Recent Advances in MXene-Based Cocatalysts for Photocatalytic Hydrogen Evolution

Abstract

While MXenes are promising cocatalysts for photocatalytic hydrogen evolution, their roles are frequently oversimplified as merely conductive electron sinks. This review begins by revisiting the classical electron-sink mechanism of Ti₃C₂ and utilizes the CdSe/Ti₃C₂ and Cd_0.5Zn_0.5S/Ti₃C₂ systems to demonstrate how intimate interfaces, Fermi-level equilibration, Schottky junctions, and directional electron-transfer pathways synergistically couple charge separation with H* conversion. The discussion is then extended to multimetallic MXenes, using ZnIn₂S₄/Mo₂TiC₂ as a model, to highlight how metal-layer composition, local interfacial coordination, work-function differences, and hydrogen-adsorption energetics jointly regulate electron extraction and surface reaction kinetics. Special attention is given to surface terminations, which can reshape interfacial energetics and enable MXenes to transition between electron-sink and hole-mediated roles depending on the interface and working conditions. Furthermore, representative Nb₂C quantum-dot and VNbC solid-solution systems illustrate how nanoscale electric fields and multimetallic sites can expand MXene functionality beyond conventional electron extraction. Ultimately, this role-oriented framework advances the development of MXene cocatalysts from simple material integration toward controllable and programmable interfacial functions.

1. Introduction

Photocatalytic H₂ production is not governed by light absorption alone; substantial energy is lost when photogenerated carriers recombine or fail to drive surface reactions. Since the photo-assisted electrolysis of water on TiO₂ reported by Fujishima and Honda [1], photocatalysis has developed toward particulate systems for solar-fuel production. Overall water splitting must be distinguished from sacrificial-agent-assisted H₂ evolution, which replaces water oxidation with donor oxidation and therefore evaluates only the reduction half-reaction [2]. Beyond the thermodynamic requirement of 237 kJ mol⁻¹, equivalent to 1.23 V, practical water splitting must overcome overpotentials and interfacial energy losses [3]. Most systems therefore remain below the efficiency required for practical implementation [4], while durability, reactor cost, gas separation, safety, and system-level performance impose further constraints [5].

At the material level, these limitations arise mainly from two coupled bottlenecks: inefficient charge utilization and sluggish surface conversion. Photogenerated carriers may recombine in the bulk, at defects and grain boundaries, or at catalyst–solution interfaces before reaching reactive sites [6]. Productive separation therefore requires both a thermodynamic driving force and selective electron- and hole-transfer pathways [7]. Photoluminescence quenching, prolonged carrier lifetimes, or enhanced photocurrent alone cannot establish productive charge transfer because these signals may also reflect trapping or altered recombination [8]. Junction construction and facet engineering can direct carrier migration [9], while internal electric fields and polarization provide additional driving forces [10]. Spin regulation may further modify carrier dynamics and reaction pathways [11]. These strategies are effective only when the separated carriers retain sufficient redox power and are rapidly consumed through the multistep reactions involved in H₂ evolution and water oxidation [12]. Slow surface conversion instead causes charge accumulation and promotes recombination, directly coupling surface kinetics with carrier utilization [13].

Considerable progress has been achieved, but usually along only one part of this process. Facet-selective cocatalyst deposition on SrTiO₃:Al produced an external quantum efficiency approaching unity under ultraviolet irradiation [14], while particulate photocatalyst sheets achieved solar-to-hydrogen efficiencies above 1% [15]. Visible-light overall water splitting was extended using an oxysulfide photocatalyst [16] and a dye-sensitized Z-scheme system [17]. A 100 m² panel reactor further demonstrated large-scale solar hydrogen production, although its conversion efficiency remained limited [18]. These results highlight the persistent difficulty of combining efficient charge transport, rapid surface kinetics, broad light utilization, stability, and scalability. Cocatalysts address this gap by extracting carriers [19], lowering reaction barriers at surface sites [20], regulating interfacial energetics [21], and coordinating complementary reduction and oxidation processes [22]. For example, coordinatively unsaturated MoS₂ edges provide active sites for hydrogen evolution [23], but their effectiveness depends on intimate coupling with CdS to ensure efficient electron delivery [24]. In Pt-PdS/CdS, Pt accelerates proton reduction [25], whereas PdS promotes hole consumption, allowing the two interfaces to operate cooperatively [26]. Oxidation cocatalysts similarly suppress surface-hole accumulation on BiVO₄ [27,28], while spatially separated reduction and oxidation sites direct electrons and holes toward different half-reactions [29]. An effective cocatalyst must therefore connect directional carrier transfer with rapid and selective surface turnover rather than merely collect charge.

MXenes, generally described by the formula M_n+1X_nT_x, are two-dimensional transition-metal carbides, nitrides, or carbonitrides, where M denotes one or more transition metals; X represents C and/or N; T_x denotes surface termination species such as O, OH, F, or Cl; n determines the thickness of the M–X slab; and x represents the variable stoichiometric amount of surface terminations. Their two-dimensional morphology, accessible transition-metal surfaces, and adjustable electronic structures enable extensive semiconductor contact and short cross-interface transfer distances [30,31]. Their interfacial properties can also be tuned across multiple length scales and chemical environments [32,33]. The exfoliation of Ti₃AlC₂ established layered Ti₃C₂T_x [34], and conductive Ti₃C₂T_x films demonstrated the rapid transport potential of MXenes [35]. However, conductivity varies with flake quality, defect density, intercalated species, and oxidation [36], while changes in composition and surface functionalization can shift some MXenes away from metallic behavior [37]. Surface terminations introduce a further level of regulation. Lewis-acidic molten-salt reactions enable more controlled surface chemistry than conventional aqueous etching [38,39], and covalent substitution has expanded MXene terminations beyond mixed -O, -OH, and -F to chalcogen, amino, halogen, and nearly termination-free surfaces [40]. These terminations alter surface dipoles and work functions [41], allowing MXenes to form electron- or hole-selective contacts depending on the coupled semiconductor [42]. Accordingly, Ti₃C₂ can accept electrons and promote H adsorption in metal-sulfide photocatalysts [43], whereas low-work-function Ti₃C₂T_x may instead mediate hole transfer in appropriately aligned interfaces [44]. MXene function is therefore determined by its actual surface and interface rather than by nominal composition or conductivity alone.

Representative Nb₂C quantum-dot and VNbC solid-solution systems further illustrate how nanoscale electric fields and multimetallic sites can extend MXene functions beyond conventional electron extraction. Early MXene photocatalysts were mainly described as conductive electron sinks, yet this captures only one interface-dependent role. Existing reviews have comprehensively summarized MXene synthesis, photocatalytic applications, morphology, interfacial chemistry, and termination-dependent H₂ evolution [45,46].

Beyond these reviews, broader studies have examined MXenes in sustainable hydrogen production [47], photocatalytic CO₂ reduction [48], and diverse energy and environmental applications [49]. Early investigations demonstrated the potential of TiO₂/MXene composites [50] and CdS/Ti₃C₂ Schottky heterojunctions [51] for improving photocatalytic H₂ evolution. Studies on MXene/Bi₂WO₆ for CO₂ reduction [52] and TiO₂/Ti₃C₂ for dye degradation, in which Ti₃C₂ was proposed to act as a hole reservoir [53], further illustrate the reaction- and interface-dependent functions of MXenes. For photocatalytic H₂ evolution, sulfur-vacancy engineering in ZnIn₂S₄/Ti₃C₂T_x [54] and surface-termination regulation in ZnIn₂S₄/Ti₃C₂T_x [55] demonstrated the importance of defect chemistry and interfacial energetics in charge separation. Beyond conventional electron extraction, Ti₃C₂T_x/CdS heterostructures combined interfacial charge separation with photothermal enhancement [56], while CdS/MoS₂/Ti₃C₂ employed Ti₃C₂ and MoS₂ as hole- and electron-accepting components, respectively, to improve charge utilization and suppress photocorrosion [57]. Further studies of CdS/Ti₃C₂ Schottky heterojunctions provided experimental and theoretical evidence for interfacial charge transfer [58], whereas in situ construction of sandwich-like Ti₃C₂/rutile-TiO₂ interfaces strengthened interfacial contact and facilitated carrier transport [59]. Collectively, these findings highlight the importance of surface morphology, termination chemistry, and interface construction in determining MXene cocatalytic functions [60].

The present review takes a different perspective by focusing specifically on how MXene interfacial functions evolve in photocatalytic H₂ evolution. Rather than comparing material classes or cataloging reported activities, we organize the discussion around the relationships among interfacial electronic structure, metal-layer composition, surface terminations, dimensionality, charge-transfer direction, and catalytic-site function. Particular emphasis is placed on the experimental evidence required to assign these roles and on whether the proposed interface is retained under working conditions. In this review, MXenes are considered cocatalytic components when they primarily regulate carrier extraction, transport, interfacial energetics, or surface reaction kinetics without serving as the dominant light absorber. This role evolution is summarized in Scheme 1.

Scheme 1. Evolution of MXene cocatalytic roles in photocatalytic hydrogen evolution. The yellow and dark-blue arrows indicate the starting point of Ti₃C₂T_x and interface/surface engineering, respectively. The green, light-blue, and orange curved arrows illustrate the conceptual links to composition, interface/active-site, and structure/stability engineering, respectively. The black arrows highlight charge-transfer and catalytic functions.

