2.1 Synthesis of MS-MXene with Various Surface Terminations

2.2 Material Characterization of Ti₃C₂T_x

The annealing treatment in Li₃N-containing molten salt electrolyte (equations 3 and 3’) induces a shift of the (002) peak from 2θ 8.0° to 7.0° (Figure 2a, blue plot), supporting an increase of the d-spacing from 11.0 to 12.5 Å; this result is another evidence that tuning the surface group is an effective route to control the interlayer distance of MXene, such as reported by Li et al. with the preparation of binary halogen terminated Ti₃C₂(ClBr) exhibiting a c-lattice parameter larger than individual Ti₃C₂Cl₂ and Ti₃C₂Br₂ MXene.¹¹ In a previous work,¹⁸ we have shown that a Li₃N annealing treatment resulted in the presence of N- and O-terminated MXene (termed as N,O−Ti₃C₂), as elemental analysis resulted in a composition of Ti₃C_1.95N_0.80O_1.11H_0.46. In our previous work,¹⁸ N₂ adsorption–desorption analysis has been conducted on both Cl−Ti₃C₂ and N,O−Ti₃C₂ to investigate changes in their specific surface area, porous volume, and pore size distribution. The results indicate that both samples exhibit similar specific surface areas (21.0 and 21.3 m² g⁻¹, respectively for Cl−Ti₃C₂ and N,O−Ti₃C₂) and total porous volume (0.065 and 0.071 cm³ g⁻¹). This reveals that the improvement in electrochemical performance of N,O−Ti₃C₂ is mainly attributed to the modification of surface group instead of the increase of the surface area and porosity.

While the nitrogen groups come from the molten salt, O-containing terminations could originate from different sources: from initial Ti₃AlC₂ MAX²¹ but also from Li₃N and LiCl partial reaction with moisture to form LiOH (see Figure S1), during the short exposure to air. In addition, for a better comparison of the elemental composition, O−Ti₃C₂ sample analysis has been performed by ICP technique and organic elemental analysis (for C, H, N, O, and S contents) (Table S2), revealing a composition of Ti₃C_1.34O_3.25H_0.12.

To further evidence the presence of O-containing surface groups, additional experiments were made to compare the spontaneous intercalation of water molecular behavior of LiF/HCl−Ti₃C₂, Cl−Ti₃C₂ and N,O−Ti₃C₂ and the results are shown in Figure 2b. LiF/HCl-MXene is well-known to be hydrophilic; it can incorporate additional structural water upon immersion in aqueous electrolyte for the stabilization of the structure and compensation of the surface charge for their ionic structural units.²² This is shown in Figure 2b, where a shift of the (002) peak to lower angle is observed for LiF/HCl−Ti₃C₂, after immersion in 3 M H₂SO₄ for 2 h, following H₂O intercalation.²³ A similar low angle shift of the (002) peak position was also observed for N,O−Ti₃C₂, suggesting a similar process. In contrast, Cl−Ti₃C₂ retains the same d-spacing suggesting spontaneous intercalation of water molecules does not occur, which might be due to the lack of hydrophilic terminations.^23b However, Figure S2 shows that O−Ti₃C₂ maintains a similar d-spacing, likely due to its already sufficient expanded d-spacing achieved during the washing process.

In addition, a N,O−Ti₃C₂ sample was annealed under Ar atmosphere at 300 °C for 10 mins. As a result, the (002) peak shifted from 7.0 to 8.8° (Figure 2a, purple plot). This could be explained by the dehydration and the loss of pre-intercalated H₂O during annealing, in agreement with Seredych et al²⁴ who showed the release of entrapped structural water in LiF/HCl−Ti₃C₂ MXene at temperature of ∼320 °C after performing thermal treatment on MXene. Cl−Ti₃C₂ has also been annealed in Li₂O-containing molten salt electrolyte, to exchange Cl and O atoms (equation 2): a shift of the (002) diffraction peak from 2θ 8.0° to 6.5° is then observed after annealing (see Figure 2a, red plot), which translated to an expansion of the interlayer distance from 11.0 to 13.5 Å. This suggests that the creation of O-terminations can induce the spontaneous intercalation of confined water adsorption and then the expansion of d-spacing. To support these assumptions, water contact angle (WCA) measurements (Figure S3) were further performed on the various MXene pellets, to stress out whether or not surface terminations are hydrophilic. The observed WCA values of 76°, 48°, and 37° for the Cl−Ti₃C₂, N,O−Ti₃C₂, and O−Ti₃C₂ samples, respectively, demonstrate a corresponding increase in hydrophilicity, consistent with the higher content of oxygen-containing polar groups.²⁵

