Noncovalent Liquid Phase Functionalization of 2H-WS2 with PDI: An Energy Conversion Platform with Long-Lived Charge Separation.

Transition metal dichalcogenides are attractive 2D materials in the context of solar energy conversion. Previous investigations have focused predominantly on the properties of these systems. The realization of noncovalent hybrids with, for example, complementary electroactive materials remains underexplored to this date for exfoliated WS2. In this contribution, we explore WS2 by means of exfoliation and integration together with visible light-absorbing and electron-accepting perylene diimides into versatile electron-donor acceptor hybrids. Important is the distinct electron-donating feature of WS2. Detailed spectroscopic investigations of WS2-PDI confirm the electron donor/acceptor nature of the hybrid and indicate that green light photoexcitation leads to the formation of long-lived WS2•+-PDI•- charge-separated states.

Introduction

In the last decade, two-dimensional (2D) materials have attracted huge interest owing to their atomically thin geometry. Tungsten disulfide (WS2), albeit less explored than molybdenum disulfide (MoS2), is a typical example of layered transition metal dichalcogenides (TMDs). WS2 adopts a hexagonal, layered structure similar to the aforementioned MoS2 with the metal atom in a triagonal prismatic coordination, sandwiched between two atomic layers of sulfur atoms. As such, strong in-plane covalent W–S bonds are combined with weak van der Waals couplings between neighboring layers. By virtue of weak interlayer interactions, bulk WS2 is easily exfoliated to yield few layer or monolayer WS2.[1]

Chemical exfoliation of TMDs by means of intercalation-assisted exfoliation in liquids commonly generates large quantities of single layers. Most notable is the well-known use of n-butyllithium dissolved in hexane.[2] Intercalation with alkali metals alters, however, the monolayers with respect to their structure and, thus, their properties. As a matter of fact, upon this treatment, the thermodynamically stable, semiconducting 2H-phase (trigonal prismatic) is converted into the metastable, metallic 1T-phase (distorted octahedral). However, semiconducting 2H-WS2 is maintained if sonication-assisted liquid-phase exfoliation of bulk WS2 is employed. In this case, suitable organic solvents or water in the presence of surfactants are needed.[3] Designed centrifugation approaches allow us to obtain monolayers with improved yields.[4]

As one of the key characteristics, monolayer 2H-WS2 is a direct-gap semiconductor (∼2.0 eV), contrary to its bulk counterpart with its indirect band gap (∼1.3 eV).[5] The electrical, optical, and transport properties of 2H-WS2 pronouncedly depend on the stacking of the WS2 sheets, which can be exploited for tuning these properties. Controlling the electronic behavior of WS2, which is governed by excitonic transitions, is mandatory for the realization of energy conversion devices. Importantly, molecular chemistry approaches are uniquely suited to address the aforementioned.[6−9]

Noncovalent interactions assist in tailoring the properties of nanomaterials. This was first explored for graphene[10] and later transferred to inorganic analogs, such as black phosphorus[11] and MoS2[12] as the representative of TMDs. Noncovalent functionalization of TMDs has predominantly been established on surfaces, except for few exceptions.[13,14] Leading examples are chemical vapor deposition grown MoS2 and tungsten diselenide (WSe2).[15] Studies in the liquid phase remain elusive. A combined approach, that is covalent and noncovalent bonds in dispersion, was shown for MoS2 and WS2via defect functionalization and electrostatic interactions with porphyrins.[16] A theoretical work on noncovalent interactions of WS2 was implemented for small fullerenes (B12 and C20)[17] as well as NH3 and H2O molecules.[18]

Herein, we report for the first time on the noncovalent liquid phase functionalization of 2H-WS. As a suitable organic ligand (OL) allowing to bind via pronounced σ (WS)−π (OL) stacking interactions, we opted for an ethylenediaminetetraacetic acid (EDTA) perylene diimides (PDI) derivative (see Figure S1, left and Scheme ), which is known from the literature.[19] What stands out among PDIs are their strong light harvesting and good electron accepting features.[20] Leading examples are PDI-based noncovalent functionalization of 2D materials like graphene, boron nitride, black phosphorous,[21] and also 3D TMDs. At the same time, the binding to WS should be facilitated by electron donor–acceptor interactions between WS and PDI, in addition to stacking. Importantly, it has been demonstrated that TMDs are complementary electron donors, which were integrated into covalently linked MoS2-PDI conjugates.[22] As a consequence, we expected a robust photoenergy conversion platform. Our WS–PDI hybrid was comprehensively characterized by spectroscopic methods, and inter alia we gathered insights into the electronic interactions between WS and PDI upon photoexcitation on the femto- to nanosecond timescales.

