Manipulating the interlayer carrier diffusion and extraction process in organic-inorganic heterojunctions: from 2D to 3D structures

Interlayer carrier transfer at heterointerfaces plays a critical role in light to electricity conversion using organic and nanostructured materials. However, how interlayer carrier extraction at these interfaces is poorly understood, especially in organic-inorganic heterogeneous systems. Here, we provide a direct strategy for manipulating the interlayer carrier diffusion process, transfer rate and extraction efficiency in tetracene/MoS2 type-II band alignment heterostructure by constructing the 2D–3D organic-inorganic (O-I) system. As a result, the prolonged diffusion length (12.32 nm), enhanced electron transfer rate (9.53 × 109 s−1) and improved carrier extraction efficiency (60.9%) are obtained in the 2D O-I structure which may be due to the more sufficient charge transfer (CT) state generation. In addition, we have demonstrated that the interlayer carrier transfer behavior complied with the diffusion mechanism based on the one-dimensional diffusion model. The diffusion coefficients have varied from 0.0027 to 0.0036 cm2 s−1 as the organic layer changes from 3D to 2D structures. Apart from the relationship between the carrier injection and diffusion process, temperature-dependent time-resolved spectra measurement is used to reveal the trap-related recombination that may limit the interlayer carrier extraction. The controllable interlayer carrier transfer behavior enables O-I heterojunction to be optimized for optoelectronic applications.


INTRODUCTION
Forming appropriate heterostructures has been a key concept in solid state science and an important tool to engineer the electronic and optical properties of semiconductors 1 . Heterostructures formed by conventional semiconductors have led to some important discoveries, such as quantum spin hall effects 2,3 and exciton-polariton condensations 4 . While, owing to the lattice-matching conditions required by the epitaxial growth, the lattice constants and crystal structures of the involved materials must be similar 5 . This severely limits the selection of materials and further development of the nanostructures.
The newly developed two-dimensional (2D) materials, such as graphene 6 , black phosphorus 7 , and transition metal dichalcogenides (TMDCs) 8 , provide diversified options to build heterostructures 9 . Importantly, TMDCs MX 2 (M = Mo, W; X = S, Se, Te) have garnered extensive attention in recent years owing to their strong light-matter interactions and extraordinary optoelectronic properties [10][11][12] . As a result, monolayer TMDCs have been integrated with other materials to constitute van der Waals (vdW) heterostructures [13][14][15] . The dangling-bond-free surface of 2D vdW materials not only can provide a platform for the efficient charge separation and transfer but also engineered the light-matter interaction and further explored potential applications for optoelectronic devices 16,17 , including light-emitting diodes 18 , photodetectors 19 , solar cells 20 , and field-effect transistors 21 .
With the rapid development of information technology, organic semiconductor materials were also an important alternative building block for vdW heterostructures, due to their advantages of diverse molecular structures, low cost, mechanical flexibility, and synthetic tunability 22,23 . Recently, organic nanostructures using vdW interaction between 2D layered systems enable new applications [24][25][26] . A phototransistor was obtained with broadband detection capability based on graphene/organic semiconductor heterojunction 27 . Various vertical organic-inorganic (O-I) heterojunctions have exhibited excellent rectifying behaviors as well as strong photovoltaic responses 13,28 . To utilize the superior lightabsorption and charge transfer (CT) properties of these vdW structures, photocarriers generated in the donor layer must transfer to the acceptor layer before recombination 12,29 . Therefore, how to effectively control the O-I structure to achieve balanced interlayer carrier diffusion and injection process to achieve efficient carrier extraction has become a key issue to promote the further development of the O-I structure 30,31 .
In this work, we illustrate the specific effects of dimensional changes in the O-I system, especially from the perspective of modulating the CT and exciton diffusion characteristics of the O-I interfaces. As a result, we have proved the controllable CT and carrier extraction process across the O-I interfaces with the heterojunction changed from 2D to 3D structures. The improved electron transfer rate and carrier extraction efficiency were obtained in the 2D O-I structure. The thickness-dependent exciton diffusion characteristics were observed in O-I heterostructures. In addition, the trap states observed from low-temperature PL spectra also play an important role in electron transfer and carrier recombination in tetracene/MoS 2 structure. Our work offers important guidance for preparing 2D vdW heterostructures with desirable band alignment, potentially leading to the realization of functional 2D photonic and optoelectronic devices.

