Multicomponent porous metal oxide semiconductors for advanced gas sensing

Electron depletion layer theory and hole accumulation layer theory

For chemiresistive MOS sensors, the sensing mechanism is primarily attributed to electrical signal changes induced by gas adsorption and desorption. At the microscopic level, when MOS materials are exposed to air, oxygen molecules adsorb onto the surface of the sensing material by capturing electrons from the conduction band, subsequently ionizing into chemically adsorbed oxygen species (such as O₂^-, O^-, or O^2- ions). The type of chemically adsorbed oxygen is strongly dependent on the operating temperature (T). At relatively low temperatures (T < 150 °C), O₂^- is the predominant oxygen species. Between 100 and 300 °C, O^- ions are primarily adsorbed onto the MOS surface. At higher temperatures (T > 300 °C), the O^2- species become dominant. The formation process of oxygen species can be summarized by^[35,36]:

Meanwhile, an EDL with higher resistance than the core region forms on the surface of n-type MOS, or a HAL with lower resistance than the core region forms on the surface of p-type MOS. This results in an increase (or decrease) in the resistance of the n-type (or p-type) MOS. When MOS materials are exposed to an atmosphere containing target gases, redox reactions occur between gas molecules and chemically adsorbed oxygen species. This further results in the release of electrons back into the conduction band of the MOS (for reducing gases) or the capture of additional electrons from the MOS (for oxidizing gases). Accordingly, the thickness of the EDL (or HAL) and the resistance of the MOS change. For instance, when n-type MOS material is exposed to reducing gases (such as acetone), the thickness of the EDL decreases, consequently reducing the resistance of the n-type MOS material [Figure 2A]^[37]. Conversely, when oxidizing gases are introduced, both the EDL thickness and the resistance of the n-type MOS further increase. For p-type MOS materials, when reducing gases (such as acetone) are introduced, the resistance increases due to the recombination of released electrons and holes, resulting in a decrease in HAL thickness [Figure 2B]. This behavior is the opposite of that observed when oxidizing gases are introduced^[38]. Following these processes, the resistance of the MOS material undergoes significant changes, exhibiting a pronounced response to the target gas.

Figure 2. (A) Schematic of the electron depletion layer theory. Mechanistic representation of acetone sensing by an n-type MOS sensor (A₁); Initial electronic levels (A₂); Changes in the electronic levels of n-type MOS upon exposure to oxygen (A₃) and target gases (A₄). Figure 2A is quoted with permission from Panigrahi et al.^[37]; (B) Core-shell configuration of p-type MOS in vacuum and air (B₁); Interaction of p-MOS in the presence of reducing gas CO (B₂) and in oxidizing NO₂ (B₃). Figure 2B is quoted with permission from Kaur et al.^[38]; (C) Schematic illustration of gas diffusion processes at different pore sizes; (D) Schematic representation of gas diffusion processes within hollow tin dioxide microspheres at different temperatures. Figure 2C and D is quoted with permission from Wang et al.^[34]. MOS: Metal oxide semiconductor.

Bulk resistance control mechanism

The bulk resistance control mechanism fundamentally relies on the resistance variation of the MOS sensors originating from phase transitions within the sensitive materials. The mechanism has a relatively limited applicability, and is only suitable for analyzing gas sensing processes involving γ-Fe₂O₃ and perovskite-type metal oxides (ABO₃), where ABO₃ denotes the general chemical formula of perovskite structures.^[30]. Wang et al. developed Fe₂O₃ sensors with distinct α-Fe₂O₃ and γ-Fe₂O₃ phase compositions by heating Fe-metal-organic framework (MOF) templates, and analyzed the differing gas sensing mechanisms of the two iron oxides^[39]. α-Fe₂O₃ possesses a relatively stable phase structure, and its sensing mechanism can be explained by a typical oxygen adsorption model. However, the resistance change in γ-Fe₂O₃ is primarily attributed to alterations in its internal phase structure. γ-Fe₂O₃ features an inverse spinel structure, wherein the octahedral voids (Fe²⁺ and half Fe³⁺ ions) typically feature a vacant cation site (□) to compensate for the increased positive charge. When exposed to a reducing gas environment, the Fe³⁺ ions within the octahedral sites are reduced to Fe²⁺, causing the material to transform into Fe₃O₄. Owing to the existence of the (Fe²⁺-Fe³⁺-Fe²⁺-Fe³⁺) structure within Fe₃O₄, electron migration becomes significantly facilitated, resulting in a material whose resistivity is ten orders of magnitude smaller than that of γ-Fe₂O₃. Upon re-exposure to air, Fe₃O₄ re-oxidizes to γ-Fe₂O₃, with its electrical resistance increasing accordingly. It should be noted that the redox reactions occurring during gas sensing are reversible, involving only simple phase transitions without any crystal structure transformation. These mechanisms are expressed as^[40,41]:

where T_h denotes the tetrahedral interstitial, □ represents a cation vacancy, O_h signifies the octahedral interstitial, and x is the reduction degree of Fe³⁺.

Gas diffusion control mechanism

In gas sensing processes, two key components are typically involved: the sensitive material and the target gas. The aforementioned EDL, HAL and bulk resistance control mechanisms primarily focus on the physical and chemical properties of the sensing materials, whereas the gas diffusion control mechanism places greater emphasis on the transport behaviors of the gas within sensing materials. The most significant factor influencing gas diffusion is the material morphology. As early as the 1990s, scientists proposed the theory that gas diffusion governs the sensitivity of MOS sensors^[42-44]. Indeed, this theory serves as a complement to the former two mechanisms, offering a more comprehensive explanation for the variations in physical parameters during gas sensing. By studying the diffusion process of gases through material surfaces, we can gain deeper insights into the operating principles of gas sensors and provide theoretical guidance for designing and optimizing novel intelligent sensors.

Wang et al. constructed pore channel models through research on SnO₂ microspheres, comprehensively explaining the relationship between surface chemical reactions (SCR) and gas diffusion, along with their impact on gas sensing performance. Based on Knudsen diffusion theory and reaction kinetics, the gas diffusion rate (D_k) and SCR rate (k) are respectively expressed by^[34]

where r denotes the pore radius, R is the gas constant (8.314 J·mol^-1·K^-1), T represents the temperature, M is the molecular weight, A is the pre-exponential factor, and E_a denotes the activation energy. Based on the two equations, k is primarily determined by T. At low temperatures (T < 180 °C), the SCR rate is relatively low, making it the rate-determining step controlling the overall reaction. Furthermore, the entire process occurs on the surface, with the external surface area the primary factor controlling gas sensing performance. At elevated temperatures (T > 260 °C), the increase in D_k is less pronounced than that of k, rendering the process diffusion-controlled, in which pore size exerts a more significant influence on gas sensing performance. At intermediate temperatures (180 °C ≤ T ≤ 260 °C), gas diffusion and SCR compete in equilibrium, with the entire system operating under dual control. Overall, as temperature increases, diffusion control progressively supplants SCR control. Based on this, this work proposes a pore channel model to analyze the diffusion of gas molecules [Figure 2C]. At low temperatures (T < 180 °C), the mean free path of gas molecules exceeds the pore diameter, thereby impeding their diffusion into the channels and causing most responses to occur outside the channels. Although both diffusion and SCR rates are relatively low, the SCR rate plays a decisive role. As temperature increases, the mean free path of gas molecules gradually diminishes to match the pore size, enabling more molecules to diffuse into the pore channels and react with oxygen species, forming a transition state. At high temperatures (T > 260 °C), the mean free path becomes shorter than the pore diameter, enabling greater molecular diffusion. However, the increase in k value far exceeds the D_k value, ultimately leading to the emergence of a diffusion-controlled state. To further elucidate gas sensing behaviors, the authors developed a hollow sphere (HS) model to describe the interplay between SCR and gas diffusion [Figure 2D]. At low temperatures (T < 150 °C), the oxygen species adsorbed on the SnO₂ surface are predominantly less reactive O₂^-, rendering SCR inactive and thus limiting the gas sensing reactions. A portion of the target gas undergoes partial oxidation on the outer surface, while the residual gas diffuses into the pore interior to perform chemical reactions, including partial oxidation, complete oxidation, and ionization of minor oxidation products. In this case, gas diffusion progressively becomes the rate-determining step. Although the gas diffusion control mechanism requires further quantitative analysis and optimization of multiple factors, it has been extensively applied to evaluate the effect of morphology on the sensing performance of MOS materials. A detailed discussion on the construction and regulation of mesoporous architectures, including pore size, morphology and hierarchical design, will be presented in Section Constructing mesoporous metal oxides.

