Simple Preparations for Plasmon-Enhanced Photodetectors

2.1 The preparation process

The WS₂ film was grown on sapphire substrate by chemical vapor deposition (CVD) method and transferred to Si/SiO₂ substrate by wet transfer method. The molecular configurations of WS₂ are shown in Figure 1a. Raman spectra (Figure 1b), PL spectrum (Figure 1c) and atomic force microscopy (AFM) files (Figure 1d) reveal monolayer feature of WS₂ film.

The 3D and cross-section view of the photodetector is shown in Figure 2a,b, respectively. The light incident normally from the top of the device. The fabrication was carried out by the following steps. First of all, a 200-nm-thick molybdenum (Mo) layer was deposited and patterned on WS₂ layer. Afterwards, the WS₂ was patterned (size: 30 × 100 μm) by photolithography and plasma etching to form active area. Finally, Au NPs solution was spun coating onto the channel and dried in air. Nearly spherical Au NPs can be seen in the low-resolution transmission electron microscopy (TEM) image (Figure 2c). The mean diameter of Au NPs is ~20 nm (Figure 2d) as shown in the statistical analysis of the TEM images. Typical low-resolution (Figure 2e) and high-resolution (Figure 2f) scanning electron microscope (SEM) images of the Au NPs are also presented. The Au NPs are well distributed on the top of the WS₂ film.

2.2 The performance of the WS₂-based photodetector

The drain-source current (IDS) under illumination at room temperature without Au NPs are shown in Figure 3a. It is concluded that IDSincrease as VDSincrease from 0 to 2 V. The irradiance power is 20.5 mW/cm² in all of these three wavelength (590, 740 and 850 nm). IDSdecrease with the increase of wavelength. The ratio of the on/off current (IDS/Idark) reached nearly 103under 590 nm light illumination. The responsivity can be calculated by

whereIPhis the photocurrent, P is the irradiance power. The responsivity at VDS=2Vis illustrated in Figure 3b. The responsivity decrease with the increase of the power which are typical for photodetectors [24, 25]. R reached 35 A/W at the wavelength of 590 nm, and reached 1.8 A/W at the wavelength of 850 nm when irradiance power is both 0.2 mW/cm². The performance of the photodetector decoated with Au NPs is shown in Figure 3c. The drain-source current are measured at the irradiance power of 20.5 mW/cm². The enhancedIDSalso reached the highest value at 590 nm and decreased with the increase of the wavelength (λ).

The current gain is defined as

where Ipeis the enhanced photocurrent of the photodetector. The current gain reveals the improvement of the device by Au NPs as shown in Figure 3d when P=0.2 mW/cm² and VDS=2V. The photoresponsivity was enhanced ~30 times and reached 1050 A/W when λ = 590 nm. The photoresponsivity was enhanced ~11 times and reached 55 A/W when λ = 740 nm. And the photoresponsivity was enhanced ~5 times at near infrared light (λ = 850 nm) and reached 8 A/W. In general, the switching behaviour, which reflect the response speed and high-frequency characteristic, is very important for photodetectors. The detectors also need to quickly refresh in some applications such as instant display. Figure 3e presents the switching behavior at near infrared light (λ = 850 nm) of the WS₂ photodetector. The photodetector shows a good repeatability during on-off cycles. Moreover, the on-off characteristic in a period is shown in Figure 3f. The rise and decay time are about 100 and 200 ms, respectively.

2.3 The theoretical mechanism

To explain the mechanism of the enhancement by Au NPs, we take finite-difference time-domain (FDTD) method to investigate the distribution of electric field of Au NPs. According to the Förster’s expression for energy W transferred from donor to acceptor [24, 25].

where Wdis the donor’s energy, fdωand kare the spectral function and wave vector of the source, σaωis the absorption of acceptor, D=q/r3is the coupling coefficient, and ris the distance. The performance could be changed by key parameters of Au NPs such as diameter d and distance between two particles s. In order to give an intuitive description, simplified models were built. The distance between the edges of two adjacent nanospheres, s, was fixed on 10 nm. The diameter d was fixed as 20 nm.

The resonant wavelength can be acquired by SPPs dispersion equation.

where k0is the vacuum wavevector, εdis the relative permittivity of dielectric, εmis the dielectric function of gold, which can be represented by Drude model. The mismatch of SPPs and incident wavevector can be compensated by the gold nanoparticles, which can be approximately presented by

where θ,λg,nare the horizontal angle of incident wave vector, the grating period, and an integer, respectively. From Eq. (5), we can see that the absorption wavelength depends only on the dimension of Au NPs.

