Balancing the Hydrogen Evolution Reaction, Surface Energetics, and

Dec 27, 2017 - Chemically exfoliated, metallic MoS2 nanosheets are functionalized with organic phenyl rings containing electron donating or withdrawin...
1 downloads 5 Views 3MB Size
Article pubs.acs.org/JACS

Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Balancing the Hydrogen Evolution Reaction, Surface Energetics, and Stability of Metallic MoS2 Nanosheets via Covalent Functionalization Eric E. Benson,† Hanyu Zhang,† Samuel A. Schuman, Sanjini U. Nanayakkara, Noah D. Bronstein, Suzanne Ferrere, Jeffrey L. Blackburn, and Elisa M. Miller* National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, Colorado 80401, United States S Supporting Information *

ABSTRACT: We modify the fundamental electronic properties of metallic (1T phase) nanosheets of molybdenum disulfide (MoS2) through covalent chemical functionalization, and thereby directly influence the kinetics of the hydrogen evolution reaction (HER), surface energetics, and stability. Chemically exfoliated, metallic MoS2 nanosheets are functionalized with organic phenyl rings containing electron donating or withdrawing groups. We find that MoS2 functionalized with the most electron donating functional group (p-(CH3CH2)2NPh-MoS2) is the most efficient catalyst for HER in this series, with initial activity that is slightly worse compared to the pristine metallic phase of MoS2. The p-(CH3CH2)2NPh-MoS2 is more stable than unfunctionalized metallic MoS2 and outperforms unfunctionalized metallic MoS2 for continuous H2 evolution within 10 min under the same conditions. With regards to the entire studied series, the overpotential and Tafel slope for catalytic HER are both directly correlated with the electron donating strength of the functional group. The results are consistent with a mechanism involving ground-state electron donation or withdrawal to/from the MoS2 nanosheets, which modifies the electron transfer kinetics and catalytic activity of the MoS2 nanosheet. The functional groups preserve the metallic nature of the MoS2 nanosheets, inhibiting conversion to the thermodynamically stable semiconducting state (2H) when mildly annealed in a nitrogen atmosphere. We propose that the electron density and, therefore, reactivity of the MoS2 nanosheets are controlled by the attached functional groups. Functionalizing nanosheets of MoS2 and other transition metal dichalcogenides provides a synthetic chemical route for controlling the electronic properties and stability within the traditionally thermally unstable metallic state. the edge sites, but not the basal sites.9,10 To increase the HER activity, basal sites have been activated by introducing vacancies,11,12 straining the 2D layers,5,13 coupling with plasmons,14−16 and by conversion to the metastable metallic 1T state.11,17−20 Although the 1T phase has high catalytic activity due to both edge and basal sites being active for HER, this phase is ultimately not stable and reverts back to the 2H phase with time or heat. In order for MoS2 to be catalytically useful, this metastable 1T phase needs to be stabilized. Ideally, the stabilization process would also allow for systematic control over the catalytic activity of these 2D metallic TMDCs to target HER and other applications. Recent reports have shown increased HER activity and stability for 2D TMDCs when interfaced with certain substrates; specifically, substrates that can donate electron charge/density to the 2D TMDC have shown increased HER activity.13,18,21 Liu et al. grew 1T MoS2 nanopatches on the surface of single-walled carbon nanotubes (SWNT) and measured higher HER activity and electrochemical cycling

1. INTRODUCTION Electrochemical reduction of protons to molecular hydrogen (H2) is a carbon-free energy conversion technology that, with increased research and development, could be a front-runner for renewable fuels. Currently, the most efficient catalyst (platinum) for H2 generation is too expensive and consequently is not produced on a large enough scale to be used as a global energy resource;1 therefore, a lower cost catalyst with high efficiency is needed. Molybdenum disulfide (MoS2) is one of the most studied 2D transition metal dichalcogenides (TMDCs) for hydrogen evolution reaction (HER) catalysis, making it a useful model system for understanding the fundamental link between TMDC atomic structure and the thermodynamics and kinetics of this technologically relevant reaction. MoS2 is a good candidate to replace platinum for HER catalysis because it is typically solution processable, earth abundant, nontoxic, and has demonstrated high catalytic activities for HER.2−8 It can also be readily reduced to twodimensional (2D) nanostructures, thus increasing the relative area of catalytically active sites. It has been shown that the more thermodynamically stable 2H (semiconducting) phase of MoS2 is catalytically active at © XXXX American Chemical Society

Received: October 21, 2017

A

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

Figure 1. (a) Schematic of the various functional groups on 1T MoS2. (b) Atomic force microscopy (AFM) image and (c) extracted line profiles that demonstrate the height and length of chemical exfoliated 1T MoS2 nanosheets. We observe a height distribution of 3−5 nm and a length distribution of 250−400 nm from our AFM topographic images.

