Introduction

In optical information transmission and processing, photodetectors, which convert optical signals into electrical signals, are critical devices, especially in optical fiber communication [1], night vision [2], infrared remote sensing [3], imaging [4, 5], biomedical [6] and spectrometer [7]. At present, conventional inorganic photodetectors based on Gallium nitride (GaN), silicon (Si) and indium gallium arsenide (InGaAs) were well developed for detecting different sub-bands within the ultraviolet to near infrared range [8,9,10]. However, the flexibility of inorganic semiconductor materials is not satisfactory. Furthermore, some photodetectors based on inorganic semiconductor materials must be cooled during operation [11], which greatly restrict its further development and practical application. Thus, organic semiconductor materials have attracted widespread attention from researchers due to their good compatibility and special physical properties [12,13,14,15]. The development of organic–inorganic hybrid structures and the optimization of multicomponent organic heterojunctions are always good ideas for achieving broad spectral response through the complementarity of absorption spectra between different materials [16, 17]. For example, Yang et al. fabricated an organic photodetector with broad spectral photo-response from 200 to 1000 nm by adopting thick polymer bulk heterojunction composed of 1, 1-bis ((di-4-tolylamino) phenyl) cyclohexane (TAPC) and C70, and achieved an external quantum efficiency (EQE) over 1000%, a specific detectivity (D*) over 1011 Jones (cm Hz1/2/W) [18]. Wang et al. reported a mixed tin–lead perovskites (MASn1−xPbxI3) photodetectors by using low-bandgap (FASnI3)0.6(MAPbI3)0.4 perovskite (FA = formamidinium) as the active layer, and achieved an EQE larger than 65% under − 0.2 V bias, a D* of 1011 ~ 1012 Jones in the wavelength range of 350–900 nm [19]. Han et al. developed a polymer photodetector by inserting cathode and anode interlayers, and achieved a D* over 1012 Jones in the wavelength range of 300–1700 nm [20]. Ma et al. utilized a ZnO/In2O3 heterojunction to improve sensing performance [21]. Zhang et al. fabricated a asymmetric supercapacitors with high energy density by using 3D hierarchical CoWO4/Co3O4 nanowire arrays [22]. And core–shell heterostructures were also used in nano structures [23]. Recently, our group has also demonstrated some broad spectral organic photodetectors based on planar heterojunction or hybrid planar-bulk heterojunction [24,25,26]. In addition, organic photodetectors have the advantages of low cost, high flexibility and large-area scalability, which make organic photodetectors have special research value and broad application prospects in the traditional optoelectronic field [27,28,29,30].

Generally speaking, the broad spectral photodetectors can be classified into three categories: phototransistors, photodiodes and photoconductors, and the structures include planar structure, bulk heterostructure and hybrid structure [31,32,33]. With further research, a series of methods have been adopted to improve the performance of organic photodetectors, such as developing new materials, optimizing device structure, doping quantum dots and inserting the inducing layers [34,35,36,37,38]. In this paper, we reported the broad spectral organic–inorganic hybrid photodiodes (OIHPDs) based on planar-bulk heterojunction of Si/C60/3,4,9,10-perylenetertracarboxylic dianhydride: copper phthalocyanine (PTCDA:CuPc) /Au. C60 presents numerous exciting chemical and physical properties and has been widely employed as an efficient trapping material in various optoelectronic applications. It is worth mentioning that the C60 film was introduced as a hole-blocking layer to enhance the barrier height for blocking hole transport. Thereby, the dark current was significantly reduced, and the spectral response was covered from visible light to near-infrared using PTCDA:CuPc as a light absorbing layer. The resulting organic photodetectors showed a specific detectivity over 1011 Jones in the spectral range of 405–655 nm.

