Silica-Pillared Mo 2 TiC 2 MXene for High-Power and Long-life Lithium and Sodium-ion Batteries

In this work, we apply an amine-assisted silica pillaring method to create the first example of a porous Mo 2 TiC 2 MXene with nanoengineered interlayer distances. The pillared Mo 2 TiC 2 has a surface area of 202 m 2 g -1 , which is among the highest reported for any MXene, and has a variable gallery height between 0.7 and 3 nm. The expanded interlayer distance leads to significantly enhanced cycling performance for Li-ion storage, with superior capacities, rate capabilities and cycling stabilities in comparison to the non-pillared version. The pillared Mo 2 TiC 2 achieved capacities over 1.7 times greater than multilayered MXene at 20 mA g -1 ( ≈ 320 mAh g -1 ) and 2.5 times higher at 1 A g -1 ( ≈ 150 mAh g -1 ). The fast-charging properties of pillared Mo 2 TiC 2 are further demonstrated by outstanding stability even at 1 A g -1 (under 8 min charge time), retaining 80% of the initial capacity after 500 cycles. Furthermore, we use a combination of spectroscopic techniques (i.e. XPS, NMR and Raman) to show unambiguously that the charge storage mechanism of this MXene occurs by a conversion reaction through the formation of Li 2 O. This reaction increases by 2-fold the capacity beyond intercalation, and therefore, its understanding is crucial for further development of this family of compounds. In addition, we also investigate for the first time the sodium storage properties of the pillared and non-pillared Mo 2 TiC 2 . of Na-ion as a promising low-cost alternative to lithium-ion we also report for the first time results for the use of Mo 2 TiC 2 in Na-ion half-cells.


Introduction
Over recent years, there has been incredible growth in the research and application of Li-ion batteries, which are now widely used in portable electronics, electric vehicles and grid storage applications. 1,2 However, further uptake of these technologies in more demanding applications, such as fast-charging electric vehicles and grid storage, requires significant improvements in high-rate charging and cycling lifetime, without compromising their energy density. Since these characteristics are determined by the electrode materials, there is an urgent need to develop new materials that can satisfy these demands. Currently used negative electrodes for Li-ion batteries often suffer from poor rate capability (for example, state-of-the-art negative electrode material graphite), and those with an impressive performance at high rates cannot achieve high capacities overall (lithium titanate only has a theoretical capacity of 150 mAh g -1 and niobium oxides have a capacity of 200 mAh g -1 , with both having little variation over different rates ) [3][4][5] .
Two-dimensional (2D) materials such as graphene have emerged as promising candidates for nextgeneration high-rate negative electrode materials due to their combination of high electrical conductivity and large 2D channels, which allow for fast electron and Li + diffusion, facilitating fast charging times. 6 However, 2D materials typically suffer from issues such as restacking of nanosheets during cycling, which can block Li diffusion channels, leading to low capacities, rate capabilities and cycling stabilities. 7 It is therefore crucial to develop methods which give rise to controlled open electrode architectures that are stable even when cycled at high rates.
MXenes are an exciting family of 2D materials which have attracted significant research attention since their discovery in 2011, especially in the field of energy storage. [8][9][10][11] However, like other 2D materials, electrode architecture plays a crucial role in their electrochemical performance, with multilayered or restacked MXenes showing poor cycling performance. 12,13 Titanium-based MXenes such as Ti2C or Ti3C2 have been by far the most heavily studied of the MXene family, despite more than 30 different MXenes having been synthesised to-date. 14 This is particularly important because Ti-based MXenes suffer from poor initial coulombic efficiencies (typically 40-60%), which severely limit their application in full cells. 15 In 2015, Anasori et al. first reported Mo2TiC2, an out-of-plane ordered MXene, with Mo occupying the outer metal layers, while the inner metal layer is exclusively Ti. 16 This allows the effect of the outer metal element to be studied, since Mo2TiC2 is otherwise analogous to Ti3C2. 16 Mo2TiC2 had several promising features for Li-ion battery applications, with delaminated Mo2TiC2 showing capacities up to 260 mAh g -1 , an initial coulombic efficiency of 86% and low average voltage. 16 Despite this, there have only been a handful of reports on this MXene, [16][17][18][19][20] with only one other reporting Liion battery performance.
Significantly, unlike Ti-based MXenes, the load curve for Mo2TiC2 displayed a plateau below 0.6 V, suggesting a different charge storage mechanism. Computational studies implied that a conversion reaction occurs between lithiated Mo-O surface groups (formed via Li intercalation in Reaction 1) and two further moles of Li as shown by Reaction 2, boosting the capacity. 16 The theoretical capacity achieved in Reaction 1 is 180 mAh g -1 , which increases to 356 mAh g -1 after the proposed conversion reaction.