stability for 1T MoS2/SWNT films when compared to pure 1T MoS2.18 They attribute this increased electrocatalytic performance to charge transfer from the SWNT to the 1T MoS2. Another example by Shi et al. highlights that doping 2H MoS2 nanosheets with Zn increases the negative charge on the MoS2, which improves the electrochemical activity and stability toward HER.22 Such studies elucidate the general importance of charge density for the HER activity of TMDC nanosheets. The electronic properties, stability, and reactivity of MoS2 can also be modified by chemical functionalization of the basal sites.2,3 Typically, sulfur vacancies or defects are used to functionalize 2H MoS2 by forming a bond to exposed Mo 4d orbitals.23 While the 2H-MoS2 basal sites are relatively inert, 1T-MoS2 basal sites are more reactive toward functionalization.7,24,25 Recently, Knirsch et al. employed a novel route to functionalize 1T MoS2 by reaction of the reduced, chemically exfoliated MoS2, with substituted phenyl diazonium salts to form an organosulfur (S−C) bond.7 This produced MoS2 with a surface attached methoxybenzene. Their studies concluded that the MoS2 was still in its metallic 1T phase after adding the functional group; the functional group inhibited conversion of MoS2 to the semiconducting 2H phase at room temperature over a month period. This approach is unique because it allows for functionalization of the MoS2 surface without relying on edge sites or defects. Voiry et al. also demonstrated functionalization of 1T MoS2 basal sites with organic groups (−CH3 and −H) and bromobenzene.24 Additional studies exploring the relationships between MoS2 functionalization, electronic structure, HER catalysis, and surface energetics are needed. In this study, we report the synthesis and functionalization of a series of chemically exfoliated MoS2 nanosheets in order to investigate the effects of surface modifiers on HER activity, electronic structure, and stability. The 1T-MoS2 nanosheets are modified with a series of substituted phenyldiazonium salts wherein the phenyl group forms a S−C bond to MoS2. X-ray photoelectron spectroscopy (XPS) is used to verify the functionalization of 1T MoS2 and to measure the workfunction of the modified surfaces. We then perform HER studies of the functionalized nanosheets in order to determine how the surface influences the reaction kinetics and activity of MoS2.

The workfunction and HER activity are correlated to the Hammett parameter (surface dipole) of the modifier. Furthermore, we explore the stability of the functionalized metallic MoS2 by measuring the chemical environment (XPS) and HER activity before and after annealing. The stability of both functionalized and unfunctionalized MoS2 is further explored during continuous HER catalysis over a 2 h period. The functional groups provide increased stability for the catalytically active 1T phase of the MoS2 nanosheets. By systematically varying the organic substituents from electron withdrawing to electron-donating groups, we show that the electronic properties and catalytic reactivity of the MoS2 nanosheets can be tuned via different functional groups and argue that such tuning results from differences in charge density of the nanosheets.

2. RESULTS AND DISCUSSION Synthesis and Characterization of 1T and Functionalized MoS2 Nanosheets. Functionalized 1T MoS2 nanosheets are synthesized following the procedure developed by Knirsch et al., as detailed in the Materials and Methods Section.7 Briefly, bulk (2H) MoS2 is solution-exfoliated via intercalation with n-butyl lithium and the subsequent reaction with water produces nanosheets that are primarily in the 1T MoS2 phase. After centrifugation/purification of these metallic nanosheets, the nanosheets are reacted with a series of five diazonium salts (in separate reactions) to form the series of functionalized MoS2. Throughout this work, we refer to pNO2Ph-MoS2, 3,5-Cl2Ph-MoS2, p-BrPh-MoS2, p-OCH3PhMoS2, and p-(CH3CH2)2NPh-MoS2 as NO2Ph, Cl2Ph, BrPh, OMePh, and Et2NPh, respectively, and the pristine 1T MoS2 nanosheets as 1T. Note that when appropriate, we include the as-purchased 2H MoS2 powder for comparison and label as bulk (2H). The five organic moieties have a phenyl ring substituted with various functional groups that are either electron donating or withdrawing (Figure 1a). This results in a range of electron-donating/withdrawing properties of the organic functional group due to substituents possessing a range of Hammett parameters, which are correlated to their dipole moments. The more negative the Hammett parameter, the more electron donating the functional group is to the B

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

Figure 2. (a) Raman spectra of pristine and functionalized 1T MoS2 along with bulk (2H) MoS2 as a comparison. The spectra are taken with an excitation wavelength of 633 nm. All spectra contain the expected MoS2 transitions: E12g (∼383 cm−1) and A1g (∼406 cm−1). The 1T MoS2 has the characteristic metallic peaks (J1, J2, and J3 located at ∼154, ∼230, and ∼326 cm−1, respectively). The functionalized MoS2 spectra have the J1 and J2 peaks but do not have the J3 peak. (b) DRIFTS spectra of MoS2 nanosheets functionalized and bare (1T). Diagnostic peaks are labeled and discussed in detail in the text.

phenyl ring and concomitantly to the MoS2 surface. Et2NPh possesses the most electron-donating group and has a corresponding Hammett parameter of −0.43. Conversely, more positive Hammett parameters correspond to more electron-withdrawing functional groups, with NO2Ph having the largest value at +0.78. Hammett parameters are taken from Hansch et al., where the value for Cl2Ph is taken as twice that of m-ClPh.26 We characterize the MoS2 nanosheets with atomic force microscopy (AFM), scanning electron microscopy (SEM), and Raman spectroscopy. The 1T MoS2 nanosheets are not monodisperse in lateral size or layer number. The AFM image in Figure 1b shows three MoS2 nanosheets that are solution deposited onto a silicon substrate and are representative of the ensemble of nanosheets (thicknesses in the range of 3−5 nm and lateral dimensions from 250 to 400 nm). The corresponding height profiles are shown in Figure 1c. To gain additional morphology information, we measure 1T and Et2NPh MoS2 with SEM (Figure S1); we do not observe any obvious changes to the morphology after functionalization. In addition, we characterize the 1T and functionalized MoS2 nanosheets with Raman, using an excitation wavelength of 633 nm (Figure 2a). All MoS2 spectra contain the expected transitions at ∼383 cm−1 (E12g) and ∼406 cm−1 (A1g). The Raman spectrum of the 1T MoS2 has distinct features at 154, 230, and 326 cm−1, which correspond to the J1, J2, and J3 modes, respectively, and is consistent with previous literature measurements.7,17,27−29 The functionalized MoS2 samples also have J1 and J2 peaks, whereas the J3 peak is not observed with appreciable signal-to-noise, indicating that the functionalized MoS2 mainly remains in the 1T phase. Interestingly in the Knirsch et al. study, the J features were not observed at room temperature for functionalized MoS2, but were observed at elevated temperature (200 °C), a result that the authors attribute to a temperature-dependent resonance enhancement effect.7

