Characteristic of the BP/ReS2 heterojunction device. Fig. 1(a) presents a schematic of our heterojunction device. All of the devices were prepared by mechanical exfoliation and site-controllable dry transfer method. The detailed fabrication processes has been provided in the Experimental Section. An AFM image of the p-BP/n-ReS2 heterojunction is shown in Fig. 1(b), with the height profile along the red solid line being shown in the inset, indicating the thicknesses of BP (~10 nm) and ReS2 (~12 nm). We first performed the electrical characterizations of the ReS2 FET and BP FET at room temperature. Fig. 1(c) shows the transfer curve of the ReS2 FET with a linear plot of the source-drain current (Ids) versus back gate voltage (Vbg) at a fixed source-drain voltage (Vds = 0.5 V). The device shows an n-type FET behaviors with electron mobility of ~29 cm2V–1s–1 (see the experimental section for the calculation details), which is consistent with the reported value. Optical microscope image of the ReS2 FET covered by ~10 nm thick BN is shown in the top inset of Fig. 1(c). The thickness of the ReS2 is about 6 nm. The Ids–Vds curves under different Vbg for the ReS2 FET have been presented in the bottom inset of Fig. 1(c). The linear Ids–Vds curves under low drain voltage indicates negligible Schottky barrier at the ReS2 and Au interfaces. Fig. 1(d) shows Ids as a function of Vbg for BP FET at room temperature and Vds = 0.5 V. This device showed a p-type FET behavior with hole mobility ~ 166 cm2V–1s–1. The top inset of Fig. 1(d) shows the optical microscope image of the BP FET covered by ~15 nm thick BN. The thickness of the BP is about 10 nm. The Ids–Vds curve of the BP FET is shown in the bottom inset of Fig. 1(d). The near-linear Ids–Vds curve shows a near ohmic contact between BP and Au.
In addition to the conductive type of the materials, the energy band structure of the materials is also very important for construction of a p-n junction. The energy band diagrams of the heterojunction in the isolated and different bias states are illustrated in Figs. 2(a)–2(d). Based on previous reports, the electron affinity and bandgap (Eg) of multilayer BP are around 4.2 and 0.4 eV, respectively[20, 22, 34]. For multilayer ReS2, the corresponding values are about 4.4 and 1.55 eV[26, 32]. Therefore, the BP/ReS2 heterojunction has a type-II band alignment because the conduction band minimum (EC) of BP lies above that of ReS2 by ?EC = 0.2 eV, whereas the valence band maximum (EV) of ReS2 lies below (?EV = 1.35 eV) that of BP. In addition, the valence band edge of BP lies below that of the conduction band edge of ReS2 by only 0.2 eV, which means the BP/ReS2 heterojunction can be easily electrostatic tuned to a broken-gap heterojunction with type-III band alignment, which can be used to explore the TFETs[35, 36]. Fig. 2(a) presents the band diagram for isolated few layers p-BP and n-ReS2. The built-in potential is formed when the two-layered materials are stacked together, which is indicated by blue arrow in Fig. 2(b). At Vds = 0 V, the minority carrier (electron) in BP crosses the small barrier and forms a small reverse current. As shown in Fig. 2(c), under positive bias, the electrons (majority carriers in ReS2) will transport to BP by overcoming the barrier height ?E1. Meanwhile, the holes (majority carriers in BP) will travel to ReS2 by overcoming the barrier height ?E2. Due to the larger ?E2 compare to ?E1, the current is dominated by the over-barrier free-electron transport (blue arrow). Fig. 2(d) shows the electrical field in the heterojunction is strongly enhanced under reverse bias. The minority carrier (electrons in BP and holes in ReS2) contributes to the conduction process.
