Defect‐Dependent Optoelectronic Properties at a Molecular p‐dopant/Monolayer WS2 Interface

Combining transition metal dichalcogenides (TMDCs) and molecular semiconductors is an attractive route for forming van der Waals heterostructures with optoelectronic properties not found in either component. Herein, the strong p‐type molecular dopant, 1,3,4,5,7,8‐hexafluoro‐tetracyano naphthoquinodimethane (F6TCNNQ), is utilized to form a van der Waals interface with a WS2 monolayer chemical vapor deposition grown on sapphire, which is characterized with angle‐resolved photoelectron and UV‐visible absorption spectroscopy. The interface formed by F6TCNNQ and the as‐grown WS2 monolayer (ML‐WS2) shows no sign of strong interaction or charge transfer. In contrast, defects formed by annealing of ML‐WS2 at 1000 K (d‐ML‐WS2) induce strong n‐type doping of the TMDC. Subsequent F6TCNNQ adsorption on d‐ML‐WS2 is then accompanied by pronounced charge transfer resulting in the formation of interface F6TCNNQ anions and an interface optical transition at 1.5 eV, possibly attributed to an interlayer excitonic state. It is shown in the results that the presence of defects within TMDC monolayers can substantially modify not only the TMDC electronic properties, but also the ground and excited states of their interface with organic molecular dopants. This demonstrates an additional degree of freedom for designing tailored (opto‐)electronic properties based on the combination of TMDCs and molecular semiconductors.


Introduction
The 2D transition metal dichalcogenides (TMDCs) with formula MX 2 (e.g., M = Mo, W and X = S, Se), particularly in their monolayer (ML) form, are semiconductors considered as potential materials for applications in next generation of optoelectronic devices, [1] for example in thin-film transistors, [2][3][4][5] photovoltaic cells, [6,7] photodetectors, [8,9] and bio-inspired devices. [10,11][17] Remarkable features like rich excitonic states, [18,19] clean surfaces, [20,21] and controllable microstructure provide the opportunities for the fabrication of heterostructures built in conjunction with organic semiconductors (OSCs) to combine the advantages of the two distinct class of material. [22,23]specially, organic molecule/ML-TMDC heterostructures can exhibit novel hybrid optoelectronic properties [22,24] and these naturally depend strongly on the groundstate electronic structure of the corresponding interface. [25,26]][26][27] For example, charge-transfer (CT) excitons were observed at tetracene/WS 2 [28,29] and zincphthalocyanine/MoS 2 [30] heterostructures and were attributed to particular CT processes occurring at the OSC/TMDC interface upon photoexcitation.Studies available to date are mostly limited to the interfacial photoexcitation dynamics at molecule/pristine ML-TMDC interfaces.However, TMDCs exhibit a range of structural defects such as sulfur vacancies (V S ), which are the most abundant defects as their formation requires the least activation energy. [31,32][35] Controlling and understanding the impact of defects on the electronic properties of TMDCs, a field of research known as defect engineering, is therefore of fundamental importance.