Effects of plasmon resonance on the low frequency noise and optoelectronic properties of Au/Cu codoped ZnO based photodetectors

We have investigated the effect of Au nanoparticles on the low frequency noise and current-voltage of Au/Cu doped ZnO nanoparticles (Nps) based photodetectors (PDs). Besides, 1/f noise of Cu doped ZnO is studied with and without Au Nps conditions and varied bias voltages. The Au/Cu codoped ZnO Nps showed much lower 1/f noise and higher normalized photocurrent to the dark current ratio (NPDR) than Cu doped ZnO Nps and undoped ZnO Nps. This enhancement may be attributed to the effect of localized surface Plasmon resonances (LSPR). Furthermore, the noise equivalent power represents the minimum optical power that a photodetector can distinguish from the noise and the detectivity (D*) is an important parameter in evaluating the ability of a photodetector to detect a weak signal. These parameters were determined using low frequency noise measurements. The observed low frequency noise performances could be used for Au/Cu codoped ZnO Nps based UV photodetectors.


Introduction
Internet of Things (IoT) and real time personal health care systems have been widely developed as a means of technology for human-computer interaction interfaces (Zumuukhorol et al. 2017;Glemža et al. 2017). In general, optoelectronic detectors, which convince this technology, are necessary components with large applications in photonic devices (Benali et al. 2019;Shewale et al. 2015;Khokhra et al. 2017). Also, for most of these applications, P-N junction and Schottky diode are used as sensors. Schottky junctions must have a Schottky barrier height (SBH) as high as possible, low leakage current, good thermal stability, and a low level of 1/f noise. 1/f noise measurements could reveal the mobility fluctuation phenomenon and the occurrence of the charge trapping/de-trapping in PDs (Zhu et al. 2016;Mukherjee et al. 2014). This type of noise directly affects Schottky contact PDs . Furthermore, the electronic noise is very important because we can determined the main parameters to evaluate the performance of PDs such as the noise equivalent power (NEP) and the detectivity (D*). Consequently, the role of 1/f noise needs to be studied to find more realistic performance information for PDs based on semiconductor junctions.
For UV PD application, ZnO Nps could be promising candidates owing to their unique properties (Choudhary and Choudhary 2018;. However, ZnO nanostructures suffer from fast recombination of excited electrons and holes, which may obstruct the general electrical performance of the PDs (Joshna et al. 2019). Much research has focused on doping ZnO with transition metal to facilitate charge separation. However, doped ZnO based photodetectors suffer from low photoresponse mainly associated with the low optical absorbance and higher noise current (Young et al. 2018;Chang et al. 2013). To overcome the above mentioned issues, noble metal nanoparticles can also be decorated with ZnO Nps, which enhances the properties due to their localized surface plasmon resonance (LSPR) property such as transfer of plasmon resonance energy from plasmonic metal Nps to nanostructures, near-field enhancement, and large absorption of incident light (Landi et al. 2013). The advantages of using Au Nps to enhance the optical properties have been confirmed (Liu et al. 2019); however, their optimization in reducing noise current is lacking in the literature. In our work, low frequency noise measurements have been used to investigate the properties of PDs and prove the impact of Au Nps for the lower noise current and high photoresponse obtained for our device. Also, we determined the noise equivalent power (NEP) and detectivity which are critical parameters to evaluate the performance of Au/Cu codoped ZnO based PDs.
The structural and morphological characterizations of our samples were carried out by X-Ray diffraction (XRD) (a diffractometer using cobalt radiation of 1.7890 A°) and transmission electron microscopy (TEM) (a JEOL 2011 microscopy operating at 100 kV). Optical properties were analyzed by diffuse reflectance spectroscopy (DRS). The FTIR spectra in the range of 4000-400 cm −1 were obtained by the infrared spectrophotometer (Spectrum two FTIR spectrometers by Perkin Elmer).
The Cu doped ZnO Nps and the Au/Cu codoped ZnO were deposited on glass substrate using spin coating method (2000 rpm for 30 s). Then, an anode (Cu) and indium gallium (In/Ga) cathode were deposited. Then, the photoresponse of the device was investigated under UV light of wavelength 375 nm at different bias voltages by 200 SCS Keithley instrument. Figure 1a shows the TEM image of ZnO Nps. Figure 1b shows the TEM image of Cu doped ZnO Nps having agglomerated circular particles. The morphology of Au/Cu codoped ZnO is displayed in Fig. 1c. It is shown that Au nanoparticles, which are recognized from the ZnO nanoparticles with dark contrast, have been scattered on the granular surface of ZnO Nps with moderately uniform size (for example in average 12 nm) and distribution. The X-ray diffraction (XRD) profiles for undoped ZnO, Cu doped ZnO and Au/Cu codoped ZnO Nps are shown in Fig. 1d. The samples are indexed to the hexagonal Wurtzite structure ZnO (JPCDS No. 36-1451). In the XRD profiles, no peak of secondary phases was noted for all the samples. The X-ray diffraction data confirm the polycrystalline structure of ZnO Nps. obtained from the diffuse reflectance spectra were fitted with the Kubelka-Munk function (Rani et al. 2018) and are presented in Fig.2(b). The optical band gaps are determined to be 3.26, 3.2 eV and 3.24 eV for undoped ZnO, Cu doped ZnO and Au/Cu codoped ZnO, respectively.

