Letter pubs.acs.org/NanoLett
Effects of Direct Solvent-Quantum Dot Interaction on the Optical Properties of Colloidal Monolayer WS2 Quantum Dots Ho Jin,† Bongkwan Baek,§,∥,⊥ Doyun Kim,† Fanglue Wu,‡ James D. Batteas,†,‡ Jinwoo Cheon,§,∥,⊥ and Dong Hee Son*,†,§ †
Department of Chemistry, Texas A&M University, College Station, Texas 77843, United States Department of Materials Science and Engineering, Texas A&M University, College Station, Texas 77843, United States § Center for Nanomedicine, Institute for Basic Science (IBS), Seoul 03722, Republic of Korea ∥ Yonsei-IBS Institute, Yonsei University, Seoul 03722, Republic of Korea ⊥ Department of Chemistry, Yonsei University, Seoul 03722, Republic of Korea ‡
S Supporting Information *
ABSTRACT: Because of the absence of native dangling bonds on the surface of the layered transition metal dichalcogenides (TMDCs), the surface of colloidal quantum dots (QDs) of TMDCs is exposed directly to the solvent environment. Therefore, the optical and electronic properties of TMDCS QDs are expected to have stronger influence from the solvent than usual surface-passivated QDs due to more direct solvent-QD interaction. Study of such solvent effect has been difficult in colloidal QDs of TMDC due to the large spectroscopic heterogeneity resulting from the heterogeneity of the lateral size or (and) thickness in ensemble. Here, we developed a new synthesis procedure producing the highly uniform colloidal monolayer WS2 QDs exhibiting well-defined photoluminescence (PL) spectrum free from ensemble heterogeneity. Using these newly synthesized monolayer WS2 QDs, we observed the strong influence of the aromatic solvents on the PL energy and intensity of monolayer WS2 QD beyond the simple dielectric screening effect, which is considered to result from the direct electronic interaction between the valence band of the QDs and molecular orbital of the solvent. We also observed the large effect of stacking/separation equilibrium on the PL spectrum dictated by the balance between inter QD and QD-solvent interactions. The new capability to probe the effect of the solvent molecules on the optical properties of colloidal TMDC QDs will be valuable for their applications in various liquid surrounding environments. KEYWORDS: Colloidal quantum dot, monolayer WS2, transition metal dichalcogenide, solvent-quantum dot interaction, photoluminescence
M
colloidal TMDC nanocrystals found applications in various surrounding environments such as imaging in biological systems and catalysis in solution phase,8,9,15,16 analogous to more common colloidal semiconductor QDs.17−20 Unlike in other QDs with dangling bonds on the surface providing the linkage for the ligands, TMDC QDs have no dangling bonds on their basal planes except at possible defect sites and edge. Therefore, the surface of TMDC QDs in solution is directly exposed to the solvent environment without intervention from the surface-bound ligands, although they can be introduced at the defect sites or edge.21−23 Naturally, QDsolvent and inter QD interaction that can modify the electronic and optical properties of the QDs are expected to be stronger in TMDC QDs than those with surface-bound ligands. This is an important issue since the common applications of colloidal
onolayer and few-layer transition metal dichalcogenide (TMDC) materials attracted much attention recently due to their unique electronic, optical and transport properties resulting from the confinement of charge carriers in twodimension (2D).1−7 TMDC nanostructures with controlled thickness have been fabricated typically via chemical vapor deposition or epitaxial growth as a thin film supported on a substrate, which is particularly useful for exploring the properties of 2D-confined exciton sensitive to the thickness and interlayer interactions. Efforts to create TMDC nanostructures with controlled lateral dimension as well as the thickness have also been made with the prospect of combining the lateral quantum confinement with properties originating from the 2Dconfined charge carriers. Such nanostructures were often created in colloidal nanocrystal form employing liquid-phase exfoliation, Li-assisted exfoliation, electrochemical synthesis and hot injection method,8−13 while the patterned structure on a substrate was also created via electron beam lithography.14 Because of the dispersibility in various liquid media, the © 2017 American Chemical Society
Received: August 7, 2017 Revised: October 26, 2017 Published: October 27, 2017 7471
DOI: 10.1021/acs.nanolett.7b03381 Nano Lett. 2017, 17, 7471−7477
Letter
