Recent Progress of Toxic Gas Sensors Based on 3D Graphene Frameworks

Air pollution is becoming an increasingly important global issue. Toxic gases such as ammonia, nitrogen dioxide, and volatile organic compounds (VOCs) like phenol are very common air pollutants. To date, various sensing methods have been proposed to detect these toxic gases. Researchers are trying their best to build sensors with the lowest detection limit, the highest sensitivity, and the best selectivity. As a 2D material, graphene is very sensitive to many gases and so can be used for gas sensors. Recent studies have shown that graphene with a 3D structure can increase the gas sensitivity of the sensors. The limit of detection (LOD) of the sensors can be upgraded from ppm level to several ppb level. In this review, the recent progress of the gas sensors based on 3D graphene frameworks in the detection of harmful gases is summarized and discussed.


Introduction
There is a huge demand for the development of simple and reliable gas sensors [1]. In many fields, such as agriculture, medical diagnosis, and industrial waste, especially in environmental monitoring, it is necessary to detect NO x (especially NO 2 ), ammonia (NH 3 ), and volatile organic compounds (VOCs), because of their possible toxicity and related risks to the ecosystem [2,3]. In many countries, air pollution is a major environmental problem caused by rapid industrialization. A large amount of NO 2 is emitted into the environment every year due to the industrial combustions and automobile emissions [4]. Therefore, the detection of NO 2 has aroused widespread concerns, because it is harmful to the plants and respiratory systems of people and animals [5]. Additionally, NO 2 can cause acid rain and photochemical smog [6,7]. Therefore, the United States Environmental Protection Agency (EPA) defines NO 2 as a typical air pollutant, and the exposure limit is only 53 ppb [8]. Ammonia (NH 3 ) is also a common dangerous air pollutant, which is produced by the industrial process, agricultural production, and manufacturing process [9,10]. Specifically, any overexposure to the high concentrations of NH 3 (>30 ppm, 10 min) can irritate the human eye, skin, and respiratory system [11][12][13]. VOCs are the hydrocarbons that exist as gases or vapor at room temperature, which can be emitted from numerous products and activities, e.g., detergents, paints, solvents, tools, clothes, toys, cleaning, and cooking [14]. Aldehyde, aromatic, aliphatic, halogenated, and terpenoid compounds are the VOCs commonly detected in commercial buildings [14,15]. Toxic VOCs that have been previously detected in air by any type of sensors include formaldehyde, acetaldehyde, benzene, toluene, xylenes, phenol, pyridine, acetone, acetic anhydride, carbon disulfide, dihydroxybenzene, and so on [14][15][16][17][18][19]. For example, phenol is a toxic VOC occurring both naturally but also from industrial processes, which can be rapidly absorbed through the skin and cause skin and eye burns upon contact [20]. It is considered as a serious pollutant because of the toxicity and persistence in the environment. The short-term exposure limit of phenol is 10 ppm, 60 min [21]. Because of the serious environmental pollution, phenol (2) Chemical reduction, which is beneficial for large-scale production, and various reducing agents can be selected [56][57][58][59][60][61][62][63][64][65]. (3) Electrochemical reduction [66][67][68]. The hydrogel prepared by this method is applied to the electrode surface and can be directly applied to the electrode materials of electrochemical instruments. (4) Vacuum filtration. A simple vacuum filtration method was developed to prepare RGO hydrogels with high conductivity, anisotropy, and responsive stimuli [69,70]. 73]. Graphene aerogel composites are usually prepared by supercritic drying of hydrogel precursors [74,75]. For example, highly compres can be obtained by freeze-drying and microwave treatment. Direction known processing technology of porous materials. This technology c the preparation of graphene aerogels [76]. Moreover, the controllable nology can also reduce GO to RGO and restore conductivity. The regu ical structure of GO can adjust the morphology and elasticity of aerog oxygen functional groups in GO have a significant effect on the morph of the gels [77].

