Plasmonic hollow gold nanoparticles induced high-performance Bi₂S₃ nanoribbon photodetector

1 Introduction

Currently, noble metal nanoparticles (e.g. Au, Ag, and Pt) have been extensively studied in a variety of field due to their localized surface plasmon resonance (LSPR) effect, which is actually caused by the collective oscillations of the excited free electrons upon light illumination with a certain wavelength [1], [2], [3]. LSPR could couple with the incident light, leading to more light trapped nearby semiconductor in plasmonic optoelectronic devices [4], [5]. This mode of manipulating light through engineering noble metal nanoparticles on semiconductors has been exploited in a number of applications, such as photocatalysis [6], [7], plasmonic nanoantenna [8], [9], solar cells [10], [11], light emitting diodes [12], and photodetectors [13, 14]. Among these light-harvesting devices, high performance photodetectors enabled by LSPR are receiving special research interest for their great potential in environment monitoring, biological analysis, industrial control, space-to-space communications and so on [15], [16], [17]. For instance, Sakurai et al developed a high performance Au nanoparticles@ZnO nanowire photodetectors. It was observed that the dark current decrease by 100 times and the on/off ratio was considerably increased after decoration of plasmonic Au nanoparticles [18]. Moreover, Hu’s group observed that a large enhancement of photocurrent response is obtained by decorating multi-layer MoS₂ with plasmonic gold nanoparticles array. Due to LSPR, the photocurrent of MoS₂ phototransistors exhibited a threefold enhancement after depositing periodic Au nanoparticles arrays [19].

One dimensional (1-D) semiconductor nanostructures have demonstrated great potential in wide ranging device applications [20], [21], [22], [23]. Bismuth sulfide (Bi₂S₃), as a typical semiconductor with a direct bandgap of 1.3 eV and a high absorption coefficient, has attracted great attention in hydrogen storage, electronic devices, and optoelectronic devices [24], [25], [26]. For instance, Konstanatos et al. reported a solution-processed Bi₂S₃ nanocrystalline photodetector with the responsitivity of about 40 A/W and a high detectivity on the order of 10¹¹ Jones [21]. Li et al. fabricated a visible light photodetector based on hierarchical Bi₂S₃ nanostructures. The photodetector showed stable photoresponse with a fast response time of 50 ms [27]. Despite these efforts, however, the device performance of the majority of the 1-D Bi₂S₃ nanostructure based photodetector ever reported was limited by relatively low photoresponsivity and external quantum efficiency, which is related to the extremely small absorption cross-section of the Bi₂S₃[28], [29]. Enlightened by the obvious positive role the plasmonic nanoparticles played in optoelectronic device, we herein reported a hollow gold nanoparticles (HGNs) induced high-performance Bi₂S₃ NRs photodetectors. It was found that the intensity of the light absorption increased after the NRs surface was modified with HGNs. Furthermore, an improved device performance with enhanced responsivity, detectivity and photoconductive gain was achieved. Theoretical simulation based finite element method (FEM) shows that the enhancement in devices performance is due to the strong LSPR effect of the HGNs.

2 Structural analysis of the Bi₂S₃ NRs

The 1-D Bi₂S₃ nanomaterial was synthesized by a modified solvothermal approach [30]. Both Figure 1(a) and b show the typical field-emission scanning electron microscope (FESEM) images of the as-synthesized Bi₂S₃ nanostructures. It is obvious that the product is composed of fiber-like nanostructures with aspect ratio (namely the length/diameter ratio) as high as 1000. The nanostructures are up to hundreds of micrometers in length and displayed high purity without appreciable contaminants, which is highly favorable for subsequent device construction. Figure 1(c) depicts the X-ray diffraction (XRD) pattern acquired from the product. All the strong diffraction peak in the range from 15–70° can be readily ascribed to orthorhombic lattices of Bi₂S₃ (JCPDS Card number: 17-320), in agreement with literature result [31]. Figure 2d shows the perspective view of the orthorhombic Bi₂S₃, in which the orange and cyan balls denote the bismuth and sulfur atoms, respectively. Further energy dispersive spectrometer (EDS) analysis shown in Figure 1e revealed the presence of B and S elements with atomic ratio of 1.9:3.1, in consistence with the stoichiometric ratio of Bi₂S₃ and the X-ray photoelectron spectroscopy analysis (Figure 1f).

