Role of native defects on the opto-electronic properties of p-type ZnO synthesized during the most straightforward method: only water

The influence of predominant native defects in forming ZnO with p-type conductivity is discussed in this work when the semiconductor is synthesized only in water. The semiconductor was prepared by dissolving a Zn-salt in deionized water at 80 °C. The powders were thermally treated at 400 °C in an air atmosphere to obtain well-defined crystalline ZnO. XRD, SEM, EDS, Raman spectroscopy, diffuse reflectance, photoluminescence, and Seebeck effect techniques were used to characterize the synthesized material. The results showed a well-crystalline semiconductor in wurtzite phase. The crystal-oriented growth was the (002) plane. The sample morphology was formed by highly ordered sticks-like. The optoelectronic characterization showed that the synthesized ZnO had a lower band gap than that reported in the literature. It was related to deep energy levels corresponding to oxygen interstitials as the predominant native defects. Raman, EPR, and photoluminescence spectra analysis corroborated the existence of native defects in the crystalline structure. The p-type conductivity of the sample was determined by Seebeck coefficient analysis. A synthesis reaction mechanism involving the formation of oxygen interstitials was proposed in this work. Understanding the effects of native defects in wide band gap semiconductors is necessary to design new materials for sensors or energy conversion applications.


Introduction
The optoelectronic properties of ZnO are sensitive even to small changes in the lattice parameter values. The parameters can change due to incorporating chemical elements in the crystalline structure of ZnO or by Zn and O vacancies formed during the synthesis or thermal treatment processes [1]. The changes in lattice parameters usually modify the surface defect states of the semiconductor, and they produce variations in optoelectronic properties of the semiconductor, depending on the deep or shallow nature of intrinsic defects [1,2]. The n-or p-type conductivity of ZnO could be a consequence of the predominant defects in the crystal lattice. The possible defects are zinc interstitials (Zn i ), oxygen anti-sites (O Zn ), zinc vacancies (V Zn ), oxygen vacancies (V O ), zinc anti-sites (Zn O ), and oxygen interstitials (O i ) [3].
Besides, the control of these defects, the lifetime of charge carriers, and conductivity type could expand the application opportunities of ZnO [4]. Finally, the study and understanding of the oxygen interstitials in the optoelectronic properties of the semiconductors open the possibility of designing materials for specific applications [5].
Nowadays, there are scientific interests in preparing the p-type ZnO in different morphologies like wires, nanotubes, nanorods, nanoflakes, and sunflowers [6][7][8][9]. The synthesis methods used for obtaining the desired morphologies are vapor-liquid-solid process, molecular beam epitaxy, sputtering, chemical bath deposition, precipitation, hydrolysis, pyrolysis, solvothermal, sol-gel, and, more recently, green chemistry [10][11][12][13][14][15][16][17]. Nevertheless, the challenge in all cases is still the control of native defects as donor (Z ni and V O ) or acceptor (V Zn and O i ) energy levels that define the conductivity type of ZnO [18,19]. Some post-synthesis treatments have been used to ensure the p-type conductivity of ZnO . However, those processes typically produce unstable semiconductors for practical applications. The formation of predominant native defects during the synthesis is likely the best procedure to obtain semiconductors with crystalline 1 3 183 Page 2 of 11 and optoelectronic stability, control of native defects, and conductivity type of the semiconductor [20,21]. The polar solvents determine the nucleation process, the formation of surface defects, and the optoelectronic characteristics of the semiconductor [22]. All the physicochemical parameters mentioned above play a crucial role in producing stable p-type ZnO.
This work shows the most facile preparation of p-type ZnO using only water as a polar solvent. The semiconductor was characterized by structural, morphological, and optoelectronic techniques. The results indicated that the intrinsic defects based on oxygen interstitials consolidated the p-type conductivity of the semiconductor.

Experimental
ZnO was synthesized in 20 ml of deionized water in a roundbottom flask connected to a recirculation system to avoid evaporation losses. 2 g of zinc nitrate hexahydrate was added to the water at 60 °C and under intense stirring. The temperature of the solution was increased to 80 °C and stirred at 350 rpm for 12 h. After that, the recirculation system was uninstalled to evaporate the liquid phase and obtain the ZnO powders. The powders were annealed in a tubular furnace (Thermo Scientific Lindberg/Blue M) in an air atmosphere at 400 °C for 2 h. The thermally treated sample of ZnO was labeled as ZnO (a) .

