Stoichiometry‐Programmed MXenes via Precursor Engineering for High‐Performance EMI Shielding and Energy Storage

ABSTRACT

Device‐level performance in MXenes is dictated by architecture—planar nanosheets are optimal for electromagnetic interference (EMI) shielding, while scrolled structures enhance ion transport for energy storage—particularly when morphology is programmed at synthesis. Whether such architectures can be deterministically encoded through precursor stoichiometry remains unresolved. Here, we demonstrate that precise carbon stoichiometry control in Ti₃AlC _x O_{2‐ x} MAX phases tunes internal lattice strain and thereby directs the emergent MXene architecture. Carbon‐rich precursors (x = 1.94) yield strain‐relieved, high‐crystalline nanosheets with metallic conductivity (∼23 300 S cm⁻¹), enabling ultrathin films with record‐high EMI shielding performances across X‐ and W‐bands (≥ 2.0 × 10⁶ dB cm² g⁻¹ at 8.2 GHz for 29 nm; 108 dB at 100 GHz for 8 µm) and robust W‐band retention after 5,000 bending cycles (r = 2.5 mm). In contrast, carbon‐deficient precursors (x = 1.71) introduce lattice compression and oxygen substitution, triggering spontaneous scrolling upon delamination. The resulting nanoscrolls offer exceptional ion accessibility, achieving 657 F g⁻¹ at 2 mV s⁻¹ with 99.4% retention over 12 000 cycles. This stoichiometry‐programmed approach establishes a synthesis‐stage lever linking MAX chemistry to MXene architecture and function, enabling application‐specific architecture design within established MAX/MXene synthesis and solution‐processing workflows for next‐generation electronics and energy storage.

Precursor carbon stoichiometry programs internal strain and oxygen substitution in Ti₃AlC _x O_{2‐ x} , deterministically switching Ti₃C _x O_{2‐ x} T _z MXenes between highly aligned 2D nanosheets and spontaneous 1D nanoscrolls. This synthesis‐stage architecture control enables ultrathin X/W‐band EMI shielding with outstanding mass‐normalized performance and bending durability, and high‐rate supercapacitor performance with robust cycling stability, providing an application‐specific design lever within established MXene processing workflows.

1. Introduction

Controlling dimensionality within a single material unlocks emergent properties that surpass those of any fixed form. For instance, 2D graphene offers high carrier mobility, mechanical flexibility, and optical transparency [1], whereas rolling it into 1D carbon nanotubes introduces chirality‐dependent metallic or semiconducting behavior—expanding its functionality in optoelectronics and energy storage [2]. Similarly, 2D van der Waals (vdW) nanosheets provide tunable interlayer spacing and band structure for electronic and catalytic applications [3], while their 1D analogues exhibit semiconductor‐to‐metal transitions and enhanced nonlinear optical responses, broadening their utility in quantum photonics and electronics [4, 5]. These examples underscore the power of deliberate dimensional design to amplify performance and diversify applications.

Since their discovery in 2011, MXenes (M_n+1X_nT _z , n = 1–4; M = early transition metal; X = C/N/O; T _z = surface terminations) have emerged as a rich family of 2D materials, typically derived by etching A‐layers (e.g., Al, Si) from MAX phases [6]. Among them, Ti₃C₂T _z is most widely studied due to its high metallic conductivity, surface tunability, and solution processability [7, 8], which enable solution‐based fabrication of transparent electrodes [9], sensors [10], and electromagnetic interference (EMI) shielding films [11]. Yet despite their performance, MXene‐based EMI shields must meet increasing frequency demands in the millimeter‐wave regimes (30–300 GHz), where legacy metal and ferritic shields fail. High‐frequency performance requires ultrathin, highly aligned, and flexible films with minimum voids and junction resistance. In this context, total shielding effectiveness (SE_T) arises from a synergy between surface reflection (SE_R) from impedance mismatch and electromagnetic absorption (SE_A) via dielectric and magnetic losses, all governed by the in‐plane conductivity, flake alignment, and defect density [12]. However, flake‐level irregularities—such as small size, wrinkles, or partial scrolling—increase interflake resistance and degrade shielding performance [13]. While recent coating and post‐alignment strategies improve morphology to some extent [14, 15, 16], they cannot fundamentally reverse poor flake quality inherited from synthesis. Achieving W‐band‐relevant performance thus demands a morphology‐first, synthesis‐stage approach that yields large, flat, high‐crystalline MXene nanosheets optimized for directional stacking and durability under strain.

In contrast, rolling MXene sheets into 1D architectures can unlock properties inaccessible to planar stacks, particularly for ion transport in electrochemical systems. Conventional MXene electrodes, comprised of densely restacked nanosheets, restrict electrolyte access and slow diffusion, limiting rate capability. While 3D structuring, spacer integration, or surface modification can improve performance [17, 18, 19], they often introduce additional steps, scalability challenges, or reduced conductivity. Scroll‐like MXenes offer an appealing alternative because they can create open and permeable pathways. However, scalable and controllable formation of such architectures remains challenging, because most reported routes are post‐synthetic and additive‐assisted. For example, ascorbic‐acid intercalation [20] can drive MXenes into scrolled fibers, but the resulting conductivity has been reported to be below 10 S cm⁻¹, likely due to residual intercalants and disrupted electronic transport. Cryo‐assisted surfactant templating [21] likewise relies on surfactant treatment followed by ice‐crystal‐driven deformation, which increases process complexity and may leave residual species that hinder conductivity and scalability. More recently, Zhao et al. reported metal‐vacancy‐induced scroll formation in V₂CT _x using TBPH‐assisted rolling, yielding non‐vdW superlattices with record X‐band shielding and high conductivity; however, this approach relies on multi‐step processing and bulky organic intercalants, and its application remains confined to EMI shielding [22]. Overall, there remains a critical need for a precursor‐guided and process‐compatible strategy that can deterministically program MXene architecture during synthesis—without the use of additives or post‐synthetic treatments—and tailor structure to functional requirements across diverse applications.

