4.4.1 Photoluminescence spectra of as-grown films
The PL measurements of the ZnO epilayers show that the emission properties vary from sample to sample, even though they were grown under nominally similar conditions. However, the main features of the PL spectra at room temperature are similar for all samples, and can be divided into three categories: the near band edge (NBE) emission, the low energy tail extending from the near band edge emission, and the deep- level emission at room temperature as shown in Fig. 4-10. The typ ical thickness of ZnO films are 1 µ m. The room temperature PL shows only the near band-edge emission with a peak intensity located at 3.3 eV for the all samples grown at various temperatures (RT ~ 800℃). The highest peak intensity of band-edge emission was obtained for the sample grown at 600℃ that is consistent with our earlier X-ray data. The X-ray diffraction pattern shows the FWHM of 0.08o for (0 0 0 2) orientation measured by DCXRD. The FWHM of the band-edge emission for this ZnO film was estimated to be 84 meV (RT), which is relatively small compare to other reports. 21) The near band-edge emission intensity of all samples shown in Fig.
4-10 is not very strong, possibly due to the dislocations or defects caused by the lattice mismatch between ZnO and c-face sapphire. These dislocations and defects could act as nonradiative centers, which reduce the intensity from the band-edge emission. We can anneal the sample of ZnO to reduce these dislocations and nonradative centers. Fig. 4-11 displays the low-temperature (10K) PL spectra of the ZnO film grown at 600℃. The band-edge emission including the emissions from the free exciton recombination located at 3.379 eV and the dominant peak (3.362 eV), with a FWHM of 5 meV, which is attributable to a transition from an exciton bound to
a neutral shallow donor (I7 line). The low energy extending tail of the band edge is attributable to emission from the longitudinal phonon replicas (LO) of free exciton (EX) and donor bound exciton (DoX), as shown in the inset of Fig. 4-11. In addition, there is an additional weak broad peaks centered at 2.97 eV, which can be determined as the natural donor-acceptor pairs recombination from the of laser power-dependent PL measurement. The spectrum also included a weak broad band with a peak at 2.23 eV, known as a deep-level emission, which is related to the oxygen vacancy. 22) We consider that the generation of oxygen vacancy in ZnO is due to the oxygen nonstoichiometry during the ablation and it should be more serious in our sample due to the lack of oxygen supply. However, our data shows that the intensity of the deep-level emission is more than two orders of magnitude less intense than the exciton emission. It is suggested that the oxygen deficiency in our sample is minimized, even in high vacuum. In other words, the ratio of ablated Zn and O element in the laser- induced plume and onto the substrate may be nearly ideal in high vacuum. To further improve the performance of our ZnO films, in-situ annealing (vacuum) and post-furnace annealing (oxygen atmosphere) are suggested.
4.4.2 Photoluminescence result of the annealed ZnO films
By DCXRD, the FWHM of non-annealing ZnO film under a condition of laser energy density ~7 J/cm2 at 2Hz rep. rate and growth temperature 600℃ is 299 arcsec and after annealing it reduce down to 217 arcsec. Now we try to use different annealing method and different annealing temperature to observe the ZnO film properties. First, we used the post- furnace annealing. Fig. 4-12(a)~(d) show PL spectra measured at room temperature of the ZnO films annealed at different temperature and different annealing time in furnace in O2 ambient and the non-annealed ZnO film under a condition of laser energy density 7J/cm2 and growth
temperature at 600℃. We observed two emission bands from all samples in Fig.4-12 (a)~(d). One emission band is near band-edge emission at UV range that is attributed to free-exciton recombination, and the other one is visible light emission, which is produced by transition of exc ited optical center from deep level to valence band, and such deep-band emission is usually accompanied by the presence of shallow donors (oxygen vacancies) and deep acceptors (Zn interstitial). Fig. 4-12 (a)
~(d) also show that no matter what the annealing temperature and time the PL intensity is improved after annealing. It is well known that the intensity of light emission was determined by the radiative and nonradiative transition. The luminescence efficiency of the light emission can be described by the following formula:
? = WR / (WR + WNR ),
where ? is the luminescence efficiency, WR and WNR are radiative and nonrradiative transition probabilities, respectively. Therefore, many lattice defects and surface defects are contained in the as-grown ZnO film. These defects produce various nonradiative centers to reduce light emission. As long as the annealing time is too long the deep- level emission is enhanced rapidly no mater what annealing temperature is. That the deep- level emission becomes strong is a result of increasing the oxygen vacancies or Zn interstitial. These phenomena also occur when the annealing temperature is too high in Fig. 4-12 (c)~(d). Therefore, at high temperature the defects may be formed in the film by re-evaporation of the O2
molecule, and following reactions may occur at high temperatures:
ZnO → ZnZn +Vo +1/2 O2
ZnO → Znj + 1/2O2
As a result, both Znj and Vo act as donors. In addition to re-evaporation of O2
molecule, formation of the spinel layers at the interface between the sapphire
substrate and the ZnO epitaxial layer by higher temperature annealing may also be the cause of the degradation of ZnO layers. From Fig. 4-13, we found the biggest ratio of near band edge emission (NBE) to the deep level emission at annealing temperature 800℃ in furnace with O2 ambient. The best ratio is about 52 in Fig.