2. Classical Starting Point: Ti₃C₂T_x as an Electron Sink

2.1. Electron Acceptance and Rapid Transport in Ti₃C₂T_x

Ti₃C₂T_x is widely regarded as a conductive electron sink in photocatalytic H₂ evolution. Favorable energy-level alignment drives photogenerated electrons from the semiconductor to Ti₃C₂T_x, while its two-dimensional conductive framework supports rapid lateral transport. The extended interface also increases the contact area and shortens the cross-interface migration distance [60,61]. O-terminated Ti₃C₂ further provides favorable H adsorption energetics, establishing a direct link between electron transfer and proton reduction, as shown in Figure 1 [43].

Figure 1. (a) Side and top views of the O-terminated Ti₃C₂ supercell. (b) Calculated free-energy diagrams for the hydrogen evolution reaction at different H* coverages. (c) Comparison of the hydrogen-adsorption free energies of O-terminated Ti₃C₂, Pt, MoS₂, and WS₂. (d) Calculated band structure. (e) Total and projected densities of states of O-terminated Ti₃C₂ [43]. Reproduced from Ref. [43] under CC BY 4.0.

Metal-sulfide systems provide representative evidence for this function. In Zn₂In₂S₅/Ti₃C₂(O, OH)_x, strong interfacial interaction promoted charge redistribution and carrier separation [62]. Growing ultrathin ZnIn₂S₄ nanosheets on both sides of Ti₃C₂T_x created shorter and more continuous transfer pathways [63]. These results show that electron acceptance depends on both the electronic structure of Ti₃C₂T_x and the quality of its contact with the semiconductor.

Intimate coupling is therefore essential. Protonated g-C₃N₄/Ti₃C₂ Schottky heterojunctions reduced the resistance of interfacial electron transfer [64], while in situ grown Ti₃C₂@MIL-NH₂ composites strengthened contact through direct interface formation [65]. Ti₃C₂T_x should thus be regarded as an interface-dependent electron acceptor rather than a universally effective conductive additive.

Electron extraction must also be followed by efficient delivery to reduction sites. In Ru/MXene/TiO₂, Ti₃C₂T_x connected the photoactive semiconductor with Ru catalytic centers [66]. Oxidized g-C₃N₄/Ti₃C₂T_x similarly combined electron collection with surface sites favorable for H adsorption [67]. In these systems, Ti₃C₂T_x acts as an electron relay that couples charge separation with surface H₂ evolution.

Morphology and surface chemistry further regulate this process. Face-to-face Ti₃C₂/g-C₃N₄ [68] and CdS/Ti₃C₂ [69] heterostructures increased the density of transfer interfaces. Plasma treatment promoted CdS nucleation and strengthened interfacial coupling [70], whereas ZnIn₂S₄ growth induced Ti₃C₂ delamination, exposed more surface area, and reduced restacking [71]. Rapid electron transport therefore requires suitable flake structure, continuous contact, favorable terminations, and efficient downstream electron consumption.

Overall, the electron-sink function of Ti₃C₂T_x relies on four linked conditions: favorable energy alignment, intimate contact, continuous transport pathways, and rapid utilization of the transferred electrons. PL quenching, enhanced photocurrent, reduced impedance, and improved H₂ evolution indicate better carrier utilization, but they do not independently establish the transfer direction.

2.2. Interfacial Electron Migration from Semiconductors to Ti₃C₂T_x

Electron migration toward Ti₃C₂T_x is governed by energy-level alignment and interfacial coupling. Fermi-level equilibration induces charge redistribution, band bending, and an interfacial electric field, driving photogenerated electrons from the semiconductor to Ti₃C₂T_x while suppressing back-transfer. Ti₃C₂T_x therefore acts as an electron-transfer intermediary that channels electrons toward reduction sites rather than as a terminal electron reservoir, as illustrated in Scheme 2 [72].

Scheme 2. Directional electron transfer from a semiconductor to Ti₃C₂T_x and subsequent electron utilization for photocatalytic H₂ evolution.

The transfer rate depends strongly on interface geometry. In Mn_0.5Cd_0.5S/Ti₃C₂, nanoscale Schottky contacts accelerated electron extraction and reduced recombination [73]. Ligand-bridged CdS/Ti₃C₂T_x further showed that molecular linkers can strengthen anchoring and improve electronic communication across the interface [74]. InVO₄/Ti₃C₂ [75] and Au/Ti₃C₂/g-C₃N₄ [76] extended this function from terminal electron acceptance to conductive relay behavior, in which Ti₃C₂ receives electrons from photoactive components and transports them toward catalytic sites.

Shortening the migration distance provides another route to faster transfer. Confining In₂S₃ within Ti₃C₂T_x layers formed compact interfaces with reduced resistance [77], while ultrathin Cd_0.8Zn_0.2S nanosheets grown on Ti₃C₂ created similarly continuous transfer pathways [78]. A recent Bi₂MoO₆/Ti₃C₂T_x heterostructure provides a direct structural example of such intimate semiconductor/MXene contact, with TEM and HRTEM resolving a well-defined heterointerface, as shown in Figure 2 [79]. What needs to be clarified here is in Figure 2d, the yellow and magenta boxes highlight Bi₂MoO₆ regions, whereas the red and cyan boxes indicate Ti₃C₂T_x regions. The corresponding enlarged images in Figure 2e reveal characteristic lattice fringes with spacings of 0.325 and 0.202 nm for Bi₂MoO₆ and 0.31 and 0.26 nm for Ti₃C₂T_x, respectively. More complex Ru/Ti₃C₂-TiO₂/CdS [80] and ZnIn₂S₄/Cd_0.8Zn_0.2S/Ti₃C₂ [81] architectures further use Ti₃C₂ to connect semiconductor excitation with spatially separated reduction sites.

Figure 2. Interfacial microstructure of the Bi₂MoO₆/Ti₃C₂T_x heterojunction: (a,b) TEM images; (c,d) HRTEM images resolving the closely contacted semiconductor/MXene interface; (e) inverse fast-Fourier-transform analysis of the interfacial region; and (f) EDS elemental mapping of the heterostructure [79]. Reproduced from Ref. [79] under CC BY 4.0.

Interfacial transport is also influenced by structural matching and dimensionality. Lattice-coherent Te/In₂O₃@MXene interfaces reduced structural discontinuity and transport barriers [82]. Electron migration is therefore controlled by the work-function difference, bonding configuration, interface continuity, dimensionality, and rate of electron consumption.

A convincing assignment of semiconductor-to-Ti₃C₂T_x electron transfer requires convergent evidence. Work-function measurements, UPS, Kelvin-probe analysis, light-irradiated XPS, transient spectroscopy, and charge-density calculations can establish the energetic feasibility and transfer direction. By contrast, PL quenching, increased photocurrent, reduced impedance, or enhanced H₂ evolution demonstrate improved carrier utilization but cannot alone prove electron migration toward Ti₃C₂T_x.

These measurements address different levels of the mechanism: energetic measurements establish whether transfer is feasible, whereas time-resolved and spatially resolved probes determine whether it actually occurs and on what timescale.

2.3. Mechanistic Evidence from Representative CdSe- and Cd_0.5Zn_0.5S-Based Systems

Our group used CdSe/Ti₃C₂ and Cd_0.5Zn_0.5S/Ti₃C₂ as model systems to establish the electron-sink function of Ti₃C₂ through complementary structural, energetic, spectroscopic, kinetic, and theoretical evidence [83,84,85].

In CdSe/Ti₃C₂, CdSe nanorods were grown directly on Ti₃C₂ nanosheets to form an intimate 1D/2D interface [84]. UPS and work-function measurements indicated electron redistribution from CdSe to Ti₃C₂ after contact, leading to Fermi-level equilibration and interfacial band bending. XPS consistently showed electron depletion on CdSe and enrichment on Ti₃C₂. Under illumination, the Cd and Se peaks shifted toward higher binding energies, whereas the Ti peak shifted toward lower binding energy, supporting photoelectron migration from CdSe to Ti₃C₂. KPFM and selective Au photodeposition further located electron accumulation on the Ti₃C₂ component. The light-induced XPS response is shown in Figure 3 [84], where the arrows indicate binding-energy shifts of +0.4 eV for Cd 3d, +0.3 eV for Se 3d, and −0.2 eV for Ti 2p under illumination.

Figure 3. High-resolution in situ XPS spectra of (a) Cd 3d, (b) Se 3d, and (c) Ti 2p for the CdSe/Ti₃C₂ composite under dark and light-irradiation conditions [84].

This transfer also produced clear kinetic consequences. The composite exhibited stronger photocurrent, lower interfacial resistance, quenched photoluminescence, and a shorter fluorescence lifetime. DFT calculations revealed electron accumulation on Ti₃C₂ and an increased density of states near the Fermi level. Ti₃C₂ also improved H adsorption energetics, showing that the extracted electrons were coupled to surface H₂ evolution rather than merely stored at the interface [84].

The Cd_0.5Zn_0.5S/Ti₃C₂ system extended this mechanism to a single-layer MXene interface [85]. Cd_0.5Zn_0.5S nanorods were uniformly anchored on Ti₃C₂ nanosheets, enabling efficient extraction at a Ti₃C₂ loading of only 0.5 wt%. Light-irradiated XPS showed electron depletion from Cd_0.5Zn_0.5S and enrichment on Ti₃C₂. Femtosecond transient absorption further identified faster depopulation of the semiconductor excited state after coupling, directly linking the interface structure to ultrafast electron transfer. The corresponding charge-transfer response is shown in Figure 4 [85]. The arrows indicate binding-energy shifts of +0.4 eV for Cd 3d, +0.5 eV for Zn 2p, +0.4 eV for S 2p, and −0.2 eV for Ti 2p under illumination. The faint gray traces represent the experimental spectra, while the colored curves show the corresponding fitted profiles.

