,close to the bulk ZnO of 34.42°,indicating a transformation to improved wurtzite structure.Although the N escaping happens right at the be-ginning of the annealing temperature at300°C,the sug-gested?N2?O has induced its lower diffraction peak position and a larger lattice parameter d value.8,10When the?lms were annealed at the temperature above300°C,the?002?peak shows a right shift,implying further nitrogen escaping and better crystallization.The grain size d of the?lms was calculated using the Debye-Scherrer formula:
d=
0.9?
B cos?
,
where B is the FWHM in radians,?is the Bragg diffraction angle,and d is the average grain size.As illustrated in Fig. 1?b?,the changes of the estimated grain size and the peak position of the preferentially oriented?002?peak indicate that the crystal quality improves as the temperature in-creases.
Local vibration modes of nitrogen in ZnO:N?lms have been observed via Raman spectra,as shown in Fig.2.The spectrum of the substrate is also measured for comparison. According to the selection rules,both E2and A1?longitudinal optical?LO??modes are expected in Raman spectra when the experiments are taken in a backscattering geometry.6The peak at437cm?1corresponds to the high-frequency E2 mode,which is characteristic of the wurtzite structure.The other two peaks can also be observed at274and580cm?1, which are considered to be from the nitrogen-related local vibration modes.11Du et al.12have argued that the A1?LO?mode at580cm?1was the result of the host-lattice defects rather than the N-related local vibration modes.In Fig.3, there is a borderline?AT?above which the integral intensity of these two peaks truly shows a similar variation trend.6,
7 FIG.1.?Color online??a?The???XRD patterns for the as-grown zinc oxynitride?lm and the samples annealed at300,400,500,600,700,and 800°C,respectively.?b?The main peak position and estimated grain size for the different temperature annealed sample.
What attracts us is that the speci?c variation trend of these two peaks below the AT is different,which could be consis-tent with the conclusion of Du et al.
Taking account of the integral intensity variation in Fig.3for the peaks of interest,upon annealing below the AT,the N escaping has weakened the 274cm ?1Raman peak.Mean-while,the recrystallization of ZnO:N ?lm has not reached the eventual accomplishment through such low temperature annealing.During this process,the crystalline disorder of the sample originating from the structure transformation and re-crystallization would increase the host-lattice defects 6and correspondingly strengthen the 580cm ?1Raman peak.When the annealing proceeds at the AT,the observed in-crease of 274cm ?1peaks would be ascribed to the N-activation process.6,7The experiment of Nakano et al.1has indicated that large amounts of point defects and grain-boundary defects appeared accompanying the N activation,which is responsible for the increase of 580cm ?1at the AT in our case.Annealing above the AT will lead to the N con-tinuative escaping,the annihilation of the point defects,and better crystallization of the ?lms.As a result,these two Ra-man peaks of our interest decrease to the extent that they are unperceivable.
In addition to the structural analyses mentioned above,a further investigation into the characteristics of the chemical bonding states is necessary.The ex situ XPS measurements were carried out.C peak,O-H,and N-H related peaks were
observed in the XPS spectra.No Ar +etching was performed because the ZnO:N ?lms would be easily denitri?ed by bom-bardment of Ar +.13The binding energies have been cali-brated by centering the adhesive carbon 1s peak at 284.6eV .
Figure 4shows the O 1s XPS spectra for the 300,500,and 700°C annealed ZnO:N ?lms named samples A,B,and C,which can represent the sample at different temperature spans:below the AT,at the AT,and above the AT.The O 1s spectra could be consistently ?tted by two Gaussian curves.As illustrated in Fig.4,the high binding energy component P2located at about 532eV is usually attributed to the pres-ence of loosely bounded oxygen on the surface,belonging to the speci?c species,such as ?CO 3,adsorbed O-H,or ad-sorbed O 2.7The lower binding energy component P1around 530eV is ascribed to the O-Zn bond.13,14The integral inten-sity ratio of P1/P2increases with the increase of the anneal-ing temperature,indicating the enhanced resistance to the moisture for the higher-temperature-annealed sample be-cause of the increased packing density caused by annealing.15
The N 1s XPS spectra for the 300,500,and 700°C annealed samples are shown in Figs.5?a ?–5?c ?,respectively.It is suggested that N could exist as substitutional diatomic molecules,such as NO,NC,and N 2on the oxygen site in ZnO wurtzite structure.16For the reactive sputtered
ZnO:N
FIG.2.?Color online ?Room temperature Raman scattering spectra for as-grown zinc oxynitride ?lm and annealed samples at different temperatures.Silicon ?100?substrate spectra for
comparison.
