1. INTRODUCTION

High-performance, self-powered, and flexible photodetectors (PDs) with ultrabroadband detection range of ultraviolet (UV) to terahertz (THz) are highly desired in a wide range of applications [1–4" target="_self" style="display: inline;">–4]. THz PDs play an important role especially in the biomedical imaging, space communications, remote sensing, imaging, and security check fields [57" target="_self" style="display: inline;">–7]. However, owing to the lack of available materials and technical methods, it is still a great challenge to realize those demands. In recent years, because of the capacity of ultrabroadband detection under zero voltage, photothermoelectric (PTE) PDs based on the Seebeck effect have attracted renewed interest with the development of new semiconductor materials [8–11" target="_self" style="display: inline;">–11]. PTE PDs based on two-dimensional (2D) materials, such as graphene, MoS2, and EuBiSe3 by converting photo-induced temperature rising into electric signals have been investigated and demonstrated to extend the detection range from the UV to the THz band [9,1214" target="_self" style="display: inline;">–14]. For PTE PDs, two effective approaches have been applied to improve their photoresponse, including increasing the photo-induced temperature gradient and the Seebeck coefficient difference of a device [1517" target="_self" style="display: inline;">–17]. For example, plasmon-enhanced photo absorption [15,18] and antenna-coupled enhanced absorption strategies in grapheme [6,7,19] have been used to build a large temperature gradient successfully. On the other hand, in order to increase the Seebeck coefficient, gate bias has been adopted to tune the Fermi level in 2D materials [14]. However, it is still hard to meet the requirement of high sensitivity for PTE PDs, due to the limitation of low photo absorbance (2.3% for single-layer graphene) and the complex preparation process of 2D materials [20,21].

Three-dimensional (3D) graphene, including in graphene foams and reduced graphene oxide (rGO) with cross-linked graphene sheets, does not merely hold the superoptical and electric properties of single-layer graphene, but also expresses higher light absorption, longer-ranging conductive networks, and stronger thermal properties [22–28" target="_self" style="display: inline;">–28]. Owing to their excellent properties, 3D graphene foams and rGO have unique advantages for PTE detection [1,21,25,29]. However, most of the current preparation methods of 3D graphene, such as chemical vapor deposition (CVD) [30] and thermal reduction by using GO solution [31], require complex process equipment, a high temperature and pressure environment, and long preparation cycles [32,33]. In recent years, a number of scientific groups have successfully manufactured reduced GO by the laser-scribed process. Laser-scribed rGO (LSG) is an attractive technology to realize any size scarce growth of rGO and has the advantages of independent chemicals or high temperature and shortened preparation cycles, from several hours to a few minutes [3436" target="_self" style="display: inline;">–36]. Moreover, this technology holds the capability of fitting into many platforms, including both flexible and corrugated substrates [34,37]. Therefore, the LSG process, with the merits of low cost, high efficiency, and flexibility has great application value for the photoelectric detection field. However, this technique is rarely applied in photodetection.

Perovskites with ABX₃ structure (A=Cs+, CH3NH3+, or CH(NH2)2+; B=Sn2+, Pb2+, and X=I, Br, or Cl) have exhibited unique performance in UV to visible (Vis) PDs and solar cells because of their high-charge carrier mobility and larger absorption coefficient [38,39]. It is worth noting that perovskite also holds superthermoelectric properties, such as a high Seebeck coefficient and ultralow thermal conductivity [4042" target="_self" style="display: inline;">–42], which could be a promising thermoelectric material for PTE PD to make up for the limitation caused by the bandgap. In addition, owing to the high carrier mobility of perovskites, they will produce an improvement in response speed for PTE PDs.

Herein, we manufacture and investigate a performance-enhanced, self-powered, and flexible ultrabroadband PTE PD, by introducing CsPbBr3 crystal as an additive in the LSG active layer. In the device, the existence of CsPbBr3 promotes light absorption and then improves the photoresponse. The LSG/CsPbBr3 PTE PDs exhibit sensitivity to an ultrabroadband wavelength range varying from the UV to the THz band, with the responsivities of 135 mA/W and 10 mA/W for the UV and THz bands, respectively. The addition of CsPbBr3 also boosts the response speed of the PTE device several times, displaying a response time of 18 ms. Moreover, the flexibility test demonstrates the excellent flexibility and stability of the PTE device. These results provide an effective way to realize high performance, flexibility, and self-powered ultrabroadband spectral detection, especially in THz detection operating at room temperature.

