To achieve staggered arrangements of metal atoms, we set BAC-BC stacking Ti₃C₂T_x/Nb₂CT_x heterostructures as the initial configurations [Supplementary Figure 5]. We then performed a full-period sliding along both the AM and ZZ directions to search for the most stable structures and to track the variation of system energy during sliding. We considered three terminations (-F, -O and -OH) to thoroughly examine the interfacial structural and electronic properties^[56,57]. To further understand the interlayer interaction, three layer-ratios between Ti₃C₂T_x and Nb₂CT_x (1:1, 1:3, 3:1) were investigated. In total, we obtained 9 basic configurations, containing 3 surface terminations and 3 layer-ratios, and studied sliding along both the AM and ZZ directions.

Hexagonal 2D crystals exhibit anisotropy in their atomic densities and periodic lengths along the AM and ZZ directions^[58-60]. To investigate this phenomenon, Figure 2 shows the evolution of Ti₃C₂T_x/Nb₂CT_x heterostructures as they slide along the AM or ZZ directions.

The upper panels of Figure 2 show the atomic configurations before relaxation. To identify the relative positions of the two layers after relaxation, Supplementary Figure 5 highlights the metal atoms in models with minimum and maximum energies. During sliding, the ground-state configurations (GCs), among the calculated models, are predominantly found at AM₀ZZ₀ or AM_δ/3ZZ₀, where the metal atoms (Ti & Nb) adopt an FCC stacking arrangement (BAC-BC or BAC-AB). Conversely, the configurations with maximum-energy configurations (MCs) among the calculated models, occur mainly at AM_2δ/3ZZ₀, featuring a twin grain boundary (BAC-CA) that forces surface functional groups to face each other directly, leading to a marked increase in system energy.

To determine the preferred relative positions of the two layers in GCs, Supplementary Table 4 summarizes the positions of interlayer Nb atoms relative to Ti₃C₂T_x. For -O terminated heterostructures, Nb atoms prefer to sit atop the nearest C atoms of Ti₃C₂T_x, whereas for -OH termination, they prefer atop the surface -OH groups. For -F termination, no clear trend is observed. Detailed structural and energetic data are provided in Appendix I of Supplementary Materials.

The middle panels of Figure 2 show the variation of d_inter during sliding. For reference, the d_inter values of the pristine Ti₃C₂T_x and Nb₂CT_x are as follows: for -F termination, 2.22 and 2.20 Å; for -O, 2.44 and 2.37 Å; for -OH, 0.52 and 0.45 Å, respectively. When sliding occurs along the AM direction, d_inter exhibits sharp increases at a displacement of 2δ/3, with increments of approximately 0.55 Å (0.58 Å) for -F, 0.38 Å (0.45 Å) for -O, 1.67 Å (1.75 Å) for -OH relative to the pristine Ti₃C₂T_x (Nb₂CT_x). In contrast, along the ZZ direction, d_inter shows saddle points at δ/2 between two peaks at δ/3 and 2δ/3, but the increase at the peaks is less than 0.2 Å relative to the pristine Ti₃C₂T_x (Nb₂CT_x) for all terminations. Thus, sliding along AM can substantially enlarge d_inter compared with pristine MXenes, whereas sliding along ZZ hardly does so.

The lower panels of Figure 2 show the variation in the system energy during sliding. Along AM, the energy peaks at 2δ/3, with energy barriers of about 80 meV/cell (-F), 100 meV/cell (-O), and 200 meV/cell (-OH). Along ZZ, two similar peaks present at δ/3 and at 2δ/3, with energy barriers of 19-23 meV/cell (-F), 23-27 meV/cell (-O), and 27-37 meV/cell (-OH). Thus, sliding along ZZ is energetically preferred over sliding along AM, but it yields a much smaller increase in d_inter.

Comparing the middle panel and the lower panel of Figure 2 reveal that d_inter and system energy follow similar variation patterns, reaching local extreme at the same sliding positions. To test the generality of this synchronous correlation, we constructed energy contour maps for global sliding [Supplementary Figures 6 and 7], which confirm that GCs always have the smallest d_inter and MCs the largest.