Nano Letters
uniform monolayer MoS2 QDs, indicating the applicability our strategy to a wider range of TMDC materials (Figures S4 and S5). Figure 1a,b shows the PL spectra of the colloidal solution of WS2 QDs prepared respectively by the previously reported
QDs utilize the photoluminescence or interfacial charge/energy transfer process directly affected by the modification of the electronic structure by the surrounding. So far, understanding the effect of QD-solvent and inter QD interaction on the optical properties in colloidal TMDC QDs has been hampered by the structural (e.g., lateral size and thickness) heterogeneity resulting in the spectroscopic heterogeneity.8,11,12 Such heterogeneity of the ensemble prohibited reliable optical characterization correlated with the structure in most of the colloidal TMDC QDs produced by top-down methods (e.g., liquid exfoliation) despite their ability to produce the QDs in a large quantity. Here, we successfully prepared the colloidal solution composed of only monolayer WS2 QDs with narrow size distribution by combining a modified liquid exfoliation method and the newly developed procedure separating the monolayer and multilayer QDs by exploiting their difference in the QDsolvent interaction. Using the solution of monolayer WS2 QDs exhibiting a well-defined spectroscopic signature distinct from the multilayer WS2 QDs, we observed the unusually large effect on the photoluminescence (PL) energy and intensity of monolayer WS2 QD by the aromatic solvent molecules beyond the simple dielectric screening effect,24 which is absent in surface-passivated colloidal II−VI QDs. This is considered to result from the direct electronic coupling between WS2 QD and aromatic solvent molecules possessing spatially and energetically accessible molecular orbital for the electronic coupling with valence band of the QDs. We also observed the effect of stacking/separation equilibrium on the PL spectrum dictated by the balance between inter QD and QD-solvent interactions. The results from this study on the effect of the solvent molecules on the optical properties of colloidal TMDC QDs will be valuable for their applications in various liquid surrounding environments. Colloidal QDs of WS2 used in this study were prepared via liquid exfoliation with a few important modifications from the previously reported method,8,15,25 which allowed us to remove the large spectroscopic heterogeneity commonly observed in the colloidal QDs of various TMDC materials. The details of synthesis are described in Supporting Information. Briefly, bulk WS2 powder in deoxygenated dimethylformamide (DMF) was sonicated for 3 h under N2 atmosphere to prevent possible oxidation by the dissolved O2. The solution was subsequently diluted in DMF and heated at 140 °C under N2 atmosphere to obtain the final product. These QDs were produced in the absence of any ligands that can potentially bind at the edge or defect sites. There are two important differences from the previous method.8 One is using the solvent after thorough “removal of the dissolved O2” for sonication by bubbling an inert gas for over 30 min instead of using as-received solvent. The other is “dilution” of the solution before the final heating step. Instead of heating the as-obtained sample solution after sonication, the solution was diluted 10 times with DMF before heating. These two modifications are crucial in obtaining monolayer WS2 QDs of uniform size exhibiting well-defined spectral characteristics by exploiting the direct QD-solvent interaction uninhibited by ligands (Figure S2). 1H and 13C NMR measurements on the concentrated solution of monolayer WS2 QDs in deuterated benzene (C6D6) showed no sign of organic ligand that may bind to the surface of the QDs, confirming the largely bare surface of the QDs (Figure S3). While we focus on WS2 QDs in this study, the same procedure applied to MoS2 powder gave colloidal solution of
Figure 1. (a,b) PL spectra of WS2 QDs synthesized by previously reported method8 (a) and new method reported here (b) at varying excitation wavelengths. (c) PL spectra of WS2 QDs in DMF before (cyan) and after (blue and green) adding pentane under 370 nm excitation. Blue and green curves are the PL in DMF and pentane phase, respectively, after the completion of solvent extraction. Inset shows a picture of the PL from WS2 QDs in DMF before adding pentane. (d) PL spectra of monolayer WS2 QDs (blue) in DMF phase and multilayer WS2 QDs (green) in pentane phase obtained via repeated solvent extraction. Inset shows the pictures of the PL from DMF and pentane phase.