[38][39] Here, we aim at understanding how the presence of defects in an ML-TMDC can impact both the ground and excited states of an OSC/ML-TMDC interface.It is reasonable to DOI: 10.1002/pssa.202300106 Combining transition metal dichalcogenides (TMDCs) and molecular semiconductors is an attractive route for forming van der Waals heterostructures with optoelectronic properties not found in either component.Herein, the strong p-type molecular dopant, 1,3,4,5,7,8-hexafluoro-tetracyano naphthoquinodimethane (F 6 TCNNQ), is utilized to form a van der Waals interface with a WS 2 monolayer chemical vapor deposition grown on sapphire, which is characterized with angle-resolved photoelectron and UV-visible absorption spectroscopy.The interface formed by F 6 TCNNQ and the as-grown WS 2 monolayer (ML-WS 2 ) shows no sign of strong interaction or charge transfer.In contrast, defects formed by annealing of ML-WS 2 at 1000 K (d-ML-WS 2 ) induce strong n-type doping of the TMDC.Subsequent F 6 TCNNQ adsorption on d-ML-WS 2 is then accompanied by pronounced charge transfer resulting in the formation of interface F 6 TCNNQ anions and an interface optical transition at 1.5 eV, possibly attributed to an interlayer excitonic state.It is shown in the results that the presence of defects within TMDC monolayers can substantially modify not only the TMDC electronic properties, but also the ground and excited states of their interface with organic molecular dopants.This demonstrates an additional degree of freedom for designing tailored (opto-)electronic properties based on the combination of TMDCs and molecular semiconductors.
assume that the presence of defect-related states in the TMDC influence also the ground-state electronic properties of the interface, and may favor the emergence of interface excited states at the OSC/ML-TMDC interface.Relating the ground-state electronic properties and the excited states of such systems by assessing their DOS and the corresponding optoelectronic transitions should therefore provide important insights in the fundamental processes taking place at these interfaces.
In this contribution, we investigated the optoelectronic properties of an interface combining a strong organic molecular p-dopant, 1,3,4,5,7,8-hexafluoro-tetracyano naphthoquinodimethane (F 6 TCNNQ), and a WS 2 monolayer (ML-WS 2 ) supported on a sapphire substrate using angle-resolved photoelectron spectroscopy (ARPES) and UV-vis absorption spectroscopy.F 6 TCNNQ, a strong electron acceptor with an EA of %5.6 eV, [40] was deposited on a chemical vapor deposition (CVD)-grown ML-WS 2 featuring initially a small density of defects, and we compare this to the situation when a CVD grown ML-WS 2 with a much higher density of defects, achieved by postgrowth annealing in ultrahigh vacuum (UHV) at 900-1000 K, is used.The ARPES spectra of the highly defective WS 2 monolayer (d-ML-WS 2 ) show an additional DOS at the Fermi level, attributed to the filling of the WS 2 conduction band (CB).In turn, for the non-annealed ML-WS 2 sample, ARPES measurements were prevented by strong sample charging, suggesting that the delocalized charge carriers observed for the highly defective ML are indeed either absent or marginal in the pristine ML-WS 2 .Consecutive ARPES measurements demonstrate the presence of ground-state CT at the F 6 TCNNQ/d-ML-WS 2 interface, and the absorption spectra show a series of peaks that are absent when F 6 TCNNQ is deposited on the pristine ML-WS 2 .One set of peaks is attributed to transitions of F 6 TCNNQ anions, which were identified by ARPES.Notably, an additional absorption peak is observed in between those of the WS 2 A-exciton and the F 6 TCNNQ anions, which is suggested as an interlayer molecule-TMDC transition.These results demonstrate that defect engineering of TMDCs can not only modify the optoelectronic properties of the MLs itself but also of their interfaces with organic molecules.