The results for the illumination at λ =590, 740, and 850 nm are shown in Figures 4a–c, respectively. It is clear that the intense electromagnetic fields were introduced by LSPR of Au NPs. The electric field near the Au NPs was significantly enhanced and stronger than the rest region, revealing that electromagnetic energy was compactly confined by the Au NPs. The electromagnetic field was enhanced more significant when the illumination was under λ = 590 nm. The enhanced electric fields could excite the generation of the carriers in the WS₂ film, resulting a prominent photoresponse. The highest responsivity obtained at λ = 590 nm (Figure 3b,d) is consistent with the most intense LSPR at λ = 590 nm (Figure 4a). The generation and transportation of the electrons are shown in Figure 4d. The photons were absorbed by WS₂ film and the excited electrons were driven by the drain-source voltage. There are more electrons around the Au NPs as Figure 4d shows.

3.1 The preparation process

Figure 5a,b shows the schematic and optical image of our designed MoS₂ plasmonic photodetector by introducing Au NPs, respectively. The few-layered MoS₂ sheet was obtained using mechanical exfoliation method. Then, we transferred the exfoliated MoS₂ to the SiO₂/Si substrate which contacted with Au/Ni electrodes. Next, we fabricated the Au NPs on exfoliated MoS₂ sheet using magnetron sputtering technique. This facile method offers the convenience of without need of template compared with other reported MoS₂ plasmonic photodetectors.

Figure 6 shows the morphology of as-prepared materials was characterized using an AFM. As shown in Section A1 in Figure 6a, b, it indicates the thickness of exfoliated bare MoS₂ is just about 6 nm (about 10 folds). In comparison, the surface morphology of the MoS₂ after depositing with Au NPs by magnetron sputtering was shown in Figure 6c–f. It clearly exhibits the physical size and particle distribution of Au NPs can be easily tuned by sputtering technique. When the sputtering current increased from 30 mA (LPP1, Figure 6d) to 35 mA (HPP1, Figure 6e) with deposition period fixed to 1 s, we obtained the controllable Au NPs with lateral size increasing from ~3 to ~5 nm and vertical height increasing from ~5 to ~8 nm, respectively. For another, if we fixed the applied current to 30 mA and prolonged the deposit period to 2 s (LPP2, Figure 6f), the Au NPs maintain almost same physical size as LPP1 but the gap of adjacent deposited Au NPs sharply drops compared with LPP1.

3.2 The structure of the MoS₂-based photodetector

In order to investigate chemical composition of the prepared materials, the X-ray photoelectron spectroscopy (XPS) was employed. As shown in Figure 7, The peaks at 229.2 and 232.3 eV correspond to the doublet of Mo 3d_5/2 and Mo 3d_3/2, respectively. And the peaks of 226.3, 162.1 and 163.2 eV of the binding energy attach to the S 2s, S 2p_3/2 and S 2p_1/2, respectively [26, 27]. For Au 4f, the peak positions of 83.6 and 87.2 eV bind to the Au 4f_7/2 and Au 4f_5/f, indicating the Au NPs are directly introduced into the exfoliated MoS₂ sheet [28]. More importantly, the banding energies of Mo and S in Au decorated MoS₂ maintain the same values as that of bare MoS₂ sheet, indicating the introduction of Au NPs has non-influence on the crystal structures of exfoliated MoS₂ sheet.

Further, we used Raman spectroscopy of 532 nm laser to confirm the structural properties of the fabricated devices. For MoS₂, the difference of E¹_2g and A_1g, Δ, corresponding to in-plane and out of plane energy vibrations, is used to index the layer number of obtained MoS₂. Figure 8a shows the Δ of bulk MoS₂ and our exfoliated MoS₂ are 27.8 and 25.3 cm⁻¹, respectively, indicating that the thickness of bare MoS₂ is about 10 layers [29], which is highly consistent with the AFM results. After decorating Au NPs with MoS₂, it exhibits the Δ maintains nearly same as bare MoS₂ but the intensities obviously increase, shown in Figure 8b, c.

