Due to the quick development of electrical automobile and portable electrical equipment, rechargeable batteries with high energy density and power density received great attention recently [1-4]. Although some novel battery systems have been developed, lithium-ion battery (LIB) is still a reliable and receivable technology for energy storage nowadays [5]. However, the commercial graphic anode for LIB cannot meet the high energy standard due to its low theoretical capacity of 372 mAh/g [5]. Specially, if considering the low tap density of graphite, its ultralow volumetric capacity (330–430 mAh/cm³) further strongly hinders the development and applications of lithium-ion battery [6]. Thus it is important to find novel anode materials with enhanced lithium-ion storage property.

Recently, layered ternary transition metal carbides or nitrides, also known as MAX phase compounds, showed their promise in lithium-ion storage [7-12]. With the repeating intercalation and de-intercalation of Li⁺ ions, the MAX particles have been exfoliated to thin nanosheets due to the structural stress generated by lithiation [7-12], which results in a significant improvement in electrochemical performance of MAX phases. Although their capacities increased with charge/discharge cycling, the close stacked layers and the high electrostatic repulsion between Ti and Li limit the lithiation of MAX phase greatly, rendering the low initial capacity and rate capability [13]. Nano-engineering in their structure is regarded as an effective strategy to increase the utilization of electrodes and achieve better electrochemical performance. For example, we decreased the size of Ti₂SnC, resulting in a high reversible capacity of 430 mAh/g, more than 5 times in comparison with its bulk counterpart [14]. In addition, both partially etched Ti₃AlC₂ [15] and oxidized Ti₃SiC₂ [16] show better specific capacity than their untreated counterparts.

According to the knowledge of fundamentals of material science, introducing defects could increase the conductivity of crystals, and the lithium could be coupled with deficient. It is reported that many metal oxides with oxygen-defects have exhibited obviously improved lithium storage capability [17-19]. MXene, a distinct representative of 2D materials, deriving from their precursor MAX, has lots of surface functional groups such as –O, –F and –OH, which show great efforts on the lithium-ion storage property because the electrical conductivity and surface redox reaction with lithium of MXene are much different with different surface terminations [20-22]. For instance, by regulation of functional groups, high capacity and ultra-long cycle capability of Nb₂CT_x MXene for lithium-ion battery have been achieved [23]. However, there are no reports concerning the surface structure of MAX phase due to the high stability of MAX phase as well as almost no functional groups on its surface. While for most non-oxide ceramics including MAX phase, there are surface oxides covered on their surface in most situations, which would probably influence the electrical conductivity and lithium ions diffusion rate. Therefore, it can be predicted that the electrochemical performance of MAX will be enhanced if the surface structure of MAX has been optimized.

Herein we report in a detailed investigation of the lithium-ion storage property of surface oxygen-deficient Ti₂SC nanosheets. By reducing the surface titanium oxide on Ti₂SC particles, oxygen vacancies are generated obviously on their surface. Due to the higher electrical conductivity and Li-ion diffusion rate derived from oxygen vacancies, our surface oxygen-deficient Ti₂SC electrode displays a superior rate performance from 50 mA/g to 2000 mA/g and holds a specific capacity of 350 mAh/g at 400 mA/g after 1000 cycles. These findings further proved the suitability of MAX phase as anode materials for LIB and provided an effective strategy for design surface structure of MAX phase to achieve enhanced electrochemical performance.

Fig. 1 shows XRD patterns and SEM images of different Ti₂SC samples, including Ti₂SC raw materials (bulk Ti₂SC), Ti₂SC collected after ultrasonic exfoliation (n-Ti₂SC), n-Ti₂SC annealed at 400 ℃ under Ar (A-Ti₂SC), n-Ti₂SC annealed with NaBH₄ at 400 ℃ under Ar (OV-Ti₂SC). According to the XRD pattern in Fig. 1a, it can be concluded that Ti₂SC has been successfully prepared by self-propagating high-temperature synthesis (SHS) method. Besides the main phase Ti₂SC, TiC is also found in this XRD pattern as an impurity phase. After treating by sonication with DMSO, the peaks' intensity became weaker, suggesting Ti₂SC particles are broken or delaminated to small particles or nanosheets. In addition, a broad diffuse hump peak in the region of 2θ = 20°−35° can be also observed, indicating Ti₂SC particles are partially oxidized during sonication. Then after heating at 400 ℃ for 2 h under Ar, the surface oxides on Ti₂SC particles were crystallized because a new peak at around 26° of anatase TiO₂ appeared. When adding NaBH₄ to Ti₂SC during annealing, the hump peak in the region of 2θ = 20°−35° is almost disappeared, showing the oxidized Ti₂SC particles have been partially reduced. Moreover, the peaks of Ti₂SC in XRD patterns of both A-Ti₂SC and OV-Ti₂SC become stronger compared to those of n-Ti₂SC, indicating a crystallization process arisen during annealing.