2. DEVICE STRUCTURE AND FABRICATION

2.1 A. Synthesis of CsPbBr₃ Crystal

Lead bromide (PbBr2, 99%) from Energy Chemical and cesium carbonate (CsCO3, 99.9%), oleic acid (OA, technical grade 90%), oleylamine (OLA, 90%), and octadecene (ODE, 90%) from Aladdin were used in this work. The CsPbBr3 solution was prepared by the following process. First of all, the cesium oleate precursor was synthesized and kept at 120°C by mixing Cs2CO3, ODE, and OA into a three-neck flask and heated under 150°C temperature and nitrogen (N2) atmosphere. Second, PbBr2, ODE, OLA, and OA were mixed in another three-neck flask and heated at 120°C for 30 min and turned to 160°C under N2 atmosphere. Then, the cesium oleate precursor was quickly dropped into the flask, and the solution was cooled by an ice-water bath to room temperature. Finally, the CsPbBr3 crystals were taken out by using a centrifugal machine.

2.2 B. Fabrication Process of the Laser-Scribed rGO/CsPbBr₃ PD

Figures 1(a)–1(e) show the fabrication process of the LSG/CsPbBr3 PD. First of all, the polyethylene terephthalate (PET) substrate with the size of 14mm×14mm was cleaned in an ultrasonic cleaner by using ethanol, acetone, and isopropanol for 15 min ordinal. The substrate was then treated with UV ozone for 10 min, shown in Fig. 1(a). Second, the graphene oxide (GO) solution with a concentration of 2 mg/mL was spin-coated on the PET substrate and was naturally dried at room temperature, as shown in Fig. 1(b). A program-controlled 450 nm laser-scribed technology with a predesigned graphic was used to reduce the GO, as shown in Fig. 1(c). The moving speed of the laser spot was 3.3 mm/s and different powers of 237.5, 208.4, 204.2, 200, and 196 mW were used in the reduction process of the GO films. After the laser reduction process, the surface presented raised and formed cross-linked folds, owing to the rapid loss of oxygen with the oxygen-containing functional groups (–OH, –COOH, C–O), which disappeared in the magnified view. Laser-reduced GO is based on the photothermal effect, which can be explained by the fact that the GO can convert light absorption into heat and then eliminate oxygen-containing functional groups, resulting in GO reduction. Moreover, the degree of GO reduction can be well regulated and controlled by the laser power. As shown in Fig. 1(g), it presents different degrees of GO reduction with different laser powers and the best degree of GO reduction shapes from 200 mW laser scribing. Then, the CsPbBr3 crystal was spin-coated on the LSG layer. The detail of CsPbBr3 crystal’s synthesis is described in the methods section. Finally, gold (Au) electrodes were prepared by thermal evaporation technology under a vacuum chamber of 10−4Pa condition via a shadow mask with a channel width of 100 μm and channel length of 2.5 mm [shown in Fig. 1(e)]. Figure 1(f) shows the schematic structure of the LSG/CsPbBr3 PD. Figure 1(g) shows the surface morphology of laser reduced GO with different laser powers.

Fig. 1. (a)–(e) Processing procedures of the LSG/CsPbBr3 PD; (f) schematic structure of the LSG/CsPbBr3 PD; (g) surface morphology of laser reduced GO with different laser powers under electron microscope view (10×40).

下载图片 查看所有图片

2.3 C. Characterization and Testing

The structure of the CsPbBr3 crystal was obtained by X-ray diffraction (XRD). The LSG/CsPbBr3 surface morphologies and energy spectrum were obtained by using a scanning electron microscope (SEM) and energy disperse spectroscopy (EDS). The photoluminescence (PL) spectrum was tested by a 374 nm laser. The photoelectric performance of the LSG/CsPbBr3 was obtained by using a Keithley 2400 source meter with LabVIEW software. The light sources were 405, 532, 808, 1064, and 1170 nm semiconductor lasers, 10.6 μm carbon dioxide (CO2) gas laser, and 118 μm THz source (FIRL 100). The switched on/off light was realized by using a shutter, which was controlled by a calculagraph. The broadband response and the absorption spectra of the device were tested using a Zolix Omni-λ 3007 spectrophotometer with Si and InGaSn PDs. The temperature distribution image was recorded by the infrared thermal imager (FLIR T630sc).