FIG.3.?Color online ?Variation of the integral intensities of Raman peaks at 274and 580cm ?1as a function of annealing
temperature.
FIG.4.?Color online ?O 1s XPS spectra for the 300,500,and 700°C annealed ZnO:N
?lms.
FIG.5.?Color online ?N 1s XPS spectra for the ?a ?300°C,?b ?500°C,and ?c ?700°C annealed samples.
?lms,two peaks related to?N2?O and?N?O have always been observed.8,17Accordingly,we have assigned the peak at403–404eV to the?N2?O.13The peaks around399eV for those three N1s spectra have been assigned to the overlap of the N-H component6and the C-N component.18The H and C incorporation are unintentional and may be due to the H and C residual in the chamber.We think that chemical state of N for the as-grown sample is similar to the300°C annealed sample because the nitrogen activation has not occurred at low temperature according to the Raman spectra,except for the escape of partial nitrogen.In this regard,we suggest that the chemical states of N dopant should mostly be N-C,N-H, and?N2?O in the as-grown and low temperature?below the AT?annealed samples.Xiao et al.have observed that the ?N?O could easily be passivated by N-H bond.7These substi-tutional diatomic molecules may lead to a larger lattice pa-rameter d value and lower energy side shift,16consistent with the illustration by XRD.In Fig.5?b?,the N has been pro-
voked at the AT where two peaks around395.03and398.30 eV appeared.It is suggested that the peaks around395.03 and398.30eV should be due to the polarized triply bounded CN19and newly activated N-Zn bond,7,20respectively.The triply bounded CN may come from the N-C and the speci?c process is unclear.However,the identi?cation of nitrogen activation by XPS is in good agreement with the Raman spectra analysis.The thermal annealing will easily induce N escaping as N2bubbles21and eliminate the passivation effect of the H.7The possible reactions related to the nitrogen are as follows:
2?N2?O+O2→2O O+2N2,?1?4?NH?O+O2→4?N?O+2H2O,?2?2?N?O+O2→2O O+N2.?3?For sample A,only the reaction?1?happens below the AT, while for sample B,both the reactions?1?and?2?happen at the AT.The reaction?2?could make part of N to be activated as a metastable N-Zn bond.21For sample C,the reactions of ?1?–?3?proceed at the temperature above the AT and exhaust
the activated nitrogen.From sample A to sample C,the FWHM of the399eV feature shows a decrease trend:2.47, 1.73,and1.53eV,which may be attributed to the exhausting of the N-H component around399eV as reactions?2?and ?3?.The residual399.4eV feature for sample C would be from the stable?NC?O because of its relatively high stability.16It should be noted that the sample C shows a N1s peak at406.15eV.The most possible origination of this peak is N-O in NO32??nitrate?induced by the high temperature oxidization.22In addition,the redshift of the binding energy of?N2?O may be due to the enhancement of the symmetry of the ZnO?lms by annealing.7
As we have suggested that the structure and nitrogen chemical states in the?lms should be affected by the different-temperature annealing,we go further to study the temperature dependence of the optical properties.In Fig.6, the photoluminescence?PL?spectra of the?lms annealed at 300,400,500,600,700,and800°C have been illustrated. Note that the PL intensities are normalized to be between0and1.Each of the PL pro?les has been multiplied by a factor
shown in Fig.6.The300°C annealed sample shows a
strong ultraviolet emission band at3.30eV relative to the
recombination of free exciton?FX?.4,23A shoulder at the low
energy side is observed,which is assigned to be the overlap
of the free exciton and different bound excitons?donor-
bound excitons D0X,acceptor-bound excitons A0X,donor-
acceptor pair D0A0?for the ZnO:N?lms.6,24From the 300°C to the400°C annealed sample,the obvious uptake
of the lower energy side shoulder is ascribed to the genera-
tion of the point defects.This phenomenon is consistent with
the increase trend of580cm?1Raman peaks in Fig.3.After
the500°C annealing,the FX emission component in the PL
spectra decreases,becoming unperceivable.The near band-
edge emission?NBE?shows a nearly symmetric peak located
at3.22eV with a FWHM of180meV.The position of3.22
eV is close to the reported value3.24eV for the A0X emis-
sion at room temperature,4which con?rms the nitrogen to be
activated to?N?O at500°C.The?N?O serves as a relatively
shallow acceptor in ZnO?lm.24However,its large FWHM
and the disappearance of the FX emission have indicated the
increase of the other point defects?except?N?O?accompany-ing the nitrogen activation.The nonradiative recombination induced by these point defects may have suppressed the FX emission,resulting in the relatively weak intensity of the NBE.?The bigger factor multiplied in Fig.6means weaker original PL intensity.25?By the way,the fairly sharp peaks above3.4eV of the spectra annealed at500and600°C originate from the resonance Raman signal.Above the AT, the annealing has induced the enhancement of FX emission and the decline of the lower energy side shoulder because of the decrease of the zinc interstitials and oxygen vacancies.26 Higher temperature bene?ts the better crystallization of ZnO:N?lms and,moreover,induces the point defects anni-hilation as the follows:
2?N?O+O2→2O O+N2,?3?2Zn i+O2→2ZnO,?4
?FIG.6.?Color online?Room temperature PL spectra of?lms annealed at 300,400,500,600,700,and800°C.The intensities of them have been normalized to be between0and1.
