Additive-Free MXene Liquid Crystals and Fibers.

The discovery of liquid crystalline (LC) phases in dispersions of two-dimensional (2D) materials has enabled the development of macroscopically aligned three-dimensional (3D) macrostructures. Here, we report the first experimental observation of self-assembled LC phases in aqueous Ti3C2T x MXene inks without using LC additives, binders, or stabilizing agents. We show that the transition concentration from the isotropic to nematic phase is influenced by the aspect ratio of MXene flakes. The formation of the nematic LC phase makes it possible to produce fibers from MXenes using a wet-spinning method. By changing the Ti3C2T x flake size in the ink formulation, coagulation bath, and spinning parameters, we control the morphology of the MXene fibers. The wet-spun Ti3C2T x fibers show a high electrical conductivity of ∼7750 S cm-1, surpassing existing nanomaterial-based fibers. A high volumetric capacitance of ∼1265 F cm-3 makes Ti3C2T x fibers promising for fiber-shaped supercapacitor devices. We also show that Ti3C2T x fibers can be used as heaters. Notably, the nematic LC phase can be achieved in other MXenes (Mo2Ti2C3T x and Ti2CT x ) and in various organic solvents, suggesting the widespread LC behavior of MXene inks.

Introduction

Liquid crystals (LCs) are thermodynamically stable mesophases that exhibit both liquid-like fluidity and crystal-like order. LC materials offer unique ordered structural characteristics, which can be tuned based on concentration, temperature, shear, and electric or magnetic fields.[1−3] Consequently, they hold great potential in various applications such as display devices,[4,5] smart glasses,[6,7] and temperature sensors.[8,9] While typical LCs are based on organic molecules with rod and/or disk shapes, colloidal dispersions of nanomaterials such as hydroxyapatite,[3] carbon nanotubes,[10−12] and graphene oxide[13,14] have also exhibited LC phases.[15] These LC nanomaterials offer simple processing routes for attaining highly ordered macrostructures for a wide range of functional devices.[16] For example, LC dispersions of carbon nanotubes and graphene oxide can be spun into strong and conductive fibers[12,17−19] for applications in flexible and wearable devices.[4,8,20,21] Thus, imparting LC behavior to highly conductive and electrochemically active nanomaterials can provide a platform for the creation of next-generation devices through macroscopic assembly of highly ordered structures.

Two-dimensional (2D) nanosheets of transition metal carbides and nitrides, known as MXenes, have received significant interest due to their unique combination of excellent mechanical properties, hydrophilic surfaces, and high metallic conductivity.[22−24] MXenes have a general formula of MXT, where M is an early transition metal (Ti, V, Cr, Mo, etc.), X is C and/or N, T represents the surface terminations (such as −O, −OH, or −F), and n is 1–4.[24,25] Ti3C2T (referred to as Ti3C2) films prepared using vacuum-assisted filtration have demonstrated high electrical conductivity (∼10 000 S cm–1),[26,27] excellent volumetric capacitance (∼1500 F cm–3),[28−30] and high efficiency in electromagnetic interference (EMI) shielding (92 dB for a 45 μm thick film),[31−33] indicating potential for use in practical applications.

The hydrophilic transition metal oxide and hydroxide bonds on the MXene basal plane impart MXene flakes with the same aqueous dispersibility as sheets of clay,[34,35] which is crucial for achieving lyotropic LC phases. A recent study utilized LC surfactant and single-walled carbon nanotubes to enable vertical alignment of Ti3C2 MXene on a substrate,[36] which led to improved ion-transport rate when used as a supercapacitor electrode. The surfactant molecules increased the packing symmetry of MXene flakes but needed to be removed prior to electrochemical testing in order to restore electrical conductivity and ion accessibility in the MXene macrostructure. Another study demonstrated the graphene oxide LC-assisted fiber spinning using Ti3C2;[37] however, the electrical conductivity of these hybrid fibers was fairly low (∼72.3 S cm–1 even at 88 wt % Ti3C2 loading). Polyurethane/Ti3C2 composite fibers have also been reported, which similarly achieved low electrical conductivity (22.6 S cm−1 at 23.1 wt % MXene loading).[68] Therefore, the development of additive-free and binder-free aqueous MXene inks and their successful integration into 3D macrostructures is a critical step toward demonstrating the full potential of MXenes for practical applications.

Here, we demonstrate for the first time that additive-free Ti3C2 MXene inks exhibit a nematic LC phase when we tune the aspect ratio of the MXene flakes and the concentration. LC MXene inks can be spun into pure MXene fibers with a high electrical conductivity of up to ∼7750 S cm–1, which is two orders of magnitude higher than the electrical conductivity of the LC graphene oxide-assisted MXene composite fibers.[37] Moreover, pure Ti3C2 fibers exhibit high volumetric capacitance of ∼1270 F cm–3 at 10 mV s–1, which is comparable to the freestanding MXene film electrodes.[29] We also show that this simple method can be extended to other members of the MXene family, using Ti2C and Mo2Ti2C3 as further examples. The LC MXenes enabled us to study for the first time the relationship between structure and properties of pure MXenes in a one-dimensional fiber format. Additionally, LC MXene fibers provide a novel platform to investigate their potential applications as energy storage electrodes in functional fabrics and heating elements in thermal comfort fabrics since they offer excellent electrical and electrochemical properties, surpassing other nanomaterial-based fibers.

