Recent Advances in Metal Oxide Semiconductor Heterojunctions for the Detection of Volatile Organic Compounds

Abstract

The efficient detection of volatile organic compounds (VOCs) is critically important in the domains of environmental protection, healthcare, and industrial safety. The development of metal oxide semiconductor (MOS) heterojunction gas-sensing materials is considered one of the most effective strategies to enhance sensor performance. This review summarizes and discusses the types of heterojunctions and their working principles, enhancement strategies, preparation methodologies, and applications in acetone and ethanol detection. To address the constraints pertaining to low sensitivity, sluggish response/recovery times, and elevated operating temperatures that are inherent in VOC sensors, several improvement methods are proposed, including doping with metals like Ag and Pd, incorporating additives such as MXene and polyoxometalates, optimizing morphologies through a fine design, and self-doping via oxygen vacancies. Furthermore, this work provides insights into the challenges faced by MOSs heterojunction-based gas sensors and outlines future research directions in this field. This review will contribute to foundational theories to overcome existing bottlenecks in MOS heterojunction technology while promoting its large-scale application in disease screening or agricultural food quality assessments.

1. Introduction

Volatile organic compounds (VOCs) generated from combustion emissions, industrial production, and biological metabolism, such as ketones, alcohols, methane, benzene, etc., have potential impacts on the environment and human health [1,2,3,4,5]. Therefore, monitoring VOCs has become one of the important tasks for environmental protection and health protection today. A variety of VOC sensors based on different working mechanisms and technologies are emerging, including metal oxide semiconductor (MOS) sensors, nanomaterials sensors, functionalized polymer sensors, and intelligent sensing systems based on machine learning and artificial intelligence [6,7,8,9,10,11,12,13,14,15,16,17].

Among these sensors, MOS sensors are widely used in various fields due to their advantages, such as simple manufacturing process, low energy consumption, and cost-effectiveness. In particular, zinc oxide (ZnO)- [18,19,20,21,22,23], copper oxide (CuO)- [24,25,26,27,28,29,30,31,32], tin dioxide (SnO₂)- [33,34,35,36,37,38,39,40], nickel oxide (NiO)- [41,42,43,44,45,46], and cobalt tetroxide (Co₃O₄)-based [47,48] materials are extensively employed as sensing materials in gas sensors. However, MOSs sensors still face some key challenges in sensing performance that need to be addressed, such as high detection limit, high operating temperature, high baseline resistance, and poor moisture resistance. Consequently, various strategies have been developed to enhance sensing performances, including constructing special morphologies, introducing oxygen vacancies, and doping impurities. Among these, constructing heterojunctions and combining them with other strategies are widely adopted approaches.

This review aims to explore the improvement of VOC detection by chemiresistive sensors through the construction of MOS heterojunctions. Specifically addressed will be the types of heterojunctions, underlying working principles, improvement measures, and preparation methodologies as well as applications in acetone and ethanol detection.

2. Types of Heterojunctions and Preparation Methods

2.1. Types of Heterojunctions

Heterojunctions can generally be defined as interface structures composed of semiconductor contacts with two different band structures. According to the relationships between the energy bands of MOSs, heterojunctions can be categorized into three types: straddling-gap-like, staggered-gap-like, and broken-gap-like (Figure 1). For a type I heterojunction with a straddling gap, the bottom of the conduction band and the top of the valence band of the semiconductor material with a small band gap (E_gap1) are in the band gap of the semiconductor material with a wide band gap (E_gap2). For a type II heterojunction with a staggered gap, the band gaps of two semiconductor materials are interlaced with each other. For a type III heterojunction with a broken gap, the bandgap of two semiconductor materials is completely staggered. Note that the band gap energy of MOSs, defined as the energetic separation between the valence band maximum and the conduction band minimum within a material, varies based on factors such as crystalline phase, purity levels, and other pertinent parameters. In the context of MOS heterojunctions, the band gap of frequently utilized materials like α-phase iron(III) oxide (α-Fe₂O₃) is approximately 2.0 to 2.4 eV, whereas titanium dioxide (TiO₂) demonstrates a band gap in the range of 3.0 to 3.5 eV [49]. As an illustrative example, ZnO with a band gap of 3.4 eV and indium oxide (In₂O₃) with a band gap of 3.6 eV can potentially form a type II heterojunction [50], showcasing a distinctive alignment of energy bands conducive to specific electronic and optoelectronic applications.

2.2. Working Principle of Heterojunction Sensors

Fabricated MOS-heterojunction-based materials can be used in the sensing materials in VOC detection. The working principle of heterojunction sensors for VOC detection is complex and diverse.

This review focuses on chemiresistive sensors, which operate based on the resistance changes that result from the adsorption and desorption processes of gas molecules, alongside the chemical reactions that occur at the surfaces of sensing materials. The sensing mechanism of chemiresistive gas sensors relies on a fluctuation in the resistance of sensor materials [51,52]. When exposed to air, oxygen molecules initially adsorb onto the surface of the sensing material (Equation (1)). After reaching an equilibrium between chemisorbed sites and the oxygen molecules, the adsorbed oxygen molecules draw electrons from the conduction band of the material and involve into various oxygen species (O²⁻, O⁻, and O²⁻; Equations (2)–(4)) [53]. This electron-trapping process increases the concentration of holes and thus reduces/increases the resistance of n-type/p-type semiconductors (R_a).