2V O +O 2→2O O .
?5?
For the 800°C annealed sample,a FX peak located at 3.28eV can be observed with the FWHM of 78meV .The most enhanced FX emission con?rms the best ZnO crystallization,as discussed by XRD pro?les in Fig.1?a ?.By comparing the FX peak position for the 300°C annealed sample at 3.30eV with that for the 800°C annealed sample at 3.28eV in Fig.7?except unperceivable FX emission at 500°C ?,we can ?nd that the FX peak undergoes a slight redshift as temperature increases from 300to 800°C,which is assigned to the in-crease of ZnO nanocrystal size with the increase of temperature.6On all accounts,the variation of the point de-fect in the annealed ?lms could be demonstrated both by the PL spectra and the Raman spectra by which the largest amounts of point defects have been shown at 500°C.
Futsuhara et al.18have observed a redshift of optical band gap E g for the N-doped ZnO,which they assigned to the decrease in ionicity caused by the formation of Zn-N bonds.It is also reported that the N doping always reduces the optical band gap because of mixing of N 2p states and O 2p states.27,28Apparently,for a thermally annealed ZnO:N ?lm,the absorption edge has been in?uenced both by the N dopant and the thermal process.In order to shed light on the different roles of the N dopant and the thermal process on the optical properties,an ex situ phase modulated spectroscopic ellipsometry ?UVISEL Jobin-Yvon ?has been employed to analyze the optical properties in the visible ultraviolet ?UV ?spectral region of 1.5–6.5eV with the angle of incidence 70°.The main bene?ts of the SE are to investigate the optical response of materials without any destruction and,particu-
larly,to extract the thickness values and the optical constants of multilayer system concurrently.In this work,a Tauc-Lorentz ?TL ?mode is used to ?t the experimental spectro-scopic data,as is shown below:
?2?E ?=?AE 0C ?E ?E g ?2?E 2?E 0
2?2+C 2E 21
E ?E ?E g ?,0?E ?E g ?,?6?
?1?E ?=??+
2P
?
?
??2???
?2?E 2
d ?.
?7?
Equations ?6?and ?7?as functions of the photon energy E are de?ned by ?ve parameters:A ?transition matrix element,re-lated to the ?lm density ?,E 0?peak transition energy ?,C ?broadening term ?,and E g ?optical band gap ?,which all have the energy unit,and ???high frequency dielectric constant ?,where P denotes the Cauthy principle part of the integral.These TL parameters,together with the layer thickness in the least-squares ?t procedure,minimize the biased estimator ?2de?ned as goodness-of-?t.28
A four-layer model is used to extract the optical con-stants of the ?lms as shown in Fig.7?inset ?.The substrate layer is Si ?100?,whose optical constants are taken from the literature and are not allowed to vary during the ?t.The interfacial layer is an effective medium approximation layer that may be composed of SiO 2,ZnO,or nonstoichiometric zinc silicate.29The ZnO:N ?lm layer is a homogeneous layer,and the rough layer is assumed to consist of voids and ZnO:N.An atomic force microscopy ?AFM ?measurement also has investigated the root-mean-square ?rms ?value to evaluate the accuracy of the best ?tting results.For each sample,the ?ve TL parameters and the ?lm thickness are determined by least-square ?tting their experimental data.Table I lists the TL parameters along with the best ?tting residual ?2.As shown in Fig.7,the pro?les correspond to the experimental ?circles ?and ?tted ?solid lines ?spectra for a representative 300°C annealed ZnO:N ?lm ?sample A ?.From these ?gures,it can be observed that an excellent agreement between the experimental and ?tted spectra for the ZnO:N ?lms has been attained,indicating that the optical constants of the ?lms can be exactly determined by the best ?tting results.The spectra of the 500°C annealed sample ?sample
B ?and 700°
C annealed sample ?sample C ?are similar pro?les ?not shown here ?.It has been reported that the rms values from AFM measurement are half the thick-ness values of the surface-roughness layer,29which is con-?rmed in Table II .By comparing the rms with the thickness of the surface-roughness layer in Table II ,the accuracy and rationality of the TL ?tting could be estimated.The
maxi-
FIG.7.?Color online ?The best ?tting spectra between the experimental ?circles ?and ?tted ?solid lines ?spectroscopic ellipsometry data for 300°C annealed ZnO:N.The inset shows the schematic structure of four-layer stack model.