Results

MXene Liquid Crystals

We synthesized Ti3C2MXene inks using LiF-HCl etching of Ti3AlC2 MAX phase, also referred as the “MILD” method.[38,39] Successful etching of Al layers and exfoliation of MXene flakes were confirmed by the disappearance of the (104) peak in the X-ray diffraction (XRD) patterns of the MAX phase and the increase in interlayer spacing noted by the downshift and broadening of the (002) peak (Figure S1). The high aspect ratio of Ti3C2 flakes (α, width/thickness, Figure a), the dispersibility of MXenes in a wide range of polar solvents, and their high surface charge indicated that it is possible to form lyotropic LC mesophase. According to Onsager theory,[40] an isotropic dispersion containing randomly oriented flakes can achieve long-range orientational ordering and form nematic phases at a critical transition concentration (Ct). The Ct for the isotropic–nematic (I–N) transition follows an inverse flake size dependence relationship as shown in Figure b, where aqueous MXene inks that consist of small flakes result in higher Ct while MXene inks containing larger flakes lead to lower Ct values.

Figure 1

Liquid crystalline (LC) behavior of MXene inks. (a) Schematic illustration of the chemical structure of Ti3C2T flakes. (b) Relationship between the MXene ink concentration (volumetric and mass) and the flake size (lateral size and aspect ratio) for isotropic to nematic (I–N) phase transformation based on theoretical calculations. The stars represent the theoretical LC transition concentrations for L-Ti3C2 and S-Ti3C2 dispersions based on their average flake size (3.1 μm and 310 nm, respectively). Inset shows the representative MXene flake orientation as the MXene ink goes through I–N transition. (c) Transmission electron microscopy (TEM) image and selected area electron diffraction pattern of a typical L-Ti3C2 flake. (d) Polarized optical microscopy (POM) images of aqueous L-Ti3C2 and S-Ti3C2 inks at various concentrations demonstrating the nematic LC formation based on flake size and concentration.

To study the aspect ratio dependence of the LC phase, we prepared aqueous Ti3C2 inks with two distinct flake size distributions. MXene inks containing Ti3C2 flakes as large as 10 μm (average flake size of ∼3.1 μm) were separated from the parent (as-synthesized) dispersion by centrifugation (herein denoted as L-Ti3C2). Dynamic light scattering (DLS) results of the aqueous L-Ti3C2 inks and the corresponding scanning electron microscopy (SEM) image are shown in Figure S2a,b, respectively. Probe sonication of the as-synthesized ink was employed to yield inks with small Ti3C2 flakes (denoted as S-Ti3C2) with an average flake size of ∼310 nm, as measured by DLS (Figure S2a). The corresponding SEM image is shown in Figure S2c. Figure c and Figure S2d are representative transmission electron microscopy (TEM) images of L-Ti3C2 and S-Ti3C2 flakes, respectively. The distinct bright spots in the selected area electron diffraction patterns of both L-Ti3C2 and S-Ti3C2 flakes along the [00l] zone axis indicate that the monocrystalline structures with a hexagonal arrangement of atoms are preserved in both samples. Figure S2e,f shows atomic force microscopy (AFM) images of representative L-Ti3C2 and S-Ti3C2 flakes, respectively. The AFM height profiles measured along the red dashed line in these figures show that both L-Ti3C2 and S-Ti3C2 flakes have the same height of ∼1.6 nm, which corresponds to a single layer of Ti3C2. However, the thickness of an individual Ti3C2 flake is reported to be ∼1 nm based on TEM studies and DFT calculations.[28,41] The increased height in AFM images is likely due to the presence of surface adsorbates, such as water molecules that are trapped under the Ti3C2 flake. It should be noted that similar observations have been previously reported for other 2D materials.[42,43] Hence, we used the thickness values obtained from TEM images to calculate the aspect ratio of MXene flakes.

Theoretical calculations based on the aspect ratio of the MXene flakes (see the Supporting Information for detailed information) suggest that the Ct for the I–N transition for S-Ti3C2 inks is between 33.3 and 83.3 mg mL–1 considering the α of MXene flakes between 500 and 200, respectively, as shown in Figure b. On the other hand, this Ct is ∼10-fold lower for L-Ti3C2 inks (from 1.7 to 16.7 mg mL–1 for α between 10 000 and 1000, respectively). It should be noted that, to study the relationship between lateral flake size of L-Ti3C2 flakes and the LC formation, we assumed that L-Ti3C2 inks contain monodisperse flakes, and this assumption has been widely acknowledged in the case of graphene oxide LCs.[14,44,45] According to Onsager theory,[40] the I–N transition is driven by a competition between rotational and translational entropies.[46,47] The translational entropy (availability of free volume) favors the isotropic state while rotational (packing) entropy promotes the formation of a nematic phase. In the dilute MXene regime, the low excluded volume (inaccessible volume due to the presence of MXene flakes) creates large translational entropy, which overcomes the rotational mobility of the MXene flakes, resulting in an isotropic dispersion. At the critical I–N transition concentration, the decrease in free volume gives rise to a high excluded volume resulting in increased packing of the MXene flakes. Consequently, the rotational entropy dominates the translational entropy, favoring long-range ordering of MXene flakes and the formation of nematic phases. The representative schematics for the I–N transition of the MXene flakes are shown in the inset of Figure b.

Figure 2

Rheological properties of S-Ti3C2 and L-Ti3C2 inks at various concentrations. Viscosity versus shear rate relationships for (a) L-Ti3C2 inks at the concentrations ranging from 3.5 to 26.5 mg mL–1 and (b) S-Ti3C2 inks at the concentration ranging from 35 to 150 mg mL–1. The black dashed lines in parts a and b indicate the applied shear rate during spinning. (c) Viscoelastic behavior of L-Ti3C2 at a concentration of 26.5 mg mL–1 (red solid symbols) and S-Ti3C2 at a concentration of 150 mg mL–1 (black hollow symbols). The square and circular symbols represent the elastic modulus (G′) and the viscous modulus (G″), respectively. (d) Frequency dependency on G′/G″ ratio for the L-Ti3C2 and S-Ti3C2 MXene inks as shown in part c. The dashed line indicates the G′/G″ ratio of 1; the right side of the dashed line suggests viscoelastic gel-like properties of dispersions.