Sensitivity

Sensitivity (S) is one of the key indices for evaluating sensor performance, reflecting the response capability of the sensing material towards the target gas. For chemiresistive gas sensors, the output is the response value (R), defined as the ratio of the sensor’s resistance under the target gas atmosphere to its resistance in air. Taking reducing gases as an example, for n-type semiconductors, R = R_a/R_g; whereas for p-type semiconductors, R = R_g/R_a, where R_a denotes the sensor’s resistance in air, and R_g represents the sensor’s resistance in the target gas. The situation is reversed for oxidizing gases. The distinction between sensitivity and response is not always clear-cut, as both describe the extent of resistance variation. Sensitivity (S) can be expressed in alternative ways:

ΔR denotes the difference between R_a and R_g.

Selectivity

Selectivity refers to a gas sensor’s ability to accurately identify specific gases in the presence of mixed gases. Traditionally, selectivity is defined as the ratio of the sensor’s response towards the target gas to its response towards interfering gases^[45]. This implies that the sensor should exhibit a distinctive response solely to the target gas, without interference from other gases. In practical applications, gas sensors often operate within complex working environments, where numerous interfering gas molecules are present during detection. Achieving selective detection of specific gases in mixed atmospheres is paramount, directly influencing the reliability and accuracy of sensors in real-world scenarios. Enhancing selectivity thus represents a significant research focus in gas sensor design and deployment.

Response/recovery time

Response and recovery times are crucial metrics for evaluating sensor performance, indicating how quickly the sensor’s signal rises upon exposure to the target gas and returns to baseline once the gas is removed. Response time (t_res) and recovery time (t_rev) are defined as the times required for a gas sensor to achieve a 90% change in resistance following gas introduction or removal, respectively^[46]. In practical applications, shorter response and recovery times indicate that sensors can respond rapidly to gas changes and return quickly to their original state, thereby ensuring the promptness and accuracy of monitoring. Therefore, optimizing response and recovery kinetics is crucial for achieving efficient gas detection, particularly in applications requiring timely responses, such as industrial safety monitoring. Generally speaking, sensing materials with higher porosity possess more gas diffusion pathways, allowing gas molecules to penetrate the material interior more readily, thereby facilitating the rapid response and recovery^[47,48].

Stability

Stability refers to the ability of a gas sensor to maintain a consistent and stable signal output after prolonged operation. This is influenced by multiple factors, including material composition, ambient humidity, and operating temperature. The stability of a sensor is one of its most crucial characteristics in practical applications, as it directly influences the sensor’s reliability and lifetime. During prolonged operation, poor sensor stability may result in fluctuating or erroneous output signals, thereby diminishing the reliability and usability. Consequently, when designing and selecting sensors, it is essential to thoroughly consider and enhance their stability to ensure long-term, consistent operation under varying environmental conditions and to deliver accurate and dependable gas detection results.

Reversibility

Reversibility denotes the ability of a gas sensor to restore its baseline state once the target gas concentration returns to normal levels. This property is essential for ensuring reliable and stable sensor operation over time. Inadequate reversibility can cause cumulative measurement errors and undermine detection accuracy, thereby affecting both operational continuity and precision. Therefore, reversibility serves as a key criterion for evaluating sensor performance and practical applicability.

Optimum operating temperature

The optimum operating temperature is defined as the temperature at which a sensor achieves its maximum response to a given gas concentration. At this point, both sensitivity and response dynamics are optimized, enabling accurate and rapid target gas detection. In practice, operating temperatures should be kept as low as possible, since elevated temperatures increase fabrication and energy costs and may induce material degradation or unstable sensing behaviors. Therefore, determining the optimum operating temperature requires a careful balance between performance, energy efficiency, material stability, and practical application demands.

Detection limit

The limit of detection (LOD) is the lowest concentration of a target gas that can be reliably distinguished from the blank under defined conditions, commonly using an S/N (Singal/Noise) ≈ 3 or 3σ criterion. In practice, the LOD is obtained from the calibration curve as LOD = 3σ/m, where m is the slope and σ is the standard deviation of the blank (or replicate low-level measurements). A lower LOD indicates superior detectability at trace levels and is critical for high-sensitivity applications. For example, food spoilage releases marker gases such as hydrogen sulfide (H₂S) at parts-per-billion (ppb) concentrations^[49], whereas breath biomarkers for early disease screening (e.g., acetone) often occur at parts-per-million (ppm) to ppb levels^[50]. Accordingly, achieving low detection limits has substantial practical significance.

Composition

Different MOS materials exhibit distinct intrinsic characteristics, including surface properties, carrier concentration, defect types, terminal groups, bandgap, and electronic structure. These factors influence gas-solid interactions in different ways, thereby governing gas sensing performance^[51,52]. Rational design and selection of appropriate material systems are therefore essential for achieving high sensitivity and selectivity. In gas sensing, n-type MOS (e.g., WO₃, In₂O₃, SnO₂, ZnO) are generally preferred over p-type counterparts^[53]. Typically, n-type MOS show more pronounced resistance variations upon gas adsorption, which is attributed to their higher density of grain boundaries^[54]. In contrast, p-type MOS usually display weaker resistance changes because parallel conduction pathways at interparticle interfaces diminish the overall modulation effect. Moreover, the sensitivity of p-type MOS sensors has been shown to scale only with the square root of that of n-type MOS sensors^[55]. Despite this limitation, many p-type MOS materials, such as CuO, NiO, and Co₃O₄, remain widely used owing to their excellent catalytic activity^[54-56-60]. For materials with wide band gaps, inter-band electron hopping becomes difficult, resulting in smaller resistance changes upon gas adsorption. Conversely, materials with narrow band gaps possess intrinsically high carrier concentrations, which reduce the relative modulation induced by adsorption, often leading to negligible signal changes. Finally, regardless of whether n-type or p-type MOS are employed, single-component materials often suffer from limited sensitivity and severe cross-sensitivity. The synergistic combination of multiple metal oxide components may also induce interfacial interactions, whose detailed roles, including heterojunction formation and interfacial charge modulation, will be systematically discussed in Section Heterojunction engineering.

Microstructure

The electrical and chemical properties of sensing materials can be tuned by tailoring their microstructural morphology, such as into nanowires, nanotubes, nanoparticles (NPs), or porous frameworks^[61]. Modifying these structures enables faster and more concentrated interactions between the sensing material and target gases, thereby markedly enhancing sensor sensitivity. In general, materials with larger surface areas exhibit higher response levels due to the increased number of active sites available for gas-solid interactions. Moreover, porous structures with interconnected channels facilitate gas diffusion within the sensing layer, improving contact between the target gas and the material, thereby enhancing reaction efficiency^[62,63]. Consequently, nanostructured and porous materials are widely employed to increase surface area and improve sensor performance.

Grain size

Grain size is a critical factor influencing variation in resistance and overall gas sensing performance. Numerous studies have demonstrated its impact on the sensing behavior of metal oxides^[64-67]. As shown in Figure 3, sensing materials consist of partially sintered grains connected by necks, which further assemble into larger aggregates through grain boundaries. At the grain surface, adsorbed oxygen molecules withdraw electrons from the conduction band and trap them in ionic form, inducing band bending and generating electron-depleted space-charge layers. When the grain size becomes comparable to or smaller than twice the thickness of the space-charge layer, sensor sensitivity increases markedly^[68]. Xu et al. explained this behavior using a semi-quantitative model^[69]. Depending on the relationship between grain size (D) and the width (L) of the space-charge layer induced by chemisorbed ions, three regimes can be distinguished. (i) D >> (indicates far exceeding) 2L. Conductivity is dominated by internal charge carriers and follows an exponential dependence on barrier height. In this case, the effect of charges generated by surface reactions is relatively minor; (ii) D ≥ 2L. Space-charge layers at grain necks form constricted conduction channels within aggregates. Conductivity is determined by both interparticle barriers and the channel cross-sectional area, making it highly sensitive to variations in surface charge; (iii) D < 2L. The space-charge layer extends throughout the entire grain, depleting most mobile carriers. Conductivity is then governed primarily by intergranular transport, and even small amounts of surface charge can induce large changes in overall conductivity. Therefore, when the grain size approaches the critical dimension, the material exhibits exceptionally high sensitivity to ambient gas molecules.

Figure 3. Schematic illustration of the effect of grain size on MOS sensor sensitivity. (A) D >> 2L; (B) D ≥ 2L; (C) D < 2L. This figure is quoted with permission from Sun et al.^[68]. MOS: Metal oxide semiconductor.