In this study, we address this challenge by tuning the carbon stoichiometry (x) in Ti₃AlC _x O_{2‐ x} MAX precursors within a narrow (<15%) window, systematically mapping its effect on lattice strain, oxygen incorporation, and the resultant Ti₃C _x O_{2‐ x} T_z MXene morphology (Figure 1). While our prior work advanced broadband EMI shielding primarily by enhancing intrinsic conductivity through X‐site N substitution in solid‐solution MAX/MXenes [8], the present study introduces a distinct synthesis‐stage lever. Here, precursor carbon stoichiometry programs internal strain and oxygen substitution, thereby deterministically switching MXene architecture (flat nanosheets vs. nanoscrolls), enabling application‐specific performance in both high‐frequency EMI shielding and electrochemical energy storage. Carbon‐rich precursors produce relaxed lattices that delaminate into large, flat Ti₃C₂T_z nanosheets with high crystallinity and electrical conductivity (∼23 000 S cm⁻¹), enabling ultrathin films with best‐in‐class EMI shielding across X‐ and W‐bands (e.g., specific shielding effectiveness per unit thickness (SSE/t) ≥ 2.0 × 10⁶ dB·cm²·g⁻¹ at 8.2 GHz for 29 nm; SE_T ≈108 dB at 100 GHz for 8 µm; robust W‐band retention after 5000 bending cycles). In contrast, carbon‐deficient precursors induce lattice compression and oxygen substitution at X‐sites, driving spontaneous scrolling upon delamination and yielding open, permeable MXene nanoscrolls that achieve a record‐level gravimetric capacitance of 657 F g⁻¹ at 2 mV s⁻¹ in 3 m H₂SO₄, with 99.4% retention over 12 000 cycles. This composition‐directed framework offers a synthesis‐stage lever for deterministic shape engineering of MXenes—flat sheets for high‐frequency EMI, scrolls for high‐rate energy storage—bridging MAX precursor chemistry, lattice physics, and device‐level performance without introducing additional post‐synthesis unit operations beyond standard MXene processing.

FIGURE 1.

Schematic illustrations of composition‐directed architecture control in Ti₃C _x O_{2‐ x} T _z MXenes and their application‐specific advantages.

2. Results and Discussion

2.1. Influence of Initial Graphite Amount on Structural Properties of MAX Phases

MAX precursors were prepared by solid‐state sintering of pressed pellets with stoichiometric Ti (3 mol) and Al (1.2 mol) and a systematically varied graphite amount (1.7–2.2 mol) at 1550°C for 2 h in a closed crucible (Figure 2a). The furnace was kept under Ar to prevent oxygen ingress. X‐ray diffraction (XRD) confirmed Ti₃AlC₂ across all conditions (Figure 2b; Figure S1). At ≤1.8 mol graphite, Ti₂AlC appeared (i.e., incomplete Ti₃AlC₂ formation), whereas ≥ 1.9 mol produced more TiC precipitates [23]. Scanning electron microscopy with energy dispersive X‐ray spectroscopy (SEM‐EDS) showed that the actual carbon fraction x increased approximately linearly from 1.71 to 1.94 as the graphite increased from 1.85 to 2.05 mol, and then slightly decreased at higher loadings due to carbon oversaturation and phase segregation (Figure 2c; Table S1). Thus, the initial graphite amount directly sets the X‐site carbon occupancy and the final MAX stoichiometry. Consistently, both a‐ and c‐lattice constants increased with graphite content, peaking at 2.05 mol (a = 3.088 Å, c = 18.633 Å at x = 1.94) and then declining (Figure 2d). Such lattice contraction has been attributed to oxygen incorporation at X‐sites in MAX phases [24, 25, 26, 27, 28]. The experimentally measured unit‐cell volumes as a function of carbon content agreed well with density functional theory (DFT) calculations, supporting assignment to Ti₃AlC _x O_{2‐ x} (Figure 2e; Figure S2). For clarity, the MAX composition is hereafter denoted Ti₃AlC _x O_{2‐ x} .

FIGURE 2.

Structural characterization of Ti₃AlC _x O_{2‐ x} precursors with varying carbon content. (a) Schematic of the MAX phase synthesis and atomic models showing the effect of excess and deficient carbon on vacancy formation. (b) Phase fractions for different graphite amounts (1.7–2.2 mol) determined by Rietveld refinement of XRD patterns. (c) Carbon content (x) measured by SEM‐EDS. (d) a‐ and c‐lattice parameters from XRD for different graphite amounts. (e) Experimental and DFT‐calculated unit cell volumes as a function of x in Ti₃AlC _x O_{2‐ x} . (f,g) SEM images of Ti₃AlC _x O_{2‐ x} precursors synthesized at (f) x = 1.94 and (g) x = 1.71 (scale bars: 5 µm). (h,i) XPS spectra of (h) Ti 2p and (i) Al 2p regions for Ti₃AlC _x O_{2‐ x} . (j) Raman spectra of Ti₃AlC _x O_{2‐ x} .

SEM images of sintered pellets clarified oxygen incorporation (Figure S3). Carbon‐deficient samples (x = 1.71 and 1.82) contained voids that open pathways for oxygen and favor Ti₃AlC _x O_{2‐ x} , whereas the carbon‐rich sample (x = 1.94) showed TiC precipitates filling intergranular voids, effectively blocking oxygen diffusion and suppressing X‐site substitution. Morphologically (Figure 2f, g; Figure S4), the sample with x = 1.94 formed compact, well‐faceted platelets with a preferred orientation, whereas x = 1.71 exhibited terraced, stepped‐layer grains. This contrast is attributed to abundant Ti‐Al liquid phase under carbon‐deficient conditions, which drives dissolution‐precipitation and layer‐by‐layer nucleation during MAX phase evolution [29].