3 Mo2TiC2O2 + 2 Li + + 2 e -→ Mo2TiC2O2Li2 (1) Mo2TiC2O2Li2 + 2 Li + + 2 e -→ Mo2TiC2 + 2 Li2O (2) The proposed mechanism is similar to the lithiation of Mo oxides, which is accompanied by large volume changes, causing significant capacity fade. 21 This could explain the relatively high fade seen in previous Mo2TiC2 studies, 16,18 demonstrating the need to develop methods that optimise the electrode architecture and increase the cycling stability of this material. There have been no reports to-date on engineered electrode architectures for Mo2TiC2 in electrochemical applications, despite the clear promise of this material. Furthermore, the lithiation mechanism for this material has not been experimentally verified yet.
Pillaring is a technique used to make porous layered materials from non-porous precursors, by inserting foreign species into the interlayer, which expands the pore space and creates stable architectures which prevent sheets from restacking. 22 This technique has recently been applied to MXenes, and has been shown to improve performance in a variety of electrochemical energy storage applications such as Li, Na and Zn-ion batteries, aqueous supercapacitors and solid-state supercapacitors. [23][24][25][26][27][28][29][30] However, these reports have been limited to titanium MXenes, with none of these techniques applied to the wider MXene family.
In this work, we developed a porous Mo2TiC2 architecture using an amine-assisted pillaring technique ( Figure 1) and obtained the largest BET surface area reported for any Mo-based MXene to-date. We tested the resulting pillared Mo2TiC2 for Li-ion storage, and obtained significantly enhanced electrochemical performance, with superior capacities, rate capabilities and cycling stabilities compared to the non-pillared version. Furthermore, the charge compensation mechanism in Mobased MXenes was investigated for the first time using a combination spectroscopic techniques, including NMR, XPS and Raman. We believe that these studies are crucial for the further development of this class of MXene materials as electrode materials in Li-ion batteries for high-rate and long-life applications. In addition, due to the growing importance of Na-ion research as a promising low-cost alternative to lithium-ion batteries, we also report for the first time results for the use of Mo2TiC2 in Na-ion half-cells.

MAX phase and MXene synthesis
The Mo2TiAlC2 MAX phase was synthesised following previously reported methods, with details given in the experimental section. 16,31 The powder X-ray diffraction (PXRD) data matches previous reports, showing the successful synthesis of Mo2TiAlC2 ( Figure 2). 16 To avoid the handling of HF, the synthesis of Mo2TiC2 was done for the first time using an adapted version of the LiF-HCl method which has been successfully used for titanium-based MXenes. 32 Details of this can be found in the experimental section. PXRD data are in agreement with previously reported diffraction data for Mo2TiC2, 16 demonstrating that the Mo2TiC2 MXene was successfully etched using the described method ( Figure 2a). The (002) diffraction peak of Mo2TiC2 has shifted to a lower angle compared to the MAX phase (to ca. 7° 2θ), and has increased in intensity, as expected from the formation of a MXene phase. 33 A small peak at 9.5˚ 2θ, which corresponds to the (002) diffraction peak of Mo2TiAlC2, indicates that a minimal impurity corresponding to the MAX phase remains. Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) were also carried out to further confirm the formation of the Mo2TiC2 MXene. SEM micrographs show the layered morphology typical of an MXene in the etched material, with some opening up between the layers also visible 6 ( Figure 2b). The flakes have lateral dimensions in the range of 1-10 μm, with thickness between 1 and 5 μm, which are similar in size to other reported MXenes, including Mo2TiC2. 16 EDS analysis shows that the MXene-like flakes contain no Al, demonstrating successful etching of the MAX phase and that Mo2TiC2 is terminated with -O and / or -OH groups (9.3 wt.%) and -F groups (1.5 wt.%), akin to its Ti3C2 counterpart ( Figure S1). The MXene has a formula of Mo2TiC2O1.75F0.25, which is consistent with the literature. 16 Having confirmed the successful synthesis of the Mo2TiC2, we then applied our previously reported amine-assisted SiO2 pillaring method to this MXene (see experimental section). 34 Throughout the paper, we refer to Mo2TiC2-Si as the pillared sample intercalated with dodecylamine (DDA) and tetraethyl orthosilicate (TEOS) and Mo2TiC2-Si-400 as the pillared sample after calcination at 400 °C  (Table S1). 19,[35][36][37] In addition to the peaks around 2.2˚ 2θ, a small diffraction peak at 4.9˚ 2θ, corresponding to a d-spacing of 1.8 nm, is also visible, which is assigned as the (004) diffraction peak, further confirming the enlarged interlayer distance. After calcination (Mo2TiC2-Si-400), the (002) diffraction peak shifts to a higher angle to 4.5˚ 2θ (Figure 2c and 2d), corresponding to a d-spacing of around 2 nm, which gives a gallery height of 1 nm. The shift suggests that the DDA template is successfully removed, which is also supported by the loss of peaks corresponding to DDA in the Raman spectra ( Figure S3). SEM studies ( BET analysis using a nitrogen isotherm at 77 K showed that the pillaring procedure resulted in a substantial increase in BET specific surface area, with the pillared Mo2TiC2 MXene obtaining a surface area of 202 m 2 g -1 compared to around 8 m 2 g -1 for the non-pillared material (Figure 3e). This is one of the largest surface areas reported for any MXene, and the largest for a non-Ti based MXene. 12,[38][39][40][41] Additionally, this is also larger than previously reported results for pillared MoS2 (Table S1). 37  adsorption-desorption curves at 77 K for the pillared and non-pillared Mo2TiC2.
