Application of candle soot CNPs-TiO₂-PVP composite in the detection of volatile organic compounds with aldehyde, amine and ketone functional groups by resistance and impedance responses

Titanium (iv) dioxide anatase nano-powder (TiO₂), N,N-Dimethylformamide (DMF) and Poly (vinylpyrrolidone) (PVP) were purchased from Sigma-Aldrich (South Africa) and white candles (Light House candles) were purchased from a local store in Johannesburg, South Africa.

2.2.1. Collection of candle soot CNPs

2.2.2. Preparation of nanocomposite sensor

2.2.3. Characterization of nano-materials

2.2.4. The set-up of gas sensing system featuring an LCR meter

Studying the fabricated sensor entailed setting up a gas sensing system with a high precision LCR meter, model: (KEYSIGHT E4980AL) with AC input signal (amplitude: 0.5 V; frequency: 25 kHz), a computer, 20 L multi-neck round bottom flask and a vacuum pump [11]. The LCR meter was connected to the computer for data display and the fabricated sensor using two insulated wires to measure the response signals during sensing. The sensor converts the chemical reaction between the analyte and the composite into an electrical response, this response is then converted into readable data by the LCR which is displayed on the PC.

The sensor was then placed inside an enclosed 20 l round bottom flask via one of the openings as illustrated in figure 1. A second opening was connected to a vacuum pump by a tube to flush out the analyte after exposure and to pump in fresh air into the camber. However, sensing performance was conducted under one bar atmospheric pressure and 42% RH. Each analyte (butyraldehyde, diethylamine and isobutyrophenone) was tested separately using volumes 1, 2, 3, 4 and 5 μl. To begin, 1 μl of an analyte was injected inside the flask and left for 10 min of exposure time. Thereafter the remaining analyte gas was flushed out using the vacuum pump for 4 min and the sensor was left to recover for 1 min. The procedure was repeated up to 5 μl analyte. Isobutyrophenone was tested using similar volumes but the exposure time was increased to 12 min, the analyte was flushed for 4 min. and the sensor was left to recover for 2 min. This is due to the higher boiling point of isobutyrophenone which is three times higher than diethylamine and two times higher than butyraldehyde.

Zoom In Zoom Out Reset image size

The concentration of the injection of each analyte was calculated in parts per million (ppm) using equation (1) below [35]

$$\begin{eqnarray}&&C=\displaystyle \frac{22.4\rho {\rm T}Vs}{273MV}\times 1000\end{eqnarray} \tag{ 1 }$$

where C is the concentration of the analyte at room temperature in (ppm), ρ is the density of the liquid analyte in (g/mL⁻¹), T is the testing temperature in (K), versus is the volume of liquid analyte in (μl), M is the molecular weight of the analyte in (g/mol⁻¹), V is the volume of the round bottom flask in (L).

The responses were reported as variations of resistance:

$$\begin{eqnarray}&&{\rm{Resistance}}:{\rm{\Delta }}R={R}_{VOCS}-{R}_{air}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&Impedance:{\rm{\Delta }}Z={Z}_{VOCS}-{Z}_{air}\end{eqnarray} \tag{ 3 }$$

where R_VOCs is the resistance of the sensor with VOCs (volatile organic compounds) and R_air is the resistance of the sensor without VOCs [36].

Figure 2 shows XRD spectra of the individually prepared components and the nano-composite material used in the fabrication of the solid-state sensor. The XRD spectrum of pure candle soot CNPs (figure 2(a)) exhibited two Bragg diffraction peaks; the first peak was observed at 2Θ value 25.21° which was a high-intensity broad peak and the second peak which was a lower intensity peak was observed at 2Θ value 44.06°. The high-intensity value corresponds to the (002) plane representing graphite as well as amorphous carbons [36, 37]. The lower intensity value corresponds to the (101) plane representing the crystalline graphitic carbon structure of CNPs or the diamond phase on the soot [38, 39].