X‐ray photoelectron spectroscopy (XPS) further supported these trends (Figure 2h,i; Figure S5). As x decreased from 1.94 to 1.71, Ti 2p spectra showed an increased Ti‐O fraction with a Ti‐C blueshift from 454.28 to 454.44 eV, and C 1s (carbide carbon) blue‐shifted from 281.18 to 281.46 eV, indicating enhanced surface oxygen and weakened Ti–C stabilization under relatively oxygen‐rich conditions. Conversely, as x increased, in Al 2p the metallic Al peak decreased, while Al³⁺ shifted from 73.97 to 73.86 eV. Raman spectroscopy corroborated structural stiffening (Figure 2j). With increasing x, ω₂ peak (in‐plane Ti‐C) intensified [30], ω₃ diminished (c‐axis preferred orientation), and ω₄/ω₆ peaks blue‐shifted (from 270 to 276 cm⁻¹ and 655 to 660 cm⁻¹, respectively), consistent with stronger Ti‐C bonds due to reduced oxygen and X‐site vacancies [31]. By contrast, ω_5a,_5b (Al‐related) peaks red‐shifted (from 630 to 624 cm⁻¹).

2.2. Carbon Stoichiometry‐Driven Morphological Transformation in MXenes

Ti₃C _x O_{2‐ x} T _z MXenes were obtained by selectively etching Ti₃AlC _x O_{2‐ x} in a mixed HF/HCl solution, followed by LiCl‐assisted delamination. After washing, a dark‐green dispersion of delaminated MXene was collected. The etching time increased from 18 h (x = 1.71) to 24 h (x = 1.94), indicating slower kinetics in carbon‐rich precursors consistent with lower oxygen content and higher structural stability (Figure S6). UV–vis spectra showed x‐dependent dispersion differences, whereas Ti oxidation states were essentially unchanged among samples by XPS (Figures S7 and S8). Although etched powders exhibited similar accordion‐like multilayers, the delaminated architectures diverged strongly with x (Figure 3a–d; Figures S9 and S10). Ti₃C_1.94O_0.06T _z (x = 1.94) produced flat nanosheets with ∼1.3 nm monolayer thickness (Figure 3e). In contrast, as x decreased, the formation of nanoscrolls increased: Ti₃C_1.71O_0.29T _z (x = 1.71) formed spiral nanoscrolls (0.5–3 µm width; ∼11 nm central thickness (∼8–9 layers via four turns)) (Figure 3f; Figure S11).

FIGURE 3.

Carbon‐stoichiometry‐driven structural transformation of Ti₃C _x O_{2‐ x} T _z MXenes. (a–c) Optical microscopy images of delaminated MXene flakes at (a) x = 1.71, (b) x = 1.82, and (c) x = 1.94 (insets: magnified views). (d) Quantification of flake morphologies (≥200 flakes per batch, 3 independent batches). (e, f) AFM images of (e) nanosheets (x = 1.94) and (f) nanoscrolls (x = 1.71), with corresponding thickness profiles (insets). (g) Raman spectra of multilayer and delaminated MXenes at x = 1.71 and x = 1.94. (h) XRD patterns of MXene films. (i,j) Wide‐angle X‐ray diffraction patterns acquired with the X‐ray beam perpendicular to the film surface for (i) x = 1.94 and (j) x = 1.71; (k) corresponding 1D profiles.

To elucidate the mechanism underlying spontaneous scrolling, XRD and SEM analyses were performed to probe the delamination pathway and interlayer coupling (Figures S12 and S13). XRD reveals that both compositions (x = 1.71 and 1.94) undergo a two‐step interlayer expansion during the transition from the etched multilayer state to the delaminated MXene. After etching, the (002) basal spacing increases to 9.97 Å for x = 1.71 and 10.04 Å for x = 1.94. Upon LiCl–induced delamination, the spacing further expands to 12.76 and 12.65 Å, respectively, consistent with progressive hydration and ion intercalation that weakens interlayer coupling. Notably, although x = 1.71 exhibits a slightly smaller basal spacing in the multilayer state, it undergoes a larger expansion upon delamination, suggesting more substantial structural relaxation once interlayer constraints are sufficiently reduced. This behavior reflects a greater release of residual strain and indicates weaker effective interlayer adhesion in the delaminated state. SEM snapshots of intermediate morphologies during delamination further reveal distinct, composition‐dependent pathways. The x = 1.94 sample primarily exhibits swelling‐driven, accordion‐like expansion of multilayer stacks followed by planar layer separation, with no evidence of edge curling. In contrast, the x = 1.71 sample frequently shows edge‐initiated curling with a typical curvature radius of ∼500 nm, occurring prior to complete flake separation. This morphology is indicative of a bending‐mediated strain‐release mechanism, which precedes and likely facilitates full scrolling. Consistently, the energetic analysis presented in Note S1 suggests that, for x = 1.71, the release of stored in‐plane elastic energy upon scrolling can outweigh the bending energy cost (Figure S14).

Raman spectroscopy captures this transformation (Figure 3g). In the multilayer state, x = 1.94 showed a sharp A_1g(Ti, C, T _z ) peak at 209.4 cm⁻¹ (full‐width at half maximum (FWHM) = 8 cm⁻¹), evidencing high crystallinity and low residual strain, whereas x = 1.71 displayed a broader feature at 209.1 cm⁻¹ (FWHM = 13 cm⁻¹). After delamination, x = 1.94 retained a relatively narrow peak at 210 cm⁻¹ (FWHM = 12 cm⁻¹), while x = 1.71 broadened and shifted to 205.8 cm⁻¹ (FWHM = 21 cm⁻¹). For the A_1g(C) mode, both compositions exhibited a blueshift upon delamination—from 721.7 to 725.8 cm⁻¹ for x = 1.71 and from 719.0 to 723.8 cm⁻¹ for x = 1.94. The pronounced peak broadening and distinct strain responses of A_1g(Ti, C, T _z ) and A_1g(C) in the carbon‐deficient sample (x = 1.71) suggest that delamination relaxes pre‐existing anisotropic strain and induces localized disorder, thereby promoting scrolling through out‐of‐plane bending. The mechanism extends beyond carbides: Ti₃AlC_1.1N_0.9 (nitrogen‐rich carbonitride MAX) exhibits > 95% nanoscrolls due to severe precursor contraction (Figure S15). Thus, X‐site chemistry functions as a primary control parameter for MXene architecture.