Optics properties
For the diffuse reflectance spectra, the other significant features in the visible region are shown in Fig. 2. The absorption bands at 566, 611 and 660 nm are associated to the d-d crystal field transitions 4 A 2 (F)-> 2 A 1 (G), 4 A 2 (F)-> 4 T 1 (P) and 4 A 2 (F)-> 2 E 1 (G), respectively. Figure 3 displays FTIR spectra of undoped ZnO, Cu doped ZnO and Au/Cu codoped ZnO Nps. The bands between 424 and 428 cm −1 are attributed to the stretching vibration of Zn-O bond in tetrahedal coordination. The peaks situated at 671 and 672 cm −1 are associated to the stretching vibrations of Zn-O bonds octahedral arrangements. It is noted that the Zn-O bands on octahedral arrangements are much lower than the tetrahedral coordination. Also, FTIR spectra demonstrates the Wurtzite structure of our samples (Ahmad et al. 2018). The peak at 671 cm −1 highlights that co-doping does not influence the band associated to octahedral coordination. Consequently, Au/Cu ions are substituted only at tetrahedral coordination in the ZnO lattice structure. The asymmetric stretching vibrations of the C=O group owing to Lewis acidity were confirmed by the band at 1385 cm −1 . While, the peaks between 1615 and 1617 cm −1 are attributed to the symmetric stretching vibrations of the C=O group owing to Bronsted acidity.
A broad absorption peak at 3450 cm −1 is due to the hydroxyl(-OH) group attributed to the water molecules. Thus, the Au/Cu codoped ZnO Nps can be dispersed easily in polar and nonpolar solvents owing to surface hydroxyl groups. Figure 4 shows the current-voltage (I-V) curves for our device based on Cu doped ZnO Nps under dark current and light illumination. The obtained values for the dark and light current are 2 × 10 − 7 and 5.5 × 10 − 7 A, respectively. The highest photocurrent value at UV illumination is due to the high surface-to-volume ratio and high crystallinity (Kwon et al. 2017). From Fig.4  and Au/Cu codoped ZnO Nps, the barrier height was determined to be 0.58 eV and 0.61 eV, respectively. This increase in barrier height is attributed to the stronger improvement electric field of LSPR.

of I-V characteristics of Cu doped ZnO Nps
The photodetector performance can be evaluated using the normalized photocurrent to the dark current ratio (NPDR), which proves direct information about the noise equivalent power and photoresponsivity resulting from the dark current. The NPDR is expressed as the following equation (Landi et al. 2013): where I phot is the photocurrent, I dark is the dark current, P opt is the optical power of incident light and R is the photoresponsivity. Figure 5 shows the NPDR versus the applied bias for the Cu doped ZnO and Au Cu codoped ZnO Nps. The plots prove that the Au Nps play an important role in the enhancement of NPDR. The Au/Cu codoped ZnO showed higher NPDR than the other device, which could be directly correlated with the variation of dark current depending on the existence of Au Nps, as shown in Fig. 4a.