Figure 4. High-resolution in situ XPS spectra of Cd 3d (a), Zn 2p (b), S 2p (c), and Ti 2p (d) of Cd_0.5Zn_0.5S and Cd_0.5Zn_0.5S/Ti₃C₂ [85].

Together, these systems provide complementary energetic and ultrafast spectroscopic evidence for the electron-transfer pathway summarized in Scheme 2.

The CdSe/Ti₃C₂ system establishes the transfer direction through UPS, KPFM, illuminated XPS, photodeposition, and DFT, whereas Cd_0.5Zn_0.5S/Ti₃C₂ provides additional ultrafast spectroscopic evidence. Nevertheless, illuminated-XPS shifts should be interpreted together with energetic and kinetic measurements rather than used alone to quantify transferred electrons [86]. Schottky-barrier height and interfacial bonding can further alter the transfer process [87], while surface terminations and adsorbed species may change the working-state role of Ti₃C₂T_x [88].

2.4. From Physical Contact to Chemically Coupled Interfaces

MXene-based interfaces have evolved from weak physical contact toward confined, chemically bonded, and atomically coupled junctions. The electron-transfer function of Ti₃C₂T_x depends not only on conductivity and work function, but also on interfacial continuity and coupling strength. Mechanical or ultrasonic mixing commonly produces discontinuous van der Waals contact and relatively high transfer resistance.

Electrostatic assembly improves dispersion and contact by exploiting the charged MXene surface. Protonated C₃N₅, for example, was assembled with negatively charged Ti₃C₂ nanosheets to form a closely contacted Schottky interface that promoted electron migration toward Ti₃C₂ [89]. Ligand bridging provides stronger anchoring and more continuous pathways across CdS/Ti₃C₂T_x interfaces [74].

In situ growth further converts geometrical contact into direct electronic coupling. In SrTiO₃/Ti₃C₂, Sr(OH)₂ partially etched surface Ti atoms while inducing SrTiO₃ nucleation. The resulting chemical bonding and Ti vacancies promoted carrier separation more effectively than physical mixing [90]. This comparison shows that interface continuity and electronic interaction, rather than simple component coexistence, determine transfer efficiency.

Interfacial bonds can also function as direct charge-transfer channels. In sandwich-like WO₃/Ti₃C₂/ZnIn₂S₄, Ti-S and Ti-O-W bonds connected Ti₃C₂ with ZnIn₂S₄ and WO₃, respectively. These bonds enhanced orbital coupling, shortened transfer distances, and enabled directional charge migration, while the sandwich architecture suppressed MXene oxidation [91]. Bonded Ti₃C₂/carbon nitride interfaces similarly showed that chemical coupling can regulate both electronic interaction and Schottky-barrier formation [87].

More precise strategies have progressed toward lattice coherence and atomic-scale bridging. Lattice matching reduces structural discontinuity and carrier scattering [81]. In Ru-Cd₁-xS/Ti₃C₂T_x, vacancy-mediated Ru-O bonds bridged the heterointerface, strengthened the built-in electric field, and accelerated charge transfer to approximately 1.3 ps [92].

MXene interface engineering has therefore progressed from weak physical contact toward increasingly confined, chemically bonded, and atomically defined junctions. Interfacial coupling is thus a key structural variable, but its effect must be considered together with MXene composition, surface chemistry, and reaction kinetics.

2.5. Design Principles for Directional Electron Transfer

The studies above show that directional charge transfer at MXene interfaces is governed by three linked factors: energetic alignment, interfacial coupling, and carrier consumption. Work-function matching and Fermi-level equilibration determine band bending and the built-in electric field, defining whether an MXene acts as an electron acceptor, transport mediator, or hole-extraction component [93]. Its role should therefore be assigned from the actual interface rather than from conductivity alone.

Once the transfer direction is established, interfacial coupling controls the transfer rate. Schottky contacts, interfacial Ti-O-W or Ti-S bonds, lattice-coherent junctions, and vacancy-mediated bridges can reduce resistance and shorten migration distances [81,90,91,92]. Effective interfacial coupling should reduce charge-transfer resistance while preserving favorable band alignment and accessible catalytic sites; the optimal barrier height and bonding configuration are therefore system-specific rather than universally maximized. The transferred carriers must also reach accessible active sites, where vacancies or single atoms can couple charge separation with proton reduction [92,94].

These principles connect the classical Ti₃C₂T_x electron-sink model with the broader role diversification discussed in the following section. Representative MXene-based photocatalysts and their reported H₂-evolution performances are summarized in Table 1, although differences in reaction conditions should be considered when comparing activities. Mechanistic assignment should therefore distinguish three levels of evidence. Work-function and band-alignment measurements establish the thermodynamic driving force, time-resolved spectroscopy resolves the transfer kinetics, and active-site or intermediate analyses determine whether the transferred carriers are productively consumed. Overall, energetic alignment determines transfer direction, interfacial coupling controls transfer kinetics, and local coordination governs carrier consumption.

Table 1. Representative performance and interfacial roles of MXene-based photocatalysts for H₂ evolution.

Photocatalyst	MXene Form/Interface Role	Construction Mode	Reported Test Conditions	H2 Evolution Rate (mmol g−1 h−1)	AQY/AQE	Reported Benchmark or Stability	Ref.
UZNs–MNs–UZNs ZnIn2S4/Ti3C2Tx	Ti3C2Tx nanosheets; sandwich-like 2D/2D Schottky interfaces	In situ hydrothermal growth	TEOA; 300 W Xe lamp	3.475	20.41% at 400 nm	6.6-fold higher than pristine ZnIn2S4; excellent stability reported	[63]
Protonated g-C3N4/Ti3C2	Ti3C2 nanosheets; electron-accepting Schottky contact	Protonation-assisted assembly	TEOA; 300 W Xe lamp	0.727	8.6% at 400 nm	Strong interfacial interaction reported; quantitative cycle retention not reported	[64]
CdS@MoS2/Ti3C2	Ti3C2 electron-transfer layer plus MoS2 HER cocatalyst	Microwave–hydrothermal construction	Lactic acid; 300 W Xe lamp	14.880	n.r.	Enhanced resistance to CdS photocorrosion and improved cycling behavior	[57]
ZnIn2S4/10% Mo2TiC2	Few-layer Mo2TiC2; tightly bound Schottky interface	In situ solvothermal growth	10 mg catalyst; simulated sunlight; sacrificial agent not reported in the original study	4.300	0.75% at 350 nm	3.8-fold higher than ZnIn2S4 and higher than ZnIn2S4/1% Pt	[95]
25% Mo2TiC2/ZnCdS	Mo2TiC2 nanosheets; strongly coupled 2D/0D Ohmic contact	Physical mixing	6 mg catalyst; 30 mL 10 vol% lactic acid; 5 W LED	17.730	16.22% at 475 nm	3.15-fold higher than ZnCdS	[96]
CdIn2S4/Mo2TiC2	Mo2TiC2 nanosheets; Schottky electron-accepting cocatalyst	Physical stirring and grinding	Detailed irradiation and sacrificial-agent conditions not reported in the original study	3.350	0.71% at 450 nm.	55.83-fold higher than CdIn2S4; activity maintained over four cycles	[97]
N-Ti3C2 MQDs/CdS	N-doped Ti3C2 quantum dots; 0D/1D charge-transfer junction	Self-assembly	Visible-light photocatalysis; detailed conditions not reported in the original study	17.094	n.r.	14.79-fold higher than pure CdS	[98]
5.5 wt% Ti3C2 QD/PGCN	Ti3C2 quantum dots; 0D/3D Schottky and photothermal interface	Electrostatic self-assembly	TEOA; 300 W Xe lamp	4.041	11.8% at 420 nm	15.24-fold higher than pristine CN and 3.53-fold higher than PGCN	[99]
ZnIn2S4/g-C3N4/Ti3C2	Ti3C2 nanosheets; Schottky electron-extraction branch integrated with a ZnIn2S4/g-C3N4 S-scheme	Hydrothermal	300 W Xe lamp (>420 nm); 10 mg catalyst; 80 mL H2O/TEOA (72/8 mL); Pyrex reactor; GC detection.	2.452	13.1% at 365 nm	1.5-fold higher than ZnIn2S4/Ti3C2 (ZM-3) and 1.6-fold higher than g-C3N4/ZnIn2S4 (CZ-3)	[100]

The reported activities should not be interpreted as a direct ranking because catalyst loading, sacrificial reagents, irradiation conditions, wavelength ranges, reactor geometry, and gas-quantification methods vary across studies. Within-study comparisons with the parent semiconductor or benchmark cocatalyst are therefore generally more informative than absolute H₂-evolution rates across different reports. AQY/AQE, MXene loading, stability, and matched controls should be considered together with mass-normalized activity. Table 1 also illustrates that similar activity enhancements can arise from fundamentally different interfacial functions. Binary Schottky interfaces mainly promote electron extraction, whereas multicomponent architectures can provide parallel carrier-transfer pathways. For example, the 2D/2D/2D ZnIn₂S₄/g-C₃N₄/Ti₃C₂ heterostructure combines an S-scheme pathway between ZnIn₂S₄ and g-C₃N₄ with a Schottky-mediated electron-transfer channel from g-C₃N₄ to Ti₃C₂, reaching an H₂ evolution rate of 2.452 mmol g⁻¹ h⁻¹. This example shows that Ti₃C₂ can function as one branch of a coordinated carrier-routing network rather than solely as a terminal electron sink.

Performance alone, however, does not establish the interfacial function of an MXene. Mechanistic assignment requires complementary evidence for energetic alignment, charge-transfer direction, carrier dynamics, and catalytic-site function. Table 2 therefore compares representative systems according to the key evidence supporting their proposed MXene interfacial roles.