To get further insights into the surface chemistry, temperature-programmed desorption coupled with mass spectroscopy (TPD-MS) has been performed on the Cl−Ti₃C₂ (black line), N,O−Ti₃C₂ (orange line), O−Ti₃C₂ (red line) MXene samples. As shown in Figure 2c, the H₂O evolution is different depending on the functionalization of the surface. Specifically, O−Ti₃C₂ has a higher content of confined H₂O, which supports our speculation that the amount of confined water increases with an increase in O-containing surface terminations, and the presence of the confined H₂O molecules is mainly controlled by the O surface termination groups. Differently, Cl−Ti₃C₂ MXene has the lowest content of water. Those results are in good agreement with the above XRD and water contact angle measurements. Moreover, the N,O−Ti₃C₂ exhibits another small water-loss peak above 400 °C. As reported by Seredych et al.,²⁴ the water loss peak below 400 °C can be attributed to the multilayer water (weak water–water interaction) while the water loss peak above 400 °C can be associated with monolayer water bound to −NH₃ and −OH groups (strong water–surface interaction). However, the water can be almost fully removed at 300 °C, which supports the conclusion that the shrunken d-spacing, as observed in Figure 2a, is associated with the removal of H₂O.

Figure S4, Figure S5 and Figure 2d show the decomposition of the surface terminations present on the Cl−Ti₃C₂, O−Ti₃C₂ and N,O−Ti₃C₂ MXene surface. Cl−Ti₃C₂ only shows a release of Cl gas over the whole temperature range, with the main release of H₂ gas occurring around 200 °C while O−Ti₃C₂ shows the presence of H₂ and Cl gas. In contrast, N,O−Ti₃C₂ exhibits a new substantial weight loss peak centered at around 400 °C, corresponding to −NH₃ or −OH groups, which is combined with a large H₂ release centered at around 500 °C. Although it is difficult to quantify −NH₃ or −OH due to the lack of standards, it seems more reliable that the peak is related to −NH₃ since the HF-MXene²⁶ with a large amount of −OH does not show as much H₂ release in the 400–600 °C. It is reported that the release of H₂ gas usually occurs in two different temperature ranges: molecular hydrogen trapped between MXene sheets in the 140–410 °C while a combination of −OH terminations at the 430–700 °C.²⁴ Therefore, it's a mainly slow dissociation of the −NH₃ group at the 200–600 °C, followed by a release of H₂ gas at the 400–600 °C due to the combination of −NH₃ groups.

A change in the MXene surface chemistry was also confirmed from zeta potential measurements in deionized water (pH=7). As shown in Figure S6, Cl−Ti₃C₂ has a positive zeta potential of ~+50 mV, whereas the zeta potential turns to negative values after surface termination exchange, with values of −40 and −48 mV for O−Ti₃C₂ and N,O−Ti₃C₂, respectively. This supports a significant change in the surface charge along with the termination groups and may result from the spontaneous chemisorption of cations on the MXene surface after immersion in the electrolyte.

A 2-probe resistance measurement set-up was used to measure the electric conductivity of MXene pellets with different surface terminations, as density functional theory²⁷ calculations have shown that surface terminations affect MXenes′ density of states and thereby their electronic conductivity. As shown in Figure S7, the electronic conductivities are 69.1 and 62.4 S m⁻¹ for the Cl−Ti₃C₂ and N,O−Ti₃C₂, respectively, while O−Ti₃C₂ exhibits a very low electronic conductivity of about 0.2 S m⁻¹. This is consistent with the report by Mitra et al., showing that the O-terminated HF-MXene exhibits the lowest conductivity compared to those with F and OH.²⁷

Overall, MXene with various surface terminations shows different properties regarding water confinement, electric conductivity, wettability, thermal stability, and even the surface charge.