Zoom In Zoom Out Reset image size

Figure 2(b) shows the two XRD peaks of PVP, one being a high-intensity broad peak and a low-intensity broad peak at 2Θ 11.88° and 2Θ 20.65° respectively, representing the polymer's amorphous nature [40, 41]. Diffraction peaks of pure commercial TiO_{2 (}figure 2(c)) were observed at 2Θ values 25.91°, 37.41°, 38.44°, 39.20°, 48.82°, 54.54°, 55.47°, 63.12°, 69.09°,70.80°, 75.45° and 83.33°; they are assigned to planes (101), (103), (004), (102), (200), (105), (211), (204), (116), (220), (215) and (224) respectively, denoting an anatase phase with no impurities and secondary phases [40, 42, 43]. As reported, [16] anatase phase TiO₂ enhances the sensor's performance at room temperature. The diffraction pattern of the composite is shown in figure 2(d). It is clear that all the TiO₂ peaks are visible, one of the two peaks of CNPs and PVP were also displayed on the graph and confirming that all materials were present in the composite used to fabricate the solid-state sensor.

The Raman spectrum of candle soot CNPs presented in figure 3(a) shows two standard carbon characteristic peaks which are very distinct, one of which is the disordered graphitic phase (D band) at 1352 cm⁻¹ that represents amorphous carbons. The second peak which is the G band at 1598 cm⁻¹ denotes crystalline graphitic carbons, confirming the results from the PXRD pattern (See figure 2(a)). The intensity ratio of the bands (I_D/I_G) which expounds the defects present in carbon material was found to be 0.85 [36, 38, 44]. Figure 3(b) shows the Raman spectrum of six active modes for anatase TiO₂, 3E_g (147, 197 and 640 cm⁻¹), 2B_1g (397 and 517 cm⁻¹) and 1A_1g (517 cm⁻¹). The symmetrical stretching of O-Ti-O is represented by E_G vibrational mode at 147 cm⁻¹ and the peak is very strong and intense while the rest of the peaks represents a highly crystalline anatase TiO₂ phase. The peaks correspond to previously reported findings [45, 46].

Zoom In Zoom Out Reset image size

BET analysis is an important aspect of sensing materials since the basics of the sensing mechanism involving the adsorption-desorption flow of the analyte gas. This experiment took interest in the pore volume (cm³ g⁻¹) and pore size (nm) (Table 1) which affects the diffusion rate, the analyte molecule and the sieving properties. The surface area (m² g⁻¹) of the materials was also determined as it affects the dissolution rate and adsorption capacity. Candle soot CNPs showed a pore volume of 0.58 cm³ g⁻¹ , a pore size of 21.73 nm and a surface area of 125.3 m²g⁻¹. The pore volume and surface area are higher than the previously reported pore volume of 0.29 cm³g⁻¹ [44] and a surface area of 53.2 m²g⁻¹ [8] and 62.18 m²g⁻¹ [11].

The TiO₂ showed a pore volume of 0.19 cm³g⁻¹, a pore size of 16.5 nm and a surface area of 46.5 m²g⁻¹.

Figure 4 shows the TEM images of the materials which give insight into their morphological properties. The image in figure 4(a) reveals the particles' fractal-like network nature and their uniform spherical structure. The spherical particles are infused with one another forming a branched-like structure which may result from diffusive bonding or diffusive adhesion between each particle thus increasing their mechanical stability [47]. The particle size ranges from 40 nm to 65 nm with an average of 52 nm denoting their smaller size characteristic. Figure 4(b) further shows the particles at a higher magnification which clearly shows fringes with a lattice spacing of 0.42 nm.

Zoom In Zoom Out Reset image size

Figure 4(c) showed the image of anatase TiO₂, the structures were very clear and pristine nanoparticles with a distinct variety of sizes and shapes. The size of the particles ranged from 34 nm to 74 nm, the two different shapes that the particles had were spherical and other particles had a heptagon-oval shape. Unlike other nanoparticles, hydrothermally synthesized anatase TiO₂ has been reported to take on a lot more different hierarchical structures depending on the temperatures and precursor material used [48]. Figure 4(d) shows the visible lattice fringes with a lattice spacing of 1.10 nm and figure 4(e) shows the TEM image of the composite, the agglomerated spherical candle soot CNPs, as well as the TiO₂ particles, are visible.

