**3. Surface modified carbon nanotubes**

were achieved at low concentrations and up to 50 ppm. Detection limits (DL), derived from the noise of the baseline and the slope obtained from the calibration curve, were found to be as low as 590 parts per quadrillion (ppq) and 1.51ppt for NO and NO2, respectively. For NH3 it was found not only that the *in situ* UV illumination reverses the direction of the changes in conductance, but it was also confirmed that it helps to improve the DL from 5.67 ppm to 27.8 ppt when tested under identical conditions. The achieved DL outperformed by several orders of magnitude the sensitivity of other CNTs-based NO, NO2and NH3 sensors that have been previously reported. This is attributed to the UV light inducing surface regenera‐ tion and actively removal of all gases adsorbed on SWCNTs surface. It is worth noticing that this *in situ* cleaning with continuous UV-light exposure without device degradation was just

**Figure 1.** SWCNTs-sensor responses to (A) NO (10 – 200ppt) under *in situ* UV illumination, (B) NO2 (40 – 1000 ppt) under *in situ* UV illumination, (C) NH3 (5 – 500 ppm) without *in situ* UV illumination, and (D) NH3 (200 – 4000 ppt) under *in situ* UV illumination. (From Chen *et al.* [40]. Copyright © 2012, with permission from Nature Publishing

Battie and coworkers used sorted semi-conducting SWCNTs as sensing film for the detec‐ tion of NO2 and NH3 [41]. The density gradient ultracentrifugation (DGU) technique was used to separate semi-conducting from as produced SWCNTs. Films of as produced and sorted semi-conducting SWCNTs were exposed to NO2 and NH3 in air. Both films showed

achieved under inert atmospheres.

340 Syntheses and Applications of Carbon Nanotubes and Their Composites

Group. )

CNTS modified with different functional groups have been used for the development of sensors for detection of volatile organic compounds (VOCs) in the environment as well as in exhaled breath. For the detection of VOCs in air, Wang *et al.* worked in the preparation of a sensor array based on MWCNTs covalently modified with different functional groups like propargyl, allyl, alkyltriazole, thiochain, thioacid, hexafluoroisopropanol (HFIP) [43] and Shrisat, *et al.* reported another one based on SWCNTs modified with different porphyrins (organic macrocyclic compounds) likeoctaethyl porphyrin (OEP), ruthenium OEP (RuOEP), iron OEP (FeOEP), tetraphenylporphyrin (TPP), among others [44]. Penza *et al*. also worked in the modification of MWCNTs with TPP for the room temperature detection of VOCs [45]. In this case, the TTP contained two different metals, Zn (CNT:ZnTPP) and Mn (CNT:MnTPP). Sensors were exposed to ethanol, acetone, ethylacetate, toluene and Triethyl‐ amine at ppm levels and all showed increase in resistance when exposed to the different gases. CNT: MnTPP showed the highest sensitivity towards all gases with respect to un‐ modified CNTs but for triethylamine and CNT: ZnTPP was more sensitive to ethylacetate.

Two different CNT-based sensor arrays have been reported for the detection and pattern recognition of VOCs present in exhaled breath samples for medical diagnosis, Tisch *et al*. presented a sensor array containing different nanomaterials including organically function‐ alized random networks of SWCNTs for the detection of VOCs related to Parkinson disease and that were present in exhaled breath collected from rats [46]. Ionescu and coworkers re‐ ported a sensor array based on bilayers of SWCNTS and polycyclic aromatic hydrocarbons (PAH) for the detection of multiple sclerosis in exhaled human breath [47]. In general, the incorporation of organic functional groups provided not only enhanced sensitivity but also provided better selectivity for each gas when compared to pristine CNTs. The use of statisti‐ cal techniques like principal component analysis (PCA), discriminant factor analysis (DFA) and linear discriminant analysis (LDA) was possible to determine the discrimination capa‐ bility of the sensors toward each VOC.

