Advanced Hub Main Navigation Menu
Operando Insights on the Degradation Mechanisms of Rhenium-Doped and Undoped Molybdenum Disulfide Nanocatalysts During Hydrogen Evolution Reaction and Open-Circuit Conditions
Abstract
Molybdenum disulfide (MoS2) nanostructures are promising catalysts for proton-exchange-membrane (PEM) electrolyzers to replace expensive noble metals. Their large-scale application demands high activity for the hydrogen evolution reaction (HER) as well as robust durability. Doping is commonly applied to enhance the HER activity of MoS2-based nanocatalysts, but the effect of dopants on the electrochemical and structural stability is yet to be discussed. Herein, operando electrochemical measurements to the structural evolution of the materials down to the nanometric scale are correlated by identical location electron microscopy and spectroscopy. The range of stable operation for MoS2 nanocatalysts with and without rhenium doping is experimentally defined. The responsible degradation mechanisms at first electrolyte contact, open circuit stabilization, and HER conditions are experimentally identified and confirmed with the calculated Pourbaix diagram of Re-doped MoS2. Doping MoS2-based nanocatalysts is validated as a promising strategy for continuing the improvement of high-performance and durable PEM electrolyzers.
1 Introduction
MoS2 is widely regarded as a Pt alternative for the hydrogen evolution reaction (HER) in acidic media due to its high catalytic activity, abundance, and low price.[1-3] However, it crystallizes in the thermodynamically stable 2H-MoS2 polytype, which only possesses high HER activity on the edge sites.[4] In order to enhance the HER performance of MoS2, exposure of the edge active sites can be maximized in nanostructures,[5] e.g., with a nanoflower morphology.[6, 7] Other strategies to optimize the HER activity include sulfur-defect engineering,[8, 9] phase engineering,[10, 11] the use of amorphous MoS2 layers,[12] or doping with transition metals.[13, 14]
In particular, Re-doped MoS2 nanomaterials (RexMo1-xS2) have shown remarkable improvement in the HER activity over MoS2 nanocatalysts.[15-18] The n-type Re dopant not only acts as an electron donor but also favors local transformation to 1T-MoS2, while both effects contribute to improved HER performance.[19-22] Although the 1T-MoS2 polytype provides enhanced conductivity and basal-plane HER active sites,[11, 23] it is a metastable phase which can revert to the 2H phase. However, Re-doping stabilizes the 1T phase and 1T-RexMo1-xS2 has shown long lifetimes on the shelf and stability under HER conditions.[15, 17] In comparison, alkali metals can also stabilize 1T-MoS2 nanocatalysts, but they are known to suffer from poor durability.[22, 24]
Indeed, beyond the HER activity, the appeal of MoS2-based nanocatalysts relies on the high stability during operation. However, such claim is usually only validated by electrochemical measurements with limited time (a few minutes to several hours) and without correlative characterization on the catalysts and electrolytes.[25] Only recently, through operando characterization using a scanning flow cell (SFC) connected to inductively coupled plasma mass spectrometry (ICPMS) has the dissolution of bulk 2H-MoS2[26] and [Mo3S13]2− [27, 28] into acidic electrolytes been studied during HER conditions. Nevertheless, the stable (dissolution-free) operation conditions for acidic HER have yet to be defined, although a wide stability window between 0 and −0.4 VRHE has been proposed from simulations.[29]
In addition, ICPMS measurements revealed high dissolution rates when the catalysts were subjected to open-circuit potential (OCP) conditions: a well-known problem for alkaline HER catalysts which was only recently reported for acidic HER catalysts.[28-32] However, so far, the dissolution during OCP was only investigated for the initial minutes,[26, 29] and its effects on the HER performance and the structure of the nanocatalysts remain unknown.
Furthermore, despite the great importance of dopants in improving the HER activity of MoS2-based nanocatalysts, their effect on the electrochemical and structural stability remains to be quantified.
Herein, we present a systematic study on the stability of MoS2 nanocatalysts for both OCP and HER conditions, with and without Rhenium dopants. We employed a combination of ex situ electrochemical degradation and operando characterization performed by SFC-ICPMS, which were correlated to the morphology and chemical composition evolution revealed by microscopy imaging and spectroscopy acquired in identical locations (IL), X-ray photoelectron spectroscopy (XPS), X-ray absorption spectroscopy (XAS) as well as calculated Pourbaix diagrams. The correlative studies unveiled different degradation mechanisms during the first electrolyte contact, OCP stabilization, and subsequent HER operation. Moreover, the operando SFC-ICPMS results demonstrated a stability window between −0.05 to −0.3 VRHE for HER operation in sulfuric acid. Furthermore, we quantitatively demonstrate that Re-doped MoS2 nanocatalysts have a higher stability against dissolution at HER conditions, as well as a broader electrochemical stability window against Re dissolution at anodic potentials, making them more durable for use in acidic electrolyzers than pure MoS2 nanocatalysts. The developed methodology and mechanistic insights will motivate the exploration of further metal dopants to simultaneously enhance the activity and stability of HER catalysts for broad-scale commercial applications.