Electronic Structure of the F 6 TCNNQ/Annealed WS 2 (d-ML-WS 2 ) Interface
Throughout this work, we present the ARPES spectra taken at the Γ-, Q-, and K-points of the ML-WS 2 Brillouin zone (BZ) for the defective d-ML-WS 2 and of F 6 TCNNQ sequentially deposited on d-ML-WS 2 .We note that although the sample is polycrystalline (in plane) and ARPES thus averages over all directions of the BZ, particularly high photoemission intensity is detected along the high symmetry directions due to the high momentum DOS (k-DOS) along those lines.Therefore, the ARPES spectra taken at any k k represent almost exclusively a superposition of the energy distribution curves (EDCs) along the Γ-Q-K and Γ-M lines at the corresponding k k value, which allows band tracking and comparison with the known band structure along these directions. [41]We also remark that ARPES measurements of the as-grown ML-WS 2 and of the F 6 TCNNQ/ML-WS 2 interfaces were precluded by pronounced sample charging.However, we note that measurements of such ML-WS 2 when transferred to a conductive substrate (e.g., HOPG) displayed the expected band structure, with the valence band maximum (VBM) at the K-point, which is responsible for the direct bandgap of ML-WS 2 . [42]Such direct bandgap on the produced sample is here confirmed by the absorption spectrum described in the next section.
ARPES measurements of F 6 TCNNQ sequentially grown on d-ML-WS 2 /sapphire were possible, because the annealed ML-WS 2 had sufficiently high conductivity to avoid sample charging upon photoionization.This is likely a result of the presence of defects (e.g., V S ), which leads to strong n-type doping of the WS 2 .Figure 1 depicts a) the secondary electron cutoff (SECO) spectra, b) EDCs at Γ, and c) the EDCs at K upon increasing nominal F 6 TCNNQ coverage.From the SECO spectra, we determined the sample work function evolution upon molecule deposition, which changes from %4.3 eV for bare d-ML-WS 2 to 5.3 eV for an (sub-) ML F 6 TCNNQ (for 1 and 2 Å nominal F 6 TCNNQ coverage), and shows an additional 0.1 eV increase for the 10 nm F 6 TCNNQ film.When inspecting the valence band spectra at Γ and K of the bare d-ML-WS 2 , we first observe that the VBM onset is located at K at 1.72 eV binding energy (BE), demonstrating electronic properties in line with those of an ML with strong n-type character.We observe a rigid shift of 0.6 eV to lower BE of the TMDC features upon F 6 TCNNQ deposition.The work function increase seen upon F 6 TCNNQ deposition indicates the formation of an interface dipole with the negative pole located in the molecular layer, correlating with a ground-state electron transfer from d-ML-WS 2 to F 6 TCNNQ.Consistently, the VBM shift upon F 6 TCNNQ deposition is attributed to de-doping of the initially strongly n-type d-ML-WS 2 .
A closer look at the electronic states in the Fermi-level region is provided in Figure 2, which depicts the EDCs of the clean and the 1 Å F 6 TCNNQ-covered d-ML-WS 2 taken at a) Γ, b) Q, and c) K points of the BZ of the TMDC.The Γ and K spectra of pristine d-ML-WS 2 are featureless in this energy region, in contrast to the spectrum at Q, which therefore indicates the presence of dispersive states at this point of the BZ.Density-functional theory calculations of the WS 2 band structure yield a CB minimum at the K-point and a local CB minimum at the Q-point which is very close in energy to the CBM. [43]Notably, the here-observed intensity at Q (while absent at K ) and its dispersive character corresponds to that of the CB minimum at Q (see Figure S3, Supporting Information).This demonstrates the presence of delocalized charge carriers within the d-ML-WS 2 [44] whose band structure is strongly renormalized, possibly by the effect of strain at the interface with sapphire and yielding a CB minimum at the Q-point. [43]We note that such band renormalization is still observed after transfer of a WS2 film grown on sapphire to another sapphire followed by annealing in vacuum as necessary to perform the ARPES measurements (Figure S3, Supporting Information).Upon deposition of F 6 TCNNQ, two interfacerelated features with peak maxima at %0.4 eV (peak A) and 1.0 eV (peak B) emerge and are visible at all k k .The absence of dispersion of these electronic states is a clear indication of their molecular character.They are arising because of electron transfer from the ML-WS 2 to the OSC, resulting in the formation of molecular anions. [36,37]Consistent with previous studies, charging of the F 6 TCNNQ molecules by CT leads to the splitting by an energy U, the Coulomb repulsion, of the neutral F 6 TCNNQ lowest unoccupied molecular orbital (LUMO) level into a singly occupied molecular orbital (SOMO) level and a singly unoccupied molecular orbital (SUMO) level. [36,45,46]In the spectra shown here, the peak A with its maximum at %0.4 eV is attributed to the SOMO.The peak B with its maximum at %1.0 eV is assigned to stem from the highest occupied molecular orbital (HOMO) level of the former neutral molecule, whose ionization energy is strongly renormalized due to the additional electron accommodated by the anion.This ground-state CT is driven by the energy difference between the low work function (4.3 eV) of d-ML-WS 2 and the high F 6 TCNNQ EA of 5.6 eV.Additionally, it is also observed that the states representative of the CBM at the Q-point are depopulated upon F 6 TCNNQ adsorption, confirming thereby the ground-state CT between the TMDC and the OSC.Finally, as shown in Supporting Information, further depleting the CB by depositing additional F 6 TCNNQ decreases the conductivity of the sample as slight sample charging starts to occur already at 2 Å coverage (see Figure S4, Supporting Information).