SWCNTs were functionalized with tetrafluorohydroquinone (TFQ) at the room temperature for detection of dimethyl methylphosponate (DMMP) at parts per trillion (ppt) levels (Fig‐ ure 2) [48]. The conductance of the TFQ-SWCNTs samples increased as function of concen‐ tration when exposed to DMMP in N2 in a concentration range from 20ppt to 5.4ppb. Sensors showed fast response and ultra sensitivity down to 20ppt when compared to un‐ modified SWCNTs sensor, which had a DL of nearly 1 ppm. The presence of TFQ clearly showed to improve the sensitivity and this is because it provides additional binding sites thru hydrogen bonds between hydroxyl groups in TQF and DMMP. In addition, TQF tailors the electronic properties of SWCNTs via hole doping.

MWCNTs were chemically treated with acid to obtain hydroxyl groups (OH) and used as sensing material for humidity sensors [51]. Changes in resistance were measured as the RH was varied from 11% to 98%. It was observed that the resistance increased as sensors were exposed to the different humidity levels. It was found that acid treated SWCNTS were more sensitive to humidity than pristine MWCNTs. The higher sensitivity of Acid treated SWCNTs is attributed to their higher surface and thus more adsorption sites that result from the acid treatment. Sensors showed fast response and to be stable. As with most humidity sensors, the recovery time was longer than response time due to slow desorption of water

Latest Advances in Modified/Functionalized Carbon Nanotube-Based Gas Sensors

http://dx.doi.org/10.5772/52173

343

Purified MWCNTs were treated with oxygen plasma or fluorine plasma and used for the detection of ethanol (Figure 3) [52]. Changes in resistance as function of time were recorded for the sensors when exposed of 50-500 ppm of ethanol vapor in air. Samples treated with oxygen plasma for 30 sec and with fluorine plasma for 60 sec showed the highest sensitivity to 100 ppm of ethanol, compared to pristine MWCNTS and other oxygen and fluorine plas‐ ma treated for different duration time. However, fluorine plasma treated samples showed the better sensitivity and reduced response and recovery time. The improvement of the fluo‐ rine plasma treated samples is explained by the difference in electronegativity between oxy‐

**Figure 3.** Responses of pristine (MC), oxygen plasma treated (O30) and fluorine plasma treated (F60) MWCNTs to

Thermally fluorinated MWCNTs (TFC) were used for NO gas detection at room tempera‐ ture [53]. The effect of thermal fluorination process was performed at various temperatures

tional groups, and even fluorine-assisted pyrolysis and fluorine-induced reorientation of the

C. TFC prepared at 200 o

C was found to be the optimum fluorination temperature. TFC sam‐

C showed a decrease of the fluorine func‐

C showed high sensitivity,

100-400 ppm of ethanol. (From Liu *et al.* [52]Copyright © 2012, with permission from Elsevier. )

molecules.

gen and fluorine.

(100 -1000 o

C) and 200 o

MWCNTs structure occurred at 1000 o

ples prepared at temperatures higher than 200 o

**Figure 2.** Representation of the possible sensing mechanism of DMMP with TFQ functionalized SWCNTs. (From Wei *et al.* [48]. Copyright © 2011, with permission from Institute of Physics Publishing. )

Wang *et al*. deposited a uniform network of SWCNT with carboxylic groups (-COOH) on a flexible poly (dimethyldiallylammonium chloride) (PDDA) modified polyimide (PI) sub‐ strate for DMMP detection at room temperature [49]. Changes in resistance as function of time were measured as the sensors were exposed to DMMP in N2 at a concentration range of 1-40 ppm. Calibration curve showed a [17] linear increase of resistance as function of con‐ centration and responses were found to be fast, stable and reproducible. Sensors showed to be selective to DMMP in presence of other volatile organic vapors like methanol, xylene and hexane, among others. Changing the carrier gas from N2 to air caused the responses to DMMP to be lower which might be due to influence of oxygen and humidity contained in air. The apparent enhanced performance of this SWCNTs (-COOH) flexible sensor when compared to sensors prepared on Si/SiO2 rigid substrates is attributed to the presence of PDDA, which is a polymer that absorbs DMMP. There is no information on the adsorption of DMMP by PDDA and its possible effect in the recovery of the sensors.

