Home Search Collections Journals About Contact us My IOPscience
Deep level transient spectroscopic study of oxygen implanted melt grown ZnO single crystal
This article has been downloaded from IOPscience. Plea scroll down to e the full text article.
2011 Semicond. Sci. Technol. 26 095016
(iopscience.iop/0268-1242/26/9/095016)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 222.29.31.176
The article was downloaded on 17/08/2011 at 14:39
Plea note that terms and conditions apply.
IOP P UBLISHING S EMICONDUCTOR S CIENCE AND T ECHNOLOGY Semicond.Sci.Technol.26(2011)095016(6pp)doi:10.1088/0268-1242/26/9/095016
Deep level transient spectroscopic study of oxygen implanted melt grown ZnO single crystal
Z R Ye1,X H Lu1,G W Ding1,S Fung1,C C Ling1,3,G Brauer2and
W Anwand2
1Department of Physics,The University of Hong Kong,Pokfulam Road,Hong Kong,
People’s Republic of China
2Institut f¨u r Strahlenphysik,Helmholtz-Zentrum Dresden-Rosndorf,Postfach510119,
D-01314Dresden,Germany垃圾分类演讲稿
E-mail:ccling@hku.hk
Received29April2011,infinal form17June2011
Published28July2011
Online at stacks.iop/SST/26/095016
种植园经济
Abstract
Deep level traps in melt grown ZnO single crystal created by oxygen implantation and
subquent annealing in air were studied by deep level transient spectroscopy measurement
between80and300K.The E C–0.29eV trap(E3)was the dominant peak in the as-grown
sample and no new defects were created in the as-O-implanted sample.The single peak
feature of the deep level transient spectroscopy(DLTS)spectra did not change with the
annealing temperature up to750◦C,but the activation energy decread to0.22eV.This was
explained in terms of a thermally induced defect having a peak clo to but inparable from
the original0.29eV peak.A systematic study on a wide range of the rate window for the
DLTS measurement successfully parated the Arrhenius plot data originated from different
traps.It was inferred that the E3concentration in the samples did not change after the
O-implantation.The traps at E C–0.11,E C–0.16and E C–0.58eV were created after annealing.十二生肖歇后语大全
The E C–0.16eV trap was assigned to an intrinsic defect.No DLTS signal was found after the
sample was annealed to1200◦C.
(Somefigures in this article are in colour only in the electronic version)
1.Introduction
ZnO is a miconductor,having a wide direct band gap(∼3.3 eV at room temperature),which has recently attracted a great deal of attention becau of its applications in short wavelength optoelectronic devices,spintronic devices,high power and high temperature devices,transparent electronic and further applications[1–4].
Reliable techniques for carrying out n-and p-type doping are esntial for fabricating miconductor devices.Despite the asymmetric difficulty in the conductivity doping of ZnO, p-type ZnO was obtained through Group-V element doping (N,As,P and Sb)and Group III–V element co-doping[5].
Ion implantation is a technology frequently ud in material and device processing.Selective area do
ping is one of 3Author to whom any correspondence should be addresd.the advantages of the ion-implantation technology.There have been reports on p-type ZnO layer fabrication achieved by ion implantation[6–16].However,some workers reported highly resistive n-type material after ion-implantation and subquent annealing(for example,[12–15]).The ion implantation process would inevitably create undesirable defects and some of the defects would persist even upon a high temperature annealing process.The would include deep traps which would compensate the material.
Other than introducing electrical doping into the ZnO material,‘magnetic doping’of ZnO could also be achieved by transition metal implantation.Room temperature ferromagnetism was reported in ZnO after ion implantation of Co,Ni,C and Fe[16–21],though the mechanism was not yet well understood.Hong et al[22]and Liu et al[23]have pointed out the influence of intrinsic defects on the magnetic property of transition metal doped ZnO.
Table1.Tabulated carrier concentration n of the O-implanted samples annealed at different temperatures as obtained by the C–V method. The estimated energy level with respect to the conduction band E C,the capture cross ctionσand the concentration N T of the deep level traps identified are also given.