To further investigate the structure of the synthesised Mo2TiC2 and the effect of the pillaring process on the MXene, X-ray photoelectron spectroscopy (XPS) was used to study the Mo, Ti and O valence states in the as-made and pillared MXene, and Si was also studied in the pillared material. Figure 4 shows the Mo 3d XPS spectra for the non-pillared and pillared Mo2TiC2. There are three main peaks MXenes, Mo2TiC2 also undergoes a slight surface oxidation either during the etching process or when exposed to ambient conditions. The Mo 6+ 3d5/2 electrons contribute to the peak centred around 233.0 eV, with an expected binding energy of 232.7 eV. The presence of a small amount of these surface oxides has also been reported previously for Mo2TiC2, where HF etching was used. 43 The surface oxide peaks do not appear to grow after pillaring and calcination, suggesting that the Ar atmosphere during calcination was sufficient to avoid extra MXene oxidation. Ti 3d XPS spectra (Figure 4c and S5) follows the same pattern, with 3+ being the dominant oxidation state and some Ti 4+ oxides also being present. XPS results for the O 1s scans are shown in Figure 4d and 4e, where clear differences between the non-pillared and pillared Mo2TiC2 can be observed. The spectrum for the non-pillared Mo2TiC2 MXene matches previous reports and shows a very broad asymmetric peak which is comprised of a variety of components as a result of multiple oxygen-containing species being present in the sample. 17 Deconvolution reveals a component centred around 530.5 eV, which corresponds to the formation of Mo and Ti oxides, supporting the Mo 3d and Ti 2p spectra. 17 The peak at 531.0 eV corresponds to Mo-O groups, while the peak around 532.0 eV reveals the presence of Mo-OH termination groups. 17 At 534.0 eV there is a small component which corresponds to surface-bound H2O molecules. 17 After pillaring and calcination, there is a significant new broad peak centred around 533.2 eV, which is a result of oxygen in silica -the pillar. 44 There is a substantial decrease in the component relating to -OH surface groups. Before pillaring, the OH:O ratio is approximately 2:1, as can be seen in 4d, but decreases significantly to 1:1 after pillaring (4e), demonstrating the direct involvement of the -OH groups in the pillaring process, as we have reported in our previous work for Ti3C2. 34 Finally, the Si 2p XPS spectra for Mo2TiC2-Si-400 ( Figure 4f) shows a broad peak at 103.9 eV, which is consistent with SiO2 being the pillar. 44 Overall, these results imply that the amine-assisted pillaring method is unaffected by the change in metal in the surface layer of the MXene, and is directly applicable to Mo-based MXenes. This suggests that the pillaring method can be applied to other types of MXenes, so long as -OH surface groups are present.