NO2 Gas Sensors
The development of a highly selective NO2 gas sensor with ppb important requirement for continuous environmental monitoring [78] of respiratory diseases [79]. However, the original graphene, RGO, a ited reaction to NO2, and could not monitor NO2 gas below 100 ppb searchers have explored a variety of methods to fabricate graphene pared with 2D RGO, 3D RGO is more sensitive to NO2 [81]. Table 1 lis ties of 3D graphene toward NO2. The response is generally given in In addition to the 3D self-assembly of graphene in a water system, the assembly of the graphene in an organic system can also be achieved by thermal solvent reduction [71][72][73].
Graphene aerogel composites are usually prepared by supercritical drying or freezedrying of hydrogel precursors [74,75]. For example, highly compressible RGO aerogels can be obtained by freeze-drying and microwave treatment. Directional freezing is a well-known processing technology of porous materials. This technology can also be used for the preparation of graphene aerogels [76]. Moreover, the controllable heat treatment technology can also reduce GO to RGO and restore conductivity. The regulation of the chemical structure of GO can adjust the morphology and elasticity of aerogels, for example, the oxygen functional groups in GO have a significant effect on the morphology and elasticity of the gels [77].

NO 2 Gas Sensors
The development of a highly selective NO 2 gas sensor with ppb detection limit is an important requirement for continuous environmental monitoring [78] and early diagnosis of respiratory diseases [79]. However, the original graphene, RGO, and RGOH had limited reaction to NO 2 , and could not monitor NO 2 gas below 100 ppb [80]. Therefore, researchers have explored a variety of methods to fabricate graphene composites. Compared with 2D RGO, 3D RGO is more sensitive to NO 2 [81]. Table 1 lists the gas sensitivities of 3D graphene toward NO 2 . The response is generally given in the form of relative change of resistance (∆R/R 0 ) or conductance (∆G/G 0 ), which was tested at a set temperature and a set concentration of NO 2 . S 3D and S 2D are the sensitivities (responses per ppm) of 3D and 2D graphene, respectively. The graphene with a 3D structure can increase the gas sensitivity to one or two orders of magnitude higher. The LOD of the sensors can be upgraded from ppm level to several ppb level.

3D Graphene
Temp. Note: The response, in the form of relative change of resistance (∆R/R 0 ) or conductance (∆G/G 0 ), was tested at a set temperature and a set concentration of NO 2 (C NO2 ). S 3D and S 2D are the sensitivities (responses per ppm) of 3D and 2D graphene, respectively. LOD = limit of detection.

3D MoS 2 /RGO
2D-layered MoS 2 is an excellent gas sensing material due to its high surface/volume ratio and excellent electronic properties [90][91][92][93][94]. However, MoS 2 nanoparticles tend to agglomerate, which limits their applications. Loading MoS 2 on 3D RGO is a good choice [95][96][97]. Chen et al. [52] reported a highly sensitive NO 2 sensor based on 3D MoS 2 /RGO composites. The composites were prepared using a novel self-assembly and hydrothermal method. The mild synthesis process enables MoS 2 to uniformly disperse on the 3D RGO framework. The agglomeration of MoS 2 was significantly alleviated, resulting in excellent low-temperature sensing performance. Figure 2 shows the fabrication process, gas responses, and sensing mechanism of the 3D MoS 2 /RGO sensor. The selectivity of the fabricated sensors was characterized using a variety of independent gases. The device shows higher sensing response towards N-based molecules (e.g., NO 2 , NH 3 ) than the other gases (CO, C 2 H 5 OH, H 2 , HCHO). When the sensor is exposed to NO 2 atmosphere, the minority charge carriers (electrons) in the MoS 2 /rGO composites will transfer to NO 2 due to the strong electron negativity of NO 2 . Thus, the width and height of the heterojunction are reduced, leading to the increased conductivity. A small variation in barrier height and width caused by gas adsorption or desorption can have a significant influence on the resistance. A superior low-temperature NO 2 sensing performance with a response of 2483% toward 10 ppm NO 2 was achieved at 80 • C.