2 Results
2.1 Effect of Rhenium in the Activity and Morphology of RexMo1-xS2 Nanocatalysts
Molybdenum disulfide nanocatalysts were grown on carbon paper (CP) substrates through a hydrothermal route[33] with various amounts of Re doping. The hydrothermal synthesis is a common method for preparation of MoS2-based nanocatalysts, which motivates a thorough stability study on these materials. CP substrate consists of interwoven carbon fibers with a diameter of ≈10 µm. The stoichiometries of RexMo1-xS2 (x = 0, 0.2, 0.3, and 0.4) were determined by ICPMS. The HER activity of the electrodes was comparatively tested by cyclic voltammetry (CV) (Figure S1, Supporting Information) in 0.5 m H2SO4 under continuous Ar purging from 0 to −0.25 VRHE before ohmic drop correction (iR correction). The results show that Re-containing materials have similar HER activity, with similar overpotential values at the same current density normalized to the metal (Mo and Re) content, which is significantly lower than the overpotential of bare MoS2/CP electrode (Figure 1a). Considering the limited activity improvement from x = 0.2 to 0.4 and the scarcity of Re, the Re0.2Mo0.8S2 composition was chosen for the stability study alongside MoS2.
Figure 1b and Table S1 (Supporting Information) present a comparison of the HER activity of the nanocatalysts in this work with literature reports of MoS2 and RexMo1-xS2 nanocatalysts. It is evident that our electrocatalysts reach state-of-the-art performance both in terms of the overpotential at −10 mA cm−2 (η10) and the Tafel slope. Specifically, the MoS2/CP and Re0.2Mo0.8S2/CP Tafel slopes were 44 and 55 mV dec−1, respectively, while their η10 values were 203 and 136 mVRHE (Figure 2a,b; Table S2, Supporting Information). Although these figures evidence an improved η10 for the nanocatalysts doped with Re, its Tafel slope is higher. The optimal rhenium composition to minimize the Tafel slope when compared to MoS2 varied across reports,[34, 35] which indicates additional factors contributing to the activity of RexMo1-xS2 nanocatalysts, e.g. the extend of additional 1T sites, sulfur vacancies or other defects.
The morphology of the MoS2/CP and RexMo1-xS2/CP electrodes was compared with scanning electron microscopy (SEM) micrographs (Figure 1c,d; Figure S2, Supporting Information). Both electrodes showed a layer of RexMo1-xS2 nanosheets covering the surface of the CP substrate. When synthesized as powders (in absence of the substrate), RexMo1-xS2 materials also exhibit a nanoflower morphology, as revealed by STEM (Figure 1e,g) and transmission electron microscopy (TEM) (Figure S3, Supporting Information) micrographs. In addition, the packing density of the nanoflowers increases with increasing Re content. High-resolution high-angle annular dark field (HAADF)-STEM micrographs revealed the distribution of high-contrast Re within the Re0.2Mo0.8S2 lattice (Figure S4, Supporting Information). XPS indicates that Re is prevalent in the Re4+ oxidation state showing binding energies that are typical for the disulfide form (Figure S1, Supporting Information). Furthermore, as a result of Re incorporation, the phase transformation from the 2H to the 1T polymorphic structure was also observed in a few areas in Re0.2Mo0.8S2 (Figure 1h; Figure S6, Supporting Information), whereas other areas remained 2H (see Figure 1f; Note 1, Figures S6 and S7, Supporting Information). The crystal structure of MoS2 was consistently found to be only 2H (Figures S7 and S8, Supporting Information).
2.2 Electrochemical and Structural Stability of RexMo1-xS2 Nanomaterials
In order to make an accurate assessment of the stability of RexMo1-xS2 nanocatalysts, the contributions of different mechanisms of corrosion and dissolution need to be decoupled. Since recent reports[26, 29] suggest that OCP and electrolyte contact can severely degrade MoS2-based nanocatalysts, the stability of the RexMo1-xS2 nanomaterials has been sequentially studied during early contact with the electrolyte, prolonged immersion at OCP conditions, and operation at HER potentials.
2.2.1 Stability at Early Electrolyte Contact (Immersion)
The dissolution and its kinetic evolution occurring to the fresh electrodes after synthesis was evaluated by immersing them in 5 mL of electrolyte for 27 h, exchanging it every 9 h, and ICPMS analyses of Mo and Re dissolved in such electrolytes during 0–9, 9–18, and 18–27 h are shown in Figure S9a (Supporting Information).
The results reveal that both samples strongly dissolved during the first 9 h of contact with the electrolyte, much more than during the subsequent 9 h periods. Thus, we separately consider the dissolution during the initial 9 h as the effect of early electrolyte contact, while the subsequent hours would be attributed to other mechanisms derived from further stabilization at OCP. To narrow down the time span of the early electrolyte contact with significant dissolution, further sets of measurements were conducted on the dissolution after every hour for the initial 6 h (Figure S10, Supporting Information). The results show that for both MoS2 and Re0.2Mo0.8S2 nanocatalysts, the majority (≈90% of the total) dissolution had happened during the first hour.