We quantify the chemical environments of functionalized and bare MoS2 via diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), XPS, and combustion analysis. DRIFTS spectra (Figure 2b) confirm the presence of the functional group and phenyl ring in all derivatized sheets except for the Et2NPh sample, which was not measured. Note that all IR peaks are assigned using ref 30. The NO2Ph spectrum shows characteristic asymmetric (1518 cm−1) and symmetric (1342 cm−1) stretches for NO2, along with the NO2 scissor mode at 853 cm−1. The OMePh spectrum shows symmetric (1463 cm−1) and asymmetric (1440 cm−1) stretches for OMe group, along with Ph-O (1291 cm−1 and 1250 cm−1) and PhO-CH3 (1025 cm−1) stretches as well as the O-CH3 rocking mode (1180 cm−1). The DRIFTS data (Figure 2b) also provide a measurement of the polar interaction between the MoS2 and the phenyl ring. The phenyl ring has four characteristic modes, split into two quadrant deformations around 1550−1600 cm−1 and two semicircle deformations around 1450−1500 cm−1. The higherenergy semicircle deformation around 1500 cm−1 is only IR active when the phenyl ring has an electron donating group on it. This is consistent with the presence of a strong feature at 1492 cm−1 in the OMePh spectrum, but the lack of a higher energy peak in the BrPh and Cl2Ph spectra (such a feature would likely be obscured by the NO2 asymmetric peak in the NO2Ph spectrum). In addition, the quadrant stretch modes are IR inactive for para-substituted phenyl rings where the substituents have identical electron-donating or -withdrawing character, maintaining a mirror of symmetry. Interestingly, the BrPh spectrum shows almost no quadrant stretching, despite the fact that the para substituents are Br and MoS2. This indicates that the MoS2 contributes a similar amount of electron density to the phenyl ring as the Br group and has a similar Hammett parameter of around 0.23. The bare diazonium salts are shown in Figure S2; and the reference spectrum for BrPh (4-Bromobenzenediazonium tetrafluoroboC

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

and Cl for Cl2Ph, data not shown). In the case of OMePh, O cannot be used as a unique identifier in the XPS experiment due to the low levels of surface oxygen contamination and/or MoOx formation, but DRIFTS confirms functionalization for this functional group. Table 1 highlights the S:Mo and S:X

rate spectrum) clearly contains this quandrant stretching when not bound to MoS2, which is around 1550−1600 cm−1. In support of our analysis, the quadrant stretches on the other three phenyl groups behave as expected: the OMePh spectrum shows the normal intensity ratio where the higher energy peak at 1594 cm−1 is more intense than the peak at 1570 cm−1; the Cl2Ph spectrum shows an inverted intensity ratio for its peaks due to extending resonance to the two Cl substituents (1562 cm−1 peak is more intense than the 1591 cm−1 peak); and the NO2Ph spectrum shows roughly equal intensity of the 1598 and 1576 cm−1 peaks due to its less-complete resonance with the phenyl ring. XPS is utilized to determine the atomic compositions of the MoS2, percentage of MoS2 that is in the 1T phase, and the workfunction of the different films. The Mo 3d and S 2p corelevel XPS results for the modified and unmodified MoS2 are shown in Figure 3a, b. Our XPS measurements on both the modified and unmodified MoS 2 films show that the concentration of impurity C and O is low. For each of the modified films, we observe the atomic species expected for each functional group (N for Et2NPh and NO2Ph, Br for BrPh,

Table 1. XPS Atomic Ratiosa (S:Mo and S:X), % of MoS2 in the 1T Phase (%1T), and Workfunction (ϕ) Energies for the Modified and Unmodified 1T MoS2 Thin Films MoS2

S:Mo

S:Xb

%1Tc

ϕ (eV)d

1T Et2NPh OMePh BrPh Cl2Ph NO2Ph

2.0 2.6 2.2 2.1 2.6 2.2

NA 2.3 NA 2.4 0.4 1.8

81 81 78 82 84 76

3.81 4.06 4.25 4.75 4.85 4.68

The atomic percentages have a ± 5% error. bX is the unique atom of the functional group, e.g., X = Cl for Cl2Ph. cThe %1T is taken from Mo 3d fits, where MoO2 is included. See the Supporting Information for more details. Estimated standard deviation of ± 8%. dThe energy uncertainty for ϕ is ± 25 meV. a

ratio, where X is the unique atom for that functional group. The ratios are determined from multiple XPS measurements, where the atomic percentages have a ± 5% error. The S:Mo ratios are close to the expected value of 2. The S:X ratios vary between samples because of differences in functional group coverage on the MoS2 from batch to batch and the functionalization activity of the different groups; therefore, we only use X as a qualitative confirmation of the functional group. In addition to the XPS results, we use elemental analysis measurements to give relative quantities between Et2NPh and NO2Ph. For the CHN combustion analysis (see Methods and Materials section for more details), we use the dried powders of the Et2NPh and NO2Ph MoS2 nanosheets. This elemental analysis confirms the presence of N in both powders (Table S1). However, the relative amount of CHN, per mole of MoS2, is reduced in the Et2NPh compared to the NO2Ph powder. This result verifies a lower functional group coverage on MoS2 for the Et2NPh group compared to NO2Ph group and is consistent with the XPS S 2p results (Figure 3b and 3c). In Figure 3b, the XPS spectra of S 2p confirms the presence of S−C bonding for the functionalized MoS2. The 1T MoS2 spectrum (bolded gray trace) clearly shows the S 2p spin orbit splitting of the metallic state (and a small contribution from the 2H state). The functionalized films have additional features at higher binding energy (162.5−165 eV), which are consistent with a S−C bond. The observation of this S−C bond gives additional evidence that the OMePh sample is functionalized, since O cannot be used as a unique identifier as discussed above. The functionalized MoS2 films with electron withdrawing groups (NO2Ph, Cl2Ph, and BrPh) have more intensity in the S−C region than the electron donating groups (Figure 3b). This suggests that the electron withdrawing groups have greater functional group coverage on the MoS2 than the electron donating groups and is consistent with the CHN combustion analysis. Fits to the core level XPS spectra are shown in Figure S3. Figure 3c plots the ratio of the areas of the 1T Mo−S bond to the S−C bond. It is clear from this figure that the more electron withdrawing groups have greater coverage on the MoS2 than the more electron donating groups.