When the sensor is transferred to the reducing gas environment at the optimal operating temperature, the target gas will undergo a reaction with oxygen ions, resulting in the liberation of free electrons into the conduction band. This process facilitates the recombination process between holes and electrons, resulting in fluctuations in the sensor resistance (R_g). Specifically, ethanol or acetone reacts with oxygen ions to produce CO₂ and H₂O and release electrons (Equations (5)–(8)) [54,55]. As indicated in Equations (2)–(4), the key oxygen species depend on the optimal operating temperature [54,55]. Within the common range of optimal operating temperatures for acetone and ethanol detection (100–300 °C), the dominant oxygen species is O⁻. For ethanol reaction, the intermediates depend on the surface properties of the materials, which could be CH₂CH₂ or CH₃CHO [56,57].

$${\mathrm{O}}_2\left(\mathrm{gas}\right)\to {\mathrm{O}}_2\left(\mathrm{ads}\right),$$

(1)

$${\mathrm{O}}_2\left(\mathrm{ads}\right)+{\mathrm{e}}^{-}\to {\mathrm{O}}_2^{-}\left(\mathrm{ads}\right)\left(\mathrm{T}<100°\mathrm{C}\right),$$

(2)

$${\mathrm{O}}_2\left(\mathrm{ads}\right)+{\mathrm{e}}^{-}\to 2{\mathrm{O}}^{-}\left(\mathrm{ads}\right)\left(100°\mathrm{C}\le \mathrm{T}<300°\mathrm{C}\right),$$

(3)

$${\mathrm{O}}_2\left(\mathrm{ads}\right)+{4\mathrm{e}}^{-}\to 2{\mathrm{O}}^{2-}\left(\mathrm{ads}\right)\left(\mathrm{T}<300°\mathrm{C}\right),$$

(4)

$$\mathrm{C}{\mathrm{H}}_3\mathrm{COC}{\mathrm{H}}_3+8{\mathrm{O}}^{-}\left(\mathrm{ads}\right)\to 3\mathrm{C}{\mathrm{O}}_2\left(\mathrm{gas}\right)+3{\mathrm{H}}_2\mathrm{O}\left(\mathrm{vap}\right)+8{\mathrm{e}}^{-},$$

(5)

$$\mathrm{C}{\mathrm{H}}_3\mathrm{COC}{\mathrm{H}}_3+8{\mathrm{O}}^{2-}\left(\mathrm{ads}\right)\to 3\mathrm{C}{\mathrm{O}}_2\left(\mathrm{gas}\right)+3{\mathrm{H}}_2\mathrm{O}\left(\mathrm{vap}\right)+16{\mathrm{e}}^{-},$$

(6)

$$\mathrm{C}{\mathrm{H}}_3\mathrm{C}{\mathrm{H}}_2\mathrm{OH}+6{\mathrm{O}}^{-}\left(\mathrm{ads}\right)\to 2\mathrm{C}{\mathrm{O}}_2\left(\mathrm{gas}\right)+3{\mathrm{H}}_2\mathrm{O}\left(\mathrm{vap}\right)+6{\mathrm{e}}^{-},$$

(7)

$$\mathrm{C}{\mathrm{H}}_3\mathrm{C}{\mathrm{H}}_2\mathrm{OH}+6{\mathrm{O}}^{2-}\left(\mathrm{ads}\right)\to 2\mathrm{C}{\mathrm{O}}_2\left(\mathrm{gas}\right)+3{\mathrm{H}}_2\mathrm{O}\left(\mathrm{vap}\right)+12{\mathrm{e}}^{-}$$

(8)

2.3. Strategies for Improving the VOC Sensing Performances of MOSs Heterojunctions

The above working mechanism provides hints for developing highly sensitive and selective VOC sensors. To improve the sensing performance of MOSs heterojunctions, the following strategies can be adopted (Figure 2):

(I) Adjust the Fermi level of the heterojunction components. The Fermi level control of MOSs can be achieved through various methods, mainly based on changing the electronic structure of the semiconductor material or environmental conditions. Specifically, this involves the following strategies:

Changing the concentration of doping elements. It is widely accepted that doping can modify the electronic properties of materials. For n-type doping, the doping element provides additional electrons. Consequently, reducing the concentration of these doping elements can lower the concentration of free electrons, thereby shifting the Fermi level towards the valence band and lowering the Fermi level. Conversely, increasing the doping concentration of doping elements can raise the Fermi level. For p-type doping, this approach is usually used to increase the hole concentration of p-type semiconductors. Changing the p-type doping concentration may indirectly affect the Fermi level distribution within MOSs structures.

Temperature control. Temperature is an important factor affecting the Fermi level. Lowering the operating temperature can shift the Fermi level of a semiconductor towards the valence band, as the thermal motion of electrons weakens at low temperatures, resulting in more states occupying lower energy levels. When the temperature is increased, the thermal motion of electrons is enhanced, which can cause the Fermi level of the semiconductor to move towards the conduction band. However, it should be noted that excessively high temperatures may lead to performance degradation in, or even the failure of MOSs materials.

External energy field regulation. By applying an external electric field, the distribution of electrons and holes in MOSs can be altered, thereby affecting the position of the Fermi level. The light irradiation may also change the Fermi level of MOSs by shifting the band structure of MOSs. For example, Liu et al. found that under 370-nanometer light illumination, the responses of a p-SmFeO₃/p-YFeO₃ planar-electrode sensor to 30 ppm of ethanol, acetone, and methanol gases were 1.35, 1.28, and 1.59 times higher than those without light, respectively, [58]. However, the effectiveness of this method may be limited by light intensity, wavelength, and MOSs’ intrinsic properties.

Interface engineering. By optimizing the interface structure, such as through interface passivation, surface treatment, etc., the interface state density and Fermi level pinning effect can be altered to regulate the Fermi level [59,60,61,62,63,64].

The band structure of MOS heterojunctions can be regulated through the above strategies, thereby influencing its gas sensing performance.