Vacuum‐filtered films exhibited dominant (00l) reflections with high phase purity (Figure 3h). The (002) peak shifted from 6.92° to 6.98° as x increased from 1.71 to 1.94. Differences in film alignment correlated with morphology in wide‐angle XRD (WAXD) measurements conducted with the incident beam oriented either parallel or perpendicular to the film plane (Figures S16). Under parallel incidence, both x = 1.94 and 1.71 exhibited sharp (00n) reflections (Figures S17). When the beam was oriented perpendicular to the plane, the x = 1.94 film displayed distinct (010) and (110) peaks, characteristic of the P6₃/mmc Ti₃C₂T _z and indicative of highly aligned, densely packed flakes lying parallel to the film plane, whereas x = 1.71 showed a dominant (00n) reflection consistent with curved or scrolled architectures that introduce out‐of‐plane periodicity (Figure 3i–k).

Although both films appeared well‐layered before annealing, differences amplified afterward (Figure S18). The Ti₃C_1.94O_0.06T _z film retained compact, ordered stacking, while the Ti₃C_1.71O_0.29T _z film developed substantial porosity and enlarged voids, attributable to hindered out‐diffusion of gaseous by‐products within curved pathways, causing localized gas buildup and larger pores [32].

2.3. EMI Shielding Performance of MXene Films

The Ti₃C_1.94O_0.06T _z film shows higher tensile strength and stiffness than Ti₃C_1.71O_0.29T _z film, consistent with superior flake alignment (Figure S19). After annealing, electrical conductivity spans 12 200 S cm⁻¹ (x = 1.71) to 23 300 S cm⁻¹ (x = 1.94) (Figure 4a), attributed to reduced inter‐flake junction resistance and a lower density of X‐site defects [33].

FIGURE 4.

Electrical properties and EMI shielding performance of Ti₃C _x O_{2‐ x} T _z MXene films. (a) Electrical conductivities of vacuum‐filtered MXene films as a function of x before and after annealing. (b) Total shielding effectiveness (SE_T) vs. thickness for spray‐coated MXene films on PEN, measured at 8.2 and 100 GHz after annealing at 150°C; dashed lines are transfer‐matrix fits. (c) W‐band EMI SE_T values of Ti₃C_1.94O_0.06T _z before and after 2000 and 5000 bending cycles (bending radius 2.5 mm). (d,e) EMI SE_T for annealed MXene films at (d) 8.2 GHz and (e) 100 GHz. (f) Comparison of EMI SE components for 8 µm‐thick MXene films in the W‐band. (g) Benchmark of SSE/t in the X‐band for Ti₃C_1.94O_0.06T _z (inset: magnified view). (h) Benchmark of SSE/t in the W‐band for Ti₃C_1.94O_0.06T _z (Table S3).

To probe ultrathin (<80 nm) MXene films, Ti₃C _x O_{2‐ x} T _z was spray‐coated onto polyethylene naphthalate (PEN) and annealed at 150°C. Film thickness was measured via the optical‐conductivity relation (σ_op = 1,041 S cm⁻¹). As expected, increasing thickness (t) lowers transmittance and raises shielding (Figure 4b; Figures S20 and S21). Because t ≪ δ (skin depth), total shielding effectiveness (SE_T) is nearly identical in X‐ and W‐bands, matching transfer‐matrix calculations (dashed lines in Figure 4b). Compositions with larger x exhibit slightly higher SE_T due to higher electrical conductivity. For Ti₃C_1.94O_0.06T _z , increasing t from 3 to 76 nm raises SE_T at 8.2 GHz from 1.4 to 28 dB; notably, 29 nm film achieves SE_T = 21 dB (≈ 99%), meeting commercial EMI requirements. In the W‐band, the reflection component (SE_R) increases from 6.9 dB (79.6% reflection at 21 nm) to 11.66 dB (93.2% reflection at 76 nm) (Figure S22), due to the high conductivity and compact stacking that enhance impedance mismatch. Owing to its highly ordered lamellar morphology, a 53‐nm‐thick‐Ti₃C_1.94O_0.06T _z film retains 23 dB after 5000 bending cycles (radius 2.5 mm), with only ∼3 dB loss, highlighting durability for flexible and wearable use (Figure 4c).

For vacuum‐filtered films, the thickness dependence of EMI SE shows a nonlinear increase (Figure 4d,e; Figures S23 and S24). Ti₃C_1.94O_0.06T _z films consistently outperform across thickness and frequency, benefiting from high electrical conductivity and internal scattering within multilayered stacks. At larger t, measured SE_T can fall below transfer‐matrix predictions—most notably for Ti₃C_1.71O_0.29T _z samples—due to porosity growth after annealing that limits shielding efficiency (Figure S18). In the W‐band, Ti₃C_1.94O_0.06T _z film reaches the equipment detection limit near 8 µm, with comparable SE_R ≈ 23 dB but much higher absorption shielding effectiveness (SE_A ≈ 85 dB) than other compositions (Figure 4f; Figure S25).

For lightweight devices, the specific shielding effectiveness per unit thickness (SSE/t) in the X‐band is a key metric (Figure 4g). SSE/t was calculated based on the apparent film density, which was approximately 3.7 g·cm⁻³ for spray‐coated films and 2.9 g·cm⁻³ for vacuum‐filtered films. Among materials with SE_T >20 dB (a common commercial threshold), the carbon‐rich MXene films achieved a record SSE/t of 2.0 × 10⁶ dB cm² g⁻¹ at 29 nm, representing one of the highest values reported to date. Table S2 benchmarks representative ultrathin MXene‐, metal‐, and carbon‐based EMI shielding films measured in the X‐band (8.2–12.4 GHz), and shows that our carbon‐rich MXene film ranks among those with the highest SSE/t values.

