Research Article www.acsami.org
Cite This: ACS Appl. Mater. Interfaces 2017, 9, 39380-39390
Ultradispersed and Single-Layered MoS2 Nanoflakes Strongly Coupled with Graphene: An Optimized Structure with High Kinetics for the Hydrogen Evolution Reaction Haoliang Huang,†,‡ Junying Huang,† Weipeng Liu,† Yueping Fang,*,† and Yingju Liu*,† †
College of Materials and Energy, South China Agricultural University, Guangzhou 510642, China Department of Chemistry, University of Southampton, University Road, Southampton SO17 1BJ, U.K.
‡
S Supporting Information *
ABSTRACT: As one of the most promising Pt alternatives for costeffective hydrogen production, molybdenum disulfide (MoS2), although has been studied extensively to improve its electrocatalytic activity, suffers from scarce active sites, low conductivity, and lack of interaction with substrates. To this end, we anchor ultradispersed and single-layered MoS2 nanoflakes on graphene sheets via a hybrid intermediate (MoOx−cysteine−graphene oxide), which not only confines the subsequent growth of MoS2 on the graphene surface but also ensures the intimate interaction between Mo species and graphene at the initial stage. Mo−O−C bond and a possible residual MoO3−x layer are proposed to comprise the interface bridging the two inherent incompatible phases, MoS2 and graphene. This strongly coupled structure together with the highly exposed MoS2 morphology accelerates the electron injection from graphene to the active sites of MoS2, and thus the hydrogen evolution reaction (HER) can achieve an overpotential of ∼275 mV at ∼−740 mA cm−2, and a Ptlike Tafel slope of ∼35 mV dec−1. Our results shed light on the indispensable role of interfacial interaction within semiconducting material−nanocarbon composites and provide a new insight into the actual activity of MoS2 toward the HER. KEYWORDS: molybdenum disulfide, graphene, hydrogen evolution reaction, electrocatalysis, nanoflakes
1. INTRODUCTION Developing inexpensive and earth-abundant electrocatalysts as platinum alternative is of great benefit to deploy clean energy technologies in sustainable hydrogen economy.1−3 Transition metal dichalcogenides, especially molybdenum disulfide (MoS2), have emerged as one of the most promising candidates for the hydrogen evolution reaction (HER) in acidic electrolyte (HER, 2H+ + 2e− → H2).4,5 Much efforts have been devoted to improving the electrocatalytic properties of MoS2 rationally. Studies from both theoretical calculations and experimental works agree that the HER activity of MoS2 is correlated to the number of exposed edge sites because the large basal plane remains catalytically inert.6−8 Recently, the active sites on the edges have been further probed to be terminal disulfide motifs via in situ X-ray absorption spectroscopy.9 Thus, numerous works aimed at multiplying the scare amount of active sites and alleviating the inherent resistance of MoS2, such as decreasing the number of layers10 or expanding their spacing,11,12 phase engineering,13,14 designing molecular mimics,15,16 and constructing edge-exposed nanostructures.8,17,18 Furthermore, nanocarbons such as carbon nanotube (CNT) and graphene, featuring vast surface area and high conductivity,19,20 act as supporting materials of MoS2 for a better electrocatalytic behavior.19,21−25 © 2017 American Chemical Society
Notwithstanding the outcomes in fabricating MoS2 structure and constructing MoS2−nanocarbon hybrids, the activity of MoS2 remains unsatisfactory, relative to the target Pt, and their disadvantages still abound. The major concern is that the interaction of MoS2 and nanocarbons is too weak to disperse MoS2 from aggregation and to supply sufficient electrical contact between these two phases. In terms of composition and structure, MoS2 and nanocarbons inherently lack the covalent bond with each other, unlike the oxygen bridge26,27 or the strong chemical attachment20 in metal oxide−nanocarbon composite, which alerts us that when it comes to designing a MoS2−nanocarbon hybrid, the role of nanocarbons should be deliberated, integrating with MoS2 as a strongly coupled nanocomposite or only acting as a current collector with a high surface area. Especially, due to the thermodynamic driving force, MoS2 tends to overgrow into large size and multiple layers to minimize its surface,8,28 which causes fewer active sites, higher sheet resistance, and a little electrical contact with substrates or nanocarbons.29−31 Received: August 13, 2017 Accepted: October 23, 2017 Published: October 23, 2017 39380
DOI: 10.1021/acsami.7b12038 ACS Appl. Mater. Interfaces 2017, 9, 39380−39390
Research Article
ACS Applied Materials & Interfaces
respectively. The common Pt counter electrode was excluded to guarantee the authenticity of measured data.5,8 All of the potentials were calibrated to reversible hydrogen electrode (RHE), and all of the electrochemical measurements were performed in 0.5 M H2SO4 with continuous purging of N2 (>99.999% purity). The polarization curves (j−V plots) were recorded from linear sweep voltammetry at a sweep rate of 1 mV s−1, where iR losses were automatically compensated by the workstation software (Zahner Z2.04). Electrochemical impedance spectroscopy (EIS) was conducted at an overpotential of 250 mV with the frequency from 100 kHz to 0.1 Hz. The electrochemical doublelayered capacitance was estimated by cyclic voltammetry (CV) sweeping between 0.1 and 0.3 V from 20 to 200 mV s−1. The longterm stability was measured by galvanostatically electrolyzing for 10 h at a constant current density of −100 mA cm−2.
