Introduction

The anomalous photovoltaic effect (APE) emerges when photoexcited electron-hole pairs undergo separation driven by spontaneous polarization within asymmetric single-phase materials1,2. APE avoids the built-in electric field established by a p-n junction and thus is highly desirable for achieving high photovoltaic conversion efficiency to surpass the Shockley–Queisser limit3,4,5,6. To induce pronounced APE, manipulating the symmetry and dimensions of the material is crucial for strengthening the polarization and enhancing the light-matter interaction7,8,9,10. In conventional three-dimensional bulk materials, intrinsic point group symmetries are strictly dictated by the lattice arrangement. APE can be caused by the microscopic grains or ferroelectric domain walls, requiring a controllable granular interface or distorted structure11,12,13,14.

Two-dimensional (2D) van der Waals (vdW) materials with reduced dimensionality intrinsically have an atomically-thin thickness and provide the opportunity to artificially design symmetries at the interfaces or surfaces15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31. For instance, by precisely aligning bilayer transition metal dichalcogenides (TMDs) with a twist angle of exactly zero, out-of-plane electric polarization occurs at the interface25,26. Heterostructures composed of atomic-layered WS2 and black phosphorus, with rationally designed stacking, exhibit interfacial in-plane polarization20. These emerging polarizations produce a prominent APE with high quantum efficiency for energy conversion. Nevertheless, this symmetry breaking is extrinsically localized to the interface between the layers, necessitating the incorporation of a multilayered structure and meticulous control over the interlayer stacking order. Furthermore, the interlayer polarization and its associated photocurrent generation are susceptible to the internal local strain or external electric field27,28,29 due to the weak nature of the vdW interactions.

An alternative approach involves atomic manufacturing of the surfaces of the 2D materials for symmetry breaking in a unit material. By using precise composition engineering through selective atom replacement on one side of the 2D TMDs, a Janus monolayer structure with an asymmetric atomic lattice can be synthesized32,33,34,35. The presence of spatially distinct atoms (for example, sulfur (S) and selenium (Se)) results in the breaking of the mirror symmetry and inversion symmetry, leading to pronounced intrinsic out-of-plane polarization32, a strong piezoelectric effect36,37 and high harmonic generation38 in MoSSe. The substantial inherent polarity in this system should also produce an APE with efficient optoelectronic conversion, which still needs to be fully explored.

Here, we demonstrate Janus MoSSe-based devices for achieving APE with giant responsivity within the 2D material family. This superior responsivity benefits from the robust intrinsic polarization from its atomic substitutional structure, the enhanced depolarization field in thin thickness down to the monolayer, and the boosted light-matter interaction involving excitons. Notably, the photovoltage of each monolayer can be easily scaled, independent of the intricate interlayer twisting or sliding.

Results

Polar structure of the Janus MoSSe monolayer

In our experiment, a local element supply method was developed to synthesize a Janus MoSSe monolayer (see Methods for details) with an intrinsic out-of-plane polarization perpendicular to the film by substituting the top sulfur (S) atoms with selenium (Se) atoms (Fig. 1a). Nondestructive optical Raman spectroscopy, photoluminescence (PL) spectroscopy, and X-ray photoelectron spectroscopy (XPS) were conducted to confirm the Janus MoSSe structure (Fig. 1b and Supplementary Fig. S1 and Fig. S2). Compared with the Raman signals of the MoS2 monolayer (with A1g at ~406 cm–1 and E2g at ~386 cm–1) and MoSe2 monolayer (with A1g at ~243 cm–1 and E2g at ~289 cm–1), MoSSe exhibits a shifted out-of-plane A1g mode located at ~290 cm–1 and an in-plane E2g mode located at ~356 cm–1 due to the broken out-of-plane symmetry and lattice constant variation. The representative PL peak of MoSSe falls between those of MoS2 (at ~1.84 eV) and MoSe2 (at ~1.52 eV), at ~1.70 eV, as illustrated in Fig. 1c. To directly determine the atomic structure of MoSSe, high-angle annular dark-field scanning transmission electron microscopy (STEM) was used to image the cross-section of a MoSSe monolayer. Since the image contrast is proportional to the atomic number, the S and Se atoms located at the bottom and top of MoSSe could be clearly distinguished, respectively (Fig. 1d).