Hard-template method

The hard-template method, also known as nano-casting, typically employs pre-synthesized mesoporous materials such as mesoporous silica or carbon as sacrificial scaffolds. The inorganic precursor is impregnated into the template channels to fill the mesopores, followed by thermal treatment to convert it into the corresponding crystalline metal oxide. Subsequent removal of the template yields the desired mesoporous metal oxide framework [Figure 4A]^[70]. This method offers several advantages. First, it avoids the use of structure-directing agents and bypasses the complex hydrolysis-condensation of inorganic precursors, making it highly versatile for synthesizing various mesoporous materials. Second, mesoporous silica and carbon templates are chemically and thermally stable, and their well-defined channels can be readily infiltrated by appropriate precursors. As a result, the hard-template strategy has been successfully applied to fabricate diverse mesoporous materials, including metal oxides^[79], metal sulfides^[80], metal nitrides^[81], and metal carbides^[82]. For example, ordered mesoporous WO₃ (mWO₃) structures have been obtained using KIT-6 as a template [Figure 4B]^[79]. Moreover, Deng et al.^[70] synthesized a series of ordered mesoporous transition metal oxides, such as Co₃O₄, NiO, Fe₂O₃, Cr₂O₃, and Mn₃O₄, by employing ordered mesoporous silicas [ Mobil Composition Of Matter N. 41 (MCM-41), Korea Advanced Institute of Science and Technology-6 (KIT-6), and Santa Barbara Amorphous-15 (SBA-15)] as hard templates.

Figure 4. (A) Schematic illustration for the casting process of a key with a mold (A₁) and nanocasting using ordered mesoporous materials as the hard template (A₂). Figure 4A is quoted with permission from Deng et al.^[70]; (B) TEM images of mesoporous WO₃ synthesized using KIT-6 as the template (B₁, B₂); HRTEM image of the ring-like structure, with the inset showing the corresponding FFT pattern (B₃); uncoupled framework model (B₄). Figure 4B is quoted with permission from Rossinyol et al.^[79]; (C) Schematic illustration of the EIAA strategy. Figure 4C is quoted with permission from Jianget al.^[24]; (D) Schematic illustration of the one-pot direct synthesis of PB-b-PDMAEMA/H₃[PMo₁₂O₄₀] nanocomposite films. Figure 4D is quoted with permission from Lunkenbein et al.^[85]; (E) Schematic illustration of the citrate ligand-assisted EISA method for synthesizing crystalline mZnO. Figure 4E is quoted with permission from Zhou et al.^[25]; (F) Schematic illustration of the synthesis routes for mesoporous metal oxides via metal carbonate intermediates. Figure 4F is quoted with permission from Eckhardt et al.^[86]; (G) TEM images of multicomponent mesoporous metal oxides prepared with acetate chelation assistance. Figure 4G is quoted with permission from Fan et al.^[87]. THF: Tetrahydrofuran; TEM: transmission electron microscopy; HRTEM: high-resolution transmission electron microscopy; FFT: fast Fourier transform; EIAA: evaporation-induced aggregate assembly; PDMAEMA: poly(2-(dimethylamino)ethyl methacrylate); H₃[PMo₁₂O₄₀]: phosphomolybdic acid; mZnO: mesoporous zinc oxide.

However, the rigid structural constraints of templates such as mesoporous silica and carbon make it difficult to flexibly tune the pore characteristics of the resulting materials, including their dimensions, types, window sizes, and connectivity. In particular, template removal typically requires hazardous chemicals such as hydrofluoric acid or hot potassium or sodium hydroxide solutions, which pose environmental risks and health hazards to operators. Therefore, the hard-template method is associated with high costs, complex procedures, and poor environmental compatibility, limiting its suitability for large-scale production.

Soft-template method

The soft-template method is considered one of the most effective and versatile strategies for fabricating ordered mesoporous materials through controlled co-assembly. Classical soft-templating is primarily realized via evaporation-induced self-assembly (EISA) or evaporation-induced aggregate assembly (EIAA) strategies [Figure 4C]^[24]. In these processes, weak interactions between inorganic precursors and organic surfactants (e.g., block copolymers) are crucial. These interactions cooperatively drive complex hydrolysis-condensation through multiple forces, including hydrogen bonding, electrostatic interactions, and covalent bonding, leading to the formation of uniform organic-inorganic micelles. As solvents such as tetrahydrofuran, acetone, or methanol evaporate, the mixed micelles pack densely to generate ordered mesoporous frameworks. Despite its advantages, synthesizing ordered mesoporous metal oxides remains challenging due to the intrinsically weak inorganic-surfactant interactions, the limited control over hydrolysis-condensation, and the structural instability of porous networks during high-temperature treatment^[83]. To overcome these limitations, numerous strategies have been developed to strengthen inorganic-organic interactions and regulate precursor hydrolysis-condensation behaviors.

Specifically, the interactions between inorganic species and surfactants can be strengthened by selecting appropriate precursors and organic polymer templates. For example, Zhang et al. employed a laboratory-synthesized amphiphilic block copolymer, polystyrene-b-poly(4-vinylpyridine) (PS-b-P4VP), as the structure-directing agent and tetrabutyl titanate (TBOT) as the titanium source to prepare nitrogen (N)-doped ordered mesoporous TiO₂^[84]. Strong chelation between the pyridine nitrogen and titanium enabled the formation of an ordered mesoporous TiO₂ framework with high crystallinity, without the need for additional additives during synthesis. Similarly, Lunkenbein et al. used the diblock copolymer poly(butadiene-block-2-(dimethylamino)ethyl methacrylate) (PB-b-PDMAEMA) and phosphomolybdic acid (H₃[PMo₁₂O₄₀]) to fabricate mesoporous films, leveraging strong electrostatic interactions between protonated poly(2-(dimethylamino)ethyl methacrylate) (PDMAEMA) segments and PMo₁₂O₄₀^3- anions [Figure 4D]^[85].

Small-molecule ligands can act as “bridges” that strengthen coordination between inorganic species and surfactants, thereby promoting organic-inorganic co-assembly. For example, Zhou et al. used citric acid to assist the formation of mesoporous ZnO by enhancing the interaction between Zn²⁺ ions and poly(ethylene oxide) (PEO) segments [Figure 4E]^[25]. Citric acid induces partial hydrolysis of zinc nitrate to form zinc citrate, which then binds to ethylene oxide segments through hydrogen bonding. Similarly, Eckhardt et al. employed citric acid as a chelating agent to form stable metal-citrate complexes, which subsequently interact with block copolymers to generate ordered mesophases, yielding ZnO and Co₃O₄ films with ordered mesopores [Figure 4F]^[86]. Fan et al. further showed that controlled coordination of metal ions with acetic acid produces nanoscale particles with uniform condensation kinetics, where acetate ligands bridge amphiphilic copolymers and inorganic species^[87]. Replacing hydrolysable alkoxy groups with acetate lowers the cross-linking density within the gel network, enabling the synthesis of multicomponent mesoporous oxides [Figure 4G]. Overall, tailored ligands serve as effective molecular linkers, guiding the formation of well-organized mesostructures.

The rapid hydrolysis-condensation of inorganic precursors, such as metal chlorides and alkoxides, often induces phase separation and agglomeration, resulting in disordered structures. To overcome these issues, various modifiers, including hydrochloric acid, nitric acid, and acetylacetone, have been introduced to regulate hydrolysis kinetics. For example, Li et al. used hydrochloric acid as a modifier and poly(ethylene oxide)-b-polystyrene (PEO-b-PS) as a template to fabricate ordered mWO₃ with a highly crystalline framework^[47]. HCl suppresses the hydrolysis of WCl₆ while protonating PEO segments, thereby strengthening hydrogen bonding between hydrophilic tungsten species and the template. Similarly, Wei et al. synthesized mesoporous Al₂O₃ using aluminum acetylacetonate as the precursor and nitric acid as a modifier^[88]. Nitric acid slows hydrolysis, whereas acetylacetonate ligands form chelated complexes that further stabilize Al species. Chelating agents also inhibit aggregation during hydrolysis. Xiao et al. employed acetylacetone to retard SnCl₄ hydrolysis, forming acetylacetonate-stabilized tin hydroxide nanoclusters (NCs) with controllable sizes suitable for subsequent co-assembly with block copolymers^[89]. An alternative strategy involves the direct co-assembly of preformed nanocrystals with amphiphilic block copolymers, thereby bypassing complications associated with precursor hydrolysis^[78]. Corma et al. demonstrated this approach by co-assembling monodisperse CeO₂ NPs with PEO₂₀-PPO₇₀-PEO₂₀ (P123) to produce mesostructures exhibiting excellent ordered hexagonal symmetry and exceptional thermal stability up to 700 °C^[90]. This strategy has been extended to other oxides such as In₂O₃^[91], TiO₂^[91], ZnO^[92], CuO^[92], Mn₃O₄^[93], ZrO₂^[94], and hybrid oxides including ITO^[91] and MnFe₂O₄^[93].

