Room Temperature Wearable Gas Sensors for Fabrication and Applications

The recent surge in demand for human–machine interaction (HMI), Internet of Things (IoTs), and artificial intelligence (AI) has created both opportunities and challenges for room‐temperature wearable gas sensors. These sensors serve as a source of perceptual information and can be easily integrated into wearable electronic devices due to their portability and miniaturization. In recent years, various types of wearable room temperature gas sensors have been developed for fields like environmental monitoring, healthcare, smart home, industrial security, food safety monitoring, and public security. These sensors not only adjust to the movements of human effortlessly but also have reduced power consumption. Therefore, room temperature wearable gas sensors hold great promise for the development of integrated intelligent gas sensing system worn on the human body. These sensors can be fabricated using various sensing materials to detect diverse target gases. This review provides a comprehensive summary of the preparation of sensing materials with extraordinary sensing capabilities at room temperature. Additionally, the article includes a brief discussion of the sensing mechanism, employing four models to explain it: oxygen adsorption, direct electron transfer, proton transfer, and ions conduction. Finally, this article discusses the various applications and future perspectives of room‐temperature wearable gas sensors.


Introduction
In recent years, the explosive development of the information era, Internet of Things (IoTs), Artificial Intelligence (AI), big Ref.[32]   rGO/MoS 1 Se 1 NO 2 1.2% [10 ppm] 58s/68s 1 ppm Ref. [226]   NiO@CuO PPTA NH 3 80% [100 ppm] 72.5s/30s 46.5 ppb Ref. [227]   MgO@CeO 2 PET LPG 32 [900 ppm] Ref. [228]   GO-PANIHs PET NH 3 10.5 [10 ppm] 102s/186s 1.39 ppb Ref. [229]   WS 2 -GONRs PI ethanol 126.5 [1 ppm] 43s/29s 1 ppm Ref. [230]   Cu-ZnO Acetone 450s/100s 1 ppm Ref. [231]   MXene/PS PVA NO 2 12.11% [5 ppm] 50 ppb Ref. [165]   systems and inevitably increase the potential risk of explosion and fire when detecting flammable and explosive gas.Recent reported wearable gas sensors have offered an effective solution to the challenges mentioned above. [15,16]With the rapid expansion of industrialization and urbanization, the emission of hazardous gases poses a significant threat to human health.For example, overexposure to NH 3 can cause irreversible severe irritation to the respiratory organs and eyes, [17] while long-term exposure to formaldehyde in the air can lead to a decline in lung function and chronic pneumonia. [18,19]Exposure to low concentrations of nitrogen oxides (NO x ) over an extended period can induce chromic pharyngitis and bronchitis. [20]Thus, it is crucial to detect hazardous gases around the human body immediately using portable wearable gas sensors in our daily lives.Moreover, these gases have also been employed as biomarkers for some diseases, enabling instant monitoring of human health in medical care to realize timely diagnosis and treatment.For instance, acetone in exhaled breath can be used to diagnose diabetes, and hydrogen sulfide (H 2 S) can be used to diagnose halitosis. [15,16]he concentration of nitric oxide (NO) is a standard to diagnose chronic bronchitis. [21]It is evident that this type of sensor has opened up new fields of applications for monitoring the condition of the human body and the environment around it in realtime.Depending on the application scenario and the detected gas type in actual breath, wearable gas sensors can be used for environmental monitoring, healthcare, smart home, industrial security, food safety monitoring, and public security.In addition, wearable gas sensors are small in size, can comfortably adhere to the human skin, and deform with human movements without altering the sensor response.
Thus, selecting and designing flexible substrates and sensitive materials has become an urgent priority to enhance the sensing performance and mechanical flexibility of gas sensors at room temperature.Compared to traditional rigid substrates, emerging polymers, and textile substrates not only exhibit bendable and stretchable properties for customization but are also lightweight and inexpensive.Table 1 shows multifarious polymer-based substrates, such as polyethylene glycol terephthalate (PET), polyurethane (PU), polydimethylsiloxane (PDMS), and polyimide (PI), which display both air permeability and mechanical flexibility.Consequently, the wearable comfort and bendability have been significantly improved. [22]To enable the real-time detection of human physiological signals and monitoring of the surrounding environment, and while being aware of potential threats, ionic liquids, polymers, carbon-based nanomaterials, 2D semiconductors, and other flexible sensitive materials can provide attractive solutions as flexible sensing materials.These materials offer an edge over others due to their extraordinary flexibility and high sensing performance at room temperature.Moreover, wearable gas sensors have another advantage over traditional gas sensors due to their lower power consumption.
Some researchers have summarized wearable gas sensors using one single type of sensing material, [23] wearable gas sensors requiring extra light activation, [24] or those designed to detect fixed single gas type. [25]However, there is no comprehensive review available to introduce and summarize different sensing materials, their properties, gas sensing mechanisms and applications of wearable gas sensors.Given the tremendous interest of researchers in improving the sensing performance of wearable gas sensors, this review provides a comprehensive introduction to the recent progress in the development of wearable gas sensors with low operation temperature, as illustrated in Figure 1.This review will benefit beginners who wish to become familiar with the operation of wearable gas sensors at room temperature, leading them to acquire a preliminary understanding and move toward a deeper mastery of the subject.

Mechanism of Wearable Sensors for Gas Monitoring at Room Temperature
[28] Among these, gas sensors based on electrical/electrochemical principles can detect the concentration of target gas by monitoring changes in electrical properties when gas molecules react with the sensor on a physical or chemical level.This sensing mechanism can be explained by the following models.

Oxygen Adsorption Model
The oxygen adsorption model has usually been employed to explain the sensing mechanism of metal oxide semiconductor (MOS) gas sensors, such as WO 3 , [29] CuO, [30] and ZnO. [31]According to this model, the resistance of the sensor varies with the amount of oxygen adsorbed on the surface of the sensitive material.At room temperature, when the gas sensor is located in the air, the chemisorbed oxygen is converted into negatively charged oxygen ions by capturing free electrons of sensitive materials. [32]he trapped free electrons are then released when the target reduction [33] or oxidization gases [34] that interact or compete with oxygen ions, resulting in a change in resistance. [23,35]The specific mechanism varies depending on the target gases.
Figure 2a,b depicts the strong oxidizing NO 2 sensing mechanism when In 2 O 3 /g-C 3 N 4 is used as the sensitive material. [36]n air, oxygen is absorbed on the porous surface of the sensing material and is converted into O 2 − due to the electron-hole pairs.Upon the introduction of NO 2 , the target gas NO 2 directly captures electrons in the sensing material or captures electrons in oxygen ions, as shown in Equations 1 and 2. [37,38] NO 2 + e − ↔ NO − 2 (ads) (1) Thus, the resistance of the sensing material sharply increases due to the trapping of a large number of electrons, which is similar to that of p-type sensor, and the mechanism of desorption is similar as well.On the other hand, O 2 − is formed by capturing free electrons when exposed to air.When the sensor is exposed to reducing gas, the resistance of the sensing layer is decreased. [39,40]Owing to the intense reducibility of NH 3 , the introduction of NH 3 can release trapped electrons back to the sensing materials, as demonstrated in Equations 3 and 4. [41] O 2 + e − → O 2 (ads) (3) Additionally, it has been reported that visible light irradiation, humidity, and the structure of sensing layer, such as specific surface area, porosity, and functional groups, could accelerate the process of response and recovery. [30,39,42]With UV illumination, a large number of electron-hole pairs are generated on the surface of the sensing material, promoting the photocatalytic adsorption of oxygen because of its weaker binding energy compared to dark condition.This contributes to the rapid reaction between negative O 2 − and the target gas acetone, as shown in Equation 5. [8] CH