To understand the origin of this correlation, we decomposed the system energy into the energies of the Ti₃C₂T_x, Nb₂CT_x components and their binding energy E_B. Variations in the individual layer energies are negligible comparing with those in E_B [Supplementary Tables 5 and 6]. For the case of 1Ti1Nb, Supplementary Figure 8 also presents the striking similarity between E_B mappings and global energy mappings. This indicates that the system energy variation is dominated by E_B. As shown in Supplementary Figure 9, E_B ranges from 15 to 40 meV/Ų for -F/-O terminated heterostructures, typical of vdW interactions between 2D materials^[61,62]. For -OH terminated heterostructures, hydrogen bonding strengthens the interlayer interaction, raising E_B to 25-55 meV/Ų.

Here we briefly summarize the effects of layer ratio and termination on sliding behavior. As shown in Supplementary Table 7, varying the layer ratio (1:1, 1:3, 3:1) has little effect on either the sliding energy barrier or the maximum interlayer spacing enlargement (Δd_inter). For example, with -O termination, the barrier changes from 103 meV (1Ti1Nb) to 110 meV (1Ti3Nb) and 93 meV (3Ti1Nb), while Δd_inter varies between 0.500 Å and 0.518 Å. Similar insensitivity is observed for -F and -OH terminations. In stark contrast, the termination dictates the sliding behavior: -OH yields barriers about twice as high (~200 meV) and Δd_inter three to four times larger (~1.7-2.0 Å) compared with -F/-O (barriers ~80-110 meV, Δd_inter ~0.5-0.6 Å). This difference arises because hydrogen bonds in –OH systems introduce strong, directional interlayer interactions that are highly sensitive to sliding displacement, whereas vdW interactions in -F/-O systems are weaker and less stacking-dependent. Thus, layer ratio plays a secondary role; tuning the surface termination is the primary lever for controlling sliding energetics and interlayer spacing in MXene heterostructures.

The mainly worked vdW interactions in our MXene heterostructure models include direct electrostatic, induction and dispersion interactions^[63]. To determine which component dominates the energy variation during sliding, we performed three sets of analyses.

First, we calculated the Bader charges and extracted charge transfer between the two layers [Supplementary Figure 10] and between the nearest atomic groups (Nb&T_x and Ti&T_x) [Supplementary Figure 11]. We then evaluated the correlation between charge transfer and system energy (and d_inter) [Supplementary Table 8, Supplementary Figures 12 and 13] to assess the role of direct electrostatic interaction.

Second, we calculated the DOS of GCs and MCs for the whole two layers [Supplementary Figure 14] and for the interlayer atomic groups (Nb&T_x and Ti&T_x) [Supplementary Figure 15], as well as the band structure [Supplementary Figure 16], to evaluate the induction interaction.

Third, we quantified the Pearson coefficients of determination (r) between -1/E_B and d_inter⁶, and between -1/E_B and D_t⁶ [Supplementary Tables 6 and 9], to reveal the role of the dispersion interaction.

We now examine each component in turn. First, Bader charge analysis [Supplementary Figures 10 and 11] shows minimal charge transfer between Nb₂CT_x and Ti₃C₂T_x during sliding (< 0.06 |e|). Similarly, negligible charge transfer (< 0.1 |e|) is observed between interlayer Ti&T_x and Nb&T_x. Quantitatively, ordinary least squares regression [Supplementary Figures 12 and 13] reveals a poor correlation between charge transfer and system energy (or d_inter). These findings contrast with previous studies on MXene heterostructures, where charge transfer played a more important role^[48,61]. Therefore, direct electrostatic interaction hardly contributes to the synchronous correlation.

Second, comparison of DOS and band structures between GCs and MCs [Supplementary Figures 14-16] reveals that all heterostructures are metallic. The differences in DOS between GCs and MCs are remarkably subtle [Supplementary Table 10, Supplementary Figures 14 and 15], both for the total DOS of individual MXene layers and for the local DOS (LDOS) of interlayer atoms (Ti&T_x and Nb&T_x), indicating minimal changes in electronic structure. The projected band for interlayer atoms [Supplementary Figure 16] similarly shows negligible difference between GCs and MCs. Furthermore, the distinct peaks in the DOS plots confirm limited electronic interaction between the two layers, consistent with the Bader charge analysis. Thus, induction interaction also barely contributes to the synchronous correlation.

