Interface-Driven Assembly of Pentacene/MoS2 Lateral Heterostructures.

Mixed-dimensional van der Waals heterostructures formed by molecular assemblies and 2D materials provide a novel platform for fundamental nanoscience and future nanoelectronics applications. Here we investigate a prototypical hybrid heterostructure between pentacene molecules and 2D MoS2 nanocrystals, deposited on Au(111) by combining pulsed laser deposition and organic molecular beam epitaxy. The obtained structures were investigated in situ by scanning tunneling microscopy and spectroscopy and analyzed theoretically by density functional theory calculations. Our results show the formation of atomically thin pentacene/MoS2 lateral heterostructures on the Au substrate. The most stable pentacene adsorption site corresponds to MoS2 terminations, where the molecules self-assemble parallel to the direction of MoS2 edges. The density of states changes sharply across the pentacene/MoS2 interface, indicating a weak interfacial coupling, which leaves the electronic signature of MoS2 edge states unaltered. This work unveils the self-organization of abrupt mixed-dimensional lateral heterostructures, opening to hybrid devices based on organic/inorganic one-dimensional junctions.

Introduction

van der Waals heterostructures (vdWHs) have emerged as novel low-dimensional systems of great potential for the development of ultrathin devices with tailored properties. In the most common approach, distinct 2D dangling-bond-free layers are used as building blocks to form vertically stacked vdWHs. The structural and electronic variety resulting from the combination of different 2D crystals can be remarkably enriched by adding dimensionality as a further parameter in the choice of single components, thus extending the vdWH concept beyond the framework of all-2D heterostructures. Combining materials with different dimensionalities in the so-called mixed-dimensional vdWH[1] has recently begun to attract interest for the possibility to significantly broaden the range of properties, functionalities, and potential applications of vdWHs. In particular, 0D–2D vdWHs formed by small organic molecules and 2D inorganic semiconductors provide a novel platform to design and study innovative proof-of-concept optoelectronic devices.[1−6] A prominent example is the vertical stacking of pentacene and MoS2 in a mixed-dimensional organic/inorganic vdWH which forms a gate-tunable photovoltaic junction with antiambipolar characteristics and long-lived charge-separated states.[3,5,6]

In parallel to the development of vertical mixed-dimensional vdWHs, the possibility to fabricate lateral (i.e., in-plane) hybrid heterostructures has been recently demonstrated.[7−10] Lateral vdWHs can form one-dimensional heterojunctions, enabling the development of atomically thin circuitry. However, the nanoscale assembly of low-dimensional materials into a functional lateral heterostructure poses important challenges, such as (i) the development of bottom-up synthesis methods providing molecular-scale control of the assembly structure and (ii) the detailed investigation of the lateral heterointerface at the molecular level. Addressing these issues requires the controlled synthesis of model heterostructures under ideal conditions, e.g., using ultrahigh vacuum (UHV) growth approaches, and their in situ nanoscale characterization by high-resolution techniques, such as scanning tunneling microscopy (STM) and spectroscopy (STS). This experimental approach has recently led to the study of lateral heterostructures between molecules and 2D materials, such as graphene,[7] MoS2,[8] and borophene.[9]

In this work, we focus on the combination of MoS2 and pentacene—which is presently the most performing molecule in organic electronics—in a mixed-dimensional lateral heterostructure. We used the Au(111) surface as a template for growing single-layer (SL) MoS2 crystals and pentacene assemblies by pulsed laser deposition (PLD) and organic molecular beam epitaxy (OMBE), respectively, under UHV conditions. In situ STM–STS, corroborated by dispersion-corrected density functional theory (DFT+D) calculations, shows the formation of in-plane pentacene–MoS2 heterostructures on the Au(111) surface. The molecular arrangement at the interface with MoS2 is driven by the interaction between pentacene and MoS2 edges. However, such an interaction does not affect the MoS2 electronic edge states, which reveal their metallic character in spatial resolved STS. Our findings provide the first molecular-scale experimental observation of pentacene/MoS2 lateral heterostructures, whose structural and electronic abruptness opens to atomically thin vdWH devices with one-dimensional heterojunctions.