Direct Adsorption Model
Copyright 2021, Wiley-VCH.b) Diagram of In 2 O 3 /g-C 3 N 4 heterojunction and the corresponding energy band.Reproduced with permission [36] Copyright 2021, Wiley-VCH.c) DOS of the ternary-NCs after NH 3 adsorption.Reproduced with permission. [46]Copyright 2022, American Chemical Society.d) Schematic of the NH 3 gas sensing mechanism illustrated by the direct electron transfer model.Reproduced with permission [46] Copyright 2022, American Chemical Society.e) Schematic illustration of the sensing mechanism of the flexible GP-PANI/PVDF sensor.Reproduced with permission. [52]Copyright 2021, Elsevier.
Among them, the sensing mechanism of most 2D sensing materials, such as MoS 2 and GO, is explained by the direct electron transfer model.The CY/RGO/MoS 2 exhibits ultrahigh sensing response to NO 2 , attributed to the synergistic effect between RGO and MoS 2 . [43]While pure MoS 2 is sensitive to NO 2 , the introduction of MoS 2 as an additional sensing material into CY/RGO increases the concentration of electrons and strengthens the transfer of electrons between RGO and MoS 2 owing to the energy levels effect. [44]Li et al. further revealed the interfering factors of the direct electrons transfer process. [45]During the gas sensing process, the sensing performance depended on the adsorption energy, distance, and number of transferred electrons between the target gas and sensing materials.For example, in the case of NO 2 , the strong absorption and interaction between NO 2 molecules and the active hybrid rGO/ZnO layer extraordinarily improved the device output signal, which was attributed to the ultrahigh adsorption energy, short distance, and a large number of electron transfer between the sensing layer and NO 2 gas molecules.
Furthermore, Du et al. [46] conducted theoretical simulations to investigate the sensing process when the direct electron transfer model was applied.As shown in Figure 2d, the electrons transfer took place between the target gas NH 3 and the Pt/polypyrrole (PPy)@CNT sensing layer.The absorbed NH 3 molecules with rich electrons directly transferred electrons to the positively charged sensitive layer, as demonstrated in Equation 6, resulting in increased resistance.
Moreover, the extraordinary absorption and catalysis effect of Pt contributed to the electrons transfer between PPy and NH 3 .And the multiwall CNTs formed electron delocalization with PPy, which boosted the transfer of electrons enriched in the PPy layer to the CNTs layer and combined with holes.As demonstrated in Figure 2c, the density of states near the Fermi level increased apparently, indicating strong interactions and a synergistic effect of Pt/PPy@CNT at room temperature, which was favorable for sensing performance. [47]The sensing mechanism of Ti 3 C 2 T xbased gas sensors is also explained by direct adsorption model.Yang et al. [48] explored the effect of Ti─O vacancies on the adsorption of the target gas NO 2 to Ti 3 C 2 T x /TiO 2 /rGO sensitive material using density functional theory (DFT) calculations.The results showed that the formation of Ti─O vacancies lowered the NO 2 absorption energy on crumpled MXene sphere, and the target gas NO 2 was easily captured by the sensitive layer.The electronic charge of NO 2 and MXene was then violently overlapped, accelerating the electrons transfer between the sensitive layer and NO 2 .

Proton Transfer Model
Proton transfer is the most commonly accepted mechanism to explain the sensing process of polyaniline (PANI)-based gas sensors at room temperature.PANI exhibits conductivity when located in an acid environment or doped with emeraldine and insulation in the de-doped form or in basic environment. [49,50]mong these, protic acid-doped PANI is often employed to explain the NH 3 sensing mechanism. [51]When doped with protic acid, H + combines with the nitrogen atom on the imine, forming PANIH + .As demonstrated in Figure 2e, when exposed to NH 3 , the proton of PANIH + is captured and converted into NH 4 + , which is known as deprotonation.It depletes the holes of PANI, resulting in insulation and a subsequent decrease in resistance. [52]n addition, PANI exhibits p-type semiconductor characteristics while most metal oxide are n-type semiconductors.Once PANI is functionalized by metal oxide, a p-n junction and depletion layer are formed due to the huge gradient of the carrier concentration, which plays a vital role in the improvement of sensing performance. [53]Upon the introduction of target NH 3 gas, H + at the interface between PANI and metal oxide is captured, decreasing the PANI doping concentration, broadening the depletion layer, and increasing the resistance.When exposed to high humidity conditions, H 2 O molecules act as proton source, which contribute to the protonation of PANI and improve the sensing response.The reactions can be described by Equations 7 and 8. [54] H 2 O ↔ H + + OH − (7)

Ions Conduction Model
The ions conduction model employs ions as carriers in the gas sensing process.For instance, the gas sensing mechanism of CuBr-based sensors is frequently demonstrated by the ions conduction model in which the carrier is Cu + .Even at room temperature, the carrier Cu + can be immobilized by the target gas NH 3 rapidly, forming Cu(NH 3 ) 2+ , which decreases the Cu + concentration and increases the resistance.With the removal of NH 3 , Cu(NH 3 ) 2+ is immediately decomposed without any extra heat or light, and the resistance of the sensing materials decreases. [55,56]dditionally, it is an effective method to eliminate the effect of humidity by introducing metal oxide such as CeO 2 during the gas sensing process. [57]Li et al. [58] designed the CeO 2 layer on top of the CuBr sensing layer to protect the sensing layer from water molecules.The sensing mechanism of organohydrogel could also be explained by the ions conduction model. [59]Due to their higher solubility in water and higher adsorption capacity for hydrogel functional groups, target gas such as NH 3 and NO 2 exhibited higher adsorption capacities compared to other VOCs.The target gas and byproducts of dissolution blocked the mobility of carriers, which were the ions in organohydrogel, resulting in higher resistance. [60] The Fabrication of Wearable Room Temperature Gas Sensors

Flexible Sensitive Materials
The flexibility of gas sensors has emerged as a bottleneck in the application of wearable intelligent devices, limiting their service ability and requiring urgent attention.Thus, the mechanical flexibility of both the sensitive layer materials and substrate should be taken into account.Ionic liquids (ILs), polymers, carbon-based nanomaterials, and transition metal compounds are commonly employed as flexible gas-sensitive materials to substitute the traditional rigid materials in smart devices.

Ionic Liquids
Ionic liquids, as room-temperature molten salts, are composed of organic cations such as N-alkylpyridinium, N,Ndialkylimidazolium, and various anions, [61] which exhibit not only high ionic conductivity but also mechanical flexibility, [62] making them ideal candidates for wearable gas sensors.Wan et al. [63] utilized 1-butyl-1-methylpyrrolidinium bis-(trifluoromethylsulfonyl)-imide ([C 4 mpy][NTf 2 ]), a room temperature ionic liquid as electrolytes to establish an electrochemical gas sensor. [64]The [C 4 mpy][NTf 2 ] formed an interface with Pt electrode, which exhibited excellent catalytic properties to identify multiple gases, including SO 2 , NH 4 , and O 3 . [65]As shown in Figure 3a, microfabrication was utilized to form disk-shaped planar gold electrodes on a porous PTFE substrate, which increased the response speed and device stability and achieved miniaturized and wearable formats.1-butyl-3-methylimidazolium-based ionic liquids [BMIM][DCA] and [BMIM][Cl] were employed to form hybrid material with gelatin and liquid crystal 4-cyano-4′pentylbiphenyl (5CB), as shown in Figure 3b. [66]It could not only monitor the humidity as a humidity sensor, but also tune the humidity impact as a humidity-tolerant VOC sensor working at room temperature, whose solid performance and miniaturized size paved a new avenue for wearable devices.Additionally, Koziej et al. [67] proposed a new strategy for detecting CO 2 between 150 and 2400 ppm at room temperature.The conductivity of Poly [p-vinylbenzyl) trimethylammonium hexafl uorophosphate] could be raised after doping with La 2 O 2 CO 3 , where the conductive channels were formed to convert chemical reactivity into an electrical signal for CO 2 sensing.