When normalized by thickness (SE/t), the X‐band performance rivals that of metals, while in the W‐band it surpasses all previously reported materials (Figure S26). Figure 4h presents a benchmark comparison of W‐band SSE/t. In the largely unexplored regime of ultrathin W‐band shielding, Ti₃C_1.94O_0.06T _z demonstrates superior shielding capability, confirming its potential for advanced high‐frequency applications (Table S3). Importantly, an expanded comparison of the limited W‐band studies that also consider mechanical durability highlights that our ultrathin Ti₃C_1.94O_0.06T _z film uniquely combines high EMI shielding effectiveness with exceptional mechanical reliability. Collectively, carbon stoichiometry and film morphology enable solution‐processable ultrathin MXene coatings/films with record‐level EMI performance up to 110 GHz.

2.4. Electrochemical Performance of MXene Films

To assess supercapacitor performance, MXene electrodes (x = 1.71–1.94) were evaluated in a three‐electrode cell (−0.6 to 0.2 V vs. Ag/AgCl) using 3 m H₂SO₄. All electrodes were prepared with an identical areal MXene loading of 1 mg cm⁻², resulting in comparable film thicknesses across compositions (Figure S27). The acidic electrolyte (smallest cation, H⁺) affords excellent ion conductivity for high electrochemical performance [34]. At 10 mV s⁻¹, cyclic voltammetry (CV) curves of all samples exhibit a pair of broad redox peaks (cathodic ≈ −0.3 V, anodic ≈ −0.25 V), indicative of pseudo‐capacitance (Figure 5a). These features are consistent with proton‐coupled pseudocapacitive processes involving reversible redox reactions at oxygen‐containing surface terminations (─O, ─OH) on MXene. The gravimetric capacitance decreases with x, from 601.7 F g⁻¹ (x = 1.71) to 363.3 F g⁻¹ (x = 1.94). Across scan rates (Figure 5b; Figure S28), the Ti₃C_1.71O_0.29T _z electrode delivers the highest capacitance. Consistently, the areal and volumetric capacitances exhibit the same trend, with the volumetric capacitance reaching 1972 F cm⁻³ for Ti₃C_1.71O_0.29T _z at 2 mV s⁻¹ (Figure S29). This improvement is primarily attributed to its nanoscrolled (more open/porous) morphology, which increases the electrochemically accessible interfacial area, shortens ion‐diffusion lengths, and provides more open transport pathways—effects that are commonly observed in porous MXene architectures (Figure S30). In contrast, the densely stacked nanosheets at higher x impose more tortuous ion pathways, limiting electrolyte access to internal active sites. Consistently, galvanostatic charge–discharge (GCD) at 10 A g⁻¹ show the same trend, with the longest discharge times at x = 1.71. The coulombic efficiencies for x = 1.71–1.94 are 92%, 89%, 92%, 92%, and 87 % (Figure 5c), indicative of highly reversible charge storage.

FIGURE 5.

Electrochemical performance of Ti₃C _x O_{2‐ x} T _z MXene films in 3 M H₂SO₄. (a) Cyclic voltammetry (CV) curves at 10 mV s⁻¹. (b) Gravimetric capacitance vs. scan rate (2–200 mV s⁻¹) (inset: schematic ion‐transport pathways). (c) Galvanostatic charge–discharge (GCD) profiles at 10 A g⁻¹. (d) Nyquist plots (inset: high‐frequency region). (e) Capacitive/diffusive charge contributions vs. scan rate. (f) CV analysis at 100 mV s⁻¹ with hatched regions indicating capacitive contributions. (g) Comparison of specific capacitances at different scan rates of Ti₃C_1.71O_0.29T _z against reported MXene‐based‐electrodes. (h) Cyclic stability at 20 A g⁻¹ over 12 000 cycles (inset: representative GCD profiles at the beginning and after cycling).

Electrochemical impedance spectroscopy (EIS) elucidates composition‐dependent kinetics (Figure 5d). Nyquist plots (0.1–10⁵ Hz) display a high‐frequency semicircle and a low‐frequency inclined line. In the high‐frequency regime, the x‐axis intercept of the semicircle corresponds to the solution resistance (R_s), while the semicircle diameter yields the charge‐transfer resistance (R_ct), reflecting interfacial electron transfer. In the low‐frequency regime, the nearly vertical line indicates capacitive behavior with low diffusion resistance. The R_s is nearly composition‐independent (0.73‐0.93 Ω·cm²), whereas the R_ct varies (Table S4): x = 1.71 and 1.76 exhibit low R_ct (0.0829 and 0.0875 Ω·cm², respectively) and steep low‐frequency branches, reflecting minimal diffusion limitations. As x increases to 1.94, R_ct rises (up to 0.1219 Ω·cm²), and the low‐frequency slope decreases, indicating that the nanosheet architecture restricts ion pathways and compromises high‐rate performance.

To assess the charge‐storage kinetics of the Ti₃C_1.71O_0.29T _z electrode, we analyzed the peak‐current (i_p ) dependence on scan rate (v) using the power–law Equation (1):

$$\begin{array}{c}{i}_p=a{v}^b\end{array}$$ (1)

where a and b are obtained from the fit, with b typically between 0.5 and 1. A b value near 0.5 indicates diffusion‐limited behavior, whereas a value near 1 indicates capacitive control [35]. The b value was determined from the slope of the log i vs. log v plot; the log‐log fit over 5–200 mV s⁻¹ yielded b close to 1, indicating predominantly surface‐controlled capacitive behavior (Figure S31). Consistently, the CV current was deconvoluted as:

$$\begin{array}{c}i\left(V\right)=k_{1}v+k_2{v}^{\frac{1}{2}}\end{array}$$ (2)

where k ₁ v is the capacitive component and k ₂ v ^1/2 is the diffusion‐controlled component. The surface‐capacitive contribution increases from 59.3% at 2 mV s⁻¹ to 93.6% at 200 mV s⁻¹, confirming that higher scan rates increasingly favor surface‐controlled charge storage (Figure 5e,f).