Methods

Sample Preparation

All experiments were conducted in an UHV system composed of three interconnected chambers for PLD, OMBE, and STM/STS characterization. Au(111)/mica substrates (Mateck) were cleaned by cycles of Ar+ sputtering (1 keV) and annealing at 700 K. MoS2 was deposited by PLD on freshly prepared Au(111) at room temperature (RT) and subsequently annealed at 730 K for 30 min. The PLD process was optimized to obtain a submonolayer coverage of well-ordered single-layer MoS2 nanocrystals. Briefly, a rotating MoS2 target (Testbourne) was ablated by KrF laser pulses (248 nm wavelength, 10 ns pulse duration) at a repetition rate of 1 pulse per second and a laser fluence of 2 J/cm2. The desired MoS2 coverage on the substrate—placed at 3 cm from the target—was achieved with six laser pulses. Pentacene (Sigma Aldrich, 98% purity) was deposited on MoS2/Au(111) samples at RT from an effusion cell (Dr. Eberl–MBE Komponenten) heated at 428 K. A mild annealing at 350 K was performed after deposition.

Scanning Tunneling Microscopy/Spectroscopy

STM/STS measurements were acquired at RT with an Omicron microscope, using homemade electrochemically etched W tips. Typical measurement parameters were in the range 0.5–2 V for bias voltage and 0.1–0.3 nA for set-point current (the specific values for the reported STM images are stated in the captions). The differential conductivity (dI/dV) was measured using a lock-in amplifier applying a modulation voltage of 30 mVrms at 6 kHz. I(V) and dI/dV(V) curves have been acquired simultaneously in open-feedback-loop conditions, using 1.5 V and 0.2 nA as set-point parameters. Line-mode STS was performed to investigate the lateral interface between different materials. Since data were acquired at RT, we had to compensate for the thermal drift to minimize the uncertainty in the tip position along the line.

Computational Methods

DFT calculations are carried out with the code VASP.[11,12] Core electrons are described with the projector-augmented wave scheme,[13,14] while H(1s), C(2s,2p), Mo(4d,4p,5s), S(3s,3p), and Au(5d,6s) are treated explicitly. We adopt the Perdew, Burke, and Ernzerhof (PBE)[15] functional. The long-range dispersion is accounted for, recurring to the D3 semiempirical scheme and the Becke–Johnson damping.[16,17] The plane-wave basis set is expanded within a kinetic energy cutoff of 400 eV. Convergence thresholds of 10–5 eV (electronic loop) and 10–2 eV/Å (ionic loop) are adopted. The sampling of the reciprocal space is reduced to the Γ point only. Dipole and quadrupole corrections to the total energy are applied along the nonperiodic direction. In order to avoid spurious interactions between replicas of the slab models, a vacuum region of at least 15 Å is included in the supercells. The monolayer (ML) MoS2/Au(111) interface is described by superposing a 10 × 10 MoS2 on a 11 × 11 Au(111) supercell as in our previous work.[18] The same 11 × 11 Au(111) supercell is also adopted for the simulation of MoS2 islands as well as for pentacene adsorption.

Results and Discussion

We prepared pentacene/MoS2 heterostructures following a two-step procedure: first, we synthesized single-layer (SL) MoS2 islands on Au(111) by pulsed laser deposition (PLD); second, after having observed the MoS2 growth by STM, we deposited pentacene by organic molecular beam epitaxy (OMBE).

We will start by discussing the main features of MoS2 on Au(111), observed before pentacene deposition. Following our previous work,[19] we developed a PLD procedure to grow SL MoS2 islands on Au(111): briefly, we deposit MoS2 precursors by ablating a stoichiometric target with a few laser pulses (inset of Figure a) and then anneal the sample at 730 K to favor the crystallization of MoS2 structures. The large-scale STM image in Figure a shows SL MoS2 islands on Au(111) obtained with six laser pulses. The growth of MoS2 islands lifts the surface reconstruction, causing the herringbone ridges to follow a more distorted pattern compared to the regular zigzag pattern of clean Au(111). The apparent height of MoS2 islands is ∼2 Å, relative to the gold terrace. The presence of higher islands, having a brighter color in Figure a, has been previously reported[19,20] and associated with SL MoS2 growing on top of monatomic Au islands, which emerge from Au terraces as a possible consequence of stress release mechanisms induced by MoS2 growth (see SI, Figure S1). Atomic resolution images (Figure b) show the surface S atoms of the MoS2 lattice (∼3.16 Å periodicity) and the hexagonal Moiré pattern (∼32 Å periodicity) generated by the mismatch with Au(111). The Moiré superlattice can be interpreted as a 10/11 MoS2/Au coincidence with no rotational mismatch, as reported in previous works.[18]

