Ping-pong intercomponent energy transfer in covalently linked porphyrin–MoS 2 architectures

Ping-pong intercomponent energy transfer in covalently linked porphyrin–MoS2 Abstract: Molybdenum disulfide nanosheets covalently modified with porphyrin were prepared and fully characterized. Neither the porphyrin absorption nor its fluorescence was notably impacted by covalent linkage to MoS 2 . It was, however, with the help of transient absorption spectroscopy that a complex ping-pong energy transfer mechanism, namely from the porphyrin to MoS 2 and back to the porphyrin, was corroborated. This study reveals the potential of transition metal dichalcogenides in photosensitization processes.


Introduction
Molybdenum disulfide, MoS2, is a typical example of layered transition metal dichalcogenides. In general, their structure is analogous to that of graphene, with an atomic layer of a transition metal sandwiched by two layers of chalcogen atoms. More specifically, each Mo atom is bound to six S atoms forming a three-atom thick monolayer. [1] Exfoliated MoS2-materials have received enormous attention in recent years due to their extraordinary optoelectronic and electrocatalytic properties, especially in the area of energy applications and catalysis. [2,3] Diverse exfoliation strategies have been developed for MoS2. Most notable is the employment of (a) strong intercalants such as Li + , which results in a phase transfer to an octahedral metallic 1T polytype, [4] and (b) non-oxidative Brønsted acids such as chlorosulfonic acid, which offers the benefit of retaining the 2H semiconducting phase. [5] Nowadays additional efforts have been placed to develop strategies for chemically modifying transition metal dichalcogenides. [6] This is meant to fully harness the properties of, for example, MoS2. In fact, incorporation of molecular dopants results in fine tuning of the electronic as well as optical properties of MoS2 and, in turn, enables broadening the spectrum of its applications. In MoS2, the S atoms in the basal plane are, however, rather inert due to saturation. Furthermore, Mo atoms are placed between S layers and, thereby challenging chemical reactivity. Still, covalent functionalization of MoS2 has been accomplished. [7] For example, the basal plane of the 1T-MoS2 polytype has been reacted with organoiodides, [8] while reaction with diazonium salts has been reported for the basal plane modification of 1T [9] as well as 2H-MoS2. [10] In this work we adopt the recently developed methodology based on the addition of 1,2-dithiolanes on S-vacant sites located at the edges of exfoliated 2H-MoS2 nanosheets. [11] In the context of the latter methodology, a true advantage is that diverse 1,2-dithiolanes are easily prepared and they react with exfoliated MoS2 to yield interesting hybrids. Pyrene and phthalocyanines were, for example, interfaced with MoS2. [11] Alternatively, ammonium moieties were grafted onto MoS2 and WS2, which facilitated the electrostatic association of carbon dots [12] and anionic porphyrins. [13] In the resulting ensembles, excited state intercomponent electronic interactions were responsible for the quenching of pyrene, carbon dots or porphyrin fluorescence. Major drawbacks of electrostatically associated ensembles are their lack of stability and of processing in organic solvents. They are also impacted by moderate bindings, especially when compared to the robust and strong bonding in covalently-linked conjugates. Therefore, it is not only imperative, but also timely, to explore the covalent modification of MoS2 with photoactive porphyrins.
Herein, we report on the modification of the edges of exfoliated MoS2 with a 1,2-dithiolane derivative 1, featuring a porphyrin (H2P). The newly prepared H2P-MoS2 2 was comprehensively characterized by spectroscopic, thermal, and microscopy means. Beyond this characterization, we gathered insights into electronic interactions between the porphyrin and MoS2 upon photoexcitation on the femto-to nanosecond timescales.
In parallel, treatment of bulk MoS2 with chlorosulfonic acid allowed the exfoliation of semiconducting nanosheets. [5] Next, reaction of exfoliated MoS2 with 1 yielded H2P-MoS2 2, as summarized in Figure 1. Filtration of the reaction mixture over a PTFE membrane (0.2 µm pore size) followed by extensive washing with dichloromethane, assured the removal of any noncovalently physisorbed 1 as revealed by UV-Vis spectroscopy ( Figure S1). Purified 2 showed reasonable solubility in DMF, benzonitrile, and isopropanol, while it is completely insoluble in dichloromethane and water. Evidence for the success in the covalent modification of MoS2 with H2P came from vibrational spectroscopy. Attenuated total reflectance infra-red ATR-IR measurements revealed for H2P-MoS2 2 ( Figure 2a) the presence of (a) C-H stretching vibration due to the alkyl chain, which connects H2P with MoS2, at 2848 and 2950 cm -1 , (b) amide carbonyl vibration at 1660 cm -1 , and (c) aromatic C=C bending at 1590 cm -1 . We further validated the origin of the IR bands related to H2P within the hybrid by performing a blank experiment in which a mixture of tetraphenylporphyrin and exfoliated MoS2 was processed under the same experimental conditions as those for obtaining 2. The presence of physisorbed tetraphenylporphyrin was not detected in the UV-Vis and emission