(II) Morphology design. For the sensing performances of chemiresistive sensors based on MOSs, one of the widely employed strategies is via designing structures with a high specific surface area and porosity through precise micromorphology control. A large surface area can offer rich active sites for the target gas, thereby enhancing the intensity of the response [65,66,67,68]. The high porosity guarantees a high diffusion rate of the gas molecules, shortening the response/recovery time. In addition to the common nanosheets and microspheres, recently developed inverse opal photonic crystals (IOPCs) with a 3D periodic macroporous array spherical symmetry structure and fractal-like structures with interconnected and cross-connected connections are also emerging as good sensing morphologies [6,69]. Some detailed examples about the morphology design of MOS heterojunction materials for acetone and ethanol detection will be provided in Section 3 and Section 4.

(III) Construction of surface defects. Creating an appropriate amount of oxygen vacancies or other defects on the surface of the material is also widely employed to improve sensing performances [55,70,71,72,73]. These defect sites can serve as active centers, promoting the adsorption and reaction of gas molecules. A suitable amount of defect sites can significantly improve the sensing performance of materials. However, excessive defects may also lead to a decrease in material stability. Therefore, it is necessary to identify the optimal equilibrium concentrations for defects.

2.4. Preparation Methods

In addition to the strategies mentioned above to improve the intrinsic properties of MOSs, optimizing the preparation process of MOSs is also an important strategy to enhance the VOC sensing performance of MOS heterojunctions.

Common preparation methods for semiconductor heterojunction materials for VOCs sensors include the hydrothermal/solvothermal method, the co-precipitation method, the sol–gel method, the calcination method, etc. (Figure 3).

Among the various synthesis methods, the co-precipitation reaction stands out as an efficient and gentle approach, which is attributed to its low energy consumption, facilitated large-scale production, abbreviated reaction duration, and ability to yield highly uniform materials. For example, CuO-ZnO heterojunction nanocomposites have been successfully synthesized by a co-precipitation method [74]. Shruthi et al. fabricated Y₂O₃-In₂O₃ heterostructure nanocrystals using a combined co-precipitation and sol–gel method for methanol sensing [75].

The hydrothermal/solvothermal method is also a common preparation method. This method uses solvation, e.g., water as the reaction medium under high temperatures and high pressures. By adjusting parameters such as reaction temperature, pressure, and material ratio, precise control of the product formation process can be achieved. Moreover, solvothermal reactions occur in a closed system, resulting in a pollution-free reaction process that meets the requirements of green synthesis. The sensing materials fabricated by this method usually present good crystallinity. For example, Sm₂O₃/ZnO/SmFeO₃ microspheres [76], cubic rhombic In₂O₃ microspheres [77], TiO₂ nanofilm [78], and α-Fe₂O₃/NiO nanorods used for VOC detection were successfully prepared by the hydrothermal/solvothermal method [42]. However, there are also disadvantages, such as high equipment costs, harsh process conditions, and difficulty in product separation and purification.

The calcination/annealing method is also commonly used for synthesizing semiconductor heterojunction materials for VOC sensors, but this method usually requires high temperatures. Typically, various precursors such as layered double hydroxides (LDHs), are prepared first, and then semiconductor heterojunction materials are obtained through high-temperature calcination/annealing (usually at 300–1000 °C) [79,80,81]. For instance, a highly sensitive acetone sensing performance based on p-SmFeO₃/n-ZnO nanocomposites was successfully synthesized by a thermal treatment method [82]. Notably, the annealing temperature may affect the phase of as-prepared nanomaterials [83].

Other than these, electrospray-assisted fabrication or the spray-drying method may also be employed [84]. For example, Li et al. fabricated ZnO-SnO₂ heterojunction IOPCs via the spray-drying approach [69]. Spray drying is liquid-waste-free. It possesses the ability to form spherical or sphere-like particles in a single-step process, exhibiting excellent scalability and a remarkable production speed (within milliseconds), thereby presenting a viable option for the rapid and scalable fabrication of IOPBs [69].

In the following sections, we will focus on the discussion of the detection of two representative VOCs, acetone and ethanol, based on MOS heterojunctions.

3. P-n Heterojunctions Sensors for VOC Detection

In the past few decades, n-type MOSs (n-MOSs) have drawn significant attention for VOC detections. N-MOSs usually present higher sensitivity to a given gas than p-type MOSs (p-MOSs). That reason for this is that p-MOS and n-MOS sensors work in a hole-accumulation layer and an electron-depletion layer, respectively. This difference yields a smaller resistance change in p-MOSs than that in n-MOSs under identical band-bending conditions. The gas response follows the equation S = exp(−qΔVs/mkT), where m, qΔVs, k, and T denote constants, band bending, the Boltzmann constant, and absolute temperature, respectively. For p-MOSs sensors, m ≥ 2, and its value rises with an increase in the degree of upward band bending, whereas m = 1 for n-MOSs sensors. However, n-MOSs also show the disadvantages of high operating temperature and sensitivity to humidity. By contrast, p-MOSs could show lower operating temperatures due to their superior redox catalytic properties to those of n-MOSs. This is because p-MOSs are able to yield a hole accumulation layer, which can chemisorb the oxygen molecules of a high concentration. In addition, investigations imply that the p-MOS sensors also exhibit low humidity dependence.

Apparently, the fabrication of p-n heterojunction structures using suitable components shows the potential to reach a low detection limit, low working temperature, low base line resistance, and strong moisture tolerance by taking advantage of each component. In addition, the formed interface at the heterojunction also provides new reactive sites and adjusts the electronic properties of the sensor materials.

