Ti2N nitride MXene evokes the Mars-van Krevelen mechanism to achieve high selectivity for nitrogen reduction reaction

Due to the challenges in their synthesis, a great amount of work was done to assure a successful synthesis was achieved prior to the NRR evaluation. The Ti₂NT_x MXene was prepared from the parent MAX phase (Ti₂AlN) via an oxygen-assisted molten salt fluoride etching technique under Ar flow, followed by exposure to air in which volume expansion occurs and color shifts from black to brown. Afterwards, the material is washed with sulfuric acid, followed by sonication in water to delaminate to single layer MXene flakes. The delamination in water is chosen to separate the exfoliated layers into a few-to-single layer morphology, which in turn will improve the catalyst’s performance in the NRR due to the increase in available active surface area. This synthesis method is summarized, and pictures of material at different steps in the synthesis are provided, in Fig. 1a–d.

(a) Schematic illustration of the synthesis of Ti₂NT_x MXene via oxygen-assisted molten salt fluoride treatment of the parent MAX phase Ti₂AlN at 550 °C for 5 h under flowing argon, then exposure to air, followed by fluoride salt removal in 4 M H₂SO₄, finally delamination is accomplished via sonication in water for 4 h. Models are not based on gathered data, only as a general guideline. In lab photographs of (b) MAX phase, (c) Multilayer Ti₂N MXene after acid washing, and (d) Single layer MXene acquired after delamination in water. (e) SEM imaging of Ti₂AlN MAX phase (black outline), molten salt treated MAX phase (blue outline), multilayer Ti₂N MXene (purple outline), and few layer Ti₂N MXene (red outline). The lateral size of the individual MXene flakes is roughly 5 μm. (f) XRD, (g) Raman, and (h) UV–Vis spectra of Ti₂AlN MAX phase (black) and single layer Ti₂N MXene (red). XRD was gathered using a zero-diffraction silicon plate with a well. Raman spectroscopy was gathered using 532 nm laser at 5% power at a 1 s exposure time. UV–Vis spectroscopy was collected using water as the matrix.

Morphological properties

It’s imperative to understand the morphological changes after the synthesis prior to the NRR characterization, as these can affect the MvK mechanism. The morphological transition from MAX to MXene due to the synthesis procedure was analyzed via SEM. Figure 1e shows the SEM images for the Ti₂AlN parent MAX phase, molten salt treated MAX phase, multilayer Ti₂NT_x MXene, and delaminated (or few-to-single layer) Ti₂NT_x MXene. It is clear to see the closely-packed layer structuring of the MAX phase (Fig. 1e black outline) is opened to the expected accordion-like structure upon completion of the oxygen-assisted molten salt fluoride process (Fig. 1e blue outline). Successful removal of the fluoride salts via sulfuric acid washing is evidenced by the thinning of the individual layers (Fig. 1e purple outline), consistent with the XRD results (Fig. 1f), and the low amount of sodium and potassium present in the electron dispersive spectroscopy (EDS) analysis (Fig. S1, supplementary information). After delamination in water, single layer flakes of lateral size 5 μm can be observed (Fig. 1e red outline). This flake size is larger than other Ti₂N MXenes synthesized before, and will be beneficial for NRR electrocatalysis^25,26. Alongside the visible changes in morphology, EDS further corroborates the successful synthesis of Ti₂NT_x material. The parent MAX phase has Ti:Al and Ti:N ratios of 2:1 for both, which is consistent with expected stoichiometry (Fig. S1, supplementary information). After oxygen-assisted molten salt treatment, the presence of Al is greatly reduced, while Na and K presence can be observed clearly (Fig. S1, supplementary information). The final delaminated MXene material shows a Ti:N and Ti:Al ratio of 2:1 and 18:1 or greater, respectively (Fig. S1, supplementary information). The large presence of Si in the EDS spectra is due to the substrate used for analysis. This major removal of Al from the parent structure will facilitate easier access to the N sites allowing for utilization of the MvK mechanism.