Figure 5(a) shows the surface morphology of candle soot CNPs which appears as clustered structures, their morphological characteristics are not clear as the magnification of the SEM was low but aggregates are visible. The sample surface composition is displayed in figure 5(b) which shows that 92.9% of the sample composition was carbon and 7.1% was oxygen atoms which are combustion products of the candle wax flame. Figure 5(c) shows the surface morphology of TiO₂ which appears as mesh-like structures with pores and the EDS displayed the samples composition of 68.1% titanium and 31.9% oxygen atoms denoting pristine TiO₂.

Zoom In Zoom Out Reset image size

Unlike other conventional studied solid-state gas sensors which are fabricated using one or two n-type SMOs or carbon material and n-type SMOs as a sensing material, this study used carbon material (CNPs), n-type SMO (TiO₂) and a polymer (PVP) to investigate the sensor performance. The reason candle soot CNPs and TiO₂ were used lies in their high surface area, their electrical conductivity as well as their novelty in gas sensing application [10]. SMOs are mostly used in gas sensing due to their high carrier mobility, conductivity and stress tolerance [49]. TiO₂ like other metal oxides has exhibited excellent properties in sensing application due to their polymorphic behaviour, stability and it is preferred to be used in chemi-resistive gas sensors which entails an efficient adsorption-desorption mechanism of reducing and oxidizing agents [49].

Here, three different vapour analytes with different organic functional groups, aldehyde (butyraldehyde), amine diethylamine) and ketone (isobutyrophenone) were studied for their response to the fabricated solid-state gas sensor. The vapour concentration was in the range of 8 ppm −68 ppm depending on the density of the analytes. The relative resistance (∆R) and impedance (∆Z) parameters were used for all three analytes to quantify the response against the vapour concentration (ppm).

In both parameters, the sensor responded well and increased as the analyte vapour concentration increased (see figures 6 and 7). As the analyte vapour interacts with the sensor's active layer, the responses of the sensor sharply increased and returned to the baseline when the analyte vapour was removed from the vessel (figure 6). This can be explained by the diffusion of analyte molecules through the exposed sensing material to occupy the free active sites. During this time the electrical property of the sensing material changes and is recorded as an increase in resistance (in this study) until most of the active sites are occupied. When the unreacted vapour molecules were removed from the vessel, by use of a diaphragm vacuum pump, the active sites became unoccupied and the sensor regenerated itself with oxygen species.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The increase in relative resistance response (∆R) of the three analytes vapour as the concentration increases indicates a linear relationship where the ∆R response is directly proportional to the concentration (figures 6(b), (d) and (f)). Although the three analytes behaved in the same manner, only the response for butyraldehyde vapour demonstrated plateaus for the first three injections (figure 6(a)).

Generally, for the butyraldehyde and diethylamine vapour, the ∆R responses were better than isobutyrophenone. The response peaks for isobutyrophenone are sharp triangle-shaped rather than a flat top bell-shaped curve. This might be due to the very low vapour pressure of 0.11 mm Hg at 25 °C of isobutyrophenone as compared to that of butyraldehyde and diethylamine, 111 mm and 237 mm Hg at 25 °C, respectively. An analyte with low vapour pressure usually has a low vapour diffusion rate which slows the analyte vapour interaction with the sensing materials resulting in longer response times than analytes with high vapour pressure.

A similar relationship between the concentration of the vapour analytes and ∆Z response was observed except for the isobutyrophenone (figure 7) where a linear relationship was observed only at low concentrations (See figure 7f inset). Generally, as the concentration increased the relative response in ∆Z increased. Also, similarly to the ∆R response, the ∆Z responses from butyraldehyde and the diethylamine vapour showed some form of bell-shaped curves, while the response curve for isobutyrophenone vapour was a sharp triangle-shape.

