Nanorod Arrays Enhanced UV Light Response of Mg-Doped ZnO Films

: Zn 1 − x Mg x O (x = 0, 0.03, 0.05, and 0.07) nanocrystalline ﬁlms were grown on silicon substrates using the sol–gel method. Furthermore, Zn 1 − x Mg x O vertically aligned hexagonal symmetrical nanorods with six reﬂection symmetries were fabricated on pure ZnO-seeded layer n-type silicon substrates via a low-temperature hydrothermal method to enhance the ultraviolet (UV) light response. The crystal microstructures and surface morphologies of nanocrystalline ﬁlms and nanorod arrays were determined by X-ray di ﬀ raction (XRD) and ﬁeld emission scanning electron microscopy (FE-SEM). Transmission spectra showed that the increasing Mg content will increase the band gap energy from to 3.46 eV. However, the current–voltage curves in the dark and under UV illumination showed that the UV response did not improve by the incorporation of magnesium. We changed the ﬂat surface of ﬁlms into symmetrical nanorod arrays and demonstrated they can signiﬁcantly enhance the normalized photo-to-dark-current ratio up to ten times.


Introduction
ZnO-based materials have attracted a lot of intensive research in recent decades due to their interesting material properties and wild applications. It is well known that they are abundant, non-toxic and low-cost semiconductors with a wide band gap and large exciton binding energy. They also exhibit piezoelectric and pyroelectric properties and a large number of nanostructures. By virtue of these advantages, various kinds of optoelectronic devices, such as thin-film transistors (TFTs) [1], solar cells [2], ultraviolet (UV) lasers [3], photodetectors [4], and surface acoustic wave (SAW) sensors [5] have been developed. For example, due to their large band gap energy (above 3.37 eV) and good electrical properties (high electron mobility and low resistance), ZnO-based materials have the great benefit to be applied in large-area TFT-LCD panels or transparent electronic circuits to increase the transparency. Because of the good optoelectronic properties in the near-UV range and their excellent chemical and thermal stabilities, they are promising materials for the next generation of short-wavelength light emitting diode (LED) and laser diodes [6]. Among ZnO-based materials, ZnMgO is a strong candidate for UV photodetectors due to its high absorption coefficient, tunable band gap energy (3.3-7.8 eV), and high radiation hardness [7,8]. Furthermore, various kinds of nanostructured ZnO-based materials, including nanodots, nanorods, nanowires and nanobelts have been realized by different methods and unambiguously demonstrate that they are probably the richest family of nanostructures among all materials [9]. Enhanced photoresponse for photodetectors and high sensitivity for high-precision pH sensors have been demonstrated using ZnO and ZnMgO nanorod arrays [10][11][12]. These unique nanostructures, not only possess high surface area and good photoresponsivity for optical detectors

Materials and Methods
The source materials for the growth of Zn 1−x Mg x O (x = 0, 0.03, 0.05, and 0.07) films were prepared with the precursors of zinc acetate dehydrate (CH 3 COO) 2 Zn·2H 2 O and (CH 3 COO) 2 Mg·4H 2 O. We used 2-Methoxyethanol as solvent to keep the concentration of metal ions at 0.05 M, with magnesium mole ratios of 0%, 3%, 5%, and 7%, respectively. We also added ethanolamine into the solution to obtain a stable precursor solution. After stirring at room temperature (300 K) for 3 h using a magnetic stirrer and letting the solution sit for 72 h, a transparent solution was obtained for the growth of Zn 1−x Mg x O films. We used n-type silicon wafers (resistance of 0.55-1.2 Ω) as substrates. A typical ultrasonic cleaning process with acetone and ethanol was used to remove possible contamination and particles; after that, the substrates were rinsed in deionized (DI) water and dried by nitrogen gas. Each layer was fabricated by dropping the sol-gel solution on the substrates which were mounted on a spin coater rotating in two steps: the first step was a spinning rate of 1000 rpm for 10 s and the second step was 3000 rpm for 30 s. After that, the samples were heated at 300 • C for 2 min to evaporate the solvent. We repeated the process ten times to grow the Zn 1−x Mg x O films. The last process was annealing the films at 600 • C for 6 h in a rapid thermal annealing (RTA) system. In order to enhance the UV light response, we grew the nanorods on silicon substrates with two layers of ZnO nanofilms as a seed layer using the hydrothermal method. We used hexamethylenetetramine as solvent and diluted with DI water to keep the concentration of metal ions at 0.02 M, with magnesium mole ratios of 0%, 3%, 5%, and 7%, respectively. The ZnO seed layer film was dipped into this solution for one hour and the temperature kept at 90 • C to grow the Zn 1−x Mg x O nanorod arrays.