Finally, the Pearson coefficients of determination (r) between -1/E_B and d_inter⁶ are all close to 1 (> 0.96 for -F/-O models, > 0.94 for -OH models) with p-value < 1E-25, whereas the r values between -1/E_B and D_t⁶ show no clear trend. We therefore conclude that the interlayer dispersion interaction plays a crucial role in the synchronous trend between system energy and d_inter.

To further examine the role of the dispersion interaction, we calculated the distribution of D_t [Supplementary Figures 17-19]. For -F/-O terminations, D_t remains close to its average value across all layer ratios, with a deviation of less than 3.3%. The average D_t values are about 2.8 Å and 2.9 Å, which are roughly twice the vdW radii of F/O^[64]. For -OH termination, D_t between interlayer atoms, including -H···-H, -O···-O and -H···-O, varies by less than 10%, except in some models with distorted -OH groups, likely due to intermolecular hydrogen bonding. Notably, for the O atoms in -OH groups, the average D_t (-O···-O) is about 3.0 Å, also close to twice the vdW radii of O^[64].

Although D_t varies by less than 5% for -F/-O terminations, d_inter can increase by up to 25% relative to its minimum [Supplementary Figure 20]. Specifically, the actual increases in d_inter are about 0.6 Å for -F and 0.5 Å for -O terminated heterostructures, respectively. For -OH termination, about 30% increase in D_t can lead to more than 300% increase (about 2.0 Å) in d_inter. We term this behavior the “decoupled interlayer correlation”, and illustrate it in Supplementary Figure 21: during sliding, Nb₂CT_x (or Ti₃C₂T_x) layer moves not only in-plane but also along c-axis, resulting in a larger d_inter while keeping D_t nearly constant. This decoupling between d_inter and D_t is the physical origin of the synchronous correlation between d_inter and E_B. For –OH terminated systems, directional hydrogen bonding alters the interaction mechanism; D_t is no longer constant (variation > 10%), so the “decoupled interlayer correlation” is not strictly applicable.

We hypothesize that the nearly constant D_t minimizes the perturbation of the interfacial electronic states during sliding, as evidenced by the almost unchanged DOS and Bader charge. This “decoupled interlayer correlation”, dominated by isotropic London dispersion interactions, ensures that the local chemical environment and electronic coupling across the interface remain largely unchanged even when the macroscopic layer spacing is manipulated by sliding. Thus, the nearly constant local coordination preserves charge transfer characteristics, explaining why the Bader charge transfer remains stable but correlates poorly with sliding energy or d_inter.

To assess the feasibility of achieving such enlarged d_inter, we mapped the full-period variation in system energy [Supplementary Figure 22]. The sliding barriers are typically ~100 meV/cell for -O and -F terminated heterostructures, and 200 meV/cell for -OH termination. For context, previously reported sliding barriers for 2D materials range from 0.0045 meV/atom to 15 meV/atom in experiments and theoretical simulations^[65-69]. In our study, the global sliding barriers for -O and -F terminated heterostructures range from 3.071 meV/atom to 5.021 meV/atom, which are likely surmountable under practical conditions. For -OH terminated heterostructures, specifically 3Ti1Nb (3 layers of Ti₃C₂T_x + 1 layer of Nb₂CT_x), 1Ti3Nb (1 layer of Ti₃C₂T_x + 3 layers of Nb₂CT_x) and 1Ti1Nb, the barriers are higher (5.38, 6.57 and 12.545 meV/atom, respectively), making sliding somewhat more challenging.

In summary, the “decoupled interlayer correlation”, arising from interatomic and interlayer vdW interactions, enables the enlargement of d_inter while keeping the distances between interlayer functional groups nearly constant. This effect, combined with the metallic conductivity of MXene heterostructures, positions Ti₃C₂T_x/Nb₂CT_x heterostructures as a novel class of metallic layered materials with low sliding energy barrier, expanding the family of 2D electronic materials.