D SnS2/RGO
Different from p-type RGO, SnS2 is an n-type semiconductor with an indirect band gap of 2.2 ev [98]. SnS2 has a high affinity to NO2 because of its much weaker electronegativity than NO2 [99,100]. However, the resistance of SnS2 is too high to be measured at room temperature, so SnS2 is not suitable for monitoring NO2 at room temperature. When SnS2 is coupled with 3D graphene, their properties are complementary to each other. Wu et al. [86] demonstrated a kind of 3D structured SnS2/RGO heterojunction, which was synthesized through a facile hydrothermal route. Figure 3 shows the microstructure, gas responses, and sensing mechanism of the SnS2/RGO sensor. The 3D structure enhances the adsorption and diffusion of small NO2 molecules. SnS2 can facilitate the electron transfer

3D SnS 2 /RGO
Different from p-type RGO, SnS 2 is an n-type semiconductor with an indirect band gap of 2.2 ev [98]. SnS 2 has a high affinity to NO 2 because of its much weaker electronegativity than NO 2 [99,100]. However, the resistance of SnS 2 is too high to be measured at room temperature, so SnS 2 is not suitable for monitoring NO 2 at room temperature. When SnS 2 is coupled with 3D graphene, their properties are complementary to each other. Wu et al. [86] demonstrated a kind of 3D structured SnS 2 /RGO heterojunction, which was synthesized through a facile hydrothermal route. Figure 3 shows the microstructure, gas responses, and sensing mechanism of the SnS 2 /RGO sensor. The 3D structure enhances the adsorption and Sensors 2021, 21, 3386 6 of 18 diffusion of small NO 2 molecules. SnS 2 can facilitate the electron transfer from RGO to NO 2 by forming a heterojunction with RGO. The depletion layer with an electrostatic field at the p-n heterojunction region could promote the dissociation of NO 2 and thus enhance the NO 2 adsorption at low temperatures. Upon NO 2 adsorption on SnS 2 , the electron will transfer from SnS 2 to NO 2 , leading to the increase both of the electron-depletion region and the hole concentration of the p-type RGO. Thus, the resistance of SnS 2 /RGO is reduced. The sensor displays impressive NO 2 sensing performance, including high sensitivity (6.1 ppm −1 ), low LOD (8.7 ppb), good linearity, and reversibility.
Sensors 2021, 21, x FOR PEER REVIEW 6 from RGO to NO2 by forming a heterojunction with RGO. The depletion layer wit electrostatic field at the p-n heterojunction region could promote the dissociation of and thus enhance the NO2 adsorption at low temperatures. Upon NO2 adsorption on the electron will transfer from SnS2 to NO2, leading to the increase both of the elec depletion region and the hole concentration of the p-type RGO. Thus, the resistan SnS2/RGO is reduced. The sensor displays impressive NO2 sensing performance, inc ing high sensitivity (6.1 ppm −1 ), low LOD (8.7 ppb), good linearity, and reversibility.

D Eu(TPyP)(Pc)/RGO
Pure RGO sensors have poor selectivity, slow response, and long recovery which limits their wide applications [89]. Chemical/physical modification with the e nal groups or atoms is an effective method [101,102]. Zhu et al. [88] reported a sandw type double-decker complex Eu(TPyP)(Pc) (TPyP = meso-tetra(4-pyridyl)porphyrin; phthalocyanine), which was in situ self-assembled on the surface of RGO driven b π-π interaction, forming a 3D RGO/Eu(TPyP)(Pc) hybrid aerogel. Eu(TPyP)(Pc) not acts as a sensor recognition unit, but also helps to enhance the amplification effect o

3D Eu(TPyP)(Pc)/RGO
Pure RGO sensors have poor selectivity, slow response, and long recovery time, which limits their wide applications [89]. Chemical/physical modification with the external groups or atoms is an effective method [101,102]. Zhu et al. [88] reported a sandwich-type doubledecker complex Eu(TPyP)(Pc) (TPyP = meso-tetra(4-pyridyl)porphyrin; Pc = phthalocyanine), which was in situ self-assembled on the surface of RGO driven by the π-π interaction, forming a 3D RGO/Eu(TPyP)(Pc) hybrid aerogel. Eu(TPyP)(Pc) not only acts as a sensor recognition unit, but also helps to enhance the amplification effect of the p-n heterojunc-tion. At the same time, it provides enough space for the efficient transmission of RGO. The resulting aerogel not only effectively integrates the gas sensing of Eu(TPyP)(Pc) and good conductivity of RGO, but also exhibits a prominent synergy effect. Figure 4 shows the microstructure, responses, and working mechanism of the RGO/Eu(TPyP)(Pc) sensor. A good linear ratio between signal response and target gas concentration is achieved in the range of 0.5-20 and 20-100 ppm.