3. RESULTS AND DISCUSSION

Figure 2(a) shows the XRD pattern of CsPbBr3 crystals. It presents typical polycrystalline structure exhibiting (100), (110), (121), (211), (220), (310) patterns and a main peak position at about 30°, corresponding to the crystal planes of (200) [43,44]. Figure 2(b) shows the Raman spectra of the GO and LSG. Two obvious peaks appear at around 1341cm−1 and 1587cm−1, corresponding to D and G bands, respectively [3]. The intensity ratio of ID/IG for LSG increased from 0.84 to 1.02 when compared with rGO, indicating graphene layer generation and a reduction in defects in the LSG. The absorption spectra of the LSG and LSG/CsPbBr3 are shown in Fig. 2(c). Both of the samples present broadband absorption characteristics, while the latter exhibits a relatively higher absorption at the same wavelength and laser power. The primary absorption spectra of the devices are contributed by LSG, and the enhanced absorption in the LSG/CsPbBr3 is caused by introducing the CsPbBr3 crystal. Moreover, an absorption peak at about 500 nm wavelength is found in the LSG/CsPbBr3, which is attributed to the presence of CsPbBr3 [43,44]. Figure 2(d) shows the PL spectra of the CsPbBr3, LSG, and LSG/CsPbBr3 devices, respectively. When the samples were exposed under a 374 nm laser, both CsPbBr3 and LSG/CsPbBr3 exhibit peaks at about 518 nm, arising from the interband optical transition in CsPbBr3; the LSG presents a flat spectrum and no peak line. Figure 2(e) shows the SEM image of the LSG/CsPbBr3 device with a scale bar of 10 μm. It presents overlapped and interconnected reduced GO foams, and the CsPbBr3 crystals are dispersed across the surface of the rGO layer. The cross section SEM diagram of the LSG is shown in the left of the inset in Fig. 2(e). It obviously presents an rGO layer, and the thickness is about 20 μm. The CsPbBr3 crystal presents separation by crystallization with a square shape size of 1 μm, as shown in the inset of Fig. 2(e). Figure 2(f) shows the sample element composition identified by the energy disperse spectrometer spectrum and confirms the existence of C, Cs, Pb, and Br elements.

Fig. 2. (a) XRD pattern of the CsPbBr3; (b) Raman characterization of the GO and LSG; (c) absorption spectra of the LSG and LSG/CsPbBr3; (d) PL spectra of the LSG, LSG/CsPbBr3, and CsPbBr3; (e) surface and cross section SEM image of the LSG/CsPbBr3; (f) EDS spectrum of LSG/CsPbBr3.

下载图片 查看所有图片

To investigate the enhanced photoelectric performance caused by CsPbBr3, quantitative analyses in the photoelectric performance of the two devices based on LSG and LSG/CsPbBr3 under a 532 nm laser illumination were carried out. The radiation spot was fully on the positive electrode, as shown in Fig. 1(f). The current-voltage (I-V) curves within ±6mV of the two devices under dark and different power densities are shown in Figs. 3(a) and 3(b), respectively. Both of the devices exhibit a positive and excellent linear relationship between the current and voltage, demonstrating good ohmic contact between the active layer and the electrodes. Obviously, under light illumination, the I-V curves shifted and a net current was generated at zero bias voltage, indicating a typical PTE effect. The currents show a significant increase, with the power density varying from 58.28 to 390.44mW/cm2 in both of the devices. Notably, the photocurrent values based on LSG/CsPbBr3 varying from 20 to 65 μA demonstrate 4 times higher than that of the LSG PD under the same light illumination. The relatively low contact resistance for LSG/CsPbBr3 junction (<0.25kΩ) when compared with LSG (>1kΩ) may be further beneficial to the photocarrier transmission in the device, resulting in a higher responsivity and faster response speed [45]. Next, the optical switching characteristics of the two devices under 932.48mW/cm2 irradiation at 0 V bias voltage are shown in Figs. 3(c) and 3(d), respectively. These two PDs show stable and repeatable photocurrents with on and off curves. In addition, the LSG/CsPbBr3 PD exhibits faster and an almost 4 times higher photocurrent response than that of LSG PD.