Commercial Pluronic surfactants composed of non-ionic PEO and poly(propylene oxide) (PPO), such as P123^[90] and PEO₁₀₂-PPO₆₅-PEO₁₀₂ (F127)^[95], have been widely used in the synthesis of mesoporous materials. However, their low thermal stability makes it difficult to obtain highly crystalline frameworks. High-temperature annealing often leads to partially crystalline porous structures or even the collapse of metal oxide frameworks. To address this limitation, amphiphilic block copolymers with higher thermal stability and molecular weight, such as PEO-b-PS^[47], PS-b-P4VP^[84], polystyrene-block-poly(2-vinylpyridine) (PS-b-P2VP)^[96], and poly(isoprene)-block-poly(ethylene oxide) (PI-b-PEO)^[97], have been employed to prepare mesoporous metal oxides with large pore diameters and high crystallinity. Lee et al. first reported a combined soft-hard template co-assembly (CASH) method using PI-b-PEO as the structure-directing agent^[97]. In this process, samples were initially calcined under an inert atmosphere. The PEO segments decomposed readily upon heating, whereas the thermally stable polyisoprene (PI) segments [each containing two s orbital and two p orbitals hybridized (sp²)-hybridized carbons] carbonized to form rigid amorphous carbon, which acted as a scaffold to support the intermediate oxide walls. Subsequent calcination in air removed the residual carbon, yielding highly crystalline mesoporous metal oxides. More recently, our group has further advanced the CASH method by employing PEO-b-PS as templates. A series of mesoporous metal oxides, including WO₃^[47], TiO₂^[98], SnO₂^[89], ZnO^[92], CuO^[92], and Al₂O₃^[88], have been successfully synthesized through gas-liquid interfacial assembly.

Furthermore, incorporating soluble phenolic formaldehyde resin (resol) and silicate species into the framework has been shown to effectively prevent structural collapse during high-temperature processing. Inspired by this concept, Li et al. developed a pore-engineering strategy using hydrophilic resol as a sacrificial carbon source to synthesize ordered mWO₃ with well-connected bimodal pores and crystalline pore walls^[99]. In this system, resol preferentially interacts with PEO segments and acts as a binder that links tungsten species with the PEO-b-PS template. During thermal treatment, both the carbonized resol and the copolymer form a rigid matrix that stabilizes the pore structure, yielding highly crystalline hierarchical mWO₃. Zhou et al. reported the controlled synthesis of silica-filled mesoporous ZnO^[100]. Citric acid was employed as a chelating agent to strengthen the weak interactions between the zinc precursor and the PEO-b-PS template. Simultaneously, tetraethyl orthosilicate (TEOS)-derived silicate oligomers interacted with citric acid through hydrogen bonding, enabling co-assembly with the zinc precursor and providing additional structural support during high-temperature processing.

These sophisticated assembly strategies, including ligand-assisted soft-templating, substitution of inorganic salts with preformed nanocrystals or NCs, sp² carbon support, and amorphous component (phenolic resin or silicate)-assisted assembly, enable more controllable construction of mesoporous MOS materials with tunable pore architectures and precisely engineered surface properties, paving the way for high-performance gas sensor fabrication.

Self-template method

MOFs represent a distinctive class of porous organic-inorganic hybrid materials constructed through the coordination assembly of metal ions or clusters with organic ligands. Owing to their tunable porosity, high surface areas, compositional diversity, and relatively straightforward synthesis, MOFs have attracted considerable attention in recent years. Representative families include the zeolitic imidazolate frameworks (ZIFs)^[101], the Materials of the Lavoisier Institute (MIL) series^[102], and the University of Oslo (UIO) series^[103]. These materials have been widely applied in gas separation, sensing, catalysis, and energy storage and conversion^[104]. Upon high-temperature calcination in air, the metal centers within MOFs are converted into crystalline metal oxides. At the same time, the organic ligands decompose into gaseous products such as CO₂ and NO₂, thereby forming mesoporous structures. Since this self-templating pyrolysis strategy requires no additional templating agents, it has emerged as an attractive alternative route for synthesizing mesoporous MOS materials.

Mesoporous MOS derived from MOFs largely retain the structural features of their parent frameworks, and a variety of mesoporous MOS sensing materials with tunable morphologies have been obtained by precisely regulating MOF precursors. Wang et al. synthesized copper-based MOF [Hong Kong University of Science and Technology-1 (HKUST-1)] crystals with distinct polyhedral shapes by adjusting the amount of lauric acid as a growth modulator, and subsequent thermal treatment produced Cu₂O/CuO sensing materials with octahedral, truncated octahedral, and cubic morphologies^[105]. Among these, the octahedral structure with the largest specific surface area exhibited the best ethanol sensing performance. Jang et al. constructed a chrysanthemum-like layered ZnO/Co₃O₄ composite by in situ coupling rod-like ZIF-67 with sheet-like ZIF-8, followed by calcination^[106]. This hierarchical structure, featuring abundant p-n heterointerfaces and high porosity, significantly enhanced acetone sensing, achieving an R_a/R_g of 29.0 to 5 ppm acetone, far exceeding that of pure Co₃O₄ rods (R_a/R_g = 1.05 at 5 ppm acetone). Li et al. further utilized MOFs as self-templates to synthesize uniform hollow La₂O₃/In₂O₃ nanotubes, which exhibited excellent triethylamine (TEA) sensing performance with a high R_a/R_g value of 458.13 to 50 ppm at 120 °C^[107]. Such excellent performance can be attributed to the synergistic modulation of the electronic structure and the abundance of acidic sites. Using bimetallic MOFs as precursors, Qin et al.^[108] prepared a series of ternary Co-M-O NSs (M = Cu, Mn, Ni, Zn) with ultrahigh surface areas and homogeneous compositions [Figure 5A], which displayed excellent CO sensing performance, including fast response-recovery rates, high stability, and good selectivity. Koo et al. developed PdO-ZnO/ZnCo₂O₄ HSs by depositing CoZn bimetallic ZIF onto polystyrene spheres, encapsulating Pd NPs within the cavities, and subsequently performing calcination^[109]. The resulting p-ZnCo₂O₄/n-ZnO heterostructure facilitated electron sensitization, thereby endowing the HSs with outstanding acetone-sensing performance. When ternary phases fail to form or stoichiometry is suboptimal, bimetallic oxide composites can still be obtained, as exemplified by In₂O₃/ZnO porous hollow nanocages derived from InZn-MOF^[110], NiO/CuO hydrangea-like composites from NiCu-BTC^[111], and NiO/NiCo₂O₄ multilayer HSs from NiCo-BTC^[112]. Moreover, noble metal NPs can be uniformly incorporated into oxide matrices by calcining catalyst-encapsulated MOFs (M@MOFs). For example, Koo et al. used ZIF-8 to support Pd NPs (Pd@ZIF-8), which were subsequently embedded into electrospun fibers and thermally treated to yield Pd@ZnO-WO₃ nanofibers (NFs) with excellent toluene sensing properties^[113]. In a follow-up study, the same group synthesized PdO-Co₃O₄ hollow nanocages by calcining Pd@ZIF-67^[114], which showed high acetone responses due to the uniform dispersion of PdO catalysts and improved gas accessibility within the hollow structures.

Figure 5. (A) Schematic illustration of ternary metal oxide nanosheet synthesis (A₁), and TEM images of Co-Cu-O (A₂), Co-Mn-O (A₃), Co-Ni-O (A₄), and Co-Zn-O (A₅). Figure B is quoted with permission from Qin et al.^[108]; (B) Schematic illustration of the synthesis of ZnO and Al₂O₃ (B₁); SEM image (B₂) and TEM image (B₃) for Zn-tannic acid coordination polymer; SEM image (B₄) and TEM image (B₅) for Al-tannic acid coordination polymer. Figure 5B is quoted with permission from Wang et al.^[116]. ZIF: Zeolitic imidazolate framework; MOF: metal-organic framework; NS: nanosheet; TEM: transmission electron microscopy; SEM: scanning electron microscope.

In addition to MOFs, plant-derived polyphenols, as a naturally abundant, nontoxic, and environmentally friendly biomass, have recently emerged as powerful precursors for constructing functional mesoporous nanomaterials^[115]. Owing to their strong chelating ability toward metal ions, plant polyphenols readily coordinate with various metal species through metal-catechol complexation, forming stable metal-phenolic coordination polymer networks. Upon subsequent calcination, the polyphenol scaffold decomposes into CO₂ and H₂O, while the coordinated metal centers undergo oxidation and crystallization, producing mesoporous metal oxides. Because the limited amount of metal species cannot fully inherit the morphology of the parent metal-phenolic framework, extensive mesopores and internal cavities are generated during pyrolysis.