3D SnO 2 /RGOH
Many studies have been devoted to improving the gas sensitivity by immobilizing SnO 2 crystals on RGO [103,104]. Li and co-workers used SnO 2 nanocrystals supported by the 3D mesoporous graphene aerogels to detect NO 2 gas at low temperature [84]. Wu et al. [85] reported a facile preparation of SnO 2 -modified graphene hydrogel (SnO 2 /RGOH) via the one-step hydrothermal method. 3D SnO 2 /RGOH was synthesized directly from Sn 2+ and GO precursors without any surfactant. The results show that it is feasible to optimize the gas sensing performance by combining reasonable material hybridization, 3D structure, and temperature modulation. The SnO 2 /RGO hybrid showed a high sensitivity to NO 2 [105]. The improved sensitivity is due to the agglomeration of SnO 2 nanoparticles and the formation of p-n heterojunction at the interface between SnO 2 and RGO [106]. The p-n heterojunction formed at the interface of RGOH and SnO 2 promotes the charge transfer. A micro-heater is integrated on the other side of the substrate to increase the substrate temperature locally, so as to suppress the interference of humidity in NO 2 sensing. Figure 5 shows the microstructure, gas responses, and selectivity of the flexible SnO 2 /RGOH sensor. When exposed to 0.5-5 ppm NO 2 , SnO 2 /RGOH showed an immediately increased conductivity.

D SnO2/RGOH
Many studies have been devoted to improving the gas sensitivity by immobiliz SnO2 crystals on RGO [103,104]. Li and co-workers used SnO2 nanocrystals supported the 3D mesoporous graphene aerogels to detect NO2 gas at low temperature [84]. Wu al. [85] reported a facile preparation of SnO2-modified graphene hydrogel (SnO2/RGO via the one-step hydrothermal method. 3D SnO2/RGOH was synthesized directly fr Sn 2+ and GO precursors without any surfactant. The results show that it is feasible to timize the gas sensing performance by combining reasonable material hybridization, structure, and temperature modulation. The SnO2/RGO hybrid showed a high sensitiv to NO2 [105]. The improved sensitivity is due to the agglomeration of SnO2 nanoparti and the formation of p-n heterojunction at the interface between SnO2 and RGO [106]. p-n heterojunction formed at the interface of RGOH and SnO2 promotes the charge tra fer. A micro-heater is integrated on the other side of the substrate to increase the subst temperature locally, so as to suppress the interference of humidity in NO2 sensing. Fig  5 shows the microstructure, gas responses, and selectivity of the flexible SnO2/RGOH s sor. When exposed to 0.5-5 ppm NO2, SnO2/RGOH showed an immediately increa conductivity.

D Porous B-and N-Doped RGOH
Recent studies have shown that doping is a feasible strategy to adjust the physi electronic, and chemical properties of graphene by creating a band gap [107]. Under stable adsorption configuration, different elements in graphene can exhibit different sensing behaviors due to the different adsorption energy and the distance between do atoms and gas molecules [108][109][110]. B-and N-doping can improve the selectivity of graphene-based NO2 sensor [111][112][113][114]. Wu et al. [87] reported a 3D porous B-and doped RGOH chemical resistor. B-and N-RGOH were synthesized using a hydrother self-assembly method with the aid of boric acid (H3BO3) and dicyandiamide (C2H4N