MWCNTs were oxidized with KMnO4 to add oxygenated functional groups, mainly (- COOH) for the detection of organic vapors [50]. Oxidized MWCNTs in form of a bucky pa‐ per were exposed to different concentrations of acetone. Variations in electrical resistance were recorded for both unmodified and oxidized MWCNTs. Oxidized MWCNTs showed higher sensitivity to acetone (2.3 vol. %) than unmodified ones, and good selectivity when sensing other oxygen containing vapors such as diethyl ether and methanol. The sensors al‐ so showed complete reversibility and high reproducibility for all tested vapors.

MWCNTs were chemically treated with acid to obtain hydroxyl groups (OH) and used as sensing material for humidity sensors [51]. Changes in resistance were measured as the RH was varied from 11% to 98%. It was observed that the resistance increased as sensors were exposed to the different humidity levels. It was found that acid treated SWCNTS were more sensitive to humidity than pristine MWCNTs. The higher sensitivity of Acid treated SWCNTs is attributed to their higher surface and thus more adsorption sites that result from the acid treatment. Sensors showed fast response and to be stable. As with most humidity sensors, the recovery time was longer than response time due to slow desorption of water molecules.

ure 2) [48]. The conductance of the TFQ-SWCNTs samples increased as function of concen‐ tration when exposed to DMMP in N2 in a concentration range from 20ppt to 5.4ppb. Sensors showed fast response and ultra sensitivity down to 20ppt when compared to un‐ modified SWCNTs sensor, which had a DL of nearly 1 ppm. The presence of TFQ clearly showed to improve the sensitivity and this is because it provides additional binding sites thru hydrogen bonds between hydroxyl groups in TQF and DMMP. In addition, TQF tailors

**Figure 2.** Representation of the possible sensing mechanism of DMMP with TFQ functionalized SWCNTs. (From Wei *et*

Wang *et al*. deposited a uniform network of SWCNT with carboxylic groups (-COOH) on a flexible poly (dimethyldiallylammonium chloride) (PDDA) modified polyimide (PI) sub‐ strate for DMMP detection at room temperature [49]. Changes in resistance as function of time were measured as the sensors were exposed to DMMP in N2 at a concentration range of 1-40 ppm. Calibration curve showed a [17] linear increase of resistance as function of con‐ centration and responses were found to be fast, stable and reproducible. Sensors showed to be selective to DMMP in presence of other volatile organic vapors like methanol, xylene and hexane, among others. Changing the carrier gas from N2 to air caused the responses to DMMP to be lower which might be due to influence of oxygen and humidity contained in air. The apparent enhanced performance of this SWCNTs (-COOH) flexible sensor when compared to sensors prepared on Si/SiO2 rigid substrates is attributed to the presence of PDDA, which is a polymer that absorbs DMMP. There is no information on the adsorption

MWCNTs were oxidized with KMnO4 to add oxygenated functional groups, mainly (- COOH) for the detection of organic vapors [50]. Oxidized MWCNTs in form of a bucky pa‐ per were exposed to different concentrations of acetone. Variations in electrical resistance were recorded for both unmodified and oxidized MWCNTs. Oxidized MWCNTs showed higher sensitivity to acetone (2.3 vol. %) than unmodified ones, and good selectivity when sensing other oxygen containing vapors such as diethyl ether and methanol. The sensors al‐

the electronic properties of SWCNTs via hole doping.

342 Syntheses and Applications of Carbon Nanotubes and Their Composites

*al.* [48]. Copyright © 2011, with permission from Institute of Physics Publishing. )

of DMMP by PDDA and its possible effect in the recovery of the sensors.

so showed complete reversibility and high reproducibility for all tested vapors.

Purified MWCNTs were treated with oxygen plasma or fluorine plasma and used for the detection of ethanol (Figure 3) [52]. Changes in resistance as function of time were recorded for the sensors when exposed of 50-500 ppm of ethanol vapor in air. Samples treated with oxygen plasma for 30 sec and with fluorine plasma for 60 sec showed the highest sensitivity to 100 ppm of ethanol, compared to pristine MWCNTS and other oxygen and fluorine plas‐ ma treated for different duration time. However, fluorine plasma treated samples showed the better sensitivity and reduced response and recovery time. The improvement of the fluo‐ rine plasma treated samples is explained by the difference in electronegativity between oxy‐ gen and fluorine.