To tackle these challenges, combining our recent success in connecting MoS2 and carbon nanotubes31 and controlling MoS2 nanostructures,18 we further construct an ideal structure of MoS2 in which ultradispersed and single-layered MoS2 nanoflakes strongly couple reduced graphene oxide sheets (UDSL-MoS2−rGO). Such a structure is statistically studied by transmission electron microscopy (TEM), X-ray diffraction (XRD) and N2 absorption. A sandwich structure of MoS2− MoO3−x−graphene with Mo−O−C bond is proposed to be responsible for this strong interfacial interaction between MoS2 and graphene, where not only can the active sites of MoS2 be increased but also can the sheet resistance of MoS2 and the contact resistance of the interface be alleviated. As a result, the actual hydrogen evolution reaction (HER) activity of MoS2 is exploited, producing a surprising electrocatalytic activity and durability, a record-low Tafel slope and a Pt-like HER mechanism.
3. RESULTS AND DISCUSSIONS 3.1. Morphological and Structural Characterization. Ultradispersed and single-layered MoS2 nanoflakes were grown on graphene sheets via a scalable and environmental-friendly method, where an aqueous solution containing three necessary precursors (molybdate, cysteine, and GO) was heated under hydrothermal condition. The elemental composition of the product is confirmed by XPS survey spectrum (Figure 1) in
2. EXPERIMENTAL SECTION 2.1. Chemicals. All of the chemicals were used as received without further purification. Graphite flake (325 mesh, 99.8%, Alfa Aesar), ammonium molybdate tetrahydrate (99.9%, Aladdin Reagent), Lcysteine (cys, 99%, Aladdin Reagent), and MoS2 crystal (99.999% purity) and a Hg|Hg2SO4 electrode (filled in saturated K2SO4 and equipped with a Luggin capillary) were used as the counter and reference electrodes,
Figure 1. XPS survey spectrum of UDSL-MoS2−rGO.
which only C, O, S, and Mo can be detected. The morphology is further confirmed by TEM images at different magnifications. In Figure 2A, UDSL-MoS2−rGO shows a two-dimensional (2D) nanostructure assembled by large graphene sheets (hundreds of nanometers in diameter) and a myriad of thin and tiny MoS2 nanoflakes with less than 10 nm in edge length on the sheets. Without the support of graphene, only MoS2 nanoflakes aggregates will be formed (Figure S1), whereas without MoS2, rGO is smooth with a high surface area (Figure S2). The ultradispersed MoS2 nanoflakes can be further proved statistically. The large scale of TEM images in Figure 2B,C show that no naked GO sheet or agglomerated MoS2 nanoflake is discernible, and all of the MoS2 nanoflakes are uniformly coated on the surface. It is worth noting that the increased contrast in some area results from folds and restacks of flexible UDSL-MoS2−rGO sheets. From HRTEM (Figure 2D,E), these ultradispersed MoS2 nanoflakes also display a single-layered feature. Ridge-like MoS2 edges, composed of one to two layers, stand in all of the direction on the surface of graphene sheets, and the monolayer rate of MoS2 nanoflakes, with more than 39381
DOI: 10.1021/acsami.7b12038 ACS Appl. Mater. Interfaces 2017, 9, 39380−39390
Research Article
ACS Applied Materials & Interfaces
Figure 2. Morphological characterization of UDSL-MoS2−rGO. (A−C) TEM images at different magnifications, (D, E) HRTEM images, and (F) the single-layered MoS2 histogram of UDSL-MoS2−rGO. (Inset of (E)) FFT pattern of the selected area (blue dotted box) in (E).