TABLE I.Parameters of the ?tting results obtained from all the samples using TL model dispersions.t :the thickness of homogenous ZnO:N layer.Sample t ?nm ?E g ?eV ???A E g ?eV ?C ?2
A 255.93±3.81 3.173±0.005 2.97±0.06187.62±10.46 3.20±0.010.634±0.001 4.20
B 292.28±4.90 3.152±0.005 2.86±0.07241.67±12.85 3.17±0.010.403±0.003 5.72C
268.90±3.67
3.165±0.003
2.96±0.06
283.51±11.90
3.14±0.01
0.391±0.003
4.97
mum discrepancy of comparison for sample B?rms22nm vs
twice the thickness of14nm?is related to the maximum
biased estimator?2value of5.72.
The parameter A in the TL model is related to the?lm
density.30The consistency between the large A value and the
high packing density has been observed in previous
works.28,31,32The parameter A values of187,241,and283,
as shown in Table I,demonstrate a rising trend,meaning that
the packing density has been enhanced by the annealing.
However,the refractive index n?including different n?and
n??for ZnO?lm with anisotropic wurtzite structure could not obtained through a single incidence of angle SE measure-
ment in our case.33,34Except for the anisotropy,the refractive
index n strongly depends on impurities,imperfection incor-
porated during growth process,polar bonds in the?lms,and
packing density.28,31,33For example,both evolution of the
C-axis preferential orientation and the N dopant will contrib-
ute to the variations of the refractive index n.35
Compared with distinguishing the refractive index n?and
n?,it is easier to estimate the optical band edge independent
of anisotropy using the extinction coef?cient spectra.Figure
8?a?shows the extinction coef?cient k calculated from the
extracted best-?tted parameters as listed in Table I.The k
spectra have been used to derive the absorption coef?cient
by?=4?k/?.The photon energy dependence of the absorp-tion coef?cient near the UV edge is employed to determine
the band structure.The optical energy gap?E g?was esti-mated by assuming a direct transition between valence and
conduction bands from the expression?h?=A?h??E g?1/2. Here,h v denotes photon energy and A denotes a constant. We determined E g by extrapolating the linear portion to ?h?=0.In this case,E g values for samples A–C are deter-mined in Fig.8?b?to be3.33,3.25,and3.28eV,respectively, close to the reported optical band gap of ZnO?lms in the literature.18,36The extracted parameters E g of3.173,3.152, and3.165eV in Table I are less than the values obtained in Fig.8,although they show a similar variation trend.This is because the parameter E g is determined by the crossing point of?h?=A?h??E g?1/2and?h?=0,which is related to the Urbach tail near the band edge.36As for the direct band gap transition,the FX emission could be estimated by h?=E g ?R*,where E g is the optical band gap and R*is the exciton binding energy.Accordingly,the band gap variation could also be compared with the position of the FX emission.37The E g extracted from the absorption edge is in good agreement with the photon energy of the FX emission in the PL spectra, considering the experiment error.A redshift of the FX emis-sion from sample A to sample C has been observed,con?rm-ing the redshift of the band edge for them in https://www.sodocs.net/doc/042927142.html,-pared to sample A,the redshifts of samples B and C are dominated by different factors.Sample B has the activated
nitrogen incorporation and bigger grain size compared to
sample A.The?N?O in sample B,other than the dominating
nitrogen chemical state?N2?O in sample A,could provoke the mixing of N2p states and O2p states.27,28Additionally,the
increase in grain size could induce the redshift of the band
edge as the quantum con?nement effect.6,31As for sample C,
the nitrogen content is the minimum in these three samples,
and no?N?O has been observed by XPS.The redshift of its
band edge is mainly ascribed to the increase of ZnO nano-
crystal size upon higher-temperature annealing.Taken to-
gether,the redshift of sample B is attributed to N-metal bond
formation and the size increase of the nanocrystal,but that of
sample C is mainly related to the latter.This interpretation
can explain the different extent of redshift in samples B and
C in Fig.8?b?,as compared to sample A,that is,the redshift
for sample B determined by two factors is more perceivable. IV.CONCLUSION
In conclusion,the N-doped ZnO?lms were prepared by
annealing sputtered zinc oxynitride?lm.According to the
XRD and Raman pro?les,the ZnO:N?lms were determined
to be polycrystalline with ZnO wurtzite structure,which has
been evolved by the annealing temperature.Based on the
variation of the integral intensity of two Raman peaks at274
and580cm?1,the nitrogen activation temperature has been
TABLE II.The?tted thicknesses of the rough layer and the rms values
extracted from AFM measurements.