The increase in interlayer spacing achieved at low energy cost can facilitate the intercalation and deintercalation of alkali cations in battery applications. Here, we investigate the potential of Ti₃C₂T_x/Nb₂CT_x 2D heterostructures for energy storage by focusing on Li/Na intercalation, due to the low cost and high energy density of these alkali metals^[15,70]. Specifically, we calculated binding energies per Li/Na atom and examined structural properties for 1L and 2L Li/Na intercalation, using both GC and MC heterostructures as hosts to explore the effect of enlarged d_inter. Three distinct high-symmetry sites in the interlayer were considered: atop terminations, atop C and atop Ti atoms of Ti₃C₂T_x^[71], corresponding to the sites A, B and C respectively. The choice of up to 2L Li/Na intercalation follows established stoichiometry for pure MXenes^[55,71,72].

Supplementary Table 11 summarizes the characteristics of the optimized heterostructures. Consistent with experimental observations of Na intercalation in Ti₃C₂T_x monolayers^[72], both the first and second intercalated Li/Na atoms tend to reside directly above the nearest C atoms of either Ti₃C₂T_x or Nb₂CT_x. Also, in line with previous studies on pure MXenes, Li/Na adsorption in -F terminated MXenes leads to local geometric distortion that curtails capacity, primarily due to the formation of metal fluorides^[71,73].

Figure 3 represents the binding energies and structural changes for 1L Li/Na intercalation in GCs, where Li/Na atoms are located at high-symmetry sites A, B and C as illustrated in Figure 3A. Data are extracted from GCs among the three intercalation sites (A, B and C). Particular attention is paid to the lattice volume change (ΔV), thickness change (ΔZ) and area change (ΔS), relative to heterostructures without intercalation, as these parameters govern the structural response and subsequent battery performance.

-O terminated heterostructures exhibit the strongest binding with Li/Na, followed by -F and then -OH [Figure 3B and F]. For -O termination, E_B values are approximately -3.11 eV/Li (1Ti1Nb), -3.07 eV/Li (1Ti3Nb) and -3.17 eV/Li (3Ti1Nb); for Na, they are approximately -2.71, -2.69 and -2.76 eV/Na, respectively. Reported binding (or adsorption) energies for Li in MXene-based layered structures range from 0.146 to -3.573 eV/Li^[74-78], and for Na, from 1.36 to -2.74 eV/Na^[75,77]. The strong binding observed here thus helps Li/Na clustering, enhancing safety and reversibility. This trend aligns with previous reports that -O functional groups are the most favorable because -O requires two electrons to bond with Li/Na^[44]; whereas -F and -OH requires only one; additionally, the H+ in -OH introduces a repulsive force on Li/Na, resulting in less favorable binding energies^{[71,74,79-81]}. For -F terminated heterostructures, E_B values are around -2.35 eV/Li (1Ti1Nb), -1.91 eV/Li (1Ti3Nb), -2.39 eV/Li (3Ti1Nb), and the E_B of Na are around -1.92 eV/Na (1Ti1Nb), -1.22 eV/Na (1Ti3Nb), -1.90 eV/Na (3Ti1Nb), respectively. While for -OH, the E_B values reach up to -0.41 eV/Li and -0.28 eV/Na, indicating much weaker binding. Although -F gives intermediate binding energies, the tendency to form metal fluorides makes -F terminated heterostructures less suitable.

When layer-ratios changes, E_B varies little, because Li/Na atoms interact primarily with the interlayer atoms of Ti₃C₂T_x and Nb₂CT_x rather than the extra layers in 1Ti3Nb and 3Ti1Nb. This underscores the dominant role of the interface. By the same logic, ΔV, ΔZ and ΔS decrease from 1:1 to 1:3 to 3:1, since Li/Na intercalation only expands the interface region, additional layers add mass without contributing to expansion, so thicker heterostructures expand less.