We also experimentally identified the I–N transition by observing the evolution of birefringence with Ti3C2 inks under polarized optical microscopy, as demonstrated in Figure S3. Nematic phases are identified as threadlike topological defects called disclinations, which are manifested by birefringent optical textures consisting of dark and bright brushes (known as schlieren textures) when observed under crossed polarizers. Although nematic domains were observed at Ti3C2 concentrations of ∼6.3 mg mL–1 for L-Ti3C2 and ∼58.0 mg mL–1 for S-Ti3C2 (Figure d), isotropic phases also existed at these concentrations. The presence of schlieren textures at ∼13.2 mg mL–1 for L-Ti3C2 and ∼66.3 mg mL–1 for S-Ti3C2 indicated the complete I–N transition. By rotating the analyzer and the polarizer together in a crossed configuration, we observed schlieren textures of Ti3C2 inks (Figure S4, L-Ti3C2 ink at a concentration of 26.5 mg mL–1 was shown as an example). Both the experimental and theoretical results are in good agreement to determine the I–N transition concentration for both L-Ti3C2 and S-Ti3C2 flakes. As shown in Figure b, the stars represent the theoretical LC transition concentrations for L-Ti3C2 and S-Ti3C2 MXene inks based on their average flake size. As a result, the difference in I–N transition concentrations for L-Ti3C2 and S-Ti3C2 offers two possible routes to achieve pure LC MXene phases, i.e., using small flakes at high concentrations or using large flakes at low concentrations.

Additionally, we demonstrated that LC MXene inks can be achieved using a wide range of organic solvents compatible with Ti3C2 MXene[48] including dimethyl sulfoxide (Figure S5a), N-methyl-2-pyrrolidone (Figure S5b), and N,N-dimethylformamide (Figure S5c) at relatively low concentrations (∼20 mg mL–1) using L-Ti3C2 flakes. We also observed nematic LC phases for MXenes other than Ti3C2, such as Mo2Ti2C3 (Figure S5d) and Ti2C (Figure S5e) at relatively high concentrations (∼100 mg mL–1) using probe-sonicated, smaller MXene flakes, which indicated the applicability of our approach to the broader MXene family.

We carried out rheological investigations to elucidate the effects of flake size and concentration on the processability of aqueous LC MXene inks. The viscosity vs shear rate plots revealed shear-thinning behavior for both L-Ti3C2 (Figure a) and S-Ti3C2 (Figure b) inks at all concentrations studied. The L-Ti3C2 inks exhibited a high zero-shear viscosity (the viscosity values measured at 0.01 s–1) of ∼5050 Pa s at ∼26.5 mg mL–1, which was observed at a much higher concentration for S-Ti3C2 inks (∼150 mg mL–1). In their isotropic phase concentrations (3.5 mg mL–1 for L-Ti3C2 and 35 mg mL–1 for S-Ti3C2), the ratio of elastic modulus (G′) to viscous modulus (G″) was less than 1 indicating the presence of a liquid-dominant viscoelastic phase (Figure S6a). On the other hand, the LC MXene inks at full nematic phase transition concentrations (26.5 mg mL–1 for L-Ti3C2 and 150 mg mL–1 for S-Ti3C2) exhibited a strong dominance of the G′ over G″ throughout the frequency range (Figure c), demonstrating gel-like properties. Corresponding G′/G″ ratios for S-Ti3C2 (150 mg mL–1) and L-Ti3C2 (26.5 mg mL–1) MXene inks over the whole frequency range are shown in Figure d, where the dashed line indicates the G′/G″ ratio of 1, and the right side of the line shows the regime where viscoelastic gel-like properties of the inks become dominant. Both of the L-Ti3C2 and S-Ti3C2 inks already demonstrate viscoelastic gel properties (G′/G″ ≫ 1).[16] Therefore, a further increase in concentration and the subsequent increase in viscosity can impose challenges during wet-spinning. Moreover, the G′/G″ ratios of L-Ti3C2 inks over the whole frequency range shown in Figure S6b indicate that all studied concentrations demonstrate spinnability at high frequencies (>10–2 Hz). These results suggest that both LC S-Ti3C2 and L-Ti3C2 inks can exhibit viscoelastic gel-like properties when their G′/G″ ratios are above 1, which is the ideal region for wet-spinning.[49]

Pure LC MXene Fibers

To demonstrate the application of additive-free aqueous LC MXene inks in assembled MXene macrostructures, pure Ti3C2 fibers were spun using a wet-spinning technique, which involved extruding aqueous LC MXene inks (as spinning dope) through a fine nozzle into a coagulation bath (Figure a). We investigated how coagulation bath composition, MXene flake size, and nozzle size (needle gauge) influence the fiber formation process and the fiber properties. We evaluated four different coagulation baths including ethanol, strong and weak acids (9 M H2SO4 and 99 wt % acetic acid in water, respectively), 5 wt % CaCl2 aqueous solution, and 0.5 wt % chitosan solution in acetic acid/water (referred to as chitosan bath).[50−52] We found that the spinnability concentration limit for aqueous S-Ti3C2 (150 mg mL–1) is higher than that for L-Ti3C2 (26.5 mg mL–1). For these concentrations, continuous extrusion was possible using ethanol, H2SO4, and CaCl2 baths, but the resulting fibers were too weak to handle after spinning. In contrast, fibers spun in acetic acid and chitosan baths demonstrated sufficient strength to be handled in the bath and transferred onto a spool. An example of 5 m of pure Ti3C2 fiber is shown in Figure b. Moreover, thermogravimetric analysis (TGA) results did not show any residue of chitosan or acetic acid in the fibers after washing and drying as shown in Figure S7.