Conductive Polymers
Conductive polymers (CPs), with conjugated principal electrons in their main chain, exhibit conductivity or obtain electrical conductivity after doping, which typically exhibit mechanical toughness. [68]Moreover, CPs are considered ideal candidates as wearable sensitive materials due to its mechanical toughness, low cost, light weight and room temperature response.This makes them highly suitable for application in wearable gas sensors working at room temperature. [69]Furthermore, CPs with adjustable bandgaps can reduce the bandgap to accelerate the generation of electrons by simple doping and compound  [65] Copyright 2013, IOP Publishing.
modification. [70]Initially, CPs are used solely as electrodes in gas sensors. [71]So far, many conductive polymers like PPy, PANI, poly(3,4-ethylenedioxythiophene): poly (styrene sulfonate) (PE-DOT: PSS) and poly(2-phenyl-1,4-xylylene) (PPPX) have been reported as sensing materials with doping conducting particle, inorganic salts and metal oxides. [33,72,73]s a type of CP, PPy has extraordinary advantages as a sensing material over original polymer due to its predominant mechanical and rigid electrical properties. [74,75]Li et al. [76] designed and fabricated porous neural network-like Au/PPy fibrous membranes for rapid NH 3 detection at room temperature by utilizing electrospinning and in situ vapor-phase polymerization.The 3D interconnected neural network structure built an electronic transmission path throughout the whole fibrous membrane, cleverly improving the signal transmission efficiency.Additionally, it was an ingenious strategy to design porous capsuleslike PPy that offered more active sites for the target gas.As an emerging conductive polymer, PPPX has been employed as a sensing material in the array of composite sensors at room temperature. [77]It was dissolved in DMF with porphyrin and then coated on the interdigitated electrode to form a sensitive layer ≈ 1 m in thickness.This work further confirms that doping has altered the electrons distribution and conductivity of conducting polymers, leading to the improved sensing performance.
PEDOT: PSS, as a well-known conductive polymer, has been widely exploited in wearable electronic equipment since its high electronic conductivity, easy processing, and commercial availability. [70,78,79]As shown in Figure 3d, a PEDOT: PSS fiber was employed as the working electrode, deposited with a Pd sensing layer, which exhibited an excellent response to H 2 at room temperature, short response time, and excellent cycling stability under a variety of mechanical bending states. [80]Lv et al. [81] utilized the same material, PSS, and observed that when PSS was added into a PANI/PVDF film, it was adsorbed by PANI NPs to form a PSS-PANI/PVDF composite film, which was used as the sensitive layer for NH 3 detection.The typical hierarchical porous structure was conducive to apace caging the target gas molecules and substantially increased the contact areas of NH 3 and PANI films, as shown in Figure 3c.The PSS-PANI/PVDF-based sensor was flexible, low power consumption, and low cost, providing new opportunities for intelligent wearable devices for real-time NH 3 detection.Liu et al. [82] functionalized PANI with WO 3 hollow spheres by a simple hydrothermal method and in situ chemical oxidative polymerization.The hollow structure of WO 3 and the p-n heterojunction between PANI and WO 3 contributed to the higher response of the sensor to NH 3 compared with pure PANI.Liu et al. [83] also adopted the conductive polymer PANI, which was functionalized by porous SnO 2 /Zn 2 SnO 4 nanosphere as the sensitive layer.The special porous structure and the formation of the heterojunction significantly strengthened the sensing performance of the NH 3 sensor.The flexible PET substrate was beneficial for the portability and wearability.To further  [33] Copyright 2017, Elsevier.b) Schematic of the spray deposition of MWCNTs on the fabric (F-MWCNTs) and PANI synthesis on MWCNTs coated fabric surface.Reproduced with permission. [73]Copyright 2017, Elsevier.c) Schematic illustration of Ti 3 C 2 T x synthesis procedure, electrode sputtering, solution deposition process, and gas-sensing device.Reproduced with permission. [85]Copyright 2017, American Chemical Society.d) Schematic diagram of the fabrication process of rGO/ZnO hybrid fibers adopting APTES as the molecular glue.Reproduced with permission. [86]Copyright 2019, American Chemical Society.e) Schematic illustration of the spinning process of MXene/GO hybrid fiber.Reproduced with permission. [87]Copyright 2020, American Chemical Society.
improve the sensing performance of the PANI-based gas sensor, Liu et al. [84] introduced Rh-doped SnO 2 to PANI by electrospinning and in situ polymerization, which improved the response and decreased the response/recovery time.The significant progress in the sensing performance was attributed not only to the p-n heterojunction but also to the formation of Rh 2 O 3 with catalytic effect.
Additionally, the sensitivity, selectivity to target gas and durability of the sensing layers can be significantly enhanced by functionalizing CPs with other types of materials, such as carbonbased 2D materials and metal. [33,72,73]

Carbon-Based Nanomaterials
[90][91] Besides, considering the demand for flexibility in wearable intelligent devices, carbon-based nanomaterials have been extensively applied in combination with polymer materials to achieve better performance and increased stability under large deformations. [92]s a unique carbon-based material, CNTs have been exploited in gas sensors working at room temperature to effectively detect the target gases, such as O 2 , [93] alcohol, [73] and NO 2 . [94]As early as 2017, Li et al. [33] adopted an effective in situ chemical solution method to prepare an MWCNI/PANI nanocomposite alcohol gas sensor, which was integrated with a PPy-based micro-supercapacitor (MSC), as shown in Figure 4a.This wearable gas sensing system displayed an excellent selectivity to alcohol and a low detection level of < 1 ppm at room temperature.Additionally, the sensitive layer of the alcohol gas sensor was fabricated by modifying the multi-wall carbon nanotube (MWCNTs) with polyvinyl alcohol (PVA), which was sprayed on cotton fabrics, as demonstrated in Figure 4b. [73]Compared with a pure MWCNTs fabric sensor, the PVA/MWCNTs-based sensor exhibited a higher selectivity, ≈ 13-fold, to alcohol at room temperature.In addition, CNTs are frequently functionalized with other materials such as PANI, [95,96] Pt, [72] TiO 2 [97] and Pt, [72] which improves their ability to react with and cage more varieties of gas molecules including NH 3 , [98] TNT, [99] NO 2 , [100] and NO. [97]This indicates that the diverse functionalized CNTs have provided the possibility for detecting a wide range of target gases. [101,102]raphene is a single-layer, 2D material with abundant hydroxyl, carboxyl, and epoxide functional groups, contributing to numerous active adsorption sites for caging target gases. [103]Consequently, it has been an ideal candidate for wearable gas sensors operating at room temperature.
Functionalizing electronic textiles with GO is a popular method to fabricate GO-based gas sensors with lightweight, flexibility, excellent mechanical durability, and reliable sensitive performance. [104,105]For instance, RGO-coated cotton yarn modified by MoS 2 exhibited an ultrasensitive response to NO 2 of ≈ 28% and extraordinary durability to washing and bending stresses with > 100 repetitive washing and 1000 bending. [106]Bovine Serum Albumin (BSA) was employed as a molecular glue to achieve electrostatic self-assembly between cotton yarn and RGO.Similarly, Li et al. [86] adopted (3-Aminopropyl)-triethoxysilane (APTES) as the linker to coat RGO on the surface of cotton and elastic thread.The resulting RGO-based e-textiles functionalized by mesoporous ZnO nanosheet could be integrated into clothes for real-time monitoring of NO 2 in advanced wearable electronics, as demonstrated in Figure 4d.
Compared with GO, MXene is a more novel class of 2D nanomaterial, and developing flexible gas sensors at room temperature using MXenes has been an appealing but challenging area of research. [107]In 2017, Kim's team [85] reported the first Ti 3 C 2 T x gas sensor with ultrasensitive capabilities for volatile organic compound (VOCs).This study was also the first to apply MXene into a wearable gas sensor working at room temperature to the best of our knowledge.In this study, the Ti 3 C 2 T x water dispersion was simply dropped on the interdigitated electrodes with a flexible PI substrate, which exhibited a sensitive response to alcohol, methanol, acetone, and NH 3 , as shown in Figure 4c.Then, Kim et al. [108] demonstrated that the Ti 3 C 2 T x displayed both high metallic conductivity and abundant analyte adsorption sites, resulting in low noise and strong signal, which benefited the high sensitivity of gas sensors.The Ti 3 C 2 T x sensor behaved an amazing detection limit of ≈ 50−100 ppb for VOC gases while working at room temperature.Furthermore, to broaden the categories of detection gases including NH 3 , NO 2 , H 2 and formaldehyde, researchers have frequently functionalized MXene with other types of materials, such as Pd, [109] CuO, [110] ZnO, [111] MoS 2 , [112] Co 3 O 4 , [113] and bentonite (Bt) nanoclay. [114]Ti 3 C 2 T x /GO hybrid fibers were fabricated by a wet-spinning process and their NH 3 sensing response was significantly increased compared with pure Ti 3 C 2 T x and GO, as shown in Figure 4e. [87]The addition of GO removed the narrow band gap defect of MXene and accelerated the adsorption of gas molecules.Chemical modification of MXene is also an effective method to improve the sensing performance of MXene-based gas sensors.Yang et al. [115] alkalinetreated the organoid structured Ti 3 C 2 T x by a sodium hydroxide solution.The fluorine group was substituted by the hydroxyl group with excellent gas adsorption properties, thereby enhancing the NH 3 and humidity sensing performance.Another practical way to enhance the gas sensitivity of MXene involves converting 2D Ti 3 C 2 T x nanosheet into 3D crumpled Ti 3 C 2 T x sphere by ultrasonic spray pyrolysis. [116]The high specific surface area of crumpled Ti 3 C 2 T x and the formation of p-n heterojunction enhance not only the response signal and reusability but also the selectivity for NH 3 .
At the same time, the fibrous structure of MXene maximizes the specific surface, enabling the sensitive material to fully capture the target gas.These researches have stimulated growing concern in promoting the application of MXene in the field of wearable gas sensors at room temperature.Thomas et al. [117] fabricated a Mo 2 CT x MXene-based CO 2 sensor, which exhibited great response and rapid response/recovery time at room temperature (30 °C).However, a porous silicon was employed as substrate, making it difficult to bend.Thus, it is an attractive strategy to apply other types MXene including Mo 2 CT x and Nd 2 NT 2 , in wearable gas sensing devices.These MXenes exhibit stable 2D structures, fully functionalized surfaces with more absorption sites, low electrical noise, and strong signals as same as Ti 3 C 2 T x , but have been rarely reported in wearable gas sensor so far. [108,118]n addition, the excellent solution processability and adjustable bandgap of MXene make it possible to be modified for enhanced sensing performance. [119,120]

