1. IntroductionDue to the stable existence of excitons at room temperature and direct wide band gap, zinc oxide (ZnO) has been considered as a promising material for short-wavelength light emitting and detecting applications.[1] High efficient optically driven light emission has been demonstrated, proving the outstanding physical property of the material.[2] However, the efficiency of the electrically pumped devices drops dramatically mainly due to the notorious difficulty in p-type doping in ZnO.[3] Such an asymmetric doping phenomenon is frequently seen in wide band gap semiconductors. As the theoretical calculation shows, the valence band maximum (VBM) energy of ZnO is very low, making the ionization of popular acceptors extremely hard.[4] Furthermore, intrinsic donor-like defects are easily formed when trying to do p-type doping in ZnO. Zinc interstitials (Zn
) and oxygen vacancies (V
) are two main intrinsic donors killing holes.[5] Therefore, high efficient p-type doping can only be achieved when a shallow acceptor is found and the compensation is suppressed.
On the way of solving the p-type doping problem, legions of researchers have tried a lot of possible elemental candidates and have developed several methods. The p-typeness has been continuously reported. However, the convincing results are rare. The sharp contrast between them again indicates the hardness of p-type doping in ZnO. Nevertheless, encouraging results from some leading groups have brought hope to the global researchers. Nakahara et al. have demonstrated the first blue ZnO homo light-emitting diode (LED)[6] and holds the power record of a single ZnO LED device.[7] The p-type films were realized by nitrogen doping in ZnO and ZnMgO, respectively. Shan et al. have developed a co-doping approach of lithium and nitrogen, and have fabricated a ZnO homo LED having a life time as long as 97 h.[8] Chu et al. have done excellent work on antimony doping in ZnO nanowires, and have applied the p-type Sb-doped ZnO to random laser.[9] These highlights demonstrate that p-type doping in ZnO, although difficult, is absolutely feasible.
However, recent theoretical calculations have complicated the origin of holes. Simple substituting defects, like group VA on oxygen, are all estimated to be deep acceptors.[10] In order to clarify the inconsistency between theory and experiment, more and more complicated acceptors have been proposed to be responsible for the observed p-type behavior. The proposed form of the acceptor could be extrinsic or a combination of extrinsic and intrinsic defects. Taking nitrogen doping for example, ammonia molecule or nitrogen molecule at zinc site [
or 
] has been calculated to have favorable formation energy as well as shallow ionization energy level.[11,12] Meanwhile, complexes involving zinc vacancy (V
) and extrinsic defects (V
–N
, 3V
–N
) or V
clusters (V
–V
, V
–V
–V
) are also suggested as the origin of hole providers.[13–17] The various theoretical predictions have indeed provided the ideas for experimentalists to artificially form the complex acceptors in ZnO. However, the difficulty can be expected since the accurate control of the desired complex form must be hard. Moreover, the experimental identification of the complex acceptors is not easy either, unless we can establish solid fingerprints for the complexes. Despite so, a lot of work awaits more inputs from global researchers in the future.
Regarding the compensating donors, the intrinsic defects and unintentionally incorporated impurities are the two origins. Until now, the origin of electron background and the ultimate suppression of it are still far from being fully understood and solved. On one hand, the intrinsic defects cannot be easily eliminated by a simple technique during growth or post-growth treatment. For example, Zn
related compensating donors are obstinate and can have various forms. An interstitial zinc can bond with another interstitial zinc, oxygen vacancy, or some extrinsic dopant like nitrogen to form donor-like complexes.[18–20] They are hard to suppress because they are sometimes energetically favorable. On the other hand, unintentional donors, like hydrogen, are also annoying since they ubiquitously exist in the growth and post-growth processes.[21] Therefore, the research on compensating defects is always of interest and requested. It is now well-accepted that alloying with magnesium (Mg) can significantly elevate the energy of the conduction band minimum (CBM) and hence the ionization energy of donors.[22] In fact, beryllium, sulfur, and tellurium (Be, S, and Te) may have similar effects to magnesium. These iso-valent elements may be beneficial to both the suppression of intrinsic donors and the easier activation of acceptors.[23–25] Therefore, p-type doping in ZnBeO, ZnMgO, ZnSO, and ZnTeO may be another route deserving further investigation.[26]
In this article, we shall review the recent progress of (i) the native defects and (ii) the p-type doping approaches of ZnO material in order to provide a comprehensive summary of the efforts made mainly in a period from 2011 till now globally. As can be seen below, people’s understanding of the behavior and control of the native defects and the desired acceptors has advanced to an unprecedented level. We must admit that the developing pace has been slower than that in 2001–2005 or 2006–2010. However, there are still remarkable results coming out. Some novel ideas may have pointed out the future direction of research on p-type doping of ZnO. We are confident that high-efficient p-type doping is applicable and a lot of research topics are pending.
2. Native defects in ZnONative defects are lattice imperfections involving only the constituent elements, including vacancies, interstitials, and antisites. Particularly in ZnO, V
, V
, Zn
, O
, Zn
, and O
are the basic forms of intrinsic defects. Some of the defects could be electrically, optically, and magnetically active, and thus would influence the electrical, optical, and magnetic properties of ZnO critically. As a fundamental topic for a material, the formation, evolution, and control of the native defects have been studied via simulations as well as experiments. However, until now, there are still a few controversial issues being debated, like the origins of the residual electrons and deep-level luminescence. Therefore, the investigation on the native defects in ZnO has not been closed yet.
For the simulation part, various papers have been published to study the formation and property of the intrinsic defects.[27–29] Traditionally, density functional theory (DFT) using local density approximation (LDA) and generalized gradient approximation (GGA) has been widely employed for the simulation. The biggest problem for this method is that the band gap value and the energetic positions for the defects have always been underestimated by a significant amount. The erroneous predictions will definitely affect the results of formation energies. Considering this, a few new approaches including the LDA+U, GGA+Uq, B3LYP, and HSE hybrid functionals have been developed to overcome the deficiencies from DFT–LDA or GGA. These methods include self-interaction corrections, and have significantly corrected the band gap values.
Figure 1 shows the general results of GGA+U calculations for native defects in ZnO with different Zn and O chemical potentials by Vidya et al.[30] As can be seen, for the donor-like defects (V
, Zn
, and Zn
), V
has the lowest formation energy, while for the acceptor-like defects (V
, O
, and O
), V
has the lowest formation energy. In general, intrinsic defects have a tendency to counteract the prevailing conductivity. In other words, in the n-type doped samples acceptor-like defects are easier to form whereas in p-type doped samples the donor-like defects are easier to form. As a result, the native defects are more likely to act as compensators rather than the majority carrier providers. The simulation results provide the formation of and transition between different covalent defects under equilibrium condition. Experimentally, it could be a basis for inducing the specific defects via tuning the chemical potential and doping level (or say Fermi level) during growth or post-thermal annealing. However, a more efficient and direct path to creating the native defects is through non-equilibrium processes, like ion-implantation and electron or laser irradiations. In this case, the formation and concentration of the desired native defects could be intentionally introduced and controlled. In the following part, we will show the basic properties of the major intrinsic defects (mainly V
, V
, and Zn
) and their complexes (like the popularly investigated Schottky and Frenkel pairs) as well as the latest theoretical and experimental progresses in these defects.