**Figure 3.** Responses of pristine (MC), oxygen plasma treated (O30) and fluorine plasma treated (F60) MWCNTs to 100-400 ppm of ethanol. (From Liu *et al.* [52]Copyright © 2012, with permission from Elsevier. )

Thermally fluorinated MWCNTs (TFC) were used for NO gas detection at room tempera‐ ture [53]. The effect of thermal fluorination process was performed at various temperatures (100 -1000 o C) and 200 o C was found to be the optimum fluorination temperature. TFC sam‐ ples prepared at temperatures higher than 200 o C showed a decrease of the fluorine func‐ tional groups, and even fluorine-assisted pyrolysis and fluorine-induced reorientation of the MWCNTs structure occurred at 1000 o C. TFC prepared at 200 o C showed high sensitivity, stability, reproducibility and full recovery, when their gas sensing properties were evaluat‐ ed towards the detection of 50 ppm NO in dry air. Interestingly, the presence of fluorine re‐ verses the electron transfer process, when compared to pristine MWCNTs, allowing them to go from NO to MWCNTs and thus causing an increase in resistance. The fluorination not only helped to enhance the sensitivity but also made the sensors insensitive to humidity changes.

groups contained in the mesoporous silica layer and the gas molecules. The silanol groups allow the mesoporous silica film to act as a diffusion barrier and allow the physi‐ cal interaction and entrapment of highly polarized molecules like H2O and NH3, avoid‐ ing them to get in contact with the SWCNTs layer. On the other hand, the sensitivity to NO2 was greatly enhanced, compared to the SWCNTs sensor. Compared to H2O and NH3, NO2 has a weaker dipole moment and its diffusion thru the mesoporous silica gel and to the SWCNTs film results easier and thus its enhanced and selective detection.

Latest Advances in Modified/Functionalized Carbon Nanotube-Based Gas Sensors

http://dx.doi.org/10.5772/52173

345

Computational studies based on SWCNTs doped with heteroatoms have been also re‐ ported. *Ab initio* (ABINIT) simulations of CNTs doped with heteroatoms like boron, oxy‐ gen and nitrogen were performed to predict the behavior of the doped CNTs and to study their application as gas sensors for Cl2, CO, NO and H2 [57, 58]. Density function‐ al theory (DFT) applied in the ABINIT code and the Generalized Gradient Approxima‐ tion (GGA) were used to perform the calculations. The calculations demonstrated that doping the CNTs with B, O, and N causes a shift in the conduction band of the CNTs. For B and O, the conduction band shifts downward and creates a p-type semiconducting material. On the other hand, N dopant causes the conduction band to shift upward and create an n-type semiconducting material. Calculations also demonstrated that Cl2, NO, H2and CO considerably affects the NTs density of states (DOS) and Fermi level as the gases become close to their surface. B-doped CNTs can detect CO, NO and H2 gas mole‐ cules, O-doped can detect H2, Cl2 and CO and N-doped can detect CO, NO and Cl2.

Similarly, Hamadanian *et al*. presented a computational study of Al-substituted SWCNTs (10, 0) (2.5% and 25%) and their use as CO gas sensor [59]. DFT calculations (local densi‐ ty approximation with ultrasoft pseudopotential) were used to study the electronic prop‐ erties of Al-substituted SWCNTs and how those properties are affected by the adsorption of CO molecules. Substitution of one carbon atom with an Al atom causes de‐ formation of the 6-membered ring and increasing the bond length. Doping with Al also alters the DOS and band structure of the CNTs. Since Al has one electron less than C in the valence shell, introduces one electron holes in the band structure, therefore the tube is changed to p-type semiconductor. Calculations showed low adsorption energy for CO on pristine CNTs and that CO does not cause significant changes in the electronic band structure and DOS when adsorbed on pristine CNTs. These results confirm that pristine CNTs are insensitive to CO as result of their weak physical interaction. When CO is ad‐ sorbed on both 2.5% and 25% Al-substituted SWCNTs, it causes severe changes in the Band structure near Fermi level. Those changes strongly depended on the site of CNTs and the direction in which the CO molecule interacts. For instance, the most stable ad‐ sorption structure is when the C of the CO interacts with the middle point of a C-C bond of CNTs. Even when the adsorption energies of CO in 25% Al-substituted SWCNTs were higher than in 5% Al-substituted SWCNTs, the fact that the conductivity of the proposed material changes, makes them suitable for their use as CO gas sensing

material.