2D Semiconducting Nanomaterials
2D semiconducting nanomaterials, including 2D metal oxides (2DMOs), transition metal dichalcogenides (TMDCs) like MoS 2 , MoSe 2 , and WS 2 , h-borazon (h-BN), and black phosphorus, have been proposed as some of the most traditional gas sensing materials due to their ease of synthesis and high response to target gases.For example, WS 2 -based gas sensors have exhibited ultralow detection limit for NO 2 , [121] and 3D structured h-BN has been used as a liquid petroleum gas sensors at 187 °C. [122]owever, high operation temperature and high power consumption have become obstacles for wearable gas sensor operating at room temperature. [23]As a result, realizing gas sensing at room temperature is the vital factor to make 2DMOs-and/or TMDC-based wearable sensors.Functionalization with metal oxide, [123,124] metal nanoparticles, [125,126] or other semiconductors to form heterojunctions has been an effective strategy to enhance the sensing performance and flexibility at room temperature, which suggests that high gas sensing performance at room temperature can be achieved by simple doping.Thus, it is an attractive avenue to further explore 2D semiconductors for wearable gas sensors.
The initial 2D semiconductor materials behave high sensing performance and solid recovery ability due to their heterostructures and large surface areas. [127,128]131] Figure 5a,b depicts a bifunctional gas sensor based on WSe 2 nanosheets fabricated by a liquid-phase exfoliation method, which had a high response to triethylamine without UV illumination and NO 2 with UV illumination. [132]What's more, adopting PI substrate made it possible to exhibit flexibility without altering sensing performance.On the other hands, adopting other types of TMOs to modify WSe 2 has also provided effective solutions for improving sensing performance at room temperature.For example, a WO x /WSe 2 hybrid film was synthesized by inductively coupled plasma (ICP) process, as shown in Figure 5c, which exhibited an ultralow detection limit of 0.3 ppb to NO x at room temperature. [133]And the sensitivity of the flexible gas sensor based on a PI substrate was only slightly decreased after a bending test over 75 °, as shown in Figure 5d.Ko et al. [134] fabricated uniform WS 2x Se 2−2x alloys by synthesizing WSe 2 and consequent sulfurization, as illustrated in Figure 5e.Remarkably, the response of WS 2x Se 2−2x alloys was about twice higher than that of WSe 2 -based gas sensors and they also exhibited low power consumption.MoS 2 is also one of the most typical TMDC materials. [135]Kim et al. [136] implemented precise control over the layer number of 2D MoS 2 through a single chemical vapor deposition (CVD) process.The Schottky barrier height varied with the number of MoS 2 layers, which decided the responsivity to NO 2 , CO, and CO 2 , while 2D MoS 2 -based gas sensors showed a minimal response to CO and CO 2 .Building upon this, Ikram et al. [137] proposed an effective hydrothermal strategy that converted the rhombic p-p MoS 2 @ZIF-8 into rod-like p-n MoS 2 @ZnO Figure 5. a) Synthesis process of WSe 2 nanosheets.Reproduced with permission. [132]Copyright 2021, Elsevier.b) Scheme of the flexible sensor based on WSe 2 nanosheets.Reproduced with permission. [132]Copyright 2021, Elsevier.c) Schematic of the low-temperature plasma-assisted selenization process of metal oxide.Reproduced with permission. [133]Copyright 2017, American Chemical Society.d) Photographs of the gas sensor on a polymer substrate.Reproduced with permission. [133]Copyright 2017, American Chemical Society.e) Schematic of the synthesis process of WS 2x Se 2−2x alloy depending on the sulfurization temperature.Reproduced with permission. [134]Copyright 2018, American Chemical Society.(f) Schematic of the formation of MoS 2 @ZIF-8 and MoS 2 @ZnO heterostructures.Reproduced with permission. [137]Copyright 2021, American Chemical Society.
heterostructures, as demonstrated in Figure 5f, which improved the gas response and lowered the detection limit of NO 2 to 10 ppb.
The types of target gas and the sensing performances of gas sensors are primarily dependent on the nature of the sensitive materials.In other words, the ability of the sensitive materials to respond to target gases at room temperature is a crucial factor to fabricate wearable electronics operating at room temperature.Consequently, those studies shed some light on exploring multiple modifications for improving the sensing performance of sensitive materials in order to further break the limitations of current wearable gas sensors.

The Classification of Gas Sensor
Wearable gas sensors that operate at room temperature with various sensitive materials have been extensively employed to detect harmful gases such as NH 3 , [138] NO x , [106,139] and SO x , [63] volatile organic compounds (VOCs) , [140,141] flammable gases including O 2 [93] and H 2 , [80] and others like TNT [99] and trimethylamine. [142]igh-temperature gas sensors have limited types of gases that can be monitored, which inevitably cause a highly potential risk of fire and explosion, when detecting flammable and explosive gases at high operating temperature.However, room temperature gas sensors do not require to consider the influence of operation temperature on sensing performance.The output signal is decided by the absorption or interaction between the target gas molecules and sensitive layer.Herein, the sensitivity is defined as the rate of ∆R/R 0 with the concentration of the target gas, where ∆R and R 0 are resistance change after introducing the target gas and the resistance in the air, respectively, unless otherwise stated.The theoretical detection limit is defined as the response signal with a signal-to-noise ratio of 3. [108,143] The actual detection limit is usually the lowest detected gas concentration during the test unless otherwise stated.The selectivity is related to the relative response between different gases compared to the target gas, [144] which is used to measure the ability to accurately identify the target gas with the coexisting of various interference gases. [145]The response/recovery time is defined as the time taken to reach 90% of the balanced response value after introducing or removing the target gas.

NH 3 Sensor
NH 3 is a highly toxic gas that is mainly used as a chemical raw material.It is also applied as a detergent, neutralizing agent and alkaloid leaching agent.Meanwhile, it can serve as a biomarker in the exhaled breath to detect kidney diseases, [146] obstructive lung disease [147] and oxidative stress schizophrenia, [148] etc. as well.In addition, NH 3 can easily disperse in water and air, causing serious environmental pollution and acute damage to human health, including headache, pulmonary damage and even death.Thus, NH 3 sensors play a vital role in environmental protection and safety care. [149]n terms of room temperature sensors for detection of NH 3 , PANI-based flexible devices have been widely reported. [150,151]ANI-based gas sensor was fabricated by electrospun and/or solid-state drawn doped with (+)-camphor-10-sulfonic acid (HCSA), performing p-type semiconductor characteristics. [152]hen exposed to NH 3 at concentration < 20 ppm, the sensitivity of the drawn fibrous sensor was reached 5.5 ppm −1 , which was slightly higher than that of as-spun fibers with 3.5 ppm −1 .These were both higher than the cast HCSA/PANI film with a sensitivity of 0.02 ppm −1 .Moreover, it was excellent that the response ratios in 700 ppm concentration NH 3 were as large as 60-fold.It is well-known that solid-state drawn results in smaller fibrous diameter and more neat molecular orientation in the drawn fibers.Thus, the drawn fibers exhibited slightly faster response time when exposed to NH 3 at concentrations ranging from 10 to 700 ppm compared with as-spun fibers.However, the difference between the two was not significant.Besides, the response/recovery time of the room temperature NH 3 sensor was ideal, at 45 and 63 s, respectively.
Wang et al. [153] recently reported excellent NH 3 response of a CeO 2 @PANI-based gas sensor for disease monitoring of patients with chronic kidney disease recently. [154]The formation of a heterojunction between CeO 2 and PANI increased the gas-sensitive effect, resulting in a clear increase ingas response.After hydrogen plasma treatment for 30 min to produce more surface defects, the actual detection limit could reach as low as 50 ppb NH 3 with a 24% response, as shown in Figure 6a, which has a promising application in clinical monitoring for CKD.And the sensitivity of the sensor could be improved to 5.68 ppm −1 (the sensitivity was defined as the ratio between ∆R/R 0 × 100% and the concentration of NH 3 in ppm, where R 0 and ∆R were same as those previously mentioned).A porous CuBr-based NH 3 sensor at room temperature consisted of loosely percolated CuBr particles with sufficient connectivity and effective conduction, which was beneficial for achieving an ultrahigh response to NH 3 . [155]Notably, the sensor displayed both an ultralow actual detection limit of 5 ppb and ultrahigh selectivity for NH 3 , as shown in Figure 6c.The theoretical detection limit was calculated to be 210 ppt.Except that, traditional sensitive materials for NH 3 such as CuO, [156] Pd-WO 3 •xH 2 O nanoparticles, [157] and MoS 2 [158] have also been frequently introduced into wearable gas sensors with room temperature operability.