3.1. P-n Heterojunctions Sensors for Acetone Detection

Studies have shown that the fabrication of p-n heterojunction structures can efficiently regulate band bending and hole concentration and, consequently, promote resistance for acetone detection [85]. Zhao et al. found that compared with Co₃O₄, the formation of a n-p heterojunction for a ZnO-Co₃O₄ heterostructure reduced both the degree of upward band bending and the surface hole concentration under the influence of electron transfer from ZnO to Co₃O₄ [85]. The reduction in the surface hole concentration demonstrated that even minimal sensing reactions (corresponding to less surface charge transfer) can cause significant resistance changes, which, along with the excellent performance of the p-MOS, resulted in high sensitivity to and low detection limit of acetone gas [85].

Introducing oxygen vacancy is an effective strategy to improve the sensitivity of reducing gas. In general, intrinsic oxygen vacancies formed at the heterojunction interface induced by lattice mismatches are thought to play an important role in improving charge transport between heterojunction structures. Three types of oxygen species may exist on the surfaces of sensitive materials: chemisorption oxygen species (O_C), vacancy oxygen species (O_V), and lattice oxygen species (O_L). Among these, O_L is regarded as having low activity and is not involved in the sensing reaction. O_C is associated with chemisorption and the presence of ionized oxygen on the surface of the sensing material. O_vs facilitate efficient gas molecular adsorption and sensing reactions by providing more active sites. Therefore, the content of O_V is regarded as an indicator for obtaining better gas-sensitive characteristics. Hu et al. found that Fe₂O₃-Co₃O₄, with a higher content of O_V than Fe₂O₃ and Co₃O₄, presented better sensing characterizations (Figure 4a–d) [86]. As demonstrated in Figure 4c, after fabricating a heterojunction structure by combining Fe₂O₃ and Co₃O₄, the sensor exhibited an improved response to acetone, accompanied by a significant enhancement in selectivity.

In another work, Wu et al. found that during a series of the oxide cage/nanofiber heterostructure composed of ZIF-67-derived Co₃O₄ and In₂O₃, the one with the highest O_V content and Co²⁺/Co³⁺ ratio delivered the best sensing performances, demonstrating an ultrasensitive response of 954 at 50 ppm acetone concentration at 300 °C, coupled with a low detection limit of 18.8 ppb (as shown in

Table 1. Comparison of the responses of sensors for acetone detection. N.A. means that the data are not available. The same applies below.

Materials	Type	Concentration (ppm)	Operating Temperature (°C)	Response (Ra/Rg)	Detection Limit (ppm)	Response/Recovery Time (s)	Reference
Fe2O3-Co3O4	p-n	100	200	91.5	N.A.	20/21	[86]
Co3O4/In2O3	p-n I	50	300	954	0.0188	4/148	[55]
Ag-NiO/SnO2	p-n II	1	190	15.7	0.05	12/25	[87]
SnO2-ZnO	n-n II	100	250	0.2	35	3/333	[88]
SnO2/ZnSnO3	n-n I	100	290	0.045	30	1/6	[89]
ZnO-CuO	p-n II	1	200	11.14	0.009	N.A./N.A.	[90]
TiO2/a-Fe2O3	n-n II	100	225	21.9	0.036	13/10	[91]
ZnFe2O4/SnO2	n-n II	100	210	120	0.1	30/197.2	[92]
ZnSnO3/ZnO/Ti3C2Tx MXene	N.A.	100	120	15.68	90	5 /12	[54]

). Additionally, it exhibited a swift response time of 4 s and favorable selectivity and repeatability, as well as robust long-term stability [55]. Besides the factor of the O_V content, the formation of the type I heterojunction results in an augmentation of the degree of upward energy band bending of In₂O₃ in the air and a larger change in the electron concentration induced by acetone, leading to a higher gas response than pure In₂O₃ (Figure 5) [55]. In addition to intrinsic O_vs induced by the lattice mismatches of heterojunction components, common methods for generating oxygen vacancies include reduction treatment methods (using reducing agents such as H₂, NaBH₄, imidazole, L-ascorbic acid, etc.), the atmosphere deoxidation method (for the atmospheres of vacuum or inert gases, e.g., Ar, N₂, He), and the ion doping method. Doping elements may react with oxygen atoms or occupy the positions of oxygen atoms, thereby indirectly introducing oxygen vacancies.

To create oxygen vacancies, another promising approach is to introduce noble metals into the sensing layer. Moreover, the introduced noble metals, such as Ag, Au, and Pd, can also facilitate oxygen dissociation. To date, various synthesis routes for decorating noble-metal nanoparticles (NPs) into MOSs heterojunction structures for gas-sensing applications have been developed. Yang et al. proposed an easy and general route for the synthesis of Ag-NiO/SnO₂ nanotubes (NTs) using a combined method of electro-spinning–coating–calcination–wet chemical reduction. In the last step, Ag NPs were introduced via a wet chemical reduction method (Figure 6a). The morphologies of one-dimensional (1D) NTs and the internal hollow structures of Ag-NiO/SnO₂ (Figure 6b,c) promoted the adsorption and the desorption of gas molecules, facilitating the availability of more active sites. The incorporation of p-n heterojunction structures into the 1D nanotubes (NTs) further expedited the charge transfer process. Subsequent decoration with Ag nanoparticles (NPs) augmented the number of surface oxygen vacancies, contributing to exceptional sensing performance (Figure 6d). The Ag-NiO/SnO₂ NT sensors exhibited remarkable gas-sensing properties towards acetone, featuring a superior response (R_a/R_g = 15.7 at 1 ppm), excellent selectivity (the response to 1 ppm of acetone exceeded that to 5 ppm of interfering gases), swift response (12 s) and recovery (25 s) times at 1 ppm of acetone, and a minimal detection limit of 50 ppb at a low operating temperature of 190 °C [87].