Fig. 1: Polar structure of the Janus MoSSe monolayer.
figure 1

a Structure of the Janus MoSSe monolayer and the generation of the electron-hole pairs (circles with + and – signs) under optical illumination (left). The intrinsic polarization (P) existing in the MoSSe monolayer results in the spontaneous separation of electrons and holes (right). Typical Raman (b) and photoluminescence (PL) (c) spectra of MoS2, MoSSe and MoSe2 monolayers. d Scanning transmission electron microscopy (STEM) image of the cross-section of the MoSSe monolayer, showing the asymmetric structure with Se (brown) on top and S (green) at the bottom. e Incident angle-dependent second-harmonic generation (SHG) intensity ratio (\({I}_{{{{\rm{p}}}}}/{I}_{{{{\rm{s}}}}}\)) between the p and s polarizations in a MoSSe monolayer. The inset shows a schematic of the out-of-plane SHG measurement of the MoSSe sample. The incident angle \(\theta\) can be tuned accordingly by moving the beam position at the objective back aperture using a translation stage. The incident light and SHG signal are denoted with \(\omega\) and \(2\omega\), respectively.

Full size image

To further reveal the intrinsic polarization properties of the MoSSe sample, incident-angle dependent second-harmonic generation (SHG) measurement was conducted (schematically shown in the inset of Fig. 1e). The incident angle could be tuned accordingly by moving the beam position at the objective back aperture using a translation stage (Supplementary Fig. S3). The SHG intensity ratio (\({I}_{{{{\rm{p}}}}}/{I}_{{{{\rm{s}}}}}\)) between the p and s polarizations of the MoSSe sample increases with increasing incident angle (Fig. 1e), which directly shows the contribution from the out-of-plane dipole of the MoSSe monolayer. By carefully analysing the angle-dependent SHG intensity ratio (Supplementary Note 1), the second-order susceptibility associated with the out-of-plane dipole can be roughly extracted, and \({\chi }_{{xxz}}^{\left(2\right)}\) is \(\sim 0.08{\chi }_{{yxx}}^{\left(2\right)}\) at the 800 nm pump wavelength.

Spontaneous photovoltaic effect in the MoSSe devices

The asymmetric atomic structure with out-of-plane polarization in the MoSSe monolayer generates a giant spontaneous photovoltaic effect. MoSSe (monolayer or few-layer)-based devices were fabricated by sandwiching graphene layers as electrodes. The entire devices were encapsulated by boron nitride (BN) films (Fig. 2a). An optical image of a typical MoSSe bilayer device is shown in Fig. 2b. The effective area of the device is marked by the red solid line. Under dark conditions, the device shows a linear current-voltage (I-V) curve crossing the zero point. Upon optical illumination (with a wavelength of 633 nm), a giant spontaneous photocurrent is observed, and the I-V curve shifts away from the origin at zero bias voltage (orange line in Fig. 2d). A device with a similar structure was fabricated using a 2H (hexagonal) stacked MoS2 bilayer for comparison. In contrast to that of the MoSSe device, the measured short-circuit I-V curve of the MoS2 device shows no spontaneous photocurrent at zero bias voltage (olive line in Fig. 2d). Notably, the symmetric I-V curves indicate a negligible p-n junction effect in our devices. By scanning the device with a focused laser spot, the measured short-circuit photocurrent mapping at zero bias voltage was found to be consistent with the effective sample area (Fig. 2b, c). The line scan results show that the spontaneous photovoltaic effect only occurs in the effective area of MoSSe with both top and bottom graphene electrodes (Fig. 2e). This observation excludes the photocurrent origin from the thermal effect induced by the laser irradiation and the p-n junction between MoSSe and graphene.