2.1. Oxygen vacancies (V
)Whether the V
is responsible for residual n-type conduction in ZnO is highly controversial. A very recent study of the V
self-diffusion process by employing a uniquely designed oxygen-isotope ZnO hetero-structures supports that the V
is +2 charged and responsible for the unintentional n-type conductivity.[31] But it is indicated by more researches that the V
cannot be the source of n-type background since it is neutral when the Fermi level is close to the CBM, making it a deep donor. However, whether the energy level is deep or shallow, it can be a source of compensation in p-type ZnO. In order to avoid incorporating the V
, an O-rich condition should be adopted during growth or annealing.
Simulations show that the 1+ charge state of V
is thermal-dynamically unstable. Utilizing this state, the V
could be created by irradiation and detected by EPR measurement. The g-factor of ∼ 1.99 is assigned as the magnetic signature of the paramagnetic V
charged state as shown in Fig. 2. The signal can only be seen after electron irradiation, which indicates that excitation is required to generate V
whereas the V
is unlikely to be present in as-grown n-type ZnO due to the predicted high formation energy.[32] The g-factor of ∼ 1.96 has previously been assigned to V
, however, the EPR signals could also been detected with doping Al, Ga, or In, and enhanced under UV illumination, which implies that more electrons have been excited into the conduction band. Nowadays, people tend to assign 
to electrons in the delocalized states (in conduction band).[33]
Regarding the optical signature of V
, the well-known green luminescence has been attributed to the excited-to-ground state of V
.[34] Heo et al. have assigned the deep-level emission (DLE) at 2.53 eV to V
.[35] Vlasenko and Watkins[36] and Evans et al.[37] have experimentally confirmed that the excitation threshold for V
is in a range of 2.0 eV–2.1 eV which is consistent with the simulated 
transition energy as shown in Fig. 1. Recently, Kaftelen et al. have observed teal and red emission in their ZnO nanoparticle sample.[38] The core-shell structures shown in Fig. 3 have been confirmed by x-ray diffraction (XRD) and Atomic force microscope (AFM). Two DLE peaks centered at 470 nm (teal) and 655 nm (red) have been detected by photoluminescence as shown in Fig. 4. Intriguingly, the relative ratio of the two peaks has the same trend as the area ratio of shell/core. Combining the EPR results, the peak at 655 nm (1.89 eV) has been ascribed to shell V
while the 470 nm (2.64 eV) peak has been attributed to core V
. Noticing that the green luminescence has also been assigned to other defects, the optical signature of V
is thus inconclusive and undetermined.
The thermal stability of V
has been checked by Janotti and Van de Walle, the calculated migration barriers for V
and V
are 2.4 eV and 1.7 eV, respectively.[39] These barrier heights indicate that V
and V
will become mobile and easily diffusive above 900 K and 650 K, respectively. Post-annealing above these critical temperatures is expected to remove the isolated V
defects.
2.2. Zinc vacancies (V
)V
is a double acceptor in ZnO. The 0/1– and 1–/2– acceptor transition energies have been calculated by DFT adopting various functionals. Before the band gap correction, the levels have been calculated to be 0.1 eV–0.2 eV and 0.9 eV–1.2 eV above the VBM, respectively.[39,40] However, after the band gap correction, Oba et al. have determined the transition energies to be 
eV and 
eV above the VBM,[29] indicating that even the V
is inactive it still acts as a hole provider. Considering the latter and more recent results giving a more valid band gap value of ZnO, it is hence inferred that the possibility of the isolated V
as a dominant shallow acceptor is questionable.
Experimental evidence of V
as a shallow acceptor has been quite rare recently. Teklemichael et al. have utilized infrared spectroscopy (IR) absorption and EPR measurement to characterize their ZnO nanoparticle samples synthesized by solid-state pyrolytic reaction process.[41] The resonance at g = 2.003 has been assigned to V
–H complex. The hole binding energy of 0.4 eV–0.5 eV for the V
–H has been obtained from the IR absorption, which is not a shallow acceptor. Khan et al. have artificially created isolated V
in ZnO bulk crystal by a 193-nm excimer laser irradiation.[42] The concentration and depth profile of the isolated V
have been determined by PAS. The PLE spectrum shows a peak at 3.18 eV, which has been assigned to the V
shallow acceptor with around 100-meV ionization energy. Wang et al. have identified the donor-acceptor pair (DAP) emission in their pulsed laser deposition (PLD)-grown ZnO films and ascribed them to the radiative transition from a shallow donor to the V
-shallow acceptor with an ionization energy of 188 meV.[43] However, The authors can never eliminate the possibility that the shallow acceptors are related to the localized states bounded to structural defects, like basal plane stacking faults.[44,45] Moreover, the structural defects are expected to be easily formed since the samples have been gone through structural destructive processes, like implantation and irradiation. As a result, more experimental efforts are required to reveal the true energetic level of V
.
In order to characterize the configuration and concentration of V
related native defects, people have constructed various relations between the measurable quantities and the V
properties. Taylor et al. have reported the EPR signals with g-factors in a range from 2.0018 to 2.056 in irradiated single crystals, and proposed that a subset of these lines could be due to V
.[46] Besides EPR, the PAS is another sensitive tool exclusively for defecting vacancy defects. Due to the relaxation of the neighboring Zn ions, V
does not result in a significant increase in open volume, and thus positron annihilation cannot detect V
, but very sensitive to V
.[5,47] The detection limitto V
in ZnO is on the order of 10
cm
.[16] More attractively, various types of V
could be differed by the line slope of the data fitting in the S–W parameter plot.
Figures 5 and 6 show the S–W plots of the ZnO samples containing different types of V
. In Fig. 5, the V
and V
have been characterized by the dash lines with different slopes with an intersection point at V
free. The transformations of the V
and V
has been controlled by annealing temperature, oxygen ambient pressure, and initial growth parameters as shown by Wang et al.[43] Moreover, Tuomisto et al. have found that nitrogen-implantation plus a flash post-annealing at 1200 °C leads to the formation of stable V
clusters and negative ion-type defects (N
). The fitting of the triangle points of the S–W plot is regarded as the signature for the stable V
clusters. It is obviously different from the fitting of the S–W data points collected from N-doped, N-implanted, irradiated, and Cd-doped ZnO samples, in which the V
clusters are volatile and unstable.[16] Therefore, the PAS is a golden standard for characterizing the concentration and even the configuration of the V
related defects and complexes at current stage.