Table 2. Critical experimental evidence required for assigning MXene interfacial functions in photocatalytic H₂ evolution.

Representative System	Interface/Assigned MXene Role	Proposed e−Transfer	Energetic Basis	Key Charge-Transfer Evidence	Active-Site Evidence	Photocatalytic Performance/Stability	Confidence	Ref.
CdSe/Ti3C2	1D/2D interface; electron acceptor/relay	CdSe → Ti3C2	Φ(CdSe) = 4.73 eV; Φ(Ti3C2) = 6.34 eV; band bending	UPS; light–XPS; KPFM; Au photodeposition; DFT	DFT-supported favorable H* adsorption	0.763 mmol g−1 h−1; AQE 1.3%; stable for 5 cycles; post-XRD/TEM	High	[84]
Cd0.5Zn0.5S/Ti3C2	1D/2D interface; electron acceptor	CZS → Ti3C2	Fermi-level equilibration/Schottky alignment	Light–XPS; fs-TA; DFT	Improved H-related energetics by DFT	15.56 mmol g−1 h−1; AQE 18.4% at 350 nm; stable for 5 cycles	High	[85]
ZnIn2S4/Mo2TiC2	Tightly coupled 2D/2D Schottky interface; electron extractor	ZIS → Mo2TiC2	Work-function mismatch; Schottky barrier	In situ/light–XPS; DFT charge redistribution	HER-site/ΔGH* evidence incomplete	4.30 mmol g−1 h−1; stability reported	Moderate	[95]
ZnCdS/Mo2TiC2	2D/0D Ohmic interface; electron-transfer mediator	ZnCdS → Mo2TiC2	Ohmic-contact alignment	XPS; DFT; carrier-separation measurements	d-band/H-adsorption regulation	17.73 mmol g−1 h−1; AQE 16.22% at 475 nm; lactic acid; stable	Moderate	[96]
ZnIn2S4/g-C3N4/Ti3C2	2D/2D/2D S-scheme/Schottky network; Ti3C2 electron-accepting branch	g-C3N4 → Ti3C2; ZIS/g-C3N4 S-scheme	Band alignment supports dual pathways	XPS/spectroscopic evidence; direct ultrafast evidence n.r.	Direct MXene HER-site evidence n.r.	2.452 mmol g−1 h−1; TEOA; cycling stability reported	Moderate	[100]

Table 2 highlights a hierarchy of mechanistic evidence. Energetic alignment establishes whether charge transfer is feasible, whereas light-induced XPS, KPFM, photodeposition, and time-resolved spectroscopy provide stronger evidence for its direction and kinetics. Active-site-sensitive measurements are further required to connect transferred carriers with H₂ evolution. Thus, PL quenching, enhanced photocurrent, reduced impedance, or increased activity alone cannot establish a specific MXene interfacial role.

Taken together, Table 1 and Table 2 show that high photocatalytic activity and reliable mechanistic assignment are complementary but distinct criteria. Rational MXene interface design therefore requires not only favorable performance, but also a clear connection among energetic alignment, interfacial coupling, carrier-transfer kinetics, and catalytic-site function. These relationships provide the basis for understanding how metal-layer composition and surface chemistry can further diversify MXene interfacial roles, as discussed in the following section.

3. Role Diversification Through Metal-Layer and Surface-Termination Engineering

3.1. Multimetallic MXenes and Surface Terminations as Electronic-Structure Regulators

The electronic properties of MXenes are jointly determined by the carbide or nitride framework and the chemistry of the exposed surface. Hart et al. showed that deintercalation and partial removal of surface species converted the apparent semiconductor-like transport of Mo₂TiC₂T_x and Ti₃CNT_x into metallic behavior [101]. Measured conductivity therefore reflects not only the intrinsic band structure, but also interflake transport, residual intercalants, and surface chemistry.

Surface terminations have consequently become controllable structural variables. Postsynthetic exchange has expanded the accessible surface chemistry beyond conventional O/OH/F terminations to N-, chalcogen-, and halogen-containing surfaces [40]. More broadly, recent screening across diverse MXene backbones and termination chemistries illustrates the large compositional space available for surface regulation, as summarized in Figure 5 [102], where the red highlights the 29 candidate M-site elements, including selected f-block elements, black denotes the X-site elements (C and N), and blue represents the 11 candidate surface-termination elements (T). The same color scheme is used in the structural models above the periodic table.

Figure 5. MXene compositional space explored across transition-metal species (M), carbon and nitrogen layers (X), surface terminations (T), and different layer thicknesses (n) [102]. Reproduced from Ref. [102] under CC BY 4.0.

Beyond expanding the accessible chemical space, increasing control over termination identity and uniformity can directly regulate charge transport. Uniformly halogen-terminated Ti₃C₂ exhibits strongly termination-dependent conductivity and carrier mobility, with reduced electronic scattering on chemically well-defined surfaces, as shown in Figure 6 [103]. In Figure 6d–f, the light-blue, gray, and purple spheres represent Ti, C, and Cl atoms, respectively, while the green and red spheres in the mixed-termination model indicate Cl and O atoms, respectively. These results show that termination identity and uniformity directly regulate electronic structure and carrier transport.

Figure 6. Charge-transport properties of uniformly halogen-terminated Ti₃C₂: (a) temperature-dependent direct-current conductivity; (b) terahertz conductivity; (c) time-resolved terahertz photoconductivity; (d) calculated structure, band structure, and density of states of Ti₃C₂Cl₂; (e,f) simulated electron transport and scattering in Ti₃C₂Cl₂ and ClO-Ti₃C₂; and (g) calculated current–voltage characteristics [103]. Reproduced from Ref. [103] under CC BY 4.0.

Multimetallic MXenes provide a complementary route for tuning the underlying electronic framework. In ordered M′2M″C₂T_x and M′2M″C₃T_x, the outer M′ layers interact directly with terminations and adsorbates, whereas the inner M″ layers regulate lattice stability and charge redistribution. Changing the M′/M″ combination reconstructs the metal d states near the Fermi level while generally retaining conductive pathways [104]. Experimental conversion of Mo₂TiAlC₂ into Mo₂TiC₂T_x further showed that phase composition and surface chemistry affect catalytic charge transport through different mechanisms [105]. Nominal stoichiometry alone is therefore insufficient to define the functional electronic structure.

At the orbital level, coupling between outer-metal d orbitals and termination O p orbitals links metal-layer composition directly to surface electronic states [106]. Computational studies likewise reveal substantial variations in conductivity, orbital occupation, and catalytic properties among ordered carbide [107], carbonitride [108], and broader multimetallic MXenes [109]. The metal layers define the d-electron framework, while the terminations determine how this framework is expressed at the interface.

These variables cannot be optimized independently. Synthesis and processing can simultaneously alter intercalants, termination distributions, defects, and structural disorder, so the measured electronic properties often reflect their combined effects rather than a single structural factor [110]. Synthetic strategies have therefore progressed from poorly defined O/OH/F surfaces toward more chemically controlled MXenes, including Lewis-acid-derived halogen terminations [38] and direct synthesis of uniformly terminated structures [111].

MXenes should thus be treated as coupled M′-M″-X-T_x electronic systems rather than generic metallic conductors. Their interfacial function depends on the specific combination of metal-layer composition, termination chemistry, and atomic arrangement.

3.2. Coupled Regulation of Interfacial Charge Transfer and HER Activity

For photocatalytic H₂ evolution, electron extraction and proton reduction are consecutive processes. An effective MXene cocatalyst must therefore combine rapid semiconductor-to-MXene charge transfer with suitable H* adsorption and H₂ release kinetics. On O-terminated MXenes, H adsorption is mainly governed by the charge state of surface O atoms, while metal composition and layer thickness further affect adsorption energetics and conductivity [112,113]. High conductivity alone is therefore insufficient.

Termination engineering directly regulates this balance. N and S incorporation into Ti₃C₂T_x creates Ti-N and Ti-S environments, reduces charge-transfer resistance, and shifts ΔG_H* toward thermoneutrality [114]. In Mo₂CT_x, mixed O/F terminations moderate the overly strong H adsorption of fully O-terminated surfaces, whereas excessive OH weakens adsorption [115]. Cl/O-terminated Ti₃C₂T_x similarly shows more favorable HER behavior than F/O surfaces, with further improvement after N or S substitution [116]. These results show that terminations regulate HER through both charge redistribution and local bonding.

Metal-layer composition provides complementary control. Co substitution in Mo₂CT_x optimizes H adsorption on neighboring O sites rather than acting only as an isolated active center [117]. Controlled Mo/Nb ordering in Mo_2+αNb_2-αC₃T_x likewise changes conductivity and HER activity through long-range electronic redistribution [118].

MXenes can also stabilize highly active catalytic species. Mo vacancies in Mo₂TiC₂T_x anchor Pt single atoms through Pt-C and Pt-O coordination [94]. Ru single atoms and clusters regulate water dissociation and H adsorption [119,120], while expanded Ti₃C₂T_x layers immobilize Pd nanoparticles and improve interfacial electron injection [121]. The MXene therefore serves as a conductive framework, anchoring matrix, and electronic regulator.

This coupled function has been demonstrated in photocatalysis. Br/O-terminated Mo₂CT_x improves charge separation in CdS/Mo₂CT_x while maintaining favorable proton-reduction kinetics, producing 17,566 μmol g⁻¹ h⁻¹ of H₂ [122]. Rational design should therefore coordinate electron extraction, MXene transport, H* adsorption, and H₂ desorption within the same interface.