Figure 1

STM/STS data of MoS2/Au(111) before pentacene deposition. (a) Large-scale STM image of MoS2 islands on Au(111) (1.2 V, 0.3 nA). Inset: schematic of the PLD process, in which intense laser pulses hit the MoS2 target, producing a plasma plume of ablated species which condenses on the surface. (b) Atomic resolution STM image showing the MoS2 lattice and the Moiré pattern (0.5 V, 0.3 nA). (c) Color map of STS data obtained along a line crossing the Au–MoS2 interface (reported in (a)). The color gradient represents dI/dV (nA/V) in log scale. The black horizontal line indicates the onset of the Au(111) surface state. The white vertical lines indicatively mark the interface region. (d) Point spectroscopy data acquired on the Au(111) (gold line), edge (green), and center (purple) of MoS2 islands. (e) Structural models of S50 (left) and S100 (right) islands. (f) Calculated PDOS of S50 (left) and S100 (right) islands. The gray shaded area is Au PDOS, and the red solid line and the blue dashed ones are MoS2 PDOS at the border and at the island center, respectively.

We investigated the local electronic properties by means of STS measurements acquired on different surface regions. The color map in Figure c shows the differential conductivity (dI/dV) acquired along a line from Au to the center of a MoS2 island. The bare Au region is characterized by the onset of the Au(111) surface state at 0.4 eV below the Fermi level (horizontal black line). At the opposite end of the line, an ∼1.5 eV band gap characterizes the MoS2 region, in agreement with previous STS measurements of MoS2/Au(111).[21−23] The border of MoS2 islands (between the two vertical white lines) shows a finite density of states (DOS) in the gap region, suggesting the influence of edge electronic states. Figure d shows the dI/dV curves of the three regions, obtained by averaging several point spectra acquired on the Au(111) (gold line), edge (green), and center (purple) of MoS2 islands. The presence of intragap states localized at the MoS2 edges is evidenced by the metallic character of the corresponding local DOS (green line), in contrast to the semiconducting behavior of the inner part of MoS2 islands. This observation agrees with previous reports of metallic edge states in MoS2 nanocrystals on Au(111).[24−26]

To analyze in more detail the observed electronic properties, we conducted DFT calculations on MoS2 islands supported on a four-layer Au(111) slab. The islands have triangular shapes, and their borders are cut along the (10 1̅0) crystallographic direction, which was proven to be thermodynamically favorable by means of DFT calculations.[27,28] Each side is composed of seven molybdenum atoms, which is a size range where the most important chemical properties are reasonably converged.[29] However, one should be aware that the islands considered in the simulations are remarkably smaller than the ones grown experimentally, and all considerations on the stability and preferential shape of these objects may depend on their size, even though this should not influence the interaction with pentacene. We considered two types of terminations, S50 (Figure e, left) and S100 (Figure e, right), differing in the loading of S atoms at the border, a parameter that is sensitive to the chemical environment of the deposition. The S50 model (Mo45S99) displays single S atoms bridging two peripheral Mo atoms and pointing downward to the Au substrate, saturating 50% of the dangling bonds at the border Mo atoms. The S100 model (Mo45S126) displays ending S2 dimers in the MoS2 lattice positions, saturating 100% of the Mo dangling bonds. During the relaxation, S50 assumes a rotated position relative to the metal substrate, while S100 remains almost aligned. In the inner region of the islands, the MoS2/Au interfacial distance resembles the one observed in the case of an extended MoS2 monolayer (ML) (2.6 Å),[18] while in corner regions there are closer S–Au contacts (2.4 Å). Both models display an interesting electronic feature, revealed in the projected DOS (PDOS) plots reported in Figure f: at variance from ML MoS2/Au(111) (see SI, figure S2), where the PDOS projected on the supported film orbitals displays a gap comparable to the free-standing case, here the region around the Fermi level is populated by several states located at the island borders. This supports the assignment of the local DOS observed by STS at the MoS2/Au interface as edge states. As noted before, S100 islands are aligned with the Au substrate, as the islands observed experimentally. Therefore, in our growth conditions the formation of islands terminated like the S100 is favored against the S50. This suggests that the growth occurs in a nondeficient S environment, in agreement with the fact that the PLD process is approximately stoichiometric.