Figure 3. XPS of (a) Mo 3d and (b) S 2p for 1T (bare) and functionalized MoS2. Because the functional group attaches to the S atom, the Mo 3d spectra are similar and the S 2p spectra differ between bare and functionalized MoS2. The Mo 3d spectra highlight that the MoS2 is dominated by the metallic phase. The 1T spectra are bolded for easier comparison. (c) Ratio of 1T Mo−S to S−C peak areas as a function of Hammett parameter. The more electron withdrawing groups (more positive Hammett parameter) have more functional groups attached to the MoS2 nanosheets than electron donating groups (negative Hammett parameter). The color scheme is the same in all panels and is defined in panel a. D

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

(requires more energy to remove an electron from the surface, deeper workfunction). Although there is some variation in the workfunction measurements of each sample type, due to differences in the amount of functional group per MoS2, there is a positive correlation between the workfunction and Hammett parameter, with a large change of approximately 800 meV across the entire series. These differences in surface energetics in turn influence the HER activity of the MoS2 nanosheets, which is discussed in the next section. HER Activity. MoS2 has been extensively studied as an electrocatalyst for the reduction of protons to molecular hydrogen. To determine the effects of chemical functionalization on HER, we drop-cast dispersions of the chemically modified MoS2 onto freshly polished glassy carbon electrodes to acquire linear sweep voltammograms (LSVs, Figure 5). Typically, LSVs are used to determine the overpotential, Tafel slope, and current density of an electrochemical reaction of interest. The overpotential is the excess energy above the required thermodynamic potential in order to achieve a particular rate for hydrogen evolution. The Tafel slope is related to the kinetic rate and exchange current of the electrochemical reaction, and its slope can provide information on rate-determining steps for the reaction. With respect to the HER, it is desirable to minimize the Tafel slope, as this typically reduces the overpotential needed to reach appreciable catalytic current densities and is indicative of facile kinetics. As the substituent on the phenyl ring is changed from the most electron withdrawing (Cl2Ph, Hammett parameter = 0.74) to the most electron donating (Et2NPh, Hammett parameter = −0.43), we observe a systematic shift to lower values in the overpotential and an increase in the catalytic rate toward hydrogen evolution. As the Hammett parameter of the substituent decreases, the overpotential required to achieve 10 mA/cm2 catalytic current density decreases from 881 mV for Cl2Ph to 348 mV for Et2NPh. This ∼500 mV shift is accompanied by a decrease in the Tafel slope from 213 to 75 mV/dec A. tabulation of the electrochemical parameters can be found in Table 2. The Tafel slope of the most electron donating functional group (Et2NPh) is slightly worse than 1T MoS2 and, the more electron withdrawing groups perform poorer than bulk 2H MoS2. This correlation between the functional groups and HER activity gives insight into the mechanism by which pendant chemical groups interact with the MoS2 surface. Figure 6 highlights the correlation between Hammett parameter and the activity of the functionalized MoS2. As the Hammett parameter is increased, so does the Tafel slope and the overpotential required to reach a catalytic current density of 10 mA/cm2. NO2Ph is not included in these graphs, as the nitro group is not stable under reductive conditions in acidic solution, presumably forming the ammonium substituted phenyl, similar to the Bechamp reduction of aromatic nitro compounds to the corresponding anilines.31 HER Mechanism. The observed HER activity is consistent with the behavior of 2D MoS2 films interfaced with electrondonating substrates/films. It has been reported that the HER activity is correlated with the electron density of the MoS2 sheet and sheets with higher reported electron density show increased activity.13,18,21,32 Chia et al. demonstrated that 2H and 1T-MoS2 HER activity could be increased when the films are reduced, and these experimental observations were supported by DFT calculations.32 The correlation of HER activity with excess charge density on the MoS2 sheet results

Mechanistically, the electron-withdrawing groups readily functionalize the negatively charged 1T MoS2 nanosheets compared to the electron-donating groups; the degree of functionalization depends on the balance of nanosheet charge density and the functional groups neutralizing the charge. This observation is consistent with previous literature results measuring the reduced negative charge on 1T MoS2 following functionalization.7,24 The S 2p peak positions of the S−C and 1T Mo−S bonds are consistent between the different functionalized films and do not have a dependence on Hammett parameter (Figure S4). High-resolution XPS core level measurements are sensitive to changes in the local atomic environment and are reported with respect to the Fermi level (i.e., electron binding energy). As expected, the functionalized films have an additional S peak because of the new bond formed between the MoS2 and the functional group. This appears at higher binding energy with respect to the Mo−S bond. As for the S− C bond peak position, this does not change with functional group because the local atomic environment is the same (specifically, the S oxidation state is the same). Therefore, the Fermi level to core level energy difference is not changing with functional group within our resolution. The Mo 3d peaks (Figure 3a) also give detailed information about the chemical environment. Like S 2p, Mo 3d also has a spin orbit splitting (3.13 eV). The Mo 3d peak contains contributions from MoS2, MoO3, and MoO2. For the S:Mo ratio, only the integrated Mo 3d area associated with MoS2 is used (fits to the spectra are shown in Figure S3). From individual fits of the spectra, we quantify the percentage of MoS2 in the metallic phase, which is ∼80% (Table 1). Importantly, Figure 3b demonstrates that the functional groups do not change the Mo 3d features and the Mo environment remains the same. This again verifies that the functional groups are bonding to the S and not to the Mo. Surface Energetics. To determine the degree to which the functional groups influence MoS2 surface energetics, we use XPS to measure the workfunction (difference between the vacuum and Fermi levels). Table 1 lists the average workfunction for the functionalized MoS2 samples, where 2− 5 films from separate reactions are measured for each functional group. There is a clear trend in how the workfunction varies with chemical functionalization (Figure 4); as the Hammett parameter increases (more electron withdrawing), the workfunction at the surface increases