Fig. 3. (a), (b) Photocurrent voltage (I–V) curves of the LSG and LSG/CsPbBr3 PDs under different 532 nm power densities irradiation; (c), (d) optical switching characteristics and time responses of the LSG PD and LSG/CsPbBr3 PD under 932.48mW/cm2 power intensity at 532 nm laser.

下载图片 查看所有图片

In general, photoresponse time is an important parameter to evaluate the performance of a PD, including the rise and fall time. The rise time is calculated as the 10% to 90% of the duration time from the light turning on to the maximum value. The fall time is defined as the 10% to 90% of the duration time from the light turning off to the minimum value [8,29]. The rise and fall times of the LSG/CsPbBr3 PD are about 32 and 18 ms, respectively, which are 10 times faster than those of the LSG, as shown in Figs. 3(c) and 3(d). The faster response time can be attributed to the fast electron mobility of CaPbBr3. Based on the comparison results of the two devices, the LSG/CsPbBr3 PD demonstrates an enhanced and superior photoelectric property that is conducive to photoelectric detection. Therefore, the photoelectric conversion performance of the LSG/CsPbBr3 PD is investigated and discussed in the following sections.

Responsivity (R), detectivity (D*), and the noise equivalent power (NEP) are the crucial parameters to evaluate a PD [1,8]. Figure 4(a) shows optical switching photocurrents of the LSG/CsPbBr3 PD under different power densities of 532 nm at 0 V bias voltage. It is obvious to see that the photocurrents of the device increase from 10 to 110 μA with the increasing power density from 58.28 to 1703.82mW/cm2. This can be explained by the PTE effect that, with the increasing incident light power density, the photo-induced temperature gradient becomes larger, resulting in the photocurrent rising. The detailed theoretical demonstration and explanation about PTE effect in this device are given below. Then the R is calculated by the following equation: where Iillu and Idark represent the currents under laser illumination and dark condition. P, Ee, and A are laser power, irradiance density, and the effective channel area, respectively. Figure 4(b) shows the R and photocurrent curves of the LSG/CsPbBr3 PD as a function of light power density Ee. The photocurrents display a positive relationship with Ee, and the R shows a negative correlation with Ee. Accordingly, the normalized detectivity D* in units of Jones (cm·Hz1/2·W−1) and NEP can be calculated by the following equations: where e is the charge of an electron. The D* and NEP curves as a function of the power density Ee are shown in Figs. 4(c) and 4(d), respectively. The D* displays a negative relationship with the Ee and a maximum value of 160×109 Jones, as shown in Fig. 4(c). Figure 4(d) shows that the NEP presents a positive correlation with the laser intensity Ee; the lowest value is 10pW·Hz−1/2.

Fig. 4. Optical-electrical response characteristics of the LSG/CsPbBr3 PD under different power densities of 532 nm illumination at 0 V bias voltage. (a) Optical switching characteristics of the device under different power intensities; (b) photoresponsivities and photocurrents curves as a function of the laser intensity Ee of the LSG/CsPbBr3 PD; (c), (d) D* and NEP curves as a function of the laser intensity Ee, respectively.

下载图片 查看所有图片

Ultrabroadband photoresponses of the LSG PD and LSG/CsPbBr3 PD are explored by multiple monochromatic lasers under 390mW/cm2 at 0 V bias voltage. Figures 5(a) and 5(b) show the broadband photoresponse characteristics of the LSG device under 1064 nm, 1177 nm, 10.6 μm, and 118 μm illumination. The LSG PD exhibits stable and repeatable on/off photocurrent curves, with the rise and fall time always slower than 300 and 100 ms, respectively. Figures 5(c)–5(e) show the broadband photoresponse of LSG/CsPbBr3 PD with the wavelengths of 405 nm, 532 nm, 808 nm, 1064 nm, 1177 nm, 10.6 μm, and 118 μm (2.52 THz). As shown in Figs. 5(c)–5(e), the device displays stable and repeatable photocurrent switching, with the incident light wavelength varying from 405 nm to 118 μm. The LSG/CsPbBr3 PD exhibits faster speed and higher photocurrents when compared with the LSG PD. These results further confirm that the photoresponse originates from the LSG layer and expresses the advantage of adding the CsPbBr3. It is obvious to see that the photocurrents show decreasing trends, with the wavelength increasing from 405 nm to 118 μm, and demonstrate the highest photocurrent of 45 μA at 405 nm and the lowest value of 1.5 μA at 118 μm. The ultrabroadband R and NEP curves as a function of laser wavelengths are presented in Fig. 5(d). It displays a wide and slow decrease trend of R, with the wavelength increasing from 405 nm to 118 μm, with the highest value 135 mA/W at 405 nm and lowest value 10 mA/W at 118 μm.