3D Porous B-and N-Doped RGOH
Recent studies have shown that doping is a feasible strategy to adjust the physical, electronic, and chemical properties of graphene by creating a band gap [107]. Under the stable adsorption configuration, different elements in graphene can exhibit different gas sensing behaviors due to the different adsorption energy and the distance between doped atoms and gas molecules [108][109][110]. B-and N-doping can improve the selectivity of the graphene-based NO 2 sensor [111][112][113][114]. Wu et al. [87] reported a 3D porous B-and N-doped RGOH chemical resistor. B-and N-RGOH were synthesized using a hydrothermal selfassembly method with the aid of boric acid (H 3 BO 3 ) and dicyandiamide (C 2 H 4 N 4 ). Figure 6 shows the fabrication process, microstructures, and gas responses of the sensors. It shows that the combination of 3D hierarchical structure and the doping of B-and N-heteroatoms can significantly improve the sensing performance. The response of B-and N-RGOH to NO 2 is more than one order of magnitude higher than that of RGOH. It is worth noting that the response of B-and N-RGOH sensors varies almost linearly with the concentration of NO 2 .
Sensors 2021, 21, x FOR PEER REVIEW 9 Figure 6 shows the fabrication process, microstructures, and gas responses of the sen It shows that the combination of 3D hierarchical structure and the doping of B-an heteroatoms can significantly improve the sensing performance. The response of B-N-RGOH to NO2 is more than one order of magnitude higher than that of RGOH. worth noting that the response of B-and N-RGOH sensors varies almost linearly wit concentration of NO2.

NH3 Gas Sensors
The ammonia (NH3) sensor is indispensable in many industries and daily life. H ever, due to its complex preparation process, strict environmental requirements, and sorption of residual ammonia molecules, the production cost is high, which hinder market acceptance [115][116][117]. The ammonia sensor might exhibit a reduced response ing the recovery process in the co-presence of ammonia and ethanol, due to the resu multistep gas adsorption and desorption processes on material surface [115]. In gen the ammonia sensing behavior of the semi-conductive graphene-based sensors is neg to that of NO2 sensing, because NH3 is the reductive gas and NO2 is the oxidative [116,117]. For the same graphene-based sensors, however, the sensing performanc ward NH3 is mostly reported as inferior to that toward NO2, for example, in 2020, W al. [89] reported a green synthesis method of 3D chemically functionalized graphen the high-performance detection of NH3 and NO2 at room temperature. They found the LOD of NH3 and NO2 were 500 and 100 ppb, respectively.

D Graphene
Generally, when a sensor is recovered at room temperature, the gas molecules ca be completely desorbed, resulting in poor stability and long recovery time [118]. Th sorption of ammonia can be promoted using the infrared light source. This is attrib

NH 3 Gas Sensors
The ammonia (NH 3 ) sensor is indispensable in many industries and daily life. However, due to its complex preparation process, strict environmental requirements, and desorption of residual ammonia molecules, the production cost is high, which hinders the market acceptance [115][116][117]. The ammonia sensor might exhibit a reduced response during the recovery process in the co-presence of ammonia and ethanol, due to the result of multistep gas adsorption and desorption processes on material surface [115]. In general, the ammonia sensing behavior of the semi-conductive graphene-based sensors is negative to that of NO 2 sensing, because NH 3 is the reductive gas and NO 2 is the oxidative gas [116,117]. For the same graphene-based sensors, however, the sensing performance toward NH 3 is mostly reported as inferior to that toward NO 2 , for example, in 2020, Wu et al. [89] reported a green synthesis method of 3D chemically functionalized graphene for the high-performance detection of NH 3 and NO 2 at room temperature. They found that the LOD of NH 3 and NO 2 were 500 and 100 ppb, respectively.