To correlate the dissolution results to the structural evolution and find the responsible mechanisms, the electrodes were investigated by means of IL-SEM (Figure 2 a–d) before and after the electrolyte immersion for 27 h. For both RexMo1-xS2/CP electrodes, small differences were spotted in some locations in the form of detached catalyst particles from the substrate (see arrows in the figures). The loss of catalysts can be correlated to the dissolution measured by ex situ ICPMS, which predominantly occurred during the first hour (Figure S9a; Figure S10, Supporting Information). On the other hand, the thickness of the remaining layer remained stable, from an average of 556 ± 20 nm before immersion to 553 ± 13 nm after immersion for MoS2/CP and from 451 ± 20 to 452 ±10 nm for Re0.2Mo0.8S2/CP (Figures S11–S13, Supporting Information).
Additionally, XPS was conducted on the electrodes to understand the changes in the surface composition of the electrodes. Figure 2e–h shows the deconvoluted Mo 3d spectra for both samples before and after immersion for 27 h. The results reveal the presence of the oxide species MoO2 and MoO3 on both electrodes as-synthesized before immersion (see Figure S14, Supporting Information for more details on the peak assignment). These oxides arise from the incomplete reaction of the molybdenum precursor reported to happen at the synthesis temperature of 200 °C, which was also correlated to enhanced HER performance.[39, 40] The presence of Mo6+ oxides is strongly reduced after immersion for both electrodes (Figures S5,S14, and S15, Supporting Information), indicating that the strong Mo dissolution observed during the initial hours of OCP (Figures S9a and S10, Supporting Information) is indeed related to partially unreacted reagents from the synthesis and/or storage of the material. This mechanism could also explain the over-stoichiometric dissolution of Mo from Re0.2Mo0.8S2 compared to Re during the initial electrolyte contact. According to XPS, Re4+ species (same oxidation state as ReS2) are predominant on the Re0.2Mo0.8S2 surface, which was maintained very stable after immersion (Figures S9b and S5, Supporting Information). To further confirm these results, XAS was conducted on both the MoS2/CP and Re0.2Mo0.8S2/CP electrodes before and after 27 h immersion in 0.5 m H2SO4. The Mo K-edges (Figure 2i,j) of both electrodes show a shift toward lower energies after immersion, suggesting reduction of Mo species, which corroborates with the dissolution of MoO3 species observed by XPS. This is also consistent with the findings of a previous in-situ XAS report for MoSx.[41] Moreover, the Mo-K edge of Re0.2Mo0.8S2/CP electrode onsets at a lower energy than that of the MoS2/CP electrode (Figure S16a, Supporting Information), which can be attributed to the coupling effect of the n-type Re-dopant on the electronic structure of the Mo atoms.[42] Finally, the Re L3-edge XANES (Figure S16b, Supporting Information) of the Re0.2Mo0.8S2/CP electrode does not show an obvious chemical shift after 27 h of immersion, again corroborating the XPS observation.
2.2.2 Stability at OCP Conditions
As observed in Figures S9a,S10 (Supporting Information), after the initial contact (0–9 h or first hour) with the electrolyte, the dissolution of the RexMo1-xS2 /CP electrodes approaches a much lower rate, which needs to be explained by other degradation mechanisms at stabilized OCP conditions. To understand the long-term effect of the OCP dissolution on the HER performance of the materials, CV measurements were performed with the RexMo1-xS2 /CP electrodes after 13 and 30 days of OCP electrolyte immersion in ≈ 5 mL of volume (Figure S17, Supporting Information). Given the small volume of electrolyte, minimal changes of 2 mV of overpotential in both materials after the 30-day treatment were observed, which is within the experimental error. These results point to the build-up of dissolved species in solution preventing further degradation of the electrodes,[43, 44] and highlight the importance of the operando measurements in the flow cell where the electrolyte is continuously refreshed to properly understand the intrinsic stability of the electrocatalysts.
Thus, to gain insights into such OCP degradation, operando electrochemical and dissolution measurements were performed with SFC-ICPMS.[45] Figure 3 shows the time-resolved dissolution of Mo and Re in MoS2 and Re0.2Mo0.8S2 as a function of the potential applied in steps of 0.1 V from the HER regime −0.1 to 0.5 VRHE, which is close to the OCP of both catalysts, as shown in Figure S18a (Supporting Information). The dissolution observed during HER potential (−0.1 VRHE) was negligible for all catalysts. When transitioning to more anodic potentials toward the OCP (0.5 VRHE), Mo atoms start to dissolve, which aligns with the ex situ ICPMS dissolution trends. Previous ex situ experimental reports suggested dissolution at 0.23 VRHE29, while our operando data provides evidence for Mo dissolution starting at potentials as low as 0 VRHE. In comparison, Re dissolution starts at more anodic potentials, becoming noticeable first at 0.3 VRHE and more pronounced toward the OCP, 0.5 VRHE.