Figure 3

Wet-spinning and formation mechanism of neat LC MXene fibers. (a) Schematic representation of the wet-spinning setup used in this work. The inset illustrates the alignment of LC MXene under the shear force in the spinneret. (b) Photograph of a 5 m long Ti3C2 fiber successfully collected on a spool. (c, d) Cross-sectional SEM images of LC MXene fibers using S-Ti3C2 flakes produced in an acetic acid bath. (e) Schematic illustration of the fast coagulation mechanism when acetic acid was used as the coagulation bath. (f, g) Cross-sectional SEM images of S-Ti3C2 fibers produced in the chitosan bath. (h) Schematic illustration of the slow coagulation mechanism when chitosan was used as the coagulation bath. The bottom (green) arrow in parts e and h represents the inward diffusion of the coagulating agent into the fiber, and the top (blue) arrow represents the outward flow of the dope’s solvent into the coagulation bath.

Examination of the fiber cross-sections by scanning electron microscopy (SEM) revealed the high alignment of the Ti3C2 flakes along the fiber axis (Figure d–g). These observations imply that the applied shear during spinning was sufficient to induce the alignment of the ordered LC MXene domains as depicted in the inset of Figure a. This aligned state was retained during the coagulation and solidification processes. We also noticed differences in the fiber microstructure at different spinning conditions. For example, fibers prepared from the S-Ti3C2 inks using the acetic acid bath had an average diameter of ∼76.2 μm (Figure c and Figure S8a), and these fibers showed an open microstructure. The distance between Ti3C2 flakes is around 100–500 nm, as measured from high-resolution SEM images (Figure d). In contrast, S-Ti3C2 fibers prepared using the chitosan bath were ∼40% smaller in diameter (∼45.3 μm, Figure f and Figure S8a) and had a densely packed microstructure (Figure g) compared to the S-Ti3C2 fibers spun using the acetic acid bath. These differences in microstructure also led to significantly different S-Ti3C2 fiber densities of ∼1.7 g cm–3 for acetic acid and ∼3.6 g cm–3 for the chitosan bath as shown in Figure S8b. Similar trends in terms of fiber diameter (Figure S8a) and density (Figure S8b) have been observed when L-Ti3C2 inks were spun using acetic acid and chitosan baths. The open microstructure of the fibers spun using acetic acid independent of the flake size could be attributed to the faster coagulation rate in the acetic acid bath compared to the chitosan bath. Acetic acid easily replaced the water in the spinning solution, as compared to chitosan, due to their differences in size. This rapid solvent exchange for acetic acid “froze” the MXene flakes and thus generated a loosely packed and open network of MXene flakes in the fiber (Figure e). On the other hand, the slow coagulation rate (Figure h) due to larger chitosan polymer chains resulted in densely packed MXene flakes in the fiber. Even though the chitosan bath led to a slower coagulation rate compared to acetic acid, the coagulation rate needs to be further decreased to improve the circularity of the fiber cross-section. We also showed that the diameter of the fibers can be controlled by changing the spinning nozzle size. For instance, the S-Ti3C2 fibers in acetic acid spun using a smaller nozzle diameter (gauge 29) were 56% smaller in diameter than those produced using the gauge 26 nozzle (Figure S8c). When gauge 29 nozzles were used during spinning, S-Ti3C2 fibers, independent of the coagulation bath, exhibited similar diameters of ∼33 μm (Figure S8c), which suggests that the effect of coagulation bath becomes less important with decreasing needle size (higher needle gauge). Moreover, to demonstrate the feasibility of the fiber spinning from other MXenes, we spun Mo2Ti2C3 MXene fibers using aqueous S-Mo2Ti2C3 inks (at ∼102 mg mL–1) into the acetone bath. Similar to the S-Ti3C2 fibers spun in the chitosan bath, the S-Mo2Ti2C3 fibers demonstrated densely packed and highly ordered MXene flakes that were aligned along the fiber axis (Figure S9).

We also conducted small-angle X-ray scattering (SAXS) and wide-angle X-ray scattering (WAXS) experiments to further study the differences in the fiber microstructure with different coagulation conditions (acetic acid vs chitosan bath) and MXene flake size (small vs large). Figure a shows the typical scattering data from Ti3C2 including the (002) peak at q = 0.55 Å–1, (010) peak at q = 2.41 Å–1, and (110) peak at q = 4.10 Å–1 for all fibers. When the intensity was integrated along the 2θ direction and normalized relative to the (110) peak that is attributed to the hexagonal space group of the Ti2C2 unit cell[53] (Figure b), we found that the S-Ti3C2 fibers spun in the chitosan bath exhibited a narrower full width at half-maximum (fwhm) value of 0.055 Å–1 compared to S-Ti3C2 fibers spun in acetic acid (0.14 Å–1) for the (002) peak suggesting larger crystallite size for the fibers spun in the chitosan bath.[54,55] However, we noticed that multiple diffraction peaks, such as (004), (006), and (008), were more pronounced for L-Ti3C2 fibers (Figure c) than for S-Ti3C2 fibers (Figure b) independent of the coagulation bath, indicating better flakes alignment in the case of large flakes. To evaluate the orientation of Ti3C2 flakes relative to the fiber axis, we plotted the intensity along azimuth (ϕ) at the (002) peak (Figure d) in order to calculate the Herman’s orientation factor (f), which is used for assessing the orientation of polymer chains.[56,57] Detailed information about the calculations is provided in the Supporting Information. The values for f range from 0 for random orientation to 1 for complete alignment. We found that the f of S-Ti3C2 fibers spun in the chitosan bath (f = 0.83) was much higher than that of S-Ti3C2 fibers spun in acetic acid (f = 0.58). These results also indicate that the slow coagulation rate in the chitosan bath led to ordered aligned MXene flakes compared to acetic acid where the coagulation rate is faster. A similar trend was also observed for L-Ti3C2 fibers spun in a chitosan bath (f = 0.72) and L-Ti3C2 fibers spun in acetic acid (f = 0.68); however, the difference in f was minimal for the S-Ti3C2 fibers spun in acetic acid and chitosan baths, which might be due to the improved alignment of large MXene flakes under shear compared to the small MXene flakes. Also, it was unexpected that L-Ti3C2 fibers spun in a chitosan bath show a lower f value compared to S-Ti3C2 fibers spun in a chitosan bath, which could be attributed to the low concentration of L-Ti3C2 flakes in the spinning dope, resulting in wrinkles and voids in the L-Ti3C2 fibers spun in the chitosan bath, as shown by SEM images (Figure S10).

