The global energy landscape is undergoing a significant transformation, propelled by the expansion of electric cars and the incorporation of clean energy systems (Luo et al., 2015; Zhang et al., 2020; Tian et al., 2021; Zhu et al., 2022; Huang et al., 2023). It is anticipated that battery technologies will play crucial roles in meeting this demand, helping to alleviate the range anxiety of electric vehicles and the intermittency of renewable energy sources (Dunn et al., 2011; Luo et al., 2015). Due to their high energy density, lithium-ion batteries (LIBs), which were commercialized in 1991, have enjoyed tremendous success in numerous sectors (Armand and Tarascon, 2008; Chou and Yu, 2017). However, energy storage battery demand will soar from 340 GWh in 2021–5,600 GWh in 2030 under the Net Zero Emissions Scenario (IEA, 2022). The scarcity of lithium and its unequal regional distribution have sparked interest in alternative battery technologies (Lin et al., 2015; Sun et al., 2015; Li et al., 2022). Given their abundant resources, low cost, environmental friendliness, and good electrochemical performance, sodium-ion batteries (SIBs) and potassium-ion batteries (PIBs) are viewed as two ideal alternatives (Yabuuchi et al., 2014; Hosaka et al., 2020; Shi et al., 2021). Efficient electrode materials are very important for the widespread use of SIBs and PIBs, especially anodes.

MXene is a family of two-dimensional nanomaterials with desirable properties for use as electrodes, including high metallic conductivity and good chemical stability (Ghidiu et al., 2014; Naguib et al., 2014; Wang et al., 2018). For instance, theoretical calculation proves that the capacities for sodium-ion and potassium-ion on Ti₃C₂M_x are 352 and 192 mAh g^-1, respectively. Their unique two-dimensional (2D) morphology and suitable interlayer spacing are conducive to lowering the ion diffusion barrier and increasing the number of ion adsorption sites (Lukatskaya et al., 2013; Liang et al., 2015; Yang et al., 2015; Zhao et al., 2015; Wu et al., 2021). Thus, they are possible to deliver high capacity and high-rate storage of metal ions. Nevertheless, MXene is still plagued by restacking and ion inaccessibility problems, which result in relatively low capacity, rapid performance degradation, and large volume expansion in SIBs and PIBs (Er et al., 2014; Xie et al., 2014; Lian et al., 2017). Reasonable electrode architecture will be beneficial to the overall performance enhancement of MXene-based anodes.

To address these challenges, researchers have devised a number of strategies, including introduce pores in the surface by chemical etching, increase interlayer spacing by surface modification, and prevent restacking of 2D nanosheets by embedding pillaring materials (Zhao et al., 2015; Ren et al., 2016; Xie et al., 2016; Kajiyama et al., 2017; Luo et al., 2017; Liu et al., 2022). Among these strategies, the melding of MXene and carbon materials has become a highly regarded approach, particularly carbon nanotubes (CNTs) with high conductivity, toughness, and elasticity. For instance, the self-assembled porous MXene/CNTs papers exhibited improved accessibility for electrolytes and showed an excellent sodium-ion storage volumetric performance (Xie et al., 2016). The knotted CNTs are able to simultaneously increase ion accessibility and decrease ion transport route tortuosity (Gao et al., 2020). These advantages make the MXene/CNTs composite a promising material for SIBs and PIBs. However, there is still a need for the development of a straightforward preparation procedure that ensures a strong coupling between MXene and CNTs.

In this study, we developed an in-situ approach to achieve superior anodes (MXene@CNTs) for SIBs and PIBs by constructing the three-dimensional (3D) structure of MXene and CNTs. The metal-organic framework (ZIF-67) was chosen as the carbon source for CNTs in-situ growth, which was strongly embedded in MXene sheets as pillaring materials to maintain the 3D structure. Additionally, nitrogen-doped CNTs also offer a high way for electron conduction and a buffer layer for volume expansion. In general, this specially designed structure can inhibit restacking of MXene, increase ion accessibility, as well as boost the overall conductivity of the electrodes. When used as anodes in both SIBs and PIBs, MXene@CNTs composites electrodes demonstrated enhanced capacity, rate capability, and cycling stability than MXene electrodes. They could deliver 286 and 250 mAh g^-1 at 0.05 A g^-1 for SIBs and PIBs, respectively, and still maintain 208 and 109 mAh g^-1 at 1 A g^-1 after cycling.

