www.acsnano.org
Downloaded via UNIV OF SOUTHERN INDIANA on July 19, 2019 at 06:28:05 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
A Scalable, Solution-Based Approach to Tuning the Solubility and Improving the Photoluminescence of Chemically Exfoliated MoS2 Myung Jin Park,† Steven Gravelsins,† Jangyup Son,‡ Arend M. van der Zande,‡ and Al-Amin Dhirani*,† †
Department of Chemistry, University of Toronto, 80 Saint George Street, Toronto, Ontario M5S 3H6, Canada Department of Mechanical Science and Engineering, University of Illinois at Urbana−Champaign, 1206 West Green Street, Urbana, Illinois 61801, United States
‡
S Supporting Information *
ABSTRACT: MoS2 are two-dimensional (2D) materials that exhibit emerging photoluminescence (PL) at the monolayer level and have potential optoelectronic applications. Monolayers of MoS2 typically achieved by mechanical exfoliation (Me), chemical vapor deposition (CVD), and chemical exfoliation (Ce) via lithium intercalation contain numerous defects that significantly reduce their PL efficiency. Several studies have reported overcoming poor PL in mechanically exfoliated and CVD-grown MoS2, but such studies for chemically exfoliated MoS2 (Ce-MoS2) have not been reported. Here, we report a solution-based method of enhancing the PL of Ce-MoS2 by reacting with molecules with suitable functional groups at high temperatures. Reaction with dodecanethiol (DDT) generates PL that is more intense than mechanically exfoliated MoS2 (Me-MoS2) with high crystallinity and has a significantly broader range of wavelengths. Based on ultraviolet−visible, Fourier transform infrared, X-ray photoemission, and PL spectroscopy as well as transmission electron and PL imaging, we propose that the present method modifies PL properties of Ce-MoS2 by simultaneously annealing, replacing molybdenum−oxygen with molybdenum−sulfur bonds, inducing strain, and generating a nanopolycrystalline structure. This work points to such defect engineering using molecules as an effective means to modify the properties of Ce-MoS2 and layered transition-metal dichalcogenides more generally. KEYWORDS: transition-metal dichalcogenide, chemically exfoliated MoS2, surface chemistry, functionalization, phase engineering, photoluminescence enhancement CVD has been overcome using p-dopants10 or passivating vacancies with oxygen bonding,11,12 respectively. A recent study reported that chemical treatment of mechanically exfoliated monolayers with a super acid results in dramatically enhanced quantum efficiency of more than 95%, potentially due to an increased sulfur-to-molybdenum ratio.13−15 Chemical exfoliation (Ce) is a widely known method that can enable the mass production of monolayers from bulk crystals and address a challenge of scalability for optoelectronic applications of MoS2.16−18 MoS2 can be exfoliated into a
M
olybdenum disulfide (MoS2) is a prototypical example of a layered transition metal dichalcogenide (LTMD) and has received significant attention recently for potential optoelectronic applications.1−4 It is semiconducting in its most stable trigonal prismatic 2H phase.5 As the number of layers decreases to one, its band gap changes from indirect (1.2 eV) to direct (1.9 eV), and MoS2 exhibits photoluminescence (PL) in this limit.6,7 Monolayers of MoS2 are typically achieved by mechanical exfoliation (Me), chemical vapor deposition (CVD), and solution-based chemical reagent assisted exfoliations.8 However, monolayers prepared using these methods have poor PL quantum yield due to an emergence of defects such as sulfur vacancies, adatoms, and impurities, which leads to nonradiative recombination.9 The poor PL of MoS2 prepared by Me or © 2019 American Chemical Society
Received: December 19, 2018 Accepted: May 30, 2019 Published: May 30, 2019 6469
DOI: 10.1021/acsnano.8b09578 ACS Nano 2019, 13, 6469−6476
Article
Cite This: ACS Nano 2019, 13, 6469−6476
Article
ACS Nano
Figure 1. Schematic illustration of the overall process for functionalizing MoS2 with dodecanethiol (DDT). (a) Transferring Ce-MoS2 flakes from deionized water (DW) to octadecene (OD) using the amphiphilic oleylamine (OA) as a ligand and phase-transfer agent. (b) Reaction between Ce-MoS2 flakes and DDT conducted at high temperatures. Interaction with MoS2 flakes can be facilitated by the cationic character of OA and anionic character of DDT. The reaction simultaneously reverts Ce-MoS2 from the metallic phase to the semiconducting phase and functionalizes MoS2 with DDT.