3D Graphene
Generally, when a sensor is recovered at room temperature, the gas molecules cannot be completely desorbed, resulting in poor stability and long recovery time [118]. The desorption of ammonia can be promoted using the infrared light source. This is attributed to the generation of charge carriers by absorbing infrared light [119]. Most of the others used heating to accelerate the desorption process, based on the principle of thermally excited gas molecules [120]. Wu et al. [121] used laser direct writing to fabricate three parallel porous 3D graphene lines on a polyimide (PI) tape to simply construct an ammonia gas sensor. In this study, the ammonia sensor is located in the middle and the two sides are used as heaters. Voltage can be applied to the heater to promote desorption, as shown in Figure 7. The recovery time is relatively stable with the increase of the number of cycles, indicating that the residual ammonia molecules in the element are almost fully desorbed.
to the generation of charge carriers by absorbing infrared light [119]. Most of the othe used heating to accelerate the desorption process, based on the principle of thermally e cited gas molecules [120]. Wu et al. [121] used laser direct writing to fabricate three paral porous 3D graphene lines on a polyimide (PI) tape to simply construct an ammonia g sensor. In this study, the ammonia sensor is located in the middle and the two sides a used as heaters. Voltage can be applied to the heater to promote desorption, as shown Figure 7. The recovery time is relatively stable with the increase of the number of cycl indicating that the residual ammonia molecules in the element are almost fully desorbe

D RGO/PPy
Recent studies have witnessed the significant progress of gas sensors based polypyrrole (PPy) conducting polymers [122][123][124][125]. However, the serious agglomerati of the nanoparticles leads to a significant decrease in the visible active surface of gas m ecules, which leads to the decrease of response sensitivity [126][127][128]. Qin et al. [129] us 3D RGO as a 3D skeleton support for the sensitive PPy nanoparticles. It has made an im portant contribution to improve the conductivity, dynamic performance, and sensing sponse, as shown in Figure 8. Highly dispersed and low agglomerated PPy nanopartic are uniformly distributed on the pore walls of 3D RGO. The response to NH3 can rapid reach sub-ppm level, 4-5 times more sensitive than that of pure PPy. The sensor has p fect stability.

3D RGO/PPy
Recent studies have witnessed the significant progress of gas sensors based on polypyrrole (PPy) conducting polymers [122][123][124][125]. However, the serious agglomeration of the nanoparticles leads to a significant decrease in the visible active surface of gas molecules, which leads to the decrease of response sensitivity [126][127][128]. Qin et al. [129] used 3D RGO as a 3D skeleton support for the sensitive PPy nanoparticles. It has made an important contribution to improve the conductivity, dynamic performance, and sensing response, as shown in Figure 8. Highly dispersed and low agglomerated PPy nanoparticles are uniformly distributed on the pore walls of 3D RGO. The response to NH 3 can rapidly reach sub-ppm level, 4-5 times more sensitive than that of pure PPy. The sensor has perfect stability.

PANI/CuO@3D-NGF
CuO nanoparticles are suitable for the gas detection due to their narrow band g and adjustable morphology [130][131][132]. However, the problem is that they can only opera effectively at high temperatures. To improve this problem, nanoparticles can be incorp rated into porous materials to construct the functional composites [102,133]. Tabar et [134] prepared a chemical-resistant ammonia sensor based on polyaniline (PANI)/Cu nanoparticles, supported on a 3D N-doped graphene framework (NGF). The sensor h an excellent response to 100 ppm NH3, with an outstanding LOD down to 50 ppb. T average response time is 30 s at room temperature. It is not sensitive to other gases a has good selectivity to NH3. The excellent sensing performance was attributed to 3D terconnected porous structure, remarkable enhancement of charge carriers, and modifi π-interactions between molecules [134].

Phenol Gas Sensors
Phenol is the natural component of many substances and can be emitted from t combustion of fossil fuels and tobacco [20]. It is also present in animal wastes and deco posing organic material. More importantly, it is a chemical product produced at a rate about 6 million ton/year worldwide [20]. The manufacture and transportation of phen as well as its many uses may lead to worker exposures to this substance with health ris [135]. In 2013, Liu et al. [136] prepared an electrochemical sensor using assembled 3D g phene as the electrode. The LOD of the phenol sensor mediated by tyrosine was succe

PANI/CuO@3D-NGF
CuO nanoparticles are suitable for the gas detection due to their narrow band gap and adjustable morphology [130][131][132]. However, the problem is that they can only operate effectively at high temperatures. To improve this problem, nanoparticles can be incorporated into porous materials to construct the functional composites [102,133]. Tabar et al. [134] prepared a chemical-resistant ammonia sensor based on polyaniline (PANI)/CuO nanoparticles, supported on a 3D N-doped graphene framework (NGF). The sensor has an excellent response to 100 ppm NH 3 , with an outstanding LOD down to 50 ppb. The average response time is 30 s at room temperature. It is not sensitive to other gases and has good selectivity to NH 3 . The excellent sensing performance was attributed to 3D interconnected porous structure, remarkable enhancement of charge carriers, and modified π-interactions between molecules [134].