The atomic model of the lamellar structure of Ti₂SC with surface oxide has been shown in Fig. 1b, which consists of two distinct alternating layers-Ti₂C and S, and surface oxide-titanium oxide with oxygen vacancy. A distinct lamellar structure can be observed in the SEM image (Fig. S1 in Supporting information) of bulk Ti₂SC obtained by SHS, which is consistent with its atomic model. After sonication and annealing, the size of as-synthesized OV-Ti₂SC and A-Ti₂SC powders have been significantly decreased, showing average size of ~100–200 nm (Figs. 1c and d). Moreover, a few ultrathin nanosheets can be also observed in Figs. 1b and c, suggesting successful delamination of bulk Ti₂SC during sonication. In addition, based on EDS results in Table S1 (Supporting information), the atomic O/Ti ratio of OV-Ti₂SC is only 0.116, decreased by 78.4% compared to that of A-Ti₂SC, indicating most surface oxide on Ti₂SC particles has been reduced by adding NaBH₄ to Ti₂SC.

Figs. 2a-d show TEM and HRTEM images of OV-Ti₂SC and A-Ti₂SC. Both samples exhibit irregular sheet-like or particle-like morphology with a size range of 100–200 nm, which are consistent with SEM images in Fig. 1. The morphology and size of OV-Ti₂SC and A-Ti₂SC are similar to n-Ti₂SC in Fig. S2 (Supporting information), indicating there are no obvious reactions between particles during annealing. In addition, it can be observed from the edge of both samples that the sheets or particles are composed of a few layers, confirming the lamellar structure of Ti₂SC. According to THE HRTEM images (Figs. 2b and d), the surface of A-Ti₂SC particles are covered by anatase TiO₂, because the spacings of lattice fringes were measured to be 0.245 and 0.478 nm, corresponding to the (103) and (002) crystal plane of anatase TiO₂, respectively [24, 25]. However, these lattice fringes of anatase TiO₂ can't be observed on the surface of OV-Ti₂SC particles, while a spacing corresponding to the (100) crystal plane of Ti₂SC was detected in Fig. 2b, suggesting that the surface oxide on OV-Ti₂SC has been partly reduced. In addition, some crystal distortion was found on OV-Ti₂SC particles, which may be ascribed to the residual surface oxide, because the oxygen element is still homogeneously distributed on the surface of OV-Ti₂SC particles according to the element distribution image in Fig. 2e.

The presence of surface oxides on Ti₂SC particles of different Ti₂SC samples is further proved by Raman spectra in Fig. 2f. Three obvious peaks located at 145, 410, 610 cm⁻¹ can be observed in the spectra of n-Ti₂SC and A-Ti₂SC, corresponding to the vibration modes of A_1g, E_2g and E_2g of anatase TiO₂, respectively [18, 26-28]. Other two peaks of anatase TiO₂ are not distinct in these spectra, which may be affected by the internal Ti₂SC samples, as the in situ fabricated TiO₂ on Ti₃C₂ MXene [29]. After reducing the surface oxide with NaBH₄, the peaks at 410 and 610 cm⁻¹ of OV-Ti₂SC were almost disappeared, which can be ascribed to the generation of surface oxygen vacancies [30, 31]. At the same time, two peaks at 376 and 460 cm⁻¹ became obvious, which are shown as shoulders of the peak at 410 cm⁻¹ in the spectra of n-Ti₂SC and A-Ti₂SC. The first one at 376 cm⁻¹ is attributed to E_2g mode of Ti₂SC, which appeared because part of surface oxide has been reduced [32]. The other one at 460 cm⁻¹ are probably from the surface functional -O group on Ti₂SC, because it is located at the region of surface functional groups of Ti₃C₂ MXenes in their Raman spectra, which inherent the metal-carbon bonds of MAX phases [33].