200 sampling capacity, is rationalized to be ∼79%. The HRTEM in Figure 2E gives an interplanar lattice spacing of 0.26 nm, whereas the existence of MoS2 is indicated by a faint diffraction ring in the fast Fourier transform (FFT) pattern (inset of Figure 2E) on the selected area, which can be indexed to the spacing of (101) facet in 2H-MoS2 (JCPDS no. 371492).33 The morphology of UDSL-MoS2−rGO is also stable: there is no significant change in the TEM image after the sample is stored under ambient condition for more than 1 year (Figure S3). The morphology of UDSL-MoS2−rGO is found to be pHdependent. Synthesized under higher pH (∼2.5), MoS2 nanoflakes in MoS2−rGO-2.5 (Figures 3A and S4A) remain a ridge-like morphology and uniform distribution on graphene sheets but grow larger (∼40 nm in edge length) and thicker. The overgrowth of MoS2 becomes more serious in the sample without acidifying the reaction (pH = ∼4; MoS2−rGO-4; Figures 3B and S4B), where the outline of rGO sheets can be no longer distinguished from petal-like MoS2 and the size of MoS2 increases to more than 60 nm. Especially, the multilayered stacking can be observed in TEM image, which can be indicated by the periodic oscillation in the gray-scale profile (inset of Figure 3B) across a selected MoS2 edge. XRD and N2 absorption−desorption further proved the single-layered and ultradispersed MoS2 of UDSL-MoS2−rGO in TEM. As in Figure 3C, all of the three MoS2−rGO samples share similar XRD patterns in interlayer-expanded 2HMoS2,11,12 but the width of (002) diffraction peak differs, which derives from the stacking of S−Mo−S atomic layers. The full width at half-maximum of this peak was measured and then the number of MoS2 layers was estimated by the Scherrer equation. Whereas the thickness of stacked layers in MoS2− rGO-2.5 and MoS2−rGO-4 is calculated to be about ∼3 and ∼4
layers, respectively, the (002) peak in UDSL-MoS2−rGO stretches out, suggesting that the layer-stacking structure of MoS2 disappears, i.e., single-layered MoS2. As for graphene, the absent diffraction peak at 2θ = 25−26° indicates the restacking of graphene back to graphite is also minimized, possibly by the formation of Mo-containing intermediate on GO before its reduction occurs. Based on the single-layered structure of both MoS2 and graphene in UDSL-MoS2−rGO, a highly exposed surface is expected. From the N2 absorption−desorption results (Figures 3D and S5), the BET surface area of UDSL-MoS2− rGO not only tops at 293.7 m2 g−1 among the as-prepared samples but also exceeds the reported values on graphenesupported 2D inorganic materials (Figure S6) and ultradispersed TiO2 nanoparticles (NPs) on graphene34 (287 and 215 m2 g−1 after annealing). In addition, the tiny size of MoS2 in UDSL-MoS2−rGO is supported by the pore size distribution (inset of Figure 3D). As the pore structure of the MoS2−rGO samples assembled by two 2D materials with different sizes, MoS2 nanoflakes and graphene sheets, this assembly creates pores in different scales, such as macropores among sheets and mesopores among nanoflakes. The lowest pore size of UDSLMoS2−rGO indicates its smallest size of MoS2 nanoflakes among the samples.35 3.2. Possible Formation Mechanism and Interfacial Structure of UDSL-MoS2−rGO. To understand the formation mechanism, the intermediate of UDSL-MoS2−rGO before the hydrothermal process was captured and the evolution of elemental chemical environment at three stages, including precursor, intermediate, and product, was studied (Figure 4A). The intermediate containing both reduced molybdenum oxide and cysteine (MoOx−cys) on graphene sheets was characterized by TEM (Figure 4B) and X-ray photoelectron spectroscopy (XPS). In Figure S7, the survey 39382
DOI: 10.1021/acsami.7b12038 ACS Appl. Mater. Interfaces 2017, 9, 39380−39390
Research Article
ACS Applied Materials & Interfaces
Figure 3. (A, B) Morphological characterization of controlled samples and (C, D) structural characterization. Typical TEM images of (A) MoS2− rGO-2.5, (B) MoS2−rGO-4, and (insets) the corresponding TEM images at lower magnification. Another inset of (B) shows the gray-scale profile across the selected MoS2 edge (blue line). (C) XRD patterns of as-prepared MoS2−rGO samples and their enlargement in a low 2θ region. (D) N2 absorption−desorption isotherm of UDSL-MoS2−rGO at 77 K and (inset of D) the pore size distribution of MoS2−rGO samples calculated from the absorption branch via DFT model.