Fig. 1

(a) Representative SEM image of the Bi₂S3 NRs. (b) SEM image of the Bi₂S3 NRs at large magnification. (c) XRD pattern of the Bi₂S3 NRs. (d) Schematic illustration of the crystalstructure of the Bi₂S3. (e) EDX spectrum of Bi₂S₃ NRs. (f) XPS spectrum analysis of the Bi₂S₃ NRs.

Fig. 2

(a) TEM image of an individual Bi₂S₃ NR. (b) Corresponding HRTEM image. (c) and (d) elemental mapping of both S and Bi.

The microstructure of the as-prepared Bi₂S₃ NRs was then analyzed by both transmission electron microscopy (TEM) and high-resolution TEM. Figure 2a shows a typical TEM image of an individual Bi₂S₃ NR, in which the NR has a very smooth surface. The HRTEM image in Figure 2b exhibits a lattice-spacing of 0.38 nm, which corresponds to the d-spacing of (101) plane. Further analysis of the HRTEM image as well as the selected area electron diffraction (SAED) pattern reveals a preferential growth direction along [001], which agrees with previous report [31]. The EDS elemental mapping shows that the constituting elements (S and Bi atoms) were uniformed distributed in the NR (Figure 2c and d).

3 Results and discussion

Next, nano-photodetector based on individual Bi₂S₃ NR as shown in Figure 3(a) was fabricated to study its potential for device application. Figure 3(b) plots the typical I – V curves of an individual NR based photodetectors in dark and under 560 nm light illumination. It can be found that the I – V curves are linear at both conditions, suggesting that good contact (Ohmic contact) was formed at the In electrode/Bi₂S₃ NRs interface. When 560 nm light was illuminated on the photodetector, the current increased abruptly from 1.46 × 10⁻³ μA (corresponding to voltage of 2 V) to 0.215 μA, yielding an on/off ratio (i.e. photocurrent/dark current) of 147. Interestingly, with the increase of bias voltage, the photocurrent was observed to saturate when the voltage is as high as ∼14 V. It should be noted that the above photoresponse is highly reproducible when the illumination was repeatedly switched on and off. As shown in Figure 3(c), the photocurrent quickly increases to the maximum value and remains at a steady state upon light illumination. This process could be repeated with the light on and off for several cycles, indicative of good reproducibility of the photodetector device. To obtain a statistical significance, the dark current and photocurrent of 5 representative devices were plotted (Figure 3d). It is noted that all the devices exhibit obvious sensitivity to light illumination, with two lines denoting the average I_dark and I_light of 1.83 × 10⁻³, and 2.36 × 10⁻¹ μA, respectively. The large photocurrent is attributed to the photoconductive properties of Bi₂S₃ NRs upon 560 nm light illumination, whose energy is larger than the bandgap of the Bi₂S₃ NRs [32].

Fig. 3

(a) Schematic illustration of the Bi₂S₃ NR device. (b) I – V curves and responsivity of Bi₂S₃ NR in dark and under 560 nm light illumination, the inset shows an SEM image of the device. (c) Photoresponsive characteristics of the Bi₂S₃ NR device. (d) Summary of both dark current and photocurrent of five representative individual Bi₂S₃ NR based device without HGNs decoration.