To further study the surface oxidation status of Ti₂SC particles, X-ray photoelectron spectroscopy has been carried and showed in Figs. 3a-c. It can be seen from high resolution XPS patterns of Ti 2p and O 1s of OV-Ti₂SC and A-Ti₂SC (Figs. 3a and b) that there are titanium oxide and carbon oxide covered on the surface of OV-Ti₂SC and A-Ti₂SC particles because the metal-oxygen and carbon-oxygen are the main bonds observed in these spectra. For Ti 2p region of A-Ti₂SC, peaks at 456.5 and 458.3 eV correspond to the different oxidation states of titanium [34-37]. In the Ti 2p region of OV-Ti₂SC, the peak of Ti-C bond (455.6 eV) appeared while the peak of Ti³⁺ (456.3 eV) increased, indicating the decrease of surface oxygen on OV-Ti₂SC flakes. In Fig. 3b, high resolution O 1s spectra of both samples show similar morphology, while the peak of O vacancy (531.0 eV) [38] of OV-Ti₂SC increased a lot compared to that of A-Ti₂SC, indicating the reduction of surface oxides and formation of O vacancy on Ti₂SC particles. The O vacancy is also observed in the EPR spectra of both samples (Fig. 3d), and particularly, the spectrum of OV-Ti₂SC shows a very sharp change at around g = 2.003, suggesting there are more O vacancies on the surface of OV-Ti₂SC particles. For comparison, OV-Ti₂SC-300 and OV-Ti₂SC-500 obtained by heating n-Ti₂SC at 300 and 500 ℃ were also measured by EPR, and the results in Fig. S3 (Supporting information) show that both samples have fewer O vacancies compared to OV-Ti₂SC. High resolution C 1s spectra of OV-Ti₂SC and A-Ti₂SC were fitted to five and three peaks in Fig. 3c, respectively. Three peaks in both samples are assigned to graphitic C-C (284.8 eV), C-O (286.3 eV), and C=O (288.8 eV) bonds [34, 35, 39]. The other two peaks only in OV-Ti₂SC correspond to internal C−Ti (281.5 eV) and CH_x (285.6 eV) bonds [40], indicating the reduction of surface oxides on Ti₂SC particles. In addition, compared to A-Ti₂SC, the peaks of oxygen-carbon bonds at 286.3 eV and 288.8 eV for OV-Ti₂SC decreased a lot, consistent with the results of O 1s spectra, further proving the reduction effect during heating with NaBH₄. Thus combining the results of XRD, EDS, TEM, Raman, XPS and EPR, the surface oxides on Ti₂SC particles are mainly composed of anatase TiO₂, and which are partly reduced after annealing n-Ti₂SC with NaBH₄ under Ar as well as generation of oxygen vacancies. Here NaBH₄ is applied as a reduction agent, which can be decomposed at high temperature and provides active hydrogen that will create oxygen vacancies on the surface of Ti₂SC. [41] In addition, as well known, XPS analysis is sensitive to surface structure while EPR displays the bulk oxygen vacancy [18, 42], thus oxygen vacancies should be located both on the surface and in the bulk of the anatase layer.

Figs. 4a and b show the initial three CV curves of OV-Ti₂SC and A-Ti₂SC anodes, which have similar morphology. In the course of the 1^st cathodic scan, the small reduction peak at around 1.65 V (vs. Li/Li⁺) associates with the establishment of Li_xTiO₂ [18, 31], while two wide reduction peaks, positioned at 1.21 and 0.72 V (vs. Li/Li⁺), suggest the setting up of a solid electrolyte interface (SEI) layer. Moreover, the oxidation peak, located at 1.90 V (vs. Li/Li⁺), conforms with the reversible processes of Li_xTiO₂ to TiO₂ with the Li-ions deintercalation from Ti₂SC [18, 31]. Compared to A-Ti₂SC, the pair of redox peaks at 1.65 and 1.90 V (vs. Li/Li⁺) of OV-Ti₂SC are sharper during all the initial three cycles, suggesting the redox reaction between Li_xTiO₂ and TiO₂ is easier on the surface of OV-Ti₂SC particles, which may be attributed to faster Li ions diffusion in the structure with abundant oxygen vacancy. In addition, an oxidation peak close to 2.34 V (vs. Li/Li⁺) can be observed in the first CV curve of OV-Ti₂SC, which may be induced by the sulfur residue arisen from the delamination solvent DMSO and exposed after reduction of surface oxides [14]. The existence of sulfur residue is also proved by high resolution S 2p XPS spectrum of OV-Ti₂SC in Fig. S4 (Supporting information) because S-S bonds at 164.4 and 165.7 eV can be observed in this spectrum [43, 44].

