The Weyl Semimetals MIrTe4 (M = Nb, Ta) as Efficient Catalysts for Dye‐Sensitized Hydrogen Evolution

The prevalent global energy crisis calls for searching viable pathways for generating green hydrogen as an alternative energy resource. Dye‐sensitized photocatalytic water splitting is a feasible solution to produce green hydrogen. However, identifying suitable catalysts has been one of the bottlenecks in driving dye‐sensitized photocatalysis efficiently. In this work, a new class of electrocatalysts is reported based on the layered Weyl semimetals MIrTe4 (M = Nb, Ta) for the Eosin Y (EY)‐sensitized hydrogen evolution reaction (HER). NbIrTe4 and TaIrTe4 exhibit HER activities of ≈18 000 and 14 000 µmol g−1 respectively, after 10 h of irradiation with visible light. Time‐dependent UV‐Vis spectroscopy and high‐pressure liquid chromatography coupled with mass spectrometry analysis shed light on the reaction dynamics and enable a deeper understanding of the observed trend in hydrogen evolution rates for MIrTe4. MIrTe4 semimetals outperform transition metal‐based Weyl semimetals in terms of catalytic HER activity using EY as photosensitizer and triethanolamine as the sacrificial agent. It is hypothesized that the topology‐related band inversion in MIrTe4 Weyl semimetals promotes a high density of M d‐states near the Fermi level, driving their high catalytic performance. This study introduces a new class of layered Weyl semimetals as efficient catalysts, and provides perspectives for designing topology‐enhanced catalysts.


Introduction
The rapidly increasing need for decentralized energy supply to meet our global society's growing energy demands has put the DOI: 10.1002/aenm.202300503 spotlight on alternative green energy resources. Hydrogen is considered as one of the most valuable green energy resources which holds potential to mitigate the current energy crisis due to its high gravimetric energy density of ≈120 MJ kg −1 . Sunlight-driven water splitting is an eco-friendly way to generate hydrogen and an emerging alternative to water electrolysis. Given the limited amount and complexity of semiconductor materials that either drive the oxygen evolution reaction (OER), hydrogen evolution reaction (HER), or overall water splitting, HER/OER electrocatalysts are conveniently screened as photocatalysts in conjunction with dyes as light absorbers. [1][2][3][4][5] In typical dye-sensitized HER, a dye photosensitizer and an electrocatalyst are employed together with a sacrificial electron donor. [6,7] One of the most commonly used photosensitizer dyes for HER is Eosin Y (EY, 2″,4″,5″,7″-tetrabromofluorescein), which undergoes photoexcitation on a comparatively fast timescale of ps-ns. [7] Next, (in case of reductive quenching of the excited state) the photosensitizer/dye accepts electrons from the sacrificial donor. [8][9][10][11] In the following step, a designated HER catalyst then transfers the photogenerated electrons from the photosensitizer to the surface-adsorbed protons/hydrogen to generate H 2 gas at longer timescales (≈ms-s), [7,[12][13][14] and consecutively the photosensitizer/dye reverts back to the ground state. Finding suitable eletrocatalysts with fast charge transfer kinetics, both internally and interfacially, is a crucial step toward efficient dye-sensitized HER.