Figure 4

Orientation of MXene flakes in fibers. (a) SAXS/WAXS scattering patterns of S-Ti3C2 and L-Ti3C2 fibers spun in chitosan and acetic acid baths. Scattering data for (b) S-Ti3C2 fibers spun in a chitosan bath and acetic acid and (c) L-Ti3C2 fibers spun in a chitosan bath and acetic acid. Data reduction is based on the integration of the SAXS/WAXS images along the 2θ (radial equatorial) direction over an interval of ±5°. (d) Azimuthal plot of scattering at (002) along the ϕ direction in the region of ±90°.

Pure LC MXene Fiber Properties

We studied how the differences in fiber morphology (e.g., open vs dense, linear density, and cross-section shape) affect the electrical conductivity of the pure LC MXene fibers. The electrical conductivity (σ) of the fibers was calculated by accounting for the noncircularity of the fiber diameter. The resistance data were normalized using the fiber cross-section area taken from multiple SEM images across the fiber length (see the Supporting Information, Figure S10). These representative SEM images illustrated the different fiber cross-sectional shapes from different spinning conditions. We found that fibers spun from L-Ti3C2 flakes have approximately 2 times higher σ than those prepared from S-Ti3C2 flakes regardless of the coagulation bath (Figure a). The highest electrical conductivity was obtained for L-Ti3C2 fibers spun in a chitosan bath (∼7748 S cm–1), which is 2.2 times higher than S-Ti3C2 fibers spun in a chitosan bath (∼3512 S cm–1). This value is close to the highest reported conductivity of the filtered freestanding film with a thickness of 4.6 μm (∼9500 S cm–1),[26] and it demonstrates 2 orders of magnitude higher electrical conductivity as compared to LC graphene oxide-assisted MXene hybrid fibers[37] and exceeds the electrical conductivity of all graphene and other carbon-based fibers.[17,58,59] When acetic acid is used as the coagulation bath, the L-Ti3C2 fibers (∼4048 S cm–1) showed more than 2 times higher σ than the S-Ti3C2 fibers (∼1942 S cm–1) (Figure a). Also, independent of flake size, the chitosan bath produced more conductive fibers than the acetic acid bath. This difference could be attributed to the denser and more packed fiber morphology of fibers produced from the chitosan bath. When we calculated the gravimetric conductivity (σG) to better understand the effect of flake size while eliminating the effect of fiber morphology, we found that the σG more than doubled for fibers spun from L-Ti3C2 flakes (∼2200 vs 1000 S cm2 g–1, Figure b). This difference highlights the importance of minimal intersheet contacts per unit length between flakes, which was observed in the literature with spray-coated Ti3C2 films[60] and reduced graphene oxide fibers.[61]

Figure 5

Electrical properties of Ti3C2 fibers spun at various conditions. (a) Electrical conductivity of Ti3C2 fibers calculated based on cross-sectional area measured from SEM. (b) Gravimetric conductivity of fibers produced using various coagulation baths and MXene flake sizes.

The superior effect of using large flakes and a chitosan bath on electrical conductivity was also observed for the mechanical properties. Figure S11 shows the tensile stress–strain data for representative samples. L-Ti3C2 fibers spun in a chitosan bath displayed the highest tensile strength of ∼40.5 MPa and strain at break of ∼1.7%, which are higher than pure freestanding Ti3C2 films (tensile strength of ∼22.1 MPa and failure strain at break of ∼1%).[62] It should be noted that these values are significantly lower than the mechanical properties of individual Ti3C2 flakes (Young’s modulus of ∼330 GPa and tensile strength of ∼17 GPa)[63] suggesting the need for enhancing the interaction between MXene flakes; thus, the overall mechanical properties are of great interest for future studies.

We also evaluated the electrochemical properties of the Ti3C2 fibers in a standard three-electrode cell using 1 M H2SO4 electrolyte, Ag/AgCl reference electrode, and carbon rod auxiliary electrode. The pure LC MXene fibers were tested without the use of a supporting current collector. S-Ti3C2 fibers spun in a chitosan bath have been chosen to demonstrate the detailed electrochemistry performance of LC MXene fibers. Cyclic voltammetry (CV) curves in Figure a displayed that the working potential window is between −0.75 and 0.2 V (scan rate of 5 mV s–1). The potential window was the same for all fiber samples, suggesting that the electrochemical behavior of LC MXene fibers is independent of the coagulation bath and the flake size. In this potential window, the CV curves at scan rates ranging from 5 to 500 mV s–1 showed the characteristic peaks for the intercalation/deintercalation of H+ ions and surface redox reactions of Ti3C2 MXene (Figure b).[29,30,36] In addition, the galvanostatic charge–discharge (GCD) curves were highly symmetric at all current densities investigated with negligible iR drop (Figure c). When we calculated the volumetric capacitance (C) of the various fibers from the CV curves, we found that fibers produced from the chitosan bath had a higher C than those produced using the acetic acid bath independent of the flake size (Figure d). For example, the C of S-Ti3C2 fibers spun in a chitosan bath (1265 F cm–3) is almost 2 times higher than that of S-Ti3C2 fibers spun in acetic acid (737 F cm–3). The difference in C could be attributed to the 2.1-fold higher fiber density of the S-Ti3C2 fibers spun in a chitosan bath (∼3.6 g cm–3) compared to the S-Ti3C2 fibers spun in acetic acid (∼1.7 g cm–3), as shown in Figure S8b. It should be noted that the C of S-Ti3C2 fibers spun in a chitosan bath is comparable to the highest reported C of pure Ti3C2 films (∼1500 F cm–3).[29] Furthermore, fibers spun using small flakes exhibited higher C and gravimetric capacitance (CG) than the fibers produced using large flakes. For instance, S-Ti3C2 fibers spun in acetic acid (434 F g–1) showed higher CG compared to the L-Ti3C2 fibers spun in the same coagulation bath (393 F g–1). This increased capacitance could be explained by the presence of more defects in small flakes, which can promote easier electrolyte diffusion and exposure to a larger number of active sites.[60]