After having studied the MoS2/Au(111) system, we deposited pentacene by molecular evaporation onto the sample at RT. The deposition was followed by a mild annealing at 350 K, to favor desorption of possible impurities resulting from the evaporation process. Figure a shows a large-scale STM image taken after deposition: the Au surface not occupied by MoS2 is now entirely covered by a monolayer self-assembly of pentacene molecules which overlays the herringbone reconstruction without lifting it. Higher-resolution images (Figure b) show that the Moiré pattern on MoS2 islands is not affected by the deposition, a sign that pentacene molecules do not lie on top of MoS2 nor intercalate between MoS2 and Au. Far from MoS2 islands, the molecular arrangement follows the same patterns which can be observed on clean Au(111), i.e., without MoS2. Indeed, from preliminary investigations, we observed that pentacene can form various ordered monolayer domains, coexisting on the Au(111) surface, where molecules are arranged in geometrically distinct molecular lattices. Some representative STM images of pentacene/Au(111) are reported in the SI (Figure S3) and are in good agreement with previous literature.[30,31] The preference for pentacene to grow only on the bare Au surface rather than on top of MoS2 leads to the formation of lateral (i.e., in-plane) interfaces between pentacene and SL MoS2 islands. The situation is depicted in Figure c, where the line profile (top) along the dashed line in Figure b is accompanied by a schematic of the sample topography (bottom). Interestingly, Figure b also shows that the presence of MoS2 islands perturbs the ordered arrangement of pentacene molecules, which assemble in different configurations close to MoS2. Therefore, the packing geometry of the molecular assembly is locally altered, and in some cases, as in the central part of Figure b, it looks like the pentacene surface density increases near MoS2 borders. This effect combined with increased disorder near the border possibly causes a lower resolution in STM imaging, which prevents us from a more quantitative analysis of the perturbed lattice. Also, we notice that the perturbation induced by MoS2 borders affects in some cases a relatively large area of the molecular assembly (as in the central part of Figure b), whereas somewhere else (as in the upper border of the large MoS2 islands in Figure b) it may affect only the molecules directly facing MoS2 termination. We can attribute this difference to the presence of MoS2 borders of different islands relatively close to each other in the central part of Figure b, which contribute to a more extended alteration of the ordered molecular arrangement. In Figure d, where two twin molecular lattices can be identified (unit cells labeled in white), we observe that the molecules change their orientation in proximity to MoS2 borders, thus losing registry with the ordered domains. A closer look at the MoS2 border regions reveals a tendency for pentacene molecules to align parallel to MoS2 edges. A distinction has to be noted with reference to Figure d: the two MoS2 nanocrystals with darker contrast grow directly on the Au terrace, whereas the other two brighter MoS2 nanocrystals grow on top of Au islands, as described previously (see also Figure S1). Close to the brighter islands, pentacene is in lateral contact with the step edge of the Au island supporting MoS2. In this case, the molecular alignment is governed by the interaction with the Au step edge, rather than MoS2 terminations. In the following, we will not consider such a situation and focus only on the lateral interface between pentacene and MoS2, e.g., formed around the darker islands in Figure d. Such an interface can be observed at higher resolution in the STM images of Figure e,f, which show that most molecules align parallel to the MoS2 border.