colloidal suspension of individual thin flakes by reacting with lithium and deionized water (DW):17,19 2H‐MoS2 + x LiC4 H 9 → 1T‐LixMoS2 + x /2C8H18
Recovery of PL for solution-based chemically exfoliated LTMDs has not been reported yet. Given success using a super acid to enhance PL of mechanically exfoliated MoS2 (MeMoS2) by reducing defect-mediated non-radiative recombination,13−15 we explore improving PL of chemically exfoliated MoS2 (Ce-MoS2) by high-temperature treatment with molecular fuctionalization agents. In this study, we report a completely solution-based, high-temperature method that begins with metallic Ce-MoS2 and simultaneously (1) functionalizes MoS2 surfaces and tuning solubility, (2) restores the semiconducting phase, and (3) strongly enhances PL. This approach affords flexibility in choice of molecules that may be used and provides a possibility of surface engineering of chemically exfoliated LTMDs by exploiting element vacancies, thereby enabling a diversity of 2D materials available for various applications.
(1)
1T‐LixMoS2 + x H 2O → 1T‐MoS2 (monolayer) + x LiOH + (x /2)H 2
(2)
In this process, MoS2 changes from its intrinsic semiconducting 2H phase to a metastable, distorted 1-T or 1-T′ metallic phase, and PL is lost. A previous study has reported that the metallic phase can be reverted back to the semiconducting phase by annealing under an inert atmosphere, and PL can be recovered.16 However, the annealing process described requires precoating substrates, which limits applicability and processability.16,20 Also, lithiation can generate sulfur vacancies in MoS2 monolayers, as described by the following formula:21 1T‐LixMoS2 + x Li+ → 2Li 2S + Mo + (2x − 4)Li+
RESULTS AND DISCUSSION Figure 1 shows a schematic illustration of method used in the present study starting with metallic Ce-MoS2 dissolved in DW. MoS2 flakes are transferred to an octadecene OD phase using oleylamine OA. OD is a high boiling point (314 °C) solvent
(3)
Such sulfur vacancies act as defect sites that can trap charges and mediate nonradiative recombination, significantly hindering PL efficiency.13,22 6470
DOI: 10.1021/acsnano.8b09578 ACS Nano 2019, 13, 6469−6476
Article
ACS Nano
Figure 2. (a) Upper images show the as-exfoliated Ce-MoS2 dispersed in DW (left) and MoS2 reacted at high temperatures with OA (labeled “OA-MoS2”, middle) and DDT (labeled “DDT-MoS2”, right), both dispersed in toluene. Images of OA- and DDT-MoS2 in toluene before and after 2 days are shown in lower images. (b) Ultraviolet−visible (UV−vis) spectra for Ce-MoS2 (black), OA-MoS2 (red) and DDT-MoS2 (orange) samples. (c) FT-IR spectra (left panel) for various samples (bare powder: blue, Ce-MoS2: black, OA-MoS2: red, DDT-MoS2: orange) showing vibration modes for symmetric and asymmetric stretching (νs,as) and bending (δ) modes. Magnified FT-IR spectra (right panel) in the S−H symmetric stretch region around 2563 cm−1 to confirm the absence of S−H related peak.