The crystal microstructures of all the Zn 1−x Mg x O films and nanorod arrays were examined using an X-ray diffractometer (Rigaku D/max-2200) with CuKα radiation. The surface morphologies and cross-section images were analyzed using a cold-field emission scanning electron microscope (FE-SEM, Hitachis-4800) at 3.0 kV. A UV-VIS-NIR absorption spectrometer (Jasco V-670) was used to investigate the band gap energies of the Zn 1−x Mg x O films. For understanding the UV light response of all Zn 1−x Mg x O films and nanorod arrays, a Keithley 2400 sourcemeter equipped with a UV lamp (254 nm) was used to perform the current-voltage (I-V) measurements in the voltage range between −5 V and to 5 V under dark and UV illuminated conditions. Figure 1 shows the XRD patterns of the Zn 1−x Mg x O (x = 0, 0.03, 0.05, and 0.07) films grown on silicon substrates. We found only one main peak located at 2θ = 34.54 degree for undoped ZnO (x = 0) and it shifted toward 34.71 degree for x = 0.07, which indicated that all the Zn 1−x Mg x O films had a c-axis preferential orientation. It can be seen that the intensity of these peaks gradually increased as the Mg content increased, which indicates the enhancement of the crystallinity of the films. In addition, all Zn 1−x Mg x O films had the same wurtzite hexagonal crystal structure with the space group P63/mc (JCPDS 36-1451). The lattice constant of the c-axis can be calculated using Equation (1) [25].

Results and Discussion
where h, k, l are Miller indices and a, c are in plane and vertical lattice parameters, respectively, and d is the distance between two adjacent planes with the same set of Miller indices. The lattice constant of c-axis can be determined to be 5.1941 Å for x = 0 and decreases to 5.1695 Å for x = 0.07. This result may be understood by the fact that Mg ions incorporated into the ZnO lattice and substituted Zn ions because the ionic radii of Mg 2+ (0.72 Å) is smaller than that of Zn 2+ (0.74 Å) [26]. This result is similar to the report of Mg-doped ZnO thin films prepared by a modified Pechini method, even though they are polycrystalline films.
Symmetry 2020, 12, x FOR PEER REVIEW 3 of 11 of all Zn1−xMgxO films and nanorod arrays, a Keithley 2400 sourcemeter equipped with a UV lamp (254 nm) was used to perform the current-voltage (I-V) measurements in the voltage range between −5 V and to 5 V under dark and UV illuminated conditions. Figure 1 shows the XRD patterns of the Zn1−xMgxO (x = 0, 0.03, 0.05, and 0.07) films grown on silicon substrates. We found only one main peak located at 2θ = 34.54 degree for undoped ZnO (x = 0) and it shifted toward 34.71 degree for x = 0.07, which indicated that all the Zn1−xMgxO films had a c-axis preferential orientation. It can be seen that the intensity of these peaks gradually increased as the Mg content increased, which indicates the enhancement of the crystallinity of the films. In addition, all Zn1−xMgxO films had the same wurtzite hexagonal crystal structure with the space group P63/mc (JCPDS 36-1451). The lattice constant of the c-axis can be calculated using Equation (1) [25].

