CNLS fitting results of the system TiO2/CNT-WT/TiO2, relaxation times and sensor response. The positive value of SR is to indicate an increase in the resistance
Design of composite materials for ammonia (NH3) sensing is important because of two main reasons: (1) NH3 is the most common substitute for chlorofluorocarbons (CFCs) in cooling systems, and (2) most sensors show long recovery times at room temperature due to the tendency of ammonia to strongly interact with many substrates. Multiwalled carbon nanotubes (MWCNTs) have been used to sense polar molecules like carbon monoxide (CO), carbon dioxide (CO2), NH3, water (H2O), and ethanol (C2H5OH) (Ong et al., 2002; Valentini et al., 2004; Varghese et al., 2001), as well as non polar gases like helium (He) and nitrogen (N2) (Adu et al., 2001). For carbon nanotubes, studies have shown that O2 molecules are electron acceptors with substantial adsorption energies and charge transference, while NH3, N2, CO2, methane (CH4), H2O, hydrogen (H2) and argon (Ar) are electron donors (Zhao et al., 2002). Adsorption in CNTs is determined by adsorption energy and availability of sites, with typically four different adsorption sites: external surface, grooves between CNTs on the bundle outside, pores inside CNTs, and interstitial channel between adjacent tubes inside the bundle (Stan & Cole, 1998; Williams & Eklund, 2000). Additionally, theoretical studies carried out by Jhi et al. about the electronic and magnetic properties of oxidized CNTs indicated their high potential as gas sensors. These studies demonstrated that the sensing mechanism of as prepared CNTs is more related to oxygen doping than to intrinsic properties, and depends on the structural defects caused by the synthesis methods (Jhi et al., 2000).
With the aim of increasing the response of CNTs to certain gases, some studies about substitutional functionalization of CNTs have been reported (Peng & Cho, 2003). Although boron (B) and nitrogen (N) doping improve the response to gases like CO and H2O, BxCy nanotubes show stronger chemisorption, in contrast with the substitutional functionalization with nitrogen (Villalpando-Paez et al., 2004). Oxidation of CNTs in acid solutions to graft oxygen functional groups [i.e., carboxyl (COOH), hydroxyl (OH), and carbonyl (CO)] on CNT walls have been widely used to diversify the sensing options. Another strategy has been the fabrication of compound materials based on CNTs and metallic oxides. Espinoza et al. reported the use of metallic oxides-CNT composites based on commercial tin (SnO2) and tungsten (WO3) oxide powders, as well as sol gel TiO2 materials, in the sensing of NO2 and CO at room temperature and 150ºC. Their results indicated a better performance for CNT/SnO2 and CNT/WO3 than for CNT/TiO2, for the titania composite the sensor response was barely sizable (Espinoza et al., 2007).
Throughout this chapter, we will present the synthesis and performance of compound materials based on MWCNT/TiO2 tested as resistor and capacitor ammonia sensors. The design aims for low cost room temperature sensors with good reversibility and fast recovery times. For sensor fabrication, both materials were prepared separately (
2.1. Film elaboration techniques
MWCNTs were obtained from Nanostructured & Amorphous Materials Inc. (90 wt.%, outer diameter<10 nm, length: 5-15 μm, tangled). To introduce oxigenated functional groups, CNTs were refluxed at 100ºC for 6 h in acid solutions of 0.5 M sulphuric acid (H2SO4) with variable concentrations of nitric acid (HNO3). Functionalized CNTs were labelled according to the concentration of HNO3 (2.5M, 7.5M, 12.5M) as CNT-2.5, CNT-7.5 and CNT-12.5, CNTs without treatment were labeled as CNT-WT, composites were labeled as CMP with the number of the HNO3 concentration used for chemical treatment. CNT inks were prepared mixing 10 mg of CNTs with 40 μL of triton X-100 in 5 mL of deionized water, the triton/CNT weight ratio was 0.6.
Sensor fabrication (ex-situ method)
To prepare TiO2 films, 100 mL sol gel baths containing 92 mL of isopropanol, 0.1 mL of hydrochloric acid (HCl), and 8 mL of titanium isopropoxide were kept at room temperature (27ºC) for 24-48 h. Films were deposited by dip coating on glass (Corning) and indium tin oxide (ITO) 1x1.5 in substrates, using 8 dipping/withdrawing cycles at 30 mm/min speed, and air annealing for 5 min at 400ºC after each immersion, and at 500ºC for 1 h after the last cycle. After TiO2 deposition, 0.5 mL of CNT ink was drop casted on top and was dried at room temperature for 24 h. The two layer film was annealed in air for 30 min at 400ºC, finally another TiO2 film (8 cycles) were deposited to make a sandwich like configuration; the whole system was annealed at 400ºC in air for 1 h.