Wang et al. developed a general metal-polyphenol self-templating strategy for synthesizing mesoporous metal oxide microspheres with diverse compositions (Al₂O₃, ZnO, Fe₂O₃, Co₃O₄, and CuO) and highly crystalline frameworks [Figure 5B]^[116]. Owing to the universal chelation capability of polyphenols toward a wide range of metal ions, this method enables the synthesis of binary, ternary, and even high-entropy mesoporous metal oxides. Using tannic acid as a universal ligand, Wang et al. further demonstrated the fabrication of binary (Fe-Co, Ni-Co, Ni-Zn oxides), ternary (Ni-Co-Zn and Ni-Co-Mn oxides), and quinary (Ni-Co-Fe-Cu-Zn oxides) mesoporous nanospheres^[117]. Depending on the intrinsic crystallization temperature and thermal evolution of different metal species, various architectures, such as solid^[118], hollow^[117], and multi-shell^[119] mesoporous structures, can be obtained. These features arise from the nonuniform shrinkage and structural reorganization of metal-phenolic precursors during calcination. Moreover, plant polyphenols exhibit rich structural diversity and tunable chemical functionality. By employing different polyphenols as organic ligands, the morphology of metal-phenolic coordination networks can be flexibly tailored into nanospheres, NFs, NSs, and polyhedra, which are subsequently inherited by the derived mesoporous metal oxides.

Mesoporous semiconductor oxides derived from metal-phenolic self-templating possess high surface area, abundant porosity, compositional tunability, and diverse morphologies, making them highly promising for gas sensing applications^[120]. For instance, Feng et al. used polyphenols as molecular glues to co-assemble glutathione-protected Au NCs with Sn species, forming hybrid metal-phenolic nanospheres^[121]. After calcination, the Au NCs were unexpectedly converted into atomically dispersed Au single atoms embedded in the mesoporous SnO₂ framework. The resulting sensor exhibited an exceptionally high response (587.3) to 2 ppm 3-hydroxy-2-butanone (3H-2B) at 50 °C, an enhancement of 183.5-fold compared with pristine mesoporous SnO₂. Additionally, the optimal operating temperature was effectively reduced from 100 °C to 50 °C. Similarly, Li et al. employed the same self-templating process to synthesize Au/Pd dual-atom-sensitized mesoporous SnO₂ for H₂ detection, achieving an ultralow detection limit (70 ppb), fast response (100 s), and strong resistance to CO, NO, H₂S, and SO₂ interference^[122]. Importantly, polyphenols serve as versatile molecular glues that interact with diverse substrates through hydrophobic interactions, hydrogen bonding, electrostatic forces, π-π stacking, and covalent coupling^[123]. After thermal treatment, mesoporous metal oxide films can be deposited directly onto different substrates, offering opportunities for in situ solution-grown, structurally stable multicomponent MOS devices. Furthermore, as naturally abundant biomass molecules, polyphenols can be produced on a large scale, providing a promising foundation for scalable manufacturing of mesoporous metal oxide materials.

Thus, the self-templating strategy has opened new avenues for the rational design of mesoporous MOS materials. By tailoring MOF and metal-phenolic precursor structures and carefully controlling synthesis parameters, the size, morphology, composition, and electrical conductivity of mesoporous MOS can be precisely tuned. Materials synthesized through this approach exhibit unique physicochemical properties and hold great promise for high-performance gas sensing applications.

Overall, the construction of mesoporous architectures remains a fundamental and highly effective strategy for enhancing the gas sensing performance of MOS. Hard-template, soft-template, and self-templating approaches each offer distinctive advantages [Table 1]. The hard-template method enables precise structural replication and well-defined pore ordering, but typically involves multistep procedures and environmentally hazardous template-removal processes. The soft-template approach provides greater versatility and tunability, allowing fine control over pore size and channel configuration. However, it requires meticulous regulation of intermolecular interactions and often faces challenges in maintaining structural integrity during high-temperature calcination. Self-templating routes, including MOF- and metal-polyphenol-derived strategies, feature simplified synthesis and rich compositional adaptability, although the final morphology inherently depends on the structure and chemistry of the precursor framework. In practical applications, the choice among these methods must balance the target pore architecture, desired crystallinity, processing complexity, and cost considerations.

Rare-earth element doping

Rare-earth elements (REs, including La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, and Yb) possess a unique 4f-electron configuration. Owing to their exceptional electrical conductivity, luminescence, and catalytic properties, they have been widely employed as effective dopants in catalysis and sensing applications^[124-126]. Rare-earth oxides typically exhibit high oxygen-ion mobility, favorable catalytic activity, and strong surface alkalinity, which enhance gas sensing performance, particularly in terms of response and selectivity. In addition, rare-earth doping can reduce grain size and increase oxygen vacancy (O_v) concentrations, thereby further improving the sensing characteristics of MOS. Extensive studies have demonstrated that rare-earth elements play a pivotal role in optimizing the gas sensing properties of MOS.

Liu et al. synthesized Ce-doped ordered mWO₃ via a straightforward in situ co-assembly approach. Ce doping effect can modulate the coordination environment of W atoms, significantly enhancing O_v and forming W^δ+-O_v sites^[127]. Therefore, the resulting Ce-doped WO₃ exhibits outstanding H₂S sensing performance at low operating temperatures (150 °C), with a high response value of 381 (for 50 ppm), rapid response dynamics (6 s), excellent selectivity, moisture resistance, and good long-term stability [Figure 6A]. Zhao et al. employed an electrospinning strategy to synthesize a series of one-dimensional In₂O₃ NFs doped with various rare-earth elements (RE-In₂O₃, RE = Y, La, Nd, Ho, and Tm)^[128]. All RE-In₂O₃ materials exhibited enhanced formaldehyde responses compared to pristine In₂O₃. Among these, Y-In₂O₃ demonstrated the highest responsivity and excellent selectivity, which was attributed to the elevated Fermi level and increased surface alkalinity of In₂O₃, resulting in enhanced oxygen chemisorption and formaldehyde adsorption. Bai et al. prepared pristine In₂O₃ and RE (Ce, Tm, Eu, Er, and Tb)-doped In₂O₃ nanotubes via electrospinning. Rare-earth doping markedly improved ethanol sensing, although the extent of enhancement varied among different dopants. Mechanistic analyses indicated that doping modifies the types and distribution of adsorbed oxygen species on the material surface, thereby boosting gas sensing activity^[129]. Wang et al. synthesized pristine SnO₂ and Tb-doped SnO₂ nanotubes with different doping concentrations using uniaxial electrospinning^[130]. Tb doping induced pronounced morphological changes, including increased surface roughness, reduced nanotube diameter, and smaller SnO₂ grain size. Notably, the 7 mol% Tb-SnO₂ nanotubes exhibited the highest response to ethanol, which was attributed to the synergistic effects of reduced grain size, defect generation, structural disorder, enhanced surface alkalinity, and increased roughness introduced by doping. Li et al. fabricated pure MoO₃ and Ce-doped MoO₃ nanoribbons via a one-step hydrothermal method^[131]. Ce doping not only increased the response toward trimethylamine (TMA) but also reduced the optimal operating temperature. The authors proposed that substitution of Mo sites by Ce atoms promotes O_v formation, which facilitates oxygen adsorption and improves sensing performance. Furthermore, Lee et al. have reported several humidity-insensitive sensing materials, including CeO₂-loaded In₂O₃ HSs^[132], Tb-doped SnO₂ yolk-shell microspheres^[133], and Pr-doped In₂O₃ macroporous microspheres^[134]. They attributed this humidity resistance behavior to the +3/+4 redox transitions of multivalent rare-earth elements (Ce, Pr, Tb), which promote the removal of surface hydroxyl groups, the regenerative adsorption of oxygen, and electron consumption [Figure 6B and C]. These processes enhance sensor stability and gas sensing performance by establishing a water-poisoning-resistant reaction pathway.