method and new method described above at varying excitation wavelengths. The shift of broad PL spectra with excitation wavelength shown in Figure 1a, commonly observed in the colloidal nanocrystals of many TMDC materials, was interpreted as the size heterogeneity. This is an inherent problem of the liquid exfoliation method, which prohibits the reliable spectroscopic characterization of TMDC QDs by ensemble measurements. In contrast, the PL spectra of QDs produced using the new method exhibit only two spectroscopically distinct species; one centered at 400 nm with multiple peaks and the other at 500 nm with a single broad peak. An important observation here is that the PL at 400 nm is very similar to the single-particle PL of monolayer WSe2 QDs from our recent study,12 which shows the same multiple-peak feature. This feature has been assigned to the vibronic coupling to the high frequency local vibrational mode not present in bulks, which may come from the edge bonds in the QDs. The presence of this vibrational mode was corroborated by Raman measurement as will be discussed later. Such feature is absent in Figure 1a due to the large ensemble heterogeneity. This suggests that the solution of QDs shown in Figure 1b contains the monolayer WS2 QDs with sufficiently narrow size distribution preserving the clear vibronic feature even in the 7472
DOI: 10.1021/acs.nanolett.7b03381 Nano Lett. 2017, 17, 7471−7477
Letter
Nano Letters
Figure 2. (a) TEM and (b) HRTEM image of monolayer WS2 QDs remaining in DMF phase after the repeated two-phase solvent extraction. Inset of (b) shows FFT pattern of the QD image. (c,d) TEM images of multilayer WS2 QDs transferred to pentane phase. (e,f) STM images of monolayer WS2 QDs deposited on Au(111). Inset in (f) shows the profile of the apparent height of the QDs on the horizontal line. (g,h) Histogram of QD diameter and apparent height.
the histogram of the QD diameter (Figure 2g) constructed from the analysis of >150 QDs indicate that the lateral size distribution of the QDs is narrow with average diameter of ∼3.5 nm. The identity of these QDs as the monolayer WS2 QDs is evidenced by the small (±0.5 Å) fluctuation of the apparent height from the average value of 2.4 Å in the STM measurement (Figure 2f,h), which is very close to that of monolayer WS2 nanoclusters directly grown on Au(111) surface.28 The observation of PL polarization anisotropy is another evidence in support of the monolayer thickness of these QDs as discussed below. The TEM image of the nanocrystals transferred to pentane phase (Figure 2c), on the other hand, shows the signature of multiple superimposed layers with the lateral size similar to monolayer WS2 QDs. The multilayered structure is more clearly seen in Figure 2d showing the side view of the QDs with the interlayer gap matching that of bulk WS2 (6.3 Å),29,30 further confirming their identity as the multilayer WS2 QDs. The monolayer and multilayer WS2 QDs each separated into DMF and pentane phase exhibit distinctly different PL spectra and PL polarization anisotropy. The monolayer WS2 QDs exhibit the vibronic feature in the PL and PL excitation (PLE) spectra appearing as the multiple peaks (Figure 3a) and linear PL polarization anisotropy (Figure 3b), which are very close to the recent observation made from single-particle study of monolayer WSe2 QDs produced via hot-injection method.12 The frequency of the dominant vibrational mode in the vibronic structure of the PL spectrum, which was estimated from the difference in the energies of the two adjacent peaks, is 1200−1300 cm−1. This mode has much higher frequency than in-plane optical phonon (E12g) and out-of-plane (A1g) modes of WS2 at 350−420 cm−1 region and is absent in Raman spectra of the bulk and monolayer sheets of WS2.31−33 However, Raman spectrum of the colloidal monolayer WS2 QDs obtained here (Figure 3e) shows a broad peak near 1300 cm−1, indicating the presence of an additional mode participating in the vibronic coupling, consistent with PL spectrum. While the assignment of