Electrochemical Testing in a Li-ion battery
To demonstrate the advantages of the pillared structure, the Mo2TiC2 materials were tested in Li-ion half-cells against Li metal, which acts as a counter and reference electrode.  (Table S2). 16,18 The low coulombic efficiency of ca. 66% in the first cycle is commonly observed in MXenes and is attributed to SEI formation and irreversible reactions between surface groups and Li + ions. 45,46 The coulombic efficiency in the second cycle is 94% and this reaches ca. 99% after 18 cycles in both MXenes. Around 80% capacity is retained between the 2 nd (316 mAh g -1 ) and 94 th (250 mAh g -1 ) cycles in the pillared MXene ( Figure 5d). By contrast, 54% capacity retention under the same conditions was observed for the as-made Mo2TiC2. Chen et al. Rate capability tests were carried out at increasing rates of 20, 50, 200, 500, and 1000 mA g -1 with five discharge-charge cycles at each rate ( Figure 5c). The pillared material shows superior performance at all rates, delivering discharge capacities of 312, 281, 229, 182 and 143 mAh g -1 , respectively. When the current was returned to 20 mA g -1 , the capacity was recovered to 292 mAh g -1 . In comparison, the nonpillared Mo2TiC2 material delivered capacities of 205, 162, 108, 79 and 59 mAh g -1 at the respective rates, with 172 mAh g -1 recovered at 20 mA g -1 . Notably, the enhancement in capacity between the pillared and non-pillared MXene increases with rate, with the pillared MXene delivering capacities 10 around 1.7 times greater than for the non-pillared MXene at 20 mA g -1 , and around 2.5 times greater than for the non-pillared MXene at 1 A g -1 . This demonstrates that the increased interlayer spacing afforded by the pillaring enables fast Li-ion transport, resulting in superior capacities at higher rates. Since the pillared MXene showed impressive capacity at high rates, its high-rate cycling stability was then tested by continuous galvanostatic cycling at 1 A g -1 (corresponding to a charging/discharging time of 8 min) after the rate capability test. After 500 cycles at 1 A g -1 , it retained a capacity of 108 mAh g -1 , a capacity retention of 80% compared to the 1 st cycle (135 mAh g -1 ), (Figure 5e). The average coulombic efficiency over these cycles was close to 100%, indicating highly reversible charge storage at this rate. This shows that Mo2TiC2-Si-400 is a very stable electrode, making it highly suitable for high-power and long-life batteries. In addition, a comprehensive comparison of our work with other Mo-based MXenes for lithium-ion battery applications demonstrates the superior performance of our pillared Mo2TiC2 ( Figure 6 and Table S2). The load curves for both materials (i.e. pillared and non-pillared) display very similar features and are markedly different from titanium-based MXenes, which typically display very linear profiles. 48 The Mo2TiC2 load curves display two clear regions on discharge after the first cycle, the first of which is between 3 and 0.6 V, which slopes with a linear profile and a second region between 0.6 and 0.01 V where there is a sloping plateau feature appearing, demonstrating that a different charge storage mechanism operates in this region. Closer inspection of the load curves reveals that the majority of the capacity is stored in the region below 0.6 V (215 mAh g -1 for the pillared material on the 2 nd cycle), but that the capacity fade also occurs mostly in that region (160 mAh g -1 , 73%, is retained after 94 cycles). This capacity loss is even more dramatic in the non-pillared MXene, where the sloping plateau feature is substantially reduced during cycling to just 90 mAh g -1 after 94 th cycles (53% retention).
Furthermore, differential dQ dV -1 plots of both non-pillared and pillared samples ( Figures S7 and S8) show a peak at 0.6 V, which rapidly decreases with cycling, confirming that this low voltage process contributes significant to the capacity during the initial cycles but it is a significant cause of capacity fade over prolonged cycling.  16 Li2O is known to be a poor electrical conductor, and it is possible that its formation, which in transition metal oxide electrodes is often poorly reversible and accompanied by a large volume change, is the main cause of the capacity fade seen in these electrodes, with several bulk oxides reporting capacity retentions between 40-50% over 100 cycles. 49 Therefore, we used a series of ex-situ spectroscopic studies to experimentally validate the charge storage mechanism for the first time.
To study the reactivity of the SiO2 pillars, we combined 29 Si NMR and Si 2p XPS data to investigate any potential lithiation of the pillars, both in the bulk and near the surface of the pillared MXene. 29 Si solidstate NMR (Figure 7c) revealed that there is only one Si environment present within the pristine pillared MXene structure, which matches well with the expected chemical shift for SiO2 (i.e. -108 ppm). 50 There is no change in the 29 Si NMR environment after discharge or subsequent charge, confirming that the SiO2 pillars are stable during cycling, and that no alloying reaction occurs. This is supported by ex-situ Si 2p XPS (Figure 7d), which also shows no significant changes in the spectra at different states-of-charge, suggesting no difference in redox activity between bulk and near-surface pillars. Additionally, no signal corresponding to LixSiy alloys (expected between 20 and 10 ppm) could be distinguished in the 7 Li NMR spectra ( Figure S9). 50,51 Significantly, this means that the improved electrochemical performance seen in the pillared Mo2TiC2 is a result of the enlarged interlayer spacing, and not due to the lithiation of SiO2.
Ex-situ O 1s XPS analysis was used to investigate the proposed mechanism of reversible Li2O formation in Mo2TiC2 (Figure 7b). At OCV there is just one broad peak visible, which is centred on 532.0 eV.