Results and Discussion
where h, k, l are Miller indices and a, c are in plane and vertical lattice parameters, respectively, and d is the distance between two adjacent planes with the same set of Miller indices. The lattice constant of c-axis can be determined to be 5.1941 Å for x = 0 and decreases to 5.1695 Å for x = 0.07. This result may be understood by the fact that Mg ions incorporated into the ZnO lattice and substituted Zn ions because the ionic radii of Mg 2+ (0.72 Å) is smaller than that of Zn 2+ (0.74 Å) [26]. This result is similar to the report of Mg-doped ZnO thin films prepared by a modified Pechini method, even though they are polycrystalline films. The full width at half maximum (FWHM) values extracted from the XRD (002) peaks and the average grain sizes as a function of Mg content are presented in Figure 2. The average grain size can be extracted using the Debey-Scherrer formula [25], D = kλ/βcosθ, where k is the Scherrer constant (k = 0.9), λ is the wavelength of the X-ray radiation, β is the FWHM in radians, and θ is the Bragg diffraction angle. The calculated average grain sizes results are 31.68, 31.34, 31.25, and 30.12 nm, for x = 0, 0.03, 0.05, 0.07, respectively. The grain size determined by the Debey-Scherrer formula is usually smaller than the size observed from their SEM images. Using SEM, one can see physical grain or particle images that may contain several crystalline domains and imperfect regions around the borders between grains. With XRD, the information about the gain size only comes from coherently scattering in the same diffracting volumes. The uncertainty around 0.5 nm is due to the uncertainty The full width at half maximum (FWHM) values extracted from the XRD (002) peaks and the average grain sizes as a function of Mg content are presented in Figure 2. The average grain size can be extracted using the Debey-Scherrer formula [25], D = kλ/βcosθ, where k is the Scherrer constant (k = 0.9), λ is the wavelength of the X-ray radiation, β is the FWHM in radians, and θ is the Bragg diffraction angle. The calculated average grain sizes results are 31.68, 31.34, 31.25, and 30.12 nm, for x = 0, 0.03, 0.05, 0.07, respectively. The grain size determined by the Debey-Scherrer formula is Symmetry 2020, 12, 1005 4 of 11 usually smaller than the size observed from their SEM images. Using SEM, one can see physical grain or particle images that may contain several crystalline domains and imperfect regions around the borders between grains. With XRD, the information about the gain size only comes from coherently scattering in the same diffracting volumes. The uncertainty around 0.5 nm is due to the uncertainty of peak position, FWHM, and angle resolution, but this approximation method still gives a reasonable reference. The average grain size decreases with Mg content which is likely due to the Mg-substitution resulting in a local lattice distortion and the formation of a small Mg doped ZnO nanoparticles. The similar results have been reported in manganese (Mn) and iron (Fe) doped ZnO by other investigators [27,28].
Symmetry 2020, 12, x FOR PEER REVIEW 4 of 11 of peak position, FWHM, and angle resolution, but this approximation method still gives a reasonable reference. The average grain size decreases with Mg content which is likely due to the Mgsubstitution resulting in a local lattice distortion and the formation of a small Mg doped ZnO nanoparticles. The similar results have been reported in manganese (Mn) and iron (Fe) doped ZnO by other investigators [27,28].  Figure 3, were counted to be around 280, 290, 300, and 310, and the diameter of each grain was estimated to get the averaged grain size, which decreased a little bit from 50 to 47 nm for x = 0 to 0.07. This 3-nm difference of grain size observed in SEM images is not obvious, but still gives a consistent trend with the XRD results.
In order to understand the Mg doping effect on the optical band gap energy, the optical transmittance spectra of Zn1−xMgxO films have been performed by using a UV-vis absorption spectrometer. Each spectrum was measured twice to check its repeatability. We found that each of the spectra measured at the second time matched well with the first one in a wavelength error below 0.2 nm (about 2 meV for the band gap energy of 3.28 eV). Figure 4a presents the optical transmittance spectra in the wavelength range of 300-800 nm. In the visible wavelength range from 400 to 700 nm, the spectra show the films are highly transparent with a high transmittance up to 90%. In the short wavelength region below 400 nm, a strong absorbance is observed. It is obvious to find that the cutoff wavelength with an intensive decrease in transmittance shifts toward short wavelength with the increase of Mg-doping concentration. This result implies the absorption edge of Zn1−xMgxO films shifts towards high photon energy direction (blue shift).
From the transmittance spectra, absorption coefficient (α) can be calculated following Bouguer  Figure 3, were counted to be around 280, 290, 300, and 310, and the diameter of each grain was estimated to get the averaged grain size, which decreased a little bit from 50 to 47 nm for x = 0 to 0.07. This 3-nm difference of grain size observed in SEM images is not obvious, but still gives a consistent trend with the XRD results.