Optical Properties of the F 6 TCNNQ/(d-)ML-WS 2 Interface
We now assess optical excitations of the F 6 TCNNQ/(d-)ML-WS2 interface from the UV-vis measurements presented in Figures 3 and 4. First, the UV-vis spectrum of the pristine (not annealed) ML-WS 2 on sapphire features the expected A-  exciton as a first intense and sharp absorption peak at % 2.01 eV.Below the absorption edge of the A-exciton, no features are observed.The UV-vis spectrum of 10 nm F 6 TCNNQ grown on the pristine ML-WS 2 mostly consists of a linear superposition of the absorption spectrum of pristine F 6 TCNNQ, with its main feature at 2.4 eV, [47] and of the bare ML-WS 2 film as shown in Figure S5, Supporting Information.Note that the small deviation from a perfect linear superposition may result from different orientation of the molecules on the reference quartz substrates and on WS 2 , but also from imperfection of the subtraction procedure.The fact that the optical characteristics of F 6 TCNNQ/ML-WS 2 are close to those of its pristine components is consistent with a weak OSC-TMDC interaction, involving no apparent ground-state CT between the components, because no (or very few) charge carriers are available in the ML-WS 2 .Annealing at 900-1000 K of ML-WS 2 for 24 h in UHV was performed to produce the defective d-ML-WS 2 sample with an expected higher density of V S . [48]he corresponding UV-vis spectrum of d-ML-WS 2 then exhibits similar features as the not annealed ML-WS 2 , both in lineshape and energy position.However, we do not observe any (sharp) defect-related feature just below (i.e., 0.2-0.3eV) the A-exciton, in contrast to previous in situ photoluminescence studies; [34] this is likely due to variations in detailed sample handling.Another noticeable difference is the observation of a broad and asymmetric intensity peaking at about 1.35 eV and extending toward the low energy side.Such intensity apparently stems from interband transitions apparently involving the presence of defects (see discussion in Supporting Information, Section 5).The UV-vis spectrum of the F 6 TCNNQ/d-ML-WS 2 sample displays a redshifted (15 meV) and broadened A-exciton peak, which is readily visible for a slightly thinner (nominally 3 nm) F 6 TCNNQ film as shown in Figure 3b.This redshift compares consistently with a more intrinsic TMDC film, which is here achieved by de-doping upon F 6 TCNNQ adsorption. [49]Furthermore, a series of sub-bandgap peaks below 1.8 eV can be seen, those being absent for all other samples.In more detail, Figure 4 shows the spectra zoomed in  the low energy range (0.9-2.0 eV) after subtracting a background due to Rayleigh scattering.In this energy range, the UV-vis spectrum of ML-WS 2 is basically featureless as is that of F 6 TCNNQ/ ML-WS 2 .For F 6 TCNNQ/d-ML-WS 2 , the UV-vis spectrum exhibits four peaks between 1.8 and 1.0 eV.As demonstrated by ARPES, ground-state CT between d-ML-WS 2 and F 6 TCNNQ leads to negatively charged molecules and concomitantly to the de-doping of d-ML-WS 2 .The formation of molecular anions is here confirmed by the series of absorption peaks at %1.02, 1.21, and 1.38 eV, the typical signature of F 6 TCNNQ À reported in Figure S6, Supporting Information, and in literature. [47,50]ore precisely, the peak at 1.02 eV corresponds to the 0-0 transition, which is followed by the two vibronic progressions at 1.21 and 1.38 eV.In addition, a further peak at %1.52 eV is observed.This peak cannot be ascribed to molecular anions or d-ML-WS 2 .Similar feature have been reported for F6TCNNQ-related doping of polymer films or carbon nanotubes and were attributed to the formation of anion dimers.In the present case, such interpretation cannot be excluded though the formation anion dimer with π-π stacking appears less likely as CT is most likely limited to the interface F 6 TCNNQ molecules, those lying flat on the surface. [37,51,52]Another explanation is that this optical transition relates to an interface feature, that is, an interlayer excitonic state, whose possible origin is discussed later.We finally remark that the defect-related absorption band below 1.5 eV of d-ML-WS 2 is quenched upon F 6 TCNNQ deposition.The quenching of this presumably defect-related band of d-ML-WS 2 , together with the emptying of the electron pocket at Q upon F 6 TCNNQ deposition thus suggests that the CB near the minimum at Q was involved in the initial emergence of the broad, defect-related, absorption band.