spectra of the blank material, while it did not show any contribution in the mass loss detected by TGA ( Figure S2-S4). Additional structural information regarding 2 came from Raman spectroscopy. Excitation at 514 nm brought the C=C and C=N bonds of H2P to light in the 1000-1600 cm -1 region ( Figure S5). Focusing on the lower wavenumber region, characteristic modes due to MoS2 were observed, especially upon excitation at 633 nm, due to coupling with the A1 excitonic transition, which in turn produces resonance Raman enhancement of the first and second order vibrational modes. Exfoliated semiconducting MoS2 gave rise to: 2LA(M) at 447 cm -1 , A1g at 406 cm -1 , and E 1 2g at 382 cm -1 (Figure 2b). The difference frequency between A1g and E 1 2g was found to be 24 cm -1 , from which we conclude the existence of few-layered MoS2. [14,15] Considering that the 2LA(M) vibration relates to the S-vacancies, [16] and comparing the Raman spectrum of exfoliated MoS2 with that of 2, the intensity ratio of A1g over 2LA(M) was found to be considerably higher for 2: i.e. 3.6 vs 1.9 for exfoliated MoS2. Complementary Raman mapping assays revealed an increased IA1g/I2LA(M) intensity ratio for 2 as compared to that for exfoliated MoS2 (Figure 2c, d). This observation is in sharp contrast to the recorded IA1g/I2LA(M) maps for the reference material prepared by physisorption ( Figure S6). Overall, this is indicative of Svacancies "healing" by the 1,2-dithiolane species in 2. [9] Likewise, the absence of features at 150, 225, and 325 cm -1 , indicative of the metallic polytype, highlights the semiconducting nature of MoS2 in 2.
Thermogravimetric analysis (TGA) provided information related to the degree of MoS2 functionalization in 2. Exfoliated MoS2 was found to be thermally stable under N2 within 100-800 o C. Notably, the mass loss observed in the modified MoS2 is directly related to the thermal decomposition of the organic addends. Hence, the mass loss of 9% for 2 ( Figure 2e) occurred up to 520 o C is ascribed to the porphyrin. Based on this mass loss, a loading of one H2P moiety per 54 units of MoS2 was calculated. This is well in line with values reported for similar MoS2 functionalization with pyrene and zinc phthalocyanine. [11] Spatially-resolved electron energy loss spectroscopy (SR-EELS) using a scanning transmission electron microscope (STEM) enables investigating, at the local scale, the morphology and chemical composition of complex hybrid nanostructures. [17] Thus, we performed SR-EELS STEM analyses of 2. Figure 3 displays an EEL spectrum-line (SPLI) recorded for a functionalized MoS2 flake -see also Figure S7. Figure 3a corresponds to the high angular annular dark field (HAADF) STEM image of the flake and the SPLI was acquired following the highlighted green line. Typical EEL spectra are displayed in Figure 3b. Each of them corresponds to the sum of four EEL spectra extracted from the two areas marked in the SPLI (red (i) and green (ii) lines, respectively). S-L2,3, Mo-M4,5, C-K and Mo-M2,3 edges are observed in both spectra. The energy loss near edge structure (ELNES) investigations of the C-K edge are displayed in the right panel of Figure S7. In the σ* region (~290-325 eV), two signatures at 291.2 and 293.4 eV can be observed for 2. Such signals are not discerned in the spectra recorded of other carbon containing materials employed as reference. Regarding the main contribution to the π* signal at ~284-290 eV, we can also observe some differences. Here, the C-K edge is dominated by aromatic C contribution (~285. 5 eV [17b-c, 18] ), which in the case of H2P is less pronounced. Furthermore, in the spectra recorded on 2, another feature at ~287.2 eV is clearly seen. This signal can be attributed to pyrrolic (C-N) contributions of H2P. [18c] Importantly, the N-K edge cannot be observed in H2P due to the Mo-M2,3 edge at the same energy range. For all these reasons, the features highlighted in the C-K edge of 2 correspond to porphyrin's contribution. Furthermore, we conclude from the EELS analyses that H2P is homogenously distributed across the surface of the MoS2 flakes in 2 ( Figure S8). Next, focusing on the optical properties of 2, the UV-Vis spectrum was registered in DMF and compared with that of 1. In particular, the UV-Vis of 1 showed a sharp absorption at 418 nm due to the Soret band, accompanied by weaker Q-band absorptions at 516, 550, 596, and 648 nm (Figure 4a). In the UV-Vis of exfoliated MoS2, spectroscopic fingerprints are discernable at 401, 494 (C-exciton), 626 (B-exciton), and 681(Aexciton) nm (Figure 4a). The UV-Vis of 2 is best described as superimposition of the individual spectra, namely 1 and exfoliated MoS2. The spectrum is, for example, dominated by the characteristic Soret-band absorption of H2P at 418 nm in addition to the MoS2 related transition at 679 nm. From their similarity we deduce that the ground-state lacks appreciable electronic interactions.
Excited-state electronic interactions also appeared suitable to probe interactions between H2P and MoS2 in 2. Specifically, upon excitation at the Soret-band absorption at 420 nm, strong fluorescence, which was centered at 651 and 715 nm ( Figure  4b), was found for 1. In TCSPC measurements, a fluorescent lifetime of 10.3 ns ( Figure S9) was determined for 1 with DMF as 10.1002/anie.201914494 Accepted Manuscript

