1. Introduction

Zinc oxide (ZnO) is an optical material at ultraviolet band due to its direct wide band gap at 3.37 eV and high efficient excitonic emission at room and elevated temperatures [1]. However, due to that the energy of the valence band maximum is quite low relative to the vacuum energy, there is always an intrinsic obstacle for efficient p-type doping [2]. The arduous problem has not been solved until now. Since the theories predicted the deep nature of simple substituting acceptors for the group-VA dopants [3], what is the real form of the shallow acceptor in group-VA-doped ZnO material has become a pending question. Theories have predicted several possible forms, which can be classified into three categories. One is solely related to the intrinsic defects, like 2V_Zn-V_O and V_Zn clusters [4,5]. The second one is related to both the intrinsic defects and the extrinsic dopants, like 2V_Zn-X_Zn, V_Zn-N_O, and 3V_Zn-X_i (X = P, As, Sb) [6–8]. The last one is solely related to the extrinsic dopants, like (NH₄)_Zn and (N₂)_Zn [9,10]. Despite the various theoretical predictions, experiments are quite rare, and no conclusion has been drawn.

The partial reason for the few experimental reports could be due to the difficulty of characterizing and identifying the acceptors in complex forms. Actually, besides a set of characterizing methods, logic inference is also indispensable, making the identification indirect. The pioneer papers by Reynolds et al and Ton-That et al have experimentally identified the V_Zn-N_O and (N₂)_Zn complexes as the forms of shallow acceptors in nitrogen-doped ZnO material [11,12]. Moreover, in our recent works, we have tried to establish completed characterization methods for the V_Zn-N_O [13], the zinc-sited nitrogen complexes [14], and the intrinsic V_Zn clusters [15]. By comprehensively considering the electrical, optical, vibrational, and paramagnetic properties, we are able to identify the proposed complex acceptors experimentally. However, these works are far from perfection to give full understanding to the identification of the potential complex acceptors. More works are required to eliminate the other factors that would affect the identification, like crystalline quality and ways to fabricating the material. More importantly, after properly identifying the complex, we should be able to control the form and ‘doping level’ of the complex acceptors by means of tuning the growth and post-growth processing parameters.

With such demands, in the current paper, we shall report on the identification of acceptors in NH₃-doped ZnO films. In our previous paper [16], we have shown that samples grown on high-quality ZnO template substrate (a 2-μm-thick ZnO epi-layer on c-plane sapphire) [17]. have better crystalline quality as compared to that grown on bare sapphire [18]. Meanwhile, nitrogen incorporation has been achieved in the high-quality ZnO films possibly due to that the ammonia is more efficient than the nitrous oxide in terms of pyrolysis. In the current paper, we continue to investigate the optical fingerprints of the acceptors as well as the donors, which have been established in the low-temperature photoluminescence spectrum and the Raman spectrum. The results could add important information to the existing database of the attribution to optical transition lines and Raman modes of the nitrogen-doped ZnO material.

2. Experimental details

2.1 Film growth

The film sample was grown by a home-built metal-organic chemical vapor deposition (MOCVD) system. Previous experience has told us that relatively low substrate temperature is indispensable for realizing efficient nitrogen doping, whereas the crystalline quality of the resulted film is relatively low [19]. In order to solve this contradiction, we have employed a ZnO template as the substrate for ZnO film growth. As mentioned in the Introduction part, the ZnO template is actually a 2-μm-thick ZnO epi-layer on c-plane sapphire. The crystalline quality and surface smoothness have been determined to be quite good. The full width at half maximum (FWHM) value of the (0002) diffraction peak obtained by x-ray rocking curve (XRC) is only 260 arc sec, while the surface root-mean-square roughness is less than 3 nm as observed by an atomic force microscope in a 5 μm × 5μm area [20].