2.3 Charge Storage Mechanism

The electrochemical performance of Cl−Ti₃C₂, N,O−Ti₃C₂, and O−Ti₃C₂ MXenes were evaluated in H₂SO₄ to understand the impact of surface terminations on the electrochemical behavior of MXene electrodes. Figure 3a shows that Cl−Ti₃C₂ has limited electrochemical activities, centered around a pair of redox peaks located at −0.58/−0.69 V vs. Hg/Hg₂SO₄. Such limited electrochemical activity could be explained by the lack of O-containing surface groups (see Figure S3 and Figure 2d), resulting in a hydrophobic surface.²⁵ In contrast, as we recently showed,¹⁸ as-synthesized N,O−Ti₃C₂ shows a drastic change in its electrochemical signature, showing a mirror-like CV with a capacitive rectangular box profile, together with one pair of sharp redox peaks located at −0.35/−0.43 V vs. Hg/Hg₂SO₄. Such an improvement in the electrochemical performance could be explained by the expanded d-spacing due to the presence of confined water between the MXene layers, thanks to the introduction of N-terminations. In addition, various surface terminations in materials affect electronic states around the Fermi energy, as indicated by DFT calculations, resulting in changes to the electronic structure of MXene layers and influencing the overall electronic properties of the material.²⁸ To support the role of confined water on the electrochemical performance, N,O−Ti₃C₂ sample was annealed at 300 °C in Ar to remove confined water. As shown in Figure 3b, the electrochemical activity of the annealed N,O−Ti₃C₂ was greatly reduced, indicating the key role of confined water on the electrochemical performance in H₂SO₄ electrolyte, which well agrees with previous simulation studies.^{26, 29}

As suggested in the XRD and TPD-MS results, O−Ti₃C₂ has a large d-spacing and a large amount of confined water. However, as shown in Figure 3a, the electrochemical performance of the O−Ti₃C₂ sample was found to be modest, in line with the very low conductivity of such material, with the presence of one pair of redox peaks at −0.55 and −0.68 V vs Hg/Hg₂SO₄, due to the presence of TiO₂ as a result of the surface oxidation of Ti, as already reported.³⁰ A similar set of redox peaks at −0.55/−0.66 V vs. Hg/Hg₂SO₄ observed for the N,O−Ti₃C₂ sample (Figure 3a) can then be assigned to the presence of O-containing surface groups. The other pair of redox peaks centered at −0.35/−0.43 V vs. Hg/Hg₂SO₄ is assumed to be associated with the contribution of N surface groups due to its absence in O−Ti₃C₂ and LiF/HCl prepared Ti₃C₂T_x. These findings evidence that the N-based functional group, −NH_x, as revealed in Figure 2d, FTIR and CHNO elemental analysis from our previous work,¹⁸ plays a key role in boosting the electrochemical performance. The −NH_x group polarity improves electrode wettability, facilitating surface redox charge transfer.³¹ As an active site, it aids electrochemical reactions and stabilizes the electrode structure through hydrogen bonding. Additionally, it modulates the material‘s electronic structure and offers a buffering effect in the acidic environment.^31a Overall, the −NH_x group optimizes ion transport, stability, and electrochemical activity in acidic conditions. Furthermore, the combination of N- and O-containing surface groups in MXenes acts in concert to offer highly accessible redox-active surface to electrolyte ions, possibly due to the presence of confined water, resulting in improved electrochemical performance in aqueous electrolyte.