ensemble PL spectrum. The identity of the species with PL centered at 500 nm, on the other hand, cannot be straightforwardly determined from the PL spectrum alone. The confirmed identity of the two species at 400 and 500 nm as the monolayer and multilayer WS2 QD, respectively, will be discussed in detail below. The presence of only two spectroscopically distinct species in newly prepared WS2 QDs prompted our attempt to exploit the potential difference in the interaction with solvent that may enable the separation via solvent extraction. Since the colloidal WS2 QDs was initially prepared in weakly polar DMF, nonpolar pentane was used as the second phase immiscible with DMF. Figure 1c compares the PL spectrum of as-synthesized WS2 QDs in DMF before adding pentane (cyan) with those in DMF (blue) and pentane (green) phase after the equilibration between the two solvent phases is complete. Surprisingly, only one species with PL centered at 500 nm transferred from DMF to pentane phase while the other remained in DMF phase, indicating that the two species have sufficiently different interaction with solvent enabling the selective extraction of one species. A complete separation of the two species, each exhibiting distinct and excitation wavelength-independent PL as shown in Figure 1d, was achieved via multiple solvent extraction (Figure S6). It is noteworthy that the same monolayer WS2 QDs were obtained in other solvents possessing similar properties to DMF, such as N-methly-2pyrrlidone, demonstrating the robustness of the synthesis procedure in different solvent environment (Figure S7). The representative TEM images of the two separated species in DMF and pentane phase are shown in Figure 2a,b and Figure 2c,d, respectively. Figure 2a shows the isolated particles of ∼3.5 nm in lateral size with clearly identifiable lattice. The highresolution TEM (HRTEM) image and Fourier transform pattern in Figure 2b show the well-defined lattice of a WS2 QD lying flat on the substrate with 2.7 Å lattice spacing in (100) direction.26,27 Scanning tunneling microscopy (STM) image of the QDs deposited on Au(111) surface shown in Figure 2e and 7473
DOI: 10.1021/acs.nanolett.7b03381 Nano Lett. 2017, 17, 7471−7477
Letter
Nano Letters
The colloidal solution of monolayer WS2 QDs prepared in this study stayed stable for several months without any change of PL spectrum. While the monolayer WS2 QDs prefer weakly polar solvent than nonpolar solvent, they could be suspended stably in a wide variety of solvents of varying polarity including pentane and methanol as long as the concentration is sufficiently low, for example, at the concentration corresponding to the optical density of 0.01 at 380 nm in 1 cm-thick cell. On the other hand, a broad PL centered at 500 nm similar to the PL of multilayer WS2 QDs appeared superimposed on the PL of monolayer WS2 QDs when the concentration of the solution was made much higher, for example, by 10 times (Figure 3f). Upon dilution and ultrasonication, the newly appeared PL at 500 nm diminished, showing some degree of reversibility. This suggests that the balance between the solvation of the individual monolayer QDs and their stacking via inter-QD interaction altering their ensemble spectroscopic properties is strongly dependent on the concentration of QDs even for relatively good solvents. Preferential solvation of monolayer QDs in DMF over pentane enabling the separation of monolayer and multilayer QDs was also observed in MoS2 QDs prepared in a similar manner as WS2 QDs (Figure S5). The dependence of the solution PL spectrum on the QD concentration was also similar in both WS2 and MoS2 