Before growth, the template substrate was pre-treated in N₂O atmosphere at 1000 °C for 5 min to obtain an adsorption-free surface. After that, in situ growth of a NH₃-doped ZnO film was carried out at 470 °C for 30 min. Dimethyl zinc (DMZn) was diluted and carried by 6N-purified N₂ gas at the flow rate of 20 SCCM (standard cubic centimeter per minute). Nitrous oxide (N₂O) and ammonia (NH₃) were employed as the oxidant and nitrogen doping source at the flow rates of 1000 and 20 SCCM, respectively. A low temperature plasma system was employed to enhance the dissociation of the N₂O and NH₃ molecules into oxygen and nitrogen ions at the relatively low substrate temperature (470 °C) for the ZnO growth and doping. The chamber pressure was maintained at around 20 kPa. In our previous paper, we have managed to achieve better crystalline (XRC FWHM of ~250 arc sec) and surface quality (RMS ~2.6 nm and averaged grain size > 400 nm) as compared to the film grown on bare c-plane sapphire [16].

2.2 Post-growth process and characterization methods

After growth, the as-grown sample was annealed. Annealing can help activate the possible acceptors [21] while further reduce the dislocation density and improve the crystalline quality. Considering the growth was done at 470 °C, the annealing temperature was thus selected between 600 to 900 °C. It has been found that annealing at temperatures higher than 900 °C would cause a severe decomposition of the ZnO material. The annealing process was done in the processing chamber of an Ecopia Rapid Thermal Processing System (RTP-1200). In order to protect the nitrogen from desorbing to the greatest extent, a 5-min rapid annealing under the N₂ (100 SCCM) and N₂O (10 SCCM) mixed ambience was carried out. Sample points of the processing temperature were selected at 600, 700, 800, and 900 °C, respectively.

The optical properties have been measured by photoluminescence. The low-temperature and temperature-dependent measurements have been done in a helium-temperature cryogenic system. A 325-nm He-Cd laser was employed to excite the samples. The resolution of the wavelength is around 0.1 nm. The Raman spectra were measured by a JOBIN YVON HR800 Raman system in backscattering geometry at room temperature. A 514.5-nm argon-ion laser was employed to excite the samples.

3. Results and discussion

3.1 Reviewing the electrical properties

In the previous paper [16], we have discussed the thermal evolution of the forms and concentration of the zinc interstitial (Zn_i) related donors of the NH₃-doped ZnO film sample on the ZnO template. In that paper, capacitance-voltage (C-V) measurement was employed to characterize the doping concentration of the samples. The reason why we use C-V instead of Hall to measure the electrical properties is due to the high resistance of the samples, where the Hall voltage signals would be very weak [22]. The C-V measurement has shown that the as-grown sample is n-type with the net donor concentration of ~10¹⁵ cm⁻³. While for the annealed samples, the conduction type has been converted to p-type with the net acceptor concentration in the order of 10¹⁵-10¹⁶ cm⁻³. In addition, the acceptor doping concentration roughly increases with increasing the annealing temperature. Obviously, the gradual decomposition of the Zn_i-related donors, including the Zn_i-N_O complex and the Zn_i small clusters, should be partially responsible for the observed results of electrical characterization [16]. However, the conversion of the conduction type cannot be achieved without forming and/or activate acceptors. Furthermore, the variation of the net acceptor doping concentration with the annealing temperature indicates that the forms and concentration of the acceptors should change as well. Therefore, we shall discuss this issue by showing and analyzing the optical characterizations in the following sections of the paper.