Deconvolution reveals that there are two main components to this peak at 531.7 eV (assigned to the carbonate oxygen from the electrolyte) 52 and 530.6 eV (which matches the Mo-Ox environment present in the powdered material in Figure 4). There is little change in these two peaks upon cycling, 13 with the shifts varying less than 0.2 eV at all states-of-charge. However, the spectrum for the electrode discharged to 0.01 V shows a new peak at 528.6 eV, which matches well with Li2O. 44 Upon charging to 3 V, this peak disappears, confirming that Li2O is reversibly formed and removed upon lithiation (discharging) and delithiation (charging) in the pillared material. with the MXene sheets shown in blue and the pillars in orange. b) Ex-situ O 1s XPS spectra. c) Ex-situ 29 Si MAS NMR spectra. d) Ex-situ Si 2p XPS spectra.
Ex-situ solid-state NMR was then used to study the evolution of the local structure of the H, Li, F environments within the bulk of the pillared MXene upon cycling. 1 H NMR showed a broad peak present in the pristine sample, which is assigned to 1 H environments in the polyvinylidene fluoride (PVDF) binder ( Figure S10). 53 After discharge to 0.01 V, new peaks appear at 3 and 4 ppm, which correspond to carbonate environments originating from retained electrolyte solvent molecules in the material. 53 In addition, a new peak at -1.5 ppm is now present, which can clearly be assigned as LiOH based on previous reports. 53 This peak disappears after charging to 3 V, suggesting that, like Li2O, the formation of LiOH is reversible, which implies that LiOH is also a discharge product.
Overall, these results support the mechanism proposed by Anasori et al (Reactions 1 and 2), 16 but imply that an additional reaction involving the formation of LiOH, most likely via a conversion reaction with terminal -OH groups, may also occur (Reaction 3).
Mo2TiC2(OH)2 + 2 Li + + 2 e -→ Mo2TiC2 + 2 LiOH This reaction could explain the second low voltage discharge peak observed in the dQ dV -1 plots for both the non-pillared and pillared Mo2TiC2 (Figures S7 and S8). An analogous conversion reaction involving the formation of LiOH has been previously observed on RuO2. 53  suggesting the cleavage of these bonds. 18 It should be noted that some partial re-oxidation is expected, which may explain the continued presence of the 260 cm -1 Raman mode. However, LiOH is also known to have a minor Raman mode at 270 cm -1 , 56,57 and thus, it could be contributing to the 260 cm -1 peak. LiOH is also known to have a major Raman mode at 320 cm -1 , which could explain the increase in intensity in this region after discharge. 56,57 However, these modes cannot be conclusively assigned due to the overlap with potential MXene peaks. After discharge, a new mode at ca. 550 cm -1 matches previous reports for Li2O, supporting the formation of Li2O as a discharge product. 56,57 Therefore, these results could support the proposed conversion reaction mechanism whereby the -O and -OH surface functional groups of MoTiC2 MXene react with Li to form Li2O and LiOH respectively.
Crucially, the Raman spectra confirm that the MXene bonding framework between Mo, C and Ti is unchanged upon cycling despite the conversion reactions unlike with transition metal oxides. 49 This should ensure superior reversibility during the lithiation process of Mo-based MXenes compared to the metal oxides. 7 Li NMR ( Figure S12) shows a broad asymmetric peak centred at -0.6 ppm. Deconvolution of this peak using a Lorentzian profile shape for the pristine sample reveals two main environments, the largest of which is assigned as pre-intercalated Li as a result of the LiF-HCl etching method used. 13 A minor peak to the right of this, centred around -1.1 ppm, can be assigned as LiF from the etching stage, which is confirmed as being present in the structure by 19 F NMR and 7 Li-19 F HETCOR NMR ( Figure S13). 54 Upon discharge to 0.01 V, the broad peak shifts to a higher chemical shift (centred at -0.1 ppm), with asymmetry now present on the left side of the peak. This reveals the existence of new Li environments, which are likely to be Li2O and LiOH based on the previously discussed XPS (Figure 7b) and 1 H NMR ( Figure S10) data and the relative shifts compared to LiF (3-4 ppm higher than the LiF component). 55 However, the broad profile of the signal, resulting from the slightly disordered nature of MXenes and the lack of separation between the 7 Li chemical shift of the different environments, 54,55 means that the environments cannot be unambiguously distinguished in the broad spectra obtained using 7 Li NMR. Nevertheless, the 7 Li NMR spectra appear to support the XPS and 1 H NMR results, confirming the contribution of conversion reactions to the charge storage mechanism. After charging, these changes are reversed, which demonstrates the reversible (de)lithiation of the Mo2TiC2 MXene.
