The schematic structure of the MgZnO PD is shown in Fig. 1a. The MgZnO PD adopts a simple planar metal-semiconductor-metal (MSM) structure, including a 100 nm Al interdigital electrode, a MgZnO functional layer, and a glass substrate. A photograph of the MgZnO PD is shown in Fig. 1b, and the fabrication process of the MgZnO PD is shown in Fig. 1d. The SEM image of the device is shown in Fig. 1c. It can be found that the surface structure of the MgZnO functional layer is homogeneous without bubbles or other obvious defects. The EDS image of the MgZnO functional layer is shown in Fig. 1e. The elemental composition of MgZnO indicates that Mg content is successfully doped. The existence of the Mg dopant is consistent with the reported study, indicating that the atomic radii of Mg and Zn are close to each other, and the stable existence of the MgZnO structure can be realized by replacing Zn atoms with Mg atoms [16]. Figure 1f shows the typical I-t curve under UV illumination (200–390 nm) at 13.05 mW/cm2 and a bias of 5 V of the device. The rise and fall times of the MgZnO PD are measured as low as 0.4 s and 2.4 s, which are faster than the ZnO-based PD. Figure 1g shows the switching ratio of the device for 6 switching operation cycles. The device has a uniform response curve with a stable switching ratio of 4000.
Figure 2a shows the current-voltage (I-V) characteristics curves at light intensities of 5.3, 1.8, 1, and 0.5 mW/cm2, respectively. The I-V characterization curves indicate that the MgZnO PD can perceive the UV light at different intensities. The obvious symmetrical rectifying behavior indicates that the classic Schottky contacts between Al and MgZnO are achieved. The stabilized photo-response curves of the MgZnO PD at 5.325 mW/cm2 are shown in Fig. 2b, indicating the photo-response will be improved by enhanced bias. Figure 2c shows the MgZnO PD response curves under different bias voltages. The left vertical axis is the responsivity, and the right vertical axis is the photocurrent. The stable photocurrent response of MgZnO PD under 5.325 mW/cm2 illumination is shown in Fig. 2d, and the stable response over a long time is shown in the enlarged inset. The dark current of MgZnO PD is measured to only 10 pA under 5.325 mW/cm2 illumination. Thus, the low dark current indicates the device’s ability to detect weak UV light. The measured variation of photocurrent with light intensity and the fitting curve of photocurrent with light intensity for the MgZnO PD are shown in Fig. 2e. The photocurrent increases with the increased light intensity in the range of 0 to 5.325 mW/cm2, indicating that the MgZnO PD has good linearity over a wide range of light intensities. Figure 2f shows the responsivity change of the device with the light intensity illumination. The responsivity gradually decreases to a limiting value as the light intensity increases in the range of 0–2 mW/cm2. When the light intensity exceeds 2 mW/cm2, the responsivity hardly changes anymore, which means the photocurrent of the device saturates.
To investigate the Mg-doping concentration on the bandgap change of MgZnO to further tune the as-fabricated PD to work in the UVC band, the optical bandgap of MgZnO with different Mg-doping concentrations is analyzed and quantitatively calculated. The six MgZnO materials with different Mg-doping concentrations are named as follows: 0MgZnO (ZnO), 1%MgZnO, 5%MgZnO, 10%MgZnO, 20%MgZnO, and 30%MgZnO for Mg-doing concentrations of 0, 1%, 5%, 10%, 20%, and 30% in mass ratio, respectively.
Firstly, the spectra of MgZnO functional materials are analyzed in the UV-vis band (200–850 nm). Figure 3a shows the absorption spectra of ZnO, 20%MgZnO, and 30%MgZnO. The inset in the upper right corner of Fig. 3a shows the transmittance spectra of the three functional materials: 1%MgZnO, 5%MgZnO, and 10%MgZnO. As the Mg-doping concentration increases, both the transmittance and absorbance spectra are blue-shifted towards the short-wave direction. From the measured spectra, the Eg of MgZnO functional materials could be derived by the Tauc-plot method [17–19]:
$${\left(\alpha h\nu \right)}^{2}=A\left(h\nu -{E}_{g}\right)$$
Where \(\alpha\) is the absorption coefficient, \(A\) is a proportionality constant, h is Planck's constant. ν is photon frequency. Then, the curves of \({\left(\alpha h\nu \right)}^{2}\) versus \(h\nu\) are plotted in Fig. 3b and Fig. 3c. From those curves, the intersections of their tangent lines and the X-axis is Eg. As shown in Fig. 3b, Eg of MgZnO functional materials increases from 3.36 eV to 3.92 eV (corresponding cut-off wavelengths decrease from 369 nm to 316 nm) as the concentration of doped Mg content increases from 0 to 10%, which agrees with reported literature (MgZnO functional materials varies from 3.38 eV to 3.84 eV as the concentration of Mg increases from 0–40%) [16]. From the analytical results, the Eg of MgZnO functional materials doped with more than 1% Mg mass fraction increases significantly, indicating a substantial blue shift of the cut-off wavelength. An Mg-doping concentration of over 10% of the MgZnO functional material is studied to achieve a shorter cut-off wavelength of the MgZnO functional material. However, single-phase segregation occurs when the Mg concentration rises to 25% [20–23]. To achieve higher doping concentration (> 10%MgZnO) thermal treatment could be utilized. 20%MgZnO and 30%MgZnO are successfully synthesized based on a 95°C preheating process during spin-coating. It is found that the key to the transition from inorganic mixture-amorphous-crystalline is combining the effects of the preheating temperature above 90°C and the annealing treatment above 300°C on the structure of the MgZnO functional materials. Compared to the high temperature of 1200°C by laser treatment [24–26], a relatively low-temperature treatment process for synthesizing 20%MgZnO and 30%MgZnO functional materials is achieved. Figure 3c illustrates that the Eg of the as-synthesized 20%MgZnO and 30%MgZnO is 4.24 eV and 4.27 eV. To be noticed, the Eg of the 20%MgZnO and 30%MgZnO is very close, which indicates that the limit of doping caused the blue-shift. From the slope of the 30%MgZnO curve in Fig. 3c, the cut-off wavelength of the 30%MgZnO is around 280 nm, blue-shifted to around 280 nm, reaching the UVC range.