3.2 The PL emission lines and their assignments

Figure 1(a) shows the excitonic region (3.30 – 3.38 eV) of the near band-edge PL spectra for the samples measured at 10 K. Two peaks can be tracked for each spectrum. According to the well-established database and utilizing the labels there, the two peaks are marked as Y-line and X-line as ticked in Fig. 1(a) [23]. The optical transitions in between the X-line region are commonly ascribed to excitons bound to shallow donors and acceptors in ZnO material [24]. It is noted that the widths of the X-line peaks (10-15 meV) are generally wider than the regular value (<10 meV) for ZnO material with similar crystalline quality [25], and the shapes of the peaks are somehow asymmetric. Therefore, the peaks could be overlapped by more than one components, and a peak deconvolution has been applied for further analysis. After the process, two components located at 3.363 and 3.355 eV have been deconvoluted. The 3.363 eV line, labeled as I₄ in the literature [23], has been ascribed to excitons bound to shallow donors (D⁰X). Applying the Haynes rule for n-type ZnO [26], the ionization energy of the responsible donor is calculated to be around 30 meV, which is consistent with the 0/1 + transition energy of the Zn_i-related shallow donors. The 3.355 eV line, whose energy is between I₉ and I₁₀ [23], is usually ascribed to excitons bound to shallow acceptors (A⁰X). Utilizing the Haynes rule for p-type ZnO, the ionization energy of the responsible acceptor has been estimated to be about 190 meV, which is consistent with the transition energy of the shallow acceptors in nitrogen-doped ZnO materials [27].

 

Fig. 1 (a) The 13-K near band-edge PL spectra for the high-quality NH₃-doped ZnO films. (b) The extracted intensity ratio between A⁰X and D⁰X. The two insets show the extracted intensity of the D⁰X and A⁰X, respectively. (c) The extracted intensity of D₁⁰X and A₁⁰X in the Y-line spectrum region. (d) The temperature dependence of energetic position for the D⁰X, D₁⁰X, and DAP transitions of the sample annealed at 600 °C, and for the A⁰X, A₁⁰X, and DAP transitions of the sample annealed at 900 °C. The black solid and dashed lines are the fitting curves by the Varshni relation [$E\left(\text{T}\right)=E\left(0\right)-\frac{\alpha T^2}{T+\beta }$], giving α ~7.1 × 10⁻⁴ eV/K, and β ~836 K. It can be seen that the D⁰X, A⁰X, D₁⁰X, and A₁⁰X show good agreement to the Varshni relation. While the DAP transitions deviate from the relation with less red shift as increasing the measuring temperature.

Download Full Size | PDF

In order to quantify the concentration of the shallow donors and acceptors, the integral intensity values of the D⁰X and A⁰X peaks have been extracted and plotted in the inset of Fig. 1(b). Although the intensity of the bound exciton emission should be proportional to the amount of the related donors or acceptors, it can be also affected by the film quality and emission efficiency. The drop of the intensity for the A⁰X from the 600 °C-annealed sample to the 700 °C-annealed one could be due to the deteriorated crystalline quality and emission efficiency induced by the desorption of nitrogen and the aggregation of isolated zinc interstitials [16]. In this case, the relative intensity ratio between the A⁰X and D⁰X would be more helpful for us to determine the corresponding change of the donors or acceptors during the annealing process. Figure 1(b) shows the relative intensity ratio between the D⁰X and the A⁰X as a function of the annealing temperature. The monotonous decrease trend reveals that the amount of the shallow donors decreases while the amount of the shallow acceptors increases with annealing temperature. This trend matches the electrical properties of the samples well, which conversely verifies the validity of the X-line assignment.

Regarding the assignment of the Y-line, there is no unified conclusion until now. Researchers have previously assigned the emission peaks in the energy range from 3.32 to 3.34 eV to excitons bound to deep donors (D⁰X), excitons bound to deep acceptors (A⁰X), free-to-shallow-acceptor transition (FA), excitons bound to the extended structural defects (D⁰X), and two electron satellites (TES) of the excitons bound to shallow donors [23,28–31]. Considering the Y-lines shown in Fig. 1(a), the shapes are also asymmetric indicative of multi-origin. A similar deconvolution process has been applied to the Y-line peaks, which gives two components located at 3.326 and 3.332 eV, respectively. In a like manner, we have plotted the integral intensity values of the two components in Fig. 1(c). As can be seen, the trend does not follow that of the D⁰X shown in Fig. 1(b), indicating that the origin is not the TES of the D⁰X. Furthermore, we have extracted the temperature dependence of the peak positions for the two components as well as the D⁰X and A⁰X and plotted them in Fig. 1(d), showing the same Varshni relation [32]. As a result, the origin of the two Y-line peaks should be excitonic. According to the electrical property trend, the component at 3.332 eV should be assigned to excitons bound to another kind of acceptors (A₁⁰X). While the ‘Λ’-shaped variation of the 3.326 eV component with the annealing temperature resembles the trend of the Zn_i small clusters as shown in the previous paper [16], implying that this emission may have some relevance to the Zn_i small clusters.