Figure 6

Electrochemical performance of Ti3C2 fibers in a three-electrode system using 1 M H2SO4 electrolyte. (a) Cyclic voltammetry (CV) curves of S-Ti3C2 fibers spun in a chitosan bath for various potential windows at a scan rate of 5 mV s–1. (b) CV curves of S-Ti3C2 fibers spun in a chitosan bath at scan rates from 5 to 500 mV s–1. (c) GCD curves of S-Ti3C2 fibers spun in a chitosan bath at current densities from 1 to 20 A cm–3. (d) Changes in volumetric capacitance of LC MXene fibers at different scan rates. (e) Nyquist plots for all spun fibers. (f) Cyclic stability of S-Ti3C2 fibers spun in a chitosan bath over 10 000 cycles at a scan rate of 100 mV s–1.

We carried out electrochemical impedance spectroscopy (EIS) to investigate the charge transfer and ion transport in the fibers (Figure e). The Nyquist plots showed nearly vertical behavior at all frequencies, suggesting fast ion diffusion in all fibers. We also estimated the equivalent series resistance (ESR) using the intercept of the semicircle at high frequencies with the real impedance axis. The L-Ti3C2 fibers spun in a chitosan bath showed the lowest ESR among the fibers investigated, indicating the lowest internal resistance (i.e., sum of the fiber electrode, the electrolyte, and the contact resistances). The ESR values for fibers from L-Ti3C2 were lower than those for the fibers made from S-Ti3C2. Fibers made using a chitosan bath had a lower ESR than when the acetic acid bath was used. These results are consistent with the higher electrical conductivity observed for large flakes and the denser packing achieved with the chitosan bath, which led to more effective interflake charge transport. The S-Ti3C2 fibers spun in a chitosan bath were tested over 10 000 CV cycles to investigate the long-term stability in liquid electrolyte. These fibers showed minimal distortion in the CV curves with a ∼100% capacitance retention and a Coulombic efficiency of ∼100% at all cycles (Figure f), indicating excellent cycling stability.

To demonstrate the multifunctionality of LC MXene fibers, we also evaluated their electrothermal properties for use in fiber-shaped Joule heating elements. Fibers with high electrical conductivity, safe operation voltage, and fast electrothermal response are desirable for use as wearable heaters.[64] We evaluated the electrothermal energy conversion properties of MXene fibers by passing a DC voltage to ∼4.5 cm long L-Ti3C2 fibers spun in a chitosan bath (resistance of ∼64 Ω cm–1) mounted between two copper electrodes, and the corresponding increase in temperature of the fiber was monitored using an infrared camera (Figure S12a). Our results showed that the L-Ti3C2 fibers spun in a chitosan bath have a rapid electrothermal response and low power consumption while being lightweight. For example, the temperature increased from ∼26 °C at 3 V to ∼68 °C at 24 V (Figure S12b). The change in fiber temperature occurred in less than ∼0.07 s, which translates to a heating rate of ∼600 °C s–1. This heating rate is higher than that of graphene fibers (∼571 °C s–1),[64] suggesting a slightly faster electrothermal response of L-Ti3C2 fibers spun in a chitosan bath. Additionally, a low power density of ∼126 W cm–2 at an applied voltage of 24 V was sufficient for these fibers to achieve a temperature of 68 °C (Figure S12c). These electrothermal properties make LC MXene fibers promising candidates for fiber-shaped Joule heating elements for car seat heaters, heated blankets, and wearable heaters, which are used commonly in the healthcare industry including self-heating garments for medical heat therapy and rehabilitation, and for controlled drug delivery.[65,66] Of course, applications in electromagnetic shielding, antennas, sensors, and other devices seamlessly embedded into textiles can be envisioned.

Conclusion

We reported the first observation of nematic LC in pure Ti3C2 MXene dispersions using both small and large MXene flakes. The outstanding dispersibility of MXene in various solvents made it possible to form LC at very high concentrations (∼150 mg mL–1) using small MXene flakes (∼310 nm) and concentrations as low as 6.3 mg mL–1 in the case of large MXene flakes (3.1 μm). This discovery also enabled the first demonstration of freestanding and highly conductive pure MXene fibers via wet-spinning the LC inks of Ti3C2, Ti2C, and Mo2Ti2C3. It also enabled us to control the microstructure of pure MXene fibers through their morphology, density, circularity, and flake alignment and, consequently, to study their effect in tailoring fiber properties. We learned from this work that MXene flake size, alignment, and stacking density influence the electrical properties of the fibers significantly. The highest electrical conductivity of ∼7750 S cm–1 achieved for pure Ti3C2 fibers with an average diameter of 34.5 μm is comparable to the highest reported literature value on 4.6 μm thick Ti3C2 films (∼9500 S cm–1). Moreover, the high volumetric capacitance of pure Ti3C2 fibers (∼1265 F cm–3 at 10 mV s−1) is on par with Ti3C2 films and also rivals other fiber electrodes. The LC MXene fibers without using any additives produced in this work introduced MXene as a new member of the family of functional fibers that offer a plethora of possibilities for a variety of smart textile applications, including energy storage, heating, and many others. We believe that LC MXenes will play a key role on the integration of MXene into various architectures and advanced applications requiring highly ordered structures.