Characterization techniques
The ZnO powders were characterized by X-ray diffraction (XRD) in a diffractometer Rigaku Model DMAX 2200 using the monochromatic Cu K-alpha radiation (1.54 Å). A vertical goniometer with an angular range from 10 to 90° was used to record the complete X-ray diffraction pattern of the hexagonal ZnO wurtzite. The micrograph and chemical composition were acquired using scanning electron microscopy (SEM) coupled with energy-dispersive X-ray Spectroscopy (EDS). The microscope operated at an accelerating voltage of 25 kV. The diffuse reflectance spectrum was recorded in a UV-3600 Shimadzu Spectrometer from 250 to 2500 nm. The Raman analysis was performed in a modular micro-Raman system, coupled to a spectrophotometer Horiba iHR550 with an 1800 l/mm diffraction grating and a laser source of 515 nm, 50 mW. The photoluminescence spectrum was obtained in an LZ4-40U600 LED Engin spectrometer coupled to a 365/385/395/405/ UV LED module 50 W and a system of the notch and bandpass fluorescence image filters for the range 345-450 nm (Thorlabs). Electron paramagnetic resonance spectroscopy (EPR) characterization was performed in a Jeol JES-TE300 Spectrometer, operating in X-band, at 100 kHz modulation frequency, with a cylindrical cavity in TE011 mode. A standard thermoelectric system with JDV_DTI software was used to investigate the conductivity type of the sample. The temperature range was from 300 to 370 K. Figure 1 shows the XRD pattern of ZnO (a) (blue line). 11 diffraction peaks corresponding to the wurtzite structure were identified in this sample. The associated planes were: The lattice parameters (a and c) of ZnO (a) were calculated for the hexagonal structure as follows:

XRD results
Bragg law (Eq. 1) was used to calculate the interplanar distances according to the three characteristic diffraction peaks of ZnO (a) . Where n represents the diffraction order [23], λ is the wavelength, d is the interplanar distance, and θ is the angle formed between the incident X-rays and the scattering diffraction of the atomic planes [24,25]. Equation (2) was used to calculate the interplanar distance between parallel planes (h k l) concerning the lattice parameters a and c of the hexagonal system. Where h, k, l are the Miller indices.
The first-order diffraction approximation (n = 1) and Eqs.
The lattice parameter a was calculated as follows: Moreover, the value of the lattice parameter c was calculated from: the calculated values were a = 3.25 Å and c = 5.20 Å for the three principal diffraction peaks (100), (002), and (101) of ZnO (a) . The values are the same as those reported in the literature [29,30]. It shows the feasibility of preparing wurtzite phase ZnO using the simplest synthesis method based on only water instead of complex and more sophisticated techniques.
where D is the crystallites mean size, λ is the incident wavelength (1.54 Å), θ is the Bragg angle, β is the full width at half maximum (FWHM), and δ is the dislocation density.
D = 0.9 cos , Table 1 shows the calculated values for the (100), (002), and (101) planes. The highest dislocation density was observed in the (002) plane, corresponding to the smallest size of crystallites in the (100) and (101) planes. In this case, the crystallite size was likely controlled by the presence of O i .
The growth of the (002) plane can be explained by considering that the O i occupies shallow energy states at the position of V Zn in the lattice structure [32]. It does not necessarily imply changes in lattice parameters. However, intrinsic native defects significantly influence the growth and orientation of the crystallites, producing changes in the optoelectronic properties of the semiconductor. It is possible to obtain the p-type conductivity of ZnO without using extrinsic dopants.
The dislocation density calculated in the three principal planes suggests the existence of native defects in all directions. The dislocation density values ensure the formation of a stable wurtzite phase. Nevertheless, the (101) plane shows the lowest dislocation density (0.03 nm −2 ) in respect of the (100) and (002) planes (1.74 and 1.81 nm −2 , respectively). In this case, the oxygen vacancies (formed in the (101) plane)   Figure 2 shows the surface morphology of the ZnO (a) formed by interconnected micro sticks, forming dense porous layers much more extensive than 5 mm. The well-defined morphology of the ZnO powders was due to the use of water as a polar solvent and the thermal treatment.

Analysis of ZnO (a) by SEM and EDS
The chemical content of Zn and O in the sample was 39.7 at% and 60.3 at%, respectively. The stoichiometric composition of the sample was likely influenced by the free oxygen content dissolved in the deionized water or the thermal treatment effect under the air atmosphere [34]. The Zn/O ratio of the sample suggests the presence of V Zn or O i in the lattice structure. The thermal treatment may increase the O i density by modifying the oxygen content in the crystalline structure.