To investigate the responsivity of the device dependence on the illumination intensities, different illumination intensities P ranging from 0 to 500 W/cm2 are applied to the heterojunction device. Fig. 3(a) shows the |Ids|–Vds curves of the BP/ReS2 heterojunction device at room temperature in the dark and under illumination with a 1550 nm laser at various excitation intensities (P = 10, 25, 50, 100, 250, 500 W/cm2). The |Ids|–Vds curve of the device in the dark exhibits a clear rectifying behavior. The rectification ratio, defined as the ratio of the forward/reverse current, is about 50 at Vds = +1/-1 V. This behavior can be attributed to the formation of p–n diode within the atomically thin BP/ReS2 heterojunction (more details see band alignment of the BP/ReS2 heterojunction). The device exhibits a low dark current (~1.0 nA) at Vds = –1 V. Under a light illumination, electron-hole pairs are generated and separated by the electric field in the heterojunction, which generate the photocurrents. Photocurrent (Iph) is defined as the difference between drain currents with and without light-illumination (Iph = |Ilight – Idark|). Under reverse bias (Vds < 0 V), Iph increases with increasing the light intensity P. Iph can also be enhanced by increasing the reverse bias (shown in Fig. 3(b)). It can be explained by the increased electrical field in the heterojunction under reverse bias (Fig. 2(d)). The increased electrical field decreases the carrier transit time, thus reduces the recombination probability of the photo-generated electrons and holes. In addition, Iph exhibits a sublinear dependence on the incident light intensity P and follows a power law of Iph ∝ Pα, where α = 0.61, 0.68, and 0.73 at Vds = 0, –0.4 and –1 V, respectively. Larger value α can be obtained by applied a larger reverse bias, which also benefits from the shorter carrier transit time. The similar sublinear behavior has also been reported for other similar structures, such as ReS2 FETs, WSe2/graphene, MoS2/WS2 heterostrctures, and several of our previous works (WS2/GaSe, InSe/GaSe, WSe2/GaSe, MoTe2 photodetectors)[39-42]. To quantify the photoresponse performance, the photoresponsivity (R = Iph/PS) of the heterojunction as a function of the illumination intensity at different applied voltages were extracted and shown in Fig. 3(c). The corresponding external quantum efficiency (EQE = hcR/eλ) of the device at different Vds (Vds = 0, –0.4, –1 V) were presented in Fig. 3(d). Here S is the in-plane area (~20 ?m2) of the device, e is the electron charge, c is the speed of the light and h is Planck constant. All of two important factors increase with decreasing light intensity, consistent with the sublinear behavior of the photocurrent. Our device exhibits a photoresponsivity of R = 1.8 mA/W with corresponding EQE = 0.14% at Vds = –1 V under the wavelength 1550 nm. The photoresponsivity is comparable to most of similar vdWs heterojunction photodetectors reported recently, such as 4.36 mA/W in MoS2/WS2.
The image of the BP/ReS2 heterojunction device shown in Fig. 4(a), where the boundaries of each layer are indicated by dotted lines with different colors. The corresponding scale bar is 10 ?m. Fig. 4(b) shows the normalized photocurrent as a function of the illumination wavelength at Vds = –1 V and illumination intensity P = 100 W/cm2, which demonstrates the BP/ReS2 heterojunction device exhibits obvious photoresponse over a broad spectral range (400–1800 nm), from the visible to near infrared. The maximum normalized Iph ~ 1 is obtained at wavelength of 550 nm, and the minimum normalized Iph ~ 0.013 is obtained at wavelength of 1025 nm. Therefore, the device can be used for broadband photodetection. ReS2 has the strongest absorption in the ultraviolet and visible band, but weaker absorption in the near infrared bands. Meanwhile, BP has a natural narrow band gap, which has a strong absorption in visible and near infrared bands. As a result, a strong absorption at 550 nm of the device is induced by the light absorption properties of ReS2 and BP. Another strong absorption at 1710 nm of the device is dominated by the light absorption properties of BP. According to the reported results, the absorption peak position for a N-layer BP can be described using tight-binding model through the expression EN,n = 1.8 – 1.46cos(nπ/(N + 1)), where the subband index n = 1…N. The absorption peaks of 550 nm (2.25 eV) and 1720 nm (0.72 eV) in our device can be explained by n = 13, 5 when L = 20. In order to explore the origin of the photoresponse, scanning photocurrent maps were measured at both zero and negative biases. The corresponding photocurrent maps at zero and negative biases with 1550 nm laser excitation (illumination intensity P = 100 W/cm2) are shown in Figs. 4(c) and 4(d), respectively. The photocurrent map shows that the photoresponse of the overlapping region is strongly enhanced compared with the non-overlapping regions. It also demonstrates the formation of a p–n junction across the area of the BP/ReS2 interface where the photo-generated electrons and holes are separated and extracted more efficiently.