The changes in chemical composition caused by the functionalization of CNTs were followed by thermogravimetric analysis (TGA, TA Instruments Q500). TGA studies were carried out under 5 mL/min oxygen flow and using 5°C/min heating rate in the 25-800ºC temperature range. Particle size and zeta potential (ζ) were measured by dynamic light scattering (DLS, Zetasizer Nano, Malvern Instruments) as a function of CNTs functionalization. Film crystallinity was monitored by X-ray diffraction (Rigaku Dmax 2200, CuKα radiation, λ=1.5405 Å), using the Debye-Scherer equation (Cullity, 1978) for crystallite size determination. Atomic force microscopy was used for surface microstructural studies (Nanosurf Easyscan, Nanosurf AG, Switzerland). The experimental setup for sensing trials for both platforms is shown in Figure 1, basically, nitrogen was passed through a bubbler containing ammonia and was injected into a flask where the resistance or capacitance of the sensors changed as they were exposed to ammonia; the baseline was measured in dry air and the sensor in the picture correspond to a resistor sensor.
For capacitor sensors, variations of capacitance were followed by electrochemical impedance spectroscopy (EIS) measurements in an Autolab PGSTAT302N potentiostat/galvanostat unit (Eco Chemie), using the configuration depicted in Figure 2. A perturbation of 5-10 mV was applied at open circuit potential in the frequency range from 105 to 10-3 Hz. Complex non linear least squares (CNLS) fitting of the experimental data was done with the Zsimpwin software (Princeton Applied Research). All the sensing trials were done at room temperature (27ºC).
3. Results and discussion
3.1. CNT characterization
Figure 3 shows the TGA curves of CNTs exposed to various acid treatments. From the variation of mass vs. temperature [Figure 3(a)], or mass/temperature ratio vs. temperature [Figure 3(b)] figures it is easy to appreciate the ~15 wt.% of metal and impurities content of as received CNTs (CNT-WT) as well as its removal by all the acid treatments. The higher oxidation temperature indicates that the treatment used for CNT-2.5 does not damage the nanotube surface, but as the aggressiveness of the functionalization increases, a sizable weakening of the CNT structures is observed, in addition to the grafting of oxygenated functional groups, as evidenced in Fourier transform infrared studies ( Sánchez & Rincón, 2009 ). DLS characterization of CNT aqueous inks prepared with a triton/CNT ratio of 0.6 and sonicated for 30 and 90 min is presented in Figure 4. Ultrasonic treatments of 90 min produced a better dispersion, especially for CNT-WT and CNT-2.5, however, for CNT-7.5 and CNT-12.5 the differences between 30 and 90 min sonication times are smaller [Figure 4(a)]. The absolute value of ζ [Figure 4(b)] shows a continuous increase as the intensity of the chemical treatment does, up to ~18 mV, which is lower than the characteristic values of stable solutions (±30 mV), this increase confirms the presence of functional groups, however in the long term these solutions would show a tendency to precipitate.
3.2. Structural characterization
The effect of CNT functionalization on the topography of the composite is shown in Figure 5. Figures 5(a-b) show the surface and edge detection of CMP-WT composite. Several longitudinal and tangled formations of CNTs bundled in ropes are evident. In contrast, Figures 5(c-d) show images of CMP-12.5 composite, where CNT/titania grains are slightly smaller, more dispersed, and with shorter ropes. Analysis at a lower scale (not shown) confirmed that a compact and thin layer (~20 nm) of titania was formed covering all the surface including carbon nanotubes, with an increasing thickness as the intensity of chemical treatment does.
Statistical analysis of the length and width of CNT ropes in TiO2/CNT composites deposited on insulator glass substrates is shown in Figure 6. This graph confirms that the chemical treatment reduces the length of ropes from ~5 μm for CMP-WT to ~2.5 μm for CMP-12.5. A slight increase in the width is observed as the intensity of functionalization increases, which can only be due to a thicker titania layer, given that thinner CNT ropes are obtained at stronger functionalizations.
X-ray diffraction studies (Figure 7) of films prepared in a multilayer configuration (system TiO2/CNT/TiO2) shows only the anatase phase and broadening of its peaks. Crystallite size determination using the peak at 2θ ≈ 48º (to avoid interference from carbon diffraction at 2θ = 25º) shows a decrease from ~25 nm for CMP-WT to ~10 nm for CMP-12.5 (a reduction of ~60%), according to some studies that report the confined growing of titania grains attached to functionalized CNTs which are smaller than the ones growing away from nanotubes (An et al., 2007; Hieu et al., 2008; Song et al., 2007; Yu et al., 2007). This figure also confirms that CNTs with more functional groups are covered with a thicker titania layer.
3.3. Electrical characterization
3.3.1. Resistor sensors
In this section the performance of CNTs and TiO2/CNT films as resistor ammonia sensors is presented. Figures 8-10 compare the effect of CNT functionalization on the dynamical response to 1 vol.% ammonia of CNT (Figure 8), CNT/TiO2 (Figure 9), and TiO2/CNT/TiO2 (Fig. 10). Some of these figures show passivation of the most reactive sites during the first pulse, and reversible responses in the subsequent pulses.
To quantify the effect of the inclusion of titania layers and the use of functionalized CNTs Figure 11 shows the comparison of the resistivity in air, the sensor response (SR) (equation 1), the adsorption and desorption times, and the reversibility (equation 2).