RESULTS AND DISCUSSION Construction and energy level arrangement of organicinorganic heterojunction
To study the interlayer carrier transfer process, O-I layered heterostructures were selected as the research platform for the 1 interlayer carrier diffusion and transfer process because of their type-II band alignment 32 , which was further confirmed by using UV − vis absorption, ultraviolet photoelectron spectroscopy (UPS) measurements. As shown in Supplementary Figs. 1-2, by extrapolating the linear portion of the Tauc plot ((αhν) 2 versus hν), where α was the absorption coefficient and hν was the photon energy, the optical bandgap energy of monolayer MoS 2 was experimentally determined to be 1.84 eV, which was consistent with the theoretical prediction 21 . In addition, Fig. 1a shows plots of the secondary electron cut-off in the UPS spectra for the MoS 2 and tetracene layer, where the work function was extracted from the difference between photon energy (21.2 eV for He I photons) and photoemission onset (17.62 eV for MoS 2 and 17.88 eV for tetracene) 33 . The values obtained from this analysis were 3.58 and 3.32 eV for MoS 2 and tetracene, respectively. Besides, from linear extrapolation of the onset of the low binding energy presented in Fig. 1b, the energy difference between the valence band edge (E V ) and E F can be estimated at 1.76 eV for MoS 2 and 1.85 eV for tetracene. The E F position of MoS 2 is closed to the edge of the conduction band, indicating highly electron-doped characteristics 34 . As summarized in Fig. 1c, the energy band diagram of the monolithically formed O-I heterojunction exhibited typical features of type-II band alignment. Figure 1d, e show the energy bandgap of 1 L MoS 2 and tetracene by first-principle density functional theory (DFT) calculations, proving that the O-I heterostructure of tetracene and MoS 2 was type-II band alignment, which may lead to the dissociation of electrons and holes to the individual constituent layers in heterojunction 15 . The electrons in the upper layer of tetracene excited by 400 nm laser can be extracted by the bottom MoS 2 layer as shown in Fig. 1f.
Building of multidimensional organic-inorganic heterostructures and their exciton dynamics Because the thickness of the donor tetracene layer in the heterostructures directly defined the length of the interlayer carrier transfer and diffusion process, building multidimensional O-I heterostructures with various thicknesses is a straightforward approach to systematically understand the interlayer carrier transfer behavior. O-I heterostructures with varied thicknesses (keeping the MoS 2 as monolayer, Supplementary Fig. 3) were controllably prepared by a two-step method (detailed information was shown in Methods). All the organic layer thicknesses in this work were measured using atomic force microscopy (AFM) images, which were shown in Fig. 2a-c. The tetracene with a thickness of ca. 0.6 nm was referred to as 1 L, while the 10 nm thick film was referred to as bulk 16 . As a result, the increased organic layer thickness changed the O-I heterojunction from 2D structure to 3D structure. In addition, Raman spectroscopy was further carried out to confirm the TMDC S and organic components of the heterojunctions ( Supplementary Fig. 4a).
The steady-state absorption spectra of the heterojunction and the isolated components of the junction were shown in Fig. 2d. The three absorption maxima of the tetracene film, which were located at 443, 474 and 522 nm, can be attributed to the absorption from H-and J-aggregates, CT states, and associated vibrational sidebands, respectively 8 . The isolated MoS 2 spectrum included the B and A excitons at 605 nm and 651 nm, respectively. Distinctly, absorption enhancement was observed in the O-I heterojunction due to the higher absorption coefficient of the inorganic layer. Typical PL spectra collected from various organic layers were compared in Fig. 2e, the increased PL intensities owing to the increment of absorption ( Supplementary Fig. 4b).