In bridging graphene and molybdenum oxide phases, cysteine plays an indispensable role. It can reduce molybdate at low pH,36,37 as measured above. Through a strong Mo−S, it can also produce a stable complex with MoV and MoVI,36,37 as shown by the S 2p spectrum (Figure S7B), in which the S chemical environment of MoOx−cys−rGO shows ∼0.6 eV positive shift from that of reported H−S−C.38 For the graphene part, cysteine can also react with GO, supported by a plunge in the epoxy C−O (the main functional group of GO) in C 1s spectrum of MoO x −cys−rGO (Figure 4B).
spectrum composes of C, N, O, S, and Mo, showing the atomic ratios of S to Mo and N to S are ∼0.7, ∼1 (S: 4.22 atom %, N: 4.13 atom %), respectively, which is consistent with that of the precursor cysteine. The core-level spectra of C 1s, N 1s, and Mo 3d confirm the presence of both cysteine and MoOx (Figures 4B, S7C, and 4C, respectively). Compared to those in GO, carboxylic group in MoOx−cys−rGO increases and C−S/ C−N bonding emerges, whereas for Mo species, the oxidation state of Mo largely remains +6 and is partially reduced to +5 (∼10%), indicating a reduced molybdenum oxide. 39383
DOI: 10.1021/acsami.7b12038 ACS Appl. Mater. Interfaces 2017, 9, 39380−39390
Research Article
ACS Applied Materials & Interfaces
Figure 4. Possible mechanism for the formation of UDSL-MoS2−rGO. (A) A schematic illustration of the possible formation process and (inset of (A)) a diagram of sandwiched structure of UDSL-MoS2−rGO, and XPS core-level spectra in each step (GO, MoOx−cys−rGO, and UDSL-MoS2− rGO) in the region of (B) C 1s, (C) Mo 3d, (D) S 2p, and (E) O 1s. For better comparison, the counts of some spectra are adapted and the magnifications are labeled on the bottom left.
MoS2−cys supports the amorphous intermediate (Figure S1). Furthermore, because the degree of polymerization of molybdate is pH-dependent,40 the increased amount of H+ can raise the immobilized amount of MoOx−cys on graphene and lower the dissolution−precipitation process of MoOx− cys−rGO when sulfurized under hydrothermal environment. Thus, this cysteine- and proton-involved process, coupled with the large surface area of graphene, spatially disperses the Mo intermediate, constrains the growth of MoS2, and, hence, produces single-layered and tiny MoS2 nanoflakes ultradispersed on graphene.
Furthermore, we excluded the molybdate precursor from the synthesis, and it produced S,N-doped rGO (S: 2.33 atom %, N: 1.61 atom %; Figure S8). We also replaced cysteine with thiourea, a sulfur source without ammine and thiol functional groups (Figure S9), where MoS2 flakes are seriously overgrown, agglomerated and detached with rGO sheets. Therefore, cysteine is integrated with both the reduced molybdenum oxide and graphene oxide. Thanks to this, under acid environment, molybdate, interfered by cysteine, prefers to polymerize randomly into an amorphous layer (MoOx−cys) on the rGO sheet, rather than crystalizing into a dense and regular MoO3.39 In addition, the irregular and near-spherical core of 39384
DOI: 10.1021/acsami.7b12038 ACS Appl. Mater. Interfaces 2017, 9, 39380−39390
Research Article
ACS Applied Materials & Interfaces
Figure 5. Electrocatalytic activity and durability for HER. Polarization curves for HER in (A) low current density and (B) high current density regions of as-prepared MoS2−rGO samples, MoS2−cys, MoS2 crystal, and Pt wire in 0.5 M N2-saturated H2SO4 electrolyte with 1 mV s−1 scan rate. (C) Galvanostatic durability test on UDSL-MoS2−rGO under −100 mA cm−2 current density and (inset) the enlargement of the selected region.
Then, the XPS spectra of UDSL-MoS2−rGO including C 1s, Mo 3d, and N 1s + Mo 3p were compared with those of the intermediate (Figures 4B,C and S10). By the hydrothermal step, the characteristic C−N and C−S bonds and the residual from cysteine are completely removed, whereas the majority of MoVI is reduced to MoIV and MoV (Figure 4C).41,42 As for S atoms, the S 2p spectrum (Figure 4D) can be deconvoluted into two chemical environments (apical and bridging S2− with S 2p3/2 at ∼162.0 eV and terminal S22− at ∼163.7 eV)43 highlighted in the MoS2 model (inset of Figure 4D). The terminal disulfideidentified to be involved in reducing protons9accounts for ∼41% of the total S atoms in UDSLMoS2−rGO, whereas ∼25% in MoS2−cys and zero in MoS2 crystal (Figures S11A and S11B). This sharp difference indicates the structure of UDSL-MoS2−rGO may provide substantial active sites to catalyze H2 evolution. In addition, because the number of terminal sulfur atoms is proportionate to the size of MoS2 (Figure S11C), compared to MoS2−cys (∼30 nm) and MoS2 crystal (