As shown in Figure 3C-E and 3G-I, structural changes occur predominantly in ΔV and ΔZ, while ΔS remains nearly constant regardless of termination. This indicates that expansion is mainly along the c-axis, reflecting the disparity between strong in-plane ionic bonds and weak out-of-plane vdW forces. For 1L Li intercalation, ΔZ ranges from 2% to 18%, corresponding to a d_inter increase of 0.76-2.97 Å; for 1L Na, ΔZ ranges from 9% to 24%, giving a d_inter increase of 1.32-3.99 Å. In comparison, experimentally reported d_inter increases for Li and Na intercalation in Ti₃C₂T_x layers are 0.80-2.25 Å and 0.25-2.28 Å, respectively^[82]. The larger increment in our work have two origins: first, the heterostructure include a large vacuum, allowing less constrained c-axis expansion than in bulk MXene; second, experimental intercalation may be incomplete (not all interlayers are filled), leading to a smaller measured increment than theoretical prediction.

Results for 1L Li/Na intercalation in MCs are shown in Supplementary Table 12 and Supplementary Figure 23 using same parameters as Figure 3. For -O terminated structures, E_B values are approximately -3.21 eV/Li (1Ti1Nb), -3.09 eV/Li (1Ti3Nb) and -3.16 eV/Li (3Ti1Nb); for Na, -2.83, -2.82 and -2.87 eV/Na, respectively. The difference in E_B between MCs and GCs is negligible, which we attributed to the “decoupled interlayer correlation”. Regarding structural changes, ΔZ for 1L Li intercalation ranges from 1% to 9% (d_inter increase 0.40-0.89 Å), and for Na from 2% to 11% (0.74-1.99 Å); ΔS again remains nearly constant. Notably, ΔV and ΔZ in MCs are generally smaller than those in GCs, and they are halved in some configurations, because the enlarged d_inter in MCs provides more space to accommodate intercalated atoms. Thus, the enlarged spacing arising from “decoupled interlayer correlation” has little effect on E_B but notably reduces volume expansion during intercalation (from 2%-24% to 1%-12%).

For 2L intercalation in GCs [Figure 4], E_B of the 2nd layer Li/Na atoms drops sharply, and ΔZ nearly doubles compared with the 1L case. This sharp decrease in E_B arise from enhanced repulsion between the positively charged alkali ions, which substantially limits the capacity^[74]. Consequently, intercalation becomes increasingly challenging with higher Li/Na loading. Both ΔV and ΔZ roughly double relative to 1L intercalation. As in the 1L case, -O termination remains the most favorable, and ΔS remain negligible. For 2L Li/Na intercalated in MCs [Supplementary Figure 24], similar trends are observed, but the expansion along c-axis decreases markedly due to enlarged d_inter (from 10%-45% in GC cases to 5%-30% in MC cases). Additionally, the E_B for Na drops slightly more, likely due to stronger Na⁺-Na⁺ repulsion. Detailed structural and energetic data are provided in Appendix II of Supplementary Materials.

Figure 5 shows charge transfer for Li/Na intercalated heterostructures. We focus on the -O terminated 1Ti1Nb heterostructure, which possesses the highest specific capacity (120.90 mAh/g) calculated by equation (4), among the studied layer ratios [Supplementary Table 13], because its 1:1 layer-ratio maximizes the active interface mass fraction, and -O termination gives the strongest binding. For 1L intercalation, similar amounts of electrons are transferred for Li and Na. For 2L case, however, Li donates considerably more charge than Na, as evidenced by isosurface areas shown in Figure 5. Electron transfer occurs predominantly from Li/Na to interlayer O atoms and surface Ti/Nb layers.

Bader charge analysis [Supplementary Figure 25] provides quantitative insight. Compared with earlier reports (0.22-0.4 |e| per Li atom)^[71], the Li/Na atoms in our heterostructures donate two to three times more electrons: ~0.9 |e| per atom for 1L intercalation (both Li and Na); For 2L intercalation, Li donates 0.7-0.9 |e| per atom, whereas Na donates only 0.1-0.5 |e| per atom. Thus, going from 1L to 2L, Li donates ~50% more electrons, while Na donation remains similar, indicating superior charge transfer capability of Li over Na. These results confirm strong Coulombic interaction between the alkali atoms and heterostructures, effectively preventing agglomeration.