Methods

Synthesis of Ti3C2 MXene

Ti3C2 MXene was synthesized by selectively etching Al atomic layers from Ti3AlC2 MAX phase (<40 μm particle size, Carbon-Ukraine). The etching solution was prepared by dissolving 1.6 g of lithium fluoride (LiF, 99%, Sigma-Aldrich) in 20 mL of 9 M hydrochloric acid (HCl, Sigma-Aldrich) while stirring for 5 min. A 1 g portion of Ti3AlC2 powder was slowly added to the etchant, and the reaction was allowed to proceed for 24 h at 35 °C. The acidic dispersion was then washed with deionized (DI) water by repeated centrifugation (Eppendorf 5810R) at 3500 rpm for 5 min per each cycle until self-delamination occurred (pH ∼ 6). In the next step, the dispersion was centrifuged at 3500 rpm for 30 min to separate delaminated flakes from multilayer MXene and unreacted MAX phase. The supernatant containing delaminated large and small Ti3C2 flakes was then collected and denoted as as-synthesized Ti3C2 ink.

Preparation of L-Ti3C2 and S-Ti3C2MXene Inks

The L-Ti3C2 ink was obtained by centrifuging the as-synthesized Ti3C2 ink at 4500 rpm for 30 min and redispersing the sediment containing predominantly large Ti3C2 flakes in water. The concentration of the S-Ti3C2 and L-Ti3C2 inks was adjusted by centrifugation at 7000 rpm for 1 h and redispersing the sediment in a carefully measured amount of water. In order to prepare S-Ti3C2 ink, the as-synthesized Ti3C2 ink was probe-sonicated (Fisher Scientific 505 Sonic Dismembrator, 500 W) for 20 min under a pulse setting (8 s on and 2 s off at 50% amplitude). An ice bath was used to prevent heating during sonication. The dispersion was then centrifuged at 3500 rpm for 10 min, and the supernatant containing small flakes was collected (S-Ti3C2).

Synthesis and Delamination of Mo2Ti2C3 and Ti2C MXenes Inks

Mo2Ti2AlC3 and Ti2AlC MAX powders were synthesized following protocols reported elsewhere.[67] Mo2Ti2C3 MXene was obtained by selective etching of the Al layers from 3 g of Mo2Ti2AlC3 (≤74 μm particle size) in 30 mL of 50% aqueous hydrofluoric acid (HF) solution (48–51%, Acros Organics). After etching at 50 °C for 96 h, the solution was washed with DI water and repeatedly centrifuged at 3500 rpm for 5 min until the supernatant reached pH ∼ 6. The sediment containing Mo2Ti2C3 multilayer was then filtered using an acetate cellulose membrane (0.45 μm pore size, Millipore). After drying under vacuum for 12 h, the Mo2Ti2C3 multilayer powder was mixed in 30 mL of 10 wt % aqueous tetramethylammonium hydroxide solution (25 wt %, Sigma-Aldrich). The dispersion was stirred for 18 h at room temperature and then washed by centrifugation at 3500 rpm for 15 min until neutral pH was achieved. The sediment was then redispersed in 40 mL of DI water and bath sonicated (Branson 2510 ultrasonic cleaner, 100 W) for 1 h while bubbling the dispersion with argon to prevent Mo2Ti2C3 oxidation. The delaminated Mo2Ti2C3 ink achieved after sonication was centrifuged one last time at 3500 rpm for 1 h, and the sediment was redispersed in a carefully measured amount of water to adjust the concentration. Ti2C MXene was prepared by etching 3 g of the MAX powder in a mixture of LiF (4.5 g, Alfa Aesar) and 9 M HCl (30 mL, Fisher) at 35 °C for 15 h. The mixture was washed with DI water and centrifuged at 3500 rpm for 5 min. The washing process was repeated 3–4 times until a dark supernatant was obtained. The first dark supernatant was decanted, and the sediment was dispersed in DI water and hand-shaken vigorously for 10 min. Finally, the mixture was centrifuged at 3500 rpm for 10 min, and the supernatant was collected and bath sonicated similarly to the Mo2Ti2C3 MXene.

Wet-Spinning of Pure MXene Fibers

Aqueous L-Ti3C2 (26.5 mg mL–1) and S-Ti3C2 (150 mg mL–1) MXene inks were prepared and used as spinning dopes. The spinning dope was then injected through a blunt-tip needle (26 or 29 gauge) into a rotating Petri dish containing the coagulation bath (Figure a). A syringe pump was used to control the spinning flow rate that was varied from 0.5 to 2.5 mL h–1 depending on coagulation bath composition and needle gauge. Ethanol 200 proof, 9 M H2SO4 solution, 5 wt % CaCl2 aqueous solution, 99 wt % acetic acid in water, and 0.5 wt % chitosan solution in water were used as coagulation baths for Ti3C2. The chitosan coagulation bath was prepared by dissolving 2.5 g of chitosan (molecular weight 310 000–375 000, Sigma-Aldrich) in a mixture of 5 mL of acetic acid (Sigma-Aldrich) and 500 mL of DI water. Aqueous S-Mo2Ti2C3MXene inks (102 mg mL–1) were spun in an acetone bath. Details of the spinning conditions are listed in Table S1. The shear rate (γ, s–1) during spinning was estimated using eq .where Q is the flow rate (m3 s–1), and R is the inner radius of the needle (m). The extrusion rates during spinning were adjusted to obtain a consistent shear rate of ∼480 s–1 for both 26G and 29G needles. The pure LC MXene fibers did not require any additional steps such as heating, drawing, or rinsing with DI water prior to winding onto a spool.