4-13. Fig. 4-14 indicates the different annealing time at the same temperature 800℃
in furnace with O2 ambient. From Fig. 4-13 and Fig. 4-14, we found the best annealing temperature and time are 800℃ and 1hr in our experimental system. The above-mentioned method is the furnace annealing with O2 ambient.
We also used the in-situ annealing to improve the structure of ZnO. Fig. 4-15 indicates the PL spectra of in-situ annealed ZnO films at 700℃ and different annealing time and the non-annealed ZnO film under a condition laser energy density 7J/cm2 at 2Hz rep. rate and growth temperature 600℃. We can easily observe the NBE intensity is strong after annealing 1hr. The intensity ratio at RT of near band edge emission to deep- level emission is 48 in Fig. 4-15. The near band-edge emission with FWHM is 60meV. Compared to the post-annealing method, the in-situ annealing will induce less O vacancies even in the long annealing time. We guessed that the thermal force is a major reason because the post-annealing in furnace is quasi-rapid thermal annealing (RTA). The force may cause O molecule evaporate rapidly. So the optimal condition for in-situ annealing is 700C for 1hr under high vacuum P=10-9 torr. In-situ annealing is a better choice because the less O vacancies will be induced. Fig. 4-16 shows photoluminescence spectra from the ZnO thin film at 10K and the film grown at 600℃ and in-situ annealing at 700℃ for one hour. The inset in Fig. 4-16 shows, in a logarithmic scale, the NBE emission at 10K. The dominant peak at 3.364eV is from the exciton bound to neutral donors (DoX) transition. The FWHM of peak is around 4 meV. The low energy extending tail of the band edge is attributable to emission from the longitudinal phonon replicas (LO)
of free exciton (EX) and donor bound exciton (DoX), as shown in the inset of Fig.
4-16. Table.4-2 shows the mechanism and the positions of spectral peak.
Fig. 4-17 shows emission spectra of the ZnO film in the temperature range from 10 ~ 300 K. At low temperature, the spectra are dominated by the neutral-donor bound exciton emission at 3.364 eV. With increasing temperature, the bound exciton emission decreases rapidly, and is hardly resolved above 140 K. At around 70 K the acceptor bound exciton vanishes. At the onset of bound exciton decay, a feature at 3.378 eV becomes stronger and finally dominates the spectra even up to room temperature. The feature is assigned to free exciton emission due to its narrow line width and the temperature dependence of its intensity.
Fig. 4-18 plots the positions of the exciton emission up to 295 K. Similar plots have been reproduced from D. M. Bagnall et al.23) We performed numerical analyses of the energy position of the free exciton as function of temperature using a model, a modification of Varshni equation for the band gap temperature dependence:
E(T) = E(0) - 3
A band gap value obtained from Eq (1) fitting to the Fig. 4-16(a) gives 3.37752 eV.
Similar energy band gap ( E0 =3.3801 eV) been produced from Boemare et al. 25,26) In Ref. [24] the authors suggest a semi-empirical expression equivalent to the Bose-Einstein model proposed by Ref. [25] that provides us with fitting and an estimative of the Debye temperature. Eq.(2) gives the best fitting to the data Fig.4-16(b), with a Debye temperature of TD=3/2T=550K, where T is the effective phonon temperature. The different values of the parameters are given in the inset of Fig. 4-16(b).
It indicates that we have determined the temperature dependence of the free exciton
peak position in the high-textured ZnO film in good agreement with the literature.25)
Within a Frank-Condon model26), the coupling of the exc iton transition to the LO phonon is expresses by the exciton-phonon coupling constant, S. The intensity distribution of the phonon replicas is determined by S. For the case of relatively strong coupling, the emission intensity of the nth phonon replica In(?) and the principle emission line (Io) is related by
From the Fig. 4-16, we can found the S factor to be 0.213 for free exciton and 0.0136 for the donor bound exciton at n=1. From the S factor, we know the free exciton couple phonon stronger than the neutral donor bound exciton.