Methods

Fabrication of Devices

The device configuration of the OIHPDs is depicted in Fig. 1, in which P-type silicon was used as the substrates, C60 films as buffer layers, organic bulk heterojunction as photosensitive layers and gold films as the top electrodes. Furthermore, the molecular structures of the organic materials are also inserted into Fig. 1 to understand the principle of the devices. For device fabrication, the p-type silicon substrates were successively cleaned by acetone, alcohol and deionized water for 10 min each, and then dried with floating N2 gas and baked in a vacuum oven at 60 °C for 20 min. Through a quartz crystal oscillator, it can monitor the different thickness of the films, and control the baffle and shadow masks, different thicknesses C60 (δ = 0, 5, 10, 20 and 30 nm, δ is the C60 thickness) films were deposited on the cleaned p-type silicon substrates by vacuum thermal evaporation. Following that, 30 nm-thick PTCDA:CuPc (weight ratio 1: 1, both purchased from Aladdin Biochemical Technology Co., Ltd.) bulk heterojunction films were vacuum evaporated on the top of C60 films. Next, the gold top electrodes were deposited on the organic films by shadow masks. During the deposition process, the chamber pressure was maintained below 3 × 10–4 Pa and the evaporation rate was kept at 0.1–0.2 Å/s. The effective area of each photodetector is 0.06 cm2.

Fig. 1
figure 1

The schematic structure of devices based on hybrid planar-bulk heterojunction. The insets are molecular structures of PTCDA, CuPc and C60

Full size image

Characterization of Devices

The absorption spectra of films were measured by using TU-1901 spectrometer. All measurements were performed using a semiconductor characterization system in a dark chamber at room temperature. Laser diodes with wavelengths of 405 nm, 450 nm, 532 nm, 655 nm and 808 nm were used as the light source, and the different optical powers were realized by using neutral density filters.

Results and Discussion

Figure 2a depict the optical absorption spectra of 30 nm-thick PTCDA:CuPc films on different thicknesses C60 films and C60 single layer. Here, the PTCDA:CuPc films show strong light absorption in the spectrum range of 200–400 nm and a uniform light absorption in the spectrum range of 450–700 nm. It is worth noting that a remarkably enhanced absorption was found from 200 to 500 nm when different thickness C60 films were inserted between the interface of the quartz glasses and PTCDA:CuPc films, which could be attributed to the absorption of C60 film dominates in the spectrum range of 200–500 nm. Figure 2b–d shows the XRD and AFM images of PTCDA:CuPc on 10 nm C60 film and single PTCDA:CuPc film. The double layer show an additional peak at 31.9°. As shown in the AFM images, the surface of the PTCDA:CuPc film is similar with or without the C60 layer. However, the roughness of the photosensitive film on the C60 layer is slightly higher. This shows that the addition of C60 does not significantly improve the performance of the device optically, or even not at all. Therefore, we believe that the improvement of device performance comes from the hole blocking effect of C60.

Fig. 2
figure 2

a Optical absorption spectra of C60 single layer (30 nm) and PTCDA:CuPc (30 nm) films deposited on C60 film of different thicknesses. All films were deposited on quartz glasses. b XRD spectrum of PTCDA:CuPc films with and without C60 film. And AFM images of c C60/PTCDA:CuPc and d PTCDA:CuPc films

Full size image

The crystallite size (D) can be calculated by

$$D = k\lambda /(\beta \cos \theta ),$$
(1)

where k is a constant, equal to 0.89, λ is X-ray wavelength of 0.15405 nm, β is full width at half maxima (FWHM) of diffraction peak, θ is diffraction angle. The FWHM of the peak at 2θ = 31.9 is 0.1617°. The calculated crystallite size is about 50 nm, which is consistent with the AFM image.

Figure 3a shows the typical I–V characteristics of the OIHPDs based on bulk heterojunction without C60 in the dark and under illumination of 532 nm laser. It is clear that the dark current is a bit large at a reverse bias voltage of − 10 V, which is as high as 1615.9 μA. The dependences of photocurrent on the reverse bias voltage are shown in Fig. 3b. It is seen that the photocurrent increases obviously with the incident optical power and gradually becomes saturated with reverse bias voltage increasing, which exhibits standard photodiode characteristics.

Fig. 3
figure 3

a The I–V characteristics of the device Si/PTCDA:CuPc (30 nm)/Au in the dark and under 532 nm laser illumination with different optical power. b The dependences of photocurrent on the reverse bias voltage under 532 nm laser illumination with different optical power

Full size image

As an important parameter of photodetectors, the photoresponsivity (R) is defined as the value of the photocurrent (Iph) which generated in the external circuit under illumination of unit incident optical power (Pin), that is [39]

$$\text{R} = \frac{{\text{I}}_{\text{ph}}}{{\text{P}}_{\text{in}}}.$$
(2)

Another important parameter characterizing photodetectors is the specific detectivity (D*). Here we ignore the power spectral density of other noise sources such as flicker or thermal noise, it can be described as [40]

$${\text{D}}^{*}=\frac{\text{R}}{\sqrt{{2}{\text{q}}{\text{J}}_{\text{dark}}}},$$
(3)

where q is the elementary electric charge, Jdark is the dark current density.