Cyclic voltammetry was used to investigate the reactions and kinetics of the system in more detail.  Figure 5). 45,46 After the first cycle, the shape of the CV plots do not notably change, apart from a clear decrease in the current below 0.6 V for the non-pillared material, demonstrating the greater fade observed in this material compared to the pillared MXene, which agrees well with the GCD tests ( Figure 5). The large peak below 0.6 V on discharge and the clear peak at 1.3 V on charge also match well with the plateaus observed on the load curves ( Figure 5) and dQ dV -1 plots ( Figure S8) and with previous reports on Mo2TiC2. 16 To investigate the kinetics of the system in more detail, the cells were then cycled at increasing scan rates of 0.5, 2 and 5 mV s -1 ( Figure S15). As the scan rate increases, both materials show no major changes, with only broadening and small shifts of their redox peaks to lower voltages, suggesting that the majority of the redox reactions are kinetically favoured. The voltammograms ran at 5 mV s -1 is much more rectangular in shape for the Mo2TiC2-Si-400 compared to the non-pillared Mo2TiC2, with increased current above 1 V. This is indicative of a greater contribution from capacitive charge storage as a result of the higher interlayer spacing and surface area in the pillared MXene, and explains the enhanced high-rate performance of this material. This is further supported by analysing the proportion of diffusion-limited (battery-like) and surfacelimited (capacitive-like) processes to the overall current. It is well known that the relationship between current and scan rate is proportional to the power half when current is diffusion-limited, whereas the relationship is linear (power of 1) when current is surface-limited. 58 This allows the formation of a simple power law to determine the proportion of current arising from diffusion or surface limited processes in a mixed mechanism system, as shown by Equation 4, where i is current, v is scan rate and a and b are fitting parameters. 58 Plotting the log of the current against the log of the scan rate gives a straight line with a gradient of b, allowing the proportion of diffusion (b= 0.5) or surface (b= 1) limited current to be quantified.
When this analysis is carried out at different voltages, the relative contribution of these processes can be studied across the voltage window on the charge and discharge sweeps ( Figure S15c and S15d). For both materials, the b-values are much closer to 1 (capacitive current) at higher voltages, but much closer to 0.5 (diffusion-limited battery-like processes) at low voltages (below 0.6 V). This is expected from the CV shapes, which are more rectangular at higher voltages, with prominent redox peaks present at voltages below 0.6 V. At higher voltages, the pillared Mo2TiC2 has a higher b-value (i.e 0.86) than the non-pillared MXene (0.6-0.8) indicting an increased capacitive contribution to the charge storage resulting from the larger surface area, as previously discussed. In contrast, at low voltages, for example, 0.01 V, the pillared Mo2TiC2 has a lower b-value (0.56) than the non-pillared Mo2TiC2 (0.68), which indicates a greater contribution from battery-like processes at these voltages. This suggests that pillaring leads to increased charge storage contribution from the Li2O conversion reaction, which explains the substantial increases in capacity compared to the non-pillared MXene.

Electrochemical performance in a sodium-ion battery
Following the promising performance of the Mo2TiC2 MXene in a Li-ion system, the non-pillared and pillared materials were further tested as Na-ion electrodes in a Na half-cell. Mo2TiC2 has so far not been reported as an electrode for Na-ion batteries. It can be seen that pillaring substantially improves the electrochemical performance, with a 2 nd cycle discharge capacity of 109 mAh g -1 compared to 74 mAh g -1 for the non-pillared material (Figure 8). By the 80 th cycle, the non-pillared MXene had retained a capacity of just 48 mAh g -1 (65% capacity retention compared to the 2 nd cycle) compared to 82 mAh g -1 for the pillared MXene (75% capacity retention). At higher rates (Figure 8d), the pillared MXene has superior capacities compared to the non-pillared material at each rate studied, retaining 40 mAh g -1 at 1 A g -1 , compared to 16 mAh g -1 for the non-pillared MXene. These capacities are much lower than what was observed for the Li-ion system and correspond to a discharge product of approximately Mo2TiC2O2Na (theoretical capacity = 90 mAh g -1 ), suggesting insertion of one mole of Na per formula unit, even in the pillared MXene. Unlike in the Li-ion system, differential dQ dV -1 plots show no peaks which would correspond to a conversion reaction for either the pillared or non-pillared Mo2TiC2 ( Figure S16). This suggests that, by analogy to the Li-ion system, Na2O does not form in the Na-ion system, which is in agreement with the DFT studies of Anasori et al. 16 These findings explain the differences in the behaviour of Mo2TiC2 in the Li and Na-ion systems. In the Li-ion system, at low voltages the conversion reaction to Li2O provides a large amount of extra capacity. In contrast, no such conversion reaction occurs in the Na-ion systems, which leads to improved cycling stability (especially in the non-pillared materials), but with capacities around a third of the Li-ion system, even when pillared. CV analysis at different rates reveals that a much greater proportion of the current is surface controlled (capacitive) compared to the Li-ion system (Figure S17), 18 for example Mo2TiC2-Si-400 has a b-value of 0.73 at 0.01 V in the Na-ion system, but 0.56 in the Li-ion system at the same voltage. In the Na system, at voltages above 1 V, the b-value for the pillared material is around 0.9, showing that capacitive contributions dominate at higher voltages. This leads to reasonable rate capability (Figure 8d), with over 40 mAh g -1 being retained even at the high rate of 1 A g -1 , which is higher than any other report for Mo-based MXenes in Na-ion systems (Table S3).