Discussion
To examine which levels could be involved in the interlayer exciton suggested earlier, a correlation the absorption features with an energy-level diagram based on the ARPES results appears useful.Schematic diagrams of the frontier energy levels of (d-)ML-WS 2 and F 6 TCNNQ, and their interfaces are shown in Figure 5. There, the BE of the levels is referred to the Fermi level, and in addition the exciton energies determined from the UV-vis measurements are provided.First, Figure 5a depicts the optical transition due to the A exciton in (d-)ML-WS 2 at the K-point, assuming an exciton BE of 0.3 eV, [53] and the filling of the CB at the Q-point.The HOMO and LUMO levels of the neutral F 6 TCNNQ "before contact," that is, assuming vacuum-level alignment with respect to that of (d-)ML-WS 2 , are further displayed in Figure 5b.As F 6 TCNNQ has an EA of 5.6 eV, [40] its LUMO (onset) is then found about 1.3 eV below the Fermi level, but well above the d-ML-WS 2 VBM.The F 6 TCNNQ electronic gap was determined by the gap between the onset of the HOMO and LUMO as observed by direct and inverse photoemission to be 1.9 eV. [40]Note that this electronic gap is actually smaller than the optical gap, probably because the absorption peak of a molecule represents the joint DOS (JDOS) of the material.Therefore, for a consistent interpretation of the UV-vis spectra with respect to the energy levels, it is preferable to consider the (single particle) HOMO and LUMO peak maxima of F 6 TCNNQ at 8.4 and 5.3 eV (with respect to the vacuum level), [40] respectively, as done in Figure 5.As the ionization energy from the VBM of (d-)ML-WS 2 is expected to remain mostly independent of electron doping, the relative energy-level alignment of F 6 TCNNQ/(d-)ML-WS 2 before contact eventually relates to that of the F 6 TCNNQ/ML-WS 2 interface, for which no CT occurs after contact due to the lack of available charge carriers.For the F 6 TCNNQ/d-ML-WS 2 interface, electron transfer from the CB at Q, leading to the formation of molecular anions, strongly renormalizes the F 6 TCNNQ energy levels at the very interface as shown in Figure 5c.The HOMO and SOMO energies are indicated with respect to their peak maxima as determined from ARPES.The SUMO is here positioned 0.6 eV above the Fermi level, by assuming a Coulomb repulsion energy U of 1 eV. [46]he diagram suggests that the 0-0 optical transition of the anion is consistent with a HOMO-SUMO transition assuming a (screened) exciton BE in the range of 0.6 eV.Because of the de-doping of d-ML-WS 2 following the CT, we found from ARPES that the WS 2 VBM and CBM after F 6 TCNNQ adsorption are then located at 1.4 and À0.8 eV BE, respectively.We can now tentatively attribute the origin of the 1.5 eV interface optical transition based on the energy difference of the F 6 TCNNQ and WS 2 energy levels, leaving aside the other interpretation of anion dimer formation as discussed earlier.Considering an interlayer exciton BE in the range of %0.3 eV, [29] we assess that an interlayer exciton could result (within 0.1 eV) from a transition between the  Finally, we comment on why an interlayer exciton could occur at the d-ML-WS 2 /F 6 TCNNQ interface but not at the pristine ML-WS 2/ F 6 TCNNQ one.First, we note that the lack of observation of an interlayer exciton at the pristine ML-WS 2 /F 6 TCNNQ may be related to the minimum energy accessible in these experiments (0.7 eV).Looking at the energy diagram and accounting for an exciton BE of 0.3 eV, one might expect an interlayer exciton with an absorption energy of %0.7 eV, which was at the limit of the experimentally accessible energy range and might have not been observed for this reason.Another possible explanation for the lack of interlayer exciton at the ML-WS 2 /F 6 TCNNQ interface is that the coupling between the organic molecules and the TMDC wave functions is so weak that it prevents the formation of interlayer excited states.In contrast, we reason that the interface dipole resulting from the CT at the d-ML-WS 2 /F 6 TCNNQ interface favors the formation of interlayer excitons, eventually with the electron located in the WS 2 and the hole in the molecule as proposed earlier.Last but not least, the (localized) defect states, though not seen here, may also play a role in the presently observed hybrid transition.