Structural and optical properties

Prior to the NRR evaluation, extensive characterization is done on the material to ensure phase purity. Analysis of the bulk structure and composition of the material is conducted via powder XRD. Figure 1f shows the progression of the XRD patterns as the material evolved from Ti₂AlN MAX phase (black) to delaminated single layer Ti₂N material (red). The XRD pattern of Ti₂AlN MAX phase matches with expectations from previous works^25,26. After the oxygen-assisted molten salt treatment process, we observe a clear shift to lower angles and a slight broadening of the (002) peak, as well as a drastic decrease in the (104) peak centered at ~ 40°, which is indicative of successful etching of the Al atoms from the 3D MAX structure to form sheets of Ti₂N (Fig. S2, supplementary information). Washing with sulfuric acid (Fig. S2, supplementary information) successfully removed the remaining fluoride salts on the structure as can be seen by the elimination of the peaks assigned to KF, LiF, and NaF (Fig. S3, supplementary information). Upon sonication, phase pure Ti₂N with a clear layered structure as evidenced by the existence of the (002) peak at ~ 9.6°, followed by (004) and (006) peaks at 19° and 28.5°, respectively. This location of the (002) peak corresponds to an interlayer spacing (d-parameter) and c-parameter of 0.920 nm and 1.84 nm, respectively. In comparison, the MAX phase exhibits an interlayer spacing and c-parameter of 0.342 nm and 0.684 nm, respectively. This tripling in the interlayer spacing indicates that clear separation of the layers from the MAX phase took place, meaning that the available surface area increased, exposing more N sites, leading to enhanced NRR selectivity and activity through the MvK mechanism.

Due to the inactivity of the parent MAX phase for NRR, it is crucial to ensure that proper etching took place. To further ensure this etching took place, Raman spectroscopy was used to investigate the vibrational modes along each step of the synthesis procedure and is shown in Fig. 1g. The MAX precursor displayed three main active Raman modes (three in-plane modes and one out-of-plane mode), which is line with reported expectations of a 211-type MAX phase. The molten salt treated MAX phase (Fig. S4, supplementary information) shows a blue shift in its active modes, which can be attributed to the reduction in the crystallinity of the material as the interlayer spacing increases due to exfoliation and intercalation of salt ions. Upon acid washing, the material transitions to having four clear features at 150, 265, 400, and 610 cm⁻¹ (Fig. S4, supplementary information) which are broader than the modes observed in the precursor MAX phase. This peak broadening stems from the material distortion of the P6₃/mmc structure due to the electron density of the surface termination groups²⁷. This peak broadening also supports the XRD and EDS results that more N sites will be available for the MvK mechanism. As the material exfoliation leads to separation of the layers, more H⁺ in solution will bond to the lattice N atoms more often than the termination groups, allowing for the MvK mechanism to be evoked, which will allow for higher selectivity of NRR over HER. After delamination, the four main modes are still present, with the main mode being the 400 cm⁻¹ peak, corresponding to the vibration of the Ti atomic planes within the structure now that the Al layers have been removed²⁶.

The acquired XRD and Raman spectra were compared against anatase, rutile, and alumina (Fig. S5, supplementary information) to confirm the stability of the newly synthesized material as no oxidation occurred during the synthesis procedure. To further confirm that no oxidation took place during synthesis, Fourier transform Infrared (FTIR) spectroscopy was taken between the MAX and MXene samples. As can be seen in Fig. S6 (supplementary information), the presence of oxygen species in the structure can be primarily attributed to surface Ti–OH groups, due to the feature at ~ 3300 cm⁻¹. The Ti₂N nitride MXene synthesized here has larger lateral size and less defects compared to the previously reported Ti₂N nitride MXene^25,26. This is because the new and improved synthesis approach employed here yields consistent, reproducible, and phase pure nitride MXenes as corroborated by the XRD, SEM, EDS, and Raman results. This method can be applied to more nitride MAX phases to make new nitride MXenes, that expands the application of MXene.

Stability of the Ti₂N nitride MXene

The Ti₂N material shows good stability in air and in aqueous media. As stability in aqueous media is one of the requirements for a good NRR catalyst, the properties and stability of the Ti₂N MXene was tested using UV–Vis spectroscopy (Figs. 1h, S7, supplementary information). Both the MAX precursor (Fig. 1h black trace) and molten salt treated MAX phase (Fig. S7, supplementary information) displayed purely metallic absorption properties, which is expected based on the atomic formula and reported properties of the materials²⁸. Upon acid washing, the multilayer material displays absorption at 250, 500, 925, and 1175 nm (Fig. S7, supplementary information), the last three of which are attributed to multilayer absorption due to their disappearance upon delamination to few layer Ti₂N material, which only displays an absorption peak at 250 nm (Fig. 1h red trace).

In practical applications, catalysts are expected to stay in solution for multiple days or weeks. Thus, to test whether or not the Ti₂N catalyst will meet this expectation, a colloidal solution with water was created, and placed into a vial for long-term colloidal stability analysis. After 28 days, no precipitation of material was observed, and no significant change was observed in the UV–Vis spectrum (Fig. S8, supplementary information), vastly outperforming their carbide MXene counterparts²⁹.