Figure S9 shows the electrochemical impedance spectroscopy (EIS) measurements for N,O−Ti₃C₂, O−Ti₃C₂, and Cl−Ti₃C₂ electrodes at OCV. All the Nyquist plots show similar features with a high-frequency semicircle, possibly attributed to the small charge transfer resistance of the fast pseudocapacitive redox process—which turns out to be higher for O−Ti₃C₂, as it is more resistive—followed by a fast increase of the imaginary part of the impedance at low-frequencies. In the low-frequency region, the sharp (almost vertical) rise of the imaginary part of the impedance indicates a capacitive-like, non-diffusion limited behavior for both N,O−Ti₃C₂ and O−Ti₃C₂ electrodes. The smaller slope obtained for the pristine Cl−Ti₃C₂ electrode suggests ion transport limitation in agreement with the CV (Figure 3a). Figure S10 shows the Nyquist plots of N,O−Ti₃C₂, at different bias potentials. In the electroactive domain, that is in the −1.0 V/Ref to −0.15 V vs Ref/, the electrochemical behavior is in agreement with a (pseudo)capacitive behavior as can be seen from the increase of the imaginary part of the impedance at low frequencies. At −1 V Vs Ref, the Nyquist plot shows a charge transfer process, ascribed to the electrolyte decomposition as a result of the low cathodic overpotential (water reduction). Staircase potential electrochemical impedance spectroscopy (SPEIS) measurements were also performed for the N,O−Ti₃C₂ electrode polarized at various potentials to differentiate double-layer (see Figure S11), pseudocapacitance, and battery-like mechanisms. The electrode is held for 10 min at each potential (25 mV step). Phase is color plotted vs f and V: the high phase (absolute value) in red indicates more capacitive behavior. As shown in Figure S11, a (pseudo)capacitive behavior is visible over a large electrochemical potential window.

Interestingly, during the 1^st discharge in 3 M H₂SO₄ electrolyte (Figure 3c), N,O−Ti₃C₂ sample does not show the reduction peak at −0.43 V vs. Hg/Hg₂SO₄; instead, a sudden cathodic current increase is visible from −0.60 V vs. Hg/Hg₂SO₄. The subsequent first anodic scan shows an oxidation peak at −0.35 V vs. Hg/Hg₂SO₄ and the following cycles show the expected set of reversible redox peaks at −0.35/−0.43 V vs. Hg/Hg₂SO₄. These results suggest that electrochemical activation occurs during the first cycle.

2.3.2 EQCM

The EQCM technique was further used in the gravimetric mode to get access to the ionic and molecular species’ fluxes upon the reduction-oxidation of N,O−Ti₃C₂ MXene coated on a quartz crystal resonator used as a working electrode during polarization in a 3 M H₂SO₄ electrolyte. Figure 4 shows the CV profile (black lines) recorded at 50 mV ⋅ s⁻¹ over the first cycle and the stable, reproducible electrochemical signature of the following 10th cycle (Figure 4b). In this plot, the blue lines correspond to the associated frequency response; the open circuit potential (OCP) indicated by the red dots is taken as the origin of the charge (Q=0). Both Figures show a frequency increase during the anodic scan and the other way around for the negative cathodic sweep, as indicated by the arrows. Similar to the previous CV curve test in Swagelok (Figure 3c), the first cycle in Figure 4a does not show any cathodic reduction peak but a sharp increase in the current density below −0.60 V vs. Hg/Hg₂SO₄ is observed during the first discharge while a small oxidation peak at −0.32 V vs. Hg/Hg₂SO₄ shows up upon the reverse sweep. In contrast, Figure 4b shows a capacitive rectangular box profile together with a pair of redox peaks at −0.35/−0.43 V vs. Hg/Hg₂SO₄ after pre-cycling. Figure S13 shows that the sharp redox peaks located at −0.35/−0.43 V vs. Hg/Hg₂SO₄ got activated during the cathodic scan below −0.80 V vs. Hg/Hg₂SO₄. This finding is consistent with the previous in situ XRD result showing the abrupt d-spacing expansion at −0.60 V vs. Hg/Hg₂SO₄ over the first discharge in Figure 3d, suggesting the intercalation of water needed to activate the redox peak at −0.35/−0.43 V vs. Hg/Hg₂SO₄.