Although the three analytes respond to the solid-state sensor similarly for a single measurement parameter, the sensitivity of the sensor towards the analytes for relative resistance and relative impedance varies. The sensitivity of the sensor is considered the gradient of the curve, $\tfrac{\partial {\rm{\Delta }}R}{\partial {C}_{ppm}}$ and $\tfrac{\partial {\rm{\Delta }}Z}{\partial {C}_{ppm}}$ for relative resistance and impedance, respectively. For ∆R response, the sensor's sensitivity towards butyraldehyde, diethylamine and isobutyrophenone vapours were 0.04, 0.07 and 0.03 Ω ppm⁻¹, respectively (See table 2). The sensor was twice more sensitive toward diethylamine vapour followed by butyraldehyde and isobutyrophenone. In the case of the impedance parameter, the sensitivity of the sensor is different from the relative resistance parameter; the sensor was three times more sensitive towards butyraldehyde vapour than diethylamine vapour. An interesting observation is that it is possible to use the same sensor to distinguish between two analytes vapour simply by changing the parameters from resistance to impedance and vice versa.

Table 2. Sensitivity of Butyraldehyde, Diethylamine, Isobutyrophenone and response-recovery time in seconds.

Analyte at 2 μl	Resistance$\tfrac{\partial {\rm{\Delta }}R}{\partial {C}_{ppm}}$(Ω ppm−1)	Impedance$\tfrac{\partial {\rm{\Delta }}Z}{\partial {C}_{ppm}}$(Ω ppm−1)	Response time(s)	Recovery time(s)
Butyraldehyde	0.04	0.14	145	132
Diethylamine	0.07	0.04	193	130
Isobutyrophenone	0.03	0.20 a	407	187

⁴ At low concentration

The response and recovery time of all the three analytes were determined from ∆R parameter and injection volume of 2 μl. The response time is defined as the time required for the sensor to reach 90% of the maximum response of the signal measured when exposed to the target analyte vapour. The time required for the sensor to recover 90% from its maximum response upon the removal of the target analyte vapour is considered as the actual recovery time. Accordingly, as table 2 shows, butyraldehyde had the fastest response time of 145 s followed by diethylamine with 193 s and finally isobutyrophenone with 407 s. The slowest response time for the isobutyrophenone vapour might be due to the lowest vapour pressure of 0.11 mm Hg at 25 °C as compared with the other two as discussed earlier. Another factor that may have led to its slower response could be attributed to a slow diffusion rate through the thin film sensing materials. This is due to the size of the isobutyrophenone as a result of its bulky structure that contains a benzene ring. A similar outcome is seen in recovery time, where isobutyrophenone showed a longer recovery time and the diethylamine was only 2 s quicker with 130 s than the butyraldehyde with 132 s. A typical response and recovery time graph is shown in (figure 8) using 2 μl (26.91) of butyraldehyde.

Zoom In Zoom Out Reset image size

The basic adsorption-desorption process of the analyte vapour molecules on the surface of the sensing materials determines the type of sensing mechanism of a sensor [50]. The analyte molecules need to diffuse through the sensing materials then interact with adsorbed oxygen species on the sensing materials. The rate of adsorption and desorption of oxygen species on the surface of n-type and p-type SMOs results in the change in response [51].

When an n-type SMO is exposed to air, oxygen molecules from the air are adsorbed on the surface of the SMO and occupy the previously formed oxygen vacancies by taking electrons from the SMO conduction band. This results in the generation of different negatively charged oxygen species (O₂ ⁻, O⁻, and O²⁻) on the surface of the SMO and the formation of an electron depletion layer (EDL) which forces the conduction band to bend upwards and the Fermi level to move down [52]. Analyte molecules react with the adsorbed oxygen species thus promoting oxidation to form CO₂ and H₂O, the electrons are then released back into the conduction band and as a result, the EDL shrinks, creating oxygen vacancies, Fermi level and the conduction band re-bounce back to normal. This leads to a decrease in resistance for the n-type SMOs. With a p-type SMO, the oxygen adsorption model is still applicable after the SMO is exposed to air. However, instead of forming an EDL, it forms a hole accumulation layer which results in a decrease in electrical resistance and during an interaction of the negatively charged oxygen species with a gas in a redox reaction, an increase in resistance occurs [53, 54].