According to the above assignments, donors and acceptors should co-exist in the samples. Commonly in this case, the emission directly from the donor level to the acceptor level, aka the donor-acceptor-pair (DAP) transition, could be observed. We indeed see the emission at around 3.258 eV for all the samples as shown in Fig. 2. Generally, the shift of the DAP peak position with the measuring temperature is different to that of the bound or free excitons, which could be utilized as a symbol to differentiate the DAP emission from the excitonic one. The difference is caused by the different strength of the Coulomb interaction at different measuring temperature. The energy position of the DAP transition can be expressed as E_DAP(T) = E_g(T) – E_D- E_A + e²/4πεr_DA, where E_D and E_A are the ionization energy of the responsible donor and acceptor, respectively, e is the electron charge, ε is the dielectric constant of ZnO, and r_DA is the averaged distance between the donor and the acceptor [33]. If the donors and acceptors (the case for highly compensated high resistance) are abundant, when the measuring temperature increases, the thermal excitation of electrons or holes on the donor and acceptor levels may not change the r_DA value obviously. In consequence, the DAP position would red shift with a slightly lower rate as compared to the excitonic emissions. Sometimes, the fluctuation of the Coulomb potential caused by large amounts of ionized donors and acceptors could make the temperature dependence of the DAP position non-monotonous. However, if the amount of the donors and acceptors are less (the case for lightly compensated high resistance), when the measuring temperature increases, the ionization of donors and/or acceptors would critically affect the r_DA valued. It could be understood that the possibility of the electron on the donor level transiting to the acceptor level would be much higher for the close-distant DAPs than the far-distance DAPs. In other words, as increasing the measuring temperature, far-distance DAPs are much easier to be ionized, leading a reduce r_DA value. In this case, the energy of the DAP may show blue shift as increasing the measuring temperature. Figure 1(d) also shows the temperature dependence of the peak position for the 3.258 eV emission line. As can be seen, the line red shifts with a slightly lower rate as compared to the X-line and Y-line, implying that the line is ascribed to the DAP emission and the samples are highly compensated.

 

Fig. 2 The 13-K PL spectra for the high-quality NH₃-doped ZnO films in the range of 2.2-3.4 eV, showing the DAP and deep-level emissions.

Download Full Size | PDF

Moreover, the intensity of the DAP emissions reaches a maximum at around 700 – 800 °C. In fact, the intensity of DAPs should be proportional to the amount of the responsible donors and acceptors. As indicated in Fig. 1(b), the concentration of the shallow donors decreases while the one of the shallow acceptor increases monotonically as increasing the annealing temperature, which would naturally result in a maximum intensity for the DAP transition at moderate annealing temperatures. This is valid only when the nonradiative recombination could be neglected. Actually, in our previous paper [16], the (0002) widths of the X-ray rocking curves (XRC) for all the samples are in the range from 0.12 ^o – 0.14 ^o, indicative of relatively high crystalline quality and low concentration of nonradiative centers. Considering a sudden drop of the A⁰X intensity has been observed for the 700 °C-annealed sample and the XRC width and Raman intensity at 276 cm⁻¹ are highest, the nonradiative recombination might be relatively stronger for the 700 °C-annealed sample. But even if the nonradiative recombination is taken into account, the ‘Λ’-shaped intensity change trend for the DAP intensity versus annealing temperature is still valid.