Angewandte Chemie International Edition
This article is protected by copyright. All rights reserved. solvent. 420 nm photoexcitation of 2 resulted, however, in no significant quenching of the porphyrin centered fluorescence relative to that of 1 (Figure 4b). For 2, the fluorescence maxima were observed at 649 and 714 nm, and the fluorescence lifetime was 10.3 ns ( Figure S9). As a complement to the aforementioned, we probed 1, exfoliated MoS2, and 2 by means of femtosecond transient absorption spectroscopy (fsTAS) using 420 nm as excitation wavelength (Figures S10 and S11, and Figure 5, respectively). For 1, ground-state bleaching at 517 and 653 nm are observed early on in the experiments. In addition, positive absorptions evolved at 442, 538, 574, 623, and 690 nm ( Figure S10). Global analysis revealed on the fsTAS timescale the presence of three exponential decays. These three decays are followed by a much slower decay, whose dynamics are outside of the timescale of our experiments. Therefore, a target model was applied to fit the transient absorption data (Figure 6, right). It involves an initial population of second singlet excited state (S2), consistent with excitation into the Soret-band absorption, which internally converts to first singlet excited state (S1) in 16 ps. The latter intersystem crosses in 10 ns to the triplet manifold (T1), which, in turn, decays back to the ground state on the microsecond timescale. Additionally, aggregates of 1 are present in the sample and they were found to decay to the ground state in 142 ps. These dynamics resemble very closely those already reported for a variety of porphyrins. [19,20]  For exfoliated MoS2, ground-state bleaching evolved at 476, 623, and 683 nm together with positive absorptions at 591, 655, and 743 nm ( Figure S11). Global analysis of the data revealed five exponential decays that follow light absorption. The first three species, which take place with 2.7, 7.3 and 250 ps, involve biexciton (be) and trion (tr) formation as well as the decay of these many-body particles, respectively, into single excitons (se). Single excitons then diffuse in 1.8 ns across the layers in the z direction (de), before they recombine to recover the ground state in 21.2 ns (Figure 6, left). These dynamics resemble those already reported for exfoliated MoS2. [21] For 2, excited states of both components are observed to participate in the decay cascade ( Figure 5). Selective excitation into the Soret-band absorption of 2 affords a photoinduced absorption at 450 nm and ground-state bleaching at 617 and 679 nm. Please note that, while the first belongs to the porphyrin, the last two are fingerprints of MoS2. Considering that, at the excitation wavelength, porphyrin absorptions in 2 are around 20 times stronger than those of MoS2 (Figure 4a), the similar intensities of the differential features at 450 and 679 nm imply an ultrafast energy transfer (<1 ps) from S2, which is populated upon 420 nm excitation, to MoS2. Upon global analysis, we noted four exponential decays on the fsTAS timescale and an additional decay, which is, however, outside of our detection range. Our interpretation is based on a kinetic target model, which is depicted in the middle panel of Figure 6. It considers an initial excited state population governed by a porphyrin-centered S2, with a minority of a MoS2-centered excited state. This is followed by a non-quantitative ultrafast energy transfer from H2P to MoS2, which results in comparable populations of H2Pcentered (70%) and MoS2-centered (30%) excited states. Afterwards, the same three species, which govern the deactivation of exfoliated MoS2, and, which result in the 10.1002/anie.201914494 Accepted Manuscript

Angewandte Chemie International Edition
This article is protected by copyright. All rights reserved.
formation of single excitons, are discernible. The times that it takes to interconvert are 2.5, 7.5 and 204 ps, respectively. Once populated, the single exciton of MoS2 is subject to a second energy transfer. This time it is, however, from the single excitons of MoS2 to the S1 of H2P. This energy transfer takes 1.75 ns. From here, the porphyrin decay proceeds via intersystem crossing to the corresponding triplet excited state in 14 ns, and ground state recovery on the microsecond timescale. This mechanism is also based on the energetics of 2. The S1 of 1 with its fluorescence at 1.73 eV (715 nm) and the MoS2 fluorescence at 1.82 eV [22] render the second energy transfer a downhill process. With the ping-pong energy transfer model we rationalize the nearly identical steady-state photophysical characteristics of 1 and 2 upon photoexcitation.

Conclusion
In summary, we accomplished the covalent grafting of porphyrin 1 onto exfoliated MoS2, forming novel H2P-MoS2 2, which was fully characterized. Full-fledged photophysical investigations based on steady-state and time-resolved measurements corroborate that the decay of the photoexcited porphyrin involves a ping-pong energy transfer to and from MoS2. Our findings suggest that transition metal dichalcogenides have great potential in photosensitization. Such hybrid materials may be useful in energy conversion applications.