Conclusions
In this work, we demonstrated that the interaction between a strong p-dopant OSCs and ML-WS 2 is strongly dependent on post-growth treatment of the TMDC and its related defect formation.Here, the interaction between an as-grown WS 2 ML and F 6 TCNNQ molecules is characteristic of a physisorptive adsorption and does not involve noticeable ground-state CT.In situ post-growth annealing of the WS 2 ML results in strong n-type doping.Because of the now available charge carriers in the annealed WS 2 ML, pronounced ground-state electron transfer to the subsequently deposited organic molecular layer is observed.The accordingly renormalized interface ground-state electronic structure then enables a modification of the optical properties of the material combination, with the notable emergence of an absorption peak possibly resulting from the formation of F 6 TCNNQ anion dimers or from an interlayer excitonic state, which deserves further studies to be clarified.This shows that the optoelectronic properties of van der Waals systems formed by TMDCs and OSCs can be tuned by defect engineering and it opens new opportunities to investigate the relation of excited states versus ground states at such interfaces.

Experimental Section
WS 2 ML films were grown by CVD on sapphire according to the procedure detailed in Supporting Information (Figure S1, Supporting Information).The employed CVD parameters yielded the formation of a continuous WS 2 ML on sapphire with beyond 90% area coverage and less than 2% multilayer contributions (see Figure S2, Supporting Information).This film morphology warranted sufficient charge-carrier percolation paths for performing ARPES measurements without sample charging, and provided a sufficiently high conductivity of the ML-WS 2 and good electrical contact using metal clamps were assured.F 6 TCNNQ (Novaled GmbH) was deposited in UHV from a resistively heated crucible in a preparation chamber that was connected to the ARPES analysis chamber.The nominal molecular coverage was determined with a quartz crystal microbalance.The ARPES measurements were performed at room temperature in an analysis chamber (base pressure of 2 Â 10 À10 mbar) equipped with a Specs Phoibos 100 hemispherical electron analyzer and using the He I radiation provided by a monochromated helium discharge lamp (consisting of an HIS-13 lamp mounted on a VUV5046 UV monochromator), both from ScientaOmicron.
Optical absorption spectroscopy was performed in air using a Lambda 950 UV/vis/NIR spectrophotometer (Perkin Elmer Inc.).Measurements in solution were performed in dichloromethane (DCM) against a DCM-filled reference cuvette.The solid-state samples were measured against an uncoated sapphire glass slide.
Before deposition of F 6 TCNNQ on ML-WS 2 , the substrate was heated in UHV at %600 K to desorb the surface contaminants and to not induce the sample n-doping.The sample was then annealed in UHV at %900-1000 K before to desorb the molecular film and to produce d-ML-WS 2 , before being measured by ARPES and UV-vis spectroscopy.Before deposition of F 6 TCNNQ on d-ML-WS 2 , the substrate was again heated more briefly at %900-1000 K to get rid of the surface contamination and potentially chemisorbed oxygen.

Figure 2 .
Figure 2. ARPES spectra of the Fermi-level region of 1 Å F 6 TCNNQ/d-ML-WS 2/ sapphire and bare d-ML-WS 2 measured at a) Γ, b) Q, and c) K of the ML-WS 2 BZ.

Figure 3 .
Figure 3. a) Wide-range UV-vis spectra of ML-WS 2 and d-ML-WS 2 before and after deposition of 10 nm F 6 TCNNQ.b) Zoom in the energy region of the A-exciton.Note that in (b) a 3 nm F 6 TCNNQ film was deposited on d-ML for a better visibility of the exciton.

Figure 4 .
Figure 4. a) Zoom in the low-energy region UV-vis spectra of ML-WS 2 and d-ML-WS 2 before and after deposition of 10 nm F 6 TCNNQ.b) Fit of the F 6 TCNNQ/d-ML-WS 2 spectra.

Figure 5 .
Figure 5. Energy-level alignment and possible optical transitions.The energy levels (solid lines) are defined by their peak maxima as determined from the (direct and inverse) photoemission.The displayed interlayer optical transitions account for an exciton binding energy of 0.3 eV.In this diagram, the possible transitions of F 6 TCNNQ anion dimer discussed in the text are not shown.

F 6 TCNNQ
HOMO and the WS 2 CB.Note that in this scenario, we did not account for the possibility of transitions between localized WS 2 defect states and F 6 TCNNQ anions because of the lack of direct observation of the former.