For the other two humps at 3.187 and 3.116 eV, their intensity variation with annealing temperature shows the similar trend to that of the DAP transition. Besides, the energy interval between the three peaks (3.258, 3.187, and 3.116 eV) is 71 meV. Therefore, the peaks at 3.187 and 3.116 eV are ascribed to the 1st and 2nd longitudinal optical (LO) phonon replica, respectively.

In the deep-level region as shown in Fig. 2, for the samples annealed above 800 °C, the green band (GB) emission gradually emerges. It should be noted that the GB emissions have fine structures, which is due to the strong coupling between the electrons and phonons. Utilizing the method discussed in Ref [34]. and applying the method to the spectrum of the 900 °C-annealed sample, a very large Huang-Rhys parameter of 6.7 could be fitted. That is why we could observe 6 LO phonon replicas on the lower energy side of the zero-phonon-line (ZPL) at 2.85 eV. According to our previous papers [35], the GB emission is actually another DAP transition. The donors are assigned to the shallow donors while the responsible acceptors are much deeper. The strong electron-phonon coupling indicates that the shallow donors are tightly bound to the deep acceptors. The emergence of the GB emission and the suppression of the DAP emission at higher annealing temperatures (> 800 °C) indicate a possible transition from the shallow acceptors to deeper ones.

3.3 The identification of the donors and acceptors

In the previous section, we have assigned all the PL emission lines. However, a more important issue is to identify the specific forms of the responsible donors and acceptors. The main form of the shallow donors is identified as the Zn_i-N_O complex. In our previous work [16], we have extensively studied the thermal evolution of the Zn_i-related donors in the high-quality NH₃-doped ZnO films. Considering the high activity and mobility of the isolated Zn_i atoms, they are impossible to stably exist while the Zn_i bound to an additional N_O could be the stable form of Zn_i-related donors in the as-grown film sample [36,37]. As increasing the annealing temperature, the oxygen-sited nitrogen atoms begin to be unstable and mobile. The nitrogen 1s X-ray photoelectron spectra (XPS) [16] have shown that some of the anion-sited nitrogen desorbs from the sample, and others may move to the cation-sites after annealing, leading to the break of the bond between the Zn_i and N_O. This process would result in the continuous reduction of the amount of the Zn_i-N_O complexes. This trend also matches the change of D⁰X intensity as shown in Fig. 1(b).

The main forms of the acceptors could be some zinc-sited nitrogen complexes and V_Zn small clusters. In the nitrogen 1s XPS after annealing [16], although the anion-sited component almost disappears, the component related to N-H bond is still prominent. Bang et al. [9] have shown theoretically that the (NH₃)_Zn defect is thermodynamically the most stable defect under O-rich conditions. The stability has been attributed to the formation of a strong dative bond of the ammonia molecule with a neighboring oxygen atom. Since we have employed molecule NH₃ doping, it is highly possible that the ammonia molecule could be doped as a whole. Furthermore, the (NH₃)_Zn defect can react with additional hydrogen, which is abundant in MOCVD-grown materials, to form a (NH₄)_Zn defect. The additional hydrogen would break the N-O bond between the NH₃ and the neighboring lattice oxygen, and thus making the defect available of accepting an additional electron. Therefore, the (NH₄)_Zn defect is an acceptor. The 0/1- transition energy of the (NH₄)_Zn complex has been calculated to be quite shallow at around 200 meV, which accords perfectly with the experimentally determined ionization energy of the shallow acceptor responsible for the A⁰X as discussed in the section 3.2.