In order to understand the Mg doping effect on the optical band gap energy, the optical transmittance spectra of Zn 1−x Mg x O films have been performed by using a UV-vis absorption spectrometer. Each spectrum was measured twice to check its repeatability. We found that each of the spectra measured at the second time matched well with the first one in a wavelength error below 0.2 nm (about 2 meV for the band gap energy of 3.28 eV). Figure 4a presents the optical transmittance spectra in the wavelength range of 300-800 nm. In the visible wavelength range from 400 to 700 nm, the spectra show the films are highly transparent with a high transmittance up to 90%. In the short wavelength region below 400 nm, a strong absorbance is observed. It is obvious to find that the cut-off wavelength with an intensive decrease in transmittance shifts toward short wavelength with the increase of Mg-doping concentration. This result implies the absorption edge of Zn 1−x Mg x O films shifts towards high photon energy direction (blue shift).
(QSE) which is highly pronounced only when the crystallite size is comparable to its Bohr exciton radius; and (3) the fact that MgO has a wider band gap energy (7.7 eV) than ZnO (3.2 eV). Since our samples were not degenerate n-type semiconductors and their grain sizes calculated from XRD data and SEM images were far beyond the quantum confinement regime, we believe that the shift in absorption edge and widening of the band gap can be attributed to the compositional incorporation and local strain in Zn1−xMgxO samples [30].    From the transmittance spectra, absorption coefficient (α) can be calculated following Bouguer-Lambert-Beer absorption law α = (1/t)ln(1/T), where t is the thickness of Zn 1−x Mg x O films and T are the optical transmittance data in function of photon energy [29]. The α over the threshold of fundamental absorption obeyed a relationship αhυ = A(hυ − Eg) n , where A is a function of refractive index and hole/electron effective masses, h is the Planck's constant, v is the photon frequency, Eg is the band gap energy, and n = 1/2 for direct band gap semiconductor. On the basis of the Tauc model in the high absorbance region, the optical band gap energy can be calculated using the intercept of (αhν) 2 versus photon energy (hν) plot as shown in Figure 4b. The calculated band gap energies of Zn 1−x Mg x O films are 3.28, 3.36, 3.41, and 3.46 eV, respectively, for x = 0 to 0.07. This result indicates that with increasing concentration of the dopant, there is a slight increase in the band gap energy. This behavior may be explained by several mechanisms such as (1) the Moss-Burstein band filling effect which originates from the lifting of Fermi level into the conduction band and usually observed in degenerate semiconductors with carrier concentration higher than 10 19 cm −3 ; (2) quantum size effect (QSE) which is highly pronounced only when the crystallite size is comparable to its Bohr exciton radius; and (3) the fact that MgO has a wider band gap energy (7.7 eV) than ZnO (3.2 eV). Since our samples were not degenerate n-type semiconductors and their grain sizes calculated from XRD data and SEM images were far beyond the quantum confinement regime, we believe that the shift in absorption edge and widening of the band gap can be attributed to the compositional incorporation and local strain in Zn 1−x Mg x O samples [30].