Commonly employed catalysts for dye-sensitized HER are noble metal-decorated wide bandgap semiconductors such as TiO 2 or ZnO, or zero-bandgap semimetals like graphene. [6,11,[15][16][17][18][19] Transition metal dichalcogenides (TMDs), namely 1T-MoS 2 , 1T-MoTe 2 , or WTe 2 , also exhibit efficient EY sensitized HER. [20][21][22] Theoretical studies envisage that TMDs with d-band density around the Fermi level may facilitate the charge transfer between the adsorbed species and the TMD, thus accelerating the catalytic process. [23][24] The efficiency of this process depends on the relative positions of the d-band center of the metal atoms and the s-/p-orbitals of the adsorbate, which determines the strength of the transition metal-adsorbate interaction (chemisorption energy). In general, a metal site with a higher (lower) d-band center exhibits stronger (weaker) affinity to adsorbates due to decreased (increased) filling of adsorbate-metal antibonding states. The d-band center of the metal atoms has been established as a useful descriptor within the theoretical framework to understand the variations in chemisorption energies of different adsorbates on transition-metal surfaces and their derivatives to design efficient catalysts. Alternatively, topological quantum materials with their robust surface states and high carrier mobilities (μ) are expected to support the charge transfer kinetics and have shown to be excellent catalysts in previous studies. [20,[25][26][27][28][29] Interestingly, 1T′-MoTe 2 , which shows promising HER activity, is a Weyl semimetal in addition to possessing Mo d-band density near the Fermi level. Hence the question arises whether d-band density is the sole driving factor for the good catalytic performance of 1T′-MoTe 2 , or whether topologically nontrivial surface states also play a (perhaps decisive) role in determining the catalytic activity of the material. Felser et al. show that 1T′-MoTe 2 is a more active HER catalyst compared to topologically trivial 2H-MoTe 2 . [20] Further, Weyl semimetals such as NbP, TaP, NbAs, and TaAs exhibit high catalytic performance while topologically trivial metallic 1T-TaS 2 does not show any activity toward EY-sensitized photocatalytic HER. [20] While these findings point to a correlation between a topologically nontrivial band structure and catalytic activity, a direct causal relationship between the two has been elusive so far. Nevertheless, the coexistence between topologically nontrivial surface states and metal d-electron density near the Fermi surface can serve as a guide in designing highperformance catalysts. [30] Following this concept, we conjectured that the ternary bimetallic TMD-type Weyl semimetals MIrTe 4 (M = Nb, Ta), which host a large density of M d-band states at the Fermi level induced by the bulk topology, could be suitable candidates for the HER. In addition, density functional theory calculations recently predicted that monolayers of NbIrTe 4 and the related TaRhTe 4 could be efficient catalysts for the oxygen reduction B. V. Lotsch Cluster of Excellence E-conversion Lichtenbergstrasse 4a, 85748 Garching, Germany reaction. [31] Hitherto, the catalytic performance of MIrTe 4 (M = Nb, Ta) Weyl semimetals has not been reported for dye-sensitized HER which motivated us to look into the catalytic performance of this class of 2D Van der Waals solids for EY-sensitized HER.
MIrTe 4 crystallizes in a noncentrosymmetric orthorhombic layered structure with space group Pmn2 1 , similar to WTe 2 as shown in Figure 1a. [32][33][34] In contrast to the cell parameters of WTe 2 , the alternating arrangement of the two metals M and Ir leads to a doubling of the unit cell along the b-direction. Along the a-axis, alternating M and Ir atoms form zigzag chains. Recently, MIrTe 4 (M = Nb, Ta) have gained significant attention due to their exciting electronic properties and exotic surface states. [32][33][34][35][36][37][38] The bulk electronic structure of MIrTe 4 (M = Nb, Ta) clearly shows that MIrTe 4 is a semimetal with a high density of states (DOS) near the Fermi level dominated by the Md and Te-p orbitals (Figure 1c,d and Figures S3-S4, Supporting Information). [34] Ternary MIrTe 4 (M = Nb, Ta) were theoretically predicted as type-II Weyl semimetals. [39][40] TaIrTe 4 has experimentally been verified as type-II Weyl semimetal by angleresolved photoemission spectroscopy (ARPES) and exhibits four Weyl points, the minimum number of Weyl points allowed for a time-reversal invariant. [35,37] The observation of Shubnikov-de Haas (SdH) oscillations, the light effective mass of charge carriers and the nontrivial Berry phase in NbIrTe 4 could be associated with Weyl fermions in NbIrTe 4 . [33,36] Electrical transport measurements show that MIrTe 4 are highly conductive materials with conductivities on the order of ≈10 4 S cm −1 . [32][33] A favorable electronic structure characterized by the metal d-band density near the Fermi level, which is enhanced by the topology-driven band inversion [41] of the Weyl semimetals, is likely to facilitate the charge transfer between the adsorbed species and the catalysts, therefore making MIrTe 4 appealing candidates for catalytic applications.