The preparation of structurally stable heterojunctions is conducive to enhancing the performance of the sensor [93,94]. Generally, diverse in situ strategies, such as in situ growth and in situ oxidation, can be employed to fabricate stable structures. This stable structure can offer tight p-n contact surfaces and robust chemical bonds, increase the carrier density, and facilitate charge transfer. Utilizing WS₂ as the primary phase, WS₂/WO₃ nanosheets with a p-n heterojunction were formed through in situ growth of WO₃. The assembled gas sensor based on the type II WS₂/WO₃ heterojunction nanosheets demonstrated acetone detection with satisfactory repeatability, selectivity, and long-term stability [95]. A similar type II heterojunction of ZnO-CuO nanomaterial with a core–hollow cube morphology was fabricated through the formation of ZnO-Cu₂O core–cube nanostructures and thermal oxidation in air (Figure 6e). The ZnO-CuO core–hollow cube nanostructure was employed as the acetone sensing material, yielding a response of 11.14 at 1 ppm acetone and 200 °C and a detection limit of approximately 9 ppb, with the detection range spanning from 40 ppb to 10 ppm. Its outstanding performance was primarily attributed to its uniform and distinctive morphology. The core–shell structure, featuring ZnO as the core and CuO as the shell, maximizes the p-n heterojunction effect and increases resistance, thereby resulting in high sensitivity and low detection limits. Additional structural factors, such as small grain size and large surface area, can also cause significant changes in resistance. Techniques such as precious metal decoration and surface functionalization can further improve sensitivity and selectivity [90]. Wang et al. directly grew a TiO₂ nano ribbon (NR) array on a ceramic electrode, and the α-Fe₂O₃ nanoscale branches were epitaxially grown on the TiO₂ NRs to obtain TiO₂/α-Fe₂O₃ hierarchical heterostructures through a simple hydrothermal method. The length of the α-Fe₂O₃ branch can be precisely controlled by the reaction synthesis time. Compared with pure TiO₂ NRs, the response and selectivity of the TiO₂/α-Fe₂O₃ heterostructures to acetone were significantly enhanced within a wide temperature range (175–275 °C) [91].

3.2. P-n Heterojunctions Sensors for Ethanol Detection

Ethanol, as a typical species of VOCs, has a wide range of applications in various fields. Monitoring the concentration of ethanol in the environment could provide important information for alcohol-driving detection, industrial production, and medical diagnosis. Hence, the study of ethanol sensors has been widely explored.

A series of MOS heterojunction composites have been constructed for ethanol detection, including n-n heterojunctions (

Table 2. Comparison of the responses of sensors for ethanol detection.

Materials	Type	Concentration (ppm)	Operating Temperature (°C)	Response (Ra/Rg)	Detection Limit (ppm)	Response/Recover Time (s)	Reference
Mesoporous In2O3-ZnO	n-n II	100	225	35	0.2	4/90	[50]
Core-double shell ZnO@In2O3 @ZnO	n-n	100	200	453	1	20/190	[99]
MoO3 nanorods @SnO2 nanosheets	N.A.	100	200	48.64	N.A.	65/230	[56]
In2O3@PW12@SnO2	n-n	100	320	22.6	0.0139	1/132	[97]
Porous Zn2SnO4/CdSnO3 nanocubes	n-n II	100	300	214.38	0.285	30/55	[57]
Hollowed-out Fe2O3-loaded NiO heterojunction nanorods	p-n I	10	150	51.2	0.5	N.A./N.A.	[98]
ZnO nanosheets @In2O3 hollow microrods	n-n	100	200	1	269.1	18/35	[100]
In2O3/SnO2	n-n II	100	200	N.A.	49.16	32/40	[101]
CuO/TiO2/MXene	N.A.	1	30	0.95	0.3	16/13	[102]
SnO2/ZnO/Ti3C2Tx MXene	N.A.	100	120	121.1	N.A.	3/141	[103]

), such as ZnO/SnO₂ [69,88], ZnO/α-Fe₂O₃ [96], and In₂O₃/SnO₂ [97], and p-n heterojunctions, such as α-Fe₂O₃/NiO [98].

To overcome the high resistance of p–n heterojunctions, some suitable additives can be added, such as doping additives with moderate catalytic activity, covering polydimethylsiloxane (PDMS), carbonaceous material, and loading noble metals [104]. Two-dimensional (2D) metal materials, such as MXene, are ideal additive materials owing to their distinctive layered morphology, high electrical conductivity, substantial surface area, superior hydrophilicity, robust thermal stability, and environmentally benign characteristics [104]. Consequently, the integration of MXene with TiO₂/CuO composites led to a reduction in the sensor’s resistance and a subsequent enhancement of its gas-sensing performance. As illustrated in Figure 7, the MXene was added during the preparation process, and it transferred electrons to the bonded p-type CuO, resulting in reduced resistance of the heterojunction. The sensor based on the p-type CuO/TiO₂/MXene composite exhibited an exceptional response value of 95% at 1 ppm of ethanol, coupled with swift response and recovery times of 16 and 13 s, respectively, a low detection limit of 0.3 ppm, and outstanding long-term stability, which were observed for the trace detection of ethanol gas at room temperature [102].