The inset of Figure 4(a) shows a typical TEM image of HGNs which were used to enhance the device performance of 1-D Bi₂S₃ NR device. In this study, the HGNs were synthesized by a sacrificial galvanic replacement reaction between Co nanoparticles and aqueous HAuC1₄[33], [34]. According to TEM analysis, their diameters are in the range from 20–40 nm, with average diameter of 30 nm (Figure 4b). What is more, the aspect ratio (i.e.: the inner-to-outer radius or d_in/d_outer) ranges from 20–80%, with a mean value of 55%. Theoretical simulation based on finite element method (FEM) reveals that for all HGNs with varied diameters (the aspect ratio is set to be 70%), the absorption spectrum exhibits strong absorption at 560 nm, which can be attributed to the LSPR of the HGNs. Moreover, the absorption intensity increases with the increasing diameter. This finding is understandable as large metal HGNs are capable of increasing the optical path length inside the nanostructures. Figure 4(d) plots the theoretical simulation of absorption of solid gold nanoparticles (SGN) and HGNs with aspect ratio ranging from 30–70%. Among the various HGNs with different diameters, SGNs show the weakest light absorption [35]. Meanwhile, when the aspect ratio increases gradually from 0 to 30, 50, 60, 70, and 75%, the wavelength of LSPR band is observed to red-shift from 500 to 505, 517, 531, 554, and 613 nm. In fact, the LSPR of the HGNs was experimentally confirmed by UV-vis absorption spectra of HGNs, Bi₂S₃ NRs and HGNs@Bi₂S₃ NRs, as shown in Figure 4(e). The absorption of pure Bi₂S₃ NRs is very weak at ∼570 nm. However, upon surface modification, the HGNs@Bi₂S₃ NRs heterojunction exhibits strong absorption at 570 nm, which corresponds to the LSPR band of HGNs.

Fig. 4

(a) Statistical distribution of the HGNs diameter. (b) The inner-to-outer radius aspect ratio of the HGNs. (c) Theoretical simulation of the HGNs with different diameters, the diameters are from 20 to 24, 28, 30, 36, and 40 nm. In addition, the inner-to-outer radius aspect ratios of all HGNs are set to be 70%. (d) Theoretical simulation of absorption of HGNs, the HGNs have inner-to-outer radius ratios ranging from 0 to 75%, the diameters of all nanoparticles (including both solid AuNPs and HGNs) are set to be 30 nm. (e) Experimental absorption spectra of Bi₂S₃ NRs, HGNs and HGNs @ Bi₂S₃ NRs.

The proof-of-concept plasmonic photodetector was fabricated by modifying HGNs on Bi₂S₃ NR surface via a simple chemical bonding method. Figure 5(a) illustrates the process to functionalize HGNs with ethylene glycol bis(3-mercaptopropionate), whose thiol groups (– SH) can bridge both Bi and Au atoms through chemical bonding. Optoelectronic study reveals that after HGNs decoration, the Bi₂S₃ NR device displays apparently enhanced sensitivity to light illumination. The photocurrent at bias voltage 2 V is about 1.66 μA, which is at least 6 times larger than that of device without HGNs [Figure 5(b)]. In addition, similar saturation in photocurrent was observed when the bias voltage was further increased. Our photoresponse study further suggests that the HGNs@Bi₂S₃ NR device is able to detect the illumination which was repeatedly switched on and off with good reproducibility [Figure 5(c)]. The dark current and photocurrent are 1.87 × 10⁻³ and 1.61 μA, respectively. Figure 5d lists the dark current and photocurrent of 5 representative HGNs@Bi₂S₃ NR devices. Compared with the device without HGNs (Figure 3d), the average photocurrent of HGNs@Bi₂S₃ NR increased from 2.36 × 10⁻¹ to 1.93 μA, while the dark current keeps nearly unchanged (I_{dark, Bi2S3}/I_{dark, HGNs@Bi2S3}:1.81 × 10⁻³ μA/1.83 × 10⁻³ μA), leading to considerable increase in On/Off ratio (from 147 to 1066).

Fig. 5

(a) Schematic illustration of the HGNs@Bi₂S₃ NR device. (b) I – V curves and responsivity of HGNs@Bi₂S₃ NR in dark and under 560 nm light illumination, the inset is the TEM image of a single NR modified with HGNs. (c) Photoresponsive characteristics of the HGNs@Bi₂S₃ NR device. (d) Summary of both dark current and photocurrent of 5 representative individual HGNs@Bi₂S₃ NR based device.