Furthermore, more and more recent researches have suggested a relation between the DLE and the V
defects. Knutsen et al. have found monotonic decrease of electron concentration and mobility with increasing the irradiation energy.[48] Two DLE components at 2.45 eV and 1.75 eV have been found in the low-temperature photoluminescence, and the intensity of the 1.75 eV emission also increases with the irradiation energy. An elaborate design of annealing experiment has been implemented to investigate the origin of these two peaks as shown in Fig. 7. The annealing in O
increases the intensity at 2.45 eV, and subsequent annealing in Zn-vapor reduces the same band. The experiment performed in the opposite sequence leads to a first decrease of the 1.75-peak followed by a recovery of the peak. These results demonstrate that the 1.75 eV peak is related to V
and the 2.45 eV peak is not related to V
. Considering the evolution trend of the 2.45 eV peak with irradiation energy and the reduced electron concentration and mobility, the emission at 2.45 eV has been assigned to relating to O
. Dong et al.[49] have investigated the V
and V
clusters in Li and N ion-implanted ZnO. Flash annealing at 1200 
C and furnace annealing at 800 °C have shown distinguished characters for the DLE emission. Figure 8 shows that the 2.0 eV peak intensity has the same trend as the S-parameter depth profile obtained from PAS, indicating that the 2.0 eV peak is closely related to V
defects and the 2.4 eV peak is unrelated to V
. In addition, the energy level of V
large clusters has been thought to be shallower than that of the isolated V
or V
small clusters. Figure 9 shows the depth-resolved cathodoluminescences (DRCLs) for N-ion implanted ZnO and the 600 °C-furnace-annealed sample. Two peaks at 1.6 eV and 1.9 eV have been assigned to isolated V
or V
clusters and V
large clusters, respectively. The gradual shift of the DLE from 1.6 eV to 1.9 eV for the DRCL of the as-implanted sample can be understood by the higher concentration of V
as the depth increases. The steady emission at 1.9 eV for the annealed sample can be ascribed to the annealing induced coalescence of isolated V
and V
small clusters into V
large clusters.
The high stability of the V
clusters and N
in the flash annealed N-implanted ZnO samples have been hypothesized to relate to the balance between the amounts of V
and V
. Bang et al.[17] have conducted a kinetic Monte Carlo simulation and have given their understandings of the presence of vacancy clusters in ZnO. Figure 10 shows the general results. The clustering starts with the formation of the Schottky pair (V
+ V
) with a very high dissociation energy of 2.5 eV, making it difficult for the pairs to dissociate. Further growth of the clusters is by attracting additional mono-vacancies. Due to the fact that the diffusive ability of V
is larger than that of V
, more V
-abundant clusters (like 2V
–V
) are formed than V
-abundant clusters (like 2V
–V
). As can also be seen from Fig. 10, the tuning of Fermi level can modify the stability of the vacancy clusters. For instance, the (2V
–V
) clusters start to appear at a much higher temperature of above 500 °C when the Fermi level rises from 1.0 eV to 1.6 eV. In consequence, the observed stable V
clusters could be V
-abundant clusters grown from the starting Schottky pair (V
–V
). Figure 11 shows another interesting effect of the 2V
–V
formation. The repulsion by the lower V
levels would push the deep V
donor level into a level above the CBM which is evidenced by the experimental observation that the electron concentration increases during annealing.[17] Therefore, this kind of V
clusters may not be useful for p-type doping in ZnO.
2.3. Zinc interstitials (Zn
)The Zn
defect could occupy the tetrahedral or the octahedral site in the ZnO wurtzite structure. The occupation at octahedral site is expected to be stable.[50] As shown in Fig. 12, the transition energies of Zn
0/1+ and 1+/2+ states are very close to the CBM, meaning that the Zn
defects act as a shallow donor. On the other hand, the formation energy of the Zn
decreases quickly when the Fermi level moves down to the VBM, making the Zn
a possible candidate as a compensator to p-type doped ZnO.
Compared with V
and V
, Zn
has a relatively low migration barrier, which has been calculated to be 0.57 eV and experimentally determined to be 0.55 eV,[51] indicating that at low temperature of 
K, the isolated Zn
could be annealed out.[52] Therefore, it is unlikely that the isolated Zn
can stably exist at room nor elevated temperatures. Although it is shallow, the isolated Zn
may not be the source of the n-type residual electron carriers. However, the isolated Zn
could be stabilized by incorporating additional atoms or defects. Look et al. have suggested that under nitrogen ambience, the complex involving Zn
and N
(Zn
–N
) is a dominant shallow acceptor with 30 meV of ionization energy in n-type ZnO.[18] Kim and Park have suggested from the LDA+U calculations that the Zn
could be stabilized in the presence of a high concentration of the V
.[19] Gluba et al. have studied the cluster consisting of a few Zn
atoms.[20] The Raman B1 mode at about 275 cm
has been assigned to the zinc related local vibrational mode, and Zn
clusters in particular. Figure 13 shows the schematic diagram of the molecular orbitals of a Zn
cluster consisting of three interstitial atoms. The stabilization of Zn
clusters is related to the participation of the O 2p orbitals, which hybridize with the Zn
–Zn
bonding (
) and anti-bonding (
) orbitals, making the charge occupying the anti-bonding orbital (
) transferred to the CBM. In this case, the Zn
clusters can be stabilized as a shallow donor complex. Therefore, although isolated-Zn
is mobile and unstable, its complexes involving N
, V
, and its clusters could be stable and act as the source responsible for residual electrons.
The magnetic resonance and optical characterizations also suggest a shallow nature of the Zn
related defects. Vlasenko and Watkins have observed a characteristic “effective mass” g-factor of 1.96 from ODMR in their electron irradiated ZnO samples and have ascribed the signal to Zn.