Conclusions
Overall, we demonstrate the application of an amine-assisted pillaring method to create porous Mo2TiC2. This leads to a large increase in interlayer spacing, achieving d-spacings up to 4.2 nm. This corresponds to a gallery height (pore size between layers) of around 3 nm before calcination, which is by far the largest for a Mo-based MXene, and larger than any reports found for other Mo-based layered materials such as MoS2. This suggests that the amine-assisted silica pillaring method could be applied to a wide range of MXenes, and perhaps other layered materials, as long as there are sufficient -OH groups present on the surface to bind to the amine. Calcination removes the DDA template and reduces the gallery height to a still expanded 0.75 nm in the final pillared material.
When tested as the negative electrode in a lithium-ion battery, the pillared material showed significantly improved electrochemical performance with respect to the non-pillared material, reaching capacities of up to 316 mAh g -1 (89% of the reported theoretical capacity, i.e. 356 mAh g -1 ) based on two moles of Li + intercalating per formula unit followed by two further moles of Li undergoing the conversion reaction), which is the highest capacity reported so far for Mo2TiC2.
Pillaring not only increases the capacity of the MXene, but also the cycling stability by providing free space for the reversible formation of Li2O and extra intercalation of Li ions without allowing layers to restack. In addition, the rate capability is also significantly enhanced, since the enlarged interlayer spacing aids the Li + diffusion to the MXene active site. The pillared material shows high coulombic efficiency (ca. 100%) and good stability even at a high rate of 1 A g -1 (under 8 min charge time) returning 80% of an initial capacity of 135 mAh g -1 after 500 cycles. The reversible formation of Li2O during cycling was confirmed by ex-situ NMR and XPS studies, while ex-situ NMR suggests that SiO2 does not undergo redox activity during cycling. Therefore, the enhanced electrochemical performance may be ascribed to the enlarged interlayer spacing, which offers the dual benefit of increasing the available sites for Li intercalation and LiO2 formation, while significantly improving the structural stability during cycling. A potential parallel conversion reaction forming LiOH from -OH surface groups was implied by ex-situ 1 H NMR and Raman results, but requires further detailed mechanistic studies to confirm the origin of any LiOH formed.

19
In a Na-ion system, the capacities of the non-pillared and pillared Mo2TiC2 MXene were up to a third lower than for the Li-ion system, which could be explained by the charge storage relying on Na + intercalation and capacitance, with no conversion reaction occurring. Nevertheless, the pillared material had superior performance compared to the non-pillared MXene, delivering reversible capacities up to 109 mAh g -1 at 20 mA g -1 over twice that of the non-pillared MXene.

Materials Synthesis and Pillaring
For the synthesis of Mo2TiAlC2, Mo (-325 mesh, 98% purity, Sigma Aldrich), Ti (-325 mesh, 99% purity, Alfa Aesar), Al (-100+325 mesh, 99.5% purity, Alfa Aesar), and C (graphite, <20 µm, 99% purity, Sigma Aldrich) powders were mixed in a pestle and mortar in a 2:1:1.1:2 molar ratio. The mixture was then heated in a tube furnace under flowing argon at 1600 °C for 4 h, with a heating rate of 5 °C min -1 . The resulting block was then crushed with a pestle and mortar and ground to give a fine grey powder.