3.3 The performance of the MoS₂-based photodetector

The photoelectric performance of the fabricated photodetector was studied at room temperature, which applied a 980 nm laser source with controllable incident power. In order to produce the laser beam pulses, we combined an oscilloscope to the incident laser source. Figure 9a shows the photocurrent plots of photodetector, where the illumination power and bias voltage are 1.60 mW and 2 V, respectively. Obviously, the Au NPs/MoS₂ heterostructure-based photodetector exhibits an ultra-high photocurrent (8.6 nA) compared with that of bare MoS₂-based photodetector (0.59 nA).

Figure 9b shows the plots of photocurrent vs. applied bias voltage ranging from 0.1 to 15 V. We obtained a photocurrent up to ~480 nA, when the applied incident laser power and bias voltage were 7.50 mW and 15 V, yielding an improved responsibility of 64 mA/W. Moreover, the I-V plots indicate the photocurrent owns a good linear relationship with applied bias voltage, when the illumination intensities tuned from 0.85 to 7.5 mW. For another, Figure 9c shows the dependence plots of photocurrent (I_ph, nA) on laser power irradiation (P_in, μW). It is found that the photocurrent follows a nonlinear dependence to the incident power intensity, aP_inb, where a and b are constant for different bias voltage. For example, when the bias voltage are 15 V, the fitting a and b are 3.37 × 1o⁻² and 1.34, respectively.

With respect to the stability of the photodetector, we performed the extended duration photocurrent measurements by periodically switching the incident laser under illumination of 3.20 mW at bias of 5 V, and the periods of both on and off state are 5 s. Figure 9d shows the photocurrents over 500 circles of continuous operation, exhibiting a well stability. Moreover, in order to characterize the response time for detecting infrared wavelengths of our designed device, we applied an oscilloscope in the process of laser excitation to produce laser pulses with a duration of 10 ms. Figure 9e shows the time-response of Au decorated MoS₂-based device under illumination of 7.50 mW at bias of 15 V. It indicates that the rise time (t_rise) and fall time (t_fall) are 2.4 and 2.6 ms, respectively, which are significantly superb to that of other reported MoS₂ photodetectors.

3.4 The operational mechanism

In order to study the potential mechanism of our fabricated photodetector, we simulated the electric field distribution of Au decorated MoS₂ using finite element method in the infrared region. We assumed that the Au NPs with a gap of 6 nm and a diameter of 5 nm under illumination of a linearly polarized plane wave with an electric field amplitude of 1 Vm⁻¹ for the simulated model. And the thicknesses of MoS₂ sheet and incident laser wavelength were 6 and 980 nm, respectively. Figure 10a, b shows the cross-section of the simulated electric field distribution of Au NPs decorated MoS₂. Benefiting from the LSPR effect excited by the Au NPs, when the diameter matched to the incident wavelengths, the intensities of electric field at interfaces of air/Au/MoS₂, up to ~3.96 × 10⁵ V/m, are obviously higher than other districts. In comparison, interfaces of air/MoS₂ show a poor intensity of electric field, only about 6.72 × 10⁴ V/m. This tendency can be also observed in the recent researches of Au NPs guided MoS₂ sheets for photo-detection [30]. We further experimentally proved the absorption by using UV-visible-NIR spectroscopy analysis. Figure 10c shows the absorption spectrum of bare MoS₂ sheet Au NPs decorated MoS₂. The normalized absorptance plots indicate that the Au NPs/MoS₂ is obviously enhanced than that of bare MoS₂ ranging from 700 to 1600 nm.

The above results and discussion clearly unveil the introduction of Au NPs plays a key role in enhancing the light matter interactions of MoS₂ with infrared wavelengths. The significantly improved sensitivity of the fabricated photodetector could be attributed to the LSPR effect, shown in Figure 10d, induced by the periodically aligned Au NPs, resulting in obvious improvement of local electric field. When the incident infrared wavelengths highly confined by the deposited Au NPs, the local electric field at the interface of Au/MoS₂ is greatly improved by the surface plasmon waves. Firstly, the Au surface plasmons effectively excite a coupling effect at the interface of Au/MoS₂ for absorbing photons, yielding a much more photo-induced carriers to improve the photo sensing [30, 31]. Moreover, the additional local electric field generated by the LSPR effect of Au NPs accelerates the photogenerated carriers to separate for producing photocurrent [32, 33]. This facile method, tuning Au NPs by sputtering method to excite LSPR effect for fabricating the unique device structure, is expected to be practical applications in other 2D materials such as WS₂ and MoSe₂ [34, 35, 36], thus offers a new route on a variety of high-performance optoelectronic devices.