.[36] Zeng et al. have detected the EPR signal for Zn
, giving g-factor of 
.[53] These values demonstrate that Zni defects are shallow donors. Zeng et al. also investigated the PLE of ZnO nanoparticle samples synthesized by a non-equilibrium technique: laser ablation in liquid.[53] Emissions centered at 415, 445, 555, and 600 nm have been excited as shown in Fig. 14. Combined with the EPR results, the violet emission has been assigned to Zn
to VBM and the 445-nm emission with fringes has been assigned to the excited Zn
to VBM. Regarding the DLE bands, the green band has been assigned to CBM to deep level acceptors whereas the yellow band has been assigned to Zn
to deep level acceptors. Chen et al.[54]and Yao et al.[55] have investigated the optical signatures of the native defects in N-doped films and nano-rods annealed or fabricated at high temperatures (950 °C–1000 °C). The dominant shallow donor has been identified as Zn
related defects while the V
related defects are responsible for the green luminescence and the famous DAP at 3.23 eV. Lautenschlaeger et al.[56] have shown that the DAP at 3.23 eV is an optical signature for N-doped ZnO. Unfortunately, they have not specified the identification of the donor nor acceptor responsible for the DAP.[54,55] Based on Chen and Yao et al.’s viewpoints,[54,55] the 3.23-eV DAP might originate from the transition involving Zn
related shallow donors as initial state and V
related shallow acceptors as final state. The doublet structure with an energy separation of 
meV observed in the green band has been ascribed to the ground-state, and the excited-state Zn
defects stemmed from the deep isolated V
acceptors. The schematic energetic diagram for various near band edge emissions(NBEs) and green band emissions have been summarized and shown in Fig. 15.
Another important issue should be stated here that two emission bands from 3.31 eV–3.32 eV and 3.33 eV–3.35 eV, which have been frequently observed and widely ascribed to be the eA
and A
X related to extrinsic acceptors previously, may have different origins. Schirra et al. have evidently observed that the 3.31 eV luminescence is emitted from the [
101] basal plane faults as shown in Fig. 16.[44] The places where the NBE is dark has very strong 3.31-eV emission. Although the origin of the peak is still an eA
, the holes are forced to localize around the faults and thus have no contribution to p-type conduction. Wagner et al. have investigated the origin of the Y band (from 3.33 eV to 3.35 eV).[57] Figure 17 shows the monochromatic CL mapping for the TES (I
) and Y
luminescence at a structural defects, indicating that the Y
line has a different origin from TES (I
) and thus I
. As I
is a D
X related to H or Al, the Y
line cannot be related to extrinsic atoms. The origin of the lines falling on the Y band could be excitons bound to structural defects (D
X), but be not necessarily the excitons bound to deep acceptors (A
X).
As the Zn
related defects are shallow donors and have very low formation energies in p-type doped samples, it is of great significance to suppress the formation of these defects. Li et al. have observed that the intensity of Raman mode at 275 cm
decreases and increases with increasing Mg and Cd alloying content, suggesting the tuning effects of Mg and Cd on the concentration of Zn
related defects.[58]
Figure 18 shows their first-principle calculated results, giving an applausive support to their experimental findings. Moreover, the electron concentration variation follows the trend of Zn
concentration, which again demonstrates that the Zn
related defects are the origin of n-type conductivity in ZnO. Tang et al. have found that Te co-doping is beneficial to the suppression of Zn
related defects.[59] A comparison of the Raman spectra recorded from a batch of N mono-doped samples with those from a batch of Te–N co-doped samples as shown in Fig. 19 reveals that the Raman mode at 275 cm
is much weaker in Te–N co-doped case, indicating a suppression effect of Te on Zn
. Furthermore, the concentrations of Zn
and V
related defects can be tuned by post-annealing temperature, which gives a possible way to control these native defects. Besides, one should avoid reducing ambience during the fabrication and processing of p-type doping, which might be advantageous for suppressing the Zn
related donors.
2.4. Other native defects and defect complexesAs reviewed from Fig. 1, the antisite Zn
also has a shallow donor characteristic with the 2+ to 1+/0 transition energy 50 meV below the CBM. Moreover, the formation energy is even lower than V
when the Fermi level is very close to VBM under Zn-rich condition, implying that the Zn
could be a potential compensating source in p-type ZnO. However, the formation energy of Zn
close to CBM is 
eV higher than that of V
under Zn-rich condition, indicating an indecisive role in providing electrons in n-type ZnO. However, like Zn
, the Zn
might be formed and donate free electrons in high-energy processes and non-equilibrium conditions. Look et al.,[59] Seager and Myers,[60] and Grossner et al.[61] have ever found very shallow donor levels in Zn-rich samples, and have ascribed the shallow donors to H, Al, and Zn
. In fact, the donor levels could also originate from Zn
. Meanwhile, the formation energy of Zn
close to VBM is higher than those of Zn
, V
, and even V
under O-rich condition, indicating that ZnO cannot be a dominant compensating defect in p-type ZnO since the chemical ambience is basically O-rich for p-type doping. For O
, the formation energy for neutral state is stable throughout the whole Fermi level range with a high formation energy of 7.23 eV. The O
shows the highest formation energy under p-type condition, but it holds the second-lowest formation energy compared with V
under n-type condition, and thus might act as a compensating center in n-type sample. Tuomisto et al. have identified a second acceptor other than V
in their electron-irradiated ZnO samples, and have assigned the acceptor to O
or O
.[63] In general, due to the high formation energies and deep energy levels of the O
and O
, they must be electrically inactive in ZnO material.
Figure 20 shows the formation energies for some intrinsic defect complexes under Zn-rich condition.[30] Except the O
+ Zn
antisite complex with extremely high formation energy, all the other complexes are donor-like. The Zn
+ V
complex could be viewed as a Zn
antisite with an off-site displacement of Zn atom from V
. Therefore, the complex has four energetic transition states and comparable formation energy similar to Zn
but a much deeper 0/1+ transition energy. The binding energy for the Zn
+ V
complex has been calculated to be a negative value, indicating that the complex is unstable. However, it is suggested that the attractive interaction between Zn
and V
stabilizes the complex.[19] All the possible Schottky and Frenkel pairs (V
+ V
, V
+ Zn
, and V
+ O
) are deep donors located at 1 eV∼2 eV above the VBM as shown in Fig. 20. The binding energies for these pairs are 0.8 eV to 0.95 eV at the VBM, also indicative of relatively unstable configuration.
According to the calculation from Figs. 1 and 20, Vidya et al. have suggested possible origins for the DLE in ZnO.[30] The red/orange emission at 1.9 eV observed in Zn-rich sample could be attributed to the ZnO since the 4+/3+ transition occurs at 1.9 eV below the CBM. The orange emission (2.1 eV) could be attributed to the 1+/0 transitions of V
+ V
and V
+ Zn
. The (V
+ V
) complex from 2+ to 1+ state could be contributed to green luminescence in addition to V
or V
alone.
2.5. A brief summary to the native defects in ZnOGenerally, native defects are undesired, which have destroyed the stoichiometry of a compound material. Therefore, provided the full understanding of the properties for the defects, it is preferred that the defects could be suppressed as thoroughly as possible. Due to the p-type doping asymmetry, the mission in exploring p-type ZnO, in particular, requires the suppression of donor-like defects and the enhancement of the acceptor-like defects. In summary, we list the main physical properties of the point native defects in ZnO in Table 1.