change color (Figure S1c) and are not soluble in DW after the reaction (Figure 2a). Both are initially soluble in toluene. After 2 days, DDT-MoS2 remains in solution, while the OA-MoS2 precipitates. This observation combined with X-ray photoemission spectroscopy (XPS) results discussed in detail below indicate that MoS2 interacts more strongly with DDT than OA. After the high-temperature reactions, ultraviolet−visible (UV− vis) spectra for both samples exhibit 2 distinct peaks around 660 and 620 nm, corresponding to A and B excitons, respectively,6,7 showing that the metallic phase at least partially successfully transforms to the semiconducting phase (Figure 2b). Data for the OA-MoS2 exhibit a strong peak in the UV region, which is attributed to OA. This peak did not disappear even after washing up to 10 times. However, data for the DDTMoS2 does not exhibit this peak, providing evidence for ligand exchange from OA to DDT. Figure 2c shows FTIR spectra for bare powder and Ce-, OA-, and DDT-MoS2 samples. Data for both OA- and DDTMoS2 exhibit reduced molybdenum oxide related peaks28 and increased C−H peaks29−31 compared with data for Ce-MoS2 sample due to removal of oxygen during high-temperature reactions with ligands. A C−N peak due to OA appears in data for the OA-MoS2, but it does not for the DDT-MoS2. Because a nitrogen N 1s peak is not detected in XPS survey spectra,
widely used to synthesize inorganic quantum dots because it allows a variety of reactions that require high temperature.23−27 It is a suitable solvent in the present study because PL of MoS2 occurs in the thermodynamically favored semiconducting phase that can be recovered from the metastable metallic phase by heating around 200 °C.16 OA serves both as a ligand and as a phase-transfer agent. The amine group (R-NH2) at the end of the alkyl chain has a cationic character that can electrostatically interact with negatively charged Ce-MoS2 flakes. The alkyl chains of OA enable individual flakes to dissolve in OD because OD is a hydrophobic organic solvent (Figures 1a and S1). OA thereby facilitates binding between Ce-MoS2 and dodecanethiol (DDT), which is introduced in the OD organic phase in the next step. DDT is chosen in the present study because it has a high boiling point (266−283 °C). It can potentially serve both as a healing agent for sulfur vacancies and as a hydrophobic ligand, making DDT-functionalized Ce-MoS2 flakes soluble in organic solvents. OA is basic and can interact with the proton of dodecanthiol (Figure 1b). To confirm that DDT bonds with MoS2 flakes, optical and elemental analyses are conducted by comparing samples after a high-temperature reaction with DDT plus OA (“DDT-MoS2”) versus with just OA (no DDT, “OA-MoS2”). Both samples 6471
DOI: 10.1021/acsnano.8b09578 ACS Nano 2019, 13, 6469−6476
Article
ACS Nano
Figure 3. a) XPS spectra corresponding to Mo 3d, S 2s, and S 2p peak regions for each sample. Mo4+3d and S 2p peaks are deconvoluted to show contributions of metallic and semiconducting phases, represented by green and red solid lines, respectively. Blue and yellow solid lines represent S 2s and Mo 6+3d3/2 peaks, respectively. (b) S-to-Mo and O-to-Mo ratios calculated based on the survey spectra for each sample. (c) Mo4+-to-(Mo4++Mo6+) and Mo6+-to-(Mo4++Mo6+) ratios for each sample. (d) 2H-to-(1T + 2H) ratio of Mo4+3d peaks for each sample. (e) TEM images for Ce-MoS2, OA-MoS2, and DDT-MoS2 samples (left, middle, and right images, respectively). Insets show fast Fourier transform (FFT) images for the entire areas of each TEM image. Scale bars represent 5 nm.