Figure 4. Relationship between workfunction and Hammett parameter. The most electron donating (Et2NPh) has the shallowest workfunction. The Hammett parameter has a positive correlation to the workfunction and varies by 800 meV over the various functional groups. E

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

Figure 5. (a) Linear sweep voltammograms (LSV) and (b) corresponding Tafel plots for glassy carbon electrodes deposited with functionalized, 1T, and bulk (2H) MoS2. LSVs are taken at 5 mV/s in 0.5 M H2SO4 with a Ag/AgCl reference electrode and vitreous carbon counter electrode. Both panels have been iR corrected. Pt and glassy carbon electrodes for HER are included in (a) as a reference.

argue that in our study the electron withdrawing (electrophilic) groups remove more electron density from the nanosheets than the electron donating (nucleophilic) groups, which ultimately determines the amount of functional group per MoS2 nanosheet. To further support this HER mechanism, we perform electrochemical impedance spectroscopy (EIS) to probe the electron transfer kinetics at the MoS2/electrolyte interface, as the low-frequency region in the EIS can be used to determine the charge transfer resistance of the interface. From the EIS data, we use the radius of the half circle to qualitatively describe the charge transfer resistance (Nyquist plots). The EIS data (Figure 7) show that the Et2NPh electrode behaves similarly to that of 1T MoS2. As the functional group becomes more electron-withdrawing, the charge transfer resistance of the functionalized MoS2 is larger than bulk 2H MoS2. The EIS and HER electrochemistry results complement each other, and support our mechanism that the functional group on 1T MoS2 is either donating or removing electron density from the nanosheet, which in turn changes the electron-transfer kinetics and HER activity. When excess charge is present on the MoS2 nanosheets, the metallic phase maintains its high sheet conductance and low charge transfer resistance; conversely, when the excess charge is removed (by electron withdrawing groups), the sheet conductance is reduced and the charge transfer resistance is increased, making the Cl2Ph and BrPh MoS2 behave worse than the bulk semiconducting phase for HER. On the other hand, we cannot rule out that the amount of functional group per MoS2 is also influencing the HER activity and electron-transfer kinetics. It is possible that the functional group is blocking the reactive S

Table 2. Electrochemical Parameters with iR Corrections for Modified, 1T, and Bulk (2H) MoS2; Tafel Slope, Exchange Current Density (j0), and Overpotential (η) at 10 mA/cm2 Are Listed for Each of the Electrodesa Hammett parameterb Cl2Ph BrPh OMePh Et2NPh 1T Bulk (2H)

0.74 0.23 −0.27 −0.43 ∼0.23c NA

Tafel slope (mV/dec) 213 170 136 75 61 186

± ± ± ± ± ±

33 7 8 3 7 54

j0 (μA/ cm2) 1.5 0.4 1.2 0.3 1.9 1.1

± ± ± ± ± ±

1.3 0.1 0.4 0.1 2.5 0.7

η, i = 10 mA/ cm2 (mV) 881 822 691 348 271 737

± ± ± ± ± ±

13 11 23 7 36 46

a

Averages and standard deviations are from three separate electrode preparations. bValues taken from Taft et al.26 cEstimated equivalent value obtained from DRIFTS data.

from improved electron-transfer kinetics. These studies provide a framework for understanding the variations in HER activity for our functionalized MoS2, where more electron-withdrawing organic substituents remove charge from the sheets and lower the activity toward HER. This argument is supported by previous reports by Voiry et al. and Knirsch et al.7,24 They measured the zeta potential of the MoS2 nanosheets before and after functionalization and concluded that the excess charge on the MoS2 nanosheets was quenched when forming a S−C bond. The change in the zeta potential resulted in the functionalized MoS2 nanosheets precipitating out of solution because of the decreased electrostatic repulsion that stabilized the 1T MoS2 dispersion in water. Therefore, we

Figure 6. Average (a) overpotential (taken at current density of 10 mA/cm2) and (b) Tafel slope for the MoS2 functionalized electrodes. Measurements are performed in 0.5 M H2SO4 with a Ag/AgCl reference electrode and a vitreous carbon counter electrode. F