Nitrogen Dioxide Sensor
Nitrogen dioxide (NO x ), one of the most common toxic irritant gases mainly produced by various combustion processes, can cause toxic damage to human health and environmental protection, such as acid rain, [159] and respiratory ailments. [160]Thus, wearable NO x sensors operating at room temperature are of great significance.One of the most notable examples of such sensors employs 2D semiconducting nanomaterials as the sensitive layer. [161]inc oxide (ZnO) nanorods hydrothermally grown on PET substrate exhibited an excellent NO 2 response of 185% for 500 ppb after gamma-ray radiation. [162]The induced defects on the surface of ZnO nanorods acted as active sites to cage the target gas after radiation.The selectivity of the sensor toward NO 2 was significantly higher than other gases such as NH 3 and CH 3 OH.An organic 2D polymer covalent triazine framework (CTF)-based flexible gas sensor was fabricated to meet the requirements of a wearable NO 2 gas sensor including room-temperature operation, high specific recognition and flexibility. [163]It exhibited an ultrahigh selectivity of 452.6 ppm −1 for NO 2 and an ultralow calculated detection limit of NO 2 down to 2.2 ppb, as shown in Figure 6b (the sensitivity was defined as the ratio between ΔG/G 0 and the gas concentration in ppm, where G 0 and ΔG were the conductance in balanced gas and the conductance change after introducing the target gas).At the same time, the solid sensing performance was demonstrated in multiple bending cycles at bending angle from 0°to 90°or after being aged in ambient air for 1-3 weeks. [164]Yang et al. [165] designed a 3D porous crumpled MXene-based NO 2 sensor which exhibited high sensitivity at ppb level for NO 2 .The polyvinyl alcohol (PVA) was employed as a flexible and water-soluble substrate and MXene was sprayed on substrate as the electrode, which could be completely degraded in medical H 2 O 2 .The sensor exhibited extraordinary selectivity for NO 2 where the response of 5 ppm NH 3 , ethanol, acetone, ethanol, and toluene were 4.35%, 1.99%, 1.39%, 2.98%, and 0.73%, respectively, significantly lower than 12.11% to 5 ppm NO 2 .The actual detection limit was as low as 50 ppb, which was extraordinary.Han et al. [139] converted the inorganic semiconductor yttria-stabilized zirconia/In 2 O 3 /graphitic carbon nitride (g-C3N4) (ZIC) into an all-flexible NO 2 sensor with sensing performance at room temperature by atomic layer deposition (ALD), subsequent vapor deposition method and electrospinning.The obtained In 2 O 3 /g-C 3 N 4 sensing layer with ultrathin thickness and YSZ substrate with small nanofiber diameter ensured synchronization and sensing stability during deformation.As shown in Figure 6d, the detection limit of this all-flexible sensor was as low as 50 ppb in response to NO 2 and it exhibited an extraordinarily high gas response with increasing gas concentrations (the detection limit here was defined as the response value > 1.5 for effective gas sensing).This work has accelerated the application of traditional inorganic semiconductor sensitive materials in wearable NO 2 sensors.Reproduced with permission. [153]Copyright 2022, American Chemical Society.b) Sensitivity curve of CTF-based gas sensor.Reproduced with permission. [163]Copyright 2020, American Chemical Society.c) Ultrahigh selectivity to NH 3 of the porous CuBr-based NH 3 sensor (A: NH 3 , N: NO 2 , T: acetone, C: CO, E: ethanol, D: acetaldehyde, and F: formaldehyde).Reproduced with permission. [155]Copyright 2022, The Royal Society of Chemistry.d) Sensing response curves of ZIC to different concentrations of NO 2 (50 ppb to 5 ppm).Reproduced with permission. [139]Copyright 2021, Wiley-VCH.e) Response characteristics fitting curves of the sensors based on MXene/Co 3 O 4 to 0.01-10 ppm HCHO at 25°C.f) Selectivity of the sensors based on MXene/Co 3 O 4 to various testing gases with a concentration of 10 ppm.Reproduced with permission. [113]Copyright 2021, Elsevier.g) Gas response as a function of formaldehyde concentration.h) Gas-sensing

Volatile Organic Compounds (VOCs) Sensor
Volatile organic compounds (VOCs) are chemicals with high vapor pressure at room temperature that evaporate readily, including but not limited to ethanol, methanol, acetone, methanal, acetaldehyde, and acetic acid. [166]VOCs have a negative impact on air quality, and they can also be employed as indicators of diseases such as lung cancer, [167] heart disease, [168] colorectal cancer, [169] etc.
Common VOCs: One example of gas sensors response to common VOCs here is the methanal sensor based on MXene and Co 3 O 4 . [113]Figure 6e depicts the identifiable response signal, where the response was increased as the concentration of methanal expanded.Moreover, it exhibited high specific recognition and an ultralow actual detection limit of 0.01 ppm to methanal, as shown in Figure 6f.The sensitivity of the sensor was 0.82 ppm −1 , where the sensitivity was defined as the rate of R a /R g and the concentration of the target gas in ppm and R a and R g were the resistance in air and the target gas, respectively.Another example of the wearable methanal sensor employed zeolitic imidazole framework (ZIF-7) nanoparticles as the molecular sieving to avoid the interference of alcohol, TiO 2 as the sensing layer, and polyethylene terephthalate (PET) as the flexible substrate. [141]This sensor showed a significant response to 5 ppm methanal and the response value (R a /R g ) was up to 1350.9. Figure 6f illustrates that the detection limit of methanal was as low as 0.0038 ppm (the detection limit here was defined as the R a /R g when R a /R g -1 > 0.2).The flexibility of the sensor, as shown in Figure 6h,i, also shed some light on wearable electronic devices.In addition, the alcohol sensor is another vital type of VOCs sensor. [140]Hassan et al. [170] designed a bionic fractal structured gas sensor fabricated by an ink printing process, where PET and graphene ink were employed as the substrate and the sensing layer, respectively.The fractal structure played a significant role in improving the sensing performances for alcohol due to the increasing specific surface area.The detection range was broadened to 5-100 ppm and the response time was effectively reduced as well.Its sensitivity was as high as 0.321 ppm −1 (the sensitivity was defined as the ratio of ∆R/R 0 × 100% to the concentration of the target gas in ppm, where ∆R and R 0 were as same as those in the previous section).The scientific research team of Fan [171] also adopted an ink printing process to fabricate a wearable wristband based on Manganese (IV) oxide (MnO 2 )/reduced graphene oxide/PEDOT: PSS and SnO 2 hybrid ink to detect alcohol and acetone.An exceptional sensitivity for alcohol and acetone was observed with the value of 13.2% to 1.54 vt% and 3.1% to 6.07 vt% respectively, where the sensitivity was defined as |∆R/R 0 | × 100% and ∆R and R 0 were as same as those mentioned previously.
Total Volatile Basic Nitrogen (TVB-N): Total volatile basic nitrogen (TVB-N) including trimethylamine (TMA) and triethy-lamine (TEA) has a pungent, strong NH 3 -like odor, which can result in severe symptoms of lung irritation. [172]It is a branch of VOCs and is produced in the process of fish and seafood deterioration.It can also be employed as a biomarker for detecting the freshness of fish and seafood. [173,174]n example of wearable TVB-N sensor based on oxygen-doped MoSe 2 nanosheets registered an ultralow theoretical limit of detection of 8 ppb for TMA, where the actual detection limit was 10 ppb.The theoretical detection limit was defined as 3.3 /S, where S and  were the slope of the calibration curve and the standard deviation of the response, respectively. [175]The sensitivity and response were significantly improved after doping of oxygen because the oxygen accelerated the process of mass transfer and charge transfer during gas detection.As shown in Figure 6j, the sensor exhibited a linear relationship between gas concentration and response current from 5 to 500 ppm for TMA, with the higher the TMA concentration resulting in greater current through the sensor.This sensor demonstrated a more rapid response time and recovery time of 32 and 25 s, respectively, for 100 ppm TMA compared with 32 s and 253 s of pristine MoSe 2 .Another wearable TVB-N sensor, capable of detecting TEA gas at sub-ppm level, employed PET as the substrate and hybrid PPy/WO 3 as the sensitive layer . [176]The detection limit was lower to 5.4 ppm TEA theoretically.The formation of p-n heterojunction and complementary effect between PPy and WO 3 contributed to the improvement of sensitivity from 0.027 to 0.076 ppm −1 and the response to 100 ppm TEA as high as 680 (Response (%) = |R gas − R air |/R air ), where the sensitivity was defined as the rate between Response (%) and TEA concentration.