Figure 2

STM images of pentacene/MoS2 on Au(111). (a) Large-scale STM image showing SL MoS2 islands surrounded by pentacene molecules (1.2 V, 0.17 nA). Inset: schematic of pentacene deposition by evaporation onto previously prepared MoS2/Au(111). (b) STM image showing the self-assembly of pentacene molecules around the border of MoS2 islands (0.8 V, 0.15 nA). (c) Top: Line profile along the dashed line in (b). Bottom: schematic of the sample topography. (d) STM image showing the perturbation of pentacene self-assembly due to the presence of MoS2 islands (2.2 V, 0.3 nA). The white parallelograms label the unit cells of twin molecular lattices far from MoS2 edges. (e),(f) STM images showing at higher resolution the MoS2 islands framed in (d) and (b), respectively. Scale bars: 3 nm. (e) 2.2 V, 0.3 nA. (f) 0.8 V, 0.15 nA.

In order to shed light on the nature and relative strength of the interactions established by pentacene with the gold substrate and the MoS2 borders, we performed DFT calculations on the three models reported in Figure a–c. In Figure a, pentacene is adsorbed on the Au(111) surface in a hollow configuration, previously reported as the global minimum in a computational study,[32] yielding an adsorption energy of −2.78 eV relative to the clean substrate and the molecule in the gas phase. This value is very close to previous dispersion-corrected DFT calculations.[33] It can be observed that the molecule lies almost perfectly flat on the metal surface. In Figure b, the pentacene molecule is adsorbed at the border of the MoS2–S100 island. The molecule is aligned parallel to the island border and is slightly tilted relative to the Au substrate. This site is remarkably more favorable for pentacene adsorption compared to Au(111), and the adsorption energy is now −3.15 eV, indicating that both the gold surface and island border atoms cooperate in binding the molecule. A very similar adsorption energy, −3.11 eV, is yielded for an analogous configuration at the MoS2–S50 island border (Figure S4a), indicating that the chemical composition of the island termination has little effect on the stabilization of pentacene, which can be justified by the dispersive and chemically nonspecific nature of the interaction. In Figure c, the pentacene adsorbed on the MoS2 single layer supported on Au is shown. Here the adsorption energy (−1.74 eV) is smaller compared to that reported for the Au surface or the MoS2 interface. It is worth noting that, if pentacene is adsorbed on free-standing MoS2 islands (Figure S5), either on top of the flat terraces or at the borders (with little changes related to the adsorption configuration and the chemical nature of the termination), adsorption energies similar to the monolayer (in the range between −1.3 eV and −1.7 eV) are obtained. This proves that the molecule stabilization at the MoS2/Au border, as highlighted in Figure d, is an interface effect. However, even though the arrangement of Figure c is less favorable compared to the interface region (Figure b) and the clean Au(111) surface (Figure a), the calculations indicate that pentacene can be adsorbed also on the inner region of the MoS2 islands, a fact which, at first glance, is in contrast with the STM images (Figure ). In fact, this result is not surprising, if one considers the universal and size-dependent nature of van der Waals forces, implying that pentacene should bind to the surface of MoS2 islands as well. However, one should also consider that molecules deposited at RT, like in the present work, will have enough thermal energy to evolve toward the thermodynamically favored sites of adsorption. The trend in adsorption strength emerging from DFT calculations (MoS2/Au interface > Au > MoS2) is also supported by recent temperature-programmed desorption measurements available for pentacene on Au(111) and basal MoS2(001), also showing a larger zero-coverage extrapolated desorption energy on Au (2.2 eV)[34] compared to MoS2 (1.2 eV).[35] In general, the presently adopted DFT-D3 approach is thus robust in depicting stability trends in weakly bound interfaces. However, the absolute adsorption energies may be overestimated with respect to accurate experimental calorimetric measurements, as also recently discussed for condensed aromatic molecules on Au(111).[33] To further investigate this aspect, we started from the structure analogous to the one reported in Figure b for the S50 termination and put a second pentacene molecule tilted at the island border (Figure e). This yielded an adsorption energy of −2.47 eV per pentacene molecule, indicating that the adsorption of further pentacene molecules in the border region is still more favorable than covering the MoS2 islands. We conclude that, even if a full pentacene coverage was reached at the MoS2/Au interface, the creation of multilayer pentacene adducts in this region would be more favorable compared to covering the inner MoS2 region. This fully accounts for the evidence from STM measurements reported in Figure . Finally, we checked the stability of a pentacene molecule adsorbed at the MoS2/Au interface orthogonally relative to the island border (Figure f). In this case, an adsorption energy of −2.89 eV is obtained, confirming thus that the parallel configuration is more favorable (see also SI, Figure S4b).