3.3. Mo₂TiC₂-Based Interfaces: The ZnIn₂S₄/Mo₂TiC₂ Model System

Mo₂TiC₂ provides a representative model for translating metal-layer engineering into photocatalytic function. Our group grew ZnIn₂S₄ nanosheets directly on few-layer Mo₂TiC₂ with thicknesses below 1.80 nm. The tightly coupled interface outperformed multilayer, mechanically mixed, and Ti₃C₂-based controls. The optimized ZnIn₂S₄/10% Mo₂TiC₂ reached 4.3 mmol g⁻¹ h⁻¹, 3.8 times that of ZnIn₂S₄ and higher than the corresponding 1% Pt-loaded system [95]. The enhancement was attributed to Schottky-junction formation, shorter transfer distances, and suppressed electron backflow.

Recent work further shows that Mo₂TiC₂T_x itself can exhibit visible-light-responsive electronic behavior relevant to hydrogen evolution. Delamination enhances the photocurrent response, while optical and electron-energy measurements reveal a band structure compatible with photoinduced reduction. The corresponding absorption, bandgap, and band-edge characteristics are shown in Figure 7 [123].

Figure 7. Electronic and optical properties of delaminated Mo₂TiC₂T_x: (a) UV–visible absorption spectrum; (b) Tauc plot, with the blue dashed line indicating the linear extrapolation used to estimate the optical bandgap; (c) LEIPS spectrum, with the blue dashed line indicating the linear extrapolation used to determine the onset energy; and (d) band-edge diagram showing the energetic positions relevant to hydrogen evolution, where the black vertical dashed lines mark the O₂/H₂O and H⁺/H₂ redox potentials [123]. Reproduced from Ref. [123] under CC BY 4.0.

This interface was later extended from electron extraction to temporary storage. In ZnIn₂S₄/Mo₂TiC₂-Ru, the Mo₂TiC₂-Ru component receives, stores, and releases photogenerated electrons. This electron-parking behavior suppresses back-transfer and promotes water dissociation and H₂ desorption. The composite produced 5.72 mmol g⁻¹ h⁻¹ under visible light and retained approximately 1.67 mmol g⁻¹ h⁻¹ after illumination ceased [124].

The same principle applies to other sulfides. Mo₂TiC₂/ZnCdS achieved 17.73 mmol g⁻¹ h⁻¹ and an apparent quantum efficiency of 16.22% at 475 nm [96]. Mo₂TiC₂ also improved charge transfer and photocorrosion resistance in Mn_0.4Cd_0.6S [125] and CdS [126], while CdIn₂S₄/Mo₂TiC₂ showed a 55.83-fold enhancement over pristine CdIn₂S₄ [97].

Mo₂TiC₂/sulfide interfaces have been described as either Schottky or Ohmic contacts, showing that junction type cannot be inferred from nominal composition alone. Terminations, intercalants, oxidation, work-function matching, morphology, and interfacial bonding can all modify the electronic behavior. Indeed, de-intercalation of Mo₂TiC₂T_x can markedly reduce interflake resistance and shift its apparent transport from semiconductor-like toward metallic behavior, as shown in Figure 8 [101]. These results highlight the need to relate the actual surface and interface state to charge transfer and reaction kinetics.

Figure 8. Influence of intercalated species on the structure and electronic transport of Ti₃CNT_x and Mo₂TiC₂T_x: (a) TGA-MS of Ti₃CNT_x showing the removal of H₂O intercalants and –OH species; (b) in situ EELS of the Ti₃CNT_x O K-edge during annealing; (c) ex situ XRD of the Ti₃CNT_x (002) peak showing decreased interlayer spacing after annealing; (d) temperature-dependent resistance of as-prepared and de-intercalated Ti₃CNT_x; (e) TGA-MS of Mo₂TiC₂T_x showing the removal of H₂O and decomposition of TBA+ intercalants; (f) temperature-dependent resistance of as-prepared and de-intercalated Mo₂TiC₂T_x; and (g) schematic illustration of intercalant-controlled conduction in multilayer Mo₂TiC₂T_x, highlighting the transition from intercalated semiconductor-like transport to de-intercalated metallic behavior [101]. In (a,e), the open circles represent the TGA mass data, and the black arrows indicate the corresponding left y-axes. The vertical dashed line in (c) marks the reference position of the Ti₃CNT_x (002) peak after annealing at 150 °C. In (d,f), the dotted and solid lines represent the as-prepared and annealed samples, respectively. In (g), the red dashed arrows indicate thermally activated interflake hopping. Reproduced from Ref. [101] under CC BY 4.0.

4. Precision Design of MXene Photocatalytic Interfaces

Although this review focuses on photocatalytic H₂ evolution, selected non-H₂ systems are included where they reveal transferable MXene interfacial functions that are difficult to isolate in conventional H₂-evolution systems. These examples are used to clarify electric-field regulation, carrier selectivity, multimetallic-site cooperation, and reaction-specific adsorption rather than as direct benchmarks for H₂ production.

4.1. Working-State Surface Chemistry and Precise Termination Control

Having established how surface terminations regulate electronic structure and HER activity, the next challenge is to control and identify these terminations under realistic synthesis and working conditions. Precise termination engineering requires control over termination identity, coverage, adsorption site, and spatial arrangement rather than reliance on nominal T_x composition. Conventional aqueous etching usually produces heterogeneous O/OH/F surfaces containing intercalated water and residual ions, so the measured properties reflect intrinsic electronic states, interlayer transport, and surface chemistry simultaneously [101]. Postsynthetic covalent exchange has expanded the accessible terminations to NH, S, Cl, Se, Br, and Te [40], while ordered O-B-O layers show that termination arrangement can be as important as elemental identity [127]. Realistic mixed surfaces further exhibit non-additive changes in stability and electronic structure [102].

These limitations have driven synthesis toward better-defined surfaces. Lewis-acid etching, controlled halogenation, and direct growth enable Cl-rich, uniformly terminated, and mixed-halogen MXenes [38,103,111]. Electrochemical etching in dilute HCl produced fluoride-free Ti₂CT_x containing O, OH, and Cl species [128], whereas molten salts yielded nearly saturated Cl terminations on Ti₂C and Ti₃C₂ and were later extended to broader MAX precursors [39,129]. Molten-salt-assisted electrochemical synthesis further enabled the conversion of Cl terminations into O- or S-containing surfaces [130].

Termination formation remains sensitive to both the MXene backbone and reaction environment. Calculations show that a species stable under one condition may be replaced or oxidized under another [131]. Operando XRD further revealed that molten-salt etching proceeds through lattice expansion, transient phases, termination grafting, and eventual over-etching [132]. Direct gas-phase synthesis offers a more defined alternative by forming the lattice and termination simultaneously. Yue et al. [133] prepared few-layer, single-phase Ti₂CCl₂ and Ti₂NCl₂ in one step, avoiding conventional etching and post-delamination. Figure 9 summarizes the synthesis route and structural evidence for phase-pure, Cl-terminated MXenes with layered morphology and high stability [133].

Figure 9. (a) Schematic diagram of the one-step gas-phase reactor. (b,c) XRD patterns and Rietveld refinements of Ti₂CCl₂ and Ti₂NCl₂. (d,e) SEM images of directly synthesized and dispersed Ti₂CCl₂. (f) Atomic-resolution HAADF-STEM image of Ti₂CCl₂. (g,h) SEM images of directly synthesized and dispersed Ti₂NCl₂. (i) Atomic-resolution HAADF-STEM image of Ti₂NCl₂. (j,k) EXAFS fitting and wavelet-transform analysis of Ti₂NCl₂. (l) EPR spectra of directly synthesized and etching-derived MXenes [133]. Reproduced from Ref. [133] under CC BY 4.0.

Reliable surface identification is equally important. Thermal or H₂ treatment can remove existing terminations and enable retermination [134], while heating redistributes O and F among nonequivalent sites [135]. UPS/XPS identified Ti-F hybridization and different O adsorption sites but found limited evidence for intrinsic OH on impurity-minimized Ti₃C₂T_x [136]. Solid-state NMR, however, detected OH- and H₂O-related environments in conventionally prepared samples [137]. Adsorbed water, surface oxides, residual phases, and handling effects may therefore be mistaken for intrinsic terminations, requiring surface- and bulk-sensitive signals to be distinguished [138].

The working surface may also differ from the as-synthesized state. H₂O can bind at Ti-Ti bridge sites and behave as an additional termination species [88], while photoexcitation can alter the stability of F, O, OH, H, halogen, and chalcogen terminations [139]. Surface reduction may further couple termination removal with electron injection and subsequent O/OH recoordination [140].

Precise termination design should therefore integrate controlled synthesis, reliable surface identification, and working-state verification. The relevant target is not the nominal M_n+1X_nT_x composition, but the termination distribution and local coordination retained under photocatalytic conditions.

4.2. MXene Quantum Dots for Nanoscale Interface and Electric-Field Regulation

Reducing MXenes from two-dimensional sheets to quantum dots creates spatially dispersed interfacial regulators. Their small size, high dispersibility, exposed edges, and rich surface chemistry increase contact density, shorten transfer distances, and reduce light shielding. In g-C₃N₄/Ti₃C₂ QD composites, the optimized system reached 5111.8 μmol g⁻¹ h⁻¹, approximately ten times the activity of the corresponding Ti₃C₂-sheet composite [141]. Ti₃C₂ QDs have also been incorporated into BiVO₄@ZnIn₂S₄ direct Z-scheme [142] and TiO₂/g-C₃N₄ S-scheme systems [143], where they extract electrons and provide additional reduction sites without disrupting the primary transfer pathway.