The motional resistance for the MXene electrodes was stable during polarization (see Figure S14a and b), thus validating the gravimetric mode of EQCM measurements and the further use of Sauerbrey's equation³⁴ (Equation (4)), to calculate the mass change based on the measured change of the quartz resonance frequency:

$$ (4)

where ▵m (g cm⁻²) is the mass change of the coated quartz, C_f is the sensitivity factor of the crystal (g cm⁻² Hz⁻¹) and ▵f is the quartz resonance frequency (Hz). In this study, the C_f was calculated to be 24.0 ng ⋅ cm⁻² ⋅ Hz⁻¹.

Figure 4c and d show the change of the electrode weight vs. the charge passed in the electrode over the reduction process during the first discharge and following stable discharge, where the arrows indicate the scan directions. In Figure 4c, the mass increases slowly during the reduction, which is in line with a counter-ion mechanism: the negative charge in the electrode is balanced by cation intercalation. The shaded areas in Figure 4c represent the two different potential regions (S1 and S2) during the 1^st reduction process as described in Figure 4a. According to Faraday's law, the average molecular weight per charge from the two potential regions (21 g ⋅ mol⁻¹ in the purple region and 14 g ⋅ mol⁻¹ in the orange region) is calculated to be about 18 g ⋅ mol⁻¹, suggesting the insertion of about one H₂O per H⁺ (H₃O⁺).

As shown in Figure 4b and Figure S15a, the starting and ending points of the frequency and mass loops overlap, implying the electrode is under steady-state conditions, after the first cycle corresponding to an electrochemical activation, the reaction process is quite reversible. Firstly, on average, the frequency decreases during the MXene reduction which is in line with a counter-ion charge compensation (cations). Then, three electrochemical stages (S3–S5) were identified during the negative sweep, according to the different observed slopes (Figure 4d): (S3) Between −0.1 and −0.3 V vs. Hg/Hg₂SO₄, the average molar weight m/z estimated from a noisy signal is low <5 g ⋅ mol⁻¹, supporting the insertion of mostly desolvated protons, which is consistent with the small d-spacing observed in Figure 3e and explains the low current density due to the lack of confined water. Below −0.35 V vs. Hg/Hg₂SO₄ in the (S4) region—the onset potential of the cathodic peak—a large increase in the m/z value (95 g ⋅ mol⁻¹) is observed, corresponding to an intercalation of 5.0 H₂O per H⁺. Such a large amount of water intercalation is very consistent with the abrupt expansion of the d-spacing previously observed in in situ XRD (Figure 3e). For potentials below −0.6 V vs. Hg/Hg₂SO₄, in the region (5), the solvation number of protons drops down, with 0.52 H₂O interacted per H⁺. This process is reversible as, during the anodic scan, the weight decrease at the onset potential of the anodic peak follows the path of the cathodic process (Figure S16b), with the de-intercalation of about 5.3 H₂O interacted per H⁺. Those results well agree with the abrupt contraction of d-spacing at the anodic oxidation peak region as observed in the previous in situ XRD result.

Figure 5 summarizes the results obtained. The shaded yellow area at high potentials shows a limited electrochemical activity of N,O−Ti₃C₂ MXene, associated with a shrunken d-spacing due to the lack of confined water. In contrast, the shaded green area in Figure 5 with improved electrochemical performance corresponds to N,O−Ti₃C₂ MXene with expanded d-spacing. Both the drastic jump/drop of the mass from EQCM and abrupt d-spacing changes from in situ XRD measurements support the evidence of (de)intercalation of solvated protons associated with a large amount (about 5) of water molecules starting from the sharp redox peak at −0.35/−0.43 V vs. Hg/Hg₂SO₄. These results reveal the importance of the presence of intercalated confined water molecules between the MXene layers, where fast access to the interlayer allows for rapid surface redox involving proton intercalation and a fast proton transfer process.

The set of results reported here also experimentally confirms the key role of the presence of confined water in improving the reaction kinetics. Beyond MXene materials, it offers an efficient strategy, based on the incorporation of structural water inside layered materials by tuning the surface chemistry, for developing the next generation of high-power electrochemical energy-storage devices (EES) operating in aqueous-based electrolytes.