Report

Identification of Active Edge Sites for Electrochemical H2 Evolution from MoS2 Nanocatalysts

See allHide authors and affiliations

Science  06 Jul 2007:
Vol. 317, Issue 5834, pp. 100-102
DOI: 10.1126/science.1141483

Abstract

The identification of the active sites in heterogeneous catalysis requires a combination of surface sensitive methods and reactivity studies. We determined the active site for hydrogen evolution, a reaction catalyzed by precious metals, on nanoparticulate molybdenum disulfide (MoS2) by atomically resolving the surface of this catalyst before measuring electrochemical activity in solution. By preparing MoS2 nanoparticles of different sizes, we systematically varied the distribution of surface sites on MoS2 nanoparticles on Au(111), which we quantified with scanning tunneling microscopy. Electrocatalytic activity measurements for hydrogen evolution correlate linearly with the number of edge sites on the MoS2 catalyst.

Progress in the field of heterogeneous catalysis is often hampered by the difficulty of identifying the active site on a catalyst surface (1, 2). In homogeneous catalysis, the active center is generally more clearly defined and quantified, with spectroscopic and mechanistic studies providing direct insight into reactive intermediates. Solid-state catalysts, however, commonly exhibit a variety of different surface sites that are difficult to identify and quantify; the scenario is further complicated when multiple sites work together in turning over a reaction. Identifying the most active site(s) is critical to designing and developing improved catalytic materials. Many useful in situ and ex situ experimental techniques, as well as computational methods, have been developed (35) to address this problem, but identifying the active site remains a challenging task.

In this study we used such methods to determine the active site of nanoparticulate MoS2 for the hydrogen evolution reaction (HER), 2H+ + 2e → H2 (6, 7), which is fundamentally important for a variety of electrochemical processes, fuel cells (as the reverse reaction), and solar H2 production (water splitting), particularly where there is a need to replace precious metal catalysts such as Pt (7, 8). In its bulk form, MoS2 is a poor HER catalyst (9). Nanoparticulate MoS2, however, is a more promising system; density functional theory (DFT) calculations indicate that the edges of MoS2 nanoparticles are active for hydrogen evolution (8), but no previous experiments have shown this conclusively.

Nanoparticulate MoS2 has been studied previously in an attempt to link activity to specific surface sites, in that MoS2 is used industrially as a hydrodesulfurization (HDS) catalyst (10, 11). Detailed insight has been gained from studies on simplified model systems in ultra-high vacuum (UHV) and by using computational methods (1215), as well as from combining reactivity measurements and ex situ characterization of industrial catalyst samples (10, 11, 16). Structural studies on the MoS2 catalyst have shown that it is composed almost entirely of flat polygons of S-Mo-S trilayers (10); depending on the synthesis conditions, these trilayers may stack in a graphite-like manner or remain as single trilayers. For single trilayers, two general kinds of surface sites exist—terrace sites, which are those on the basal plane, and edge sites, which lie at the edge of the nanoparticles. DFT studies suggest that the active site for HDS is on the edge of the MoS2 nanoparticles. This result is supported by adsorption studies of thiophene using scanning tunneling microscopy (STM) (17). Despite numerous studies on this material, there is a call for studies that uniquely link the well-defined structures of the model system to catalytic activity under standard reaction conditions (18).

To provide an experimental elucidation of the active site for the HER, we prepared MoS2 samples in UHV of deliberately chosen nanoparticulate morphologies such that the fractions of the terrace and edge sites were systematically varied, then characterized by STM. All of the MoS2 samples in this study were synthesized on a clean Au(111) substrate by physical vapor deposition of Mo in a background of H2S (19), followed by annealing, according to the approach in (13). Three samples were annealed at 400°C, two were annealed at 550°C, and a “blank” sample was synthesized without the deposition of Mo and annealed to 400°C. The Au(111) substrate serves to disperse the MoS2 nanoparticles by its herringbone reconstruction and is not particularly active for the HER (20). To maintain discretely separated single trilayer particles, we purposely synthesized the samples with low area coverages of MoS2, less than one-fourth ML (i.e., 0.25 nm2geometric/nm2MoS2).

Immediately after deposition, each sample was vacuum transferred to a second UHV chamber for STM imaging (Fig. 1). The crystallized, single-layered MoS2 nanoparticles can be described as flat polygons with a conducting edge state, seen as bright lines along the particle perimeter. Comparison of representative images of samples annealed at 400°C (Fig. 1A) and 550°C (Fig. 1B) shows how particle size increased after sintering at the higher temperature. The particles annealed to 400°C are consistent with similarly prepared MoS2 nanoparticles on Au(111) (13). Besenbacher et al. have shown that the dominant edge structure of MoS2 nanoparticles is that of a sulfided Mo edge and that this edge is particularly favored by larger-sized particles (12, 18). We also observe the predominance of the sulfided Mo-edge in our samples, regardless of annealing temperature. Thus, controlled sintering allows us to change the ratio of basal plane sites to edge sites without changing the nature of the edge. This sulfided (1 0–1 0) Mo-edge is the same structure predicted by DFT calculations to be the active site for H2 evolution (8).

Fig. 1.

A series of STM images of MoS2 nanoparticles on Au(111). The particles exhibit the typical polygon morphology with conducting edge states and are dispersed on the Au surface irrespective of coverage and annealing temperature (400°C or 550°C). (A) Low coverage (0.06 nm2MoS2/nm2geom.), annealed to 400°C (470 Å by 470 Å, 1.2 nA, 4 mV). (B) High coverage (0.23 nm2MoS2/nm2geom.), annealed to 550°C (470 Å by 470 Å, 1.2 nA, 1.9 V). (C) Atomically resolved MoS2 particle, from a sample annealed to 550°C, showing the predominance of the sulfided Mo-edge (19, 20) (60 Å by 60Å, 1.0 nA, 300 mV).