monolayer QDs. These observations indicate that not only the uniformity of the QD size and thickness but also good solvation of individual QDs is crucial in utilizing the colloidal TMDC QD as the spectroscopic reporter probing the surrounding environment it is interacting with. Having established the method to prepare the colloidal solution of pure monolayer WS2 QDs with narrow size distribution exhibiting well-defined spectroscopic behavior, we investigated the effect of solvent-QD interaction on the PL peak energy and intensity. We focus on the effect of dielectric screening and potential direct electronic coupling between the QD and solvent molecules. Lin et al. reported the shift of PL peak energy (EPL) of monolayer MoS2 nanosheets deposited on SiO2/Si substrate in contact with liquid solvent of varying dielectric constants (εr).24 The blueshift of EPL with the increase of εr of the solvent (ΔEPL < 10 meV for A exciton from Δεr of 31) was explained by the dielectric screening of the Coulomb potential modifying the band gap (Eg) and exciton binding energy (Eb).24,36,37 Compared to the effect of varying εr of solvent on Eg and Eb, EPL in the monolayer MoS2 sheet varied more weakly due to the opposite dependence of Eg and Eb on εr partially canceling their contributions to EPL. Nevertheless, such dielectric screening effect should be present in the colloidal QDs, possibly with stronger influence since the solvent molecules are surrounding the QDs rather than making a contact on one side of the basal planes. Direct electronic coupling between the QDs with surface-bound ligand that modifies the energy level and quantum confinement of the charge carriers are well-known in colloidal quantum dots of II− VI and IV−VI semiconductors.38−41 In the colloidal monolayer WS2 QDs without the surface-bound ligands, one may anticipate possible electronic coupling with the solvent molecules and the surface of the QD directly exposed to the solvent environment. Here, we obtained the PL spectra of monolayer WS2 QDs in various organic solvents having different dielectric constants (εr) to examine the effect of dielectric screening and possible direct electronic coupling between the QD and solvent molecules. The solvents of two different categories, non-
Figure 3. (a,c) Solution PLE (black) and PL (red) spectra of monolayer (a) and multilayer (c) WS2 QDs in DMF and pentane, respectively. (b,d) Polar plot of the ensemble PL polarization anisotropy for monolayer (b) and multilayer (d) WS2 QDs deposited on the glass substrate under linearly polarized 370 nm excitation to 0° direction. (e) Raman spectra of bulk WS2 powder and colloidal WS2 QDs. (f) Normalized PL spectra of monolayer WS2 QDs before (black) and after (red) increasing the concentration by 10 times. The difference spectrum is shown in green.
this new mode is not clear, we conjecture it is the local vibrational mode at the edge based on the gradual disappearance of this vibronic feature in the PL with increasing QD diameter in the previous single-particle study on WSe2 QDs.12 The monolayer WS2 QDs obtained here exhibits the unique linear PL polarization anisotropy, which is very similar to what was observed in monolayer WSe2 QDs.12 Figure 3b show the polarization angle dependent PL intensity from the monolayer WS2 QDs deposited on a glass substrate under linearly polarized excitation to 0° direction, where the direction of excitation polarization coincides with that of PL emission at both single-particle and ensemble level. In contrast to the monolayer WS2 QDs, multilayer WS2 QDs exhibit much more red-shifted PL and very weak PL polarization anisotropy. These behaviors resemble those of aggregated monolayer WSe2 QD obtained from hot-injection method12 (Figure S8), where complete loss of PL anisotropy and red-to-near-IR PL were observed presumably from the interlayer electronic coupling analogous to the multilayer TMDC sheets.34,35 The above results demonstrate that the separation of monolayer and multilayer QDs through their difference in the QD-solvent interaction is highly effective in preparing pure monolayer WS2 QDs via liquid-phase exfoliation. 7474