Oxygen-implanted ZnO samples
As-O-implanted350◦C750◦C900◦C1200◦C
n=2×1017cm−3n=3×1017
cm−3
n=1×1017cm−3n=4×1017cm−3n=1×1018cm−3
E C–0.29eV E C–0.29eV E C–0.29eV E C–0.30eV No peak
(σ∼10−16cm2,N T= 1015cm−3)(σ∼10−16cm2,
N T=1015cm−3)
(σ∼10−15cm2,N T∼1015
cm−3)
(σ∼10−15cm2,N T∼1015
cm−3)
E C–0.16eV(σ∼10−18cm2,
N T∼1016cm−3)
百分折E C–0.16eV(σ∼10−17cm2,
N T∼1016cm−3)
E C–0.11eV(σ∼10−18cm2,
N T∼1015cm−3)
E C–0.58eV(σ∼10−17cm2,
N T∼1016cm−3)
All the aspects indicate the importance of a thorough understanding of defects in ZnO for the development of the corresponding ion-implantation technology.However, the defects in ZnO are still not well understood and many controversies remain.
In this study,deep level transient spectroscopy(DLTS) was ud to study the deep traps induced in the melt grown ZnO single crystal induced by O-implantation.Thermal evolution of the deep traps upon annealing in air was also studied.The rate window( t)−1dependence for the DLTS spectra was studied systemically over a wide range from t=0.086to 430ms.
2.Experimental details
The ZnO starting material was a single side polished melt grown n-type,nominally undoped ZnO single crystal(10×10×0.5mm3)obtained from Cermet Inc.with an electron concentration of8×1016cm−3and an electron mobility of 217cm2V−1s−1.
The samples from this material were implanted by oxygen ions on the polished side.The energy of implantation was 150keV and thefluence was1014cm−2.The samples were kept at temperature of300◦C during implantation.TRIM [24]calculation showed that this resulted in an O-implantation profile with the peak at∼280nm.
The isochronal annealing step was carried out in a tube furnace in air for a period of30min.After annealing,contacts for DLTS measurement were fabricated.The large area ohmic contact was fabricated by evaporating a50nm Alfilm onto the non-polished side of the sample.The Schottky conta
ct was fabricated onto the ion-implanted side of the sample by evaporating Au contacts with a thickness of50nm after pre-treatment by hydrogen peroxide(details are given in[25,26]).
DLTS measurements were carried out on the samples across the Schottky and the ohmic contacts.The contact quality was verified by I–V and C–V measurements carried out on each of the diodes using the HP4145A miconductor parameter analyzer and the HP4275A multi-frequency LCR meter,respectively.The DLTS measurements were performed using the Sula DLTS system with temperature varying from 80to300K.
The electron emission rate e n from the trap is given by
e n=σn v th N C exp[−(E C−E a)/kT](1) whereσn and E a are the capture cross ction and the activation
energy of the deep trap,respectively.N C and v th are the state
density of the conduction band and the velocity of free electron, respectively.As v th∼T1/2andσn∼T3/2holds,the values of E a andσn were obtained from the linearfitting of the Arrhenius
plot ln(e n/T2)as a function of T−1[27].The concentration of
the trap N was related to C/C0by
N T=2N d
C
C0
一年级阅读理解
(2) where N d is the donor concentration, C is the capacitance transient produced with the rate window( t)−1and the applied rever bias of V R during the emission period,and C0is the equilibrium junction capacitance with the rever bias of V R.
3.Results and discussions
DLTS spectra from unimplanted samples were obtained
as control.For the as-grown sample,the DLTS spectra
had a dominant peak having the activation energy,capture
cross ction and trap density of E a=0.31eV,σn∼10−16cm2and N T∼1015cm−3,respectively.This trap has parameters agreeing with the commonly found deep trap E3 in ZnO materials[10,15,28–35].Another peak with E a= 0.10eV,σn∼10−17cm2and N T∼1013cm−3was also identified with a much weaker intensity(20times less than E3). This trap has the activation energy and the capture cross ction clo to the previously obrved E2trap[28,30–32].The E3 intensity dropped with annealing temperature,but persisted at 1200◦C annealing.The E a=0.10eV level anneals out at 900◦C.