Results and Discussion
MIrTe 4 (M = Nb, Ta) was synthesized by a high temperature solid-state diffusion process, described in detail in the experimental section of the Supporting Information. The phase purity of the as-synthesized MIrTe 4 samples was examined by Xray powder diffraction (XRPD) and the graphical results of the final Rietveld refinements are presented in Figures S1 and S2, Supporting Information. All the reflections in the XRPD patterns of the finely powdered MIrTe 4 samples can be indexed in the orthorhombic crystal system (space group Pmn2 1 ) of MIrTe 4 . Rietveld [42] analysis of the XRPD of TaIrTe 4 and NbIrTe 4 shows the presence of significant preferred orientation effects due to their pronounced plate-like morphology and anisotropic peak broadening, in particular of the non-00l reflections caused by stacking fault disorder in the system, which is further verified by microscopic analysis. Back-scattered scanning electron microscopy (SEM) reveals a plate-like morphology of the MIrTe 4 samples with a lateral size distribution between ≈1 and 25 μm (Figures S5a and S6a, Supporting Information). Energy dispersive X-ray (EDX) analysis further confirms the composition of the synthesized MIrTe 4 powders (Figures S5b and S6b, Supporting Information), which is very close to the theoretical composition of NbIrTe 4 and TaIrTe 4 . We note that EDX reveals 3-5 mol% of oxygen, which may indicate the presence of thin layers of We performed X-ray photoelectron spectroscopy (XPS) analysis in order to acquire information about the surface oxidation states and chemical environment of MIrTe 4 ( Figure S10, Supporting Information). Nb-3d XPS spectra show an intense Nb 3d 5/2 and 3d 3/2 doublet peak centered at 206.4 and 209.2 eV, which is best ascribed to the expected +4 oxidation state of Nb. [43] Similarly, the Ta 4f 7/2 and 4f 5/2 spin-orbit doublet peak at binding energies of 26.5 and 28.4 eV is attributed to the +4 oxidation state of Ta. [43] Minor peaks in the Nb-3d and Ta-4f XPS spectra (-Nb 3d 5/2 peak centered at 203.2 eV; Ta 4f 7/2 peak centered at 23.7 eV) are best assigned to the +2 oxidation states of Nb and Ta in NbO and TaO, most likely formed due to exposure of the MIrTe 4 samples in the air. The anticipated Ir +4 is detected with its distinctive 4f 7/2 peak centered at 61.1 and 61.6 eV binding energy for NbIrTe 4 and TaIrTe 4 respectively. [44][45] There are two doublet peaks in the Te 3d XPS spectra with similar intensity. Te 3d 5/2 peak centered at 572.5 eV for MIrTe 4 best represents the −2 oxidation state of Te in tellurides, while the characteristic Te 3d 5/2 peak at a higher binding energy of 575.5 eV in MIrTe 4 is best attributed to the +4 oxidation state of Te in TeO 2 . [43][44][45] These results show that Te is prone to surface oxidation and forms a thin oxide layer on the surface of MIrTe 4 .
To check the activity of the MIrTe 4 toward dye-sensitized photocatalytic HER in the batch measurement method, Eosin Y (EY) dye is used as photosensitizer together with triethanolamine (TEoA) as the sacrificial electron donor. In a typical experiment, x mg (x = 0.6-5 mg) of MIrTe 4 and y mg (y = 2.5-12.5 mg) EY  [20] for EY sensitized HER.