MWCNTs were modified with amino groups for the detection of formaldehyde at ppb level [54]. Changes in resistance as function of time were measured as the sensors were exposed to formaldehyde in a concentration range between 20 and 200 ppb. Sensors con‐ taining MWCNTs with higher amino group content (18%) were 2.4 and 13 times more sensitive to formaldehyde than samples containing 5% amino groups and pristine CNTs, respectively. Short response times are due to a chemical reaction between the aldehyde and amino group. For the same reason, the recovery times are longer, since chemical de‐ sorption is a slow and irreversible process. SWCNTs with 18% amino groups showed to be selective to formaldehyde when tested against interferences like acetone, CO2, ammo‐ nia, methanol and ethanol.

Silicon (Si) nitrogen (N), and phosphorous-nitrogen (P-N) were used to modify MWCNTs and study their gas sensing properties for hydrogen peroxide, sodium hypo‐ chlorite and1, 2-dichloromethane, nitrogen, and ammonia [55]. Samples of Si-MWCNTs, N-SWCNTs, and P-N-SWCNTs were prepared by aerosol chemical vapor deposition. It is known that the incorporation of heteroatoms in the CNT structure changes its morpholo‐ gy and thus the reactivity. To evaluate the gas sensing properties of the prepared materi‐ als, changes in resistance as function of time were recorded when they were exposed to the different gases. Exposure to N2 caused the removal of physisorbed water molecules and thus a decrease in the resistance values. Sodium hypochlorite and dichloroethane caused decrease in resistance of pristine MWCNTs and Si-MWCNTs due to charge trans‐ fer (electrons) from CNTs to Chlorine atoms and increase in resistance of N-P MWCNTs. Ammonia showed the opposite effect in resistance. These results demonstrate the p-type semiconductor behavior for pristine MWCNTs and Si-MWCNTs and n-type of N-MWCNTs and N-P-MWCNTs. All sensors recovered in 10 min for all gases with the ex‐ ception of ammonia that exceeded 1 hour.

In an effort to enhance the selectivity of SWCNTs-based vapor sensors, Battie *et al*. worked in the preparation of SWCNTs covered with a mesoporous silica film [56]. Sen‐ sors were fabricated by covering a SWCNTs film with a mesoporous silica film via by sol-gel deposition technique. Characterization of the sensors was done by measuring changes in resistance when exposed to 200 ppm of NO2, NH3, and H2O in dry air. A sen‐ sor of SWCNTs without the mesoporous silica film prepared and tested under the same conditions. While the SWCNTs sensor showed to be sensitive to the three gases, the sen‐ sor based on SWCNTs film covered with mesoporous silica film showed to insensitivity to H2O, and its sensitivity for NH3 was considerably reduced. These observations can be explained considering the polarization capabilities and dipole moments of the silanol groups contained in the mesoporous silica layer and the gas molecules. The silanol groups allow the mesoporous silica film to act as a diffusion barrier and allow the physi‐ cal interaction and entrapment of highly polarized molecules like H2O and NH3, avoid‐ ing them to get in contact with the SWCNTs layer. On the other hand, the sensitivity to NO2 was greatly enhanced, compared to the SWCNTs sensor. Compared to H2O and NH3, NO2 has a weaker dipole moment and its diffusion thru the mesoporous silica gel and to the SWCNTs film results easier and thus its enhanced and selective detection.

stability, reproducibility and full recovery, when their gas sensing properties were evaluat‐ ed towards the detection of 50 ppm NO in dry air. Interestingly, the presence of fluorine re‐ verses the electron transfer process, when compared to pristine MWCNTs, allowing them to go from NO to MWCNTs and thus causing an increase in resistance. The fluorination not only helped to enhance the sensitivity but also made the sensors insensitive to humidity