Figure 6. (A) Sensing responses of the Ce-X/mWO₃ sensor to 50 ppm H₂S at various temperatures (A₁); Sensing mechanism of Ce-X/mWO₃-based gas sensors for H₂S detection (A₂). Energy band structure and electron-transfer process of Ce-X/mWO₃ sensitive material exposure in air and H₂S-air mixture (A₃). Figure 6A is quoted with permission from Liu et al.^[127];( B) Gas responses of pure In₂O₃ (B₁) and 11.7 Ce-In₂O₃ (B₂) hollow spheres exposed to 20 ppm acetone (Ace), CO (C), ammonia (A), H₂ (H), and toluene (T) at 450 °C in dry and humidity (r.h. 80%); Polar plots of gas responses of pure In₂O₃ (B₃)and 11.7 Ce-In₂O₃ (B₄) hollow spheres exposed to various reducing gases in dry (red) and r.h. 80% (blue). C: Self-refreshing of the In₂O₃ sensing surface by the CeO₂ nanoclusters. Water vapor inflow (C₁), chemisorption (C₂), desorption (C₃), and oxygen ion regeneration (C₄). Figure 6B and C is quoted with permission from Yoon et al.^[132]. ppm: Parts-per-million.

Transition metal element doping

Transition metals constitute another important class of dopants for enhancing the gas sensing performance of MOS. Owing to their comparable atomic radii to those of common MOS hosts (e.g., W, In, Sn, Zn), many transition metals can be readily incorporated into the lattice through substitutional doping. Such doping not only modulates lattice parameters and oxygen adsorption but also alters defect and carrier concentrations, thereby regulating the overall sensing behavior. Previous studies have demonstrated the effectiveness of various transition metal-based dopants, including V^[135], Cr^[136,137], Mn^[138], Fe^[139], Co^[140,141], Ni^[142,143], Cu^[144], and Mo^[145,146].

Chen et al. synthesized Co-doped SnO₂ sensing materials with a 3D anti-opalescent structure^[141]. Among these, the Co-SnO₂ sensor with a Co/Sn atomic ratio of 1:24 exhibited the best performance, showing ultrahigh sensitivity, fast response, and excellent selectivity toward low concentrations of formaldehyde (200 °C) and acetone (225 °C), respectively. Mechanistic investigations revealed that cobalt doping narrows the bandgap, elevates the Fermi level, and introduces vacancy defects through the reactions

where, $$ {Co}_{Sn}^{\prime \prime} $$ reperents Co²⁺ on Sn site, $$ {Co}_{Sn}^{\prime} $$ means Co³⁺ on Sn site, $$ V_{O}^{\cdot \cdot} $$ is oxygen vacancy with +2 charge, $$ V_{O}^{\cdot} $$ is oxygen vacancy with +1 charge, $$ O_{O}^{X} $$ is lattice oxygen, and increases the amount of surface chemisorbed oxygen, collectively enhancing the sensing performance.

The electronic sensitization effect induced by transition metal doping can also occur in p-type MOS. Li et al. synthesized pure NiO and Cr-doped NiO NPs via a hydrothermal method and systematically investigated their gas sensing properties^[137]. The 25 at% Cr-NiO sensor exhibited high sensitivity (R_g/R_a = 10.6 vs. 100 ppb), excellent selectivity, and good stability for benzyl mercaptan detection. As a p-type semiconductor, NiO conducts through holes. Substitutional Cr³⁺ doping at Ni²⁺ sites follows a charge compensation mechanism, as given below:

where $$ Cr_{N i}^{\cdot} $$ represents the occupation of Cr³⁺ in Ni²⁺ site, $$\quad O_{O}^{X}$$ indicates the lattice oxygen, $$\quad V_{N i}^{\prime \prime}$$ is the double ionized Ni vacancy, $$\quad N i_{N i}$$ signifies the oxidized Ni species, and $$\quad O_{i}^{\prime \prime}$$ is the negatively charged oxygen at interstitial (or surface) site. Increasing the Cr³⁺ content leads to hole consumption, resulting in lower hole concentration and higher baseline resistance in air, which is beneficial for improving gas response.

Simultaneously, oxygen molecules oxidize Ni²⁺ to Ni³⁺, generating abundant reactive O^- species that amplify carrier concentration variations (Δp). Moreover, ionic species such as Ni²⁺ act as oxygen adsorption sites, forming reactive oxygen species that catalyze surface redox reactions, thereby further increasing Δp. From the perspective of electronic sensitization, the t_2g orbitals of Ni overlap with p orbitals of oxygen, causing Pauli repulsion and facilitating partial electron transfer between metal centers and lattice oxygen. Due to the O_h symmetry of the octahedral MO₆ unit, Ni²⁺ and Cr³⁺ have valence configurations of (t_2g)⁶(e_g)² and (t_2g)³(e_g)⁰, respectively. Cr³⁺, with unfilled t_2g orbitals, acts as an electron acceptor, whereas Ni²⁺, with a fully occupied (t_2g)⁶ configuration, induces e^--e^- repulsion with the oxygen p orbitals, leading to electron transfer from Ni²⁺ to Cr³⁺. This process promotes the formation of Ni³⁺ species and enhances their electron affinity. The authors proposed that surface high-valence Ni plays a pivotal role in gas redox reactions, while doped Cr regulates carrier concentration and Cr-O-Ni charge transfer, synergistically improving gas sensing performance. Similar phenomena have been reported for Fe doping. Wang et al. synthesized a unique nest-like Fe-NiO structure, whose sensor exhibited a rapid response rate (26 s), short recovery time (5 s), and excellent selectivity toward TEA^[139]. At 260 °C, its response to 50 ppm TEA reached 63, a 21-fold enhancement compared with pristine NiO. The improved performance was attributed to the distinctive nest-like architecture and the incorporation of Fe ions into the NiO nanocrystals.

Non-metal doping

The incorporation of non-metallic impurity atoms (e.g., C, N, Si, and S) into MOS frameworks has also been demonstrated to significantly enhance gas sensing performance. Compared with metallic dopants, non-metallic elements can induce the formation of amorphous species, improve thermal stability, enhance selectivity and sensitivity, and lower operating temperatures. Therefore, the rational design of non-metal doping strategies is of great importance for optimizing the performance of MOS sensing materials.

Carbon, an essential non-metal widely used in catalysis and energy conversion, can act as an effective dopant to modulate the micro/nanostructure and physicochemical properties of sensing materials, including localized electron distribution, surface active sites, and electrical conductivity. Raghu et al. synthesized C-doped TiO₂ with abundant (101) crystal facets using urea as the carbon source via a mild pyrolysis process [Figure 7A]^[147]. The resulting material exhibited a markedly improved ethanol sensing response (34.8%) compared with pristine TiO₂ (8.5%). Both substitutional and interstitial carbon doping generated abundant active sites, O_v, and chemisorbed oxygen species, which accelerated interfacial catalytic reactions. Moreover, graphitic carbon improved electrical conductivity and reduced the bandgap, thereby lowering the ethanol sensing operating temperature from 250 °C to 150 °C. Similarly, Zhao et al. developed a facile template-free solvothermal route to synthesize C-doped TiO₂ NPs (~ 30 nm) without an additional carbon source^[148]. The C-doped TiO₂ exhibited outstanding alcohol sensing performance, with sensitivity increasing alongside carbon chain length. In particular, its response to n-pentanol was ~ 5.4 times higher than that of undoped TiO₂. This improvement was attributed to enhanced conductivity, accelerated electron transfer, and increased surface-adsorbed oxygen, which collectively lowered the operating temperature and boosted sensitivity.

Figure 7. (A) Time-dependent ethanol gas sensing performances of TiO₂ and C₄₀-TiO₂ sensor performances at 150 °C (A₁); Ethanol gas sensing mechanism on pristine TiO₂ and C₄₀-TiO₂. Figure 7A is quoted with permission from Raghu et al.^[147]; (B) molecular structure of N-doped In₂O₃ (B₁); SEM (B₂) and TEM (B₃) image of N-doped In₂O₃; The response transients of sensors to 10 ppm NO₂ (B₄); The linear fitting to the relationship between response and concentrations for In₂O₃ and N-doped In₂O₃ (B₅). Figure 7B is quoted with permission from Mo et al.^[150]; (C) Acetone sensing performance and mechanism of 3D cross-stacked Si-doped WO₃ nanowire arrays. Figure 7C is quoted with permission from Ren et al.^[153]; (D) Schematic illustration of the synthesis of the Ti₃C₂T_x MXene, sulfur-doped Ti₃C₂T_x MXene, and preparation of the MXene sensors (D₁); Dynamic response curve of sulfur-doped Ti₃C₂T_x sensors for toluene at room temperature (D₂). Figure 7D is quoted with permission from Shuvo et al.^[156]. VOC: Volatile organic compound; TEM: transmission electron microscopy; SEM: scanning electron microscope; ppm: parts-per-million.