NRR performance via MvK mechanism

After ensuring that the synthesized Ti₂N is phase pure and that its morphology and compositions are favorable for the MvK mechanism, we proceed to evaluate its NRR performance. We also characterize Ti₃CN and Ti₃C₂ carbonitride and carbide MXene, respectively, for comparison and validation of the MvK mechanism. The catalytic NRR activity for the delaminated Ti₂N, Ti₃C₂, and Ti₃CN samples was investigated within an H-cell under ambient temperature and pressure. Figure 2 shows the cyclic voltammograms (CV) and linear sweep voltammetry (LSV) curves for these catalysts recorded in Ar- and N₂-saturated 0.1 M HCl electrolyte. We used 0.1 M HCl as the electrolyte due to its heavy use in previously reported NRR papers, and to ensure that a steady stream of H⁺ are available for the protonation of the N₂ molecules^8,17,18.

Cyclic voltammograms for (a) Ti₂N, (b) Ti₃CN, and (c) Ti₃C₂ MXenes conducted in Ar-saturated (black) and N₂-saturated (red) 0.1 M HCl electrolyte. All scans conducted using a scan rate of 20 mV/s. Linear sweep voltammograms for (d) Ti₂N, (e) Ti₃CN, and (f) Ti₃C₂ MXenes conducted in Ar-saturated (black) and N₂-saturated (red) 0.1 M HCl electrolyte. All scans conducted using a scan rate of 5 mV/s.

When comparing between the carbide MXene and the carbonitride and nitride MXene counterparts, it is clear to see that as the ratio of N atoms to Ti atoms within the lattice structure increase, the electrocatalytic performance under NRR conditions is enhanced. While the ratio of N to Ti in the structure may not be the only factor for this enhancement (termination groups and interlayer spacing could also play a role), based on the MvK mechanism, it is reasonable to assume that a large portion of the enhancement is due to this heightened ratio. Namely, the CVs for the carbonitride and nitride grow wider when switching from Ar to N₂, indicating more charge storage capability under N₂-saturated electrolyte. Whereas the carbide merely experiences a downward shift of its CV while still maintaining the same area. This indicates that the nitride and carbonitride MXene materials are able to take on more current as the reaction conditions switch, which will allow for enhanced NH₃ production. The lack of changes in the CV of the carbide MXene when switching to N₂-saturated electrolyte provides preliminary evidence of limited performance for NRR, which is in alignment with previous NRR studies of Ti₃C₂²³. In terms of the LSVs, the carbonitride and nitride experience a much greater anodic shift in their onset potential when switching from Ar-saturated to N₂-saturated, while the carbide experiences next to no shift when switching between reactive conditions. The lack of shift in the LSV for the carbide is once again indicative of a material whose selectivity and performance towards NRR is extremely limited. The anodic shift in the LSV for the nitride and carbonitride catalysts indicate that switching to N₂-saturated electrolyte leads to a drastic improvement in catalytic activity, due to easier NH₃ production through NRR. It is also clear to note that the Ti₂N material is able to display an almost ten-fold greater current density in the LSV when compared to the carbonitride and carbide. This greater current density observed for the Ti₂N can be attributed to the greater amount of N sites and improved conductivity and reactivity^30,31,32. This initial evidence of the MvK mechanism in Ti₂N is further supported by a N₂-free experiment, which shows that in Ar-saturated electrolyte, the Ti₂N consistently draws moderate current densities over 50 cycles implying that some reaction could be occurring within the crystal lattice structure of the material. This cycling experiment also indicate the stability of the catalyst under these environments.

For the Ti₂N nitride MXene, the onset potential for NRR was observed to be around − 0.25 V versus RHE, so the voltage window of catalytic performance was based around this area. Figure S9 (supplementary information) displays the time-dependent chronoamperometry (CA) data gathered during the experimental runs on Ti₂N, Ti₃CN, and Ti₃C₂, respectively. The CA curves show good stability throughout the 4-h NRR experiment. Through use of the indophenol blue method and Watt–Chrisp method, the yields of ammonia and hydrazine, respectively, were estimated. No significant yields of hydrazine were detected for any of the catalysts during any of the experimental runs (Fig. S10, supplementary information). Similar maximum absorbance values from the indophenol blue method were observed for applied potentials of − 0.25 and − 0.45 V versus RHE (Fig. 3a). Through use of the generated calibration curve (Fig. S11, supplementary information), the concentration, and therefore yield, of produced NH₃ was estimated to be about 11.33 and 11.38 μg/cm² hr, respectively. When taking into account the selectivity, or FE, of Ti₂N towards NRR over HER, the optimal NRR experimental run was observed at − 0.25 V versus RHE with a FE of 19.85% compared to 0.03% for the − 0.45 V versus RHE run. This drastic drop off in FE occurs due to the predominance of HER at more cathodic potentials as protons are drawn more towards the termination groups that catalyze HER instead of the edge sites where the nitrogen atoms are more easily accessed³³. This preference for HER at more negative potentials requires that these lower potentials be used for future studies. These performance metrics outperform pristine carbide MXenes²², ruthenium-platinum alloys³⁴, and MoS₂³⁵ in both yield and selectivity; modified carbide MXenes²³ in terms of yield; and ruthenium single atom catalysts³⁶ in terms of selectivity. Extra attention should be paid to the vast improvements in FE that we observe, as this can be directly related to the MvK mechanism allowing for heightened selectivity towards NRR over the troublesome HER competition.