The I-V characteristics of Zn 1−x Mg x O films were measured separately under dark (dark current, I dark ) and UV illumination (photocurrent, I photo ) between applied bias range +5 V and −5 V, as shown in Figure 5. The measured dark currents (I dark ) were 4.21, 4.12, 4.02, and 3.80 µA while the UV illumination current (I photo ) were 5.86, 5.52, 5.12, and 4.84 µA (at +5 V bias) for Zn 1−x Mg x O films with x = 0 to 0.07, respectively. For photodetector applications, the UV response can be evaluated by the ratio of photo-to-dark-current (R CV ) defined as (I photo − I dark )/I dark . The measurement data and calculation results are listed in Table 1  The I-V characteristics of Zn1−xMgxO films were measured separately under dark (dark current, Idark) and UV illumination (photocurrent, Iphoto) between applied bias range +5 V and −5 V, as shown in Figure 5. The measured dark currents (Idark) were 4.21, 4.12, 4.02, and 3.80 μA while the UV illumination current (Iphoto) were 5.86, 5.52, 5.12, and 4.84 μA (at +5 V bias) for Zn1−xMgxO films with x = 0 to 0.07, respectively. For photodetector applications, the UV response can be evaluated by the ratio of photo-to-dark-current (RCV) defined as (Iphoto − Idark)/Idark. The measurement data and calculation results are listed in Table 1    In this study, we have grown Zn 1−x Mg x O nanorod arrays on pure n-type ZnO seed layer by hydrothermal method to enhance the UV light response. Figure 6 shows top-view and cross-sectional FE-SEM images of Zn 1−x Mg x O nanorod arrays. From the top-view images in Figure 6a-d, Zn 1−x Mg x O nanorod arrays have clearly hexagonal structure almost along the full length and good uniformity with diameters in the range between 50 to 110 nm. The cross-sectional images in Figure 6e-h show that well-aligned Zn 1−x Mg x O nanorod arrays were grown preferentially in the vertical direction, which had the same result as the XRD patterns with a c-axis preferential orientation. In addition, it can be observed from the cross-sectional view that when the Mg content x increases from 0 to 0.07, the length of the nanorod arrays increases from 1.27 to 2.23 µm. This result suggested that, compared with the pure ZnO nanorod arrays (x = 0), with increasing Mg content, the growth rate along the c-axis has been improved obviously.
We measured the dark and UV illuminated I-V characteristics of Zn 1−x Mg x O nanorod arrays between applied bias range +5 V and −5 V. The results are presented in Figure 7 dotted line for dark and solid line for UV illumination. The measured dark currents (I dark ) were 72.5, 69.6, 61.6, and 58.7 µA while the UV illumination current (I photo ) were 326, 283, 239, and 216 µA (at +5 V bias) for Zn 1−x Mg x O nanorod arrays with x = 0 to 0.07, respectively. These data and calculated values of (I photo − I dark )/I dark are also collected in Table 1. It is worth mentioning that the nanorod arrays structures highly enhance the UV response about ten times (from 27 to 287 for x = 0.05). We observed that the increase of Mg content did not continue improving the UV response, but the change in structure from films to nanorod arrays resulted in a great improvement in the UV response. We believe the nanorod arrays increased the efficiency of adsorption of incident light that struck on the nanorod arrays because multiple reflections created by the nanorod arrays. It can be imagined that as a ray of light struck a nanorod, a part of it may be reflected from the nanorod array's surface following Snell's Law, and the reflected light could easily be incident on another. Each nanorod acts as a light scattering center and enhances the multiple scattering for the UV light [32].

Conclusions
Zn1−xMgxO nanocrystalline films were successfully fabricated using a sol-gel process with x = 0, 0.03, 0.05, and 0.07 and their crystal structure and band gap energies have been studied. FE-SEM images and XRD data demonstrated the films were grown preferentially in the vertical direction with a uniform thickness and lattice constant of c-axis has been changed from 5.1941 Å for x = 0 to 5.1695 Å for x = 0.07. Compositional incorporation of Zn1−xMgxO films with x = 0 to 0.07 shifts the band gap energies from 3.28 to 3.46 eV. We also grew Zn1−xMgxO nanorod arrays with symmetrically hexagonal shape along the full length and good uniformity. The large enhancement of the normalized photo-todark-current ratio resulting from nanorod array structures demonstrates their potential in UV light detection applications.

Conclusions
Zn 1−x Mg x O nanocrystalline films were successfully fabricated using a sol-gel process with x = 0, 0.03, 0.05, and 0.07 and their crystal structure and band gap energies have been studied. FE-SEM images and XRD data demonstrated the films were grown preferentially in the vertical direction with a uniform thickness and lattice constant of c-axis has been changed from 5.1941 Å for x = 0 to 5.1695 Å for x = 0.07. Compositional incorporation of Zn 1−x Mg x O films with x = 0 to 0.07 shifts the band gap energies from 3.28 to 3.46 eV. We also grew Zn 1−x Mg x O nanorod arrays with symmetrically hexagonal shape along the full length and good uniformity. The large enhancement of the normalized photo-to-dark-current ratio resulting from nanorod array structures demonstrates their potential in UV light detection applications.