Characterization

LC MXene fiber cross-section morphology was studied using a field emission scanning electron microscope (SEM, Zeiss SUPRA 55-VP). A transmission electron microscope (TEM, JEM-2100) was used to observe the delaminated Ti3C2 layer, and its crystal structure was investigated using selected area electron diffraction (SAED). X-ray diffraction (XRD) patterns of the MAX powders, MXene films, and wet-spun fibers were obtained using a powder diffractometer (Rigaku SmartLab) equipped with a Cu Kα radiation X-ray source at a 2θ scan step of 0.02° and a scan time of 14 s per step. The size distributions of MXene flakes in the dispersions were estimated using the dynamic light scattering (DLS) technique on a Zetasizer instrument (Malvern Instruments Nano ZS) from an average of five measurements for each sample. The birefringence of MXene dispersions was observed under a polarized optical microscope (Nikon Eclipse 80i) with a homemade cell, shown in Figure S3. Atomic force microscopy (AFM) images of Ti3C2 flakes were obtained using the ScanAsyst scan technology mode of AFM (Bruker, MultiMode 8-HR) with a SCANASYST-AIR tip (tip radius of ∼2 nm). Ti3C2 flakes were deposited on a silicon wafer and were dried under vacuum at room temperature for 6 h before measurement. Thermal analysis was carried out in a thermogravimetric analyzer (TA Instruments Q50 TGA). Approximately 5 mg of samples was heated for each test under nitrogen atmosphere, and the heating rate was at 5 °C min–1. The small- and wide-angle X-ray scattering (SAXS and WAXS) data of the LC MXene fibers were collected on the SAXS/WAXS beamline at the Australian Synchrotron (Melbourne, Australia), using 18.1 keV beam energy (wavelength 0.685 Å) and a Pilatus3-2 M detector. The sample–detector distance was set to ∼420 mm. The data were processed using XMAS (X-ray microdiffraction analysis software) to identify the anisotropic scattering.

The rheological properties of aqueous MXene inks were investigated using a rheometer (TA Instruments AR-G2) with a cone-shaped geometry (angle, 2°; diameter, 40 mm). Approximately 600 μL of ink was loaded with great care to prevent shearing or stretching the sample. Viscosity change as a function of shear rate was measured at shear rates between 0.01 and 1000 s–1 using logarithmic steps. The viscosity of the inks at 0.01 s–1 was taken as the zero-shear rate viscosity, as measuring viscosity at lower shear rates (<0.01 s–1) can be challenging for dilute samples.[49] The viscoelastic properties of Ti3C2 inks were studied by measuring the elastic modulus (G′) and viscous modulus (G″) as a function of frequency at a constant strain amplitude of 0.1%.

The cross-section area of pure LC MXene fibers was estimated from cross-sectional SEM images using ImageJ software. The average cross-section area was used to calculate conductivity and mechanical properties of the fibers. The electrical conductivity of the MXene fibers was measured using a source meter unit (Keysight B2901A) by the aid of a custom-built four-point probe setup with a 2.5 mm probe spacing. The resistance (R) of the fiber was measured from the slope of the I–V curve and then converted to conductivity (σ) using the probe spacing as length (l) and fiber cross-section area (A) through eq .The mechanical properties of the fibers were measured using a universal tensile testing system (Instron 30 kN tensile tester) with a 5 N load cell. Fiber samples were fixed between the grips with a gauge length of 5 mm and tested at a crosshead speed of 0.5 mm min–1 (0.1% min–1).

The electrochemical properties of the fiber electrodes were studied in a three-electrode cell using 1 M H2SO4 electrolyte, a Ag/AgCl reference electrode (3.5 M KCl), and a carbon rod (diameter, 6 mm; length, 2 cm) as the counter electrode. Cyclic voltammetry (CV) and galvanostatic charge–discharge (GCD) curves of the Ti3C2 fibers (∼0.5 cm long) were recorded using an electrochemical workstation (BioLogic SP-300) at scan rates ranging from 5 to 500 mV s–1 (CV) and current densities ranging from 1 to 20 A cm–3 (GCD). Cyclic stability of the MXene fiber was studied over 10 000 cycles at a scan rate of 100 mV s–1. The electrochemical impedance spectroscopy (EIS) measurements were carried out at an open circuit potential (OCP) by applying an alternating current (AC) voltage with an amplitude of 10 mV in a frequency range from 10 mHz to 200 kHz.

The volumetric capacitance (C) and the gravimetric capacitance (CG) of the Ti3C2 fiber were calculated from the CV curves according to eqs and 4, respectively.where I is the current, v is the potential scan rate, ΔU is the potential window, V is the volume, and M is the mass of the LC MXene fiber electrode. To calculate the C, SEM images were used to determine the cross-sectional area of the fibers since it takes into account the irregular fiber surface area, which is not accounted in the case of optical microscopy measurements. The capacitance retention of the fiber electrode (Cr) was calculated from the cyclic stability test through eq by using the specific capacitance (C) in cycle i and the specific capacitance in the first cycle (C1).The electrothermal properties of the L-Ti3C2 fibers spun in a chitosan bath (4.5 cm in length) were measured by an infrared thermal camera (FLIR T430sc) using the setup shown in Figure S11a. Silver paint was used to establish the connection between the fiber and the copper tape to reduce the contact resistance. The power density (PA) of electrothermal conversion was measured using Joule’s law from eq , assuming that the electrical energy was converted to thermal energy completely.where, U is the input voltage, A is the surface area of fiber, and R is the fiber resistance.