After imaging, we transfer the samples from UHV into an electrochemical cell to measure HER activity (21). Polarization curves (iE) within a cathodic potential window, and corresponding Tafel plots (log iE), are shown in Fig. 2. Current densities are normalized to the geometric area of the exposed face of all samples.

Fig. 2.

Polarization curves and Tafel plots in a cathodic potential window for the five different MoS2 samples as well as a blank sample. Samples annealed to 400°C are dark blue, samples annealed to 550°C light blue. (A) Polarization curve showing H2 evolution on all samples. (B) Tafel plot (log current versus potential). All of the MoS2 samples have Tafel slopes of 55 to 60 mV per decade irrespective of annealing temperature and coverage. Sweep rate: 5 mV/s.

The most inherent measure of activity for the HER is the exchange current density, i0 (6, 7, 22, 23), which is determined by fitting iE data to the Tafel equation (6), yielding Tafel slopes of 55 to 60 mV/decade and exchange current densities in the range of 1.3 × 10–7 to 3.1 × 10–7 A/cm2geometric for all MoS2 samples (table S1). In Fig. 3, we plot the exchange current density for each sample versus two sample parameters, the MoS2 area coverage (Fig. 3A), and the MoS2 edge state length (Fig. 3B). The data points fall on a straight line only when plotted versus edge length. Although the points show some scatter around this trend, they are described by a best-fit linear relation with a slope of 1.67 × 10–20 A/nmMoS2-edge.

Fig. 3.

Exchange current density versus (A) MoS2 area coverage and (B) MoS2 edge length. In both figures, open circles are samples annealed to 400 °C, filled circles are samples annealed to 550 °C. The exchange current density does not correlate with the area coverage of MoS2, whereas it shows a linear dependence on the MoS2 edge length. Exchange current densities are extracted from the Tafel plot in Fig. 2. The edge length was measured on all imaged particles and normalized by the imaged area.

Because the rate of reaction is directly proportional to the number of edge sites for all samples, regardless of particle size, we conclude that the edge site is indeed the active site (24). Bearing this in mind, we note in Fig. 3A that the exchange current densities of the samples sintered at 550°C are significantly lower than those prepared at 400°C, per MoS2 coverage, exactly as one would expect considering that the sintered samples have less edge length per area of MoS2.

We also compared nanoscale MoS2 to other materials that catalyze the HER on a per active site basis (1). For this direct site-to-site comparison, we used the 1.5 × 1015 sites/cm2 for the Pt(111) face as the basis for comparison as Pt is the archetypical HER catalyst (25). An exchange current density of 4.5 × 10–4 A/cm2 for this face (26) yields a turnover frequency (TOF) of 0.9 s–1 (table S2). In general, TOFs of transition metals range over 10 orders of magnitude (Hg, for instance, has a TOF as low as ∼10–9 s–1) (22). Given the slope in Fig. 3B, we have calculated the TOF of the MoS2 edge to be 0.02 s–1, indeed in the high range of TOFs for metals.

For further insight into the catalytic nature of the MoS2 edge, we have added our data for nanoparticulate MoS2 to a recent version of the volcano-type relations observed for HER catalysts (Fig. 4), in this case for the Gibbs free energy for atomic hydrogen adsorption (ΔGH) (22, 23). These volcano relations ultimately reflect the Sabatier principle, which accounts for optimal surfaces as ones that exhibit moderate binding energies of reaction intermediates, hydrogen adsorption in the case of the HER. In Fig. 4, the exchange current density is shown as a function of the DFT-calculated free energy of adsorption of hydrogen, which was recently determined to be +0.08 eV for the MoS2 edge (8). To add MoS2 to this figure, we converted the TOF of nanoparticulate MoS2 to its exchange current density per 1.5 × 1015 sites/cm2, which yields 7.9 × 10–6 A/cm2 (table S2). This value surpasses those of the common metals and lies just below those of the precious Pt-group metals. When plotting this experimentally determined activity of the edge site versus its DFT-calculated ΔGH (8), we see that it follows the volcano trend (23). This agreement validates the predictive capability of this DFT model as well as its applicability beyond metal catalysts.

Fig. 4.

Volcano plot of the exchange current density as a function of the DFT-calculated Gibbs free energy of adsorbed atomic hydrogen for nanoparticulate MoS2 and the pure metals (23). As seen, MoS2 follows the same trend as the pure metals. The MoS2 exchange current density is normalized to the atomic site density of Pt for comparison. Samples are polycrystalline unless otherwise noted.

After identifying the active site and comparing it with typical metal catalysts, we may consider how to improve its activity. The DFT-calculated ΔGH of the MoS2 edge site is slightly positive at +0.08 eV, with calculations suggesting an H coverage of only one-quarter on the edge under operating conditions (8). Thus, only 1 in 4 edge atoms evolves molecular H2 at a given time, unlike Pt(111) which operates at a H-coverage of ∼1 ML (7, 26, 27). If all MoS2 edge sites could be made to adsorb H, activity could be increased by a factor of 4. This might be accomplished by appropriately tuning the electronic structure of the edge to increase the bond strength of the adsorbed H (23). Such a modification could simultaneously improve the inherent turnover of each edge site, further improving the overall activity of the material toward that of Pt-group metals.

Supporting Online Material

www.sciencemag.org/cgi/content/full/317/5834/100/DC1

Materials and Methods

SOM Text

Fig. S1

Tables S1 and S2

References

References and Notes

View Abstract

Navigate This Article