In order to obtain enough reactive sites for the enhanced resistive ability of heterojunction sensing materials, researchers have been engaged in constructing diverse morphologies with high specific surface areas. Finely designed and prepared three-dimensional (3D) shapes such as hollow structures, nanoflowers, mesh-like and sponge-like structures, etc., could maintain a high specific surface area while also exhibiting excellent mechanical properties and stability. The hollowed-out Fe₂O₃-loaded NiO heterojunction nanorods assembled by porous ultrathin nanosheets (Figure 8a,b) prepared by Li et al. showed superior humidity adaptability during the process of ethanol monitoring. They exhibited an outstanding response of 51.2 toward 10 ppm of ethanol at 80% relative humidity (RH) (Figure 8c), along with exceptional selectivity for ethanol (Figure 8d) under conditions of high relative humidity. This was achieved at a lower operating temperature of 150 °C and with a low detection limit of 0.5 ppb. The excellent selectivity could be attributed to the favorable specificity of the NiO material towards ethanol and the hollowed-out morphology, which allowed massive ethanol gas molecules to adhere to the water molecules to spread into the inside of the material. Moreover, Fe₂O₃@NiO had a higher content of adsorbed oxygen than its components, which was also beneficial to gas-sensing reactions [98].

Moreover, the porous structure is also well-designed, since it not only presents a large specific surface area, but also owns rich channels to facilitate species transport. To acquire porous heterostructures, one of the most extensively employed approaches is to use porous precursors, such as metal–organic frameworks (MOFs) [105,106,107,108,109,110,111,112]. MOFs contain metal centers, which are linked by organic ligands. MOFs feature high porosity and specific surface areas, and their pore size and structure can be regulated by adjusting the types of metal ions and organic ligands and synthesis conditions. When MOFs are calcined at an appropriate temperature, oxides can be obtained, which can inherit the porous structure with rich channels from MOFs. For example, the commonly used n-ZnO, n-type α-Fe₂O₃, p-Co₃O₄, and p-CuO can be obtained from the calcination of zeolitic imidazolate frameworks (ZIF) ZIF-8, ZIF-67, MIL-101(Fe), and HKUST-1 [55,113]. Doan et al. constructed n-ZnO/p-Co₃O₄ heterojunctions derived from ZIF-8/ZIF-67 for the detection of low concentrations of ethanol. The assembled sensor showed great ethanol selectivity in eight interfering gases and a low detection limit [113]. Li et al. constructed Co₃O₄-TiO₂ porous heterojunction nanosheets by calcining ZIF-67/MIL-125 precursors. The sensing results showed that the fabricated Co₃O₄-TiO₂ porous nanosheets exhibited superior ethanol-sensing performances compared to MOF-derived TiO₂ nanotablets, owing to the formation of p–n heterojunctions and unique porous nanosheet nanostructures, which possess a high specific surface area [114].

4. N-n Heterojunction Sensors for VOC Detection

4.1. N-n Heterojunction Sensors for Acetone Detection

In addition to the construction of p-n heterojunction, a few works suggested that the fabrication of n-n heterojunction composites could also improve the sensing properties. The n-n heterojunctions can modulate the energy band and carrier concentration to enhance sensing performances.

Using MOSs as precursors can not only prepare the above-mentioned p-n materials, but also n-n materials. Zhang et al. fabricated a hollow polyhedral SnO₂-ZnO composite from MOFs (Figure 9a) [115]. The optimized 6% SnO₂-ZnO NPs showed remarkable acetone sensing with a 0.82 ppb detection limit and linear response up to 10 ppm [115]. ZnO-SnO₂ heterojunction inverse opal photonic bandgap materials, characterized by a regular three-dimensional ordered macroporous structure, were successfully synthesized through a straightforward spray drying method followed by an impregnation–annealing process. The uniformly distributed ZnO and SnO₂ formed the connected skeleton. The special structure resulted in a five-times-higher response (R_a/R_g) of IOPBs (40.3 to 50 ppm acetone) in comparison to the performance of the pure SnO₂ sensor [69] (Figure 9c). He et al. synthesized porous ZnFe₂O₄/SnO₂ core–shell spheres for high-performance acetone by combining SnO₂ nanospheres with ZnFe₂O₄. The optimal porous ZFO/SNO composite exhibited a high response value of 120 to 100 ppm of acetone at a temperature of 210 °C, accompanied by excellent stability. Furthermore, its detection limit could reach as low as 0.1 ppm, which was attributed to its abundant heterojunctions, porous structure, and small nanoparticle size [92].

As discussed in Section 3.2, in addition to being additive in p-n type heterojunctions, MXene can also act as the support for n-n type heterojunctions. Su et al. fabricated 1D ZnSnO₃/ZnO nanofibers via the electrospinning technique and subsequently attached these nanofibers to 2D Ti₃C₂T_x MXene sheets using electrostatic self-assembly technology, resulting in the formation of 1D/2D heterostructured ZnSnO₃/ZnO/Ti₃C₂T_x MXene composites. The accordion-like MXene slices were uniformly and tightly cross-covered with ZnSnO₃/ZnO nanofibers, forming a strong bond. As n-type semiconductor materials, ZnO and ZnSnO₃ established an n-n heterojunction, which subsequently formed a p-n heterojunction between the ZnSnO₃/ZnO component and the p-type semiconductor material Ti₃C₂T_x MXene within the ZnSnO₃/ZnO/Ti₃C₂T_x MXene composite. The successful construction of multiple heterojunctions and the synergistic interactions among these three components facilitated real-time and efficient detection of acetone gas. Compared to the ZnSnO₃/ZnO samples, the ZnSnO₃/ZnO/Ti₃C₂T_x MXene composite exhibited a response value of 15.68 in 100 ppm acetone gas, representing a 3.5-fold increase. Additionally, the optimal operating temperature was notably decreased to 120 °C [54].