Fig. 4c shows the cyclic profiles of OV-Ti₂SC and A-Ti₂SC electrodes at 400 mA/g. The 1^st lithiation capacity of OV-Ti₂SC is 247 mAh/g, while the 2^nd lithiation capacity is decreased to 148 mAh/g. This capacity loss can be ascribed to the irreversible setting up of a SEI layer. After the initial a few charge/discharge cycles, the specific capacity begins to gradually grow with cycling and then a stable reversible capacity of 350 mAh/g has been reached after ~820 cycles. The coulombic efficiency of the first cycle is 58%, then achieved to close 100% after a few cycles and persisted unchangeable in the following cycles. This increase in lithium ions storage with cycling can be assigned to the thinner nanosheets exfoliated by constant insertion of Li⁺, bringing about much more electrochemical active sites on the surface for Li⁺ insertion and alloying [7, 8, 14]. On the other hand, the A-Ti₂SC anode with less oxygen vacancy displays a capacity of 170 mAh/g after 1000 charge/discharge cycles, which is far smaller than that of the OV-Ti₂SC anode. The higher capacity of OV-Ti₂SC anode could be attributed to its specific surface structure with abundant oxygen vacancies, facilitating the diffusion and intercalation of lithium ions into OV-Ti₂SC particles. Moreover, OV-Ti₂SC electrode also displays excellent long-term cyclic performance at a high current density of 4 A/g, and a stable reversible capacity of 130 mAh/g is still preserved even after 3000 charge/discharge cycles (Fig. S5 in Supporting information). The stability of OV-Ti₂SC electrode in the process of cycle performance is further proved by the EPR results of different charge/discharge cycles, showing in Fig. S6 (Supporting information). The intensity of EPR peaks is increased after 100 charge/discharge cycles, and then kept stable even after 1000 charge/discharge cycles, indicating bulk O vacancies are increased with initial a few cycles and then remained almost no change during following cycles.

Fig. 4d shows the rate performance of OV-Ti₂SC after 1000 charge/discharge cycles at different current densities. At the current density of 100 mA/g, the OV-Ti₂SC electrode delivers a reversible specific capacity of 480 mAh/g. When the current densities were increased to 200, 400, 1000, 2000 and 4000 mA/g, the OV-Ti₂SC electrode exhibits reversible capacities of 432, 377, 303, 245 and 190 mAh/g, respectively. In addition, the OV-Ti₂SC electrode recovers a specific capacity of ~466 mAh/g once the current density was restored to 100 mA/g, exhibiting a splendid rate performance. Comparison of the rate performance between fresh OV-Ti₂SC and A-Ti₂SC without any previous charge/discharge cycles is shown in Fig. S7 (Supporting information). The capacities of fresh OV-Ti₂SC are much higher than those of A-Ti₂SC, particularly in high current density, indicating the higher electrical conductivity and the faster lithium ion diffusion in the OV-Ti₂SC electrode. In addition, the rate and cyclic performance of OV-Ti₂SC-300 and OV-Ti₂SC-500 are also presented in Figs. S7 and S8 (Supporting information), respectively. Due to their fewer O vacancies, both samples show poorer electrochemical performance compared to OV-Ti₂SC. Thus it can be concluded that the number of oxygen vacancies has great impact on the electrochemical performance of Ti₂SC.

Moreover, electrochemical impedance spectroscopy (EIS) was performed to analyze the electrochemical behavior of both Ti₂SC anodes. It can be seen from Fig. S9 (Supporting information), both Nyquist plots are mainly composed of two parts: a semicircle in the high frequency region is considered as the resistance at the Ohmic surface layer, and an oblique line in the low frequency region represents an ion diffusion process. To quantitatively asses the two processes, the Nyquist plots were fitted using the modified Randles equivalent circuit (the inset in Fig. S9). On the basis of the fitting results, the OV-Ti₂SC anode exhibits a lower charge-transfer resistance of 291.7 Ω, while A-Ti₂SC shows a higher transfer resistance of 410.3 Ω, indicating the oxygen vacancy enhances the electrical conductivity significantly and as a result, OV-Ti₂SC anode reaches a faster charge-transfer procedure. This finding is in a good agreement with the improvement of electrochemical performance of OV-Ti₂SC anode.