Supplementary Figures 26-29 summarize the electron transfer from Li/Na atoms to the Ti₃C₂T_x and Nb₂CT_x components in both GCs and MCs for various terminations and layer-ratios. We highlight three key factors:

(1) alkali identity: Li donates more electrons than Na, especially for -OH termination. Na donates electrons more evenly to the two MXene layers. Li also tends to form fluorides with -F, leading to imbalanced electron transfer.

(2) Number of layers: From 1L to 2L, Li donates 50% more electrons, whereas Na donation remains nearly unchanged, indicating limited electron-donating capability of Na at higher loading.

(3) Configuration (GC vs. MC): The total electron donation is similar for GCs and MCs, showing that enlarged d_inter has minimal effect on Coulombic interaction.

The energies for configurations with different intercalation sites (A, B, C) differ negligibly (typically < 1 eV/cell) for both Li and Na and for both 1L and 2L intercalation [Supplementary Figure 30]. This suggests that the intercalated Li/Na atoms can easily move among high-symmetry sites, implying high mobility. This is consistent with previous reports of low diffusion barriers for Li/Na on MXene monolayers^[71,72,81].

Layer ratio barely affects charge transfer but plays a vital role in capacity. The specific capacity (gravimetric) decreases monotonically with increasing number of inactive layers: 1Ti1Nb -O has the highest value (120.9 mAh/g for 2L Li), followed by 3Ti1Nb (~63.6 mAh/g) and 1Ti3Nb (~59.4 mAh/g) with larger thickness, consistent with reported trends in Supplementary Table 13. Our capacities are lower than those of typical pure MXenes (e.g., 250 mAh/g for Ti₃C₂O₂)^[55,74] because only the Ti₃C₂T_x/Nb₂CT_x interface stores Li/Na; additional layers add mass without providing new storage sites. Accordingly, the volume expansion (ΔZ) upon intercalation follows the same order: 1Ti1Nb expands the most (up to ~18% for 1L Li), while 3Ti1Nb expands the least (up to ~8%). The binding energy (E_B) remains nearly constant across layer ratios (variation < 0.2 eV/atom for -O/-OH models), confirming that the interlayer chemical environment is dominated by the immediate interface.

In addition to electrochemical performance, the application of flexible devices also requires the evaluation of their mechanical properties. Based on standard energy-strain method as detailed in prior MXene studies^[52,83], we evaluate the in-plane stiffness (Y^2D, in N/m) of the ground-state Ti₃C₂T_x/Nb₂CT_x heterostructures [Supplementary Tables 14 and 15]. To avoid the ambiguity of defining a “thickness” for heterostructures with a vacuum layer, we focus on Y^2D rather than the bulk-equivalent Young’s modulus E (GPa) in our discussion, following the convention recommended for 2D materials^[83].

The calculated Y^2D values for -O terminated heterostructures are higher than those for -F and -OH. For instance, Y^2D of the robust -O terminated 1Ti1Nb heterostructure reaches 583 N/m (along y-direction), while its -F and -OH counterparts give 445 N/m and 467 N/m, respectively. This trend, with O-termination yielding the highest stiffness, agrees well with previous findings on pristine MXenes^{[52,54,83,84]}. This difference could be attributed to the robust O–M bonds, as –O exhibits higher coordination and obtains more electron (~1.1 |e|) with metal atoms due to its higher coordination compared to -F (~0.75 |e|) and -OH (~0.74 |e|) [Supplementary Table 16], as evidenced by the larger M–O bond stiffness^[47,83]. Notably, Y^2D for -F and -OH terminated structures are consistently similar across all stacking configurations, matching the reported statistical equivalence of F- and OH-termination strengthening effects in pure MXenes^[83].

Furthermore, the Y^2D values of 1Ti1Nb heterostructures (436-584 N/m) are obviously lower than those of high-stiffness pure MXenes like Nb₄C₃O₂ (605.99 N/m)^[52]. This is expected because the heterostructures are effectively multilayers with weak vdW interlayer coupling, leading to a higher bending rigidity and structural thickness, while their in-plane load-bearing capacity (Y^2D) is still dominated by the sum of individual stiff layers. The observed trend confirms that the in-plane stiffness of MXenes is a strong function of the overall layer count and surface termination composition, as determined by the number and strength of the M–T bonds within the framework^[52,83], the flexibility diminishes as the atomic layer thickness of MXene increases, aligning with earlier studies^[52,84].