More severe dissolution at OCP conditions with respect to HER was further confirmed by operando measurements under other electrochemical protocols, including CV and chronopotentiometry measurements in the galvanostatic mode (Figure S19, Supporting Information). Longer measurements maintaining OCP for ≈1 h revealed that Mo dissolution starts to decay after an initial increase, while the Re dissolution slightly increased with time (Figure S20, Supporting Information). Such result is in agreement with the previous ex situ ICPMS measurements of the electrolyte after every 9 hours at OCP. This further indicates that with more time of stabilization at OCP, Mo and Re dissolution would approach to the stoichiometric ratio (Figure S9b, Supporting Information).
To investigate the long-term effects of OCP on the chemical composition and structure of the RexMo1-xS2 nanocatalysts, IL-STEM measurements were conducted before and after 13 days of immersion in 0.5 m H2SO4 (Figure 4). The electrochemical measurements are performed directly on Au/C TEM finder grids that are numerated to allow the analysis of the same area of the specimen before and after electrochemical treatment, which provides information on the mechanisms of morphological degradation of the catalysts.[46-48] The OCP of Re0.2Mo0.8S2 nanocatalysts dispersed on the Au/C TEM finder grid in the identical location set-up (Figure S21, Supporting Information) was determined to be ≈0.6 VRHE, which is comparable to the OCP of RexMo1-xS2/CP electrodes (Figure S18b, Supporting Information).
As shown in Figure 4a–d, there are no apparent changes in the morphology of MoS2 and Re0.2Mo0.8S2 nanocatalysts after 13 days of OCP beyond the small movement of the nanoflower layers (Figures S22 and S23, Supporting Information). By tracking identical locations of the nanocatalysts, systematic changes in the chemical composition were detected by STEM-energy dispersive X-ray spectroscopy (EDS). The Re content in Re0.2Mo0.8S2 increased from an average of 21.3 at% (normalized to Re+Mo) to 23.4 at% after 13 days of OCP (Table S3 and Figure S24, Supporting Information). This is consistent with the initially higher amount of Mo dissolution at OCP compared to Re dissolution (Figure S9, Supporting Information).
EDS measurements were also conducted at identical locations before and after 13 days of immersion to monitor the oxygen content of the nanocatalysts (Table S3, Supporting Information). The results show an increase in the O content from 10.1 at% (normalized to O + S) to 12.6 at% for MoS2 (Figure 4e; Figure S25a,b, Supporting Information) and 6.6 at% to 7.9 at% for Re0.2Mo0.8S2 (Figure 4g; Figure S25c,d, Supporting Information). Oxidation of both samples after long-term immersion is further verified by conducting electron energy loss spectroscopy (EELS) at identical locations, as shown in Figure 4f,h. For both catalysts, the intensity of the O-K edge increased after 13 days of immersion. Moreover, the oxidation state of Mo also increased after immersion. The intensity ratio of the Mo M2,3 white lines (IM3/IM2) is lower in more oxidized states, as was demonstrated for MoO2 and MoO3 standards.[49] The IM3/IM2 decreased from 1.76 to 1.45 for MoS2 and from 2.29 to 1.97 for Re0.2Mo0.8S2, confirming an increase in the oxidation state of Mo in both nanocatalysts after 13 days immersion. Although MoS2 and MoO2 both have an oxidation state of Mo4+, the higher electronegativity of oxygen compared to sulfur could also account for the decreased IM3/IM2 ratio.
2.2.3 Stability at HER Conditions
To evaluate the stability of electrocatalysts, typical tests include CV for thousands of cycles or chronoamperometry measurements for tens to hundreds of hours.[50, 51] In this study, we opted for a high-cycle CV instead of long-term chronoamperometry measurements in order to alleviate the disturbance derived from bubbling from H2 evolution.[52] To study the electrochemical stability of MoS2/CP and Re0.2Mo0.8S2/CP through CV, the electrodes were first immersed in 0.5 m H2SO4 for 32 h prior to HER conditions to completely remove the oxides from synthesis/storage, which would cause severe dissolution (see Section 2.2.1). After the treatment, 4000 CV cycles were subsequently performed in a fresh batch of electrolyte (0.5 m H2SO4) free of the dissolution products from the previous immersion step. Most CV cycles were conducted at a scan rate of 100 mV s−1, and slower cycles at a scan rate of 1 mV s−1 were performed every 1000 CV cycles. For each cycle, the potential was scanned between 0 and −0.25 VRHE (before iR correction), reaching a current density of −16 mA cm−2 for MoS2/CP and −35 mA cm−2 for Re0.2Mo0.8S2/CP. The electrodes showed remarkable HER performance stability during the CV cycles, with negligible changes in η10 (2 and 0 mV for MoS2/CP and Re0.2Mo0.8S2/CP, respectively), which are within the error of the analysis (Figure 5a; Figure S26 and Table S2, Supporting Information). Moreover, the changes in Tafel slopes are also minor (6 mV and 3 mV dec−1 for MoS2/CP and Re0.2Mo0.8S2/CP, respectively). The HER stability demonstrated by the Re0.2Mo0.8S2/CP electrodes is in agreement with previous reports: For example, η10 of Re0.55Mo0.45S2 nanocatalysts only increased by 2 mV after 3000 CV cycles.[15]
Additionally, the dissolution of the electrodes during these 4000 CV cycles was quantified by ex situ ICPMS (Figure S9a, Supporting Information). Compared to the integrated dissolution during the ≈8 h (4000 CV cycles) at HER conditions, the dissolution at OCP conditions is higher, which is consistent with the operando SFC-ICPMS measurements (Section 2.2.2). However, it is important to note that such ex situ measurements possess limitations that hinder a fully quantitative analysis. First, although fresh electrolytes free of dissolved Mo and Re were exchanged prior to the CV cycles at HER conditions, ex situ experimental protocols cannot avoid some time (at least of the order of minutes) when the electrodes are immersed in conditions closer to OCP. Secondly, while the previous immersion step could remove most oxides formed during the synthesis or storage, it would also introduce new oxides when approaching OCP (Section 2.2.2). These two issues can cause overestimation of the dissolution related to HER due to the degradation of oxides formed on MoS2. Thirdly, the possible buildup of dissolved species in the stagnant electrolyte and the presence of a Nafion membrane in the H-cell can cause underestimation of the dissolution.[43, 44] As a result of these three considerations, the integrated dissolution during the CV cycles would lead to an inaccurate HER stability assessment.