To further explain the reasons in detail for the splendid electrochemical behavior, particularly the rate behavior of the OV-Ti₂SC electrode, the kinetic mechanism analyses were performed according to its CV measurement. Figs. S10a and b (Supporting information) exhibit the CV profiles of OV-Ti₂SC and A-Ti₂SC at the scan rates from 0.1 mV/s to 2 mV/s, respectively. On the basis of the model of Dunn [45], the current in CV curve is divided into two fractions, associating with two different mechanisms, the surface-induced capacitive process and the diffusion-controlled insertion/extraction process. The formula: i = aν, in which both a and b are constants, representing the power law relationship between current (i) and the scan rate (ν), is applied to qualitatively examine the contribution of the two processes. The b value can be calculated by the slope of the fitted line log(ν)-log(i) (i = peak current). A b value of 0.5 means a typical diffusion-controlled mechanism, while b value of 1.0 is considered as surface-controlled capacitance behavior [46]. The b values of cathodic/anodic process are calculated to be 0.829/0.842 and 0.822/0.824 for OV-T₂SC and A-Ti₂SC, respectively (Figs. S10c and d in Supporting information). Compared with A-Ti₂SC electrode, the larger b values obtained in OV-T₂SC electrode are associated with more capacitive-like behaviors, befitting to faster reaction kinetics, which is useful for achieving superior rate performance. The quantitative capacitive contribution can be further analyzed according to the equation i(V) = k₁ν + k₂ν^0.5, where k₁ν is ascribed to surface-controlled capacitive contribution, k₂ν^0.5 is associated with diffusion-controlled insertion process, both k₁ and k₂ are constants, ν is the scan rate, and i is the current at a fixed potential (V) [45]. Thus the contribution of capacitive-like behavior for the OV-T₂SC electrode is as high as 78.2% at 0.5 mV/s (Fig. S10e in Supporting information), which is bigger than that for the A-T₂SC electrode (Fig. S10f in Supporting information), showing a better rate performance as a result. In addition, the capacitive contribution of A-T₂SC and OV-Ti₂SC at other scan rates was also calculated and shown in Fig. S11 (Supporting information). The fraction of capacitive contribution increases with the scan rate, and the overall contribution of OV-Ti₂SC at all scan rates is higher than A-T₂SC, further indicating the fast redox reaction behavior of OV-Ti₂SC.

Interestingly, OV-Ti₂SC electrode renders a high capacity of ~350 mAh/g after 1000 charge/discharge cycles at a current density of 400 mA/g, which doubles the values of A-Ti₂SC and our previous result of submicron Ti₂SC [7], further confirming the lithium-ion storage potential of MAX phase compounds. The Li-ion uptake in MAX phase compounds was proved to be originated from the intercalation of Li-ion into the layer structure of MAX phase because there are strong interactions of MX-Li and A-Li, specially for A elements (Si, Sn, S and so on) having alloying effects with lithium [9]. Assuming two Li ions can intercalate next to S atom in MAX phase, theoretical capacity value of 383 mAh/g can be expected for Ti₂SC, which is in good agreement with the OV-Ti₂SC electrode. The increased specific capacity and better rate performance of OV-Ti₂SC may attributed to the partly reduced surface oxides and generation of O vacancies on Ti₂SC particles, which provide better electrical conductivity and surface-induced capacitive contribution. In addition, it's worth mentioning that the high specific capacity of 480 mAh/g at 100 mA/g is highly comparable with those of previously reported MAX materials, showing in Fig. S12 (Supporting information). Especially at a high current density of 2000 mA/g, OV-Ti₂SC electrode has the highest specific capacity of 245 mAh/g among all reported MAX materials.

In summary, we prepared surface oxygen-deficient Ti₂SC by annealing the sonication delaminated Ti₂SC with NaBH₄. Benefiting from reduction of surface oxides and generation of oxygen vacancies, the as-prepared oxygen-deficient Ti₂SC (OV-Ti₂SC) shows a higher lithium-ion storage capacity and a better rate capability compared to Ti₂SC with more surface oxides. A specific capacity of 350 mAh/g for OV-Ti₂SC, along with impressive cycling performance and excellent rate capability were achieved, showing the promise of MAX phase for Li uptake. Considering there is a large amount of MAX phase compounds, this research provides a promising, facile surface engineering strategy for exploring the lithium-ion storage property of 2D materials from this large family.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by the National Natural Science Foundation of China (Nos. 21671167 and 51602277), Qinglan Project of Jiangsu Province. In addition, the authors would like to thank Shiyanjia Lab (www.shiyanjia.com) for the support of Raman and EPR analysis.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.05.014.