Feasible reaction mechanism for the synthesis of ZnO in water
The facile synthesis of ZnO (a) in deionized water and the formation of oxygen interstitials can be explained according to the following reaction mechanism: The synthesis involved the solvation of Zn 2+ and NO − 3 obtained from the dissolution of Zn NO 3 [37,38].
Oxygen anti-sites are likely generated due to the H + interstitials located between zinc and oxygen atoms during the phase formation [39]. Additionally, H 2 O , H + , and NO − 3 are released from the semiconductor when it is thermally treated at 400 °C in an air atmosphere. The vacancy sites ( H + ) are quickly occupied by oxygen originating the oxygen interstitials formation. Figure 3a shows the diffuse reflectance spectrum of the ZnO (a) sample. The results were used to calculate the optical band gap (E g ) according to the Kubelka-Munk relation for weak adsorption processes (Eq. 8) [40]. The results were compared with those obtained for ZnO (ref) (Fig. 3b).

Diffuse reflectance analysis of ZnO (a)
where R ∞ = R sample ∕R reference , R ∞ is the reflectance of a sample of infinite thickness, K is the molar absorption coefficient, and S is the dispersion coefficient.
The E g value was calculated using Eq. (9). The energy absorption corresponds to the electronic excitation from the valence band to the conduction band [41].
where α represents the semiconductor absorption coefficient, h is the photon energy, C 1 is a proportional constant, E g is the optical band gap, and n is a constant associated with the transition type [42]. There is a substantial difference between the samples because the E g of ZnO (a) was calculated as 3.08 eV (Fig. 3c), while 3.3 eV was obtained for commercial ZnO (Fig. 3d). The E g value of the synthesized semiconductor could likely be related to small changes between the valence band maximum (VBM) and conduction band minimum (CBM) due to predominant native defects.
The positions of the valence band maximum and the conduction band minimum were determined according to Eqs. (10 and 11) [43]: The results indicate that the ZnO sample synthesized in water has a lower electronegativity than the commercial sample. This is consistent with the predominant native defects in each semiconductor. The band gap edge variations should be investigated accordingly with new electrochemical applications based on hetero-structured materials, such as sensors or solar cells. Figure 4a shows the Raman diffraction peaks at 325. 20 (a) . The peaks at 325.20 and 377 cm −1 are vibrations corresponding to the symmetrical harmonic optical phonons indicated as A 1 high and E 2 high-E 2 low. The peak at 435.81 cm −1 indicates the non-polar mode E 2 (high), related to the presence of Zn and O in the lattice interface [44,45]. E 2 (high) also corresponds to vibrations occurring at the heavy sub-lattice of Zn, and E 2 (low) represents the vibrations involving oxygen atoms. The peak at 519.81 cm −1 also corresponds to E 2 (high), and it is associated with the optic phonon of oxygen atoms [46,47]. The peak at 563.92 cm −1 corresponds to A 1 (LO), related to the characteristic mode of the wurtzite ZnO [47] or the E 1 (LO) mode [48,49]. Peaks at 1102.26 and 1142.16 cm −1 are related to 2LO(A 1 ) modes, and they are characteristic harmonic phonons of ZnO [42]. The complete vibrations of the Brillouin zone are 2A 1 + 2E 1 + 2B 1 + 2E 2 [47].

Analysis by Raman spectroscopy
The emission at 563.92 cm −1 is associated with the presence of O i , where the widening of the diffracted peak is directly proportional to the O i concentration in the crystalline lattice. The E 2 (high) mode at 519.81 cm −1 could probably shift towards the red color region of the visible spectrum as the native defect concentration increases in the crystal. This effect was discussed elsewhere [49] when ZnO was doped with Sb 3+ .
On the other hand, TA+LO signals represent the positions of oxygen vacancies, and they are related to the diffraction peak appearing at 643.22 cm −1 in the Raman spectrum. But in this sample, this characteristic peak is negligible, indicating a low V O content. The native defect concentration percentage (C(V O )) was estimated according to Eq. (12), using the relative intensity of the diffraction peaks at 563.92 (I 563.92 ), 435.81, and 643.22 cm −1 .
The V O concentration in ZnO (a) was 27.9%. It also proves that the low E g value of the synthesized semiconductor cannot be attributed to native defects based on V O . Figure 4b shows a typical Raman spectrum of commercial ZnO as an n-type conductivity semiconductor.