The sensing materials synthesised in this study were composed of TiO₂, CNPs and PVP, of which the oxygen reactive species can be found on the surface of both TiO₂ and CNPs. According to the EDS result, the CNTs are composed of 7% oxygen and 93% carbon atoms (See figure 5(b)). Since those oxygen atoms on the surface of the CNPs can transform into reacting oxygen species (O₂ ⁻, O⁻, and O²⁻), they are responsible for the total decomposition of VOCs [11, 12]. Although one of the sensing materials in the fabricated solid-state sensor is an n-type SMO, we observed that the sensor performed as a p-type sensor for all three VOCs [11, 12], 54].

Mani et al, [55] proposed that the sensing mechanism of nanostructured ZnO:Co thin film for ammonia vapour liberated N₂ and H₂O molecules. In a separate study by Calestani et al, [56], a high conductance change for similar concentrations (ppb) of propionaldehyde (or propanal) and acetaldehyde (or ethanal) was observed using zinc oxide (ZnO) tetrapods based gas sensor. They proposed that the propionaldehyde molecule was completely oxidized to CO₂ and H₂O molecules and as a result higher number of electrons were transferred to the ZnO. Similarly, Kim et al, [57] revealed that by optimizing Pt- and Pd-functionalized ZnO nanowires (NW) gas sensors, they could selectively detect toluene and benzene gases, respectively. The sensing mechanism is also explained based on the reaction between the volatile gas molecules adsorbed onto the surface of the sensor and the highly reactive oxygen species, which liberates CO₂ and H₂O.

Therefore, the sensing mechanism of the solid-state gas sensor designed in this study for each of the three analytes was summarized as follows:

For the Diethylamine:

$$\begin{eqnarray}&&2{C}_{4}{H}_{11}{N}_{(g)}+13\displaystyle \frac{1}{2}{O}_{2(ad)}^{-}\to {N}_{2}+8C{O}_{2}+11{H}_{2}O+27{e}^{-}\end{eqnarray} \tag{ 4 }$$

For the Butyraldehyde:

$$\begin{eqnarray}&&{C}_{4}{H}_{8}{O}_{(g)}+12{O}^{-}\to 4C{O}_{2}+4{H}_{2}O+12{e}^{-}\end{eqnarray} \tag{ 5 }$$

For the Isobutyrophenone

$$\begin{eqnarray}&&{C}_{10}{H}_{12}{O}_{(g)}+25{O}^{-}\to 10C{O}_{2}+6{H}_{2}O+25{e}^{-}\end{eqnarray} \tag{ 6 }$$

Table 3 shows the ∆R of the sensor at 25 ppm of the three vapour analytes. Interestingly, the sensor resistance for the diethylamine and isobutyrophenone interaction was 2.9 and 2.5 Ω, respectively. However, for butyraldehyde vapour, the sensor's resistance was lower than the two analytes which were less than half of Isobutyrophenone, 1.1 Ω. This change in the sensor's resistance can be explained from the point of the hole accumulation layer theory. During sensing, the sensor's resistance increased as the concentration of the analytes increase, such characteristic is a p-type sensor behaviour (See figures 6 and 7). Therefore, our sensing materials captured electrons to form absorbed oxygen species and hole accumulation layers. The interaction between analytes vapour and the charged oxygen species from the surface of the sensing materials led to the oxidation of analyte vapour by the negatively charged oxygen species thus generating electrons that reduce the hole concentration in the shell layer which results in an increase in the resistance.

Interestingly, if we look at the decomposition reactions equation for diethylamine and isobutyrophenone, higher numbers of electrons were transferred to the sensing materials. However, for butyraldehyde vapour, it was less than half of the Isobutyrophenone, only 12e^- were transferred to the sensing materials [56]. This makes it consistent with the relative resistance of the sensor and the electron injection back to the sensing materials.