In order to quantitatively evaluate the performance of the present and other Bi₂S₃ nanostructures based devices, several key parameters including responsivity (R), conductivity gain (G), and detectivity (D) are calculated using the following equations:[36], [37]

Where I_p is the photo-excited current, P_opt the incident light power illuminated on the device, h the Planck’s constant, c the velocity of the light, q the elemental charge, λ the incident light wavelength, η the quantum efficiency, A the effective area of the PD and I_d the dark current, respectively. Based on the above equations, R is calculated to be 1.4 × 10² A/W (pure Bi₂S₃ NR) and 1.09 × 10³ A/W (HGNs@ Bi₂S₃ NR) under the illumination of 560 nm light at a bias of 2 V. G is estimated to be 2.68 × 10² (pure Bi₂S₃ NR) and 2.31 × 10³ (HGNs@ Bi₂S₃ NR). What is more, by assuming η = 1, D is determined to be 2.45 × 10¹² and 2.78 × 10¹³ Jones or Bi₂S₃ NRs and HGN@Bi₂S₃ NRs, respectively. Table 1 summarizes the several important metrics of our devices and other Bi₂S₃ nanostructures based photodetectors. Apparently, the key values of HGNs@Bi₂S₃ NRs were not only higher than photodetectors based on pure Bi₂S₃ NRs, but also than other Bi₂S₃ nanostructures based devices [21]. This result suggests that the utilization of the plasmonic HGNs is a promising method for improving the optoelectronic devices.

In addition to the high responsivity, conductive gain, and detectivity, the photocurrent of the plasmonic photodetector is highly dependent on the intensity of the excitation light. Figure 6 plots the photocurrents as a function of light intensity, from which one can find that the photocurrent increases almost linearly with increasing intensity. Carful fitting of both values finds that their relationship could be described by a power law [39]: I = AP^θ, where I, A, P and θ represent the photocurrent, the constant depending on the wavelength, the light power intensity and the exponent determining the response to light intensity, respectively. By fitting the curves, the θ for HGN@Bi₂S₃ NR is estimated to 0.96, slightly larger than that of pure Bi₂S₃ NR (0.91). Such a relatively large exponent is probably due to the passivation effect of the ethylene glycol bis(3-mercaptopropionate) during surface modification process. These nearly integer exponents for both devices are much larger than that of device composed of Bi₂S₃ nanocrystalline film (0.77),¹⁷ which means that the density of trapping center in both devices is very low according to Rose’s theory [40].

Fig. 6

The fitting of the relationship between the photocurrent and light intensity for devices based on both pure Bi₂S₃ NR and HGNs@Bi₂S₃ NR.

The large enhancement of HGN@Bi₂S₃ NRS may be related to the strong LSPR, which is verified by FEM simulation. Figure 7 illustrates the electric field intensity distribution of the HGNs under light illumination with wavelengths of 350, 560 and 740 nm. Obviously, when shined by 560 nm light illumination, the HGNs can induce large-area hot spots with high field intensity. The peak dipolar electric energy according to the color bar, is more than 10-fold, which is much stronger than the ∼2.7-fold (350 nm) and 5.6-fold (740 nm), respectively. In addition, the field enhancement of HGNs under 560 nm light is much stronger than HGNs with the same diameter (5.1-fold). Such superb optical properties can be ascribed to their unique geometry: unlike the conventional solid gold nanoparticles with only one exterior surface, the HGNs have one more surface within the cavity. The two faces are able to facilitate coupling at light illumination, thus leading to relatively large field enhancement [41]. which is highly beneficial for light detection. Without question, compared with traditional plasmonic photodetectors using solid noble metal nanoparticles, the utilization of HGNs for enhancing nano-photodetector is advantageous in the following two aspects. (1) Large enhancement at low cost. Although the HGNs consume relatively few gold in comparison with SGNs, their field enhancement in contrast is much stronger. (2) Tunable LSPR band. As discussed above, the LSPR of HGNs can be easily adjusted in a range from 500–610 nm by tuning the aspect ratio. This convenience will make it possible to develop ultrasensitive nano-photodetectors with tunable spectral selectivity.

Fig. 7

The electrical field distribution of both solid AuNP and HGNs under light illumination of 350, 560, and 740 nm. The solid AuNP has a diameter of 30 nm, and the HGN has a diameter and aspect ratio of 30 nm and 70%, respectively.

When the HGNs with strong LSPR effect were modified on the surface of Bi₂S₃ nanoribbons, the photocurrent, responsivity, and detectivity can be enhanced by the following three mechanisms: (1) Increased capability of light absorption by field enhancement. As a result, more photo-generated electron-hole pairs were generated, leading to enhanced photocurrent. (2) Accelerated separation of electron-hole pairs owning to the enhanced electromagnetic field surrounding the plasmonic HGNs [42], [43], [44]. (3) Direct electron transfer (DET). When irradiated by light illumination, the energetic hot electrons with energy as high as 1–4 eV will directly inject into the conduction band of the nearby semiconductor nanostructures [45], [46]. As a result, the photocurrent and responsivity were considerably increased.