binding to Mo in particular, Figure 3c shows an analysis of Mo4+ and Mo6+ peak areas in Figure 3a. Mo4+ and Mo6+ peak areas probe Mo−S and Mo−O binding, respectively.16,36 The anti-correlation in Figure 3c between Mo4+/(Mo4+ + Mo6+) and Mo6+/(Mo4+ + Mo6+) shows similar behavior to that obtained for S/Mo and O/Mo from survey spectra (Figure 3b). These results indicate that oxygen that is bonded to molybdenum in Ce-MoS2 can be substituted by sulfur in DDTMoS2. To analyze phase compositions of the OA- and DDT-MoS2 in Figure 3a, bare powder is used as a reference for semiconducting peaks. Mo4+ 3d5/2 and Mo4+ 3d3/2 peaks appear at 229 and 232 eV in the bare powder, respectively. These peaks are shifted to lower binding energies by 1 eV in the as-exfoliated Ce-MoS2, which indicates significant transformation to the metallic phase. The fact that peaks appear broadened compared to those of the bare powder also suggests a presence of a mixture of mostly metallic and a small amount of 2H phases (Figure 3d). The metallic phase almost completely reverts back to the 2H phase in OA-MoS2. These results indicate that the present convenient approach of annealing Ce-MoS2 with OA in a solvent with high-boiling point can also be used to restore the semiconducting phase. In contrast, the metallic phase contributes significantly to the composition of DDT-MoS2. In fact, PL has been previously reported for surface functionalized metallic 1T MoS2.30
OA-MoS2 binding is negligible (Table 1 in Figure S2). As a result, most OA is likely washed away during the purification step, leading to MoS2 precipitating from toluene as shown in Figure 2a. We attribute peaks corresponding to C−N and C− H in the FTIR spectrum (Figure 2c) to OA physical adsorption on MoS2. A S−H peak, which should appear at 2563 cm−1 if present, is not observed in the DDT-MoS2, indicating that the thiol group is successfully covalently bonded to the surface of the Ce-MoS2 flakes.30,31 Figures 3a−d and 3e show XPS and transmission electron microscopy (TEM) data, respectively. To analyze elemental and phase compositions of the OA- and DDT-MoS2, we deconvoluted XPS data given positions of semiconducting and metallic Mo 3d, S 2s, and S 2p peaks.16 Figure 3b shows an analysis of oxygen and sulfur content for various samples calculated from survey spectra (section S2 of the Supporting Information). The ratio of sulfur to molybdenum (S/Mo) decreases from 1.68 ± 0.04 for bare powder to 1.56 ± 0.01 for Ce-MoS2, which indicates that sulfur is depleted during the lithiation process. These vacant sites have a potential to be fuctionalized by DDT during the high-temperature reaction, as evidenced by an increased S/Mo ratio of 1.66 ± 0.04 for DDTMoS2. This value is very close to that for bare powder. The ratio of oxygen to molybdenum (O/Mo) shows opposite behavior: as S/Mo decreases, O/Mo increases and vice versa. To test whether this anti-correlation is related to S vs O 6472
DOI: 10.1021/acsnano.8b09578 ACS Nano 2019, 13, 6469−6476
Article
ACS Nano
Figure 4. (a) Un-normalized and (b) normalized photoluminescence (PL) spectra for Me-MoS2 (blue), CVD-MoS2 (sky blue), OA-MoS2 (red), DDT-MoS2 (orange), and Ce-MoS2 (black). Labels A and B in panel b, respectively, correspond to different exciton peaks arising from K point of the Brillouin zone. (c) PL peak intensities vs energies for various samples. (d) The laser-power-dependent PL intensity for each sample. Dashed lines show the power law fits.