Scheme 1

Illustration of the Sonication-Assisted Liquid Phase Exfoliation of Bulk 2H-WS (Composed of Three-Atomic S–W–S Sandwich Layers) with Sodium Cholate (Shown as Blue Spheres) and the Subsequent Noncovalent Functionalization of the Purified WS Nanosheets with PDI (Core in Red and Side Groups in Gray) in an Aqueous Solution Resulting in WS–PDI

Results

Exfoliated WS nanosheets were produced by probe sonication of commercially available WS powder in water, utilizing sodium cholate (SC) as a stabilizing surfactant. The resulting polydisperse dispersion of WS nanosheets was purified via multiple centrifugations with varying rotational speed, as previously described.[4] The few layer material was further processed by successive washing with water to remove the SC surfactant and followed by centrifugation-induced precipitation. The WS nanosheets were then redispersed in an aqueous solution of PDI, stirred for 1 h, and finally centrifuged to afford WS–PDI (see Figure S1, right and Scheme ).

PDI (see Figure S1, left) is water soluble under slightly basic conditions and was dissolved in a potassium hydroxide (KOH) solution (pH ≈ 10).[19] Absorption spectroscopy corroborates the aggregation of PDI in solution due to intermolecular π–π stacking interactions. In Figure S2 (black absorption spectrum with the typical features of PDI), the ratio between the (0,0)- and (0,1)-absorptions at 542 and 501 nm, respectively, is around 0.55, suggesting predominantly aggregated species in solution.[19] The absorption spectrum of WS (see Figure , black curve) is characterized by three excitonic transitions, namely, the A, B, and C excitons at 623, 520 (shoulder), and 337 nm, respectively.[23] The absorption spectrum of WS reveal the WS excitonic bands, but the absence of any PDI characteristics (see Figure , red curve). As a matter of fact, the PDI-centered absorbance is only discernible in the form of an intensity increase of the 520 nm shoulder. Interestingly, in addition to the excitonic WS features, WS–PDI gives rise to a new transition at 571 nm. We assign this to a charge-transfer (CT) band, which is 29 nm red-shifted compared to the (0,0)-absorption of the PDI reference. This suggests strong electron donor/acceptor electronic interactions in the ground state.a Please note that covalently linked MoS2-PDI conjugates fail to exhibit similar CT bands or any evidence for intercomponent ground-state electronic coupling.[22] A likely rationale is centered around the noncovalent nature of WS–PDI. Here, PDIs are accommodated at positions, where the electronic interactions with WS are maximized. According to the second derivative of the absorption spectrum (see Figure , inset), the WS excitonic bands lack any appreciable shifts upon functionalization. Similar conclusions were drawn for a previously reported covalent MoS2–PDI conjugate.[22] It signifies that the degree of functionalization is insufficient to significantly alter Coulomb interactions between electrons and holes within WSvia dielectric screening.[24]

Figure 1

Normalized absorption spectra of WS (black) and WS–PDI (red). Inset: second derivative of the absorption spectra.

Concerning the polydispersity of the WS nanosheets in lateral size and thickness, it has recently been shown that absorption spectra allow for the simultaneous quantification of the nanosheets dimensions in dispersion.[25] The average length ⟨L⟩ (longest dimension) and layer number ⟨N⟩ of the exfoliated WS were determined to be ⟨L⟩ = 60 nm (Ext ratio 235/295 = 1.5) and ⟨N⟩ = 3.5 (A exciton centered at 624.5 nm), respectively. By means of high-angle annular dark-field scanning transmission electron microscopy (HAADF STEM) measurements, we probed the morphology of WS–PDI (see Figure S3). Even though the flakes exhibit a rather broad size distribution from 20 to 100 nm, the average layer number was determined to be ⟨N⟩ = 3.5.

Strong interactions between WS and PDI were further corroborated by Raman spectroscopy. Of great importance is the strong quenching of the PDI fluorescence in WS–PDI. The Raman spectrum (see Figure , red spectrum) shows the WS characteristics, namely, the 2LA(M) mode at 352 cm–1 (overlapping with the first order 1E2g(Γ) mode at 356 cm–1) and the A1g(Γ) mode at 418 cm–1, in line with the WS reference (see Figure , black spectrum).[26] Additionally, the PDI vibrational modes in WS–PDI appear at 1301/1380 cm–1 (in-plane ring “breathing”), 1456 cm–1 (ring deformation), 1573/1588 cm–1 (in-plane C–C stretching), and 1699 cm–1 (imide bending), as well as the imide Raman bands of the branched EDTAs between 2600 and 3200 cm–1. All of the aforementioned proves the successful functionalization of WS and the formation of WS–PDI.