Their function can be further tuned through defects, doping, and terminations. Partially reduced Ti₃C₂ QDs contain Ti³⁺ and oxygen-vacancy sites that combine electron trapping with reactant activation [144]. Nitrogen doping strengthens coupling with CdS nanorods and increases the H₂ evolution rate to 17.09 mmol g⁻¹ h⁻¹ [98]. More selective carrier routing was achieved by coloading O-terminated Ti₃C₂ QDs and N-doped carbon dots on CdS, where the two components acted as electron- and hole-accepting units, respectively, giving 5.64 mmol g⁻¹ h⁻¹ and an apparent quantum efficiency of 45.7% at 450 nm [145]. Ti₃C₂ QDs may also provide photothermal promotion, although this contribution should be separated from charge-transfer effects [99].

A more distinctive function of nanoscale MXene interfaces is local electric-field regulation. A recent non-H₂ system provides direct evidence for this principle. At a Cl-terminated MXene/perovskite interface, MXene-induced dipoles modified the local work function and generated n-type character at the buried interface. Combined with an opposite surface potential gradient, this produced dipole-driven band bending that promoted carrier extraction and suppressed recombination, as shown in Figure 10 [146]. In Figure 10a, the differently colored spheres distinguish the atomic species in the MX-Cl/MAPbI₃/4-FPEAI interface, while the red arrows indicate interfacial work-function shifts. In Figure 10b, the horizontal and vertical arrows in the insets indicate the short-circuit current density (JSC) and open-circuit voltage (VOC), respectively. Although demonstrated in a photovoltaic interface, this result provides a transferable model for understanding how MXene surface chemistry and spatial confinement can regulate local electrostatic fields.

Figure 10. MXene-induced interfacial field-effect regulation: (a) optimized atomic structure of the MX-Cl/MAPbI₃/4-FPEAI interface and the corresponding plane-averaged electrostatic-potential profile, showing asymmetric interfacial work functions; and (b) simulated J–V characteristics of the reference, 4-FPEAI-modified, and MX-Cl/4-FPEAI devices, with corresponding JSC and VOC values [146]. Reproduced from Ref. [146] under CC BY 4.0.

This field-engineering principle is directly relevant to photocatalytic MXene QDs. The Nb₂C QD/COF system further demonstrated that highly dispersed QDs can act as nanoscale dipoles rather than simple electron sinks. The QDs strengthened the built-in electric field, accelerated charge separation, and optimized O₂ adsorption, enabling sacrificial-agent-free H₂O₂ production from air and seawater [147]. KPFM, light-induced XPS, transient absorption, and theoretical calculations linked the interfacial potential difference with carrier kinetics and reaction energetics, as shown in Figure 11.

Figure 11. (a,b) Surface potentials of PY-DHBD-COF and NQPD-3 measured by KPFM. (c) Zeta potentials of PY-DHBD-COF and NQPD-3. (d) Photoluminescence and time-resolved photoluminescence spectra. (e) EIS Nyquist plots and transient photocurrent-response curves. (f–i) In situ light-irradiation XPS spectra of NQPD-3. In (g–i), the red arrows indicate the light-induced binding-energy shifts of N 1s (−0.2 eV), O 1s (−0.2 eV), and Nb 3d (+0.3 eV), respectively [147].

Recent multicomponent systems extend MXene QDs from terminal acceptors to electronic relays. In Ni₂P/Zn_0.5Cd_0.5S@Ti₃C₂ QDs, the QDs facilitate electron transport, while Ni₂P provides the main HER sites [148]. Ti₃C₂ QDs have also been coupled with BaTiO₃ polarization fields, combining piezoelectric charge separation with QD-mediated electron extraction [149].

Dimensional reduction, however, simultaneously changes size, terminations, defect density, oxidation state, and work function. Improved performance should therefore not be attributed to quantum confinement alone. Reliable identification requires matched sheet/QD controls together with work-function measurements, KPFM, light-induced spectroscopy, ultrafast dynamics, and charge-density calculations. The design objective is to use MXene QDs as distributed electronic nodes that integrate short-range charge extraction with local field and active-site regulation.

4.3. Emerging Compositionally Complex MXenes for Cooperative Active-Site Design

Solid-solution MXenes extend composition control beyond discrete single- and double-metal structures. In Ti₂-gammaNbgammaCT_x, Ti₂-gammaVgammaCT_x, and V₂-gammaNbgammaCT_x, continuous M-site substitution produces composition-dependent changes in conductivity and optical absorption [150]. TiVCT_x and Mo_xV_4−xC₃T_x further show that the metal ratio also affects termination chemistry and surface redox behavior [151,152]. Similar Nb-containing solid solutions confirm that transport, surface chemistry, and stability can be regulated through composition [153].

High-entropy MXenes further broaden this space by integrating multiple transition metals within one lattice. Four-metal TiVNbMoC₃T_x and TiVCrMoC₃T_x established the feasibility of high-entropy atomic layers [154], followed by five-metal M₂CT_x, M₄C₃T_x, and M₅C₄T_x structures with wider distributions of local coordination and strain [155,156]. Medium-entropy Ti_1.1V_1.1Cr_0.4Nb_1.4C₃T_x also shows that stable complex compositions need not be equimolar [157]. The objective should therefore be reaction-specific composition design rather than simply increasing the number of metals.

Average composition alone does not define the active surface. Controlled Mo/Nb occupation in Mo_2+αNb_2−αC₃T_x changes both conductivity and HER activity [118]. Entropy-driven transitions from short-range order to greater disorder have also been observed in multimetallic carbides [158], while high-entropy MXenes may contain multihyperuniform arrangements that differ from ideal random models [159]. Active-site models must therefore consider local ordering, segregation, and coordination.

The catalytic value of compositional complexity lies in combining neighboring sites with complementary functions. In Ti-V-Mo-Cr high-entropy MXenes, different local environments provide a distribution of H* adsorption strengths. Short-range Mo ordering creates favorable migration pathways, allowing H spillover between strong- and weak-binding sites and lowering the Volmer–Tafel barrier [160]. High-entropy MXenes can also regulate supported catalytic centers. In V-Co₂P/high-entropy MXene interfaces, electronic coupling optimizes water and intermediate adsorption, and lowers the barriers of both HER and OER [161].

The VNbC/COF system provides a photocatalytic example. Alloyed V-Nb sites promoted interfacial charge transfer and strengthened styrene adsorption, enabling nearly complete conversion with 99% epoxide selectivity during coupled H₂O₂ photosynthesis and alkene epoxidation [162]. The VNbC/COF structure and cooperative V-Nb sites are shown in Figure 12. This result demonstrates that compositionally complex MXenes can coordinate carrier transfer and reaction-specific adsorption rather than merely increase the number of nominal active sites.

Figure 12. (a) Schematic illustration of the synthesis of COF-TpBpy@VNbC, where the dark-green, dark-blue, light-green, gray, and blue spheres represent V, Nb, Al, C, and O atoms, respectively. (b) TEM image of VNbC. (c) TEM image of TBNV-3, with the inset showing the selected-area electron diffraction (SAED) pattern. (d) HRTEM image of TBNV-3, where the orange dashed outlines highlight the VNbC and COF-TpBpy regions. (e) Aberration-corrected STEM and (f) HAADF-STEM images of VNbC, with the orange dashed box in (f) highlighting the local atomic arrangement. (g) EDX elemental mappings of TBNV-3, with C, N, O, V, and Nb shown in red, cyan, purple, green, and yellow, respectively. (h,i) XANES and (j,k) EXAFS spectra of TBNV-3 and the reference samples. Wavelet-transform EXAFS plots at the (l) V K-edge and (m) Nb K-edge, where the color scale from blue to red indicates increasing signal intensity [162].These systems also illustrate how MXene composition shifts the dominant interfacial function. Ti₃C₂T_x provides the classical termination-sensitive platform for electron extraction and transport [43], whereas Mo₂TiC₂ couples multimetallic electronic regulation with H-adsorption control [123]. At reduced dimensions, Nb₂C QDs can act as local dipoles that strengthen interfacial electric fields [147], while VNbC introduces cooperative multimetallic sites for carrier transfer and reaction-specific adsorption [162]. These distinctions are not intrinsic labels, however, because the observed role also depends on termination state, dimensionality, semiconductor partner, and reaction environment. Direct applications of high-entropy MXenes in photocatalytic H₂ evolution remain limited. Future work should identify defined local ensembles for electron extraction, adsorption, intermediate migration, and bond activation using atomic-resolution mapping, operando spectroscopy, and calculations based on realistic metal and termination distributions. The principal structure-encoded variables and their functional consequences are summarized in Table 3.

4.4. Reaction-Specific Role Design and AI-Assisted Materials Discovery

MXene cocatalysts should be designed for the target reaction rather than assigned a fixed electron-sink role. Replacing sacrificial donors with productive oxidation reactions enables simultaneous use of photogenerated electrons and holes. Ti₃C₂T_x/CdS, for example, couples H₂ evolution with the selective oxidation of furfuryl alcohol to furfural [163]. Figure 13 shows the coupled reaction pathway, the dependence of product yields and substrate conversion on Ti₃C₂T_x content, and the cycling stability of the optimized composite [163]. CdS-Ti₃C₂T_x similarly integrates H₂ evolution with bioethanol conversion to 1,1-diethoxyethane [164], while MXene/Zn_xCd_1−xS combines H₂ production with PET degradation [165]. In these systems, the MXene function is defined within the complete redox process rather than by proton reduction alone.