MoS2 (Figure 4b). Given that the electron transition mainly originates from molybdenum’s d-orbital,34,35 a presence of oxygen bonding with molybdenum as confirmed by XPS data likely affects the transitions. In DDT-MoS2, some oxygen is substituted by sulfur, and the B exciton is suppressed (Figure 4b). These results indicate that the optical properties of MoS2 flakes can be modified by choice of molecules used to functionalize MoS2. Figure 4b shows that both OA- and DDT-MoS2 exhibit significantly broadened PL spectra, with particularly pronounced shoulders at lower energies, compared to Me- and CVD- MoS2. Previous studies have reported that tensilestrained samples can exhibit red-shifted and broadened PL spectra because strain-induced distortions to band structure can lower gap energies.37,38 This mechanism may contribute to the broadening at lower energies observed here, as atomic force microscopy (AFM) images of OA- and DDT-MoS2 show that these materials possess a crumpled structure (Figure S4). To test for variations, we show in Figure 4c peak intensities and energies obtained at various locations on Me-, OA-, and DDT-MoS2 samples. These data are consistent with peak broadening exhibited by OA- and DDT-MoS2: peak shifts have a significantly wider distribution for OA-, and DDT-MoS2 compared with Me-MoS2. Overall PL spectra of DDT-MoS2 exhibit the strongest intensities of all the various types of samples (see Figure 4c). Improved PL efficiency can be demonstrated also by measuring PL intensity versus laser power.13 Figure 4d shows that DDT-MoS2 exhibits higher PL
TEM analysis is used to analyze the structure of the various types of samples in more detail. Figure 3e shows a TEM image of Ce-MoS2 and reveals stripped features, as is also indicated by distorted hexagonal bright spots in the fast Fourier transform image (FFT). OA-MoS2 exhibits an ordered atomic lattice, and this regularity is confirmed by a hexagonal pattern of bright spots in the FFT. DDT-MoS2 exhibits a polycrystalline atomic structure phase, as confirmed by multiple sets of hexagonal bright spots in the FFT. Panels a and b of Figure 4 show the un-normalized and normalized PL spectra, respectively, for the various types of monolayer samples prepared in this study. Mechanically exfoliated and CVD-grown samples are used as reference samples to determine peak positions in PL spectra because both types of samples have good crystallinity mainly composed of the trigonal prismatic 2H phase. The 2H phase typically exhibits two distinct PL peaks arising from the direct transition at the K point of the Brillouin zone because broken inversion symmetry and spin−orbit interactions split the valence bands by 160 meV.32,33 The Ce-MoS2 does not exhibit PL, while distinct peaks are observed in both OA- and DDT-MoS2. In fact, the PL intensities of functionalized samples with different ligands in the present study are much higher than those of mechanically exfoliated and CVD-grown samples: PL peak intensity of a DDT-functionalized MoS2 monolayer is around 7 times higher than that of the crystalline-mechanically exfoliated sample (Figure 4a) and ∼4 times higher than that of a OAMoS2 monolayer. The PL spectrum of OA-MoS2 exhibits a larger B exciton feature compared with the spectrum of DDT6473
DOI: 10.1021/acsnano.8b09578 ACS Nano 2019, 13, 6469−6476
Article
ACS Nano
Figure 5. Bright-field and photoluminescence (PL) images of Ce-MoS2 (panels a and b), OA-MoS2 (panels c and d), and DDT-MoS2 (panels e and f), respectively. PL images are captured through a 630 nm emission filter using 560 nm excitation filters. Scale bars represent 50 μm. (g) Histogram corresponding to the intensity for the entire area of each sample.
respectively. DDT-MoS2 exhibits a ∼4-fold improved PL intensity with a suppression of B exciton feature compared with OA-MoS2 and a ∼7-fold increase in PL compared with Me-MoS2. The high-temperature treatment and choice of molecule used to functionalize MoS2 play important roles in improving PL. DDT-MoS2 exhibits PL at a broad range of wavelengths, which is likely due to strain effects and inhomogeneity arising from grain sizes that are small and broadly distributed. This work provides insights into mechanisms and benefits of engineering of 2D sheets using straightforward and scalable solution-based methods. It demonstrates additional functionalities (such as tunable solubility and strongly improved PL) beyond those intrinsically available.
efficiency compared to the other samples even at low laser powers (