were suspended in a 0.56 M aqueous solution of TEoA (10 ml). The resulting suspension was irradiated with AM1.5G solar simulator equipped with a ≥ 420 nm filter for different intervals of times. The headspace of the reactor was sampled consecutively at discrete intervals to analyze the evolved amount of hydrogen by gas chromatography (described in detail in the experiment section, Supporting information). Both TaIrTe 4 and NbIrTe 4 show high activity toward dye-sensitized HER (Figure 2 and Figure  S11, Supporting Information). In order to identify the optimum dye-to-catalyst ratio, we initially fixed the dye concentration to 5 mg and then varied the catalyst amount from 0.6-5 mg in 10 ml aqueous solution (Figure 2a and Figure S11a,c, Supporting Information). MIrTe 4 produced the highest amounts of H 2 /g of the catalyst for 1 mg catalyst in 10 ml 0.56 M aqueous solution of TEoA containing 5 mg of EY, while the hydrogen evolution activity decreases with further increase in catalyst concentration (>1 mg). This fact can be attributed to the light-shielding effect of the catalysts. [46] In the next step, we varied the dye concentration keeping the catalyst amount fixed to 1 mg in aqueous electron-donor solution. The amount of the evolved hydrogen initially increased with increasing the dye concentration, reaches a maximum and then decreases (Figure 2b,c, Figure S11b,d, Supporting Information). The optimum catalyst-to-dye ratio for NbIrTe 4 is 1: 7.5 (w/w) ( Figure 2b) and for TaIrTe 4 the optimum catalyst to dye ratio is 1: 10 (w/w) (Figure 2c), for optimized HER activity. NbIrTe 4 exhibits slightly better performance than TaIrTe 4 : NbIrTe 4 exhibits HER activity of ≈18 000 μmol g −1 after the 10th h of illumination, while TaIrTe 4 exhibits HER activity of ≈14000 μmol g −1 after the 10th h of illumination. This result is also in accordance with a recent report, which shows that the Weyl semimetal NbP exhibits slightly better HER activity compared to TaP (which is also a Weyl semimetal). [20] The observed trend in HER activity with varying EY concentration is in agreement with earlier reports [18] and is rationalized as follows: Initially with the increasing dye concentration, the number of excited dye molecules and effective transfer of the photogenerated charges to the catalysts increases, overall enhancing H 2 evolution. With further increase in dye concentration, incident light is increasingly attenuated while intimate contact between the sensitizer molecules and catalyst particles decreases, causing self-quenching of excited dye molecules. This phenomenon inhibits the efficient use of incident light and the photocatalytic activity declines accordingly. The catalytic performance of NbIrTe 4 for other dyes such as Fluorescein or Erythrosin B sensitized HER ( Figure S12, Supporting Information) shows that NbIrTe 4 is an active catalyst in all cases, with the highest activity for EY. Notably, in a control experiment without a metal catalyst, EY exhibits poor HER activity (66 μmol for 10 mg EY dye after 12 h of irradiation, Figure S13, Supporting Information) under similar conditions. This finding suggests that MIrTe 4 provide binding sites for H atoms (see discussion below) and effectively transfer the photogenerated electrons from the photosensitizer to the adsorbed H atoms producing H 2 , therefore driving HER efficiently which has been further verified by photoelectrochemical measurements ( Figure S14, Supporting Information).
Further, we compared the turnover frequency (TOF) of MIrTe 4 with other Weyl semimetals for EY-sensitized HER in Figure 2d. [20] For calculating the TOF, which represents the number of moles of H 2 evolved per hour and per mole of catalyst used, we considered all the sites of MIrTe 4 as active for HER, which strongly underestimates the actual TOFs given that only the surface sites will be active for catalysis. [20,47] Figure 2d shows that MIrTe 4 outperforms other Weyl semimetals for EY dye-sensitized HER. [20] We also compared the hydrogen evolution activity and rate in EY-sensitized HER for MIrTe 4 and recently developed nonprecious metal catalysts based on topological materials, transition metal chalcogenides, MXenes, and oxides (Table S3, Supporting Information). These results suggest that a combination of topology-enriched band structure and metal d-band densitywhich may be interrelated-near the Fermi level in MIrTe 4 may indeed be beneficial for the charge transfer process in catalysis, albeit a causal relationship cannot be established at this stage.