Figure 13. (a) Photocatalytic conversion of furfuryl alcohol to furfural coupled with H₂ evolution, with the arrow indicating the reaction under visible-light irradiation. (b) Product yields and furfuryl-alcohol conversion over CdS and Ti₃C₂T_x/CdS composites with different Ti₃C₂T_x contents under visible-light irradiation for 4 h, where the left-pointing arrows indicate the H₂ and furfural yields corresponding to the left y-axis, and the right-pointing arrow indicates furfuryl-alcohol conversion (dashed line) corresponding to the right y-axis. (c) Recycling performance of 0.5% Ti₃C₂T_x/CdS [163].

Table 3. Structure-encoded design variables and their functional consequences in MXene cocatalysts.

Design Variable	Structural Implementation	Primary Electronic/Interfacial Effect	Cocatalytic Consequence	Key Descriptors for Comparison	Representative Refs.
Metal-layer composition and atomic ordering	Select outer/inner transition-metal layers and control layer-resolved occupation in ordered double-transition-metal MXenes.	Reconstructs d states near the Fermi level, conductivity, metal–termination orbital coupling, and adsorption energetics.	Couples rapid electron transport with tunable H* binding; separates backbone transport from surface-reaction functions.	Work function, DOS and orbital occupation; d-band descriptor, ΔGH*, and layer-order parameter.	[104,106,118]
Solid-solution and medium/high-entropy composition	Continuous M-site substitution or multimetal lattices with controlled short-range order rather than assuming an ideal random alloy.	Creates distributions of local coordination, charge, strain, and adsorption energies that are not captured by average composition.	Enables complementary strong- and weak-binding sites, H* migration/spillover, and reaction-specific local ensembles.	Local composition and pair correlations, short-range-order parameter, strain distribution, ΔGH* distribution, and migration barrier.	[150,154,158,160,166]
Termination identity, coverage, and spatial distribution	Regulate O, OH, F, Cl/Br and other anionic terminations; control mixed-surface ratios and adsorption sites.	Changes surface dipoles, work function, charge redistribution, metal–Tx bonding, and interfacial band alignment.	Tunes electron/hole selectivity, back-transfer suppression, water/proton interaction, and H* adsorption strength.	Termination coverage and site occupancy, work function, surface potential, Bader charge, ΔGH*, and termination stability.	[101,114,115,116]
Thickness, dimensionality, and synthesis/processing route	Control layer number, lateral size, delamination, direct CVD growth, or one-step gas-phase formation.	Modulates sheet and interflake resistance, carrier-scattering length, accessible area, oxidation susceptibility, and contact geometry.	Controls extraction distance, density of interfacial nodes, light shielding, and batch-to-batch reproducibility.	Layer number, lateral size, conductivity, interflake resistance, oxidation fraction, and effective interface area.	[61,110,111,133]
Vacancies and substitutional sites	Introduce metal/cation vacancies, second-shell defects, or isolated substitutional metals while retaining the MXene framework.	Produces local coordination asymmetry, charge polarization, anchoring sites, and modified reactivity of neighboring terminations.	Can strengthen interfacial fields, stabilize single atoms, and tune adjacent O-terminated HER sites without making the dopant the direct H* site.	Defect formation energy, coordination number, Bader charge, local DOS, and vacancy concentration and spatial distribution.	[92,117]
Supported single atoms, clusters, and nanoparticles	Immobilize Pt or Ru single atoms, Ru clusters, or Pd nanoparticles through MXene C/O/heteroatom coordination.	Generates metal–support charge transfer, suppresses aggregation, and differentiates water-dissociation and hydrogen-binding sites.	Adds kinetically specialized centers for water activation, H* formation, recombination, and H2 desorption.	M–C/O coordination; oxidation state, dispersion, metal loading, water-dissociation barrier, and ΔGH*.	[94,119,120,121]

Table 3. Structure-encoded design variables and their functional consequences in MXene cocatalysts.

Design Variable	Structural Implementation	Primary Electronic/Interfacial Effect	Cocatalytic Consequence	Key Descriptors for Comparison	Representative Refs.
Metal-layer composition and atomic ordering	Select outer/inner transition-metal layers and control layer-resolved occupation in ordered double-transition-metal MXenes.	Reconstructs d states near the Fermi level, conductivity, metal–termination orbital coupling, and adsorption energetics.	Couples rapid electron transport with tunable H* binding; separates backbone transport from surface-reaction functions.	Work function, DOS and orbital occupation; d-band descriptor, ΔGH*, and layer-order parameter.	[104,106,118]
Solid-solution and medium/high-entropy composition	Continuous M-site substitution or multimetal lattices with controlled short-range order rather than assuming an ideal random alloy.	Creates distributions of local coordination, charge, strain, and adsorption energies that are not captured by average composition.	Enables complementary strong- and weak-binding sites, H* migration/spillover, and reaction-specific local ensembles.	Local composition and pair correlations, short-range-order parameter, strain distribution, ΔGH* distribution, and migration barrier.	[150,154,158,160,166]
Termination identity, coverage, and spatial distribution	Regulate O, OH, F, Cl/Br and other anionic terminations; control mixed-surface ratios and adsorption sites.	Changes surface dipoles, work function, charge redistribution, metal–Tx bonding, and interfacial band alignment.	Tunes electron/hole selectivity, back-transfer suppression, water/proton interaction, and H* adsorption strength.	Termination coverage and site occupancy, work function, surface potential, Bader charge, ΔGH*, and termination stability.	[101,114,115,116]
Thickness, dimensionality, and synthesis/processing route	Control layer number, lateral size, delamination, direct CVD growth, or one-step gas-phase formation.	Modulates sheet and interflake resistance, carrier-scattering length, accessible area, oxidation susceptibility, and contact geometry.	Controls extraction distance, density of interfacial nodes, light shielding, and batch-to-batch reproducibility.	Layer number, lateral size, conductivity, interflake resistance, oxidation fraction, and effective interface area.	[61,110,111,133]
Vacancies and substitutional sites	Introduce metal/cation vacancies, second-shell defects, or isolated substitutional metals while retaining the MXene framework.	Produces local coordination asymmetry, charge polarization, anchoring sites, and modified reactivity of neighboring terminations.	Can strengthen interfacial fields, stabilize single atoms, and tune adjacent O-terminated HER sites without making the dopant the direct H* site.	Defect formation energy, coordination number, Bader charge, local DOS, and vacancy concentration and spatial distribution.	[92,117]
Supported single atoms, clusters, and nanoparticles	Immobilize Pt or Ru single atoms, Ru clusters, or Pd nanoparticles through MXene C/O/heteroatom coordination.	Generates metal–support charge transfer, suppresses aggregation, and differentiates water-dissociation and hydrogen-binding sites.	Adds kinetically specialized centers for water activation, H* formation, recombination, and H2 desorption.	M–C/O coordination; oxidation state, dispersion, metal loading, water-dissociation barrier, and ΔGH*.	[94,119,120,121]

Reaction-specific design can also extend MXene functions toward oxidation. In La-MOF@MXene, interfacial electron withdrawal shifts the valence-band position and promotes hole migration [167]. Ti₃C₂ can likewise serve as a hole-transfer and accumulation component in an appropriately aligned MOF interface [168]. Our VNbC/COF system combines carrier transport with styrene adsorption and selective epoxidation [162], while Nb₂C QDs regulate the built-in electric field and O₂ reduction pathway during COF-based H₂O₂ production [147]. MXenes can therefore be tailored to regulate carrier transfer, reactant adsorption, and intermediate conversion according to the required reaction sequence.

The growing combinations of metal layers, terminations, defects, and interfaces make exhaustive experimental screening impractical. Early data-driven studies screened ordered binary MXenes using stability, conductivity, and hydrogen-adsorption free energy, while identifying geometric and elemental descriptors for HER activity [109]. Machine-learning and DFT workflows later predicted hydrogen-adsorption free energies across 4500 MM′XT₂ configurations [169]. For CO₂ activation, interpretable models identified the d-band center, surface-metal electronegativity, and valence-electron count as key descriptors [170]. These approaches accelerate screening but remain largely focused on isolated adsorption properties.

Photocatalysis-oriented screening requires broader criteria. Janus MXenes with asymmetric metal layers or terminations can generate intrinsic dipoles, realign band edges, and spatially separate reduction and oxidation reactions [171]. The MXgap model predicts metallic or semiconducting behavior and bandgaps, enabling rapid selection based on optical absorption and band alignment [172]. Deep-learning screening has further incorporated thermodynamic stability, band alignment, light absorption, carrier effective masses, and nonadiabatic electron–hole recombination [173]. Figure 14 presents the screened composition and termination space, four representative MXene configurations, and the high-throughput workflow used to evaluate 23,857 structures and identify 14 candidate photocatalysts [173].

Figure 14. Deep learning-enabled high-throughput screening of MXene photocatalysts, integrating stability prediction, band alignment, light absorption, carrier properties, nonadiabatic electron–hole recombination, and built-in electric-field evaluation [173].

Reaction-specific discovery must also consider multiple intermediates and realistic local environments. Models covering H₂O, CO₂, CO, H₂, O₂, OH, O, and H adsorption show that different species and terminations can be treated within a common predictive framework [174]. In high-entropy TiVNbMoC₃O₂, H adsorption depends mainly on the first-neighbor metal environment around surface O sites rather than on average composition. Machine learning further identified the mean covalent radius of neighboring metals as an effective local descriptor [166]. Future models should therefore represent adsorption-site geometry, local coordination, termination distribution, and competing intermediates.

AI-assisted MXene discovery should therefore be reaction-defined and multiobjective, considering charge-transfer direction, adsorption energetics, kinetic barriers, selectivity, stability, and light absorption. Candidate structures must ultimately be verified under working conditions because reconstruction, oxidation, solvent adsorption, and charge redistribution may alter their predicted functions. The goal is to connect target reactions with experimentally verifiable MXene interfacial roles.