Figure 3d shows the measured Eg of six MgZnO functional materials doped with different Mg concentrations and the simulated Eg based on Silvaco. Clearly, the experimental values match the simulated Eg well. The horizontal trend of the curve at Mg-doping is larger than 30%, which indicates that the Eg tuned by the doped Mg is reaching the limit. For 30%MgZnO, the Eg (4.27 eV, corresponding to 285nm) is suitable for UVC detection. Figure 3e shows the mechanism of the doped Mg on increasing the Eg of MgZnO functional materials. According to the reported Moss-Burstein shift [27, 28, 28–30], when the element Mg is doped, the MgZnO functional materials show the change of the Fermi energy level towards the conduction band, which causes an increase in the measured Eg. The Fermi energy level increases with increasing Mg-doping concentration. The more the Mg-doping concentration increases, the more the Fermi energy level shifts. Figure 3f-3h shows the working mechanism of MgZnO PD. As shown in Fig. 3f, Φm is the energy required for the electron to jump to the conduction band, and ΦAl is the energy needed for the electron to reach the vacuum energy level. Ec, Ef, and Ev are the positions of the conduction band, the Fermi energy level, and the valence band, respectively. The MgZnO materials form two crystalline states, w-MgZnO, and c-MgZnO, which are similar to ZnO and MgO, respectively. The traveling of thermal-induced free electrons in the MgZnO materials is hindered by crystalline barriers. The electrons can hardly flow through the interface state, which effectively reduces the dark current. Figure 3g shows that under an applied bias voltage, a few electrons jump through the potential barrier and flow across the interfacial state to produce a dark current with a total electron current density of Jn. Under an external bias voltage and UV illumination, the photon energy excites many electrons to jump across the potential barrier and form an induced photocurrent with a total electron current density of Jp, as shown in Fig. 3h.
To verify that the Eg of MgZnO materials could be modulated by Mg-doping concentration, UV PDs built from 0%MgZnO, 1%MgZnO, 5%MgZnO, 10%MgZnO, 20%MgZnO, and 30%MgZnO materials are tested. As shown in Fig. 4a, the responsivity of MgZnO PDs changes with the wavelength of UV illumination. The simulated responsivity curves for three typical 1%MgZnO, 5%MgZnO, and 10%MgZnO PDs are well agreed with the measured results. For all PDs, as the incident light wavelength increases, the responsivity of the devices decreases. There are cut-off wavelengths for the devices with different Mg-doping concentrations, and the cut-off wavelength decreases with the increase of Mg-doping concentration. Clearly, the cut-off wavelengths of 1%MgZnO, 5%MgZnO, and 10%MgZnO PDs change from 336 nm to 316 nm, which is in good agreement with the calculated Eg (Fig. 3d). Specially, from Fig. 4a, the 1%MgZnO PD achieves a measured responsivity of up to 108 mA/W under UVC illumination (254 nm) and a bias of 5 V. The suppression ratio of UVC (254 nm) and non-UVC (310, 365, and 395 nm) signals (UVC responsivity vs. non-UVC responsivity) can be as high as 50 times. The inset of Fig. 4a shows that when the Mg-doping concentration increases from 1–10%, the responsivity of non-UVC decays gradually.
The responsivity of MgZnO PDs changes with Mg-doping concentration is further verified as shown in Fig. 4b. For 0%-30%MgZnO PDs, when under 254nm UV illumination, the responsivity first increases then decreases with Mg-doping concentration, and 1%MgZnO PD has the highest responsivity. Under 310nm, 365nm, and 395nm illumination, the responsivity of the PDs all decreases and keeps a constant small value for all PDs. To see the doping-caused wavelength cut-off effect, the PDs with 1%MgZnO, 5%MgZnO, and 10%MgZnO are tested in 254nm and 310nm UV illuminations. As in Fig. 4c, for 254nm illumination, the PDs with 1%, 5%, and 10% Mg-doping concentration all have high responsivity, which means 254 nm is within the devices’ UV detection spectra. Under 310nm illumination, 1%MgZnO PD has a large responsivity of 2 mA/W, while 5%MgZnO and 10%MgZnO PDs have a much smaller responsivity of 0.2 mA/W, which means the 310 nm beyond 5%MgZnO’s and 10%MgZnO’s UV detection range.
Figure 4d shows the responsivity of the devices with three different Mg-doping concentrations under 365nm and 395nm illumination. Clearly, the responsivity of the devices under the two non-UVC illuminations (365, 395 nm) is low because the wavelength cut-off values of those Mg-doped devices are shorter than the illumination wavelengths. This further confirms that the MgZnO functional materials fabricated from the low-temperature sol-gel method could effectively turn cut-off wavelengths to form UVC detectors.