344 Syntheses and Applications of Carbon Nanotubes and Their Composites

MWCNTs were modified with amino groups for the detection of formaldehyde at ppb level [54]. Changes in resistance as function of time were measured as the sensors were exposed to formaldehyde in a concentration range between 20 and 200 ppb. Sensors con‐ taining MWCNTs with higher amino group content (18%) were 2.4 and 13 times more sensitive to formaldehyde than samples containing 5% amino groups and pristine CNTs, respectively. Short response times are due to a chemical reaction between the aldehyde and amino group. For the same reason, the recovery times are longer, since chemical de‐ sorption is a slow and irreversible process. SWCNTs with 18% amino groups showed to be selective to formaldehyde when tested against interferences like acetone, CO2, ammo‐

Silicon (Si) nitrogen (N), and phosphorous-nitrogen (P-N) were used to modify MWCNTs and study their gas sensing properties for hydrogen peroxide, sodium hypo‐ chlorite and1, 2-dichloromethane, nitrogen, and ammonia [55]. Samples of Si-MWCNTs, N-SWCNTs, and P-N-SWCNTs were prepared by aerosol chemical vapor deposition. It is known that the incorporation of heteroatoms in the CNT structure changes its morpholo‐ gy and thus the reactivity. To evaluate the gas sensing properties of the prepared materi‐ als, changes in resistance as function of time were recorded when they were exposed to the different gases. Exposure to N2 caused the removal of physisorbed water molecules and thus a decrease in the resistance values. Sodium hypochlorite and dichloroethane caused decrease in resistance of pristine MWCNTs and Si-MWCNTs due to charge trans‐ fer (electrons) from CNTs to Chlorine atoms and increase in resistance of N-P MWCNTs. Ammonia showed the opposite effect in resistance. These results demonstrate the p-type semiconductor behavior for pristine MWCNTs and Si-MWCNTs and n-type of N-MWCNTs and N-P-MWCNTs. All sensors recovered in 10 min for all gases with the ex‐

In an effort to enhance the selectivity of SWCNTs-based vapor sensors, Battie *et al*. worked in the preparation of SWCNTs covered with a mesoporous silica film [56]. Sen‐ sors were fabricated by covering a SWCNTs film with a mesoporous silica film via by sol-gel deposition technique. Characterization of the sensors was done by measuring changes in resistance when exposed to 200 ppm of NO2, NH3, and H2O in dry air. A sen‐ sor of SWCNTs without the mesoporous silica film prepared and tested under the same conditions. While the SWCNTs sensor showed to be sensitive to the three gases, the sen‐ sor based on SWCNTs film covered with mesoporous silica film showed to insensitivity to H2O, and its sensitivity for NH3 was considerably reduced. These observations can be explained considering the polarization capabilities and dipole moments of the silanol

changes.

nia, methanol and ethanol.

ception of ammonia that exceeded 1 hour.

Computational studies based on SWCNTs doped with heteroatoms have been also re‐ ported. *Ab initio* (ABINIT) simulations of CNTs doped with heteroatoms like boron, oxy‐ gen and nitrogen were performed to predict the behavior of the doped CNTs and to study their application as gas sensors for Cl2, CO, NO and H2 [57, 58]. Density function‐ al theory (DFT) applied in the ABINIT code and the Generalized Gradient Approxima‐ tion (GGA) were used to perform the calculations. The calculations demonstrated that doping the CNTs with B, O, and N causes a shift in the conduction band of the CNTs. For B and O, the conduction band shifts downward and creates a p-type semiconducting material. On the other hand, N dopant causes the conduction band to shift upward and create an n-type semiconducting material. Calculations also demonstrated that Cl2, NO, H2and CO considerably affects the NTs density of states (DOS) and Fermi level as the gases become close to their surface. B-doped CNTs can detect CO, NO and H2 gas mole‐ cules, O-doped can detect H2, Cl2 and CO and N-doped can detect CO, NO and Cl2.