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

results, we study the modified and unmodified MoS2 films before and after annealing. When samples are annealed (150 °C, 24 h) under lab atmosphere, the MoS2 undergoes conversion to MoO3 and very little MoS2 remains, as determined by XPS core level analysis (Figure S5). It is evident from the spectra that the 1T MoS2 cannot withstand aerobic conditions at elevated temperatures. However, when the films are annealed in a N2 glovebox at 150 °C for 24 h, the MoS2 does not convert to MoO3. Figure 8 shows the XPS core level data of Mo 3d and S 2p before and after annealing for the 1T, OMePh, and Et2NPh MoS2. The 1T MoS2 film converts from being predominantly in the metallic phase (∼80% 1T) as determined by Mo 3d peak fitting) to a predominance of the semiconducting phase (∼75% 2H) under these conditions (Figure 8a, d). In comparison, the XPS data for the functionalized MoS2 does not show a change in the Mo 3d and S 2p peaks for OMePh, BrPh, Cl2Ph, and NO2Ph (BrPh, Cl2Ph, and NO2Ph annealed spectra are similar to OMePh and are shown in Figure S6). The spectra show no appreciable changes after annealing, indicating that the metallic nature of MoS2 is preserved for functionalized MoS2, relative to 1T MoS2, for relatively aggressive annealing. This result suggests that functionalization through a S−C bond and/or removing some amount of excess charge inhibits conversion back to the 2H state. The Et2NPh films do change before and after annealing but to a smaller degree than the 1T MoS2. This conversion could be due to excess charge remaining on the MoS2 nanosheets following functionalization with the electron donating (nucleophilic) Et2NPh group and/or relatively poor coverage of Et2NPh. The mechanism for 1T to 2H conversion is not well studied and is an area of interest that is being explored. However, the stability brought on by functionalization fits with our proposed mechanism by which the functional groups influence the HER mechanism. Specifically, the more electron-withdrawing (electrophilic) functional groups remove excess charge from the nanosheets, while “locking in” the 1T

Figure 7. Electrochemical impedance spectra (Nyquist plots) for functionalized, 1T (bare), and bulk (2H) MoS2 on glassy carbon electrode at −0.29 V vs RHE.

site, and as more functional groups are present (electronwithdrawing groups) the number of active sites decreases. However, we propose that the amount of functional group attached to the MoS2 nanosheets is a balance of MoS2 nanosheet charge and the Hammett parameter. Additional studies need to be performed to determine the relation between the amount of functional group on MoS2 and the corresponding HER performance but is beyond the scope of this work. Stability. Although 1T MoS2 is more catalytically active for HER than 2H MoS2, the 1T phase is thermodynamically unstable and with time or heat reverts back to the 2H phase. Understanding how to preserve the metallic state of MoS2 is important to maintaining the catalytic activity of the basal sites. Knirsch et al. showed that when the reduced metallic MoS2 was functionalized with OMePh, the metallic phase was preserved in ambient conditions for over a month compared to the bare metallic MoS2.7 To broaden and generalize these

Figure 8. XPS of (a−c) S 2p (left) and (d−f) Mo 3d (right) spectra before (blue) and after (red) annealing for (a, d) 1T and (b, c, e, and f) functionalized MoS2. The nanosheets are annealed in a N2 atmosphere for 24 h at 150 °C. The functional groups protect the MoS2 from undergoing conversion from the 1T to the 2H state. (c, f) The Et2NPh MoS2 undergoes some conversion to the 2H phase. G

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

Figure 9. HER stability tests for Et2NPh and 1T MoS2. (a) LSV for glassy carbon electrodes deposited with Et2NPh, 1T, and bulk (2H) MoS2. The scans are performed for as prepared electrodes and then electrodes following an anneal in a N2 glovebox at 150 °C for 24 h. LSVs are taken at 5 mV/s in 0.5 M H2SO4 with a Ag/AgCl reference electrode and a vitreous carbon counter electrode. (b) Overpotential is measured as a function of time to maintain a current density of 10 mA/cm2. Et2NPh outperforms 1T MoS2 for H2 generation within ∼7 min.

3. CONCLUSIONS Our results demonstrate a strong correlation between the electron donating strength of substituents and the surface energetics, electron transfer resistance, and the HER catalytic activity of functionalized MoS2 nanosheets. The functionalized nanosheets are more stable to the thermally initiated phase transformation from the metallic 1T phase to the semiconducting 2H phase. Furthermore, we show for an exemplary functionalized sample (Et2NPh-MoS2) that functionalization leads to better stability and long-term performance under HER conditions. These results provide a framework for understanding and controlling the balance between catalytic activity and stability for these unique 2D materials. Formation of S−C bonds via covalent surface functionalization protects the catalytically active, but metastable, 1T phase. However, the HER catalytic activity is compromised for functional groups that remove appreciable electron density from the MoS2 nanosheets and have more functional groups per MoS2 nanosheet. Thus, there is ultimately a balance to be struck between catalytic activity (optimized initially for 1T, relative to 2H and functionalized MoS2) and stability (using a functional group that forms a S−C bond to kinetically protect the metastable 1T phase). We believe our findings encourage further exploration of novel functionalization strategies that offer the ability to balance the catalytic activity and stability over a broad class of 2D materials.