Besides, in the nitrogen 1s XPS of the 900 °C-annealed sample [16], a peak related to the N-N bond could be observed. Considering that the zinc-sited double nitrogen [(N₂)_Zn] is also an energetically favorable shallow acceptor as shown by Ton-That et al. [12] and our group [14], the emergence of this component in the XPS might be due to the formation of the (N₂)_Zn shallow acceptors for the high-temperature-annealed sample. The process of forming the (N₂)_Zn complexes requires the diffusion of N_O or N_i to a V_Zn. Liu et al. have simulated the minimum energy path from N_O to N_Zn. Based on their result, a potential barrier of 1.6 eV exists [30]. This result could explain the reason why the process requires high-temperature annealing. Nevertheless, the formation of the (N₂)_Zn complex may not be achieved without an additional bond to the lattice oxygen and the assistance of hydrogen. Both binding to adjacent lattice atoms and hydrogen are beneficial to reducing the total energy of the system [14].

In the Raman scattering spectroscopy, we have also found some evidences on the existing of the nitrogen-related complexes. Figure 3(a) shows the Raman spectra for all the samples in the wavenumber range from 100 – 200 cm⁻¹. A mode at around 160 −170 cm⁻¹ can be clearly observed for the sample annealed at 900 °C. This mode is not related to any modes of ZnO, but has been reported to be the N-H vibration in the ammonia molecule [38]. In fact, in the Raman spectra of some previous works discussing the ammonia doping in ZnO material, the mode at the identical wavenumber range could be observed, although the authors have not noticed and discussed [39,40]. Therefore, the emergence of this mode supports the possible existence of the (NH₄)_Zn complexes. Moreover, in the high-wavenumber range of the Raman spectra as shown in Fig. 3(b), weak modes at around 3100 cm⁻¹ and 2250 cm⁻¹ can also be observed for the 600 °C and 900 °C-annealed samples. These two modes have been assigned as the vibrations of the N-H and N-N, respectively [41,42]. This result also supports the existence of the (NH₄)_Zn and (N₂)_Zn complexes in the samples.

 

Fig. 3 The Raman backscattering spectra of the as-grown, 600 °C-annealed, and 900 °C-annealed samples in the range of (a) 120-220 cm⁻¹, (b) 2150-2300 cm⁻¹, and (c) 3000-3200 cm⁻¹ wavenumber regions.

Download Full Size | PDF

For identifying the forms of the shallow acceptors, the V_Zn clusters is also a possible candidate. Tuomisto et al. have shown that the incorporation of nitrogen into ZnO material would lead to the formation of stable V_Zn clusters as well as negative ion-type defects [43]. Bang et al. have shown that the V_Zn clusters are grown by the diffusion of an additional V_Zn to an existing V_Zn-V_O complex, forming a 2V_Zn-V_O defect, which can stably exist at high temperature [4]. The V_Zn clusters have been commonly seen in ion-implanted and flash-annealed nitrogen-doped ZnO materials, characterized by the positron-annihilation spectroscopy [5]. Therefore, the V_Zn clusters can be formed and stably exist at high-temperature processed nitrogen-doped ZnO materials with no doubt. In fact, we have observed the enhanced GB emissions for the samples annealed at higher temperatures (> 800 °C). A large amount of V_Zn is thus available, which is a prerequisite for the clustering of the isolated V_Zn. As the V_Zn clusters are shallow acceptors, we think that the V_Zn clusters could be another source of the shallow acceptors in the samples annealed at high temperatures.

As abovementioned, another donor (D1) and acceptor (A1) with much deeper transition energies are responsible for the Y-line emission. Regarding the identification of D1, the intensity variation shows a ‘Λ’ shape, which is consistent with the change of the concentration of the Zn_i small clusters [16]. We have also simulated the 0/1 + transition energy of both the isolated Zn_i and the Zn_i small clusters, showing that the ionization energy for the clusters are higher than the isolated Zn_i. Combining the theoretical and experimental results, the deep donor D1 is attributed to the Zn_i small clusters. For the identification of A1, by utilizing the Haynes rule for p-type ZnO, the ionization energy of the A1 is estimated as round 420 meV, which is similar to the reported value of isolated V_Zn acceptor [44,45]. Moreover, the intensity variation shows a similar trend to that of the GB emission. Therefore, the A1 is attributed to the isolated V_Zn deep acceptors.