I–V and C–V measurements were carried out on every sample annealed at different temperatures in order to check the quality of the rectifying contacts.The carrier concentrations for each of the samples were also calculated and are tabulated in table1.The carrier concentration did not undergo a significant change(n∼1017cm−3)after the O-implantation, and post-implantation annealing up to900◦C.However,the
1200◦C post-implantation annealing incread the carrier
concentration to∼1018cm−3.
DLTS spectra of the O-implanted samples were obtained
between80and300K in as-O-implanted state,and after
different post-implantation annealing(350◦C,650◦C,750◦C,
1200◦C).The rever bias,filling pul voltage and period
werefixed at V R=−1V,V p=0V and t p=1ms, respectively,for all the DLTS measurements,while the rate
window( t)−1varied over a wide range from t=0.086to
430ms.The application of V R=−1V to the sample with n=1017cm−he O-implanted samples with annealing temperature 900◦C)implied a depletion width of∼120nm, which was within the oxygen implanted region as determined by the TRIM simulation and the SIMS measurement.For the sample annealed at1200◦C,the corresponding depletion width dropped to38nm as the carrier concentration incread to∼1018cm−3.
For the samples annealed at temperatures 750◦C,the corresponding DLTS spectra contained a single major carrier peak,implying that the corresponding deep trap was an electron trap.One such typical spectrum ries with different t s is illustrated by the spectra obtained from the sample annea
led at750◦C and is shown infigure1(a).The peaking temperatures and the heights of the DLTS peaks incread with decreasing t.The DLTS spectra of the900◦C annealed sample taken with different t s are shown infigure1(b),from which three peaks,labeled as peak1,2and3,were obrved. Peak1had a much weaker intensity than the other two.For the1200◦C annealed sample,no peak was obrved in all the DLTS spectra taken with different rate windows.
The corresponding Arrhenius plots of the DLTS spectra obtained from the as-O-implanted,750◦C and900◦C annealed samples are shown infigure2(a).The activation energy E a of each trap from different samples was obtained by carrying out the linearfitting to the Arrhenius plots according to equation(1).The results are shown infigure3.
Fromfigure3,a single trap with E a∼0.29eV was identified in the as-O-implanted sample.It was found that its E a decreas with increasing annealing temperature and drops to a value of0.22eV at750◦C annealing temperature. After further increa of the annealing temperature to900◦C, three peaks with E a s of0.11,0.16and0.37eV,respectively, were obrved.
It is also noticed that the Arrhenius plot data of the750◦C and the peak3of the900◦C data significantly deviate from the straight lines obtained from the linearfitting process. However,for the as-O-implante
d sample,thefitting straight linefitted excellently well with the experimental data.This is clearly illustrated infigure2(b),which shows the Arrhenius plots of the relevant peaks in a larger scale.
It is worth investigating the drop of the activation energy for the0.29eV trap,as identified in the as-O-implanted sample, when the annealing temperature increas.One possible explanation was that a new trap having a peaking temperature clo to that of the0.29eV trap was formed by thermal annealing and the resultant obrved peak is the unresolved combination of the two signals.
The DLTS measurement was performed by monitoring the DLTS C/C signal)at different temperatures
while Figure1.DLTS spectra of(a)750◦C and(b)900◦C annealed
O-implanted melt grown ZnO samples.The spectra were taken with V R=−1V,V p=0V and t p=1ms.The rate window( t)−1of the measurements varied systemically over a wide range from t= 0.086ms up to430ms.
temporarilyfixing the rate window.For afixed rate window, each of the traps would have its carrier emission dominant at different temperatures,and thus is revealed by a peak in the DLTS spectrum.The peaking temperature is dependent on the trap’s emission rate e n,activation energy E
中孝介花海a,capture cross ctionσn and the rate window adopted.The Arrhenius plot data for a trap can thus be obtained by monitoring the peaking temperatures with the rate window being varied.Two traps(with different E a s)having their peaking temperatures too clo to be parated would have their emissions dominant at different temperatures.For example,for the Arrhenius plots of the750◦C annealed samples(figure2(b)),the data at the lower temperature 1000/T>5.2)corresponded to the emission dominantly contributed from the trap with the lower E a,and tho at the higher temperature region (1000/T<5.2)reprented tho from the trap having a higher E a.