Table 1.
Table 1.
 | Table 1.
A brief summary to the physical nature of the native point defects in ZnO.
. |
As reviewed above, the V
exhibits the lowest formation energy among the donor-like native defects. Due to its deep donor level in the band gap, it is unlikely to contribute electron carriers in the n-type case. However, in the p-type case, the V
is a major source to provide compensation for acceptors. For Zn
and Zn
, they might not only provide electrons in the n-type case but also compensate for acceptors in the p-type case due to their shallow donor characteristics. The isolated Zn
is quite mobile and diffusive, indicative of instability at room and elevated temperatures. However, the Zn
has been considered to be stable with the help of N
, V
, and self-clustering. The stability of Zn
has not been studied. If some mechanism could stabilize the Zn
, the role of Zn
could be similar to Zn
. It has been suggested that growth and process of p-type ZnO with oxidizing ambience, rather than reducing ambience, is favorable for the suppression of the compensation donors. Moreover, the incorporation of isovalent atoms, like Cd, Mg, and Te, has been found to influence the formation of the Zn
related donors. It should be emphasized here that the Zn
and, possibly, Zn
defects and their relavent complexes must be suppressed to a great extent before p-type ZnO with low compensation level can be realized. Therefore, more efforts are required to investigate this issue.
Provided that the extrinsic p-type doping in ZnO has given no substantial progress, the idea that the native acceptor-like defects are uaed to solve the p-type problem has thus been anticipated, and people have great expectations of V
related defects, which is representative and has the lowest formation energy among the acceptor-like native defects (V
, O
, and O
). On one hand, the transition energy level of V
is still controversial, and the 0/– transition might be shallow. On the other hand, some mechanisms supports that V
can be stabilized by forming 2V
–X
, 3V
–X
, (
, As, Sb), V
–N
, V
–H
, nV
–V
, and V
cluster complexes, which might have a shallower acceptor ionization energy level in the p-type doped case. As a result, the design and experimental realization of existing and novel V
related acceptor-like complexes could be a future hot topic for ZnO. The electrical properties for the Schottky and Frenkel pairs are all donor-like with deep transition levels in the band gap. Thus, some of the DLEs could be attributed to the pair complexes. Moreover, it is suggested the binding within the pairsis weak, indicating that they are not stable and inactive electrically.
3. p-Type doping in ZnO: Theory and experiment3.1. TheoryThe direct and simplest way to realize p-type doping in ZnO is to incorporate the elements from group-IA, IB, and VA. At first, theoretical calculation based on tight-binding theory[64] and density functional theory (DFT) with local density approximation (LDA) or generalized gradient approximation (GGA)[65,66] has shown that the substitutional defects (IA at zinc or VA at oxygen) can indeed act as shallow acceptors in ZnO. Moreover, the complex 2V
–
(
elements), with being regarded as the origin of free holes, has been calculated to have even lower formation energy than the X
.[67] However, the prediction is obviously over-optimistic according to the experimental facts that measurable p-type conduction is extremely difficult. The self-interaction error and the underestimation of the band gap have caused the unreliable values of the dopant/defect energy levels.[68] Recently, a corrected calculation method based on the spin-polarized Kohn–Sham theory with hybrid functionals of Heyd, Scuseria, and Ernzerhof (HSE) has been developed.[69] Utilizing this method, the band gap has been corrected and the acceptor levels have been determined to be much deeper.
Lyons et al.[68] and Tarun et al.[70] have calculated the energy level of nitrogen on oxygen site (N
) and have concluded that the N
is actually an acceptor as deep as 1.3 eV above the valence band maximum (VBM). Petretto and Bruneval have conducted a comprehensive ab initio study regarding the doping of group-VA elements in ZnO.[10] Figures 21 and 22 show the calculated formation energies of the substitutional and complex defects for X (
, P, As, and Sb), respectively. As can be seen, the acceptors [
, 
–2V
, and (N
] are all very deep with the 0/– transition energy larger than 1.5 eV. However, the ionization energy of the (N
complex is somehow controversial. Lambrecht and Boonchun[71] have utilized first-principles calculations to predict that the (N
complex is a shallow double acceptor and could be identified experimentally by the donor-acceptor-pair recombination. Further experimental evidence for the shallow nature of (N
has been reported by Ton-That et al.[12] Furthermore, when Fermi level is close to VBM, the formation energies of the complex donors [
–V
and (N
become lower, resulting in severe compensation. A similar conclusion has been drawn by Cui and Bruneval that the (N
would energetically prevail over the N
and N–D–N cluster acceptors (
, Al, and Ga), although the (B–N
) complex has been thought to be shallow (0.18 eV).[72]
Another complex involving N
and V
(N
–V
) has been proposed by Liu et al. to have low formation energy during ZnO growth on Zn-polar surface.[13] The complex is thought to evolve from the metastable N
–V
double donor. However, the evolvement faces a barrier of 
eV to overcome. Amini et al. have recently conducted another calculation of this complex. They have concluded that the N
–V
acceptor could be in the form of H–N.
–V
, where H plays an important role in lowering the formation energy of the N
–V.
complex as shown in Fig. 23.[14] However, the additional H atom also occupies the shallower hole level of the N
–V
complex, leaving only two empty acceptor states deep in the band gap (
eV as shown in Fig. 23) making the H–N
–V
complex a deep acceptor. Moreover, Zhang et al. have proposed another possible form of shallow acceptor in Zn-rich-grown ZnO films, which is Zn
–2N
.[73] The introduction of the complex theoretically can form an impurity band above the VBM, resulting in a decrease in the acceptor ionization energy and an improvement in the stability of p-type nitrogen-doped ZnO film.
Besides the substitutional and antisite configuration for the group-V elements, Puchala and Morgan have identified stable interstitial-vacancy complexes in the form of 
–3V
(
, As, or Sb) in ZnO.[15] However, these complex centers are calculated to be too deep to create high concentration of delocalized holes. Very recently, Bang et al. have found that ammonia molecule substituted for Zn has extremely low formation energy as compared with N
–V
, V
, and (N
acceptors.[11] Figure 24 shows the calculated formation energy for various Zn-site and O-site N-related defects. Under O-rich ambience, the formation energy of (NH
is 7 eV lower than that of the N
–V
. More importantly, by capturing a interstitial H, the formed (NH
is a shallow acceptor, which might be the origin of holes in NH
-doped ZnO samples.