Figure 4a shows the dependence of photoresponsivity R on the wavelength at a reverse bias voltage of − 10 V and an incident optical power of ~ 0.1 mW. In the broad spectral response range from 405 to 808 nm, the responsivities of the OIHPD without C60 film (C60-free OIHPD) are all greater than 0.68 A/W. Especially, at the wavelength of 655 nm, the highest responsivity value is as high as 2.57 A/W. In addition, photoresponsivity generally follows the absorption spectrum of heterojunction films. As shown in Fig. 4b, under different wavelengths of light, the photoresponsivity decreased linearly with the incident light power increased in the logarithmic coordinate. Moreover, the maximal photoresponsivity is 8.13 A/W under 655 nm wavelength illumination with an incident optical power of 0.003 mW, and the minimal photoresponsivity is 0.32 A/W under 808 nm wavelength illumination with an incident optical power of 3.29 mW.

Fig. 4
figure 4

a The dependence of photoresponsivity R on the wavelength. b The logarithmic dependences of R on the incident optical power Pin. All results were obtained at a reverse bias voltage of − 10 V and an incident optical power of ~ 0.1 mW

Full size image

Figure 5 shows the I–V characteristics of the OIHPDs with different thicknesses C60 films (C60-OIHPDs) in the dark and under 532 nm laser illumination. All measurements were conducted in the same reverse bias voltage range with different incident optical power. It is observed that the dark currents were 92.1 μA (Fig. 5a), 25.6 μA (Fig. 5b), 149.0 μA (Fig. 5c) and 179.8 μA (Fig. 5d), which corresponded to the C60 thickness of 5 nm, 10 nm, 20 nm, 30 nm, respectively. Additionally, the C60-OIHPDs exhibited much lower dark current (Fig. 3a vs Fig. 5), and the minimum dark current of C60-OIHPDs was 25.6 μA when the C60 film thickness is 10 nm, which is about 63 times smaller than it of the C60-free OIHPD. Moreover, it is interesting to find that the dark current decreased first and then increased with increasing C60 thickness. Considering the energy level, a mechanism explanation as shown in Fig. 6 is proposed to support the phenomenon of the above OIHPDs. It is noteworthy that the energy levels of valence band (Ev) (5.24 eV) of p-type silicon, the highest occupied molecular orbital (HOMO) (5.3 eV) of CuPc and the work function of the gold electrode (5.1 eV) match well for hole transport in the absence of the C60 layer. Therefore, the holes can reach p-type silicon easily under a very low voltage bias, which causes a large dark current in the C60-free OIHPD. Fortunately, the HOMO (6.4 eV) of C60 is higher than the HOMO (5.3 eV) of CuPc, which means it will match better (Δ = 1.1 eV) for blocking hole transport. This is consistent with the experimental results that a high dark current of 1615.9 μA in the C60-free OIHPD and a low current of 25.6 μA in the C60-OIHPD. However, as an excellent electron transport material with high electron mobility (> 1.3 cm2 V−1 s−1) [41], a small amount of electrons can still easily reach the lowest occupied molecular orbital LUMO (4.2 eV) of C60 from the guide band Ec (4.05 eV) of p-type silicon. Thus, as the thickness of C60 film increasing, we infer that the injection of electrons is enhanced. In this regard, it has been reported that the strain relaxation in the multilayer film is related to the film thickness [42, 43]. Consequently, it is reasonable to assume that a thicker C60 film can provide broader space for the multilayer film, thereby more effectively alleviating the interplanar stress. In addition, the promotion of electron transport by the C60 film has become the dominant factor affecting dark current, which may be due to the increase of the dark current when the thickness of the C60 film exceeds 10 nm.