Infigure2(b),the as-O-implanted sample data were well fitted by a straight line with E a=0.29eV.For the data of the
Figure2.(a)Arrhenius plots of the all the peaks found in the
as-O-implanted,750◦C annealed and900◦C annealed ZnO samples. The straight lines are obtained from thefitting so a single straight line and the corresponding E a are shown in thefigure;(b)magnified Arrhenius plots in the high temperature range so that the single peaks data of the as-O-implanted and750◦C annealed sample,as well as tho of peak3of the900◦C annealed sample,could be shown in a larger scale.Single straight linefitting to the data was found in the as-O-implanted sample(e dotted line).However,for the750◦C and900◦C sample annealed sample data,significant deviation from single linefitting was obrved(e dotted lines). The solid lines are thefitted lines assuming two
temperature regions in thefitting process with details described in the text.
750◦C annealed sample,the data could only be wellfitted by two straight lines with the breaking point at1000/T∼5.2and the corresponding E a s were0.16and0.29eV.A similar two gment linefitting was also carried out on the Arrhenius data of peak3for the900◦C annealed sample with the breaking point at1000/T∼5.8,which yielded the E a values of0.30and 0.58eV,respectively.The twofitted lines also well described the data(figure2(b)).
The prent obrvation thus suggests that only one trap with E a=0.29eV exists in the as-O-implanted and the350◦C annealed samples.However,if the annealing temperature is incread to750◦C,the obrved0.22eV peak is indeed the combined signal from0.29and0.16eV traps,respectively. And if the sample was annealed at900◦C,two unambiguously parated peaks1and2,having an E a of0.11and0.16eV
碎刘海发型图片女
,Figure3.Activation energy E a as a function of the annealing temperature T of the identified single peak in the O-implanted and the fourfolded N-implanted samples as the single component peak. The drop of E a with increasing annealing temperature 750◦C was the effect of merging of an annealing-induced peak(0.16eV)with the original peak with E a∼–0.29eV.New peaks were induced in the 900◦C annealed sample spectra.The0.37eV peak consisted of0.29 and0.58eV,and the other two peaks had a single component.
respectively,were identified.In addition,the obrved0.37eV peak consisted of0.30and0.58eV traps,respectively.The findings of the DLTS study in O-implanted ZnO samples are also summarized in table1.
A similar DLTS study was also performed on the Schottky contact formed on an n-type N-implanted ZnO sample. The nitrogen implantation process involved a fourfolded implantation with energies of80,180,310and500keV, which produced a1μm depth box-shaped region with a nitrogen concentration of∼6×1018cm−3.A similar trap with E a=0.31eV was also identified in the as-nitrogen-implanted sample.The annealing behavior of its E a is included in figure3for comparison to demonstrate that a similar drop of E a from0.31to0.25eV in the750◦C annealed sample was also obrved after N-implantation.Moreover,it is worthy to point out that as compared to the TRIM simula
五年级写景作文
ted O-implanted depth profile peaking at∼280nm,the120nm depletion width corresponding to the rever bias V R=−1V during the emission period only barely overlapped the O-ion implantation.It was thus plausible to suggest that the induced defects were formed primarily by the kinetic of vacancy and knock-on in this depletion rather than with the involvement of the implanted oxygen.
The∼0.29eV trap(σ∼10−16cm2)found in the as-grown sample and the O-implanted samples had E a andσn in good agreement with tho values of the previously reported deep trap E3[28,30,32].E3is the deep trap commonly found in ZnO materials irrespective of the growth method. In this study,E3was the major deep trap found in the as-received melt grown sample having N T∼1015cm−3.Its density remained unchanged with the O-implantation and the subquent thermal annealing up to900◦C.The E3deep trap