Fig. 5. (a) Temporal photocurrent responses of the LSG device under 1064 and 1177 nm illumination at 390mW/cm2; (b) temporal photocurrent responses of the LSG device under 10.6 and 118 μm (2.52 THz) illumination at 390mW/cm2; (c) temporal photocurrent responses of the LSG/CsPbBr3 device under 405, 532, and 1064 nm illumination at 390mW/cm2; (d) temporal photocurrent responses of the LSG/CsPbBr3 device under 10.6 and 118 μm (2.52 THz) illumination at 390mW/cm2; (e) multiwavelength optical switch photocurrent curves from 405 nm to 118 μm; (f) ultrabroadband R and NEP curves of the device with the wavelength range from 405 nm to 118 μm at 0 V bias voltage.

下载图片 查看所有图片

Obviously, the relationship between the R and wavelength is consistent with the absorption spectra. The NEP curve presents increased trends with the wavelength increasing, with the minimum value 20pW/Hz1/2 at 405 nm and the maximum value 0.03nW/Hz1/2 at 118 μm. The photoresponses under different laser wavelengths demonstrate that the rGO/CsPbBr3 PD could work as a ultrabroadband PD covering the UV to THz range. Table 1 summarizes the reported high-performance PDs based on graphene and other 2D/3D materials [1,8,9,25,46]. Compared with the reported results, the rGO/CsPbBr3 photodetector displays the broadest operation wavelength range and highest responsivity. Moreover, the response speed is faster than most reported results.

Now we will fully discuss the theoretical mechanism of PTE effect in the LSG/CsPbBr3 PDs. The mechanism schematic for the PTE effect and the photocurrent generation process of the device are shown in Figs. 6(a) and 6(b), respectively. Figure 6(a) shows the PTE potential difference generation during the whole device under the condition of the laser illuminating on one side of the electrodes. It is known that the PTE effect is based on the photo-induced temperature difference and then results in a potential difference owing to the hot carrier transit during the operation of a whole device. Therefore, the PTE potential can be calculated by where ∇S is the Seebeck coefficient gradient and ∇T is the temperature gradient. Detailed analysis is shown in Fig. 6(b). It displays that temperature difference is generated throughout the whole device under the illumination of the 532 nm laser on one side of the electrodes. Based on the PTE effect, the hot carrier diffuses from the hot end to the cold end and produces the PTE voltage. Different from the photovoltaic effect, in which the photo-excited electron–hole pairs are separated by the built-in potential and transported to the external electrode under an appropriate bias, in the PTE effect, hot carrier dynamics generally dominate photocurrent generation because of inefficient cooling of electrons with the lattice [11,47]. The Fermi energy levels of the CrPbBr3 and rGO are inclined during the movement of hot carriers, and the Fermi energy difference between the two ends is equal to the PTE potential difference.

Fig. 6. (a) Mechanism schematic for PTE effect; (b) schematic of photocurrent generation process of the device; (c) temperature profile of active location under dark and 532 nm illumination; inset, infrared imaging temperature distribution map of the device under 532 nm illumination; (d) increased temperature profile of the device under 532 nm laser illumination; (e), (f) current voltage (I–V) characteristics of the device under 532 and 1177 nm laser irradiation, respectively; (g) photocurrent and temperature variation curves of the device under 532 nm laser illumination.