Fig. 5
figure 5

The I–V characteristics of the device Si/C60/PTCDA:CuPc (30 nm)/Au in the dark and under 532 nm laser illumination with different optical power. The thicknesses of C60 films are a 5 nm, b 10 nm, c 20 nm and d 30 nm, respectively

Full size image
Fig. 6
figure 6

The energy level diagram and charge generation kinetics and transport mechanisms

Full size image

In order to better understand the effects of C60 thickness on device performance, the relationship curves about the dependences of photoresponsivity R on the thickness of C60 films is plotted in Fig. 7a. Under the wavelength of 405–532 nm, there is an overall trend that R firstly increased and then decreased with C60 thickness increasing, and the maximum value of R was reached with 10 nm thick C60 film. It is indicated that C60 film does enhance optional absorption from 200 to 500 nm, which is consistent with the absorption spectrum of C60 single layer (Fig. 2). But the thicker C60 film will also lead to smaller photocurrent and smaller R, this is probably because C60 film hinders the transmission of portion of the photogenerated carriers. As for the wavelength of 450–532 nm, R significantly decreased when the thickness of C60 increasing from 0 to 5 nm, we speculate that the inhibitory effect of the C60 film on the photocurrent is greater than the enhancement effect at this time. Under the wavelength of 655–808 nm, R decreased continuously as the thickness of C60 increasing, since the C60 film has no significant enhancement effect at the wavelength of 655–808 nm. Figure 7b shows the dependences of D* on the thickness for different wavelengths. The D* of device under different wavelengths illumination firstly increased with C60 thickness increasing and reach a maximum value at 10 nm, then decreased. The results above manifest that the optimized C60 thickness is 10 nm, which makes the D* reach the maximum. For the OIHPD with 10 nm C60 film (10 nm-C60 OIHPD), the D* over 1011 Jones in the wavelength range of 405–655 nm, and the highest D* value is 1.96 × 1011 under the illumination of 450 nm wavelength. Compared with C60-free OIHPD, the D* of 10 nm-C60 OIHPD is almost one order of magnitude larger in the wavelength range of 405–808 nm (details in Table 1).

Fig. 7
figure 7

The dependences of a photoresponsivity R on the different C60 thicknesses, b specific detectivity D* on the thickness for different wavelengths. All results were obtained at a reverse bias voltage of − 10 V and an incident optical power of ~ 0.1 mW

Full size image
Table 1 Comparison of the photosensitive performance of broad spectral OIHPDs with different C60 thicknesses
Full size table

Figure 8a shows the dependence of photoresponsivity R on the wavelength for an incident optical power of ~ 0.1 mW at − 10 V. It is worth mentioning that the peak value of responsivity of 10 nm-C60 OIHPD changes to around 450 nm from 655 nm of C60-free OIHPD (Fig. 8a vs Fig. 4a) due to the addition of C60 film, which enhanced the absorption spectrum from 200 and 500 nm (Fig. 2). Furthermore, significantly reduced responsivity values are obtained at the wavelength of 655 nm and 808 nm (Fig. 8a vs Fig. 4a) due to C60 film which hinders the transmission of portion of the photo-generated carriers. As shown in Fig. 8b, it is seen that the 10 nm-C60 OIHPD has the best responsivity under the illumination of 450 nm because of the influence of C60 film. And the maximal responsivity is 4.53 A/W at an incident optical power of 0.004 mW of 450 nm, the minimal responsivity is 0.30 A/W at an incident optical power of 3.29 mW of 808 nm, respectively.

Fig. 8
figure 8

a The dependence of photoresponsivity R on the wavelength. b The logarithmic dependences of R on the incident optical power Pin. All results were obtained at a reverse bias voltage of − 10 V and an incident optical power of ~ 0.1 mW

Full size image

Conclusion

In summary, the photodiodes based on hybrid planar-bulk heterojunction with different thicknesses C60 films were fabricated and characterized. The broad spectral region response from visible to near-infrared demonstrated that using C60 films as hole-blocking layer can effectively enhance the performance of broad spectral OIHPDs. Specifically, the OIHPD with 10 nm-C60 film exhibited the optimized performance with a much lower dark current of 25.6 μA, which is about 63 times smaller than that of C60-free OIHPD. A schematic energy level model combined with experimental measurements is well capable of explaining the origin of decreased dark current. Furthermore, the D* of the 10 nm-C60 OIHPD was almost one order of magnitude larger than the C60-free photodiode, and a notable enhancement of over 1011 Jones was obtained due to the fiercely reduced dark current.