Flammable and Explosive Gas Sensor
Flammable and explosive gases refer to the gases that can be ignited once contact with fire, heat, or oxidizer, including hydrogen (H 2 ), carbon monoxide (CO), methane (CH 4 ), and hydrogen sulfide (H 2 S). [177]Consequently, it is essential to fabricate gas sensors that can operate at room temperature for detecting flammable and explosive gases in order to prevent the occurrence of explosion accidents.
Palladium (Pd)-based H 2 sensor is one of the most commonly used gas sensors for detecting H 2 owing to its low powerconsumption and high sensitivity to H 2 at room temperature.A typical example of Pd-based H 2 sensors is the Pd-based sensor functionalized by Ti 3 C 2 T x . [178]The sensitivity of this H 2 sensor was 41.7 ± 2.1% when exposed to 40% H 2 , where the sensitivity was defined as ΔI/I 0 ×100 (%), and ΔI and I 0 represented the current change after introducing H 2 and the initial current in air, respectively.Figure 6k displayed an eximious sensing performance of the Pd/Ti 3 C 2 T x -based sensor with an obvious response signal when the sensor was exposed to 0.5% H 2 at room temperature.The response signal and sensitivity intensified gradually transients of TiO 2 -based sensor to 5 ppm formaldehyde under various bending conditions at 23°C.i) Flexibility of TiO 2 -based sensor.Reproduced with permission. [141]Copyright 2021, Springer Nature.j) Real-time response of MoSe 2 -200 to various concentrations of TMA in ppm level.Reproduced with permission. [175]Copyright 2020, Springer Nature.k) Real-time response/recovery curves of MXene@Pd CNC film to high concentration of H 2 (from 0.5 to 40 v/v%).Reproduced with permission. [178]Copyright 2020, Elsevier.l) Time-dependent response of the sensor to 1% O 2 in three experimental cycles.m) Dynamic response curves to 1% O 2 at 0%, 25%, 50%, and 100% tensile strains.n) Quantitative responses to 1% O 2 in the states of pristine, self-healed, 180°bending, 25%, 50%, and 100% tensile strains.Reproduced with permission. [181]Copyright 2022, Springer Nature.
as the concentration of H 2 increased, until the concentration of H 2 reached 40%.Notably, the sensor was flexible and could be bent from 0°to 90°without altering the sensor sensitivity.What's more, as CH 4 is a flammable and explosive gas, it is of critical importance to detect it at room temperature in the industrial production process.A WO 3 /CNFs-based CH 4 sensor was designed to detect 10-250 ppm CH 4 , providing rapid response and recovery within 200 and 40 s, respectively.The sensitivity was 8.25% for 250 ppm CH 4 , which was improved by six times compared with pure CNFs, and the sensitivity was defined as ∆R/R 0 × 100% (where ∆R and R 0 were the same as those mentioned previously).
CO is an especially dangerous flammable and explosive gas because it is colorless and odorless, making it difficult to be detected immediately when it leaks. [179]A wearable CO gas sensor based on a hybrid graphene/zinc oxide (ZnO) coated on cotton fabrics that operated at room temperature has been reported. [180]The addition of GO effectively improved the sensing performance of ZnO by enabling the rapid electron transfer from graphene to ZnO.As a result, the sensor exhibited a broad detection range from 10 to 90 ppm and excellent selectivity while the response to CO was 40.26% higher than that of NO, ethanol, acetone, and methanol (i.e., 5.48%, 1.22%, 0.9%, and 0.45%).At the same time, it also exhibited a low actual detection limit of < 10 ppm at room temperature.

Oxygen
Detecting oxygen concentration is of particular importance in many branches, including medical, industry and daily life applications.As a result, of a portable O sensor has become increasingly appealing.However, the majority of oxygen sensors usually need to operate at high temperature, which blocks the development of O 2 sensors for use in wearable electronic devices.Consequently, the primary problem that needs to be solved is the operation temperature in the fabrication process of wearable oxygen sensors.
Liang et al. [181] employed a polyacrylamide-chitosan (PAM-CS) organohydrogel as the sensitive layer to fabricate an O 2 sensor with exceptional sensitivity at room temperature.The theoretical limit of detection was down to 5.7 ppm at room temperature, where the detection limit was defined as 3.3 /S, with  and S having the same representation as mentioned previously.Figure 6l demonstrated the ultrahigh response of up to 3200 to 1% O 2 and excellent repeatability.The unique structure of the organohydrogel enabled the sensor to work under mechanical deformations, including stretching and bending, as shown in Figure 6m,n.Besides, a room temperature ionic liquid-based gas sensor has also been reported for O 2 sensing. [182,183]It was reported that the sensitivity for O 2 could reached 0.2%/ppm, where the sensitivity was defined as the ratio between ∆I/I 0 ×100% and O 2 concentration in ppm.

The Applications of Wearable Gas Sensors
Compared with traditional rigid gas sensors, wearable gas sensors offer more flexible installation options, such as being worn on the human body or integrated in clothes, rather than being limited to stationary locations by rigid sensors.Wearable gas sensors can be seamlessly adhered on human skin or integrated in clothes as electronic fabric, which can bend freely with human movements without altering the gas sensing performance.Importantly, wearable gas sensors can immediately perceive surrounding environmental information and physiological signals of the human body, and send warnings at the first sign of harmful gas concentrations or breathing markers exceeding the health standard, by the use of IoTs or deep learning related technologies.The comfort and portability of wearable gas sensors make it possible to wear and monitor all day in real time.Consequently, wearable gas sensors have been found to play an essential role in environmental monitoring, [184] healthcare, [185] smart home, [186] food safety monitoring, [187] and even public security, [188] where the real-time monitoring and immediate warnings are required.

Environmental Monitoring
The emission of hazardous gases into the environment is known to cause damage to the ecological system, which can slow down the growth speed of forests and crops, cause water acidification and threaten the survival of both humans and other organisms.Additionally, the hazardous gases can also irritate the respiratory tract and even caused pneumonia.To address this issue and protect human health, wearable hazardous gas sensors have been developed to monitor the ambient air quality.
Palomeque-Mangut et al. [189] have designed and fabricated an air quality monitoring platform that combined a metal oxide semiconductor (MOS) gas sensor for perceptual information, a PM sensor and a smartphone that used Bluetooth Low Energy (BLE) communication to gather information.Meanwhile, a smartphone application called aQtracer employed Flutter framework and managed the operating system by plugins such as GPS, Bluetooth, or internet connection, as shown in Figure 7a.The collected data was transmitted to a smartphone terminal and uploaded to the cloud through the internet for machine learning and model prediction, such as using multi-layer perceptron (MLP) feedforward artificial neural network (ANN).In addition, a smart face mask (SFM) has been designed to identify diverse gases for environmental monitoring as well. [190]The SFM integrated different gas sensors with three different color LEDs (red green and blue) respectively, a power supply system and an integrated circuit, as demonstrated in Figure 7b.When the target gases were introduced, the conductivity of sensor varies, resulting in the emission of various light colors to warn the user, which exhibited great potential for monitoring the environmental gases.