5. Conclusions and Perspectives: Toward Programmable MXene Interfacial Cocatalysts

The preceding sections show that MXene cocatalytic functions emerge from the interplay of electronic structure, interfacial coupling, surface chemistry, dimensionality, and reaction conditions. This section therefore consolidates these mechanistic insights, identifies unresolved questions under working conditions, and outlines a path toward reaction-specific and programmable MXene interfacial cocatalysts.

5.1. Mechanistic Convergence and Role Diversification

The Ti₃C₂T_x electron-sink model remains the starting point for understanding MXene-assisted photocatalytic H₂ evolution. In early metal-sulfide/Ti₃C₂ systems, Fermi-level equilibration, conductive transport, and O-terminated HER sites jointly enabled electron extraction, transfer, and proton reduction [43]. An effective MXene cocatalyst must therefore connect charge separation with surface electron consumption rather than act only as a conductive additive.

This role is conditional rather than universal. In ZnIn₂S₄/Mo₂TiC₂, metal-layer composition and local coordination regulate both electron extraction and H adsorption [123]. Our VNbC/COF system further shows that multimetallic sites can couple carrier transfer with reaction-specific adsorption [162]. Terminations, defects, and interfacial bonding can therefore reshape work functions, orbital coupling, and adsorption energetics [93].

MXenes may consequently receive, transport, store, or consume charge. Suitable band alignment can promote hole transfer instead of electron extraction [168]. Vacancies can stabilize isolated HER centers [94], while Mo₂TiC₂-Ru enables temporary electron storage and delayed release [124]. MXene quantum dots can further regulate local electric fields and reactant activation [146,147].

MXenes should therefore be defined by their working interfacial function rather than nominal composition. Their dominant role depends on the structure of the interface and the rate-limiting step of the target reaction.

5.2. Unresolved Questions in Working-State MXene Interfaces

A central uncertainty is whether the characterized structure represents the interface operating during photocatalysis. Water, pH, temperature, illumination, and adsorbates can alter terminations, coordination, and oxidation states. Ti₃C₂T_x readily oxidizes in liquids [175], particularly under acidic or heated conditions [176]. Water can also adsorb on surface Ti, weaken Ti-C bonds, dissociate, and extract Ti atoms [177]. The working surface may therefore differ substantially from the nominal O/OH/F structure [88,139].

Dynamic surface chemistry also complicates charge-transfer assignment. Similar Mo₂TiC₂/sulfide interfaces have been described as either Schottky or Ohmic contacts [123,125]. In situ KPFM further shows that high-work-function MXenes favor electron acceptance, whereas OH-rich surfaces may promote hole transfer [178]. Nominal composition, PL quenching, photocurrent, or impedance alone cannot establish the carrier pathway.

The catalytic site is equally uncertain. H* adsorption may occur on O terminations [112], while substitutional metals can regulate neighboring O sites [117]. Vacancies may anchor isolated metals [94], and adsorbed water, exposed metal atoms, reconstructed O/OH species, or oxide phases may also participate [88,140]. Electron accumulation on MXene therefore does not identify the proton-reduction site.

Surface quantification introduces further ambiguity. XPS signals may contain contributions from terminations, water, oxides, residual phases, and adventitious species [136,137]. Conventional analysis can overestimate Ti vacancies and underestimate terminal atoms, particularly F [179]. Reliable assignment requires corrected XPS combined with XAS, NMR, SIMS, or atomic-resolution microscopy.

Practical evaluation presents a separate challenge. Many high-performing MXene photocatalysts still rely on sacrificial donors and Cd-based sulfides, so high mass-normalized H₂-evolution rates do not directly translate to overall water splitting or scalable solar H₂ production. Meaningful comparison therefore requires standardized reporting of photon flux, catalyst loading, AQY/AQE, long-term stability, and post-reaction MXene composition. Solar-to-hydrogen efficiency should be used when true overall water splitting is demonstrated, rather than being extrapolated from sacrificial-agent-assisted H₂ evolution.

Because synthesis and processing often alter multiple variables simultaneously, future studies should use matched controls to connect the as-synthesized structure with working-state reconstruction, local potential, carrier dynamics, intermediates, and products. The central question is which evolving structure performs each interfacial function under reaction conditions. The corresponding validation requirements are summarized in Table 4.

Table 4. Working-state questions, minimum evidence requirements, and validation criteria for assigning MXene interfacial roles.

Working-State Question	Major Confounding Factors	Minimum Evidence Set	Acceptable Validation Criterion	Representative Refs.
Does the MXene remain structurally intact during photocatalysis?	Water, dissolved O2, pH, temperature, irradiation, and defects can oxidize or dissolve the carbide and generate electronically coupled oxide phases.	Time-resolved or operando structural/chemical-state measurements, supported by pre/post-reaction XRD, Raman, XPS/XAS, TEM, and dissolved-metal analysis.	Distinguish intact MXene from partially oxidized MXene and MXene-derived TiO2/NbOx/MoOx; stability cannot be inferred from activity retention alone.	[175,176,177]
What are the identity, coverage, and adsorption sites of the working terminations?	Intrinsic terminations overlap with adsorbed/intercalated water, surface oxides, etching residues, and adventitious species; heating and photoexcitation can redistribute them.	XPS/UPS combined with at least one chemically independent probe such as solid-state NMR, XAS, controlled desorption, or atomic-resolution microscopy; report fitting constraints and uncertainty.	Assign termination identity, coverage, site occupancy, and environmental dependence rather than describing the surface only as nominal O/OH/F-Tx.	[88,135,136,137,138,139]
Does an irradiated-XPS binding-energy shift establish charge-transfer direction and magnitude?	Differential charging, surface photovoltage, final-state screening, calibration choices, and peak-fitting variability can mimic or modify shifts.	Paired dark/light spectra with internal calibration and repeated measurements, integrated with work-function/band-alignment data and an independent spatial or time-resolved carrier probe.	Use irradiated XPS as one element of a convergent evidence chain; do not derive transferred-electron numbers or a junction type from peak shifts alone.	[86]
Which working-state site actually performs proton reduction?	Surface O/OH, exposed metal atoms, vacancies, supported species, adsorbed water, or reconstructed oxides may all bind H*; electron accumulation does not identify the active site.	Calculations on realistic mixed/reconstructed surfaces plus site-sensitive experiments, such as isotope labeling/KIE, selective poisoning, operando vibrational or XAS measurements, and rate correlation.	Connect the proposed site to a specific intermediate and elementary step, and explain activity changes under matched site perturbations.	[112,113]
Does synthesis or operation induce phase and termination reconstruction?	Molten-salt etching, over-etching, reduction, retermination, and electron injection can generate transient phases and surfaces absent from the nominal starting material.	Operando or time-resolved phase/chemical-state tracking, quenched intermediate samples, and mass/composition balance across initial, intermediate, and working states.	Establish a continuous pathway linking the initial structure, intermediate states, and working structure before assigning the catalytic role to the as-synthesized phase.	[132,140]

5.3. Design Roadmap Toward Programmable Cocatalytic Platforms

Programmable MXene cocatalysts should be designed from the target reaction. Required functions such as carrier extraction, back-transfer suppression, adsorption, intermediate stabilization, and product desorption can be encoded through metal composition, terminations, defects, dimensionality, and interfacial bonding [40,103,127,147,158]. Their successful design ultimately requires verification under working conditions.

Reliable design requires structure–function relationships under operating conditions. Static models cannot capture termination reconstruction, oxidation, solvent adsorption, or changes in interfacial potential. Computational screening should therefore be combined with corrected surface analysis, operando spectroscopy, and local-potential measurements [173,178,179]. AI-ready datasets should link synthesis, structure, spectra, reaction conditions, negative results, and uncertainty [180].

Closed-loop experimentation provides a practical route through this design space. A mobile robotic platform completed 688 photocatalytic experiments and identified a formulation with sixfold higher activity [181]. Machine learning-guided screening also enriched active candidates within large ternary heterojunction libraries [182]. For MXenes, optimization should include efficiency, selectivity, stability, oxidation resistance, and reproducibility rather than H₂ evolution rate alone.

Working-state characterization should be integrated into this feedback loop. Operando quick XAS combined with machine learning has resolved dynamic coordination and phase evolution [183], illustrating how working-state information can refine theoretical models and guide subsequent synthesis.

Active learning is well suited to the limited and heterogeneous datasets available for MXenes.

The A-Lab synthesized 36 of 57 targeted inorganic compounds by integrating computation, literature knowledge, robotics, characterization, and active learning [184]. Another closed-loop study identified a high-performing four-metal catalyst from 30 initial samples and ten additional experiments [185]. These closed-loop platforms have not yet reached comparable maturity for MXene photocatalysts and are discussed here as transferable methodological models rather than established MXene-specific workflows. MXene datasets should therefore record terminations, oxidation states, interlayer species, morphology, and interface quality together with composition.

Programmability must finally survive scale-up. Automated screening of molecular nanojunction photocatalysts was transferred to continuous-flow production at approximately 200-fold scale, as shown in Figure 15 [186]. Similar translation for MXenes will require reproducible etching, delamination, termination control, semiconductor growth, and oxidation management.

Figure 15. (a) Pipeline for the accelerated discovery, scale-up production, and computational rationalization of molecular nanojunction photocatalysts. (b) Chemical structures of the molecular acceptors and donors used to construct the combinatorial library [186]. Reproduced from Ref. [186] under CC BY 4.0.

A practical design strategy should connect reaction requirements with controllable structural variables, working-state verification, and multiobjective optimization before scale-up. Programmable MXene cocatalysis will depend not on reporting more materials, but on reproducibly controlling defined interfacial functions for specific reactions.