The detailed synthesis procedure of MXene@CNTs is schematically depicted in Figure 1A. Initially, the exfoliated MXene nanosheets were produced by immersing the MAX powder (Ti₃AlC₂) in HF aqueous solution (Supplementary Figures S1 A, B). Second, the cleaned and dried MXene nanosheets were dispersed in a methanol solution to serve as the base for ZIF-67 growth. At the same time, the cobalt precursor and 2MI were successively poured into the above solution. The first added cobalt ions can be deposited on the negatively charged MXene nanosheet through electrostatic interaction, which is convenient for the subsequent reaction with 2MI ligands. The generated ZIF-67 polyhedral particles were affixed to the surface of MXene after aging, and their size was smaller than that it would have been without MXene substrate (Supplementary Figures S1C-F). Finally, considering the close decomposition temperatures, 437 °C for ZIF-67°C and 443 °C for uric acid, we selected the uric acid as the vapor-phase carbon source to catalyze the nitrogen-doped CNTs in-situ growth (Kone et al., 2020).

The morphology and microstructure of MXene@CNTs were further analyzed using electron microscopy. As shown in Figure 1B, MXene@CNTs still display a similar delaminated structure as MXene nanosheets, but their surface is covered with velvet-like architecture, forming a unique 3D structure. In the magnified SEM image, the detailed “velvet” can be identified as an aggregate of numerous interconnected nanotubes (Figure 1C). The TEM image also verifies this morphology of nanotubes-coated MXene, and these nanotubes also formed a coating in the interlamellar position of MXene, not just in one part (Figure 1D). According to the high-resolution TEM image (HRTEM), these nanotubes can be determined as carbon nanotubes (CNTs) with approximate 0.35 nm interlayer spacing (Figure 1E). These CNTs show a bamboo-like hollow morphology, which is derived from the catalytic growth of cobalt nanoparticles (Wang et al., 2020). Such a composite structure is capable of providing numerous ion transportation channels, facilitating electrolyte infiltration, and mitigating volume expansion of electrodes. The EDS element mapping reveals an even distribution of C, N, and Ti, which demonstrates the introduction of nitrogen doping by uric acid (Figure 1F).

The successful composite of MXene with ZIF-67 was verified by XRD patterns of MXene@ZIF-67 (Supplementary Figures S2). The crystal structure of MXene@CNTs was further analyzed using XRD. As displayed in Figure 2A, the peaks at 7.1° and 14.5°correspond to the (002) and (004) reflections of delaminated MXene sheets (Naguib et al., 2012). The (002) peak of MXene@CNTs was located at 6.1°, and the corresponding layer spacing increased from 1.24 nm to 1.45 nm. The weaker (002) peak well confirmed the inhibition of MXene restacking by CNTs. In Raman spectrum of Figure 2B, the MXene@CNTs exhibits the characteristic peaks of MXene and carbon, indicating that they have been successfully composited. Two peaks of 211 and 622 cm^-1 correspond to the out-of-plane vibrations of Ti and C atoms, while three peaks of 273, 405, and 513 cm^-1 are in-plane vibrations of Ti, C, and surface functional groups (Yu et al., 2018; Sun et al., 2019). Additionally, the two peaks at 1,361 and 1,572 cm^-1 contributed to the D-band and G-band of carbon, respectively (Ferrari et al., 2006). Furthermore, the surface area and pore size distribution were analyzed by nitrogen adsorption-desorption isotherms (Figure 2C). The MXene@CNTs shows a type IV adsorption/desorption isotherm with a hysteresis loop, suggesting the coexistence of both mesopores and macropores. The Brunauer–Emmett–Teller (BET) surface area and total pore volume of the MXene@CNTs are 117 m² g^-1 and 0.354 cm³ g^-1, respectively. Both are considerably higher than 11 m² g^-1 and 0.019 cm³ g^-1 of pure MXene nanosheets. The increased surface area and pore volume further corroborate the significance of CNTs for 3D structure, which is anticipated to increase the ion storage sites and shorten the ion diffusion distance, thereby enhancing rate performance of electrodes.