The calculations of group-IA elements are also not positive. McCluskey et al.[26] and Vidya et al.[74] have concluded from their calculation that Li
, Cu
, and N
are deep acceptors, and compared with other acceptors, the Li
acceptor is relatively shallow as shown in Fig. 25. Besides, the Li
–Li
[75] and Li
–(OH
[76] complexes are energetically favored and will critically compensate for the Li
acceptors. Only in very rigorous O-rich ambience, could the formation energy of the Li
–Li
cluster acceptor be slightly lower than the passivated complex Li
–Li
as shown in Fig. 26.[74] Considering that Li atom is very diffusive and mobile, the mono-doping of Li to solve the p-type doping problem in ZnO may not be the correct direction.
Despite the fact that the mono-doping of IA and VA elements is discouraging, various co-doping methods are still promising. Duan et al.[77] and Huda et al.[78] have revisited the acceptor–donor–acceptor (ADA) co-doping mechanism, and have proposed that the electrically passivated acceptor–donor pairs can change the VBM characteristic by inducing fully occupied impurity bands with the N 2p character above the VBM. The further doping of the acceptor in ZnO would lower the acceptor level as compared with the isolated substitutional acceptors. Besides the ADA co-doping, an acceptor-isovalence (AI) co-doping technique has attracted much more attention. The alloys of S, Se, and Te in ZnO could significantly elevate the VBM energy, and hence make the acceptors much shallower.
Figure 27 depicts the band diagram of the ZnO
S
alloys calculated by Persson et al., in which the N
acceptor energy becomes smaller when x increases from 0 to 0.19.[23] Hu and Chen have calculated the N–S co-doped ZnO system and have found that the 3N–S cluster can enhance the nitrogen solubility and make the acceptor level shallower.[79] Except for isovalent elements on oxygen, the ones on zinc are also beneficial to enhancing p-type doping. Li et al.[80] and Tang et al.[24] have shown that the acceptor transition energies of N
-
Zn
(
–4) and V
–O
can be reduced provided Zn being replaced by isovalent Mg or Be and O being replaced by F due to the reduced anion-cation kinetic p–d repulsion. Table 2 shows the calculated transition energies for various complexes. As more Mg or Be elements substitute for Zn elements, the transition energy could further decrease. The 3
kind of the co-doping scheme is the dual-acceptor (AA) co-doping. Calculations of Li–N and K–N co-doped ZnO system have been conducted by Gai et al.[81] and Wu et al.,[82] and the results demonstrated the beneficial effect of AA co-doping. The dominant acceptor has been proposed to be Li
–N
–Li.
, where the N
functions as the passivator to Li
, and bonds with the effective acceptor Li
. The complex is stable and shallow (0.12 eV above VBM). The results of K–N also provides a shallow acceptor level at 
eV. To sum up, the calculations on the co-doped ZnO seem to give a bright future, however, experimental realization of the designed complex and/or cluster acceptors will be quite challenging due to (i) the incorporation of more elements would lead to a critical degradation of the crystalline quality and a more complicated material system, (ii) the growth technology to induce the desired acceptors is difficult to master, and (iii) the dopants may exist in various undesired forms. Therefore, meticulous experimental work is required for verifying the calculations.
Table 2.
Table 2.
| Table 2.
Calculated binding energy of the defect complex and transition energy levels in ZnO.[80]
. |
3.2. Experiment3.2.1. Mono-doping approachWithin the mono-element doping approach to p-type conduction, the group-IA, IB, and VA elements have been indefatigably studied. Table 3 and 4.[83–122] summarizes the results regarding the mono-doping attempts with group-I and V elements in past five years. Generally, more groups have transferred their efforts from thin films to nano-structures. The results have shown that the nano-structured ZnO is relatively easy to realize the p-type conduction. However, as can be seen from the tables, the relatively high resistivity and extremely low hole mobility have not been substantially improved, which indicates that the p-type material is still highly compensated for by relatively poor crystalline quality.
Table 3.
Table 3.
| Table 3.
Hall data for p-type ZnO mono-doped by group-I elements.
. |
Table 4.
Table 4.
| Table 4.
Hall data for p-type ZnO mono-doped by group-I elements.
. |
For the scheme of mono-doping group-I elements, p-type conduction has been observed by doping with Ag, K, Li, and Na through using various growth methods as shown in Table 3. As mentioned above, the doping of group-I elements suffers serious compensation from undesired form of dopants due to the diffusivities of the elements. Tay et al. have studied the K-doped ZnO grown by a low-temperature aqueous solution technique.[85] Figure 28 shows the schematic diagram of the potassium doping mechanism. They have found that in K-poor condition, the K
–H
is the main passivator to K
, while in K-rich condition, the K
and K
–K
complex would act as the acceptor-killers. The conclusion would be applicable to other group-I dopants and other growth techniques, and therefore, the p-type conduction can only be observed through minimizing the amounts of the K
and K
–K
complex, while the hole concentration is critically limited by the existence of the K
–H
complex. For the Li-doping approach, Lee et al. have synthesized vertically oriented p-type Li-doped ZnO nanowires by using a simple hydrothermal technique with lithium nitrate serving as a doping source.[86] Their results have shown that lithium in the as-grown sample can occupy the empty cages of the wurtzite structure at octahedral sites, and a thermal annealing is required to assist the substitution of Li at Zn sites.
Figure 29 depicts the output and transfer properties of a single-nanowire Li-doped ZnO FET, in which the p-type conduction of the nanowire has been undoubtedly proven. For the Na-doping approach, the group from Zhejiang University in China have done a series of researches on Na-doping in ZnO.[91–100] They have found that non-polar faces are beneficial to the realization of p-type conduction and thus their researches mainly use the non-polar substrates.[95] Furthermore, some novel attempts have been made. Liu et al. have utilized a periodic delta-doping to realize Na-doped p-type ZnO.[94] He et al. have realized a coaxial growth of the Na-doped ZnMgO shell material surrounding an n-type ZnO core by tuning their PLD parameters with low chamber pressure.[92] The p-typeness has been verified by characterizing the single-wire FET as shown in Fig. 30.