Healthcare
At present, exhaled gas detection is an effective and common noninvasive diagnostic method to diagnose certain diseases and provide treatment options timely.For example, the diabetes and neonatal jaundice can be diagnosed by detecting the concentration of acetone and CO, respectively [191] and the concentration of NO in exhalation is related to the bronchial asthma. [192]or the diagnosis of diabetes, Jaisutti et al. [193] demonstrated room-temperature gas sensors with a sensing resolution down to  [189] Copyright 2022, Elsevier.b) The picture of multi-functional face mask with integrated wearable gas sensors.Reproduced with permission. [190]Copyright 2017, Springer Nature.c) Schematic to illustrate the deep-learning-assisted adaptive respiratory monitoring system.Reproduced with permission. [195]opyright 2022, Wiley-VCH.d) Response/recovery curves of the NH 3 sensor with the introduction of healthy person's breath and the simulated breath of a patient with H. pylori infection.Reproduced with permission. [58]Copyright 2022, Elsevier.e) Schematic of the chamber for testing NH 3 gas sensor with the existence of spoilage meat.Reproduced with permission. [196]Copyright 2017, Springer Nature.f) Experiment setup of the MSS 2 used to monitor the spoilage fish and beef meat at room temperature.Reproduced with permission. [197]Copyright 2023, Wiley-VCH.g) Modification of NFC tag of turn-on sensor.RIC, C, L, and R denote the integrated circuit, capacitor, antenna, and resistor, respectively.Reproduced with permission. [198]Copyright 2016, American Chemical Society.sub-ppm level, which was sufficient to distinguish the diabetes patients and healthy individuals.Su et al. [194] proposed a novel wirelessly powered strategy based on both chemisorption and triboelectrification to diagnose diabetes non-invasively, which further upgraded wearable gas sensors.Once exhaled, the polytetrafluoroethylene (PTFE) membrane started vibration and was separated from the nylon film under airflow to output electric signals for the acceleration of chemical detection of acetone.To improve the accuracy of diagnosis, Fang et al. [195] framed an onmask sensor network to collect more accurate exhaled breath signals using deep-learning and IoT technology.As demonstrated in Figure 7c, the signal of exhaled breath collected by the gas sensor was amplified and filtered by an amplifier and low-pass filter, respectively.The exhaled information was then extracted from data features and automatically recognized by deep learning of a three-layer 1D-CNN.Finally, it contributed to HMI through a Bluetooth module and smartphone application, achieving noncontact disease diagnosis.
Additionally, detecting the concentration of NH 3 can also be used to diagnose the infection of Helicobacter pylori.Figure 7d depicted the different response signals between healthy and infected individuals.The exhalation of an infected person caused a noticeable signal fluctuation of the gas sensor due to the excess NH 3 in their exhaled breath, while that of healthy individual exhibited a smooth response curve. [58]

Industrial Security
The emission of industrial by-products like CO and SO 2 , as well as toxic raw materials and auxiliary materials, can have serious health implications for industrial workers.It is reported that an increasing number of humans are died from inhaling harmful gases each year.With the rapid fusion of wearable technology and the IoT, an increasing number of researchers are exploring the application of gas sensors for detecting the leakage of toxic gases during industrial production.The collected information is transmitted to intelligent data terminals.Once the detected gas concentration exceeds the threshold, the alarms activate through the IoT to protect humans working in the industrial environment.Swager's team [198] from Massachusetts Institute of Technology conducted a thorough investigation into gas sensing technology for detecting the leakage of SOCl 2 during the industrial process of using Li-SOCl 2 backup batteries.Their efforts resulted in the development of a wireless gas sensing near-field communication (NFC) tag that enabled non-contact detection by a smartphone.Figure 7g demonstrated that the NFC switched between readable and unreadable states when the NFC antenna (L) was in series with a resistance > 2.2 or < 1.0 kΩ, respectively.First, the NFC integrated with SOCl 2 sensors with 3.6 kΩ resistance exhibited unreadable.However, with the introduction of SOCl 2 , the resistance dramatically decreased to 0.79 kΩ and was then able to be read by a smartphone.In this instance, Ganeshan et.al [199] proposed an intelligent system that integrated monitoring, control, and protection functions.This system not only monitored the surrounding environment through gas sensors, but also utilized the ZigBee module for data communication.Different from other research, this system could store monitoring logs in data terminals and upload above information to the cloud.In addition, the researchers also utilized the CNN algorithm to identify and update relevant information immediately.

Food Safety
The consumption of spoiled food and food contaminated with pesticides, bacteria, fungi and viruses can lead to food poisoning, causing gastrointestinal discomfort, fever, and even death. [200]To prevent these potential risks, portable wearable gas sensors worn on the human body have been developed to detect the freshness of food according to the emission gases produced by the food.Thus, exploring the potential applications of wearable gas sensors in food safety has become an urgent priority to avoid the fatalities and injuries caused by foodborne diseases.
It is widely recognized that protein-rich meat releases nitrogenous gases, such as NH 3 , during the spoilage process.Thus, the concentration of NH 3 can serve as a marker to monitor the freshness of meat. [201]Matindoust et al. [196] fabricated a PANI-based NH 3 sensor, which was calibrated by a commercial gas sensor named TGS 826. Figure 7e demonstrated the examination and correction process of the PANI-based sensor, showing that it displayed a higher sensitivity than the commercial sensor.As NH 3 was released from spoilage meat, the change of current was measured by PicoLog technological software, which exhibited a functional relation between NH 3 concentration and current change.In addition to NH 3 , H 2 S is also produced in the spoilage process of protein-rich foods and monitoring the concentration of H 2 S can be used to detect the freshness of meat as well. [187,202]igure 7f illustrated the sensing performance in the process of spoilage, where the modified-sonicated sensors (MSS), a commercial gas sensor from Figaro, a datalogger and food were set up in the gas chamber. [197]The MSS exhibited high specific recognition and could be integrated with IoT to achieve long-range monitoring of food freshness.
Compared with protein-rich food, vegetables, and fruit produce diverse product gases in the spoilage process, such as alkenes, esters, aldehydes and alcohols.Moreover, sugar-rich fruits often produce large quantities of alcohol. [203,204]Hassan et al. [170] combined the fractal structured alcohol sensor and sen-sor data processing/transmission module for real-time monitoring the freshness of strawberry, as shown in Figure 8i. Figure 8a,b demonstrated the solid performance of this device during the monitoring process.The response curve displayed a smooth transition when the strawberry was fresh, but once it was spoilage, the response as a resistance variation changed dramatically.This device paves the way for improving the efficiency of monitoring food freshness and ensuring food safety.

Smart Home
With the advancement of science and technology and the rapid improvement of people's living standard, the emergence and rapid development of smart home is inevitable.As the important source of perceptual information, the environmental information anywhere in the home can be detected by gas sensors that move with humans.Compared to gas sensors placed in fixed positions, gas sensors worn on humans can eliminate the need to install multiple sensors in the home, allowing a single sensor to ensure the safe of family members and property.Wearable gas sensors can flexibly detect indoor conditions in a smart home system, which has been gradually optimized by researchers in recent years using not only wearable gas sensors but also IoT technology.
For example, Song et al. [205] built an intelligent gas sensor network for real-time monitoring and automatic alarm of flammable gas leakage in households.By using machine learning algorithms like Principal component analysis (PCA) and Support Vector Machine (SVM), the self-powered integrated nanostructuredgas-sensor (SINGOR) systems could accurately identify different types of gas.Additionally, the signals collected by the SINGOR system were transmitted to a smartphone via Bluetooth, enabling the visual indoor condition monitoring in long-range, as demonstrated in Figure 8c.Meanwhile, Figure 8d showed another function of the system, where the locations of gas leakage were rapidly identified using a gas diffusion model and localization algorithm based on the locations of the intelligent sensor network.Furthermore, Hsu et al. [186] fused multiple sensors, including gas sensors and motion sensors, with wearable intelligent technology and AI to build a multifunctional intelligent smart home system.the system's intelligent fire detection was operated by detecting CO concentration.The obtained signals were classified by a probabilistic neural network (PNN) to monitor the indoor condition, as shown in Figure 8e.If the information processing module detected dangerous or alarming conditions, the exhaust fan would be opened under control commands from the decision module.

Public Security
As the number of motor vehicles surges, the incidents of drunk driving have become a serious criminal activity that endangers people's safety.At the same time, terrorism, explosions, and toxic gas leakages have seriously endangered lives and caused property damage as well. [208,209]The gas sensors worn on security personnel can find mobile threats immediately by monitoring the toxic or flammable gases, thereby preventing the safety accidents.The gas sensors worn on drivers can monitor their alcohol Reproduced with permission. [170]Copyright 2021, American Chemical Society.c) Visual real-time gas monitoring with the mobile app.d) Four SINGOR systems deployed at the corners of the kitchen and the locations of gas leakage source with red circle flashing shown in the mobile app.Reproduced with permission. [205]Copyright 2021, American Chemical Society.e) Block diagram of the intelligent fire detection and alarm algorithm.Reproduced with permission. [186]Copyright 2017, MDPI.f) Real-time alcohol concentration analysis/display in the case of unknown gas concentration (no gas, low, middle and high concentrations).Reproduced with permission. [206]Copyright 2017, Elsevier.g) User interfaces of Drunk Mate, consisting of Blynk IoT platform and the LINE Notify messaging platform for real-time alarming notification.h) Schematic illustration of the smart wristband for real-time alcohol monitoring.Reproduced with permission. [207]Copyright 2022, MDPI.i) Optical image of the Hilbert sensor on the strawberry box connected to a sensor module.Reproduced with permission. [170]Copyright 2016, American Chemical Society.concentration in real-time and send the information to the traffic management department through the internet to prevent the occurrence of drunk driving accidents in time.And the possibility of wearing these sensors all day is attributed to their comfortable design, ensuring that the wearers can use them without discomfort.
It is apparent that the development of portable wearable gas sensors for detecting drunk driving, and explosive/toxic gases at room temperature holds significant potential.
Li et al. [206] fabricated a wearable in situ integrated analysis system, wherein the flexible alcohol sensor was integrated on a PCB.
The electric signal was rapidly amplified by a filter and then collected by a Microcontroller (MCU).The signal was then transmitted wirelessly to a smartphone by Bluetooth, enabling long-range visualization of detecting drunk driving, as shown in Figure 8f.Besides, Khemtonglang et al. [207] designed an intelligent wristband with an alarm function for remote drunk driving detection, as demonstrated in Figure 8g.The obtained output signal was uploaded to Blynk application, as shown in Figure 8h, through IoT technology and was converted into visual information on the smartphone.The collected signal was classified into different alarming levels by programming calculation, providing a warning as necessary.These remote detection methods for drunk driving enable traffic management departments to stop drunk driving crimes in real-time.Furthermore, the detection of explosives is equally important.Chen et al. [210] developed a TNT sensor with the sensing resolution down to 8 ppb, which was close to the TNT threshold set by the U.S. Occupational Safety and Health Administration.