In order to understand these high HER activities of the  Figure 1c,d, Figure S3-S4, Supporting Information). For all other adsorption sites involving Ir, the ΔG H is relatively large, suggesting weak catalytic activity. In addition, the small DOS contribution of Ir near the Fermi energy is expected to lead to overall insignificant p-d interactions. This is in line with previous calculations [20] for the active H-adsorption sites in TaAs and TaS 2 -related materials, where the most active sites are all closely related to the metal atoms showing large d-band DOS near the Fermi energy. Besides, previous calculations [31] show that the most active reaction sites for oxygen reduction reaction in NbIrTe 4 are very similar to the m1 and t1 sites for HER as shown here. The average ΔG H of the two special sites m1 and t1 is ≈0.77 eV for NbIrTe 4 , and 0.89 eV for TaIrTe 4 . The average ΔG H of NbIrTe 4 is slightly lower than that of TaIrTe 4 , which may explain the slightly higher HER activity of NbIrTe 4 compared to that of TaIrTe 4 .
We hypothesize that the large M d-band DOS near the Fermi energy, which is related to the Weyl semimetal phase, is the driving force to improve the catalytic performance in MIrTe 4 . The band topology involves the band inversion between the valence and conduction bands, [41] which are strongly hybridized Nb/Ta-d and Te-p orbitals in the current case. Therefore, the band inversion forces the existence of a large density of d-states near the Fermi energy. We point out that the band inversion is robust in both bulk and thin layers and insensitive to SOC, although the band anti-crossing gap and position of Weyl points are sensitive to SOC. In this sense, it is not surprising to find that m1 (Nb/Ta site) and t1 (Te site, but strongly mixed with Nb/Ta-d states) sites are the most active ones on the surface, which are not sensitive to SOC. We thus propose a mechanism inherited from the bulk topology where the band inversion comes with large d-band DOS near the Fermi energy. This mechanism is independent of but does not necessarily exclude the possible contribution from topological surface states.
To confirm the reproducibility and long-term stability of the HER activity of MIrTe 4, we repeated the experiment for different batches of MIrTe 4 samples (Figure 4a,b). From these measurements, we calculate the standard deviation between different measurement batches and conclude that the HER activity of MIrTe 4 can be reproduced within a maximum standard deviation of ≈ 8%. Further, upon the addition of freshly prepared dye and sacrificial agent to the solution, MIrTe 4 resume their HER activity (within a maximum standard deviation of ≈10%), as probed for a total of 45 h (Figure 4c,d). Detailed SEM and EDX analysis of MIrTe 4 (M = Nb, Ta) samples do not show any significant change in their morphology and composition after catalysis (Figure S16-S17 and Table S4-S7, Supporting Information). However, we observed a small increase in the oxygen content in the EDX (5-8%) after catalysis that could also arise from the dye and the SiO 2 substrate. These findings ensure the stable and reversible HER performance of MIrTe 4 (M = Nb, Ta). The dynamic plots of HER activity with time ( Figure 4 and Figure S11, Supporting Information) show that initially the catalytic activity of MIrTe 4 (M = Nb, Ta) increases sharply with illumination time. However, after 1 h the HER activity becomes sluggish. In order to obtain mechanistic insights into the hydrogen evolution rate (R H2 ), we have dynamically recorded the HER rate of the MIrTe 4 with optimum catalyst/dye ratio using a continuous flow reactor that has been optimized to obtain kinetic information (Figure 5 and Figure S18, Supporting Information) as previously reported by us. [48][49] The decay in the hydrogen evolution rate curve is fitted across the entire time domain by an exponential function and by mathematically disentangling transient times (discussed in detail in the Supporting Information, Tables S8 and S9, Supporting Information). We find that the induction time for NbIrTe 4 (t 0 ≈ 0.66 h) is slightly shorter than that of TaIrTe 4 (t 0 ≈ 0.71 h), which also contributes to a slightly higher HER activity of NbIrTe 4 . The maximum hydrogen evolution rate R max from both experiments is extracted at the initial point where the exponential starts at t 0 with an average value of ≈ 8999.96 μmol h −1 g −1 for NbIrTe 4 and ≈ 7778.98 μmol h −1 g −1 for TaIrTe 4 . The asymptotic hydrogen evolution rate (R ∞ ) is 2189.32 μmol h −1 g −1 for NbIrTe 4 and 2019.81 μmol h −1 g −1 for TaIrTe 4 . In general, this result confirms that NbIrTe 4 is catalytically slightly more active than TaIrTe 4 , which is in good agreement with the data obtained from the discrete batch measurements.