To overcome the limit of ex situ measurements, the stability of MoS2 and Re0.2Mo0.8S2 nanocatalysts under different HER potentials was investigated using the operando SFC-ICPMS set-up. MoS2 and Re0.2Mo0.8S2 nanocatalysts were drop casted onto FTO substrates and rinsed with 0.1 m H2SO4 prior to SFC-ICPMS measurement. The rinsing step significantly reduced the dissolution derived from the initial flowing electrolyte (0.1 m H2SO4) contact through the SFC. Once the dissolution signal decreased below the background level, the electrodes were subjected to HER conditions by chronoamperometry measurements at −0.05, −0.1, −0.2, −0.3, and −0.4 VRHE. As shown in Figure 5b, there is negligible dissolution from MoS2 and Re0.2Mo0.8S2 between −0.05 and −0.3 VRHE. Starting from −0.4 VRHE, dissolution of Mo was observed from both MoS2 and Re0.2Mo0.8S2, while Re remained stable against dissolution. Thus, these operando measurements confirm the HER stability of both MoS2 and Re0.2Mo0.8S2, as long as the applied potential does not exceed −0.3 VRHE, which is already a high overpotential for HER operation.
To specifically compare the stability of MoS2 and Re0.2Mo0.8S2 electrodes, the universal stability metric Stability number (S-number)[53] was used, which is a dimensionless figure defined as the ratio between the molar quantity of reaction products (H2 for HER)[29] and the molar quantity of dissolved ions from the electrode (See Note 2, Supporting Information). At −0.3 VRHE, within the stable range of HER operation, the S-number is evaluated to be at least 106 for both MoS2 and Re0.2Mo0.8S2. This number is comparable to noble IrO2 catalysts for acidic oxygen evolution reaction (104–107),[53] and way above photocatalysts such as BiVO4 (102).[54] At −0.4 VRHE, where the Mo dissolution is clearly above background, the S-number of Re0.2Mo0.8S2 (4.8·105) shows a higher stability than bare MoS2 (3.4 ·105). Notably, Re dissolution was not observed even at −0.4 VRHE, indicating a higher stability of Re than Mo in the HER regime. These quantitative results confirm that in addition to providing higher HER activity, Re-doping is a valid strategy to improve the electrochemical stability of MoS2 nanocatalysts.
Finally, the electrochemical results were correlated with the morphological evolution of MoS2/CP and Re0.2Mo0.8S2/CP electrodes before and after 4000 CV cycles by means of IL-SEM. No visible changes were observed by comparing the SEM micrographs acquired at the same CP fibers, as the coverage and thickness of the nanocatalyst layer remained constant (Figures S27 and S11, Supporting Information). The average thickness of such layer changed from 553 ± 13 to 555 ± 18 nm after HER for MoS2/CP, and 452 ± 10 to 452 ± 10 nm for Re0.2Mo0.8S2/CP, confirming that negligible dissolution took place. In addition, our recent report of IL-STEM on Re0.2Mo0.8S2 nanocatalysts confirmed that the morphology and the composition were maintained down to the nanometric scale after 4000 CV cycles in the HER regime.[48]
3 Discussion: Mechanisms of Dissolution
Considering the results from the electrochemical and structural stability studies, the following mechanisms of degradation of RexMo1-xS2 nanocatalysts can be established. We also separate the discussion into three stages, consisting of the initial contact with the electrolyte, stabilization at OCP, and the HER conditions.
The initial electrolyte contact causes most of the dissolution, which is especially severe in the case of Mo. This can be attributed to the dissolution and physical detachment of MoO3 and MoO2 stemming from the synthesis and storage (Figure 2; Figures S9 and S14, Supporting Information). After most of the oxides are dissolved at the initial contact with the electrolyte, the OCP stabilizes at≈0.5 VRHE, leading to a steady rate of Mo and Re dissolution.