Notably, carbon doping can also modify crystalline phases in WO₃, thereby improving gas sensing performance. Wang et al. synthesized three-dimensionally ordered macroporous/mesoporous C-doped WO₃ using polystyrene spheres as templates followed by post-heating^[149]. Carbon atoms diffused into interstitial sites of the WO₃ lattice, inducing the formation of ε-phase WO₃. Benefiting from this unique crystal structure, the C-doped WO₃ sensor exhibited high responsivity and selectivity toward acetone.

Beyond carbon doping, nitrogen incorporation into porous metal oxides can also markedly improve their gas sensing performance through distinct mechanisms. Mo et al. reported a novel N-doped indium oxide gas sensor, which shows a remarkable response to 10 ppm NO₂ at room temperature, with outstanding long-term stability (over 30 days) and rapid response/recovery times (5/16 s) [Figure 7B]^[150]. Abbasi et al. employed density functional theory (DFT) calculations to investigate the adsorption behavior of NO₂ molecules on pristine and N-doped anatase TiO₂ NPs^[151]. Their results showed that NO₂ adsorption on N-doped TiO₂ is energetically more favorable than on pristine TiO₂. Specifically, the nitrogen atom in NO₂ preferentially binds to O_v and doped nitrogen sites on the TiO₂ surface, while the oxygen atoms exhibit higher affinity for five-coordinated Ti atoms. Moreover, NO₂ molecules are more easily oxidized to NO₃^- due to Ti-O bond cleavage between the TiO₂ lattice and NO₂. These factors collectively result in superior sensing performance of N-doped TiO₂ compared with undoped NPs. Zhang et al. further reported a simple EIAA method for the in situ synthesis of N-doped ordered mesoporous TiO₂^[84]. The incorporation of nitrogen atoms into the TiO₂ lattice led to smaller grain sizes and increased O_v concentrations, both of which enhance gas sensing activity. The resulting ordered mesoporous N-TiO₂ sensor exhibited a response of 17.6 toward 250 ppm acetone, an 11.4-fold improvement compared with undoped TiO₂. These improvements were mainly attributed to the open pore structure, large specific surface area, and abundant O_v. Additionally, nitrogen doping introduces impurity energy levels that lower the Fermi level of TiO₂^[152], thereby facilitating electron transfer from adsorbed target molecules to N-TiO₂.

Silicon (Si)-doped metal oxides exhibit markedly enhanced gas sensing performance. Notably, ε-WO₃, a metastable crystalline phase with exceptional acetone selectivity, can be stabilized by incorporating Si atoms into the framework. A representative example is the versatile directional orthogonal assembly strategy reported by our group^[153]. In this method, PEO-b-PS is co-assembled with various Keggin-type heteropolyacids, followed by calcination-induced structural transformation. This approach enables the fabrication of heteroatom-doped MOS 3D multilayer cross-nanowire architectures featuring well-interconnected frameworks and uniform nanowire spacing. Si-doped WO₃ nanowires synthesized through this method exhibit the ε-WO₃ phase owing to the in situ introduction of Si from silicotungstic acid. Gas sensors based on these nanowires demonstrate excellent acetone sensing characteristics, including high sensitivity (LOD = 10.0 ppb), outstanding selectivity, fast response and recovery, and good stability. This superior performance is attributed to the combined effects of large surface area, abundant catalytic active sites, efficient electron transport along the nanowires, and the high electric dipole moment of ε-WO₃ [Figure 7C]. Besides, our group developed a dynamic co-assembly method to synthesize Si-doped WO₃ (Si-WO₃) with controllable nanostructures, including hollow hemispheres, NPs, and nanowires^[154]. The obtained Si-WO₃ hollow hemispheres were used to fabricate gas sensing devices, exhibiting excellent gas sensing performance toward acetone, thanks to the high specific surface areas, abundant O_v, and the Si-doping induced metastable ε-phase of WO₃. In addition, Si doping inhibits particle sintering and crystal growth during high-temperature processing, and the doping concentration strongly influences sensing behaviors. Güntner et al. prepared pure and Si-doped MoO_x NPs using flame-spray pyrolysis^[155]. At 1.5 wt.% SiO₂, the crystal size determined by X-ray diffraction (d_XRD) decreased from 147 nm (pure MoO₃) to 65 nm, and the particle size calculated by Brunauer–Emmett–Teller method (d_BET) decreased from 272 nm to 83 nm, indicating improved thermal stability of Si-doped MoO₃. The 3 wt.% SiO₂-MoO₃ sensor showed optimal sensitivity and selectivity toward NH₃. When the SiO₂ content exceeded 1.5 wt.%, the precipitated SiO₂ narrowed local conduction pathways, increasing resistance and thereby enhancing NH₃ sensitivity.

Sulfur (S) doping represents another effective strategy for enhancing the gas sensing performance of semiconductors, primarily by promoting charge transfer between oxides and sulfides. Shuvo et al. reported a S-doped Ti₃C₂T_x MXene gas sensor, which shows a greater toluene gas sensing performance compared to their undoped counterparts with unique selectivity and long-term stability [Figure 7D]^[156]. S doping leads to a 214% and 312% increase in sensing response for 1 ppm and 50 ppm toluene, respectively, compared to undoped materials. Simultaneously, it exhibits a pronounced response to extremely low concentrations of 50 ppb toluene, demonstrating its exceptionally low detection limit. Li et al. synthesized biomorphic mesoporous SnO₂ using biomass carbon as a template and subsequently introduced sulfur through chemical vapor deposition to obtain S-doped SnO₂^[157]. The S-terminated sites of the resulting SnS₂ phases act as active reaction centers for NO₂ adsorption, thereby improving sensing performance. The S-doped SnO₂ sensor exhibited an excellent response of 57.38 toward 100 ppm NO₂, with a fast response time of 1.60 s and a detection limit of 10 ppb at room temperature. This performance enhancement was attributed to the formation of S-Sn-O bonds, which serve as electron-transfer bridges, reduce interfacial state density, increase carrier concentration, and generate more chemisorbed oxygen, thus accelerating NO₂ reaction kinetics at room temperature. Xu et al. further achieved S-doped surface modification of SnO₂ by sintering flower-like SnS₂^[158]. The obtained flower-like mesoporous S-doped SnO₂ exhibited ultrahigh sensitivity to NO₂ at low operating temperatures (50 °C), with an R_g/R_a value of 600 toward 5 ppm and a detection limit of 50 ppb (R_g/R_a = 11). The enhanced performance was attributed to the modified surface chemistry induced by S doping, which endowed the SnO₂ surface with higher catalytic reactivity. In situ Raman spectroscopy and DFT calculations confirmed that S-doped SnO₂ possesses superior catalytic activity and strong adsorption capability, and that its sensing performance is closely related to the S doping level.

Elemental doping represents a powerful strategy for tailoring the intrinsic physicochemical properties of MOS materials to achieve high-performance gas sensing. Different dopants operate through distinct mechanisms [Table 2]. Rare-earth elements, characterized by their partially filled 4f orbitals and variable oxidation states, are particularly effective in modulating band structures, enhancing surface alkalinity, and improving moisture tolerance, thereby enabling superior sensitivity under humid conditions. Transition-metal dopants, whose ionic radii are often comparable to those of host cations, can be readily incorporated into the lattice, significantly enhancing sensitivity by modulating carrier concentration, defect chemistry, and O_v. Non-metal dopants primarily influence surface electronic states and local electron distribution, offering notable potential for reducing operating temperatures by facilitating surface reactions and charge transfer. Future research should deepen the mechanistic understanding of the coupled dopant-host-target gas interactions, enabling targeted doping strategies that maximize sensitivity, selectivity, and stability.

p-n junctions

The p-n junction is widely regarded as the most effective heterostructure for gas sensors because the work function difference between the two MOS components drives electron transfer from the higher Fermi level to that with a lower Fermi level. Upon heterojunction formation, Fermi level mismatch further induces charge redistribution, thereby significantly modifying the electronic structure of the nanocomposite. As illustrated in Figure 9A, electrons migrate from the n-type to the p-type MOS, while holes move in the opposite direction until the Fermi level equilibrium is reached^[46]. This process generates a space-charge layer and band bending on both sides of the interface, creating a potential barrier that narrows the electron transport channel. Oxygen adsorption suppresses electron transfer and reduces the effective cross-section for charge conduction, thereby increasing resistance. Exposure to reducing gases triggers reactions with adsorbed oxygen species, releasing electrons back into the MOS and lowering resistance. The large resistance changes before and after gas exposure enhance sensing performance.