Fig. 2: Spontaneous photovoltaic effect in the MoSSe devices.
figure 2

a Schematic of the photovoltaic device. The heterostructure consists of a MoSSe film sandwiched between two graphene electrodes, which is further encapsulated by BN films. b Optical image of the MoSSe bilayer device. The effective area (overlap area of the MoSSe bilayer, top and bottom graphene electrodes) is marked by the red solid line. The MoSSe bilayer and graphene electrodes are outlined by orange and gray dashed lines, respectively. c Short-circuit photocurrent mapping result of the MoSSe bilayer device, corresponding to the area marked by the red solid line in (b). d Short-circuit photocurrent of the MoSSe bilayer and MoS2 bilayer devices as a function of bias voltage under the same illumination conditions. The nonzero photocurrent at zero bias voltage indicates spontaneous photovoltaic responses in the MoSSe device. e Line scan photocurrent result of the MoSSe device indicated by the horizontal white dashed line in (b) and (c). The different photocurrent results in and out of the effective area with different configurations help to demonstrate the origin of the spontaneous photocurrent. f Layer dependence of the short-circuit current and open-circuit voltage in the MoSSe device. With increasing layer number, the photocurrent remains nearly unchanged, while the voltage increases accordingly.

Full size image

The prominent spontaneous photovoltaic effect can be scaled up with increasing MoSSe layer number. In contrast to 3R (rhombohedral) MoS2, which exhibits an interfacial spontaneous photovoltaic voltage under a strict stacking order, the intrinsic polarization in MoSSe increases the open-circuit voltage with additional layers, irrespective of interlayer twisting or sliding. The out-of-plane polarization aligns in the same direction, and the stacking does not diminish the polarization of the crystal (Supplementary Fig. S4 and Fig. S5). Figure 2f shows the layer number-dependent open-circuit photovoltage and short-circuit photocurrent of three different devices with 1-3 layers of MoSSe, respectively (Supplementary Fig. S6, S7). Remarkably, in the MoSSe monolayer device, a significant spontaneous photocurrent (~125 nA) is generated with an open-circuit photovoltage of approximately 0.3 mV. For the three-layer MoSSe device, the photovoltage increases to nearly 0.9 mV and the photocurrent is almost unchanged.

Ultrafast response of spontaneous photocurrent in MoSSe devices

In addition to the photoresponsivity, the photoresponse time is also a vital device parameter, because it determines the intrinsic bandwidth (\(f\)) of a photodetector. We used ultrafast time-resolved photocurrent autocorrelation spectroscopy to determine the photoresponse time39,40. In our experiment, two successive laser pulses were applied to the MoSSe device at the same point with a time delay (\(\triangle t\)), as shown in Fig. 3a. When two pulses overlap in time (\(\triangle t=0\)), the photocurrent reaches its minimum value due to the saturation behavior at higher incident powers. When the time delay is apart from zero (\(\triangle t < 0\) or \(\triangle t > 0\)), the photocurrent gradually increases. Specifically, the dip of the photocurrent curve reflects the detector’s ability to distinguish two successive pulses. An exponential fit to the rising curve yields a typical response time of \(\tau=\,\)~50 ps (Fig. 3b, c), corresponding to an intrinsic bandwidth of \(f=0.55/\tau=11\) GHz, which is independent of the pump power. Regarding the noise characteristics, the noise level of the device was obtained to be ~10−21 A2 Hz−1 with zero bias voltage at room temperature (Supplementary Fig. S8). This value is expected given that the device resistance is relatively small (~2 kΩ). When operating at high frequency, the noise level can be substantially suppressed. Therefore, the MoSSe-based devices show the advantage at high frequency with the intrinsic ultrafast photoresponse.

Fig. 3: Ultrafast response of the spontaneous photovoltaic effect in the MoSSe devices.
figure 3

a Schematic of the ultrafast time-resolved photocurrent measurement. The two ultrafast pulses with a time delay Δt are focused collinearly on the device. The generated photocurrent is measured by a lock-in amplifier. b Autocorrelation photocurrent of the device as a function of the time delay between the two pulses at zero bias voltage, with increasing excitation power from bottom to the top. The circles are the experimental data, and the solid lines are the exponential fits. The inset black curves at the top represent the pump and probe pulses, and the black arrow denotes the time delay. c Response time constant of the MoSSe-based device at different incident power, which is extracted from the exponential fitting of the photocurrent-delay curves in (b). Here, the error bars show the standard errors of the fitting parameters.

Full size image

Mechanism of photocurrent generation in MoSSe devices

In the followings, we discuss the origin of the giant spontaneous photovoltaic effect in the MoSSe devices. In an asymmetric polar structure, APE can be caused by three different spontaneous photovoltaic mechanisms. First, the shift current and injection current are second-order nonlinear optical effects governed by the quantum geometric properties and occurs in materials with broken inversion symmetry41,42, which have notable features and can be distinguished from other mechanisms, including unique polarization and power dependence. Second, due to the residual uncompensated polarization charge at the polar material/electrode interface, the depolarization field can also drive the separation of the electron-hole pairs and lead to prominent APE, especially in the thin polar layer43 (Fig. 4a).