To understand the origins of such OCP degradation, we conducted Pourbaix analysis to calculate the diagram of Re0.2Mo0.8S2 (Figure S28, Supporting Information). Figure 6 presents the calculated aqueous decomposition free energy (ΔGpbx) for MoS2 and Re0.2Mo0.8S2 as a function of the applied potentials. A larger ΔGpbx indicates greater instability of the given species. These calculations show that the sulfide species remain stable at anodic potentials until MoS2 transforms into MoO2 at≈0.35 VRHE, while ReS2 is maintained stable up until≈0.4 VRHE. These findings also correspond with our experimental results obtained through operando SFC-ICPMS, where the onset of Re dissolution is delayed relative to that of Mo when approaching OCP potentials (Figure 3). Furthermore, the dissolution rate of both elements reaches a stoichiometric ratio once the OCP is stabilized, as experimentally shown in Figures S9 and S20 (Supporting Information). In addition, such calculations are also consistent with the increase in oxidation state observed at identical locations (Figure 4).
When subjecting the RexMo1-xS2 electrodes to cathodic HER conditions, Pourbaix analysis predicts a wide stability range for Mo at pH 1 from≈+0.35 until≈−0.45 VRHE, where a transformation from MoS2 to Mo is predicted to occur (Figure 6). This is further confirmed experimentally by SFC-ICPMS, although the stability range is narrowed down to a range of −0.3 to 0 VRHE (Figure 5b).
In the case of Re, no experimental dissolution was observed operando at HER potentials as low as −0.4 VRHE (Figure 5b), despite the Pourbaix diagram predicting the transformation from ReS2 to Re0 at≈−0.2 VRHE. Given that no dissolution was detected with the high-sensitivity SFC-ICPMS technique, we hypothesize that a higher overpotential would be required to drive such reaction, kinetically stabilizing the RexMo1-xS2 material. Note that Re+ and Re− species were not considered for the RexMo1-xS2 Pourbaix diagram calculations based on experimental considerations, see Note 3 and Figure S29 (Supporting Information) for further details.
Considering the operando results, we note that the degradation observed at HER conditions when performing ex situ analysis of the electrolyte after 4000 CVs from 0 to –0.25 VRHE (prior to iR correction) (Figure 5a; Figure S9a, Supporting Information), can be attributed to the degradation of the MoOx produced during OCP and/or remaining from synthesis instead of the inherent instability of the MoS2. Specifically, according to the Mo Pourbaix diagram, MoO2 dissolves as Mo3+ at pH 1 at potentials negative than 0.1 VRHE.[55] This highlights the need for pretreatment of the electrodes to eliminate the present oxides that can lead to misleading stability assessments. Based on such mechanisms, it also becomes evident that start-stop cycles intermixing OCP conditions (“stop”, sulfide species transform into oxides) and HER operation (“start”, oxides dissolve) can decrease the lifetime of MoS2-based nanocatalysts for electrolyzers and should consequently be avoided. With these precautions, MoS2 and Re0.2Mo0.8S2 can be operated at high stability for hydrogen production in the wide range of potentials from −0.05 to −0.3 VRHE, with the Re-doped material surpassing the MoS2 in both activity and stability.
4 Conclusion
The electrochemical and structural stability of MoS2 and RexMo1-xS2 nanocatalysts was investigated through a combination of operando electrochemical measurements, correlative spectroscopy techniques, and electron microscopy analyses at identical locations.
Based on the electrochemical stability and the structural evolution tracked by IL-SEM, IL-STEM, XPS, and XAS, we propose different degradation mechanisms of RexMo1-xS2 nanocatalysts at first electrolyte contact, OCP stabilization, and HER conditions. Most of the Mo and Re dissolution was measured upon first contact with the electrolyte, derived from a heavy loss of molybdenum oxide species (MoO2, MoO3) introduced from synthesis and storage. OCP stabilization led to the surface oxidation of RexMo1-xS2, resulting in their dissolution. Finally, HER conditions led to the dissolution of oxides formed during OCP stabilization, whereas the dissolution of sulfides remained negligible until −0.4 VRHE when Mo started dissolving.
By eliminating most of the surface oxides from RexMo1-xS2, and using operando SFC-ICPMS, we could demonstrate a stable operation range for HER experimentally (stability number over 106), between −0.05 and −0.3 VRHE. To inhibit the dissolution of Mo, it is advisable to operate electrolyzers within the stability window (not exceeding −0.3 VRHE), as well as to avoid going toward anodic potentials (e.g., toward OCP). Moreover, we confirmed that Re-doped MoS2 nanocatalysts not only had higher HER activity than their undoped counterparts, but they also showed higher stability during HER (outside the stability range), and higher onset of dissolution at anodic potentials toward open-circuit conditions, which are favorable for stable operation in an electrolyzer.