Typically, 3 g of Mo2TiAlC2 were slowly added to 30 ml of 9 M HCl with 3 g of pre-dissolved LiF. The mixture was heated to 60 ˚C and stirred for 5 days. The powder was recovered by centrifuging cycles, with DI water added after each cycle until the pH ≈6. The sample was then analysed by PXRD, which showed that significant amounts of unetched MAX phase remained in the sample. Therefore, the partially etched sample was re-dispersed in a fresh etching solution using the same volumes and concentrations used previously. After four days, the solid was collected via centrifuging, using the same protocol as described above. A washing step, where the powder was dispersed in 1 M HCl for 3 h at ambient temperature, was used to remove any salt impurities resulting from the etching step. A NaOH washing step was also attempted, but this caused the dissolution of the majority of the powder in approximately 30

Material Characterisation
Powder X-ray diffraction (PXRD) was carried out in a Smartlab diffractometer with a 9 kW rotating anode (Rigaku, Tokyo, Japan) using Cu Kα radiation (wavelength of 1.54051 Å) operating in reflection mode with Bragg-Brentano geometry. Prior to PXRD characterisation, all samples were dried in a heated oven at 80 °C for 18 h. The powders were then ground and placed on a silica sample holder and pressed flat with a glass slide.
Scanning electron microscopy (SEM) was performed in a JEOL JSM-7800F (JEOL, Tokyo, Japan), and energy-dispersive X-ray spectroscopy (EDS) was carried out using an X-Max50 (Oxford Instruments, Abingdon, UK) with an accelerating voltage of 10 kV and a working distance of 10 mm. The dried powder samples were dry cast onto a carbon tape support, which was placed on to a copper stub for analysis.
Gas sorption isotherms were measured on a Micromeritics 3 Flex 3500 gas sorption analyser using high purity nitrogen gas at 77 K. BET surface areas were calculated over a relative pressure range of 0.05-0.15 P/P0. Pore size distribution analysis was calculated using the NLDFT (non-linear density functional theory) method with a slit pore model using 3Flex Micromeritics software.
Raman spectroscopy was carried out on a Horiba Lab Raman Spectrometer (Horiba, Minami-ku Kyoto, Japan) with an EM-cooled Synapse camera. Spectra were collected using a 100x, 0.90 NA microscope objective. For each measurement, three scans were collected, with a total measurement time of 30 min. The dried powder was sandwiched between two glass microscope slides which were pressed together to give flat MXene particles. One of these slides was then discarded, with the other slide placed flat under the diode green laser (532 nm, 200 μW, 1% intensity) for measurements. For ex-situ measurements, the extracted discharged electrodes were washed with dimethyl carbonate (DMC), dried in the anti-chamber of an argon-filled glovebox, and transported to the spectrometer in a sealed container. The spectra were collected under air.

Electrochemical characterisation
Pillared and non-pillared Mo2TiC2 were tested in coin cells (CR2032 type) in a half-cell configuration using lithium or sodium metal (Tob Energy, China) disks as the counter and reference electrodes and 1 M LiPF6 or NaPF6 in EC/DEC (1:1 weight ratio) as the electrolyte. The MXene (active material) was 21 mixed with carbon black (super P) as a conductive additive and PVDF as the binder in a 75:15:10 weight ratio respectively. The mixture was added to a few ml of NMP to make a slurry, which was then cast onto a Cu foil used as current collector, from which electrodes with a diameter of 16 mm were punched. The active mass loading of each electrode was ca. 3.2 mg cm -2 . Coin cells were constructed in an argon-filled glovebox (O2 and H2O levels < 0.1 ppm) using Whatman micro glass fibre paper as the separator. Galvanostatic tests were carried out on a Neware battery cycler (Neware Technology Ltd., China) at a current density of 20 mA g -1 in the potential range of 0.01-3 V vs. Li + /Li for 94 cycles for the tests in the Li-ion half-cells and in the potential range of 0.01-3 V vs. Na + /Na for 80 cycles for the tests in the Na-ion half-cells. For rate capability tests, the cells were cycled at a current density of 20 mA g -1 for 1 cycle to stabilise the cell before 5 cycles were run at each current density of 20, 50, 200, 500 and 1000 mA g -1 before returning to 20 mA g -1 . For the long-term high-rate cycling test on the pillared MXene, 500 cycles were run at a rate of 1 A g -1 after the rate capability test. Cyclic voltammetry (CV) measurements were conducted using an Ivium potentiostat (Ivium Technologies BV, The Netherlands) with increasing scan rates of 0.2, 0.5, 2 and 5 mV s -1 for 2 cycles at each rate in the potential range of 0.01-3 V vs. Li + /Li and Na + /Na for the tests in Li-ion and Na-ion half-cells, respectively. In each case the final cycle at each scan rate was used to calculate the b-values. Ex-situ Solid-state Nuclear Magnetic Resonance Spectroscopy (NMR). 19 7 Li heteronuclear correlation spectrum was obtained using a cross-polarisation based sequence with the contact pulse ramped for 19 F. This spectrum was used as an internal reference for the 7