Similarly, Hamadanian *et al*. presented a computational study of Al-substituted SWCNTs (10, 0) (2.5% and 25%) and their use as CO gas sensor [59]. DFT calculations (local densi‐ ty approximation with ultrasoft pseudopotential) were used to study the electronic prop‐ erties of Al-substituted SWCNTs and how those properties are affected by the adsorption of CO molecules. Substitution of one carbon atom with an Al atom causes de‐ formation of the 6-membered ring and increasing the bond length. Doping with Al also alters the DOS and band structure of the CNTs. Since Al has one electron less than C in the valence shell, introduces one electron holes in the band structure, therefore the tube is changed to p-type semiconductor. Calculations showed low adsorption energy for CO on pristine CNTs and that CO does not cause significant changes in the electronic band structure and DOS when adsorbed on pristine CNTs. These results confirm that pristine CNTs are insensitive to CO as result of their weak physical interaction. When CO is ad‐ sorbed on both 2.5% and 25% Al-substituted SWCNTs, it causes severe changes in the Band structure near Fermi level. Those changes strongly depended on the site of CNTs and the direction in which the CO molecule interacts. For instance, the most stable ad‐ sorption structure is when the C of the CO interacts with the middle point of a C-C bond of CNTs. Even when the adsorption energies of CO in 25% Al-substituted SWCNTs were higher than in 5% Al-substituted SWCNTs, the fact that the conductivity of the proposed material changes, makes them suitable for their use as CO gas sensing material.


Considering that one of the problems of MWCNTs is their aggregation and agglomeration, they were oxyfluorinated under different conditions in order to obtain a better dispersion in aqueous solution and it was found that the better dispersion was obtained with the MWCNTs with the highest oxygen content. The oxyfluorinated MWCNTs were then mixed with aniline and ammonium persulfate (APS) and other chemicals, for in-situ polymeriza‐ tion. Changes in resistance as function of time were used to characterize and evaluate the resulting PANi/MWCNTs composite was for the detection of ammonia (NH3) in a concen‐ tration range of 1-50 ppm. PANi/MWCNTs composite with the highest oxygen content had a uniform composition, improved thermal stability and highest and faster response for 50 ppm of NH3. The composite was able to detect 1 ppm and showed excellent repeatability for cycling exposures to 50 ppm and it needed heat treatment to accelerate the NH3 desorption and thus the recovery of the sensor. A possible drawback is the selectivity to NH3 among

Latest Advances in Modified/Functionalized Carbon Nanotube-Based Gas Sensors

http://dx.doi.org/10.5772/52173

347

) from PANi.

**Figure 4.** Illustration of the steps to obtain oxyfluorinated CNTs modified with PANI. (From Yun et al. [68]. Copyright

Mangu *et al.* also worked in the preparation of PANi-MCNTs as well as Poly (3, 4-ethyl‐ enedioxythiophene) poly (styrenesulfonate) (PEDOT:PSS)-MWCNTs composites for the detection of 100 ppm of NO2 and NH3 [69]. This group studied the effect of dissolving the polymers in different solvents for the gas sensing properties. PANi was dissolved in dimethyl sulfoxide (DMSO), *N*, *N*-dimethyl formamide (DMF), ethylene glycol (EG) and 2-propanol. PEDOT: PSS was dissolved in DMSO, DMF and 0.1M sodium hydroxide (NaOH). Each polymer solution was spin-coated in plasma treated MWCNTs and evalu‐ ated as sensing material. All PANi-MWCNTs composited showed an increase in resist‐ ance for NH3 and a decrease in resistance for NO2, which is typical of p-type semiconducting composite films. All PANi-MWCNTs composites were selective to NO2. However, better sensitivities were achieved when PANi was dissolved in 2-propanol and DMSO for NH3 and for NO2, respectively. On the other hand, all PEDOT:PSS-MWCNTs composites were also excellent for the detection if both NO2 and NH3. PE‐ DOT: PSS-MWCNTs (prepared without any solvent) showed to be more sensitive to

Sayago, *et al.* have worked on the preparation of different composites using polymers with small percentages of CNTs as sensitive layers for surface acoustic wave (SAW) gas sensors [70-72]. Composites of polyisobutilene (PIB), polyepichlorohydrin (PECH) and polyetherur‐

gases that can extract protons (H+

© 2012, with permission from Elsevier. )

NH3 and PEDOT:PSS dissolved in NaOH to NO2.

\*Lowest tested concentration

**Table 1.** Summary of metal oxide NPs used to modify CNTs for gas sensor applications.