Figure 2

Raman spectra of WS reference (black) and WS–PDI (red).

Furthermore, fluorescence spectroscopy provided valuable insights into the electronic communication in the excited state within WS–PDI. Dispersed WS nanosheets in water (see Figure , left) show weak photoluminescence (PL) at 612 nm (2.03 eV), when photoexcited between 340 and 460 nm. This is attributed to the direct-gap PL of a WS monolayer.[27] It should be noted that the quantum efficiency of the WS PL is dependent on the number of layers. The direct excitonic transition at the K point is energetically favored for monolayers. With the increasing number of layers, the indirect transition from the T point of the conduction band to the Γ point of the valence band becomes energetically favored and leads to a dramatic decrease of the PL quantum efficiency.[27] Here, the WS PL in the reference spectrum indicates the presence of a high monolayer content in the dispersion. Interestingly, after the addition of PDI to the exfoliated WS, we observed a strong fluorescence quenching of the WS PL (see Figure , right). This quenching was seen for heterojunctions comprising a 3,4,9,10-perylene tetracarboxylic dianhydride thin film deposited onto a WS monolayer.[28] The PL quenching rate of WS excitons was determined to be 81 ± 2%, which is highly consistent with the data presented by Liu et al.[28] Our PL quenching is attributed to a charge transfer process across the WS–PDI interface (vide infra).

Figure 3

Top: 2D PL mapping of WS reference (left) with a PL maximum at 612 nm and WS–PDI (right) with a quenched PL. Bottom: magnification around the 612 nm region, for WS reference (left) and WS–PDI (right).

The steady-state fluorescence spectrum of PDI shows maxima at 543 and 586 nm (see Figure S2). Due to aggregation, the fluorescence quantum yield is with 0.26 notably lower than that for nonaggregated PDI.[29] In stark contrast, WS–PDI (see Figure , right) is nearly nonfluorescent. This is a consequence of strong intercomponent electronic interactions in the excited state, which opens additional nonradiative decay channels absent in PDI.b

We then turned to femtosecond transient absorption spectroscopy (fs-TAS) in water and at room temperature. fs-TAS of the PDI reference upon 500 nm photoexcitation leads to strong ground-state bleaching at around 480 nm as well as photoinduced absorptions at 539 and 600 nm, which reach all the way to 1000 nm. Global analysis of the data was based on a three-species sequential kinetic model (see Figure S4). The respective lifetimes are 3.8 ps, 271 ps, and 3.3 ns. The spectral shape of the first two species is fairly similar with a 480 nm minimum and 538, 600, 700, 785, and 817 nm maxima. Both states are ascribed to PDI aggregates: the first species is a vibronically excited hot-S1 and the second species is the vibronically relaxed S1. The third species, which is ascribed to the fluorescent S1 of nonaggregated PDI, features minima at 471, 540, and 590 nm as well as maxima at 700 and 930 nm. The minima correspond to the ground-state absorption and stimulated emission, respectively.[30]

In the case of the WS reference, 500 nm photoexcitation leads to ground-state bleaching of exfoliated WS at 420, 517, and 620 nm together with the formation of positive absorptions at 465, 540, and 640 nm (see Figure S5). Global analysis was carried out, using a model, which resembles the dynamics already established in the analysis of other TMDs, such as MoS2.[31,32] By virtue of this model, five exponential decays with lifetimes of 0.4, 2, 21, and 129 ps were deconvoluted. Notably, the fifth and final species is persistent on the timescale of our fs-TAS experiments. The aforementioned states include, respectively, the exciton generation (eg), biexciton (be), and trion (tr) formation as well as the decay of these many-body particles into single excitons (se) followed by excitons that diffuse across the layers in z-direction (de) before they recombine and recover the ground state.