To further corroborate the charge‐storage mechanism, ex situ XRD and XPS were performed at five representative potentials along a single CV cycle (10 mV s⁻¹) (Figure S32). XRD reveals a reversible shift in the (002) reflection, indicating potential‐dependent and reversible modulation of the interlayer spacing. Notably, the magnitude of interlayer expansion/contraction is greater for Ti₃C_1.71O_0.29T _z than for Ti₃C_1.94O_0.06T _z (Figure S32b,c), suggesting a stronger structural response in the former. In contrast, the Ti 2p spectra remain essentially unchanged across the sampled states (Figure S32d), indicating that variations in Ti oxidation states are not the dominant contributor to charge storage under the tested potential range. Together, these findings support a mixed charge‐storage mechanism, comprising electric double‐layer charging and a pseudocapacitive contribution that is coupled to a reversible interlayer structural response.

Figure 5g benchmarks the specific capacitance and rate capability of the Ti₃C_1.71O_0.29T _z against previously reported MXene‐based supercapacitor electrodes, demonstrating its superior performance [18, 34, 35, 36, 37, 38, 39, 40]. Cycling stability of Ti₃C_1.71O_0.29T _z electrode, assessed at 20 A g⁻¹ over 12 000 charge–discharge cycles, shows 99.4% capacitance retention, indicating excellent durability. Post‐cycling characterization shows that the electrode surface remains essentially unchanged, with no obvious cracking or delamination (Figure S33). Table S5 summarizes the specific capacitance and cycling stability of representative MXene‐based electrodes. Among them, Ti₃C_1.71O_0.29T_z shows competitive capacitance and robust long‐term cycling stability. This comparison further underscores that many prior studies rely on post‐synthetic templating, hybridization, or additional structural engineering steps to enhance capacitance. In contrast, our electrodes achieve high gravimetric and volumetric capacitance through a precursor‐guided, synthesis‐stage morphology‐programming strategy, without the need for external additives or post‐processing. Overall, these results demonstrate that precursor stoichiometry‐programmed morphology control enables the formation of nanoscrolled MXene architecture with enhanced ion accessibility and transport properties, leading to improved electrochemical performance within the established capacitive (pseudo‐capacitive) charge‐storage framework.

Finally, a consolidated comparison across the full x range (Figure S34) reveals a clear trade‐off between the two functions: higher x enhances electrical conductivity and flake alignment, thereby improving EMI shielding performance, whereas lower x promotes scrolling and open transport pathways, resulting in higher capacitance. Notably, intermediate compositions define a practically relevant regime in which both shielding effectiveness and capacitance remain at functionally acceptable levels, further validating the stoichiometry‐programmed architecture framework.

3. Conclusion

In summary, we present a composition‐programmed morphology control strategy that enables deterministic switching between flat 2D Ti₃C₂T_z nanosheets and spontaneously scrolled 1D nanoscrolls by precisely tuning the carbon stoichiometry (x) in the Ti₃AlC _x O_{2‐ x} MAX precursor. This synthesis‐stage approach is fundamentally different from prior methods relying on post‐synthesis treatments or templating, offering a process‐compatible and generalizable pathway to architecture control. The resulting MXene films demonstrate best‐in‐class performance across two distinct application domains: ultrathin EMI shielding, achieving up to 108 dB at 100 GHz (with SSE/t ≥ 1.0 × 10⁶ dB·cm²·g⁻¹), and nanoscrolled electrodes delivering 657 F g⁻¹ at 2 mV s⁻¹ with 99.4% retention over 12 000 cycles. Beyond advancing MXene synthesis and processing, this study establishes a general framework for encoding macroscopic functionality through precursor stoichiometry—bridging atomic‐scale chemistry, nanoscale architecture, and device‐level performance. This paradigm opens new opportunities for the rational design of 2D materials tailored to high‐frequency electronics, energy storage, and multifunctional applications.

4. Experimental Section

4.1. Materials

Ti (‐325 mesh, 99.5%), Al (‐325 mesh, 99.5%), graphite (‐325 mesh, 99.9995%), and HF solution (48%–51%) were purchased from Thermo Scientific. LiCl (powder, 99%), HCl solution (37%), and nafion solution (5 wt.%) were purchased from Sigma–Aldrich. All materials were used as received without further purification.

4.2. Synthesis of Ti₃AlC _x O_{2‐ x} MAX Phase

For Ti₃AlC _x O_{2‐ x} synthesis, Ti, Al, and graphite powders were weighed at molar ratios of 3: 1.2: (1.7–2.2), respectively. The powders were mixed with ethanol and ball‐milled at 250 rpm for 24 h, followed by vacuum filtration and air drying for 12 h. The dried mixtures were cold‐pressed into 15 mm‐diameter pellets and sintered in lidded alumina crucibles under flowing argon at a heating rate of 5°C min⁻¹ up to 1,550°C for 2 h. The resultant pellets were ground and sieved through a 400‐mesh screen. For Ti₃AlC_1.1N_0.9, 3 mol Ti, 0.3 mol Al, 0.9 mol AlN, and 1.1 mol C were used, and the same synthesis procedure was applied.