Sample Rough layer?nm?rms?nm?
A22.36±0.6311
B22.19±1.0814
C18.22±3.210
FIG.8.?Color online??a?The extinction coef?cient k of ZnO:N thin?lms
for samples A?300°C?,B?500°C?,and C?700°C?.?b?Plots of??h??2
vs.h?for samples A–C.
deduced to be500°C.Below the AT,the integral intensities
of them show different variation trends,which supports that
the Raman peak at580cm?1originates from the host-lattice
defects generated by the N dopant rather than the N-related
local vibration modes.The XPS spectra found diatomic mol-
ecule states for the doped N and unintentionally doped C as ?N2?O,?NH?O,and?NC?O,and further con?rmed N activa-tion at the AT,where the N-Zn bonds have been detected in
N1s spectra.In the PL spectra,the sample at the AT exhibits
a symmetric room temperature NBE peak at3.22eV,which
is assumed to be the?N?O A0X emission,and its low intensity,
which is due to the point defects caused by the activation.SE
investigation has been employed to analyze the optical prop-
erties of samples within the temperature span below,at,and
above the AT.The TL formula has been employed to?t the
experimental data.The redshifts of the optical band edge
have been attributed to different factors,such as nitrogen
dopant and the grain growth.Eventually,the high tempera-
ture oxidative annealing treatment could be used to adjust the
N chemical states and optical properties. ACKNOWLEDGMENTS
This work was supported by the National Natural Sci-
ence Foundation of China?Grant No.10674138?and the
Natural Science Foundation of Anhui Province?Grant No.
070414196?.
1Y.Nakano,T.Morikawa,T.Ohwaki,and Y.Taga,Appl.Phys.Lett.88, 172103?2006?.
2C.H.Park,S.B.Zhang,and S.-H.Wei,Phys.Rev.B66,073202?2002?. 3I.V olintiru,M.Creatore,W.H.van Helvoort,J.L.Linden,and M.C.M. van de Sanden,Appl.Phys.Lett.89,022110?2006?.
4B.S.Li et al.,J.Mater.Res.18,8?2003?.
5Y.Yan,S.B.Zhang,and S.T.Pantelides,Phys.Rev.Lett.86,5723?2001?.
6J.G.Ma,Y.C.Liu,R.Mu,J.Y.Zhang,Y.M.Lu,D.Z.Shen,and X.W. Fan,J.Vac.Sci.Technol.B22,94?2004?.
7Z.Y.Xiao,Y.C.Liu,J.Y.Zhang,D.X.Zhao,Y.M.Lu,D.Z.Shen,and X.W.Fan,Semicond.Sci.Technol.20,796?2005?.
8B.Yao,Z.Z.Zhang,X.H.Wang,Z.P.Wei,B.H.Li,Y.M.Lv,X.W. Fan,L.X.Guan,G.Z.Xing,C.X.Cong,and Y.P.Xie,J.Appl.Phys.99, 123510?2006?.
9T.M.Barnes,K.Olson,and C.A.Wolden,Appl.Phys.Lett.86,112112?2005?.
10N.Y.Yuan,J.H.Li,L.N.Fan,X.Q.Wang,and Y.Zhou,J.Cryst.Growth 290,156?2006?.
11A.Kaschner,U.Haboeck,M.Strassburg,M.Strassburg,G.Kaczmarczyk,
A.Hoffmann,C.Thomsen,A.Zeuner,H.R.Alves,D.M.Hofmann,and
B.K.Meyer,Appl.Phys.Lett.80,1909?2002?.12G.T.Du,Y.Ma,Y.T.Zhang,and T.P.Yang,Appl.Phys.Lett.87,213103?2005?.