XPS analysis was performed to elucidate the chemical environment state of MXene@CNTs. The survey XPS spectrum reveals the existence of C, N, O, F, Ti, and Co (Supplementary Figures S3). F is derived from the surface functional groups formed by etching, while Co content of about 1 at% is derived from ZIF-67. As demonstrated in Figure 2D, the C 1s can be deconvoluted into four peaks at 284.8, 285.9, 286.9, and 288.3 eV, corresponding to C-C, C-N, C-O, and O-C-O species, respectively. Notably, the N content in MXenen@CNTs was calculated to be as high as 9.63 at%. The high-resolution N 1s spectrum can be fitted well with four different peaks located at 396.2, 398.8, 401.0, and 403.2 eV, which correspond to the Ti-N, pyridinic (N-6), pyrrolic (N-5), and graphitic N (N-G), respectively (Figure 2E). The small N-Ti peak confirms the nitrogen doping of MXene, which is further confirmed by the Ti-N peak in the Ti 2p spectrum (Figure 2F) (Bao et al., 2018). The nitrogen modification in CNTs and MXene is expected to promote the adsorption of sodium/potassium-ion.

The electrochemical performance of MXene@CNTs in sodium/potassium-ion storage was comprehensively evaluated through a series of electrochemical tests. When applied as anodes for SIBs, MXene@CNTs show a broad reduction peak below 1.0 V on the first CV curve, which can be ascribed to the irreversible chemical reaction between sodium-ion and surface functional groups, along with the development of solid electrolyte interphase (SEI) film (Figure 3A). Then, the near coincidence of the CV curves in the second and third cycles indicates a high reversibility of MXene@CNTs. As shown in Supplementary Figure S4, the sloping GCD profiles indicate the dominating capacitive sodium-ion storage behavior of MXene@CNTs. The initial Coulombic efficiency (ICE) is only 41.1%, resulting from the combination of irreversible reaction and SEI film, which is consistent with the finding of CV analysis. Moreover, a gradual increase in the capacity of subsequent cycles implies the presence of an electrochemical activation process in the MXene@CNTs electrode (Wang et al., 2016). The rate performance of MXene@CNTs and MXene are compared in Figure 3B. MXene@CNTs shows capacities of 286, 246, 223, 188, 157, 124, and 80 mAh g^-1 at current densities of 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g^-1, respectively. Significantly greater than 88, 73, 63, 51, 38, 27, and 13 mAh g^-1 for MXene. The capacity of MXene@CNTs is 6.1 times that of MXene, which further confirms the significance of the modification of CNTs for electron conduction and ion transportation. As for cycling performance, it is worth noting that MXene@CNTs retains 208 mAh g^-1 after 500 cycles at 1.0 A g^-1, with a capacity retention rate of 107% relative to the second cycle (Figure 3C). This also proves the stability of the composite structure.

To further investigate the kinetics of ion transportation, a half-cell was scanned from 0.2 to 1.0 mV s^-1 (Figure 3D). The contribution ratios of capacitive effect (k₁ v) and diffusion control (k₂ v ^1/2) were separated by the Dunn method (Brezesinski et al., 2010). With the increase of scanning rate, the contribution ratio of capacitive effect increases gradually, reaching 92.3% at 1.0 mV s^-1. This suggests a capacitive-dominated sodium-ion storage mechanism in MXene@CNTs (Figure 3E). The capacitive contribution of MXene is only 82.4%, indicating the enhancement of active sites by 3D structure and nitrogen doping (Supplementary Figure S5). EIS analysis was further performed to investigate the charge transfer behavior. In the high-frequency region, the semicircle can be attributed to the charge-transfer resistance (R _ct), whereas the Warburg impedance is represented by the straight line observed in the low-frequency region. As shown in Figure 3F, the R _ct of MXene@CNTs is lower than that of MXene, indicating a quicker charge transfer process. In addition, the Warburg impedance also demonstrates a faster ion transport of sodium-ion in MXene@CNTs.