The PL spectra acquired from tetracene and MoS 2 peaked at 532 nm and 668 nm, respectively ( Supplementary Fig. 4c). In general, carriers in the donor layer can be efficiently extracted by the extraction layer during its lifetime which brings about dramatic quenching of the PL ( Supplementary Fig. 5) 35 . Comparing the spectral changes of the organic layer from 2D (0.6 nm) to 3D (bulk), no obvious peak broadening and shift are observed, indicating that the multi-dimensional organic structure has no obvious change of crystallinity affecting the optical properties ( Supplementary Fig. 6a-b). Furthermore, the full width at half maxima measurement was also performed to exclude the 17 possibility that the PL intensity quenching originates from the defect states in tetracene layers of various thicknesses due to the laser irradiation-induced thermal decomposition 36 . These results proved that the defect states do not change significantly with the thickness change. Distinct PL behaviors in different tetracene layers can be well interpreted by the excitonic dynamics processes. The TRPL measurements based on neat tetracene with different thicknesses were performed to accurately verify the potential CT process in the O-I heterojunction (Fig. 2f) 37 . The lifetime data were fitted with the biexponential Eq. (1). For the O-I heterostructures, τ 1 and τ 2 are considered to be nonradiative and radiative lifetimes 16 . The detailed information was shown in Table  1 (Supplementary).
Interlayer exciton diffusion and carrier transfer process with various organic layer thickness For these type II O-I heterostructures, electron carriers transfer from the upper layer (tetracene) to the bottom layer (MoS 2 ) after excitation 38 . The DFT charge distribution calculations of 2D and 3D O-I heterostructures were shown in Fig. 3a, which proved that the organic tetracene layer was an excellent charge donor. Compared to the isolated tetracene with different thicknesses, the average lifetimes of tetracene (donor layer) in O-I heterostructures were decreased, which validates the CT process in the O-I system. For the 2D O-I heterojunction, significant fluorescence and lifetime quenching were observed (green and red lines in Fig. 3b). While, after the tetracene layer changed to the 3D structure (yellow and purple lines in Fig. 3b), the corresponding lifetime decreased from 318 ps to 294 ps at 532 nm and slight fluorescence quenching was observed in Supplementary Fig. 5b. These results indicated that the existence of the faster and more efficient electron transfer process may originate from the 2D O-I system with the optimized nanoscale exciton separation. The related electron transfer rates (k ET ) of these blends were calculated using Eq. (2) 39 .
where τ neat and τ HS are the lifetime of neat tetracene and O-I heterostructures, respectively. For the 2D O-I heterojunction, the k ET value of 6.07 × 10 9 s −1 was obtained, which was much longer than that of 3D structure (2.57 × 10 8 s −1 ), as presented in Fig. 3c, which manifests the more efficient exciton dissociation and thus more sufficient CT state generation at the donor/acceptor interfaces. To further study the interlayer carrier transfer process, the carrier extraction efficiency was calculated as η ¼ τneatÀτHS τneat 100% (Fig. 3c). To make a fair comparison of O-I structures, we used 4.0-10.21 μw excitation light to explore the variation of extracted parameters. The results showed that the electron transfer rates and extraction efficiency remain stable with the change of excitation power (4.0-10.21 μw) for both 2D and 3D O-I structures (Supplementary Fig. 7). Supplementary Fig. 8a exhibited the significant overlap between the normalized absorption spectrum of tetracene and PL spectrum of MoS 2 , demonstrating the potential existence of Förster resonance energy transfer (FRET) in theory. In addition, we also probed the fluorescence decay time at 668 nm that was longer in the 2D O-I system than that in neat tetracene film, shown in Supplementary Fig. 8b. Therefore, we speculated that FRET occurred between the tetracene layer and MoS 2 layer 40 .