Fig. 4: Depolarization field-dominated photocurrent generation in the MoSSe devices.
figure 4

a Schematic of the band alignment (top) and the photocarrier transfer process (both electrons and holes) in the device (bottom). The intrinsic polarization (black arrow) in the MoSSe layer induces charges in the graphene electrodes that do not fully compensate for the polarization charges. The existing depolarization field drives the photocarriers to the interface and leads to a photocurrent (gray arrow) in the direction opposite to the polarization direction. b Excitation wavelength-dependent photocurrent responsivity. This peak corresponds to the exciton resonance of MoSSe. The dots are the experimental data and the solid line is used for eye guidance. c Excitation power-dependent spontaneous photocurrent. In order to characterize the slope clearly, logarithmic coordinates are used in the figure representation. The dots are the experimental data and the dashed lines correspond to the linear and square-root fit, respectively. d Schematic of the polarization-dependent photocurrent measurement. The polarization of the incident light can be tuned by combining a linear polarizer with a \(\lambda /2\) waveplate or a \(\lambda /4\) waveplate. Linear polarization direction (e) and circular ellipticity (f) dependence of the photocurrent by varying the angle of the \(\lambda /2\) or \(\lambda /4\) waveplates, \({\theta }_{1}\) or \({\theta }_{2}\), respectively. The horizontal dashed lines are used for eye guidance.

Full size image

In the MoSSe devices, the generated photocurrent is closely related to the optical absorption of the MoSSe film (Fig.4b). The photocurrent responsivity spectrum shows two peaks at ~700 nm and ~625 nm, which correspond to the absorption of the A and B excitons of MoSSe, respectively (Fig. 4b and Supplementary Fig. S2c). With further decreases in the excitation wavelength, the photoresponsivity reaches above 3 mA/W with a laser spot of approximately 1 μm2 at 450 nm. Regarding the power dependence of the photocurrent, the generated spontaneous current linearly increases with the incident laser power at a relatively low excitation power (Fig. 4c). Notably, the curve shows a clear deviation from linear behavior towards saturation at higher excitation power.

The spontaneous photocurrent in MoSSe devices is independent of the polarization of the incident light under normal incidence. By varying the orientation direction of the waveplates (\({\theta }_{1}\) and \({\theta }_{2}\) are the angles between the polarization direction of the incident light and the fast axis of the \(\lambda /2\) and \(\lambda /4\) waveplates, respectively), the polarization of the incident light upon the sample can be continuously modulated (Fig. 4d). As expected, the photocurrent remains nearly unchanged under different linear polarization directions (Fig. 4e) and different ellipticities (Fig. 4f). While for the scheme of oblique incidence, the photocurrent evolves with the waveplate angle and reaches its maximum when the light polarization is totally in-plane, due to absorption anisotropy (Supplementary Fig. S9). Together with the power-dependent results, it’s indicated that the photocurrent is not dominated by the shift current mechanism, but more like the depolarization field mechanism.

In the depolarization field-induced APE scenario, the intrinsic out-of-plane polarization (\(P\)) existing in MoSSe induces surface charges (\(\sigma\)) with opposite signs in the top and bottom graphene electrodes. The electrons in the electrodes can screen the surface charges, and two dipoles with the same direction form at the two electrode‒sample interfaces. Each dipole is responsible for an electrostatic potential drop (\(\Delta V\)) in the same direction (Fig. 4a and Supplementary Fig. S10). Since the induced charges cannot fully compensate for the polarization charges in MoSSe, the resulting depolarization field inside the material drives the separation of the electron–hole pairs, which can be collected by the graphene electrodes via interfacial charge transfer, thus forming a spontaneous photocurrent. In general, the depolarization field established is proportional to the induced potential difference and inversely proportional to the film thickness. Therefore, a thinner polar film will lead to a stronger depolarization field and more efficient APE. In our experiment, the magnitude of the depolarization field can be estimated by the band diagram and related open circuit voltage, which gives the upper bound value on the order of ~104 kV/cm (Supplementary Note 2).