13M.Futsuhara,K.Yoshioka,and O.Takai,Thin Solid Films322,274?1998?.
14Y.Nakano,T.Morikawa,T.Ohwaki,and Y.Taga,Appl.Phys.Lett.87, 232104?2005?.
15D.Niu,R.W.Ashcraft,and G.N.Parsons,Appl.Phys.Lett.80,3575?2002?.
16S.Limpijumnong,X.Li,S.-H.Wei,and S.B.Zhang,Appl.Phys.Lett.86, 211910?2005?.
17C.L.Perkins,S.-H.Lee,X.N.Li,S.E.Asher,and T.J.Coutts,J.Appl. Phys.97,034907?2005?.
18M.Futsuhara,K.Yoshioka,and O.Takai,Thin Solid Films317,322?1998?.
19J.Y.Feng,C.P.Long,Y.Zheng,F.W.Zhang,and Y.D.Fan,J.Cryst. Growth147,333?1995?.
20S.J.Jiao,Z.Z.Zhang,Y.M.Lu,D.Z.Shen,B.Yap,J.Y.Zhang,B.H. Li,D.X.Zhao,X.W.Fan,and Z.K.Tang,Appl.Phys.Lett.88,031911?2006?.
21P.Fons,H.Tampo,A.V.Kolobov,M.Ohkubo,S.Niki,J.Tominaga,R. Carboni,F.Boscherini,and S.Friedrich,Phys.Rev.Lett.96,045504?2006?.
22J.Torres,C.C.Perry,S.J.Brans?eld,and D.H.Fairbrother,J.Phys. Chem.B107,5558?2003?.
23M.Joseph,H.Tabata,H.Saeki,K.Ueda,and T.Kawai,Physica B302-303,140?2001?.
24W.Z.Xu,Z.Z.Ye,Y.J.Zeng,L.P.Zhu,B.H.Zhao,L.Jiang,J.G.Lu, H.P.He,and S.B.Zhang,Appl.Phys.Lett.88,173506?2006?.
25C.-C.Yang,C.-C.Lin,C.-H.Peng,and S.-Y.Chen,J.Cryst.Growth285, 96?2005?.
26F.K.Shan,G.X.Liu,W.J.Lee,G.H.Lee,I.S.Kim,and B.C.Shin, Appl.Phys.Lett.86,221910?2005?.
27R.Asahi,T.Morikawa,T.Ohwaki,K.Aoki,and Y.Taga,Science293, 269?2001?.
28G.He,L.D.Zhang,G.H.Li,M.Liu,L.Q.Zhu,S.S.Pan,and Q.Fang, Appl.Phys.Lett.86,232901?2005?.
29Z.Song,B.R.Rogers,and N.D.Theodore,J.Vac.Sci.Technol.A22, 711?2004?.
30A.Fontcuberta i Morral,P.Roca i Cabarrocas,and C.Clerc,Phys.Rev.B 69,125307?2004?.
31H.T.Fan,X.M.Teng,S.S.Pan,C.Ye,G.H.Li,and L.D.Zhang,Appl. Phys.Lett.87,231916?2005?.
32M.Liu,Q.Fang,G.He,L.Q.Zhu,and L.D.Zhang,J.Appl.Phys.101, 034107?2007?.
33Y.S.Park and J.R.Schneider,J.Appl.Phys.39,3049?1968?.
34R.Schmidt,B.Rheinl?nder,M.Schubert,D.Spemann,T.Butz,J.Len-zner,E.M.Kaidashev,M.Lorentz,A.Rahm,H.C.Semmelhack,and M. Grundmann,Appl.Phys.Lett.82,2260?2003?.
35S.Venkataraj,O.Kappertz,R.Jayavel,and M.Wuttig,J.Appl.Phys.92, 2461?2002?.
36N.R.Aghamalyan,I.A.Gambaryan,E.Kh.Goulanian,R.K.Hovsepyan, R.B.Kostanyan,S.I.Petrosyan,E.S.Vardanyan,and A.F.Zerrouk, Semicond.Sci.Technol.18,525?2003?.
37S.Heitsch,G.Zimmermann,D.Fritsch,C.Sturm,R.Schmidt-Grund,C. Schulz,H.Hochmuth,D.Spemann,G.Benndorf,B.Rheinl?nder,Th. Nobis,M.Lorenz,and M.Grundmann,J.Appl.Phys.101,083521?2007?.