The resistivity in air [Figure 11(a)] shows a decrease up to ~80% when CNTs are covered with titanium dioxide, affecting also the value of S. In titania-free materials, the effect of functionalization is to increase the resistivity in air, while in those containing titania the resistivity decreases and this drop is more notorious in sensors with a higher number of titania layers. With respect to the adsorption/desorption times a clear reduction is observed in the time required for adsorption in functionalized materials, whereas the effect of titania layers is difficult to appreciate. For desorption times, the presence of TiO2 layers causes up to one order decrease (~90%) and it is far more important than CNT functionalization. The faster sensors correspond to a combination of functionalized CNTs and the presence of TiO2 layers, see systems CNT-2.5/TiO2 and TiO2/CNT-2.5/TiO2.
Some observations from AFM studies can help to explain the electrical behavior of these composites. Dip coating is a technique that applies strong pressures on the surface as films are deposited (Brinker & Hurd, 1994), therefore it results in the deposition of compact and continuous titania films as was observed by AFM and aids to the formation of new paths for charge carriers. These processes are enhanced if functionalized CNTs are used, giving composites with lower resistivity. There are also some studies that report a larger work function for the TiO2 than for CNT (Ou et al., 2006; Wang et al., 2005), opening the posibillity for electronic transfer from CNT to TiO2 (Figure 12), increasing the composite conductivity through the increase of titania conductivity and dedoping of CNTs.
3.3.2. Capacitor sensors
EIS studies of multilayer systems TiO2/CNT-X/TiO2 prepared by dip coating and drop casting using CNT-WT and CNT-2.5 are presented in Figures 13 and 14, respectively. Cole-Cole plots are shown in Figures 13(a)and 14(a), the variations of phase angle vs frequency are shown in Figures 13(b) and 14(b), and the results of fitting experimental data with equivalent circuits are presented in Tables 1 and 2. The equivalent circuit used to fit the experimental data of TiO2/CNT-WT/TiO2 composite [Figure 13(a)] consists of a resistance R0 in series with three subcircuits: higher frequencies (R1C1), intermediate frequencies (R2Q2), and lower frequencies (R3C3). The value of R1 (few ohms) decreases as the material is exposed to ammonia, in contrast to the values of R2 and of R3 (several kohms) which show an increase. Capacitance values do not show a unique tendency either, C1 increases, C2 it is almost constant, and C3 decreases with ammonia exposure. Differences among capacitances values are of several orders of magnitude suggesting that adsorption processes are taking place in different sites.
In contrast to the multilayer sensor based on non-functionalized CNTs, EIS results of TiO2/CNT-2.5/TiO2 required only two subcircuits to fit the experimental data depicted in Figure 14. The subcircuit at higher (R1C1) and lower (R2Q2) frequencies are evident in air, but during ammonia exposure only the process at lower frequencies dominates. R1 values have been related to TiO2 grain boundaries in TiO2-CNT composites, although the value of ~101 Ω is lower than the typical values (~103 Ω) ( Sánchez et al., 2009 ), this difference could be related to processes taking place at the CNT/TiO2 interface, supporting the improved connectivity of the materials and the direction of charge transfer from CNT=>TiO2. With respect to the subcircuits at intermediate and lower frequencies in composites based on CNT-WT, changes in capacitances and resistances when the films are exposed to ammonia suggest the presence of two carbon types or two different adsorption sites. In contrast, for composites based on CNT-2.5, amorphous carbon and metal catalysts are removed during functionalization, defining the surface of CNTs as the main adsorption site. Reduction of R2 in air as well as in ammonia, suggest doping of both TiO2 and CNTs, because the electronic transference from CNTs to TiO2 might be facilitated by the formation of covalent bonds during functionalization. Given the tendencies described, we propose that subcircuit 1 is the
response of titania grain boundaries, subcircuit 2 is the response of CNTs, and subcircuit 3 is related to the response of impurities in CNT-WT, these correlations are summarized in Table 3. The removal of impurities (i.e. elimination of the third subcircuit) and the decrement in R values reduce the relaxation times up to three orders of magnitude, from seconds for composites with CNT-WT to miliseconds for composites with CNT-2.5. Another benefit of the functionalization is that the capacitance response computed from equation 3 increases up to ~2.5 (i.e., 150%, Table 2) in composites with CNT-2.5, which is about one order higher than the response of resistors [~1.2, i.e., 20%, Figure 11(a), Tables 1 and 2].
|R0||R0||Resistance due to external connectors|
|R1C1||R1C1 (only in air)||TiO2 grain boundaries|
|R3C3||Amorphous carbon and impurities|
4. Conclusions and future directions
Design of composite materials based on combinations of nanoparticled titanium dioxide and multiwalled carbon nanotubes, using small amounts of nanotubes for low cost room temperature ammonia sensors have been demonstrated. For sensor fabrication,
We acknowledge the felloship provided by the Consejo Nacional de Ciencia y Tecnología, CONACYT-México (M. Sánchez), we thank to M.L. Ramón García for the XRD analysis and to P. Altuzar Coello for TGA analysis.