Conclusion
Wearable gas sensors that work at room temperature have been explored widely as a main source of perceptual information in the information era.Based on the present research situation and prospects, of the integration of flexible gas sensors that work at room temperature in wearable intelligent systems holds great potential to achieve HMI, healthcare, and environmental detection.As demonstrated herein, several sensitive materials, such as ions liquids, conductive polymers, carbon-based nanomaterials and 2D semiconductors, have been employed to provide effective solutions to obtain ultrahigh sensing performance and mechanical flexible in gas sensors that can operate at room temperature without extra heating.Different gas sensing mechanisms have been unraveled by four kinds of sensing models, including oxygen adsorption, direct electron transfer, proton transfer, and ions conduction models, which explain the potential change due to the different interactions between diverse target gases and sensitive layers.
However, despite numerous restrictions accompanied with new opportunities still persist, and a large number of problems must be addressed before these technologies can be implemented in practical applications.First, sensitive materials used in wearable gas sensors often exhibit lower response and longer response/recovery times compared with those that work at high temperatures.Thus, it is urgent to study the doping and structure of existing sensitive materials to enhance their sensing performance at room temperature.In addition, the stability of roomtemperature sensing materials, such as Ti 3 C 2 T x and PANI, is relatively poor as they are prone to oxidation and have poor longterm stability, limiting their further application in wearable gas sensors.Second, in terms of device manufacturing, it is vitally important to ensure the stability of gas sensing performance under large mechanical deformations and achieve miniaturization and integration to meet the requirements of wearable intelligent devices.Finally, developing the practical applications of gas sensors that are combined with big data, AI, machine learning, etc., is a brilliant research direction, which can accelerate the intelligence, integration, and digitization of wearable gas sensors.
Geyu Lu is a professor, doctoral supervisor, and dean of the College of Electronic Science and Engineering of Jilin University.His research mainly focused on gas sensors, advanced sensing materials, and high-performance sensing systems.

Figure 1 .
Figure 1.The fabrication and application of wearable gas sensors working at room temperature.

Figure 2 .
Figure 2. a) Schematic diagram of the NO 2 detection process demonstrated by the oxygen adsorption model.Reproduced with permission.[36]Copyright 2021, Wiley-VCH.b) Diagram of In 2 O 3 /g-C 3 N 4 heterojunction and the corresponding energy band.Reproduced with permission[36] Copyright 2021, Wiley-VCH.c) DOS of the ternary-NCs after NH 3 adsorption.Reproduced with permission.[46]Copyright 2022, American Chemical Society.d) Schematic of the NH 3 gas sensing mechanism illustrated by the direct electron transfer model.Reproduced with permission[46] Copyright 2022, American Chemical Society.e) Schematic illustration of the sensing mechanism of the flexible GP-PANI/PVDF sensor.Reproduced with permission.[52]Copyright 2021, Elsevier.

Figure 4 .
Figure 4. a) Schematic of the as-fabricated MSC integrated with the MWCNI/PANI alcohol gas sensor.Reproduced with permission.[33]Copyright 2017, Elsevier.b) Schematic of the spray deposition of MWCNTs on the fabric (F-MWCNTs) and PANI synthesis on MWCNTs coated fabric surface.Reproduced with permission.[73]Copyright 2017, Elsevier.c) Schematic illustration of Ti 3 C 2 T x synthesis procedure, electrode sputtering, solution deposition process, and gas-sensing device.Reproduced with permission.[85]Copyright 2017, American Chemical Society.d) Schematic diagram of the fabrication process of rGO/ZnO hybrid fibers adopting APTES as the molecular glue.Reproduced with permission.[86]Copyright 2019, American Chemical Society.e) Schematic illustration of the spinning process of MXene/GO hybrid fiber.Reproduced with permission.[87]Copyright 2020, American Chemical Society.

Figure 6 .
Figure6.a) Response/recovery curve of H30-CeO 2 @PANI for low concentration NH 3 .Reproduced with permission.[153]Copyright 2022, American Chemical Society.b) Sensitivity curve of CTF-based gas sensor.Reproduced with permission.[163]Copyright 2020, American Chemical Society.c) Ultrahigh selectivity to NH 3 of the porous CuBr-based NH 3 sensor (A: NH 3 , N: NO 2 , T: acetone, C: CO, E: ethanol, D: acetaldehyde, and F: formaldehyde).Reproduced with permission.[155]Copyright 2022, The Royal Society of Chemistry.d) Sensing response curves of ZIC to different concentrations of NO 2 (50 ppb to 5 ppm).Reproduced with permission.[139]Copyright 2021, Wiley-VCH.e) Response characteristics fitting curves of the sensors based on MXene/Co 3 O 4 to 0.01-10 ppm HCHO at 25°C.f) Selectivity of the sensors based on MXene/Co 3 O 4 to various testing gases with a concentration of 10 ppm.Reproduced with permission.[113]Copyright 2021, Elsevier.g) Gas response as a function of formaldehyde concentration.h) Gas-sensing

Figure 7 .
Figure 7. a) Block diagram with the operating scheme (left) and the screen capture (right) of the developed app.Reproduced with permission.[189]Copyright 2022, Elsevier.b) The picture of multi-functional face mask with integrated wearable gas sensors.Reproduced with permission.[190]Copyright 2017, Springer Nature.c) Schematic to illustrate the deep-learning-assisted adaptive respiratory monitoring system.Reproduced with permission.[195]Copyright 2022, Wiley-VCH.d) Response/recovery curves of the NH 3 sensor with the introduction of healthy person's breath and the simulated breath of a patient with H. pylori infection.Reproduced with permission.[58]Copyright 2022, Elsevier.e) Schematic of the chamber for testing NH 3 gas sensor with the existence of spoilage meat.Reproduced with permission.[196]Copyright 2017, Springer Nature.f) Experiment setup of the MSS 2 used to monitor the spoilage fish and beef meat at room temperature.Reproduced with permission.[197]Copyright 2023, Wiley-VCH.g) Modification of NFC tag of turn-on sensor.RIC, C, L, and R denote the integrated circuit, capacitor, antenna, and resistor, respectively.Reproduced with permission.[198]Copyright 2016, American Chemical Society.

Figure 8 .
Figure8.a) Real-time monitoring of alcohol released from the strawberry during day 1 (fresh strawberry) and b) day 7 (rotten strawberry).Reproduced with permission.[170]Copyright 2021, American Chemical Society.c) Visual real-time gas monitoring with the mobile app.d) Four SINGOR systems deployed at the corners of the kitchen and the locations of gas leakage source with red circle flashing shown in the mobile app.Reproduced with permission.[205]Copyright 2021, American Chemical Society.e) Block diagram of the intelligent fire detection and alarm algorithm.Reproduced with permission.[186]Copyright 2017, MDPI.f) Real-time alcohol concentration analysis/display in the case of unknown gas concentration (no gas, low, middle and high concentrations).Reproduced with permission.[206]Copyright 2017, Elsevier.g) User interfaces of Drunk Mate, consisting of Blynk IoT platform and the LINE Notify messaging platform for real-time alarming notification.h) Schematic illustration of the smart wristband for real-time alcohol monitoring.Reproduced with permission.[207]Copyright 2022, MDPI.i) Optical image of the Hilbert sensor on the strawberry box connected to a sensor module.Reproduced with permission.[170]Copyright 2016, American Chemical Society.

Table 1 .
The selection of flexible sensitive materials and substrates, and the gas sensing performances of common wearable gas sensors that work at room temperature.