Discussion

Compared with other photovoltaic materials, Janus MoSSe has numerous figures of merit. The polarization state of MoSSe originates from the intrinsic asymmetric atomic lattice that can stably exist at the thickness of ~0.67 nm. It’s expected to be the thinnest photovoltaic devices with giant photocurrent responsivity within the entire family of 2D material (Supplementary Table S1 and Table S2). Moreover, the enhanced depolarization field and the reduced diffusion length of photocarriers increase the efficiency. On the other hand, MoSSe-based devices are also limited by the typical disadvantages such as lack of polarization switching, existence of strain and extrinsic contamination, and the challenge of mass production. Overall, Janus TMDs are promising candidates to facilitate the development of miniaturized, high-integrated, and flexible thin-film optoelectronic and photovoltaic devices.

In conclusion, by atomic substitutional engineering of the surfaces of 2D TMDs for symmetry breaking, giant APE arises in a Janus structure with an asymmetric atomic lattice and an intrinsic out-of-plane polarization. Unlike conventional photovoltaics, the enhanced depolarization field-induced photocurrent provides an alternative mechanism for developing photovoltaic devices and photodetectors based on thin 2D layered polar materials. Combined with the ultrafast photoresponse time, potential applications in future optical microsensing, microactuation, and high-speed optical detection will be developed for next-generation compact and flexible optoelectronic devices.

Methods

Sample preparation and device fabrication

The Janus MoSSe structure was prepared by the postprocess of as-produced MoS2 film on an a-plane sapphire substrate. The as-grown MoS2 and the ZnSe crystal plate were placed in a face-to-face manner using two pieces of mica. The Se substitution was performed at 870 °C and the system pressure was kept at ~1 torr with 120 sccm Ar and 5 sccm H2. This postprocess time was kept for 30 minutes and the system was naturally cooled down to room temperature after substitution. The devices were fabricated by using the dry transfer method under a microscope. BN and graphene flakes were mechanically exfoliated onto the SiO2/Si substrates and transferred using poly(propylene) carbonate (PPC). Janus MoSSe films were extracted from the substrates and transferred using poly(methyl methacrylate) (PMMA). The Au electrodes were transferred onto the device using PPC.

Device Characterization

Optical images were taken with an Olympus BX51M microscope. Raman and PL spectra were measured by the WITec alpha300R system with a laser excitation wavelength of 514 nm. The cross-sectional transmission electron microscopy samples were prepared using the focused ion beam technique (Thermo Fisher Helios G4 UX). The atomically-resolved STEM experiments were performed in an FEI Titan Themis G2 300 operated at 300 kV. The X-ray photoelectron spectroscopy (XPS) measurements were performed using an ESCALAB 250X system (Thermo Fisher Scientific) and excited by monochromatic Al Kα radiation.

Incident angle-dependent SHG measurement

The SHG was excited by femtosecond pulses from a Coherent laser (~150 fs, 250 kHz) generated by a Ti:sapphire oscillator. A collimated p-polarized pulse laser beam (with a 2 mm spot size) was guided into the back aperture of a Nikon objective (50x, NA = 0.8). The incident angle was tuned by moving the beam position at the objective’s back aperture using a retroreflector on a translational stage. The excitation wavelength was 800 nm, and the SHG signal was collected using the same objective after filtering the excitation laser, and analyzed by a polarizer along p polarization to exclude s polarized SHG induced by the in-plane dipole. In order to extinct the in-plane SHG component along p polarization, the sample was rotated to a specific crystal angle. The experiment setup is shown in Supplementary Fig. S3.

Time-resolved photocurrent measurements

Two cross-polarized pulsed beams (minimize interference near zero delay, each with ~100 fs pulse width and 710 nm wavelength, sourced from a Ti:sapphire 80 MHz oscillator) were focused on Janus MoSSe devices. The devices were operated without external bias for a spontaneous response assessment. Photocurrent modulation within the device was induced by one beam pulse chopped at 1300 Hz. This modulation was sensitive to the interaction with the second beam pulses, delayed through a motorized delay line by \(\Delta t\), and then the photocurrent’s dependency on the delay time was analyzed. The response time is the time constant τ determined through the exponential fitting.