state. When excess charge remains on the sheets, like with Et2NPh, the nanosheets have a lower thermodynamic barrier for the 1T to 2H conversion. Theoretically and experimentally, it has been demonstrated that 2H MoS2 can be converted to 1T MoS2 by electron beam bombardment via in situ scanning transmission electron microscopy,33,34 where the excess charge drives the conversion. This argument is further supported by theoretical calculations by Gao et al.; they calculated minimum energy pathways from 2H to 1T (or vice versa) with different charge states. 35 Their calculations highlight that the thermodynamic barrier separating the 1T and 2H phases decreased when there is more charge on the MoS2. This calculation along with transmission electron microscopy experiments, suggest that removing excess charge in the 1T phase stabilizes this phase. To test the stability of the functionalized MoS2 under HER conditions, we perform two separate experiments. First, the LSVs are measured for both “fresh” electrodes (consistent with those presented in Figure 5) and electrodes that are annealed at 150 °C for 24 h under N2 environment (similar to conditions in Figure 8). The effect of annealing on HER performance is shown for the Et2NPh and unfunctionalized 1T MoS2 in Figure 9a. The unfunctionalized 1T MoS2 significantly degrades and approaches that of the bulk (2H) MoS2 electrode. This result is consistent with the XPS annealing data (Figure 8a, d). The Et2NPH MoS2 electrode is more resilient to this annealing step and only slightly degrades, which is again consistent with the XPS annealing data (Figure 8c, f). Second, we compare the durability of the Et2NPh and unfunctionalized 1T MoS2 electrodes under HER conditions over a 2 h period (Figure 9b), where the overpotential required to maintain 10 mA/cm2 is monitored as a function of time. As can be seen in Figure 9b, the unfunctionalized 1T MoS2 performance quickly degrades and becomes worse than the Et2NPh electrode within 7 min. Over the 2-h period, the 1T MoS2 electrode requires an additional 0.139 V to maintain the current density. This is very different from the Et2NPh electrode, which degrades only slightly over this time period. The Et2NPh electrode only requires an additional 0.014 V to maintain 10 mA/cm2 over 2 h. The enhanced stability of Et2NPh compared to the unfunctionalized 1T MoS2 is very encouraging for utilizing these types of functionalized nanosheets for realistic long-term catalysis (e.g., HER) applications.



MATERIALS AND METHODS

MoS2 Preparation and Functionalization. MoS2 powder was obtained from Sigma-Aldrich and vacuum-dried at 100 °C overnight prior to use. The chemically exfoliated 1T MoS2 was prepared in a similar method as Knirsch et al.,7 wherein 5 mL of nBuLi (2.5 M) in hexanes is added to a suspension of 500 mg (3.1 mmol) of MoS2 in 5 mL of dry hexanes and allowed to stir under an inert atmosphere for 48 h. Reaction is then quenched with ∼100 mL of Milli-Q water. After hydrogen evolution ceases, the resulting suspension is then washed twice with ∼100 mL of hexane to remove organic impurities and then tip sonicated at ∼120 W for 1 h in an ice bath. The solution is centrifuged at 800 rpm for 90 min to remove unreacted material. The solution is then decanted off and subjected to three centrifugations at 13200 rpm (21 400 g) for 90 min at 20 °C to remove small amounts of MoS2 and LiOH. The MoS2 was functionalized by suspending ∼100 mg of 1T MoS2 in Milli-Q water and adding dropwise ∼5 mL of a 10 mg/mL solution of the corresponding tetrafluoroborate salt: 4-p-diazo-N,N-Diethylaniline Fluoborate (MP Biomedicals), 4-Methoxybenzenediazonium H

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society



tetrafluoroborate (Sigma-Aldrich), 3,5-dichlorophenyldiazonium tetrafluoroborate (Sigma-Aldrich), 4-bromobenzenediazonium tetrafluoroborate (Sigma-Aldrich), or 4-nitrobenzenediazonium tetrafluoroborate (Sigma-Aldrich). The solutions were then allowed to stir overnight. The resulting precipitate was collected by filtration and washed twice with 20 mL of water to remove any unreacted material. The resulting material was then dried under vacuum. Characterization. 1T and functionalized MoS2 films were prepared for the various characterization experiments. The films were made by suspending the 1T MoS2 in DMF or suspending the modified MoS2 in DMF or anisole and then drop-casting solutions of the modified and unmodified MoS2 nanosheets onto different substrates (Si substrate, AFM and Raman; Au substrate, XPS; glassy carbon substrate, electrochemistry). All films are stored under a flowing N2 environment (atmospheric pressure) or vacuum until being removed and exposed to ambient air for limited time. CHN analysis of the Et2NPh and NO2Ph powders were performed by Midwest MicroLab (Indianapolis, IN). Photoelectron Spectroscopy. XPS data were obtained on a Physical Electronics 5600 system using Al Kα radiation. The XPS setup was calibrated with Au metal, which was cleaned via Ar-ion sputtering. The energy uncertainty for the core level data is ± 0.05 eV and for the workfunction measurements are ± 0.025 eV. In order to measure XPS on our series of MoS2 nanomaterials, thin films were made on Au substrates by solution deposition. All samples were checked for and did not exhibit charging, which was verified by X-ray power dependence measurements. Atomic percentages have ± 5% error. Electrochemistry. Electrochemical measurements were controlled by a CH Instruments 600D potentiostat coupled with a Pine analytical rotator. Measurements were taken in 0.5 M H2SO4 with a Ag/AgCl reference electrode and vitreous carbon counter electrode. For a typical measurement, 10 μL of a 1 mg/mL solution of the nanosheets suspended in DMF was drop cast onto a freshly polished 5 mm diameter glassy carbon electrode. All LSVs were performed at 5 mV/s and 1600 rpm and the electrolyte was degassed for 15 min with N2 prior to experiments. Impedance measurements were carried out with the same setup with no rotation and measured at frequencies ranging from 1 GHz to 10 Hz at a constant overpotential of −0.29 V vs RHE. Confocal Raman. An inVia Renishaw confocal Raman microscope with a Coherent HeNe 633 nm laser was used for characterizing the Raman signatures of MoS2 film samples with and without functionalization. The samples were scanned by a 100× objective lens with 5% laser intensity (∼0.5 mW) and dispersed by 1800 lines mm−1 in air under ambient conditions. DRIFTS. Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) spectra were acquired on a Bruker ALPHA FTIR Spectrometer using the DRIFTS sampling accessory. The samples were deposited by drop-casting onto aluminum-coated polished silicon wafer fragments (roughly 5 mm x 5 mm), and the instrument was baselined against fragments from the same wafer. The instrument settings for both baseline and sample were 128 scans, 2 cm−1 resolution, from 360 to 7000 cm−1. Atomic Force Microscopy. Atomic force microscopy (AFM) was used to image the 2D MoS2 flakes that were solution deposited onto a silicon substrate. The ambient environment AFM uses a Park AFM XE-70 controller and is housed inside an acoustic box that is located on top of a vibration isolation table. Budget sensors silicon cantilevers (Tapp300G, ∼300 kHz) were used to image surface topography images in intermittent contact mode. Scanning Electron Microscopy. Scanning electron microscopy (SEM) was used to image the MoS2 nanosheets that were solution deposited onto a silicon substrate in a concentration that is consistent with the electrodes. All SEM images were taken in a FEI Nova 630 system.