The interplay between V
and Na has been studied and the V
has been concluded to facilitate the substitution of Na acceptors with inhibiting interstitial Na donors.[123] Considering the fact that the emission intensity of p-type ZnO is not so good as un-doped and n-type doped ZnO, Chen et al. have deposited Pt nanoparticles on Na-doped p-type ZnO films. By utilizing the coupling between the surface plasmon resonance and the exciton emission. The PL intensity can be enhanced by 10 times.[99]
For the scheme of mono-doping with group-V elements, nitrogen and antimony are two popularly investigated elements as shown in Table 4.[101–122] For N-doping approach, Chavillon et al. have reported that the ammonolysis below 350 °C of ZnO
can produce pure wurtzite N-doped ZnO nanoparticles with up to 20% Zn vacancies.[106] Figure 31 shows the phase diagram reporting the compositions of all the samples as well as the electrochemical and photoelectrochemical characterizations of the ZnO nanoparticle samples prepared at different solution temperatures, suggesting that Zn-poor samples are generally of p-type. They have concluded that the combination of a high concentration of zinc vacancies plus the insertion of nitrogen and the coexistence of oxide and peroxide groups can lead to the stabilization of positive charge carriers. The Zn-poor precursors have also been employed by Herring et al.[108] They have utilized a facile microwave irradiation method to rapidly heat the oleylamine solution with ZnO
/zinc acetate and urea precursors and have produced stable p-type N-doped ZnO nanostructures. Figure 32 shows the electrochemical measurement on the undoped and N-doped ZnO nanostructures, giving clear p-type characters. The inset shows that the nanostructures can be in the form of spherical nanoparticles, nanorods, and nanoprisms by varying the dose of the zinc precursors. Since the substituting nitrogen is not shallow, these researches suggest that zinc vacancies play an important role in the p-type nature of N-doped ZnO material. Furthermore, the undesired forms of nitrogen will critically reduce the efficiency of nitrogen doping and compensate for the acceptors. Hoffmann and Pettenkofer have studied the chemical nature of N-incorporated ZnO film,[124] and they have found that the N
and NO molecules at oxygen sites are two kinds of compensating complexes. Gao et al. have found that H
and V
can passivate N
within the whole typical annealing temperature range.[125] As a result, the compensation must be greatly suppressed before high-quality p-type conduction can be achieved.
Unlike N-doping, due to the large radius of Sb ion, Sb tends to occupy Zn site. Recent papers discussing Sb-doped ZnO have all assigned the Sb
–2V
complex to the nature of p-type conduction.[126] Yankovich et al. have synthesized the Sb-doped ZnO nanowires from a zinc acetate ethanol solution.[114] From the Z-contrast scanning TEM observations, they have found that all the Sb ions in the nanowires are incorporated into head-to-head basal plane inversion domain boundaries (H–H b-IDBs) as shown in Fig. 33. The calculations have suggested that the extra basal plane of oxygen per H–H b-IDBs makes them acceptors. Kang et al. have grown Sb-doped ZnO/n-ZnO nanowire p–n junction.[115] They have observed stacking faults in the Sb-doped ZnO region as shown in Fig. 34. They have thus assigned the p-type conduction to the easy formation of Sb
–2V
complex by combining Sb
with zinc vacancies formed near the stacking faults. Despite of various reports on p-type Sb-doped ZnO, Liu et al. have pointed out a possible donor behavior of Sb in ZnO.[127] They have concluded that Sb ion predominantly occupies Zn position and acts as a donor in ZnO film with low Sb concentration. At Sb content 
at.%, the Sb
acceptor compensating defects, rather than the Sb
–2V
complexes, are formed. It reminds us that the real acceptor in Sb-doped ZnO is still not defined. Although no big advancement has been achieved in terms of the p-type performances of group-I and V-doped ZnO as shown in Tables 3 and 4, most of the device applications (LEDs, LDs, and PDs) have been realized by p-type N- or Sb-doped ZnO or ZnMgO,[7,9
128] which proves the feasibility of doping ZnO p-type via incorporating N and Sb.
3.2.2. Co-doping approachDue to the fact that the p-type properties of the simply mono-doped ZnO film are not satisfactory, various co-doping approaches have been proposed and experimentally implemented to enhance the p-type parameters. In Table 5 listed are the recent Hall results on the co-doped ZnO material, which have been classified as three categories, i.e., ADA, AA, and AI co-doping schemes.[22,25,129–165] The ADA approach follows the theoretical concept that solubility of the acceptor element can be enhanced by forming an A–D–A complex.[166] Some recent novel modifications to the typical ADA scheme might be interesting and enlightening. Balakrishnan et al. have introduced two acceptor elements (As and N) and have co-incorporated them with aluminum into the ZnO samples fabricated by RF magnetron sputtering.[136] A low resistive p-type film (
cm) has been fabricated. The low resistance has been ascribed to the combined effect from As
–2V
and Al
–2N
. Li et al. have investigated the p-type conduction realized in In–N co-doped ZnCdO thin films.[138] Figure 35 shows their basic findings that the p-type stability is directly related to i) the Raman intensity of the mode at 275 cm
and ii) the band gap. Indicated by these experiments and a theoretical calculation, the formation of passive (–Cd
–O
–In
–N
–) complex can lead to an impurity band above the VBM, and thus result in a decrease in the ionization energy of the N
acceptor beside the complex. Huang et al. have developed a simple sol-gel spin-coating technology to fabricate In–N co-doped p-type ZnO films on n-GaN.[139] Figure 36 depicts the schematic diagram for the technique. The In–N co-doping was realized by the diffusion from InN layer plus a post-annealing in NH
ambience. The resulting p-type ZnO generally had a mobility from 100 cm
/V
s∼200 cm
/V
s. Senthil Kumar et al. have employed an Li–Ni co-doping technique and successfully fabricated stable p-type conduction.[140] The Ni
ions have been found to be favorabble for increasing the lithium solubility. They have also suggested that not only Ni, but also other transitional metal ions (Mn, Co) may have a similar effect.
Table 5.
Table 5.
| Table 5.
Hall data for co-doped p-type ZnO.
. |
The tentative idea of the AA co-doping approach is to enhance the acceptor concentration as compared with mono-acceptor doping. Actually, there are some results showing the enhancement of hole concentration via dual acceptor co-doping. Duan et al. have directly compared the electrical properties of an Ag-doped, an N-doped, and an Ag–N co-doped ZnO samples. The hole concentration has increased by 1∼2 orders of magnitudes by co-doping.[141] However, the mechanism of AA co-doping may not be that easy. A possible mechanism is that the p-typeness can be enhanced by the elevated VBM. In the Li–N co-doping scheme, Zhang et al. have demonstrated that the Li
–N.