4.3. Synthesis of Ti₃C _x O_{2‐ x} T _z

MXene nanosheets were synthesized by HF/HCl etching of Ti₃AlC _x O_{2‐ x} and Ti₃AlC_1.1N_0.9, followed by Li⁺‐assisted self‐delamination, as previously reported. For Al removal from MAX phase, 1 g of the precursor was slowly added to a mixed etchant consisting of 4 mL HF, 13 mL HCl, and 13 mL deionized (DI) water, and the suspension was stirred at 35°C for a composition‐dependent duration. The resulting etched mixture was transferred into a 250 mL conical tube, diluted with DI water, and washed by centrifugation at 3500 rcf for 5 min until the supernatant reached pH ≈ 6. The sediment was then resuspended in a LiCl solution prepared by dissolving 2 g LiCl in 10 mL DI water and vortex‐mixed for 1 h to promote Li⁺ intercalation. Subsequent centrifugation at 3500 rcf for 5 min produced a dark supernatant, indicating spontaneous delamination of MXene flakes. Residual multilayer material was removed by centrifugation at 2500 rcf for 20 min, and the dark supernatant was collected in a 250 mL flat tube. The dispersion was concentrated by repeated centrifugation at 8000 rcf for 20 min until the supernatant became sufficiently clear. The final concentrate yielded high‐purity, few‐layer MXene.

4.4. Fabrication of MXene Films

Spray coating was used to prepare ultrathin films (<80 nm) on polyethylene naphthalate (PEN), while vacuum‐assisted filtration produced freestanding, flexible films with thicknesses of 0.5–15 µm. Spray‐coated films on PEN were annealed at 150 °C to remove residual solvents and reinforce interlayer cohesion without damaging the substrate.

4.5. Computational Methods

DFT calculations were performed with QuantumATK [41] to compare Ti₃AlC _x O_{2‐ x} compositions over varying x. For all systems, geometry optimizations employed a linear combination of atomic orbitals (LCAO) basis, norm‐conserving PseudoDojo pseudopotentials, and the generalized gradient approximation (GGA) in the Perdew‐Burke‐Ernzerhof (PBE) form [42, 43]. Convergence thresholds for the total energy and forces were set to 10⁻⁴ eV and 0.01 eV/Å, respectively. Ti₃AlC _x O_{2‐ x} structures were constructed using a 2 × 5 × 1 supercell of the hexagonal Ti₃AlC₂ primitive cell, and carbon atoms were randomly substituted by oxygen to realize x = 1.6, 1.7, 1.8, 1.9, 1.95, and 2.0. The real‐space density mesh cutoff was 125 Hartree, and a 5 × 2 × 2 Monkhorst‐Pack k‐point grid was used for geometry optimization [44].

4.6. Characterizations

XRD patterns were recorded on a Rigaku D/MAX2500V/PC with Cu K_α radiation (40 kV, 200 mA) at a step size of 0.02°. Rietveld refinement was performed using FullProf suite with the WinPLOTR interface to determine phase fractions and lattice parameters. Sample morphologies were examined by SEM (SU7000, Hitachi) equipped with an EDS detector. Thicknesses of vacuum‐filtered films were measured from cross‐sectional SEM images using ImageJ. High‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) and high‐resolution transmission electron microscopy (HRTEM) were performed using an aberration‐corrected TEM (FEI Titan G2 60–300) operated at an accelerating voltage of 200 kV. Raman spectra of MAX and MXene were acquired on a WiTec alpha300S with a 532 nm excitation (1 mW). AFM images were obtained in tapping mode using a Bruker Dimension system. XPS was performed on a ThermoFisher ESCALAB 250XI with a micro‐focused monochromatic Al X‐ray source under ultrahigh vacuum. Sheet resistance was measured using a four‐point probe (CMT2000N, AIT) using highly p‐doped Si as a standard substrate. Optical properties were recorded using UV–vis NIR spectrometer (Cary 5000). WAXD was performed at the 6D UNIST‐PAL beamline of Pohang Light Source‐II (Pohang Accelerator Laboratory, Korea). X‐rays (18.986 keV) from a bending magnet were monochromatized with a Si(111) double‐crystal monochromator and focused by a toroidal mirror. Diffraction patterns were collected on a charge‐coupled device (CCD) detector (MX225‐HS, Rayonix, USA) at a sample‐detector distance of 242.9 mm with 100 s exposure, adjusted to avoid detector saturation. Diffraction angles were calibrated against the NIST SRM 660b LaB₆ standard. EMI shielding was evaluated using a two‐port vector network analyzer (VNA, N5247A, Agilent), a millimeter head controller (N5261A, Agilent), a frequency extender (N5260‐60003, Agilent), and rectangular waveguides for the X‐ and W‐bands, with standard Thru, Reflect, and Line (TRL) calibration.

4.7. EMI Shielding Calculations and Transfer Matrix Simulations

The vector network analyzer (VNA) provides the complex scattering parameter S _ij, from which the power reflection (R) and transmission (T) coefficients are calculated as:

$$\begin{array}{c}R={\left|S_{11}\right|}^2={\left|S_{22}\right|}^2\end{array}$$ (3)

$$\begin{array}{c}T={\left|S_{21}\right|}^2={\left|S_{12}\right|}^2\end{array}$$ (4)

The total shielding effectiveness (SE_T) is the sum of reflection, absorption, and multiple internal reflections:

$$\begin{array}{c}\mathrm{S}{\mathrm{E}}_T=\mathrm{S}{\mathrm{E}}_R+\mathrm{S}{\mathrm{E}}_A+\mathrm{S}{\mathrm{E}}_{MR}\end{array}$$ (5)

When SE_T > 15 dB, the contribution from multiple reflections (SE _MR ) can be neglected. In this case, the three components are given by:

$$\begin{array}{c}\mathrm{S}{\mathrm{E}}_{\mathrm{T}}=-10log\left(\mathbf{T}\right)\end{array}$$ (6)

$$\begin{array}{c}\mathrm{S}{\mathrm{E}}_{\mathrm{R}}=-10log\left(1-\mathbf{R}\right)\end{array}$$ (7)

$$\begin{array}{c}\mathrm{S}{\mathrm{E}}_{\mathrm{A}}=\mathrm{S}{\mathrm{E}}_{\mathrm{T}}-\mathrm{S}{\mathrm{E}}_{\mathrm{R}}=-10log\left(\frac{T}{1-R}\right)\end{array}$$ (8)