Our results give a comprehensive view on the degradation mechanisms of RexMo1-xS2 and explain the superior HER activity and stability of Re-doped MoS2 catalysts compared to undoped MoS2, opening the door to the rational design of doped-MoS2 for optimized stability of PEM electrolyzers.
5 Experimental Section
Materials
The synthesis of RexMo1-xS2 nanoflowers was conducted using the reagents ammonium heptamolybdate tetrahydrate ((NH₄)₆Mo₇O₂₄·4H2O, 99.98%, Sigma-Aldrich), ammonium perrhenate (NH4ReO4, >99%, Merck), thiourea (SC(NH₂)₂, >99.0%, Sigma-Aldrich) and ethanol (EtOH, >99.8%, Carl Roth). Hydrophilic carbon paper (CP, HCP030N, Hesen) was employed as electrode substrate for the growth of RexMo1-xS2 nanocatalysts. The electrolyte of choice for all electrochemical measurements was diluted sulfuric acid (H2SO4, Suprapur, Sigma-Aldrich, and MilliQ water).
Synthesis of RexMo1-xS2 Nanocatalysts and RexMo1-xS2 /CP Electrodes
In a typical synthesis, 35.3 mg of (NH₄)₆Mo₇O₂₄·4H2O and 80.7 mg of SC(NH₂)₂ were dissolved in 10 mL of distilled water, adding 13.4, 45.3, or 134 mg of NH4ReO4 for Re0.2Mo0.8S2, Re0.3Mo0.7S2, and Re0.4Mo0.6S2 materials, respectively. The clear solution was transferred to a Teflon liner including a 1 cm × 2 cm CP with an area of 1 cm × 1 cm covered with Teflon tape, thus limiting the RexMo1-xS2 deposition to the 1 cm2 remaining exposed. The solution with the substrate was heated at 200 °C for 20 h in an autoclave and subsequently let to cool down at room temperature before removing the Teflon tape and rinsing the RexMo1-xS2/CP electrodes with water. Independently of Re content, the resulting electrodes showed a characteristic black color in the area with deposited material. If no substrate was introduced inside the Teflon liner and the concentration of all reagents was increased 2.7 × 103 times, black suspensions of RexMo1-xS2 nanocatalysts were obtained as a powder to analyze in ILSTEM. These products were washed by centrifuge-redispersion cycles in water and ethanol before evaporating the ethanol solvent at 110 °C for 4 h. Black RexMo1-xS2 powders were subsequently ground in an agate mortar. The MoS2, Re0.2Mo0.8S2, Re0.3Mo0.7S2, and Re0.4Mo0.6S2 stoichiometries were determined by ICP-MS.
Raman Spectroscopy, XPS, and XAS Characterization
Raman spectroscopic investigations were conducted with a WITec Raman system, operated with a green solid-state laser (λ = 532 nm), a ×50 objective lens, an 1800 l mm−1 diffraction grating, and a Peltier-cooled charge-coupled device (CCD) detector. The Raman spectra were collected for 2 s with 10 accumulations each, with an incident laser power of ca. 0.02 mW on the sample surface to avoid any potential laser-induced oxidation.[56]
The electrodes were studied ex situ before and after 27 h of electrolyte contact at OCP by XPS using monochromated Al Kα 1486.6 eV radiation. Spectra were recorded with a pass energy of 55 eV and a step size of 0.2 eV. Binding energies were calibrated by positioning the main graphite C 1s peak at 284.5 eV.
All the presented XANES spectra at Mo K- and Re L-edges were collected in fluorescence mode using the beamline 10C of the Pohang Accelerator Laboratory in Korea. Reference spectra were simultaneously measured using Mo and Re metal foils for energy calibration.
Electron Microscopy Characterization
SEM measurements were performed on a Hitachi S-4800 at 15 kV and a Zeiss Gemini at 1.5 kV. HRTEM experiments were conducted at 300 kV on a Titan Themis (Thermo Fisher) microscope equipped with an aberration corrector for the objective lenses. STEM measurements were done at 300 kV on a Titan Themis microscope (Thermo Fisher) equipped with an aberration corrector for the condenser lenses. EELS and EDS measurements were recorded using a Quantum ERS spectrometer (Gatan) and a SuperX detector (Thermo Fisher), respectively.
Electrochemical Characterization
The RexMo1-xS2/CP electrodes were electrochemically characterized using a Metrohm Autolab PGSTAT204 potentiostat with a reversible hydrogen electrode (RHE, Hydroflex) as the reference and glassy carbon as the counter electrode. The electrodes were stabilized for 32 h at OCP and the electrolyte was changed before starting HER. Such pretreatment was introduced to decouple the contributions of first electrolyte contact. If such contributions are not decoupled, a decay in the HER response after 4000 CV could be mistakenly attributed to HER corrosion, even if it is instead caused by the first electrolyte contact dissolution. HER stability was evaluated by 4000 cycles of CV at a scan rate of 100 mV s−1 in a 0.5 M H2SO4 electrolyte saturated with Ar gas through continuous purging. Three cycles were recorded at a scan rate of 1 mV s−1 every 1000 cycles. Electrochemical impedance spectroscopy was also acquired at –100 mVRHE from 105 to 10 Hz to apply 100% IR correction to all CV scans during data analysis. Chronopotentiometry measurements were conducted at 0 A for 20 h to evaluate the OCP of the electrodes using glassy carbon as counter electrode and RHE as reference electrode. To understand the effect of OCP on the HER activity, pretreated electrodes with 32 h of OCP were measured for HER for 3 cycles at a scan rate of 1 mV s−1 at this initial state (Day 0), and after 13 days (Day 13) and 30 (Day 30) days of electrolyte contact at OCP. The electrolyte (5 mL) was exchanged after the 13 days measurement.