To further study the relationship between the carrier transfer and extraction, the carrier diffusion dynamics of tetracene layer in O-I heterostructure was modeled by calculating the population and distribution of the carriers in tetracene layers using a one-dimensional diffusion equation as following where D is the diffusion coefficient of tetracene, and τ represents the exciton lifetime across the sample, which was obtained by fitting the TRPL curve. n(x, t) is the carrier population at x nm depth in the tetracene layers (x = 0 at the upper surface), t the time delay, n(0, t) = 0. Theoretical calculation of carrier density in the organic layer was shown in Note 1 (Supplementary). Based on this model, the TRPL curves of tetracene in heterostructures with different thicknesses were well fitted. The fitting results of D ranged from 0.0027 to 0.0036 cm 2 s −1 (Supplementary Fig. 9), which were consistent with the consequences of conventional organic materials 43 . To better explore the relationship between the exciton diffusion and the organic layer thickness, we further investigated the diffusion length in 2D-3D O-I heterostructures as Eq. (4) 44 : The diffusion length L D was defined as a function of the diffusion time τ D and the diffusion coefficient D. In addition, the diffusion time τ D was the effective value of the intrinsic tetracene carrier lifetime, which can be obtained by fitting the TRPL curve of tetracene in the heterostructure with a bulk thickness (>40 nm, Supplementary Fig. 10) 44 . When the thickness of the O-I heterostructure increased over 40 nm, the carrier extraction effect can be ignored. The results showed that the effective value of τ D was 422 ps. On the basis of Eq. (4), the results of diffusion length ranged from 10.7 nm to 12.32 nm were obtained (Fig. 3d). For the 20 nm organic layer thickness of heterojunction, the shortest diffusion length (10.7 nm) was observed. It was worth noting that for the 3D O-I heterojunction, the diffusion distance was significantly smaller than the thickness of the upper donor layer, which results in the unbalanced interlayer carrier diffusion and the extraction process, in good agreement with the results proved in Fig. 3c. In addition, the exciton-exciton annihilation effect may be increased as the thickness of the organic layer increased, which may limit the exciton diffusion and carrier transfer process in the O-I system 45 .
Thermal field regulated interlayer carrier transfer behaviors The interlayer carrier transfer behavior was also sensitive to the temperature. By decreasing the temperature, the intensities of PL characteristics increased due to the decreased phonon interaction.
In the temperature-dependent PL spectra of MoS 2 ( Supplementary  Fig. 11a), an obvious blue shift was observed, which may be attributed to the broadening of the energy bandgap and exciton confinement by the weak phonon interaction 46 . The temperaturedependent energy bandgap extension can be described as E g ¼ E 0 À αT 2 =ðT þ βÞ, where E g was the energy bandgap, E 0 was its value at 0 K, α and β were proportionality constants 47 . Figure 4a, b show the temperature-dependent steady-state PL spectra of the pristine 1 L (0.6 nm) tetracene and 3D tetracene, respectively. For the pristine tetracene structure, new PL peaks in the range of 700-760 nm were observed, indicating the formation of trap states at low temperature ( Supplementary Fig. 11b) 8 . The bound excitons originated from trap states were caused by impurities and defects on the surface, resulting in the broadening of the PL spectrum.
To better analyze the influence of the trap states of different O-I structures, the temperature-dependent time-resolved spectra were measured, as shown in Fig. 4c, d. Compared with the tetracene lifetime at room temperature, a significant lifetime increase was observed for all organic layer thickness under lowtemperature conditions. Besides, clearer biexponential decay was observed with the temperature decreased due to the trap-related recombination. Compared to the isolated tetracene with different thicknesses, the average lifetimes of tetracene in O-I heterostructures were decreased at the same temperature, which proves the CT process in the O-I system. For the O-I structure with the 2D tetracene layer, we observed a more significant lifetime quenching phenomenon of the donor layer, which is consistent with the aforementioned results at room temperature (Fig. 3b). In addition, the increased lifetime of the organic layer may be originated from exciton fission 48 , which is expected to slow the electron-hole recombination. Like Eq. (2), the related electron transfer rates of these hybrid O-I heterojunctions were measured to describe the temperature-dependent CT process (Fig. 4e). By changing the temperature from 160 to 340 K, the electron transfer rates increased from 2.60 × 10 8 s −1 to 9.53 × 10 9 s −1 for the 2D O-I heterostructure. For all thicknesses of heterojunctions, the lowest electron transfer rates were obtained at 160 K. Besides, the trap states also limited the carrier transfer and carrier extraction at the donor/acceptor interfaces leading to the decreased carrier extraction efficiency at lower temperatures (Fig. 4f). The detailed information about the temperature-dependent lifetime with various organic layer thicknesses was shown in Table 1 (Supplementary).