下载图片 查看所有图片

To further confirm the PTE mechanism in the LSG/CrPbBr3 device, we conducted a series of experiments. First, temperature variation of the device is obtained by carrying out the infrared temperature measurement using an infrared imager, as shown in Fig. 6(c). The inset shows an infrared imager photograph, and the active area is marked with a black box. We define a linear temperature region along the direction of the two electrodes during the active area. To facilitate analysis, the temperature profiles under dark and 532 nm illumination are plotted, as shown in Fig. 6(c). The temperatures of the two electrodes are consistent with each other in darkness, presented by the black line. The temperature gradient of the active area changes visibly when the laser aims at one end of the electrodes shown in the red curve. Obviously, the increase in temperature at the laser exposure end is higher than that at the nonexposure end, and the specific temperature increment is calculated, as shown in Fig. 6(d). It exhibits the temperature gradient of the whole device and a temperature peak at the position of about 160 μm. The maximum temperature occurs at the active layer and the electrode interface, which is the location of the laser.

Next, I-V curves under 4 mW and different laser wavelengths (532 and 1177 nm) illumination are exhibited in Figs. 6(e) and 6(f), respectively. The curves present good linearity, indicating a good ohmic contact between the active layer and electrodes. Moreover, the curves under 532 and 1177 nm laser illuminations shift downward, with no resistance variation when compared to that under the dark condition, producing and generating obvious short-circuit photocurrents and open-circuit photovoltages, which correspond to the PTE currents and PTE voltages, respectively.

It is noted that the PTE potential and current at 532 nm illumination are higher than those at 1177 nm illumination. This can be attributed to the absorption differences of the device on the two different lasers, as shown in Fig. 2(c). Finally, the temperature variation (ΔT) and the corresponding photocurrent (ΔI) (λ=532nm, P=1.6mW) measurement as a function of time are performed, as shown in Fig. 6(g). The ΔT and ΔI exhibit the same variation tendency with the laser switched on and off. The above results confirm that the photocurrent generation in the LSG/CrPbBr3 PD can be attributed to PTE effect.

Flexible photoelectric stabilities measurement of the device was carried out under 532 nm irradiation before and after 1000 bending cycles. The bending diameter was 9 mm, with front and back bending states as shown in the Fig. 7(a) inset. Figure 7(a) shows the I-V curves of the LSG/CsPbBr3 device under 532 nm irradiation (Ee=58.28mW/cm2) before and after bending with different bending states. I-V curves of the flexible device remain nearly unchanged in a voltage range from −6 to 6 mV under different bending states, indicating that the performance of the flexible device was not influenced by the external stress under light illumination conditions. Figure 7(b) shows the temporal photocurrent characteristics of the device under different bending states before and after a bending test for 1000 times at 532 nm irradiation. After 1000 times of the bending test, the device still expresses repeatable optical switching characteristics. The photocurrent slightly declines, from 9.8 to 9.1 μA. The results give reliable evidence for the fabrication of a flexible device based on an LSG/CsPbBr3 composite. In addition, due to the flexibility of the preparation method, the device can be easily manufactured as flexible arrays for ultrabroadband, highly sensitive, and flexible sensor applications in the future.

Fig. 7. (a) I–V curves of the LSG/CsPbBr3 device under 532 nm irradiation (Ee=58.28mW/cm2) before and after bending with different bending states; (b) temporal photocurrent curves of the LSG/CsPbBr3 device before (solid lines) and after (dotted lines) a bending test for 1000 times under 532 nm laser illumination (Ee=58.28mW/cm2) at 0 V voltage.

下载图片 查看所有图片

4. CONCLUSIONS

In summary, we developed a novel high-performance, self-powered, and flexible ultrabroadband PTE PD based on an LSG/CsPbBr3 composite. Owing to the excellent thermoelectric characteristics and enhanced absorption properties of LSG/CsPbBr3, the device displays ultrabroadband photoresponse covering the UV to THz range at room temperature. High performance with the maximum photoresponsivity of 135 mA/W, the lowest NEP of 0.002nW/Hz1/2, and 18 ms response speed is also achieved. The bending test for the LSG/CsPbBr3 PD on the PET substrate demonstrates the superior flexibility of the device. In addition, the PTE effect dominating the photocurrent generation was confirmed and fully discussed. This work opens up a new avenue towards high-performance, self-powered, and flexible ultrabroadband photodetection, especially in THz range at room temperature.