donor-acceptor complex can be formed, leading to an impurity band above the VBM, and thus can result in a decrease in the ionization energy of the acceptor.[143] The formation of the complex can also suppress the formation of the compensating (N
. Figure 37 shows the plots of the calculated DOSs versus energy for un-doped and Li–N co-doped ZnO, showing the clear elevation of the VBM. Similar reason has been applied to the explaination of the enhanced p-type conductivity in P–N-co-doped ZnO. Sui et al.[147] have demonstrated that in the P–N co-doped ZnO system, a neutral passive P
–3N
complex can be easily formed, which may induce an additional fully occupied impurity band above the VBM. Regarding the assignment of the real acceptor, in the scheme of Li–N co-doping, the Li
is regarded as an active acceptor, and the effect of nitrogen is actually to passivate the compensating interstitial Li by forming the above-mentioned Li
–N
complex.[144] In the scheme of N–P co-doping, the P
–4N
is regarded as a complex shallow acceptor.[147]
The AI co-doping approach has been widely and popularly investigated in recent years as shown in Table 5.[22,25,149–165] Briefly, the isovalent dopants include Be, Mg, S, Se, and Te. Various experiments have been designed to verify the theoretical predictions as well as to study the mechanism of the co-doping technologies. For Mg, Akasaka et al. have successfully realized an MgZnO film with a low residual donor concentration.[22] By appropriately tuning the growth parameters, especially the Zn/O ratio during the ZnO MBE growth, the background donor can be controlled to be as low as 
cm
. The resulting MgZnO film is ready as the starting material for p-type doping. The enhanced p-type conduction has been observed by Cao et al. in their Mg-Ag co-doped ZnO films. The alloyed Mg can have some effects on the increase of the activation energy of the intrinsic donors and the suppression of the oxygen-related defects.[149,150]
For Be, a little work has been done by Chen et al.[155] The Be–N co-doping technique has been studied in detail. Figure 38 shows the comparison of the SIMS depth profile between N-mono-doped and Be–N co-doped ZnO films, indicating clearly the effect of Be on stabilizing N incorporation.[155] They have recently demonstrated a p–i–n structured stable UV LED by utilizing a Be–N-co-doped ZnO film as hole injecting layer.[159] Zhu et al. have grown p-type Be-doped ZnO films with 3% of Be, and it have been found that V
intrinsic defects can be greatly suppressed by incorporating a small amount of Be into ZnO, and the resulting p-type conduction has been thus ascribed to the elimination of V
.[156] Further co-doping with nitrogen has resulted in an enhancement of the hole density. Besides the N stabilizing effect, the enhancement has also been partially ascribed to the much shallower acceptor ionization energy of the nBe–N
complex. More interestingly, the calculated ionization energy of nBe–N
is even smaller than that of nMg–N
, making Be a perfect co-doping element for N–ZnO. Chen et al. have demonstrated that the p-type window can be widened by co-doping with Be.[157] For S, its co-doping with Ag, Cu, and P has been investigated. An enhanced p-type conduction has been observed by Sun et al.[150] and Xu et al.[152] The XPS of Ag 3d line has demonstrated the existence of both Ag–O and Ag–S bonds as shown in Fig. 39, suggesting the formation of Ag
–nS
complex. They have also calculated the ionization energy for the Ag
–nS
complex, showing that the ionization energy decreases with increasing n. As a result, the mechanism of S co-doping might be similar to those of Be and Mg. By incorporating heavy doping of Cu into ZnOS alloy, the Cu acceptor level has artificially expanded to a band and formed a ‘degenerated’ p-type material, and the hole concentration can reach 
cm
.[154] For Te, Park et al.[161] and Tang et al.[25,162] have realized p-type conduction in Te–N-co-doped ZnO samples.
Figures 40 and 41 show the basic electrical properties. The hole concentration is in a range of 10
cm
–10
cm
. Based on the investigation, the Te–N co-doping technique has the following beneficial effects: (I) the solubility of nitrogen can be enhanced; (II) the acceptor ionization energy might be smaller; (III) donor-like defects, like interstitial zinc and carbon-related impurities could be suppressed; (IV) the surface smoothness could be improved due to the possible surfactant effect from tellurium. The biggest obstacle for Te–N co-doping is that the optical quality has degraded a lot, which requires more efforts to solve the issue.
As shown by the above progress of p-type doping in ZnO, the difficuly of long-standing p-type doping is far from being ultimately solved. However, the solution is still expected in the future. At this stage, there is no unified conclusion on what is the best doping strategy because the well-acknowledged milestone progresses of various doping approaches has been made, like mono-Sb,[9] mono-N,[6] Mg–N,[7] Li–N,[8,167] etc. More efforts should be devoted by summarizing the past failures and improving the existing success. One thing should be mentioned here is that the Hall-effect measurement with using van der Pauw’s configuration has been repeatedly proven to be as a problematic way to determine the p-type conductivity in ZnO.[168] However, this characterization technique is still the most popularly used way, which, as before, yields unbelievable high hole concentration. Besides, the rectified I–V curve and the very weak light emission from the alleged ‘p–n’ junctions should be carefully explained. It might be due to the formation of Schottky or MIS junctions. The electroluminescence might be from the high-field induced optical transition. In order to eliminate the fake p-typeness, it is suggested that more characterizations, as partially highlighted in this section, like the 
–B curve, C–V measurement, photoelectrochemical method, Seeback-effect, and device verifications (PD, FET, LED, LD, etc), should be presented for a credible report of a p-type realization.
4. Conclusions and perspectivesIn Sections 2 and 3, the recent research progress of the native defects and the p-type doping of ZnO is reviewed briefly. We show that the theoretical investigation on the native defects has been refined by employing hybrid functionals. The band gap underestimation can be thus corrected. Utilizing the very accurate calculation methodology, various defects including simple point defects and complicated complexes have been studied extensively. Their formation energy as well as respective transition energy between different valences has been determined. Relying on these results, experimentalists are able to design and modify the growth and post-growth process conditions to deliberately modulate the formation of a specific defect. Also, they could use the calculated ionization energy as a reference to experimental identification of the defects. For the realization of high efficient p-type doping, Zn
and V
related intrinsic defects and their combined complexes are the most important compensating donors, some of which are very obstinate to suppress because of their high stabilities. The methods to suppress them are under investigation. Generally, tuning the growth condition, employing proper post-growth treatment, and alloying with some iso-valent elements are three chief means to prevent the compensating donors from forming.
Regarding p-type doping attempts, searching for shallow acceptors is still an uncompleted task. The mainstream theories have shown that shallow acceptors in group-VA-doped ZnO could be in a complex form. The complex involves intrinsic defect, mainly V
, and extrinsic dopants, like nitrogen, phosphorus, antimony, etc. On the contrary, group-IA dopants could be shallow, but the small atomic radii make them favorably stay as interstitials. The interstitials must be passivated before prominent p-type nature being observed. Traditional acceptor-like dopants in ZnO alloyed by iso-valent elements, like magnesium, beryllium, sulfur, selenium, and tellurium, may have shallower transition energy levels. Considering the fact that the isovalent elements can help elevate the energy of either CBM or VBM, compensation could be reduced while the ionization of acceptors could be much easier. No matter what the approach is, the main mission we are facing in a near future is to investigate the introduction, characterization, and control of the desired acceptors with true shallow transition energy levels. Like the McCluskey et al.’s viewpoint,[26] we believe that the study of the p-type doping in isovalent element alloyed ZnO may be the right path to ultimately solving the p-type hurdle.