At first glance, the transient absorption spectra of WS–PDI, similar to the ground-state absorption, are dominated by WS-related features in the visible range. In the near-infrared range, a closer look at the 3D differential absorption heat map assists in revealing a sharp positive signal at 986 nm (see Figure , top left panel). This feature is PDI centered and of utmost importance as it is the fingerprint absorption of the one-electron reduced form of PDI, PDI.[23] It is subject to a slow decay throughout the femtosecond and nanosecond timescales (please see nanosecond transient absorption spectroscopy (ns-TAS) experiments in Figures , S6, and S7). From its presence, we conclude the coexistence of WS-centered excited states, on the one hand, and charge-separated (CS) states, on the other hand. To account for this fact, target analysis required an elaborated kinetic model to fit the data (see Figures and 5). The model is based on the parallel decay of two different branches. At the origin of these two branches is the WS (eg) state, which is formed upon light absorption. It consists of the WS-related features (see Figure , bottom left panel, black curve), namely, minima at 430, 517, 620 nm, maxima at 480, 550, 657 nm, and a broad feature stretching from 990 to 1400 nm. It is within 0.5 ps that the WS (eg) state deactivates in parallel via two different branches—vide infra.

Figure 4

Top left: 3D differential absorption heat map obtained upon 500 nm photoexcitation in fs-TAS with WS–PDI, in water and at room temperature. Top right: Concentration evolution as a function of time. Bottom left: Species-associated spectra of the WS-related states in WS–PDI. Bottom right: Species-associated spectra of the CS states in WS–PDI.

Figure 5

Kinetic models used to fit the fs-TAS and ns-TAS data of PDI, WS, and WS–PDI on the left, in the center, and on the right, respectively.

One of these two branches relates to WS-centered excited states and dynamics thereof. It features three different states with lifetimes of 5.4 ps (blue), 21 ps (green), and 110 ps (purple), respectively (see Figure . Their spectral shapes as well as their lifetimes, bottom left and top right panel) are in sound agreement with be, tr, and se, respectively. Our assignments are based on the conclusions, which were drawn from the experiments performed with the WS reference—vide supra. Importantly, no sizeable PDI-related absorptions are discernible across the visible range for these excited states. This confirms their nature as an exclusive WS-centered species. The second branch consists likewise of three species (see Figure , top right panel). In this instance, the respective lifetimes are 0.6 ps (red), 1.3 ns (yellow), and >10 ns (cyan). What is common to all three of them is their spectral signatures (see Figure , bottom right panel). However, they differ from those observed for the WS- and PDI references. The most prominent characteristics are minima at 480 and 568 nm, a broad maximum around 730 nm, and a sharp maximum at 986 nm. Notable is the fact that the 568 nm minimum mirror images the CT absorption seen in the ground state, while the remaining features resemble the fingerprint absorption of the one-electron reduced form of PDI.[33] In other words, our target analysis underscores the formation of WS–PDI CS states evolving between the electron-donating WS and the electron-accepting PDI.

To study the charge-recombination dynamics in detail, ns-TAS was deemed necessary. Figure S7 shows the 3D differential absorption heat map and global fits of the data. Two species with nearly identical differential spectra were deconvoluted. Their lifetimes are 6 and 162 ns, respectively.c Both species are in agreement with the fs-TAS analyses and assigned to the WS–PDI CS states. It is interesting to note that only the latter species reinstates the ground state via charge recombination.

Single wavelength analyses involving the fingerprint absorptions of PDI reveal a multiexponential decay on a timescale of up to several hundred nanoseconds (see Figures S6 and S7). Such a finding is rationalized on grounds of further separating the charges in the WS–PDI CS state. Very likely is that the positively charged hole on the WS electron donor is subject to charge-shifting via hopping. In light of the latter, the initially formed CS state (CS1) transforms via, for example, a charge shift into a second CS state (CS2), in which the charges are more spatially separated from each other. As such, CS1 yields sequentially CS2 and the third CS state (CS3) within 0.6 and 1.3–6.0 ns, respectively, before charge recombination takes over. Remarkable is that it takes 162 ns until the ground state is recovered from CS3 (see Figure S7). Given the almost linear decay of the 986 nm signal, it is likely that other different CS states with different electron–hole distances and therefore also different lifetimes are present in WS–PDI.d

Conclusions

In conclusion, we succeeded for the very first time in synthesizing a WS–PDI hybrid by linking electron-donating, exfoliated WS to PDI, which serves as a light harvester and electron acceptor. Thoroughly performed characterization revealed the successful formation of WS–PDI. In the ground state, a CT absorption, which stems from strong electronic communication, is only discernible in WS–PDI, but not in the individual components. These strong electronic interactions result in the excited state in the population of a long-lived (162 ns) PDI–WS CS state. Our results prompt to explore WS–PDI hybrids in catalytic schemes. To this end, the long-lived nature of the CS states should be employed to drive subsequent charge shifts to catalytically active reaction centers.