The calculated Y^2D values [Supplementary Table 14] increase systematically with total layer count: 3Ti1Nb (four layers total) shows the highest stiffness (e.g., 1,260.65 N/m for -O), followed by 1Ti3Nb (1,142.99 N/m for -O) and 1Ti1Nb (583.69 N/m for -O). This trend is expected because Y^2D (in N/m) is a width-normalized force constant; adding more layers of stiff MXene in parallel increases the force required to achieve a given strain. When normalized by effective thickness (i.e., equivalent 3D modulus), the multilayers become softer due to weak interlayer coupling - a point already discussed above.

The synchronous correlation between interlayer spacing (d_inter) and binding energy (E_B) during sliding arises from a “decoupled interlayer correlation”: the distance between nearest functional groups (D_t) remains nearly constant, while d_inter varies by up to 25% for -F/-O termination. This decoupling, governed by isotropic London dispersion forces, allows remarkable tuning of interlayer spacing (~0.5 Å for -F/-O, ~1.7-2.0 Å for -OH) with minimal perturbation to electronic structure and low sliding barriers (3-12 meV/atom). Therefore, for applications requiring large interlayer expansion (e.g., ion intercalation), -OH termination is advantageous despite its higher barrier; for low-energy sliding manipulation, -F or -O termination is preferred. The weak dependence on layer ratio simplifies device fabrication because the interface chemistry dominates.

Practical MXene surfaces inevitably contain defects and mixed terminations, and operating temperatures vary^[85]. Encouragingly, a recent high-throughput study^[86] on 230 2D materials confirmed a universal negative correlation between interlayer binding energy and equilibrium distance, which was further validated by room-temperature atomic force microscope (AFM) on C, BN, and In₂Se₃. These findings support that the synchronous correlation we discovered is robust against moderate temperature variations and realistic surface imperfections, reinforcing its relevance for real-world devices.

Compared with sliding in some MXene homo-structures (barriers ~80-130 meV/atom) and MXene/MoS₂ heterostructures (barriers ~15-20 meV/atom)^[69], the barriers in Ti₃C₂T_x/Nb₂CT_x (3-12 meV/atom) are substantially lower, while the unique decoupling between d_inter and D_t has not been reported in those systems. Unlike homogeneous MXene sliding, where functional groups face identical chemical environments, the heterostructure introduces asymmetric interfaces that stabilize the constant D_t across a wide sliding range. This work thus expands the “slidetronics” paradigm to all-MXene heterostructures.

Turning to energy storage applications, for Li/Na storage, Ti₃C₂O₂/Nb₂CO₂ offers strong Coulombic interaction (0.7-0.9 |e| per Li) and reduced volume expansion in MCs, but the specific capacity (~120 mAh/g for 2L Li) is moderate due to the heterostructure’s large molar mass. A 1:1 layer ratio maximizes capacity, while asymmetric ratios (e.g., 3Ti1Nb) provide higher in-plane stiffness (~1,261 N/m) for structural reinforcement.

Before concluding, we acknowledge several simplifications in our study. The assumption of ideal FCC termination ordering is a pragmatic starting point, yet real MXene surfaces typically exhibit a mixture of -F, -O, and -OH groups. Static calculations of Li/Na intercalation provide binding energies but do not capture finite-temperature diffusion dynamics; future work should therefore employ kinetic Monte Carlo or molecular dynamics to assess rate performance. Similarly, the 0 K sliding barriers reported here may be overestimates, as thermal activation could facilitate interlayer motion under operating conditions. On the experimental side, in situ X-Ray diffraction (XRD) or transmission electron microscope (TEM) characterization during sliding would be invaluable for verifying the predicted decoupling. More broadly, we suggest that the “decoupled interlayer correlation” is not unique to MXenes and may be exploitable in other vdW heterostructures, such as transition metal dichalcogenides, for slidetronics or energy storage.