Wave transport through the multilayer system was modeled using a 2 × 2 transfer‐matrix model (TMM) under normal incidence. An alternating air/MXene stack was represented by a product of interface and propagation matrices, yielding a total matrix 𝑀 that connects the electric‐field amplitudes at the entrance and exit planes:

$$\begin{array}{c}\left[\begin{array}{c}{E}_{out}^{+}\\ E_{out}^{-}\end{array}\right]=M\left[\begin{array}{c}{E}_{in}^{+}\\ E_{in}^{-}\end{array}\right]=\left[\begin{array}{cc}A& B\\ C& D\end{array}\right]\left[\begin{array}{c}{E}_{in}^{+}\\ E_{in}^{-}\end{array}\right]\end{array}$$ (9)

Here, E⁺ and E⁻ denote forward‐ and backward‐propagating complex electric‐field amplitudes. A radiation boundary condition at the exit enforces $E_{out}^{-}=0$.

For normal incidence, the Fresnel and transmission amplitudes at an air (a)‐ film (m) boundary are:

$$\begin{array}{c}{r}_{am}=\frac{{\eta }_m-{\eta }_a}{{\eta }_m+{\eta }_a},t_{am}=\frac{2{\eta }_m}{{\eta }_m+{\eta }_a}\end{array}$$ (10)

The corresponding interface matrices are:

$$\begin{array}{c}{T}_{a\to m}=\frac{1}{t_{am}}\left[\begin{array}{cc}1& r_{am}\\ r_{am}& 1\end{array}\right],T_{m\to a}=\frac{1}{t_{ma}}\left[\begin{array}{cc}1& r_{ma}\\ r_{ma}& 1\end{array}\right]\end{array}$$ (11)

Propagation through a layer k of thickness d is expressed as:

$$\begin{array}{c}{P}_k=\left[\begin{array}{cc}{e}^{-{\gamma }_k{d}_k}& 0\\ 0& e^{+{\gamma }_k{d}_k}\end{array}\right],{\gamma }_k={\alpha }_k+j{\beta }_k\end{array}$$ (12)

In the good‐conductor regime applicable to MXene films:

$$\begin{array}{c}{\gamma }_m\approx \left(1+i\right)\sqrt{\pi f\mu \sigma },{\eta }_m\approx \left(1+i\right)\sqrt{\frac{\pi f\mu }{\sigma }}\end{array}$$ (13)

For air:

$$\begin{array}{c}{\gamma }_a=j\frac{2\pi f}{c},{\eta }_a=\sqrt{\frac{{\mu }_0}{{\epsilon }_0}}\approx 377\mathrm{\Omega }\end{array}$$ (14)

To match the implementation used, each air(L_a )–MXene(L_m ) period is described as:

$$\begin{array}{c}{M}_{per}=P_a(L_a)T_{a\to m}{P}_m\left(L_m\right)T_{m\to a}\end{array}$$ (15)

The overall transfer matrix for N repeating periods is:

$$\begin{array}{c}{M}_{total}={\left(M_{per}\right)}^N\end{array}$$ (16)

Applying the exit boundary condition $E_{out}^{-}=0$ yields the input‐referenced field amplitudes:

$$\begin{array}{c}r=-\frac{M_{21}}{M_{22}},t=M_{11}-\frac{M_{12}{M}_{21}}{M_{22}}\end{array}$$ (17)

The corresponding power coefficients are then:

$$\begin{array}{c}R={\left|r\right|}^2,T={\left|t\right|}^2\end{array}$$ (18)

Finally, R and T are substituted into Equations (6)–(8) to obtain SE _R , SE _A ,  and SE _T .

4.8. Electrochemical Measurements

The MXene working electrode was prepared by drop‐casting an MXene dispersion onto carbon paper. An MXene dispersion (5 mg mL⁻¹) was mixed with 15 µL of 5 wt.% Nafion using vortex mixing to ensure homogeneous distribution, and the mixture was then drop‐cast and dried at 60°C for 12 h. Measurements were performed at room temperature in a three‐electrode configuration using a potentiostat (Ivium Technologies), with a graphite rod counter electrode and an Ag/AgCl reference electrode. The electrolyte was 3 M H₂SO₄, and the potential window ranged from −0.6 to 0.2 V (vs. Ag/AgCl). The exposed electrode area was 1 cm² with a mass loading of 1 mg cm⁻². CV, GCD, and EIS were performed under the same setup; EIS was measured over 0.1–10⁵ Hz. Cycling stability was evaluated by GCD cycling under the same three‐electrode configuration and testing conditions as described above.

The gravimetric capacitance (C_g ) was calculated from the CV curves using the following equation:

$$C_g=\frac{1}{mv\mathrm{\Delta }V}\int I\left(V\right)dV$$

where m is the mass of active material (g), v is the scan rate (V s⁻¹), and ΔV is the potential window (V). The areal capacitance was calculated as $C(A)=C(g)\times \left(\frac{m}{A}\right)$, where $\left(\frac{m}{A}\right)$ is the MXene mass loading (g cm⁻²). The volumetric capacitance was estimated as $C_V=C_g\times \rho$, where ρ is the apparent film density (g cm⁻³). Based on the measured film thickness at a fixed areal loading, ρ was approximately 3.0 g cm⁻³, with negligible variation across compositions.

4.9. Batch‐to‐Batch Reproducibility

Batch‐to‐batch reproducibility was assessed for each target composition (x = 1.71, 1.82, and 1.94) using three independent synthesis batches (n = 3), all prepared following the identical protocol described above (Figure S35 and Table S6). All characterizations were conducted under identical sample preparation and measurement conditions. Data were reported as mean ± standard deviation (s.d.) across batches, and relative standard deviation (RSD, %) was calculated as (s.d./mean) × 100.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma72218‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.