Figure 9. (A) Schematic illustration of the band structure of a p-n heterojunction before and after contact. Figure 9A is quoted with permission from Mondal et al.^[46]; (B) Schematic illustration of the band structure (B₁) and surface oxygen vacancy sites (B₂) of the In₂O₃/NiO heterostructure. Figure 9B is quoted with permission from Xie et al.^[169]; (C) NO_x gas sensing mechanism on ordered mesoporous WO₃/ZnO n-n heterojunction. Figure 9C is quoted with permission from Han et al.^[174]; (D) NH₃ gas sensing mechanism on CuO/α-Fe₂O₃ p-p heterojunction. Figure 9D is quoted with permission from Bhuvaneshwari et al.^[175]; (E) Synthesis route diagram (E₁) and ethanol sensing performance (E₂) of multilevel dendritic mesoporous TiO₂-SnO₂ composites. Figure 9E is quoted with permission from Zhao et al.^[76]. PEO: Poly(ethylene oxide).

Wang et al. reported a highly selective and sensitive 3H-2B sensor based on Co₃O₄ NP-modified 3D radial ZnO nanorods^[168]. The ZnO/Co₃O₄ composite exhibited ultrahigh sensitivity (R_a/R_g = 550 at 5 ppm), an ultralow detection limit (10 ppb), and excellent selectivity at 260 °C, attributed to the p-n heterojunction modulated depletion layer and the radial structure that facilitates gas diffusion. The p-n junction also provides additional O_v and active adsorption sites, further improving sensing performance. For example, Xie et al. fabricated a NiO-functionalized 3D ordered macroporous In₂O₃ thin-film sensor via atomic layer deposition (ALD)^[169]. This approach precisely controlled O_v concentration and formed p-n heterojunctions within the In₂O₃/NiO composite, effectively tuning surface chemistry and electrical properties [Figure 9B]. The optimized sensor exhibited a response of 532.2 to 10 ppm NO₂ at 145 °C, representing a 26-fold improvement over pristine In₂O₃.

Surface reactions at p-n heterojunction interfaces also play a critical role in achieving high sensing performance. Xiao et al. fabricated mesoporous p-n heterojunction MOS based on n-type WO₃ and p-type NiO^[170]. Oxygen adsorption creates an electron-depleted surface layer before H₂S exposure. Upon contact with H₂S, both adsorbed oxygen and metal oxides react with the gas, releasing abundant electrons and sharply decreasing resistance. During sensing, SO₂ and conductive sulfides (WS₂ and NiS) are formed, further reducing resistance and enhancing performance. The mesoporous structure, with high porosity and abundant active sites, facilitates gas diffusion and adsorption. At the same time, ultrafine NiO nanocrystals embedded in the WO₃ framework provide numerous p-n junctions, thereby collectively yielding outstanding H₂S sensing performance. Similarly, Shao et al. demonstrated that decorating n-type SnO₂ nanowires with p-type CuO NPs significantly improves H₂S detection^[171]. In air, the extended EDL induced by CuO NPs increases resistance, whereas exposure to H₂S leads to the formation of CuS, which eliminates nanoscale p–n junctions and results in a sharp resistance drop.

Heterojunctions also influence selectivity. Jeong et al. converted Co₃O₄ HSs into Co₃O₄-SnO₂ core-shell HSs via electrochemical replacement^[172]. While pure Co₃O₄ showed selectivity toward ethanol, the core-shell structures exhibited high selectivity to xylene and toluene at 275 °C and 300 °C, respectively, owing to the synergistic catalytic effects of SnO₂ and Co₃O₄ combined with the pore confinement effect. Moreover, the microstructure and exposure configuration of heterojunctions are crucial. Woo et al. investigated ZnO-Cr₂O₃ nanowire heterojunctions for TMA detection^[173]. They found that decoration with discrete Cr₂O₃ NPs enhanced the response through catalytic activity and p-n junction formation, whereas uniform Cr₂O₃ shell coating led to p-type conduction dominance and diminished response compared to pure ZnO nanowires.

n-n or p-p junctions

Band bending can also occur in n-n and p-p heterojunctions due to energy level differences between two n-type (or p-type) metal oxides. For n-n heterojunctions, variations in conduction band energies drive electron transfer across the interface, leading to depletion in the material with the higher conduction band energy and accumulation in that with the lower energy [Figure 9C]^[174]. In p-p heterojunctions, holes serve as the majority carriers [Figure 9D]^[175]. Differences in valence band energies cause holes to migrate from the material with the higher valence band energy to the one with lower energy, resulting in hole depletion and accumulation regions, respectively^[176]. Both n-n and p-p junctions arise from work function differences, which induce carrier migration to maintain equilibrium and amplify electrical signal variations^[30]. Additionally, heterojunction formation promotes oxygen adsorption, generating abundant O_v and providing extra active sites for sensing reactions^[177]. The mesoporous structure of MOS composites further facilitates gas adsorption and desorption, collectively enhancing sensor sensitivity and response speed.

Zhao et al. constructed multi-scale dendritic mesoporous TiO₂-SnO₂ nanocomposites with well-defined n-n heterojunctions for highly sensitive ethanol detection [Figure 9E]^[76]. The hydroxyl-rich surface of SnO₂ nanocrystals forms hydrogen bonds with the PEO segments of the PEO-b-PS template, enabling their homogeneous dispersion within the mesoporous TiO₂ matrix and generating numerous TiO₂-SnO₂ n-n heterojunctions. Electron transfer from TiO₂ to SnO₂ leads to surface electron accumulation on SnO₂, which is subsequently depleted by oxygen adsorption, increasing the interfacial energy barrier and enhancing the gas response. Additionally, the multi-level dendritic mesoporous structure, featuring large pores, abundant active sites, and short diffusion paths, accelerates gas diffusion and reaction, thereby improving sensitivity and response speed. Similarly, Wang et al. synthesized highly crystalline mesoporous TiO₂/WO₃ n-n heterojunctions via an acid-base pair-regulated self-assembly route^[178]. The material exhibits large pores (~ 21.1 nm), crystalline frameworks, and high surface area, enabling efficient oxygen adsorption and rapid acetone diffusion. Importantly, the TiO₂/WO₃ n-n heterojunction promotes the formation of an EDL, thus amplifying the sensing response and ensuring fast response/recovery characteristics. Other n-n heterojunction systems, including SnO₂-ZnO hollow dodecahedra^[179], porous MO₃/SnO₂ NSs^[180], In₂O₃@SnO₂ core-shell NFs^[181], and 3D anti-opalescent ZnO-In₂O₃ multilayer films^[182], have also demonstrated significantly enhanced gas sensing performance.

The formation of p-p heterojunctions can also markedly enhance gas sensing performance. Alali et al. fabricated porous Co₃O₄/CuO p-p heterojunction nanotubes by electrospinning and demonstrated that visible light irradiation significantly improves their response to dimethyl methylphosphonate (DMMP), a simulant of sarin nerve agent^[77]. The Co₃O₄-CuO heterojunction lowers the interfacial potential barrier, and oxygen chemisorption increases hole concentration, leading to reduced resistance in air. Light excitation further promotes the generation of holes by transferring electrons from the valence to the conduction band, thereby decreasing resistance. In a reducing DMMP atmosphere, electron donation to the sensing layer results in a pronounced resistance increase. Suh et al. synthesized NiO-modified Co₃O₄ nanorods via multi-step grazing-incidence deposition for efficient benzene detection^[183]. The conformal NiO coating induces numerous p-p heterojunctions, lowering hole concentrations in Co₃O₄ and thus increasing baseline resistance, which amplifies the sensing response compared to pristine Co₃O₄. Additionally, modified surface lattice orientation and the catalytic activity of NiO further enhance gas sensing properties. Other p-p heterojunction systems, including mesoporous NiO@CuO flower-like hierarchical composites^[184], Cu₂O/CuO core-shell nanowires^[185], and NiO/NiCr₂O₄ nanocomposites^[186], have also demonstrated significantly improved sensing performance.

Overall, heterojunction engineering provides a versatile and powerful toolbox for tailoring the electronic and chemical properties of MOS sensing materials. By carefully selecting material pairs and controlling interface nanostructure, it is possible to achieve unprecedented levels of sensitivity, selectivity, and response speed. The design of high-performance heterojunction-based MOS sensors relies on several fundamental principles. The band alignment between two semiconductors determines the direction and degree of interfacial charge transfer, which directly modulates the width of the depletion layer and the height of the potential barrier. The presence of mesoporous frameworks can amplify interfacial effects by providing abundant exposed junctions and facilitating rapid gas diffusion. The selection of semiconductor components with complementary catalytic activities and adsorption affinities enables synergistic surface reactions. These principles collectively provide a rational foundation for engineering heterojunctions with optimized electronic structures, enhanced reaction kinetics, and improved sensing performance. The following section on applications will showcase how these designed heterojunctions address specific detection challenges across various fields.