1/f noise characteristics
Noise control plays an essential role for both sensors and detectors. The noise in semiconductor devices is divided into four principal types; thermal noise, shot noise, generation recombination noise and 1/f noise. In general, 1/f noise is considered an important issue for various semiconductor devices. Also, it is important to investigate the low frequency noise to understand and probe the electronic transport in semiconductor devices. In Schottky contact, the importance of 1/f noise can be explicated by Hooge's model relatively than the Mc Whorter's model (Noothongkaew et al. 2018). Recently, based on Mc Whorter's model, the origin of 1/f noise is the trapping of charge carriers and it is attributed to the carrier's number fluctuations and the surface state in the diode. In fact, we studied the current noise power spectral density SI of undoped ZnO, Cu doped ZnO and Au/Cu codoped ZnO devices at bias magnitude ranging from 0.5 to 6 V, respectively. According to Hooge's empirical relation, 1/f noise can be calculated by the following equation   where α, I, f, β and γ are the noise amplitude, DC current, the frequency and constant exponents, respectively. Power low fits to the experimental data indicate that for undoped ZnO, γ = 1.12-1.18, for Cu doped ZnO, γ = 1.14-1.21 and for Au/Cu codoped ZnO, γ = 1.15-1.21. The γ values close to 1 confirmed the 1/f-type noise performance. The decrease of amplitude noise with the decrease of the bias voltage is explained by the numbers fluctuation theory, which is with regard to the Mc Whorter's model.
Next, we studied the 1/f noise of our devices with Au and without Au and extract a comparison for the noise amplitude. Figure 6 shows that the noise density of undoped ZnO (1.19E-18 A²/Hz) is obviously higher than that for the Cu doped ZnO (1.33E-21 A²/Hz) and Au/Cu codoped ZnO (4.38E-25 A²/Hz). Consequently, the Au/Cu codoped ZnO shows a much lower noise density as compared to the other samples. After decoration with Au Nps, there was an increase in the width of the depletion region and surface electric field owing to the Schottky barriers localized at the interface of Au and ZnO. Accordingly, these Schottky barriers brought about a smaller dark current and compelling partition of separation of photogenerated holes from electrons to accelerate the desorption of oxygen particles (see Fig. 7a). To better comprehend the surface plasmon coupling between Au Nps and S I = I f Fig. 6 Plots of current noises spectra as function frequency under different bias voltages for the samples (a-c), the noises spectra for the different samples (d)

Table 1
Comparison of the most important UV photodetector parameters between our work and others studies ZnO, an energy band diagram is plotted in Fig. 7b. The nanoparticles of Au adsorbed some broadband emissions of deep level defects into ZnO by surface plasmon resonance since the resonance energy of the Au Nps is like the emission energy of ZnO deep level defects. In fact, these emissions produced the surface plasmon resonance in Au Nps and significantly improved the electromagnetic field near the Au Nps. Consequently, large quantities of electrons in the Au Nps are excited and then transferred to the conduction band of ZnO. As a result, the Au Nps concentrate free space electromagnetic waves powerfully by localized surface Plasmon resonances (LSPR). Consequently, more carriers are excited with the aid of LSPR. So, at higher total carrier's number, the relative fluctuation of carrier's number is logical to be lowered.
The NEP is an important parameter for the evaluation of PDs ). Another important parameter for PD is the detectivity (Kumar et al. 2016). The calculated NEP and D * for Au/Cu codoped ZnO are 1.6 × 10 −14 W/H 1/2 and 3.6 × 10 12 Jones, respectively, which shows much lower NEP and much higher D * as compared to Cu doped ZnO (3.6 × 10 −10 W/H 1/2 and 3.6 × 10 9 Jones) This means that our proposed PD could be used for low light detection. Table 1 shows the collected data of our samples as compared to results found in the literature. The performances of our devices indicate the possibility of using Au/Cu codoped ZnO Nps to develop photodetectors that demonstrate improved performance using a simple, and low cost elaboration method.

Conclusion
We fabricated Cu doped ZnO and Au Cu codoped ZnO MSM PDs and studied the impact of Au nanoparticles on their low frequency noise and electrical properties. According to their above analysis, the NPDR of Au/Cu codoped ZnO PD showed a lower dark current than the other device. Therefore, the NEP for the codoped sample was 1.6 × 10 −14 W/ H 1/2 , which is much lower than the other devices. It was found that the key factors for the enhancement of the optoelectronic properties for the proposed device were the formation of the Schottky barrier at Au/ZnO interfaces, near field enhancement and transfer of Plasmon resonance energy from plasmonic Au Nps to ZnO. Our device could be used for imaging sensors in cell phone cameras as well as for PDs in pulse oximeters for low light detection.