Photoluminescence spectrum of ZnO (a)
The photoluminescence spectrum of ZnO (a) is shown in Fig. 5a. An asymmetric signal with a maximum peak at 535.39 nm (corresponding to the yellow-orange color range) was the characteristic optical print of the p-type ZnO (a) . The Origin PRO-2021 Software was used to analyze the photoluminescence spectrum. The Lorentzian deconvolution algorithm from 250 to 800 nm (Fig. 5b) was used to fit the experimental response. The deconvolution process of the photoluminescence spectrum provided information about native defects in the semiconductor. The photoluminescence spectrum was fit with a pure Lorentzian signal. It indicates the existence of physical or chemical phenomena that do not necessarily correspond to the crystalline structure of the semiconductor. In the Lorentzian signal, the native defects were not generated from external dopants but intrinsic  structural changes during the synthesis. The spectrum of the original signal and its Lorentzian fit are shown in black and blue lines. Six low-intensity signals were obtained from the deconvolution process of the photoluminescence spectrum. The signals with maximum emission peaks at 371 and 424 nm correspond to the purple color. It represents the energy range where the native defects V Zn and O i are located in acceptor sites at the band gap [50,51]. It is also possible to observe a signal with maximum luminescence in the green color at 490 nm. It is a characteristic emission of ZnO associated with the presence of V O [49], but in this case, the signal has low emission that likely reveals the V O to O i conversion when the sample was thermally treated. The emission peak at 535.39 nm corresponds to the yellow-orange range of the visible spectrum, and it is related to the formation of Oi [52][53][54][55][56][57]. The signal with a high-intensity emission at 565.46 nm is related to luminescence produced by V Zn [57]. In this case, the predominant presence of V Zn and O i could determine the p-type conductivity of the ZnO (a) . Additionally, the signal with the maximum emission peak at 654.71 nm corresponds to the presence of atomic oxygen [58]. Figure 5(c) shows the photoluminescence spectrum of ZnO (ref) . The emission was within the ultraviolet (UV) range. It is a typical response of n-type ZnO but very different from that of ZnO (a) , where the response of the synthesized semiconductor is predominantly due to native defects based on oxygen interstitials or zinc vacancies. Figure 6a shows the isotropic EPR spectrum of ZnO (a) . It was possible to identify a resonance signal with a g value of 1.95644. It is a characteristic signal defining the hexagonal wurtzite structure of ZnO and represents the presence of native defects (V O ) located close to the surface [59]. In addition, three V O signals with g-factor values of 1.96014, 1.91294, and 1.86491 were observed in this plot. All of these were associated with point defects located close to the surface of the material. In this case, the low definition of these signals likely indicates that these native defects were not predominant in ZnO (a) .
On the other hand, it is possible to observe well-defined resonant signals with g-factor values of 2.15113, 2.10085, and 2.00466. These signals are associated with shallow point defects located above the valence band based on V Zn [60]. These signals are also related to recombination processes between the valence and conduction bands [59]. The results indicate that shallow point defects of V Zn are predominant in ZnO (a) , which determines the p-type conductivity of this material. Figure 6b shows the EPR spectrum of ZnO (ref) . The characteristic resonant signal of hexagonal wurtzite ZnO with a g-factor of 1.95853 can be identified. Similarly, three other signals with g-factor values of 1.9637, 1.987868, and 2.00298 were identified in this plot. They are associated with the presence of V O , such as point defects located close to the surface of the material, and induce the n-type conductivity of commercial ZnO [61]. Even though the signals with g values of 1.987868 and 2.00298 were assigned to shallow V O defect states, the fact is that this kind of signal at a g value around 2.0 is still under scientific discussion. Figure 7a shows the Seebeck coefficient behavior used to identify the conductivity type of the ZnO (a) . The experiment was performed by applying a temperature variation (ΔT) between the tips and measuring the surface voltage change. The Seebeck coefficient of the sample was calculated as + 75 mV/K. It may be related to the formation of O i . The Seebeck coefficient value was obtained at + 0.1 K above the reference temperature (305.15 K), and it was associated with the band bending process caused by the thermal excitation of the intrinsic charge carriers (holes). The charge carriers were directly correlated with the oxygen interstitials because the Seebeck coefficient gradient decreased quickly until an asymptotical tendency for ΔT values higher than 1.5 K, indicating a fast depletion of free electrons due to the activation of intrinsic charge carriers by temperature effects. It is a characteristic response of a semiconductor with low free electrons, meaning a p-type ZnO. Figure 7b shows the Seebeck coefficient plot of the commercial ZnO. The coefficient was − 1 mV/K. This result corresponds to the n-type conductivity mainly controlled by predominant native defects based on V O .

Conclusions
During the synthesis in water, the predominant native defects formed p-type ZnO. They did not necessarily modify the lattice parameters but influenced the crystallite growth orientation and were observed as point defects close to the surface. Raman, photoluminescence, EPR, and diffuse reflectance analyses suggested that the conductivity was due to predominant native defects based on oxygen interstitials or zinc vacancies. It is necessary to continue research to understand the influence of the experimental conditions during the synthesis or the presence of native defects that could define the optoelectronic properties of ZnO. It would be more practical to prepare stable wide-band-gap semiconductors by controlling native defects instead of using ex situ doping procedures, as typically performed for this type of semiconductor.