4 Conclusion

In summary, we have demonstrated a simple strategy to improve the performance of nano-photodetector. The as-assembled HGN@Bi₂S₃ NR photodetector is highly sensitive to 560 nm illumination light with good reproducibility. Moreover, the HGNs decorated photodetector exhibits enhanced responsitivity, conductivity gain and detectivity in comparison with pure Bi₂S₃ NR and other Bi₂S₃ monocrystalline based devices. Theoretical simulation based FEM reveals that the enhancement in device performance is related to the LSPR effect of the HGNs, which may increase the light absorption of the HGNs@Bi₂S₃ NR heterojunction. This study suggests that plasmonic HGNs are highly reliable and efficient platforms to boost the device performance of 1-D semiconductor nanostructures based optoelectronic devices.

5 Methods

Synthesis and structural analysis of Bi₂S₃ NRS and hollow gold nanoparticles: The Bi₂S₃ NRs were synthesized via a simiple solvothermal approach. Briefly, NaOH, Bi(NO₃)₃, and Na₂S₂O₃ were added sequentially into a mixed solution of glycerol and water (2:1 by volume), leading to a solution with concentration of 50 mM of Bi³⁺, 2 M of NaOH, and 100 mM of Na₂S₂O₃. The mixed solution was then transferred into a Teflon-lined autoclave (50 mL capacity) which was sealed and maintained at 160°C. After keeping for 20 h, dark-gray precipate floating on the top of the solution was observed. The product was centrifuged and washed with distilled water for several times to remove possible contaminant and ions. The HGNs were prepared by a sacrificial galvanic replacement reaction between Co nanoparticles and aqueous HAuCl₄. The Co nanoparticles were firstly synthesized by the following process: 1 M sodium borohydride solution (100 μL), 0.1 M trisodium citrate dehydrate (400 μL), 0.4 M cobalt hex-ahydrate solution (100 μL) and ultra-pure water (100 mL) were put into a three-neck flask and stirred to form a homogenous solution. To avoid the oxidation of the Co, the reaction solution was purged with N₂ gas flow during the whole reaction process. After the formation of Co nanoparticles, 50 μl 0.1 M of chloroauric acid trihydrate solution was dropped into the above solution. Finally, the HGN was collected by centrifuging the HGN solution at 8000 rpm for 20 min. The phase of the Bi₂S₃ NR was examined using an X-ray diffraction (XRD, Rigaku D/Max-yB with Cu Kα radiation). The morphology and crystal structure were carried out on a SIRION 200 FEG scanning electron microscopy (FESEM), transmission electron microscopy (TEM), and high-resolution transmission electron microscopy (HRTEM, JEOL JEM-2010, at 200 kV). The optical absorption properties were performed on a UV-Vis absorption spectrophotometry (CARY 5000). The chemical composition of the product was analyzed using energy dispersive X-ray spectrometer (EDS), and X-ray photoelectron spectrometer (XPS).

Device construction and characterization: The Bi₂S₃ NRs based device was fabricated via the following procedures. Firstly, the as-synthesized Bi₂S₃ nanoribbons were dispersed on the SiO₂ (300 nm)/Si substrate with a desired density. Then two parallel In electrodes were de-fined on the NR through a photolithography and e-beam evaporation method. Finally, to further decorate with hollow gold nanoparticles, the as-assembled device was immersed into an ethylene glycol bis(3-mercaptopropionate) solution for 1 h and HGN aqueous solution for 12 h sequentially. The optoelectronic properties of the fabricated photodetectors were measured on a Keithley 4200-SCSI-V system equipped with a monochromatic light source system (Princeton, ARC-SP-2155).

Theoretical simulation: The theoretical modeling was carried out by using a finite element method (FEM) to study the unit cell. A periodic boundary condition is used along x direction and transverse magnetic (TM) polarized light was assumed to be normally incident from the top of the structure. The calculated region is bordered by perfectly matched layers (PML) to reduce the influence of light reflection. The optical data of Au were from Sopra S. A. company database.