The GITT was performed to gain a deeper comprehension of the kinetics pertaining to the sodium-ion diffusion coefficient (D _Na). Figure 3G shows the voltage response of MXene@CNTs and MXene during the sodiation/desodiation process, and the corresponding diffusion coefficients are calculated. MXene@CNTs exhibits higher D _Na than that of MXene, implying faster kinetics performance, which is consistent with the above rate and EIS analysis (Figures 3H,I). All of them show a decreasing D _Na during the sodiation process, which may be a repulsive force caused by embedded ions on unembedded ions. Correspondingly, the contribution of repulsive forces during desodiation process also causes the D _Na to be higher at the beginning and lower later.

Additionally, the electrochemical performances of MXene@CNTs in PIBs were evaluated. Similar to the above SIBs section, the CV curves also show a broad reduction peak below 1.0 V on the first cycle because of the irreversible reaction and SEI film (Figure 4A). From the second cycle onward, the CV curves show a similar shape, suggesting good reversibility of MXene@CNTs electrode during potassium-ion storage. Accordingly, the first GCD curve reveals an initial Coulombic efficiency of 47.4% (Supplementary Figure S6). The sloping charge/discharge profiles are also consistent well with the CV curves, and all display a capacitive-dominated potassium-ion storage mechanism through adsorption/desorption. As for rate performance, the MXene@CNTs can deliver reversible capacities of 250, 210, 170, 127, 84, 53, and 38 mAh g^-1 at current densities of 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g^-1, respectively (Figure 4B). Much higher than 143, 94, 74, 55, 43, 32, and 19 mAh g^-1 of MXene, demonstrating the significant effect of CNTs on the rate performance. When cycled at 1.0 A g^-1 for 100 cycles, MXene@CNTs still attain a reversible capacity of 109 mAh g^-1 and a Coulombic efficiency of 99.9%, which is better than 11 mAh g^-1 and 96.6% of MXene, respectively (Figure 4C). Compared to SIBs, the relatively rapid attenuation may be attributable to the destruction of CNTs caused by the intercalation mechanism of large-size potassium-ion, resulting in the deterioration of cycling performance.

In terms of ion transportation kinetics, the contributions of capacitive effect and diffusion control were also studied utilizing CV curves analysis (Figure 4D). As shown in Figure 4E, the capacitive contribution displays an upward trend with the escalation of scan rates, reaching 81.5% at 1.0 mV s^-1. It is possible that nitrogen-doped CNTs introduce the intercalation storage mechanism of potassium-ion, resulting in a relatively lower capacitive contribution ratio than that of MXene (Supplementary Figure S7). In EIS spectra, the R _ct of MXene@CNTs is smaller than that of MXene, indicating a faster charge transfer process after decorating with CNTs (Figure 4F). The GITT was used to explore the kinetics of potassium-ion storage, and the corresponding diffusivity of potassium-ion (D _K) was evaluated (Figures 4G–I). The D _K of MXene@CNTs is higher than that of MXene in both the potassiation and depotassiation processes. In addition, it could be observed that D _K decreased as the degree of potassiation increased and that it decreased gradually during depotassiation. Just like in the SIBs scenario, this tendency is also due to the repulsive force formed by the pre-embedded potassium-ion, which hinders subsequent ion embedding and pushes ion out.

In summary, we have synthesized MXene@CNTs by decorating MXene nanosheets with in-situ growth of nitrogen-doped carbon nanotubes. Nitrogen-doped CNTs are utilized to inhibit the stacking of MXene nanosheets to form 3D structures, thereby increasing the active storage sites of ions. This rationally designed anode exhibits significant sodium/potassium-ion storage performance in terms of capacity, rate capability, and cycling stability. The outstanding performance of the MXene@CNTs may be attributed to remarkably improved conductivity through the integration of nitrogen-doped CNTs, which also facilitates electron conduction and ion adsorption. In particular, MXene@CNTs can sustain 80 mAh g^-1 for SIBs and 38 mAh g^-1 for PIBs when cycled at 5.0 A g^-1, which is 6.1 and 2.0 times that of MXene, respectively. Kinetic analysis reveals that capacitive contribution is primarily responsible for the high-rate capability. These features highlight this MXene@CNTs electrode as an attractive option for advanced energy storage devices, and further promote the practical applications of SIBs and PIBs.