Article

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b11242.



Experimental details and additional characterizations (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Eric E. Benson: 0000-0001-6364-6472 Jeffrey L. Blackburn: 0000-0002-9237-5891 Elisa M. Miller: 0000-0002-7648-5433 Author Contributions †

E.E.B and H.Z. contributed equally to this work

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank Todd G. Deutsch and Jao van de Lagemaat for fruitful discussions. We would also like to thank Bobby To for all SEM images and Al Hicks for the Table of Contents figure. The DRIFTS data and analysis as well as the preliminary studies were supported by the Laboratory Directed Research and Development (LDRD) Program at the National Renewable Energy Laboratory (NREL). SAS was supported by the U.S. Department of Energy, Office of Science, Office of Workforce Development for Teachers and Scientists (WDTS) under the Science Undergraduate Laboratory Internship (SULI) program. The majority of the work was supported by the U.S. Department of Energy, Office of Science, Basic Energy Science, Division of Chemical Sciences, Geosciences and Biosciences, under Contract DE-AC36-08GO28308 to NREL. The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a nonexclusive, paid up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for U.S. Government purposes.



REFERENCES

(1) Platinum Quarterly Q2 2017; World Platinum Investment Council: London, 2017. (2) Wang, H.; Yuan, H.; et al. Chem. Soc. Rev. 2015, 44, 2664−2680. (3) Guo, Y.; Xu, K.; et al. Chem. Soc. Rev. 2015, 44, 637−646. (4) Yu, X.; Sivula, K. ACS Energy Lett. 2016, 1, 315−322. (5) Lee, J. H.; Jang, W. S.; et al. Langmuir 2014, 30, 9866−9873. (6) Shi, J.; Ma, D.; et al. ACS Nano 2014, 8, 10196−10204. (7) Knirsch, K. C.; Berner, N. C.; et al. ACS Nano 2015, 9, 6018− 6030. (8) Sun, Y.; Wang, R.; et al. Appl. Phys. Rev. 2017, 4, 011301. (9) Jaramillo, T. F.; Jørgensen, K. P.; et al. Science 2007, 317, 100− 102. (10) Seh, Z. W.; Kibsgaard, J. Science 2017, 355 (6321), eaad4998. (11) Yin, Y.; Han, J.; et al. J. Am. Chem. Soc. 2016, 138, 7965−7972. (12) Li, H.; Tsai, C.; et al. Nat. Mater. 2015, 15, 48−53. (13) Lei, Y.; Pakhira, S.; et al. ACS Nano 2017, 11, 5103−5112. (14) Cui, J.; Jiang, R.; et al. Small 2017, 13, 1602235. (15) Shi, Y.; Wang, J.; et al. J. Am. Chem. Soc. 2015, 137, 7365− 7370. (16) Kang, Y.; Gong, Y.; et al. Nanoscale 2015, 7, 4482−4488. I

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society (17) Lukowski, M. A.; Daniel, A. S.; et al. J. Am. Chem. Soc. 2013, 135, 10274−10277. (18) Liu, Q.; Fang, Q.; et al. Chem. Mater. 2017, 29, 4738−4744. (19) Voiry, D.; Salehi, M.; et al. Nano Lett. 2013, 13, 6222−6227. (20) Ambrosi, A.; Sofer, Z.; et al. Small 2015, 11, 605−612. (21) Maitra, U.; Gupta, U.; et al. Angew. Chem., Int. Ed. 2013, 52, 13057−13061. (22) Shi, Y.; Zhou, Y.; et al. J. Am. Chem. Soc. 2017, 139, 15479− 15485. (23) Chou, S. S.; De, M.; et al. J. Am. Chem. Soc. 2013, 135, 4584− 4587. (24) Voiry, D.; Goswami, A.; et al. Nat. Chem. 2015, 7, 45−49. (25) Tang, Q.; Jiang, D.-e. Chem. Mater. 2015, 27, 3743−3748. (26) Hansch, C.; Leo, A.; et al. Chem. Rev. 1991, 91, 165−195. (27) Calandra, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88, 245428. (28) Jiménez Sandoval, S.; Yang, D.; et al. Phys. Rev. B: Condens. Matter Mater. Phys. 1991, 44, 3955−3962. (29) Nayak, A. P.; Pandey, T.; et al. Nano Lett. 2015, 15, 346−353. (30) Lin-Vien, D.; Colthup, N. B., et al., The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press: San Diego, 1991. (31) Mévellec, V.; Roussel, S.; et al. Chem. Mater. 2007, 19, 6323− 6330. (32) Chia, X.; Ambrosi, A.; et al. Chem. - Eur. J. 2014, 20, 17426− 17432. (33) Ryzhikov, M. R.; Slepkov, V. A.; et al. J. Comput. Chem. 2015, 36, 2131−2134. (34) Lin, Y.-C.; Dumcenco, D. O.; et al. Nat. Nanotechnol. 2014, 9, 391−396. (35) Gao, G.; Jiao, Y.; et al. J. Phys. Chem. C 2015, 119, 13124− 13128.

J

DOI: 10.1021/jacs.7b11242 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX