Gas Sensors Based on Chemi-Resistive Hybrid Functional Nanomaterials

1Monitoring and recording chemical stimulus or variations in the environment are increasingly important in future production and daily life of human health. Achieving this goal relies on the availability of high-performance sensor units that are capable of detecting gas analytes, such as volatile/semi-volatile organic compounds (VOCs/SVOCs) highly rich regarding critical information for the detection, monitoring and closed-loop control in many fields, including medicine, food industry, environmental monitoring, public security, and agricultural production.

2An ideal gas sensor requires high responsivity, good selectivity, fast response/recovery, great stability/repeatability, room-working temperature, low cost, and easy-to-fabricate for practical applications. To meet those requirements, many types of gas sensors with different transduction forms, e.g., chemi-resistor, field-effect transistor (FET), solid-state electrochemical sensor (SSES), quartz-crystal microbalance (QCM), gas capacitor, surface acoustic wave (SAW), have been well studied and developed. Among them, since the 1960s, a chemi-resistor that contains an active sensing layer bridging a pair of electrodes became a promising candidate due to its advantages including easy-to-fabricate, use of very small quantity (milligram level) active materials, wide adoption of sensitive materials, and simple sensing data, which ensure its success in certain commercialization opportunities. However, it is rare to find chemi-resistors that can meet these specific requirements.

3An emerging approach in chemi-resistors to meet these needs relies on hybrid materials, viz. materials that integrate 2+ single constituents at the nanometer or molecular level, to achieve new and/or enhanced sensing properties. In this progress report, we review the advances of the hybrid material-based gas sensors concisely and comprehensively. The hybrid materials-based chemi-resistive gas sensors are distinguished, understood, and introduced based on the generic mechanisms rather than simply dividing them as different types of combination of materials. Then, the report, in detail, focuses on the research and development (R&D) aspects of hybrid gas sensors, while presenting and discussing the sensing performances of different types of hybrid materials, and associated enhanced sensing mechanisms. Promises and challenges toward the future development of each elements are deeply thought and discussed. Critical thinking and ideas regarding the orientation of the development of hybrid material-based gas sensor in the future are also discussed.

4Chemi-resistors for gas sensing include three main processes: diffusion/molecule capture unit, surface reaction unit (including charge transfer), and charge carrier transport unit (Fig. 1a). To date, most of these sensors and/or sensor arrays utilize sensing elements that are based on single material or transduction mechanism, of which intrinsic sensing activity or additional thermal/photonic energy are usually employed as the driving force to stimulate the sensing effects of target gases (Fig. 1b). The hindrances are unavoidable at several levels: (i) not satisfying long-term stability and sensitivity of organic chemi-resistors due to the high affinity of conductive polymers (CP), such as polyaniline (PANI), polypyrrole (PPy), and polythiophene (PTh), toward volatile organic compounds (VOCs) and humidity existed in the atmosphere; (ii) high operating temperatures (usually > 200 °C), baseline drift, limited selectivity, and oxidation/decomposition of VOCs in the case of inorganic materials (especially metal oxide materials, e.g., ZnO, SnO, TiO2, SnO2)-based chemi-resistor. A reliable solution for these drawbacks is the design and utilization of new gas sensitive materials based on hybrid inorganic–inorganic, organic–organic, and inorganic–organic materials.

5Using hybrid materials as sensitive transducer offers several obvious advantages, compared with the single constituent. First, the inexhaustible abundance of hybrid materials (in both the complex constituents and novel nanostructures) makes it possible to involve an almost infinite continuum of variable factors (surface-dependent factor, interface-dependent factor, and structure-dependent factor) to generate new sensing behaviors (Fig. 1c). Second, with hybrid material, more chemical/physical processes with different enhanced mechanisms could be introduced to precisely design, regulate, and enhance the sensing performance mainly through catalytic reaction with analyte, charge transfer, charge carrier transport manipulation/construction of heterojunctions, molecular binding/sieving, and their combinations.

6Hybrid materials can perform improved sensing characteristics via one or a combination from five typical hybridizing forms which are categorized into three sensing-dependent factors (Fig. 2a). The first combination relies on catalysis reactions (normally noble metal catalysts, e.g., Pt, Pd, Au, and Ag) between analyte gas and decorated catalysts on host semi-conductive materials (categorized as surface-dependent factor). The second relies on a fast charge transfer process, viz carrier withdrawal or donation, electron acceptor or acceptor between guest additives and the host material (e.g., carbon nanotubes (CNTs)), reduced graphene oxide (rGO) (categorized as interface-dependent factor). The third relies on regulating the charge carrier transport in a conductive/semi-conductive materials (e.g., single-wall carbon nanotube (SWCNT)-metallo-supramolecular polymer (MSP), gold nanoparticles (GNPs)-thiols, N,N’-diphenyl perylene tetracarboxylic diimide (PTCDI-Ph)/para-sexiphenyl (p-6P)) upon exposure to gas analytes (categorized as interface- and structure-dependent factor). The fourth relies on manipulation/construction of the heterojunctions such as n–n, p–n, p–p, p–n–p heterogeneous semi-conductive materials (categorized as interface-dependent factor). The last one relies on semiconductors coated by gas molecular sieving/binding layers or ligands/complexes for selective gas detection (categorized as surface- and structure-dependent factor). In the following section, we provide more details on each of these combinations. It should be noted that, according to the understanding of authors on chemi-resistors (1. measuring the resistors directly; 2. measuring the current when the device is applied a constant bias voltage; 3. measuring the partial voltage on the device in parallel with a constant resistance when the resistance and device is applied a constant bias voltage), we will review here only resistive-change sensing devices in which the contact of sensitive materials and electrodes is ohmic (good linearity of I–V curves under DC bias, Fig. 2b); therefore, we exclude devices in which the I–V curves under DC bias are nonlinear despite that they exhibit similar resistance changes (e.g., chemical diodes, proton/ions types).

7Catalytic effects of hybrid functional nanomaterials contribute to high response, fast speed, and low operating temperature via chemical/electronic sensitization, which is usually accompanied by synergistic effects, complementary behavior, and porous structures. In addition, the exposed facets (morphologies) of matrix nanograins (facet-dependent chemical activity) and catalytic additives can greatly enhance the sensing properties of hybrid nanomaterials. For example, introduction of Cr dopants to WO3 polydedra can not only control the specific exposed facets and activation energy, but bring catalytic effects to the matrix. The combining effects led to improved sensitivity and reduced operating temperature. Further hybridization with catalytic Au nanocrystals—to form Au/Cr–WO3 hybrids—contributed to high sensitivity, fast speed, and reduced working temperature to acetone and benzene due to Au/Cr co-catalysts-enhanced surface reaction (Fig. 3a, b). The advantages of co-catalysts can improve even further the hybrid materials. A recent example of this approach is Pd/Sb nanocrystals modification of SnO2 that Sb and Pd functioned as anti-humidity and catalytic sites, respectively, which remarkably reduced humidity interference and improved responses toward H2 (Fig. 3c).

8The catalytic effects of loaded catalysts on host-sensitive materials are associated with the contact between catalyst and gas. A gas diffusion-favoured structure can provide additional exposed surface areas and fast speed via a combination of surface- and structure-dependent factors. The introduction of catalytic Ag NCs via e-beam evaporation and calcination into quasi-1D heterostructures significantly enhances the response and selectivity to ethanol (Fig. 3d). For even further performance, quasi-1D nanostructures with both porosity and sensitive nanobuilding blocks, namely mesoporous 1D nanofibers/tubes (meso-NF/NTs), have been reported. By introducing sacrificial polymeric colloids and protein-templated catalysts to the solutions, meso- and macro-porous Pt-decorated SnO2 NTs have been fabricated by electrospinning and sintering in sequence (Fig. 3e, f). The combined effects of porous nanostructures, fully depleted sensing areas and uniformly distributed Pt nanocatalysts on SnO2 NTs allow a highly selective detection of acetone (R5ppm = 192). Similarly, bimetallic PtM (M = Pd, Rh, and Ni) catalysts can be introduced to meso-WO3 NFs that are then highly selective detectors of acetone and H2S gas. Sensors array combined with pattern recognition methods (so-called e-nose) based on three different PtM-decorated meso-WO3 NFs can accurately detect and discriminate the breath of a simulated biomarker through principal component analysis (PCA, Fig. 3g).

9On the basis of intensive works on porous 2D ZnO nanostructures, Pd NCs have been deposited on porous 2D ZnO nanoplates (host materials) transformed by Zn5(CO3)2(OH)6 nanoplates to form 2D Pd/ZnO hybrid nanoplates (Fig. 3h); these acquire enhanced sensing properties. Similarly, catalytic Pt can be used to decorate the surface of BP, which enables RT detection of H2 by Pt/BP at RT (Fig. 3i). When the structure of the host materials is further upgraded to 3D hierarchical porous (hp) nanostructures, a good gas diffusion platform is obtained with a large loading area of catalysts. By taking the advantage of the opals/polymeric beads or mesoporous silica/carbon/polymers templating method, hp-MOX thin films (3D hp nanostructures) with certain additives (catalysts) have been developed. In this simple approach, hp-SnO2-inverted opal thin films loaded with mono-dispersed Pt catalyst (of uniform size of ~5 nm) were prepared (Fig. 4a). The improved sensing responses of Pt-doped SnO2-inverted opal thin films were achieved due to increased porosity, electronic sensitization, and synergism (Fig. 4b). Similarly, hp–Pt–WO3 or hp–Cr–WO3-inverted opal thin films have been successfully prepared (Fig. 4c). Due to the catalytic activation of N–H bond dissociation and effective gas diffusion within the macro-porous structures, excellent NH3 responses were obtained for the hp–Pt–WO3 sensor. As another example, well-controlled self-assembly of block copolymers such as poly(ethylene oxide)-blockpolystyrene (PEO–b–PS) could generate a perfect template with a highly ordered structure. Mesoporous WO3/Pt with a highly ordered and porous structure (inset of Fig. 4e) could be obtained by using this template and a 2-step pyrolysis process (Fig. 4d, including first treated in inert atmosphere and finally calcinated in air). Contributed to high surface areas (112–128 m2 g−1), large pore size (13 nm), and well-dispersed catalytic Pt NCs (~ 4 nm), the WO3/Pt-0.5 sensor had the highest response and fastest response-recovery speeds (Fig. 4e). Another category of nanostructures of host materials is multi-shell, yolk–shell, and multi-yolk–shell with hollow nanochamber. With the hollow host nanochamber (SnO2) loading catalytic Pd through spray pyrolysis (Fig. 5a), and showing the formation of double-shelled Pd–SnO2@Pd–SnO2@Pd–SnO2 yolk–shell spheres (Fig. 5b), both sensitivity and selectivity were enhanced due to the unique hierarchical porous structure and uniformly exposed Pd catalysts. The representative works are summarized in Table 1.

10As aforementioned, greatly improved sensitivity and speed have been realized by well design on the dispersion of catalysts/co-catalysts and gas diffusion-favored structures of hybrid gas sensors based on catalytic effects. The remaining problems of operating temperature and selectivity might be resolved by further combination with charge transfer (Sect. 3.2) and molecule sieving layers (Sect. 3.5), respectively. Moreover, the newly developed single atomic metal and/or metal cluster-based catalysts with better catalytic effect might bring new understandings and chance in such areas.

11Charge transfer happens between decorations and the host materials (good conductivity), which could vary the conductivity of the hybrid materials. This process improves sensitivity to the analysts at low temperature or even at room temperature (RT), accompanying fast response and recovery properties. Discrete and uniform SnO2 NCs-decorated multi-walled CNTs (MWCNTs) (Fig. 6a) gave high performances (response of ~ 180% to 100 ppm of NO2) at RT due to the abundance of active sites and easy electron transfer under the assistance of the well-matched work functions of SnO2 and MWCNTs. Liu et al. used rGO instead of the carbon nanotubes. This researcher successfully synthesized SnO2 QDs/rGO hybrids by a one-step solvent thermal reaction at 180 °C (oleic acid and oleylamine as capping agents) (Fig. 6b). Due to co-effects of excellent gas adsorption of QDs, effective charge transfer between SnO2–rGO interfaces and the superb transport capability of rGO, the sensor responses in 2 s with fully recovery properties upon exposure to 33 to 50 ppm of H2S at RT (Fig. 6c). Meanwhile, the SnO2/rGO-based sensor showed an obvious enhanced response compared with the responses of pure SnO2- or rGO-based sensors toward H2S at 22 °C, in which the rGO acted as a host transducer material. By combining both advantages, 2D MoS2 sheets were hybridized with 2D graphene to form rGO/MoS2 aerogel with large surface areas, porous structure, and high electrical conductivity (Fig. 6d). Efficient and rapid charge transfer across the interface ensured enhancement and fast detection of NO2 than bare rGO or MoS2 (Fig. 6e). Ascribing to the high specific surface area of porous Cu2O nanowires networks and improved conductivity via effective charge transfer, rGO–Cu2O mesocrystals had much higher sensitivity to NO2 at RT, surpassing the performance of stand-alone systems of Cu2O and rGO sheets (Fig. 6f).

12CPs could also be applied in charge transfer hybrids by replacing the inorganic components. For instance, graphene was combined with PANI to form a hybrid thin film that had improved, reversible, and stable NH3 sensing (Fig. 6g, h). The fast electron transfer between hybrids and NH3, assisted by π–π interactions of PANI and rGO with low electron transfer energy barrier, led to more electrons transfer from PANI to rGO; this effectively improved the responsivity and response time (Fig. 6i). Up to date, the detecting gases are limited to strong reducing/oxidizing molecules such as NO2, NH3, and H2S, which hinders the widely application of such hybrid gas sensors. The critical points to overcome this shortcoming may be depicted as: (1) the chemical/electronic modification of the reported charge transfer based hybrids to improve the sensitivity and expand the types of detectable gases (i.e., Pt–SnO2/rGO, details in Sect. 3.6); (2) the development on the new candidate of chemi-resistive decorations with desired absorption–desorption process and well-tailored energy level/energy band gap structure, which always shows low thermal activation energy (< 0.5 eV, i.e., electronic conductive metal–organic frameworks (EC-MOFs), for details see Sect. 3.5). The representative works are summarized in Table 2.

13Different from charge transfer that simply uses high charge transport capability of the highly conductive component, hybrid gas sensors based on regulation of charge carrier transport can manipulate the sensing properties by changing carrier concentrations, transportation mode, and/or pathways of charge transport.

14A simple and effective way of enhancing responsivity relies on controlling the charge transport by tuning carrier concentrations. For example, a PTCDI-Ph/p-6P ultrathin film was fabricated with a thickness of only 6 nm, of which 5 nm was attributed to (p-6p/p type) asymmetric thickness and 1 nm to (PTCDI-Ph/n type) (Fig. 7a). Electrons in the PTCDI-Ph were deprived by NO2, which simultaneously released the restricted hole in p-6p, and thus influenced the transportation of p-6p; this generated a NO2 sensing signal at RT (Fig. 7b). Chi et al. thermal-deposited a high-quality crystalline terrace-like TIPS-pentacene film on p-6P that can easily be positively charged (Fig. 7c). The efficient charge transport ability and low original carrier concentration gave superb NO2 sensing in terms of both response/recovery speed (Fig. 7d) and responsivity/sensitivity (Fig. 7e). Impressively, when the transport direction of charge carriers changes from horizontal to vertical in a vertical diode (containing top/down electrodes and VOPc/F16CuPc layers, Fig. 8a), the sensor responded remarkably well to 0.5–5 ppm NO2 at RT (Fig. 8c), with an acceptable sensing stability (Fig. 8b) and wide linear working region. More interestingly, when the voltage bias is raised from 0.2 to 1.5 V, the sensing ability weakened dramatically (Fig. 8d), being ascribed to the transportation change from ohmic to space charge limited current (SCLC) mode (Fig. 8e), which may give guidance as to how to choose the bias to control the charge transportation for gas sensors to get them to work under the best conditions.

15Another strategy to manipulate the properties of sensors based on charge transport is by regulating the conductive pathways of conductive-insulate hybrid materials via physical cracks, chemical bindings, or the phase of the component. More specifically, this approach relies indeed on regulating the electron hopping barrier, the interspace between conductive materials, phase-change or their combination. Insulating polymers can be combined with sensitive inorganic materials to fabricate highly sensitive and selective chemi-resistors (Fig. 9a). Hybrid thin films as sandwiched PMMA/Pd/PMMA (PMMA = poly(methyl methacrylate)) were prepared on a flexible substrate using sputtering and spin coating in sequence (Fig. 9a). Hybrid thin films with nanogaps formed by 25% mechanical stretching have very selectivity and sensitivity in detecting H2 against O2, ascribed to the selective penetration of H2 in PMMA membranes and the density reduction of the cracks formed in the trilayer of the hybrid thin films (Fig. 9b). Adoption of similar principles, but with higher effects, relies on films of GNPs coated with monolayers of thiols. These structures were a good solution for VOCs sensing due to the swelling and shrinkage of molecular chains interacting with VOCs (Fig. 10a). Assembly of the GNP-based chemi-resistors with a wide variety of functional groups creates sensor arrays with different resistances that can be further varied after interacting with VOCs. The transport of electrons, expressed in electrical resistance, can be dually regulated by controlling the interspaces between GNPs after applying strain to the GNPs-based film, which further influence sensitivities (Fig. 10b). CNTs are also used as the host with the surface coverage of molecular (MSPs) (inset of Fig. 10c). Such MSPs could create sensory devices with a dosimetric (time- and concentration-integrated) increase in electrical conductivity triggered by electrophilic chemical substances (Fig. 10c).

16In summary, for cases where the resistance decreased upon exposure to target gas, the depression of the off current via carrier concentrations reduction, transportation mode changes and/or physical cracks might be the most effective way to realize high sensitivity. Simultaneously, as presented above, the component used for the controlling charge transport of host materials can further contribute to improved speed, long-term stability, and excellent selectivity. The representative works are summarized in Table 3.

17Heterojunction is defined as the interface between two dissimilar semiconductors (one is the host, and the other one is the guest) that form a junction (n–n, p–p, p–n) linked with energy band structure due to the alignment of their fermi level. Notably, although the broad definition of heterojunctions covers all types of composites forming a junction in the interface, it is not clear enough for the well understanding of the complicated sensing mechanisms of composites sensing materials. Therefore, in this work, the narrow definition of manipulating/constructing heterojunctions is used, which excludes cases of catalytic effects, charge transfer, and charge carrier transport. According to the definition, the junction changes the interface potential energy barriers and regulates the transfer and/or injection of electrons and holes in a precise manner when it interacts with gas analytes. For example, n–n heterojunctions made of In2O3 hollow spheres (acetone-sensitive host) coated with CeO2 nanoparticles (humidity-sensitive guest) were synthesized and characterized as a chemi-resistive film. Exposing the layer to various gas analytes has shown selective detection of acetone in the presence of water, taking advantage of the chemical interaction between CeO2, In2O3, and water vapor, which greatly reduces the interfering effects of humidity (Fig. 11a). By modulating interface potential energy barrier between n–n junctions, as in the case of Fe2O3/TiO2 tube-like quasi-1D nanostructures synthesized through a multi-step hydrolysis (Ostwald ripening & thermal reduction), the corresponding sensing performance could be greatly improved (Fig. 11b). Combining modulation of electron transfer over the energy barrier at the perfect SnO2/ZnO heterojunction—fabricated by atomic layer deposition—and UV light generated electron–hole pairs, the sensitivity to NO2 could be improved using the SnO2/ZnO core-sheath nanowires (Fig. 11c). By introducing narrow band gap into the junction, such as in the case of In2O3 NCs to ZnO, a good response at visible-light conditions to gas analytes (e.g., formaldehyde) at RT (R100 ppm = 419%) was attainable.

18The unique morphology (good compatibility with the devices), nanoscale thickness, and high surface area of 2D nanostructures make them promising as the host materials for chemi-resistive gas sensors. Hybrids of SnO2 NCs-decorated MoS2 nanosheet (MoS2/SnO2) were synthesized via hydrolysis-pyrolysis processes (Fig. 11d) with air stability. The SnO2 NCs not only enhanced the stability of MoS2 nanosheets in dry air, but served as strong dopants for MoS2, leading to the changes of conduction channels in the MoS2 nanosheets (Fig. 11e). For further improvements in the sensing performance, introduction of porosity, such as in the case of WO3 lamella-based films loaded with mono-dispersed SnO2 QDs (~ 4 nm) (Fig. 11f), could reach high levels. Experimental results show that the porous lamellar-structured WO3–SnO2 hybrid films could achieve high response to NO2 gas, ascribing the effective insertion of QDs into lamella stacks as a strong electronic sensitization.

19Compared with n–n heterojunctions, p–n heterojunctions provide a stronger manipulation on interface potential energy barriers, build-in electric field and additional catalytic effects in some unique cases. For example, exposing p–CuO nanoparticles loaded on CuO–SnO2 p–n nanowires to H2S transformed it to highly conductive CuS (Fig. 12a), resulting in depleted region change (the p–n junction breakup) and second-order effects (the oxidation of H2S by absorbed oxygen) after p-CuO was reversibly generated and removed on the SnO2 surface. Without generating new chemical compounds, simply tuning the thickness of in situ oxidation layer, rich Te–Te or TeO2/TeO2 bridging point contacts and additional p–n heterojunctions (Te/SnO2) contributed to further excellent sensing performances (to CO and NO2) of the brush-like heterostructures (Fig. 12b). The nanorods of p-type coating layer can be replaced by continuous layer to form core-sheath hybrids processing radial modulation of potential energy barriers, for instance, both n-ZnO/p-CoPc (cobalt phthalocyanines, Fig. 12c) and n-SnO2/p-Cu2O (Fig. 12d). Core-sheath NRs have better sensitivity of the target gases.

20Conductive polymers (CPs, e.g., PPy, poly(3,4-ethylenedioxythiophene) (PEDOT), PANI) that are p-type components of diverse types of p-n heterojunctions can work at RT or low operating temperature with different working principles. First, PPy-ZnSnO4, p–n hybrid nanoparticles, can enhance the NH3 sensing performance (3–4 times higher) compared with pure PPy and ZnSnO4 (Fig. 13a). The concentration of NH3 can be quantitatively detected (Fig. 13b) with shorter time of response (26 s) and recovery (24 s) (Fig. 13c). The overall improved performance has been ascribed to the p–n heterojunction, in which the holes at high concentration in PPy and the electrons in Zn2SnO4 diffuse into each other to form a built-in electric field of a depletion layer (Fig. 13d). Interaction between Zn2SnO4–PPy and NH3 broadens the depletion layer, which determines the response, and the speed of response/recovery. When the p–n junction was reinforced as dual p–n junctions (p–n–p) in the hybrids of the hollow In2O3 nanofibers (NFs) and PANI (Fig. 13e), the performances were further enhanced (Fig. 13f). For CP-based chemi-resistive heterojunctions, unsatisfied sensitivity (response) and long-term stability will be two challenging issues for researchers.

21Obvious advantages of heterojunction-type chemi-resistive hybrids-based gas sensor can be summarized as: (1) higher sensitivity due to manipulations of the potential energy barrier formed by band bending of different components (e.g., Fe2O3/TiO2 tube-like quasi-1D nanostructures (n–n), n-ZnO/p-CoPc, and n-SnO2/p-Cu2O core-sheath NRs); (2) improved selectivity to some gases (e.g., CuO–SnO2 p–n nanowires to H2S); (3) promising, although limited so far, to anti-interference gas (e.g., CeO2–In2O3 hollow spheres with anti-humidity properties); (4) avoiding UV-introduced ozone and performance degradation (for example, when narrow-band guest material hybrids with semi-conductive host materials, e.g., CdS-ZnO, ZnO-CdS, CdSe-ZnO, the room operation temperature can be achieved by the visible-light-driven gas sensing). The representative works are summarized in Table 4.

22In summary, the critical points to achieve better performance in heterojunction-based gas sensor are depicted as: (1) maximum effective contact areas of the interfaces via surface and structure design; (2) matched band structure to facilitate the manipulation of potential energy barrier; (3) additional capability of catalysis of guest additives to host materials; (4) visible-light-driven photocatalytic abilities and good charge carriers separations for light-driven n–n or n–p hybrids; (5) selectivity-improvement-purposed heterojunction design based on specific interaction (e.g., Pd–H2, CuO–H2S, PPy–NH3) or n–p response-type reversion (Co3O4–SnO2 p–n junctions for H2).

23Functionalization, coating or doping in/on the sensing materials, i.e., introduction of the sensing probe, was demonstrated as an effective way of improving the selectivity and specificity through a one-lock one-key binding or structure similarity-based combination. Both the inorganic and organic probes have been well developed. The sieving of interferon, especially the humidity in the environment, is an alternative way of improving specificity.

24A 3D sulfonated rGO hydrogel (S-RGOH)-based gas sensor combining chemical functionalization and porous structures was synthesized in a one-step hydrothermal reaction (Fig. 14a). Addition of a NaHSO3 probe dramatically enhanced the response (Fig. 14c) of NO2 with fast recovery (Fig. 14b), assisted by the porous structures of the graphene host. Self-assembled monolayers (SAMs) with suitable alignment of the gas–SAM frontier molecular orbitals (Fig. 14e) with respect to the SAM–NW Fermi level (Fig. 14f); this led to high selectivity and sensitivity to analyte gas. SnO2 NWs were modified with amine-terminated SAM and applied as light-driven chemi-resistors working at RT, achieving good NO2 sensing performance, the schematic mechanism of which can be found in Fig. 14d. This concept was extended to porous MOX nanostructures for further enhancements of their sensing properties. APTES-modified porous WO3 nanotubes (P-WO3 NTs (10%)@APTES) performed the best sensitivity and selectivity (Fig. 14a), which can be ascribed to the large surface area and high gas diffusion rate provided by P-WO3, and selective reaction between NO2 and surface SAM with APTES (Fig. 15b, c). The existence of SAM on the surface of inorganic materials (except 2D nanomaterials) limits the working temperature, which greatly weakens the sensing performance, although it could be resolved by UV irradiation. Using conductive polymer as the host material with surface SAM functionalization by the “1-stone 2 birds” strategy was promising and novel (Fig. 15d, e). Superb sensing performances were achieved by combining RT sensitivity of CP and good selectivity of SAM (Fig. 15f).

25Figure 16a shows the low cross-sensitivity to humidity and other interferon gases by refreshing the regenerative surface involving the interaction between facile redox fair (Tb3+/Tb4+) and surface OH group (or water vapor) on SnO2. This 5 Tb–SnO2-based chemi-resistor achieved high response to acetone exposure. The oleic acid SAM also was effective in screening the effect of humidity (< 350 ppm) when it was layered on PANI surface (Fig. 16b, c). Although it is not enough for practical application, this demonstration is still valuable in pointing to a promising way to eliminate the interferon of humidity.

26Recently, MOF materials are great opportunity in generating sub-nanometer or nanometer pores with high uniformity. Neat MOFs chemi-resistors were prepared based on hydrophobic MOF (ZIF-67), which showed selective response to VOCs, with slightly interfering effects of humidity (Fig. 17a). ZnO@ZIF-8 core-sheath NRs powders were synthesized by hydrothermal reaction using a self-template strategy (Fig. 17b). The chemi-resistive gas sensor based on the thick film of ZnO@ZIF-8 hybrids had satisfactory sensitivity and response time to 100 ppm formaldehyde, even under interfering humidity (Fig. 17c). Mixing the CP with molecularly imprinted polymer (MIP) molecular (Fig. 17d) was another effective approach to improve not only the responsivity, but selectivity (Fig. 17f). Responses of interferon (2,4,6-TNT) and analyte (2,4-DNT) were suppressed and enhanced, respectively, although they are of very similar molecular structure and functional group. However, the speed of response and recovery showed no obvious change (Fig. 17e). Instead of using functional MOFs as filter film coating on sensing materials to provide additional selectivity and/or sensitization, EC-MOFs are novel emerging materials with regular porosity and conductivity, which are promising for chemi-resistors with high sensitivity and selectivity. Unlike MOX and MOX-MOFs, which still need additional thermal or photonic energy as the trigger source to activate the sensing reaction, EC-MOFs can be directly used as sensitive materials based on their regular micro-porosity, selective frameworks, high electronic conductivity, and RT activity. Therefore, EC-MOFs are promising components for hybrid gas sensors and will be powerful competitors for the new generation of gas sensors.

27As mentioned above, the introduction of molecular probing and sieving effect can effectively overcome the poor selectivity problem of chemi-resistors. Up to date, only a few organic/inorganic probe or porous materials with molecule sieving effects have been applied to chemi-resistors to realize simple guest–matrix interaction (e.g., -NH2 group with NO2, –NO2 group with NH3, NaHSO3 with NO2) or molecule rejection (e.g., anti-humidity, gas molecules with large kinetic diameter). Such cases are ultra-small fraction of the state-of-art gas molecule adsorption and separation areas. More guest–matrix interactions (e.g., van der Waals interactions, hydrogen bond, π–π interactions, weak acid–base interactions) and gas separation design (e.g., channel traffic effects, framework flexibility), that have been well studied on rGO, polymers (e.g., metal-induced ordered microporous polymers (MMPs), covalent-organic frameworks (COFs), conjugated mesoporous/microporous polymers (CMPs)), MOFs, can be introduced to chemi-resistors for advanced sensing performances. The representative works are summarized in Table 5.

28In many cases, multi-forms working on the hybrid materials can simultaneously and dramatically improve sensitivity and responsivity. When the heterojunction barrier (the SnO2/rGO heterostructure interface) was combined with catalytic Pt, sensitivity of the SnO2/rGO hybrids to H2 was greatly enhanced (Fig. 18a–e). Hydrogen ranging from 0.5 to 3% in air could be quantitatively detected at near RT with response and recovery times of 3–7 and 2–6 s, respectively. Furthermore, when the catalytic effect was co-working with the p-n heterojunctions and porous structure (Fig. 19a), Co3O4–PdO loaded on n-SnO2 hybrid hollow nanocubes (Fig. 19b) reached selectivity and response amplitudes for the detection of acetone superior to those MOF-derived metal oxide sensing layers previously reported. Accordingly, the sensor arrays (Co3O4-loaded n-SnO2 HNCs and Co3O4–PdO-loaded n-SnO2 HNCs) can clearly distinguish 1 ppm humid acetone molecules among the seven interfering analytes (Fig. 19c). The reason is electron migration from n-SnO2 to PdO or p-Co3O4 in the multi-junctions significantly influencing the electron depletion regions, which leads to the superb sensitivity (Fig. 19d, e).

29When the catalytic effect co-worked with the molecular sieving effect, improved selectivity (anti-interferon) with enhanced sensitivity can be achieved by coating a layer of hydrophobic and thermally catalytic bimetallic ZIF-CoZn thin film on ZnO to form core-sheath MOX@MOFs nanowire arrays (NWAs) (Fig. 20a). The bimetallic ZIF-CoZn MOF sheathes gave good thermal stability (ZIF-8(Zn)) and excellent thermal catalytic ability on ZnO (ZIF-67(Co)), as well as hydrophobic channels. By combining their advantages, the ZnO@ZIF-CoZn preparation showed greatly enhanced performance on selectivity (good anti-humidity, Fig. 20b) and also on its response, response and recovery behavior and working temperature (Fig. 20c). More complicated hybrid nanostructures containing MOX, plasmonic/catalytic NMs, and hydrophobic MOFs, i.e., the dual-functional Au@ZnO@ZIF-8 Janus structure (Fig. 20d, e), have been fabricated. Au@ZnO@ZIF-8 hybrids had enhanced selective adsorption, detection and oxidation of HCHO and prevented interference from gases such as H2O and toluene (Fig. 20f), where Au NRs helped to generate charge carriers on a ZnO surface under visible-light irradiation. The representative works are summarized in Table 6.

30The current progress report reviews advances and the advantages of the chemi-resistive hybrid nanomaterials compared with the single constituent, according to five main sensing mechanisms: manipulating/constructing heterojunctions, catalytic reaction, charge transfer, charge transport, molecular binding/sieving, and their combinations. Table 7 lists typical chemi-resistive materials for hybrid gas sensors categorized by types of materials and conductivity.

31As the first commercial gas sensor, metal oxide-based chemi-resistors still occupy a leading role in both fundamental researches and commercial devices. Various commercial chemi-resistive gas sensors based on single or hybrid materials have been developed for the detection toward target gases (toxic, flammable, VOCs, explosive, H2, etc.) ranging from sub-ppm to saturated vapor, which are widely used in fields including environment monitoring, medical care, food industry, agriculture production, and public security. The versatile commercial chemi-resistive gas sensors are introduced but not limited as follows.

32Some e-nose systems comprised of chemi-resistive sensor arrays have succeeded in the application of medical care. Commercial e-nose PEN3 (Airsense Analytics GmbH, Schwerin, Germany) made up of a gas sampling unit and a sensor array (10 different metal oxide thick film sensors (MOS)) can screen colorectal cancer (CRC) and polyps. Another 14 commercial gas sensor-integrated e-nose system could generate characteristic “breath fingerprints” by exhalation components and could diagnose the lung disease through pattern recognition of a “breath fingerprint.” Those sensors categorized as MOS, hot wire gas, catalytic combustion gas, and electrochemical gas sensors are produced by the manufacturers, Hanwei (Fig. 21a), Figaro (Fig. 21b), Winsen, Nemoto and Alphasense. Aeonose in (Fig. 21c) is a CE-certified, handheld, and battery-powered e-nose device designed by a Zutphen Company in Netherlands. The aeonose comprises three micro-hotplate metal oxide sensors and a pump to detect gastric cancer from exhaled breath. Sunshine Haick Ltd. have successfully designed the sensor arrays to diagnose lung and gastric cancer via pattern analysis of exhaled VOCs, which has made great and has a perfect perspective. Other representative commercial e-nose in clinical diagnosis of complex regional pain, diabetics, head and neck cancer, dyskinesia, and prostate cancer are summarized in Table 8.

33Commercial e-nose has acted as an indispensable instrument for the rapid, accurate, and overall-process assessments of food health and quality aim at adulterated counterpart, contamination and spoilage. An PEN-2 e-nose (WMA Airsense Analysentechnik GmbH, Schwerin, Germany) composed of 10 different metal oxide sensors was utilized to monitor the adulteration of milk with water or reconstituted milk powder. Also, the PEN-2 is used to monitoring the change in volatile production of mandarin during different picking-date. Meanwhile, PEN-2 is used to characterize espresso coffees brewed with different thermal profiles. MOS sensors manufactured by Figaro (Figaro Inc., Japan) were used to recognize odors emitted from different stages in a waste water treatment plant. Tagushi gas sensor based on metal oxide semiconductor from Figaro Engineering Inc. is used to classify the tea aroma. Cyranose 320 in Fig. 21e that consists of an array of 32 thin-film carbon-black conducting polymer sensors was used to identify odor emitted from dairy operations. The portable e-nose based on thin-film semiconductor (SnO2) sensors (Hanwei Sensors) in Fig. 21f can perform early detection of wine spoilage thresholds in routine tasks of wine quality control. An e-nose system (Fig. 21g) was used to detect detergent powder in raw milk. Representative applications are summarized in Table 9.

34The application of commercial e-nose to monitor volatile compounds in the environment both indoor and outdoor provides a reliable solution. Single semiconductor gas sensor GGS 10331 (produced by Umwelt Sensor Technik) was made with a semiconductor sensing layer on Al2O3 substrate to predict the concentration of ammonia under humidity interference. Tagushi (TGS) gas sensor (Figaro Engineering Inc.) was applied to detect NH3, CO, H2, C2H6O, C4H10, C3H8, CH4, alcohol, and solvent vapors and the accuracy was 100%. MQ-7 (Fig. 21j) is a commercial electronic nose for monitoring CO. Portable electronic noses in Fig. 21d were used to classify pollutants in water. Similarly, the commercial e-nose is widely used to identify the toxic wastes, soil/water pollution, indoor volatile organic compounds, etc. Table 10 summarizes the recent applications of e-nose for monitoring environment.

35E-nose is widely used in agricultural to analyze growth, classify seeds, detect the maturity, monitor quality, which promoted agricultural modernization and saved labor. Eight MOS sensors produced by FIS (Osaka, Japan), MQ (Hanwei, China), and TGS (Figaro Engineering Inc.) were applied for classifying cumin, caraway, and other seeds. Similarly, e-nose based on MOS TGS and FIS sensors (Fig. 21h) were distinguished Iranian Rosa damascena essential oils. An e-nose FOX 4000 (Alpha MOS, Toulouse, France) was chosen to analyze ginseng at different stages. An e-nose FOX 3000 (Fig. 21i) was applied to characterize and classify seven Chinese robusta coffee cultivars. Commercially available chemical sensors intended for agriculture are summarized in Table 11.

36Some commercial e-noses are attempted to detect explosives with ultra-low saturated concentration (parts-per-billion or below). Figaro Engineering Inc. produce an e-nose comprised of eight MOS sensors to discriminate and quantify different chemical warfare agents mimics. More expectations in applications are also possible in the future.

37Although excellent advances in both e-nose system and chemi-resistive sensory unit have been reached in the field during the last few years, there is still room for improvements. Below is a summary of the main rules for improving the performance of hybrid material-based gas sensors (details see the summary paragraph of each section):

38For sensors based on heterojunctions (potential energy barrier manipulation), the more uniform and the larger the contact area of heterojunctions and charge transfer hybrids, the higher the response, resulting in faster speed/lower operating temperature, e.g., Fe2O3/TiO2 tube-like quasi-1D nanostructures (n–n), n-ZnO/p-CoPc, CeO2–In2O3 hollow spheres with anti-humidity properties, and CdS–ZnO.

39For sensors based on catalytic effect assistance, the higher the potential energy barrier tuning, the higher the response, e.g., Pd/Sb–SnO2.

40For sensors based on charge transfer, the more dispersion uniformity and the smaller size of catalysts on host-sensing materials, the higher the response, giving faster speed and lower operating temperatures, e.g., SnO2 QDs/rGO hybrids, rGO/MoS2 aerogel, and PANI/rGO.

41For sensors based on regulation of charge carrier transport, the thinner and the more defect-rich of the hybrid film (e.g., suppression of original gas-off current in current-increased gas sensor), the higher and faster the responses obtained, e.g., PTCDI-Ph/p-6P ultrathin film, sandwiched PMMA/Pd/PMMA, and MSP-covered CNTs.

42For sensors based on molecular binding/sieving, the more selective and uniform dispersion of molecular binding/sieving guests, the higher the selectivity, e.g., APTES-modified porous WO3 nanotubes, the oleic acid SAM-modified PANI, the ZnO@ZIF-CoZn NWAs P-O3 NTs (10%)@APTES.

43Improving the performance requires better understanding of the mechanism. Recently, most sensing mechanism represented in the research articles is “possible mechanism” based on the results of comparative tests instead of direct observations. Exactly, the latter one is more trustable and solid results to support the mechanism study, e.g., in situ FTIR, in situ Kelvin probe, in situ STM, in situ TEM). In addition, Theoretically studies (such as DFT simulation), are also helpful for researchers to understand the interaction between the gas analyte and sensitive materials, the succedent electronic structure changes, or band gap regulation in heterojunction, or charge transfer, etc., which can guide the orientation of materials design. Otherwise, learning theoretical studies toward hybrid catalyst designs can inspire the further researches on hybrid gas sensing due to the similar surface physical/chemical science, band gap theories, and charge transfer process.

44Controlling the fluidic behavior of gas, enhancing the anti-interferon ability by loading novel porous sieving materials (e.g., MOF, COF), screening the cross-talk (such as deformation) by special micro-/nanostructures, deeply mining the features of sensing signal (e.g., response, area of peak, and speed), and enhancing catalysis effect using small NPs, clusters, or even single-atom catalyst are the long-term challenges of hybrid gas-sensing materials to adapt the applications under real-world conditions. The advances in knowledge in all our endeavors can be a foundation and useful experience for sensing technology, surface science, catalysis, fluidic mechanics, and microelectronics.