Phenol Gas Sensors
Phenol is the natural component of many substances and can be emitted from the combustion of fossil fuels and tobacco [20]. It is also present in animal wastes and decomposing organic material. More importantly, it is a chemical product produced at a rate of about 6 million ton/year worldwide [20]. The manufacture and transportation of phenol as well as its many uses may lead to worker exposures to this substance with health risks [135]. In 2013, Liu et al. [136] prepared an electrochemical sensor using assembled 3D graphene as the electrode. The LOD of the phenol sensor mediated by tyrosine was successfully achieved down to 50 ppm. In 2016, Guo et al. [137] reported RGO/metal-oxide p-n heterojunction aerogels as efficient 3D sensing frameworks for phenol detection. Upon the detection of phenol at room temperature, the sensor has good sensitivity, repeatability, and stability. The linear relationship is in the range of 10-80 ppb. Compared with the detection results of ethanol, toluene, and methanol, the gas sensor based on RGO/SnO 2 composite aerogel has much higher sensitivity to phenol with the LOD down to 5 ppb. In the same year, Gao et al. [138] reported highly sensitive electrocatalytic determination of phenols based on coupled cMWCNT/cyclodextrin edge-functionalized graphene composite. The sensor has excellent performance towards trace detection of three typical phenols (4-aminophenol, 4-AP; 4-chlorophenol, 4-CP; 4-nitrophenol, 4-NP). Under optimal conditions, the current responses of 4-AP, 4-CP, and 4-NP are linear to concentrations over two different ranges, with the LODs of 0.019, 0.017, and 0.027 ppm (S/N = 3), respectively. In 2019, Qi et al. [139] reported a facile synthesis of 3D S/N co-doped graphene derived from GO hydrogel. This newly fabricated sensor was used in the simultaneous detection of catechol and hydroquinone, with the LODs of 0.28 and 0.15 ppm, respectively.

Conclusions
Compared with the 2D graphene nanosheets, the signal transduction of 3D graphene frameworks are more than ten times higher in most gas detection, due to the increase of the interaction surface area and the number of active adsorption sites [89]. In addition to the original 2D structural properties of graphene, its 3D porous frameworks are more favorable. It is good for adsorption and diffusion of gas molecules. In general, the LOD can be improved from ppm level to several ppb level. This has a great impact on the prevention of toxic and harmful air pollution to the atmosphere.
In addition, 3D graphene has a strong mechanical strength and a high temperature resistance, capable of using in harsh environments. These properties are very promising for practical application. It is a potential research direction to combine 3D graphene with flexible substrate to make wearable flexible sensors.
However, 3D graphene is still limited in wide applications due to its limited gas sensing types. The main toxic pollutants in the air are NO x (main NO 2 ), NH 3 , CO, CH 2 O, and phenol. Compared with traditional metal oxide sensing materials, graphene is superior due to the much lower operating temperature and much lower resistance. It requires a low energy consumption in operation [83,87,115]. As shown in Table 2, traditional semiconductor-based gas sensors generally work at 200-400 • C [98,[140][141][142]. Some commercial gas sensors could work well at room temperature, but with some performance loss of the LOD [143,144]. Compared to individual MoS 2 , SnS 2 , organic compounds (e.g., PPy), and semiconductor metal oxides (e.g., SnO 2 ), the sensors based on 3D graphene composites could see an increase of the sensitivity.
At present, 3D graphene is mainly sensitive to NO 2 and NH 3 , and more gas sensors need to be discovered. The adsorption and desorption of 3D graphene need to be accelerated so as to reduce the response and recovery times of the sensor. Finally, 3D graphene sensors are still in the stage of laboratory investigation, and more work is needed to put forward these developments to the commercialization stage.  Informed Consent Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.