Figure 3

Top (above) and side (below) views of the DFT structures for pentacene adsorbed (a) on Au(111), (b) at the MoS2–S100/Au border (parallel configuration), and (c) on the monolayer MoS2/Au. (d) Pentacene adsorption energy for (a), (b), and (c) cases. (e) Adsorption of two pentacene molecules at the MoS2–S50/Au interface. (f) Pentacene adsorbed at the MoS2–S100/Au border (orthogonal configuration).

To assess the electronic properties of the pentacene/MoS2 system, we performed STS measurements across the heterointerface. dI/dV curves were acquired as the tip was moved from pentacene to MoS2 along a straight line (reported in Figure a). The data are shown in the color map of Figure a. The DOS in the pentacene region is characterized by a small gap approximately from −0.5 to 0.5 eV, in agreement with reported STS of isolated pentacene molecules on Au.[36] The interface region is characterized by the electronic features of MoS2 edge states, closing the energy gap at around ∼7 nm, as also shown by the corresponding dI/dV curve in Figure c (green line). The band gap is then restored in the MoS2 region (∼8–12 nm). The spectral feature at about −1.2 eV is shown by all the spectra in Figure c and is not present in the MoS2 spectrum of Figure d: it could be due to the influence of tip states during the acquisition of STS measurements. The observation of MoS2 edge states suggests that pentacene is not strongly bound to MoS2 terminations, which then retains its original electronic structure.

Figure 4

(a) STM image of pentacene/MoS2 on Au(111) showing the interfacial regions of the lateral heterostructure (0.8 V, 0.15 nA). (b) Color map of STS data obtained along a line crossing the pentacene/MoS2 interface (reported in (a)). The color gradient represents dI/dV (nA/V) in log scale. The white vertical lines indicatively mark the interface region. (c) Point spectroscopy data acquired on the pentacene layer far from the MoS2 borders (gold line), edge (green), and center (purple) of MoS2 islands. Calculated PDOS for (d) pentacene/Au and (e) pentacene/MoS2–S100/Au. Corresponding structural models are reported on the left. The gray shaded area is Au PDOS; red solid lines are pentacene PDOS; and the black dashed line in (e) is MoS2 PDOS at the border.

The calculated PDOS, reported in Figures d and 4e, provide some useful information. First of all, one can notice that, for pentacene on Au(111), the peak related to the molecule HOMO lies at −0.33 eV relative to the Au Fermi energy, while the LUMO is at +0.89 eV. This is in reasonable agreement with ultraviolet photoelectron spectroscopy (UPS) measurements on the pentacene/Au interface (−0.5 eV for HOMO and +1.0 eV for LUMO).[37] The situation at the MoS2/Au interface is different: besides the small positive shift of the pentacene HOMO, in the energy range corresponding to the MoS2 band gap, there are many interfacial states related to the bonding between the unsaturated atoms at the MoS2 borders and the metal substrates,[29] at variance relative to that observed for an infinite MoS2 film supported on Au(111).[18] Simulations corroborate the experimental observation that the lateral assembly of pentacene molecules does not affect significantly the electronic structure of MoS2 borders. This behavior determines the abrupt transition of the local DOS across the lateral interface.

Conclusions

In conclusion, we reported a molecular-resolution investigation of lateral pentacene/MoS2 nanoheterostructures, supported by Au(111). STM/S measurements and DFT calculations showed that interfacing pentacene molecules with single-layer MoS2 result in a unique self-assembly where molecules are aligned along MoS2 edges. The local electronic DOS undergoes an abrupt transition across the in-plane interface, characterized by the contribution of MoS2 metallic edges states. Our findings shed light on the nanoscale physics of the system, highlighting the stabilizing role of MoS2 terminations in driving the interface assembly of pentacene molecules. This work shows the potential of molecular-scale investigation of hybrid lateral heterostructures, opening to further investigations and future engineering of atomically thin mixed-dimensional devices.