4.2. N-n Heterojunction Sensors for Ethanol Detection

Compared to single-oxide sensing materials, MOS composites typically demonstrate superior sensitivity, improved stability, and a reduced detection limit. Zhang et al. developed hierarchical ZnO-In₂O₃ heterostructures assembled from nanosheets for the purpose of ethanol detection [116]. Yan et al. synthesized ZnO-In₂O₃ porous nanosheets for the application of ethanol sensing; it is noteworthy that the optimal operating temperature required for these nanosheets was relatively high (350 °C) [117]. Recently, Jiang et al. successfully designed and synthesized a distinctive hierarchical In₂O₃-ZnO compound through a hydrothermal calcination process, which exhibited high reproducibility for the detection of ethanol (Figure 10). The gas sensor containing 15 at. % In₂O₃-ZnO achieved a lower detection limit for ethanol (200 ppb) and a reduced baseline resistance (~1 Mohm at 225 °C) compared to pure ZnO. The improved moisture tolerance suggested that an appropriate doping concentration of In³⁺ in ZnO could effectively enhance gas sensing performance [50]. This enhancement is primarily attributed to the smaller size of In₂O₃ nanoparticle clusters, which reduce baseline resistance via electron transfer and lower the detection limit by optimizing humidity tolerance. Additionally, Gao et al. linked the improved gas sensing performance of ZnO@In₂O₃-2 heterostructures to their unique heterogeneous structure construction, abundant O_vs, substantial specific surface area, and internal electric field [100].

Similar results have also been observed in other n-n MOSs heterojunctions, such as MoO₃ (1D) @SnO₂ (2D) core–shell heterostructures [56], an irregular particle-shaped SnO₂/In₂O₃ composite [101], and In₂O₃/Fe₂O₃ shuttle-like structures [118]. Wang et al. found that an ethanol gas sensor using the optimal MoO₃ (1D) @SnO₂ (2D) core–shell heterostructures as the sensing materials exhibited a higher response than its components with 48.64 at the optimum operating temperature (200 °C) to 100 ppm ethanol. The enhanced gas sensing mechanism of MoO₃@SnO₂ can be attributed to its abundance of high-activity sites, narrow band gap, and rich concentration of surface vacancies [56]. Similarly, the formation of n-n heterojunctions, an augmentation in oxygen vacancies, and its distinctive structural characteristics also induced In₂O₃/Fe₂O₃ to show excellent ethanol sensing performances [118]. The response to 100 ppm of ethanol gas achieved a value of 67.5 at an optimal operating temperature of 200 °C, while the response/recovery time was 9 s/236 s [118].

In addition to the commonly used n-type MOSs, such as ZnO, SnO₂, In₂O₃, etc., some other novel sensing materials have also been explored. Spinel oxides with a formula of AB₂O₄ show great potential for gas sensing application [119]. Zn₂SnO₄ is a promising candidate due to its heightened chemical sensitivity, minimal visible light absorption, and outstanding optical electronic properties. Compared with the nanoparticles of SnO₂/Zn₂SnO₄, Zn₂SnO₄ and SnO₂, hierarchical porous SnO₂/Zn₂SnO₄ nanospheres, synthesized via a one-step hydrothermal method, exhibited superior gas sensing properties towards ethanol. At the optimal operating temperature of 250 °C, the sensor fabricated from these porous SnO₂/Zn₂SnO₄ nanospheres demonstrated a peak response of 30.5 to 100 ppm of ethanol [120]. Recently, Ma et al. fabricated hollow Zn₂SnO₄/SnO₂ nanocubes with the n-n heterojunction between Zn₂SnO₄ and SnO₂ grains and found that the Zn₂SnO₄/SnO₂ sensor possessed good responses (R_a/R_g) toward ethanol and formaldehyde compared to other VOC gases [121]. Following these works, superior ethanol-sensing performances were achieved by the in situ synthesis of a porous Zn₂SnO₄/CdSnO₃-nanocube-based n-n heterostructure [57]. Taking Zn₂SnO₄/CdSnO₃ nanocubes as an example, the sensing mechanism is as follows: CdSnO₃ and Zn₂SnO₄ are ubiquitous n-type semiconductor perovskite materials, with electrons serving as the primary charge carriers. Consequently, the oxygen adsorption/desorption model theory can be employed to elucidate the underlying mechanisms of their gas=sensing mechanism for Zn₂SnO₄/CdSnO₃, through Equations (1)–(4), (7), and (8). Firstly, the pore volume and specific surface area of Zn₂SnO₄/CdSnO₃ were higher than those of Zn₂SnO₄ and CdSnO₃, which facilitated the adsorption of reactive oxygen species from the atmosphere, enabling their participation in the gas-sensing reaction and enhancing the overall gas sensing performance. Secondly, the abundant O_vs in the Zn₂SnO₄/CdSnO₃ heterostructure promoted oxygen adsorption, resulting in a thicker oxygen space charge layer and a more pronounced electron depletion layer and electron depletion layer based on the XPS and electron paramagnetic resonance (EPR) spectra. As depicted in Figure 11b, the intensity of the Zn₂SnO₄/CdSnO₃ was much higher than those of the other two samples, revealing the highest content of O_vs in Zn₂SnO₄/CdSnO₃. In n-type MOSs, oxygen vacancies act as electron donors, which is favorable for the adsorption of oxygen molecules onto the surface and the subsequent formation of a space charge layer. Zn₂SnO₄/CdSnO₃ has higher O_vs than pure CdSnO₃, causing greater electron depletion during oxygen adsorption, leading to higher R_a and R_a/R_g. The n-n heterojunction between Zn₂SnO₄ and CdSnO₃ induces electron transfer due to work function differences, creating an electron depletion layer on Zn₂SnO₄. This increases CdSnO₃ carrier concentration, facilitating oxygen adsorption and oxygen species space charge generation [57].