In order to understand the observed trend in the hydrogen evolution rate, which shows a steep increase followed by an exponential decrease ( Figure 5), we characterized the sample solution before and after catalysis. The fact that MIrTe 4 samples do not show any significant change in their morphology and composition after catalysis, confirmed by SEM and EDX analysis (Figure S16-S17 and Table S4-S7, Supporting Information), may suggest that the surface states remain intact during catalysis, which is an important prerequisite for their possible involvement in the charge transfer process. On the contrary, EY undergoes chemical trans-www.advancedsciencenews.com www.advenergymat.de formation under irradiation as revealed by UV-vis spectroscopy and high-pressure liquid chromatography (HPLC) coupled with mass spectrometry (MS) analysis. First, we characterized the aqueous solution of the EY dye only by HPLC-MS and UV-vis spectroscopy ( Figure S19, Supporting Information). A prominent absorption at 516 nm appears due to the -* transition in the EY system ( Figure S19c, Supporting Information). [18,50] The observed shoulder at ≈480 nm in the absorption spectrum is characteristic of the aggregated xanthene dye species. Next, we recorded the time-dependent UV-vis spectra of the solution containing EY, catalysts, and TEoA for different illumination time (Figure 5c,d). The absorption spectrum of the solution shows a gradual shift toward shorter wavelengths with increase in illumination time, indicating a change in the chemical structure of the chromophore under illumination. Most interestingly, the change in the absorbance maximum of the dye with increasing irradiation times (inset of Figure 5c,d) mirrors the dynamics of the HER rate and thus explains the observed trend in the HER rate for the MIrTe 4 catalysts (Figure 5a,b). To further understand the fate of EY under illumination, we performed HPLC-MS analysis for the solution containing only Eosin Y dye before illumination ( Figure S19a,b, Supporting Information) and the solutions containing the MIrTe 4 catalysts, Eosin Y dye, and TEoA after dif-ferent illumination times ( Figure S20, Supporting Information). HPLC-MS spectra ( Figure S20, Supporting Information) confirm that under illumination and in the presence of TEoA, EY undergoes gradual de-bromination and ultimately forms a completely debrominated product (fluorescein). Formed by additional reduction steps, as shown in Figure S20a, Supporting Information, reduced forms of the debrominated compounds, e.g. dihydroflurescein, were also found in the reaction mixture in different relative amounts, depending on the irradiation time. This observation is in agreement with earlier reports on photodegradation of the xanthate dyes. [19,51]

Conclusion
In conclusion, we explore the EY-sensitized photocatalytic HER activity of the layered Ir-based transition metal chalcogenides, MIrTe 4 (M = Nb, Ta), which combine metal d-electron density at the Fermi level with nontrivial band topology. We report high activities of ≈18 000 μmol g −1 for NbIrTe 4 and ≈14 000 μmol g −1 for TaIrTe 4 after 10 h of irradiation with visible light for the EY sensitized photocatalytic HER. MIrTe 4 thus outperforms most related catalysts based on transition metal chalcogenides and Weyl semimetals for EY dye-sensitized HER. Our results further substantiate previous reports where high catalytic activities go hand in hand with a favorable binding energy of the surfacebound H atoms according to d-band theory. We point out that the d-band density of states can be naturally enhanced by the bulk band inversion in topological semimetals and topological insulators, which may be a general guiding principle to design topologyenriched catalysts. We note that while we observe a correlation of catalytic activity with the above salient features in the band structures of MIrTe 4 (M = Nb, Ta), a clear causal relation between nontrivial topology and catalytic activity cannot be established at this stage. Nevertheless, we have introduced a new class of layered Weyl semimetals for dye-sensitized catalysis, which strengthens the link between the still largely disjunct fields of topology and catalysis.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.