The excellent stability is an important condition for the application of photodetectors. For investigate the stability of the device, the device was illuminated by a laser with different wavelength (or under different Vds) and light intensity P = 50 W/cm2. Fig. 5(a) presents the time dependences of Ids during incident light switched on/off at Vds = –1 V with different wavelength (λ = 400, 532, 1064, 1320, 1550 nm), which shows the device can steady working over a broad spectral range, from visible to near infrared. In addition, the photoswitch ratio (Ilight/Idark) of the device is about 23 at 400 nm, 25 at 532 nm, 12 at 1064, 13 at 1320, and 7 at 1550 nm. Source-drain current Ids as a function of time with photoswitching at different Vds (Vds = 0, –0.2, –0.4, –0.6, –0.8, –1 V) shown in the Fig. 5(d). A stabilized photoresponsivity ~ 0.4 mA/W was achieved at λ = 1550 nm under zero bias, which shows the device can work as a self-driven near infrared photodetectors. Moreover, after many cycles, the photocurrent still responds in a similar fashion to the light, which exhibits excellent operation reversibility and stability.
To examine the polarization sensitivity of the device, the device was illuminated by linearly polarized light with different polarized angle. The photocurrent as a function of polarized angle under different illumination wavelength is shown in Fig. 6. The optical microscope image of the BP/ReS2 heterojunction device is exhibited in Fig. 6(a). The angles of zero degree and 90 degree represents the rotation direction of the linearly polarized light. The anisotropic response in Iph plotted in the polar coordination at visible light wavelength of 532 nm (Fig. 6(b)) and 650 nm (Fig. 6(c)). The visible light intensity P = 25 W/cm2. Figs. 6(d)–6(f) presents the evolution of the Iph plotted in the polar coordination at near infrared light wavelength of 1064 nm (Fig. 6(d)), 1550 nm (Fig. 6(e)), and 1750 nm (Fig. 6(f)). The near infrared light intensity P = 100 W/cm2. The results show the device can act as a polarized photodetector over a broad spectral range (from 532 to 1750 nm). The double symmetry has been observed. The direction of the maximum photocurrent is perpendicular to the direction of the minimum photocurrent. The polarization sensitive absorption of the BP/ReS2 heterojunction photodetector is induced by crystal structure anisotropy. Because of the intrinsic low carrier mobility and high electron-hole recombination rate in ReS2, the polarization sensitivity of ReS2 is relatively low. Thus, the polarization sensitive absorption of the BP/ReS2 heterojunction device is mainly induced by crystal structure anisotropy of BP. The absorption of armchair-polarized light is always stronger than zigzag-polarized light for BP with all thicknesses, thus the two directions of the maximum photocurrent difference correspond to the orientation of BP crystal[45, 46]. It is worth noting the polarized photocurrent ratio Iph.max/Iph.min ≈ 2.98 at 532 nm, 3.39 at 650 nm, 6.44 at 1064 nm, 5.5 at 1550 nm, 4.3 at 1750 nm, which is larger than some other polarized sensitive anisotropy photodetectors, such as GeSe (1.09 at 532 nm) and BP vertical p-n junction (3.5 at 1200 nm). The electrons transition process must meet the transition selection rule of momentum conservation, which can be expressed by the formulas: Eg = ?ω and ?ki ± ?qphonon = ?kf, where Eg is the bandgap of the semiconductor, ki and kf are the electron wave vectors at the valence band maximum (VBM) and conduction band minimum (CBM), respectively, ω is the angular frequency of the incident photon. In addition, the electric dipole transition probability P(ω) for photon absorption per unit time is introduced by Fermi Golden Rule: P(ω) = 2π/?|<φf|HeR|φi>|2δ(E(kf) – E(ki) – ?ω), where ? is the reduced Plank constant, HeR is the interaction of electromagnetic excitation. Here the transition occur from the initial states |φi> to the finial states |φf> with corresponding energy level E(ki) and E(kf). Moreover, the distribution and the overlap of VBM and CBM wave function is different along zigzag and armchair direction in BP, which will result in the difference of the |<φf|HeR|φi>|2 value and wavelength dependent polarized photoresponse. In the near infrared region, BP has a large photoresponse to the polarized light along the armchair direction while the photoresponse of polarized light along the zigzag direction is small and changes little. Thus the large polarized photocurrent ratios are observed in the near infrared region and the maximum polarized photocurrent ratio is 6.44 at 1064 nm.