DOI: 10.1021/acs.nanolett.7b03381 Nano Lett. 2017, 17, 7471−7477
Letter
Nano Letters
Table 1. PL Peak Energy (EPL) of Monolayer WS2 QDs in Nonaromatic and Aromatic Solvents of Varying Dielectric Constant (εr) nonaromatic solvent
EPL (eV)
εr
QY (%)
aromatic solvent
EPL (eV)
εr
QY (%)
pentane diethyl ether ethyl acetate 2-propanol acetone methanol acetonitrile
3.086 3.088 3.099 3.105 3.098 3.101 3.100
1.84 4.33 6.02 17.9 20.7 32.7 37.5
17 18 15 19 17 16 16
hexafluorobenzene (HFB) 1,2,4-trichlorobenzene (TCB) benzene (BZ) toluene (Tol) bis(trifluoromethyl)benzene (TFMB) 1,2-dichlorobenzene (DCB) 1,2-difluorobenzene (DFB) benzonitrile (BZN)
3.064 3.024 3.050 3.055 3.082 3.028 3.055 3.052
2.05 2.24 2.27 2.38 5.98 9.93 13.8 26.0
15 12 10 11 13 8 46 9
Figure 4. (a,b) PL spectra of monolayer WS2 QDs in various nonaromatic (a) and aromatic (b) solvents. DFB, 1,2-difluorobenzene; HFB, hexafluorobenzene; TFMB, bis(trifluoromethyl)benzene; BZ, benzene; TCB, 1,2,4-trichlorobenzene; BZN, benzonitrile; Tol, toluene; DCB, 1,2dichlorobenzene. (c) EPL of monolayer WS2 QDs in various solvents as a function of solvent dielectric constant (εr) for nonaromatic (black) and aromatic (red) solvents. (d) EPL of CdS/ZnS QDs passivated with oleylamine in various solvents as a function of solvent εr for nonaromatic (black) and aromatic (red) solvents.
optical properties of monolayer WS2 QDs cannot be explained by dielectric screening alone. Among the aromatic solvents, bis(trifluoromethyl)benzene (TFMB) shows EPL, PL QY yield and spectral line shape closest to those of nonaromatic solvents. The largest difference from the PL in nonaromatic solvents is observed in the two halogenated benzene derivatives; difluorobenzene (DFB) and dichlorobenzene (DCB). DFB shows the highest PL QY (46%) and the largest deviation of the PL line shape from that of nonaromatic solvents. DCB exhibits the lowest PL QY (8%) and the largest deviation of EPL (∼70 meV) from that of nonpolar solvent having similar εr. In the mixture of nonaromatic (ethyl acetate) and aromatic (DCB) solvents, EPL varies almost proportionally to the composition of the two solvents as shown in Figure 5. This indicates that despite the stronger influence of aromatic solvents on EPL of monolayer WS2 QDs, both solvents have equal access to the surface of the QDs without preferential interaction with aromatic solvent molecules. Nevertheless, the significant change of EPL, PL line shape, and QY observed in
aromatic and aromatic solvents exhibiting distinctly different behavior, were tested as summarized in Table 1. For aromatic solvents, benzene derivatives with different substituents were used. Figure 4a,b compares the PL spectra of monolayer WS2 QDs in nonaromatic and aromatic solvents, respectively. The concentration of the QDs was kept the same and sufficiently low to ascertain the complete dispersion of the monolayer QDs in this comparison. The PL peak energies (EPL) in the two groups of solvents are also compared in Figure 4c. The different response of the PL spectrum to the change of the solvent in nonaromatic and aromatic solvents is apparent in the intensity and line shape of the PL as well as EPL. In nonaromatic solvents, PL spectrum of monolayer WS2 QDs does not change significantly. The PL quantum yield (QY) is scattered in a narrow range of 15−18% and the spectral line shape is nearly identical in all nonaromatic solvents used (Figure S9). The PL peak energy (EPL) blueshifts weakly as εr increases from 1.84 (pentane) to 37.5 (acetonitrile), exhibiting