To perform ILSTEM, MoS2, and Re0.2Mo0.8S2 nanocatalysts were drop cast on Au/C finder grids (Plano). To measure OCP, the grids were connected to the working electrode of a Gamry Reference 600 potentiostat through an Au wire (>99.99%, Redoxme) inserted in the grid. A RHE and a glassy carbon electrode were used as the reference and counter electrodes, respectively. The grids were left in contact with the electrolyte and were analyzed in the microscope fresh and after 13 days of electrolyte exposure. ILSEM measurements were conducted in a cross-section of the edge of the electrodes before and after 32 h of contact with the electrolyte.
For SFC-ICPMS measurements[57] a Gamry Reference 600 potentiostat, a saturated Ag/AgCl reference electrode (Metrohm), and a Pt wire counter electrode (99.997%, Alfa Aesar) were used. The 0.1 m H2SO4 electrolyte was analyzed online with a NexION300X spectrometer for Mo and Re elements. Y and Ir were added to the electrolyte in a 0.1 m H2SO4 solution as they served as internal standards to quantify Mo and Re, respectively.
Computational Methods
Spin-polarized density functional theory calculations were performed using the Vienna Ab initio Simulation Package and employed the projected-augmented wave method.[58, 59] The exchange-correlation interaction was described using the strongly constrained and appropriately normed (SCAN) functional.[60] The DFT+U method was adopted by applying a Hubbard U parameter of 2.05 eV for Mo.[61] The plane wave energy cutoff was set to 520 eV. The convergence criteria for the electronic total energy and the atomic force on atoms during all structural relaxations were set to 10−5 eV and 0.02 eV Å−1, respectively. The Brillouin zone was sampled with a k-point grid density of 100 per reciprocal space. All crystal structures for the chemistries under study were sourced from the Materials Project[62] and optimized using the SCAN functional. The energetics obtained were employed to construct the Pourbaix diagram, utilizing a modified version of the method used in the Materials Project.[61, 63] The stable domain in the Pourbaix diagram was identified by computing all possible equilibrium redox reactions in an aqueous solution. These reactions are represented by the equation: [reactants] + H2O ↔ [products] + mH+ + ne−. The Nernst equation connects the Gibbs free energy change of these reactions to the surrounding pH and external potentials. Quantitatively, the aqueous stability of a catalyst was determined by comparing its chemical potential to that of the stable domains on the Pourbaix diagram under operating conditions.
Statistical Analysis
The EDS data was evaluated in 3 identical locations before and after electrochemical treatment and the averages and standard deviation were compared. The original dissolution data of the SFC-ICPMS is shown in each graph and the smoothed data is plotted on top for clarity.
Acknowledgements
We acknowledge Petra Ebbinghaus for the initial Raman measurements, Benjamin Breitbach for the XRD measurements, Bettina V. Lotsch for providing the infrastructure for electrochemical measurements, Thomas Warkentine for proofreading the manuscript. R.A.-A. is grateful for financial support from the International Max Planck Research School for Interface Controlled Materials for Energy Conversion (IMPRS-SurMat). Z.W. acknowledges the funding support from City University of Hong Kong Start-up Grant 9020004. M.R. work was funded by the Deutsche Forschungsgemeinschaft under Germany's Excellence Strategy – EXC 2033 – 390677874 – RESOLV. F.P. gratefully acknowledges UK Research and Innovation (UKRI) funding under the grant reference EP/X027449/1. This research was supported by the Nano·Material Technology Development Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (NRF-2022M3H4A4097520). The authors acknowledge the support of the National Research Foundation of Korea (NRF) with a grant funded by the Korean government (MSIT) (No. NRF-2021R1F1A1061943 and RS-2024-00345717). This research was supported by Global- Learning & Academic Research Institution for Master's, Ph.D. students, and Postdocs (LAMP) Program (No. RS-2023-00301850) and Basic Science Research Program (RS-2023-00271205) of NRF grant-funded by the Korean Ministry of Education. This research was supported by the International Research & Development Program of the NRF funded by the MSIT (RS-2024-00439825). The support under the framework of the international cooperation program managed by the National Research Foundation of Korea (NRF-2023K2A9A2A22000124) and the Deutsche Forschungsgemeinschaft (DFG: ZH1105/3-1) is gratefully acknowledged.
Open access funding enabled and organized by Projekt DEAL.
Conflict of Interest
The authors declare no conflict of interest.
Open Research
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