In summary, we have observed the controllable interlayer carrier transfer and extraction process successfully in multidimensional type-II band alignment vdW heterostructures with tetracene and monolayer MoS 2 . Apart from the controllable carrier transfer research, the thickness-dependent exciton diffusion (e) (f) Fig. 4 Temperature-dependent interlayer carrier dynamic characteristics. a Steady-state PL spectra of 2D tetracene layer from 160 to 320 K. b Steady-state PL spectra of 3D tetracene layer from 160 to 320 K. c The results of the time-resolved spectra of tetracene in isolated 1 L tetracene and 2D tetracene/MoS 2 structures at 160 K, 240 K and 320 K, respectively. d The results of the time-resolved spectra of tetracene in isolated bulk tetracene and 3D tetracene/MoS 2 structures at 160 K, 240 K and 320 K, respectively. e The results of the temperature-dependent electron transfer rates with different organic layer thicknesses. f The results of the temperature-dependent carrier extraction efficiency with different organic layer thicknesses.
process was also observed by ultrafast spectroscopy. Due to the balanced interlayer carrier diffusion and injection process, the highest electron transfer rate (9.53 × 10 9 s −1 ) was obtained in the 2D O-I structure. Furthermore, the electron transfer process was also sensitive to the temperature due to the formation of trap states at low temperatures, which caused the trap-related recombination and limited carrier extraction. Our findings provide a facile and general way to manipulate the light-matter interaction in 2D materials and to control the CT across O-I interfaces in hybrid systems, enabling new possibilities and innovations in future optoelectronic applications.

METHODS Materials
Monolayer MoS 2 on SiO 2 /Si substrates was purchased from SixCarbon Technology Shenzhen Inc. Tetracene was purchased from tetracene TCI Technology.

Preparation of organic-inorganic heterostructures
Tetracene/MoS 2 O-I heterojunction was fabricated in two steps. The monolayer MoS 2 that we used in this work was grown on a 300 nm Si/SiO 2 substrate by the chemical vapor deposition (CVD) method. The tetracene films with different thicknesses were subsequently grown on the monolayer MoS 2 via the physical vapor deposition method. The deposition rate was 0.01 nm s −1 and the vacuum was held to 2.0 × 10 −4 Pa at 215°C. The thickness of bulk-tetracene was over 10 nm. All the tetracene layer thickness in this work were measured using the AFM images.

DFT calculation
DFT calculations were carried out using the Quantum Espresso package. Projector-augmented-wave potentials were adopted to describe the electron-ion interactions, and the generalized gradient approximation in the form of the Perdew-Burke-Ernzerhof functional was employed to treat the electron exchange-correlation. The energy cut-off for plane-wave basis was set to 500 eV. The convergence threshold for the iteration in the selfconsistent field and the maximum force component for geometry optimizations were set to 10 5 eV and 0.01 eV Å −1 , respectively. The k-point sampling of the Brillouin zone was carried out using a 3 × 3 × 1 and 7 × 7 × 1 k-point grid for structural optimizations and energy calculations, respectively.

Characterization
The absorption was characterized with the TU-1900 UV-vis dual-beam spectrophotometer (PG Instruments, Ltd.). The absorption spectra of MoS 2 were obtained by transferring the monolayer MoS 2 to the quartz substrate. The steady-state PL spectra and TRPL spectra were acquired through a confocal optical microscope (Nanofinder FLEX2, Tokyo Instruments, Inc.) equipped with time-correlated single-photon counting (TETRACENESPC) module (Becker & Hickl, SPC-150). All the PL spectra were measured using a charge-coupled device sensor (DU420A-OE, Andor) to select the different regions across the samples. Topography images were obtained using AFM in tapping mode (Solver P47 PRO, NTMDT Co.). Raman measurement was conducted using Micro Raman spectrometer LabRAM HR800 (Renishaw plc).