NH₃ yield and Faradaic efficiency of (a) Ti₂N, (b) Ti₃CN, and (c) Ti₃C₂ MXenes at differing potentials after 4-h chronoamperometry experiments. An experiment was carried out in Ar-saturated electrolyte on Ti₂N MXene to provide evidence of MvK mechanism. All bars correspond to yield values (read to the left) while lines with markers correspond to FE values (read to the right).

When comparing the catalytic capabilities of the carbide, carbonitride, and nitride MXenes, it is clear to see a large distinction between materials, especially in terms of FE. Following a similar strategy as that applied to the nitride MXene catalyst, a maximum yield and FE for the carbonitride was observed at an applied potential of − 0.55 V versus RHE, and was found to be 3.352 μg/cm² hr with an efficiency of 0.03% (Fig. 3b), significantly lower than that for the Ti₂N catalyst. The carbide MXene catalyst exhibits a maximum yield and FE of 12.90 μg/cm² hr and 0.58%, respectively (Fig. 3c). Interestingly, these maximum values were observed at different potentials (− 0.35 V for yield and − 0.25 V for FE) compared to occurring at the same potential like what was observed with the nitride and carbonitride indicating that a trade-off between yield and FE could exist for the carbide due to it requiring an associative/dissociative mechanism to form NH₃—rather than an MVK mechanism, like in the case of Ti₂N and Ti₃CN. While the carbide MXene maximum yield rate is slightly higher than that of the nitride, the extremely low FE towards ammonia (~ 0.02%) production significantly hinders its application in this system. Once again, this low FE occurs due to the predominance of HER at more cathodic potentials. Whereas, in the N-based catalysts, the MvK mechanism is found predominant, which suppresses the HER.

The stability of the catalyst was analyzed through long-term CA experimentation at the highest performing reaction conditions (− 250 mV), and the Ti₂N catalyst is stable across this entire time span (Fig. S12, supplementary information). To further ensure stability of the catalyst, SEM imaging was conducted on the material before and after 4-h NRR conditions by transferring the drop-cast material to carbon tape. The layered morphology before and after are similar (Fig. S13, supplementary information), further indicating that the material is stable under the NRR conditions.

To ensure accuracy of our results and rule out the possibility of contamination in the gas inlet, several control experiments were conducted. For these experiments, blank electrodes were run through the system at experimental operating conditions, and it was found that no formation of ammonia or hydrazine took place, indicating that any product formation observed from experimental runs came solely from the catalytic MXene material present (Fig. S14, supplementary information). MXene material was then analyzed by performing 4-h CA experiments at standing potential (0.0311 V vs. RHE), and again, no formation of either product was observed, confirming that the material is not catalytically active under no applied charge. The final experiment that was performed was placing the Ti₂N MXene material in an Ar-saturated electrolyte under an applied charge of − 0.25 V versus RHE to establish preliminary results towards the support of the MvK mechanism on this nitride catalyst. Under these experimental conditions, it was consistently observed that a noticeable yield of ammonia (5.24 μg/cm² hr) was achieved. Due to the lack of nitrogen gas in the system, and confirmation that ammonia production only stems from the catalyst material, this produced ammonia could only derive from the nitrogen within the catalyst structure, which lends heavily towards the possibility of the MvK mechanism occurring. Further experimentation is currently being conducted on the MvK mechanism, including isotopic experimentation, further post-production characterization/catalyst stability, and in-situ spectroelectrochemical analyses; and will be reported at a later time. Nevertheless, the findings reported here constitute the first evidence for an MvK mechanism in Ti₂N and Ti₃CN catalysts; the implications of which goes beyond MXene catalysts.