Note that in the above work, the annealing temperature of the CdSn(OH)₆ precursor was 700 °C and produced Zn₂SnO₄. When the annealing temperature decreased to 500 °C, it would produce ZnSnO₃ [57]. This material is also a good candidate for ethanol sensing [122]. In addition to obtaining ZnSnO₃ through calcination, it can also be fabricated through other methods, such as magnetic sputtering. Cao et al. prepared a single-layer SiO₂ microsphere template using the self-assembly method, and used this template to prepare a porous thin film of SnO₂-and-ZnSnO₃ composite by magnetron sputtering. Next, a double-layer porous-film ethanol gas sensor was prepared, and a detection limit of 50 ppm was reached at 290 °C, with a response value of 11.5 [123].

To overcome the limit of the high electron–hole pair recombination rate of n-type MOSs, which restricts their performances in gas sensors, the incorporation of additives, such as MXene and polyoxometalates (POMs), into the fabricated heterojunctions has been demonstrated to be an effective strategy. POMs, which are metal–oxygen cluster ions rich in transition elements like W, Mo, and V within a specific skeletal structure, can function as electron acceptors. They participate in the synthesis of composite materials to acquire and store electrons, thereby enhancing the gas-sensing response [97]. Taking into account the distinct energy levels of In₂O₃ and SnO₂, along with the electron capture capabilities of POMs [97], Zhang et al. designed and successfully synthesized In₂O₃@PW₁₂@SnO₂ core–middle-shell nanofibers (NFs) through a one-step coaxial electrospinning technique. The interfaces between In₂O₃/PW₁₂ and PW₁₂/SnO₂ are arranged in a tandem fashion, resulting in the formation of 1D continuously distributed tandem heterojunctions (Figure 12b). The addition of an appropriate amount of PW₁₂ electron acceptor and the tandem heterogeneous interfaces between 1D POM and MOS accelerated electrons transferred and reduced the recombination of electron–hole pairs. Consequently, the ethanol sensing performances were enhanced. The optimal sensitivity response to 100 ppm of ethanol gas was approximately 4-fold higher than that observed in the control samples (In₂O₃@SnO₂ and In₂O₃/3%PW₁₂/SnO₂ sensors) (Figure 12c) [97].

On the other hand, MXene can act as an electron doner in n-type composites. Mei et al. discovered that the work function of SnO₂ exceeds that of Ti₃C₂T_x MXene, resulting in the transfer of electrons from Ti₃C₂T_x MXene to SnO₂ upon contact. This electron transfer induced a downward bending of the SnO₂ band until Fermi level equilibrium was achieved. In the process, Schottky junctions were formed at the interface, and the Ti₃C₂T_x MXene became electron depletion centers [103]. The assembled sensor based on the n-type SnO₂/ZnO/Ti₃C₂T_x MXene composites with the optimal Ti₃C₂T_x quantity of 3 wt% exhibited a superior response value of 121.1, a faster response time of 3 s, and favorable selectivity characteristics, as compared to pure Ti₃C₂T_x or SnO₂/ZnO nanomaterials (Figure 13b,c). Note that the SnO₂/ZnO/Ti₃C₂T_x MXene composite also presented a response toward acetone, but the response was much lower under the same detection conditions. However, the syntheses of In₂O₃@PW₁₂@SnO₂ and SnO₂/ZnO/Ti₃C₂T_x both required multiple steps (Figure 12a and Figure 13a).

5. Conclusions and Outlook

As summarized and extensively discussed in this review, a substantial number of studies have been conducted on the detection of acetone and ethanol by various heterojunction materials.

To enhance the sensing performance of MOS-heterojunction-based sensing materials, various strengthening methods have been proposed, including doping (Ag, Pd, etc.), the incorporation of additives (MXene, POMs, etc.), novel morphology design, and self-doping via oxygen vacancies. The content of O_V was strongly related to the sensing performances. The heterojunction MOSs with higher O_V contents commonly presented improved resistant ability. The aforementioned strategies have significantly enhanced the sensing capabilities of MOS heterojunction composites. Nevertheless, improving the selectivity of MOS sensors remains a critical challenge that constrains their practical applications. Specifically, differentiating between ethanol and acetone poses difficulties due to their analogous chemical properties. To date, only limited research has addressed this particular issue [88].

To enhance sensitivity, the integration of multiple sensors is employed through sensor array technology [4,124]. This approach amalgamates the response patterns from various sensors to improve sensitivity toward low-concentration target compounds. Simultaneously, it enhances selectivity for complex VOCs and diminishes responses to interfering substances. Furthermore, due to its rapid response time, this technology facilitates real-time monitoring [125].

Additionally, in order to fabricate MOS heterojunctions with superior performances, current synthesis methods predominantly involve multi-step processes that allow for precise control over product structure and morphology. However, these intricate reaction steps complicate the regulation of reaction parameters. Consequently, developing a universal method that is easy to operate, controllable, and suitable for large-scale production presents a significant challenge that must be addressed for the industrial application of such sensors. The implementation of data-driven material design methodologies and the optimization of synthesis parameters may serve as crucial strategies in overcoming this issue [126,127,128,129,130].

In summary, MOS heterojunction materials utilized as gas sensors are expected to evolve in the following directions: enhancing sensitivity and selectivity while minimizing operating temperatures; advancing intelligence and integration; achieving miniaturization and portability; ensuring environmental protection and sustainability; and broadening their application domains in the future. These developments will propel continuous advancements and innovations within gas sensor technology, providing reliable and efficient monitoring solutions across fields such as environmental protection, medical health care, and industrial safety.