The donor and acceptor responsible for the DAP emission have been determined to have the same origin for the A⁰X and D⁰X emissions as evidenced by the intensity variation trend discussed in the section 3.2. Actually, according to the ionization energies of the shallow donor (30 meV) and acceptor (190 meV), the DAP transition energy at 10 K is calculated to be 3.21-3.29 eV depending on the strength of the Coulomb interaction (0 – 80 meV). The PL energy position of the DAP emission (3.258 eV) accords well with this range, which further supports the conclusion. The GB emission is ascribed to the tightly bound Zn_i-V_Zn complexes as already proven in previous literatures [32,35]. In fact, considering the limited concentration of the isolated Zn_i in the samples annealed at higher temperatures, the formation of the complex at higher annealing temperatures could be due to the small displacement of the lattice zinc, which naturally results in a closely-bound Zn_i-V_Zn Frenkel pair. Since it is a DAP emission, we can also estimate the energy position from the respective ionization energy of the isolated Zn_i (30 meV) and V_Zn (420 meV), which gives the energy position in the range of 2.98 ~3.06 eV (Assuming the Coulomb interaction item as 0 – 80 meV). This range is 0.1 ~0.2 eV higher than the ZPL energy (2.85 eV). This inconsistency could be due to that the lowest transition energy level of the isolated V_Zn has been occupied by the electron from the 1+/2 + energy level of the isolated Zn_i, making the actual transition happen from the 0/1 + level of the isolated Zn_i to the 1-/2- level of the isolated V_Zn. The energy of this transition is obviously lower than that to the 0/1- level of the isolated V_Zn.

4. Summary

In this paper, we have investigated the optical properties of the high-quality NH₃-doped ZnO films annealed at different temperatures. The optical fingerprints have been established for the donors and acceptors. The Zn_i-N_O complex has been assigned to the form of the shallow donors with an 0/1 + transition energy at 30 meV. The (NH₄)_Zn complex, (N₂)_Zn complex, and V_Zn small clusters have been thought to be the possible shallow acceptors with the 0/1- transition energy at around 190 meV. At current stage, we are unable to tell which one plays the dominant role. However, all the three shallow acceptors are strongly correlated with the V_Zn. Besides these shallow defects, a deep acceptor in the form of the isolated V_Zn and a deep donor in the form of the Zn_i small clusters have been found to co-exist in the material, which are responsible for the Y-line emission. The transition from the donor level of the Zn_i-N_O to the acceptor level of the shallow acceptors results in the DAP emission. The transition from the shallow donor level of the isolated Zn_i to the deep acceptor level of the isolated V_Zn within a Frenkel pair (Zn_i-V_Zn) gives rise to the fine-structured GB emission. A schematic diagram has been drawn in Fig. 4 to make these assignments illustrative.

 

Fig. 4 The schematic diagram showing the main optical transition paths and the defects responsible for the transitions.

Download Full Size | PDF

Furthermore, the post-growth annealing temperature could be utilized as a parameter to tune the concentration of the defects. The concentration of the Zn_i-N_O monotonically decreases as increasing the annealing temperature due to the desorption of the N_O. However, the amount of the Zn_i small clusters would increase first and then decrease due to the aggregation of the isolated Zn_i at moderate annealing temperatures. The concentration of the shallow and deep acceptors monotonically increases as increasing the annealing temperature. The formation of the NH₄ complex is due to its low formation energy at zinc site. The formation of the (N₂)_Zn, V_Zn small clusters, and the Frenkel pair (Zn_i-V_Zn) could be due to the enhanced diffusive ability of atoms by the additional energy supplied from the high temperature annealing. This study has deepened the understanding to the possible forms of shallow acceptors and compensating donors and has provided a way to control the defects in nitrogen-doped ZnO material.