Mechanistic roles of substitutional Fe dopants on catalytic acetylene-sensing process of flame-made SnO₂ nanoparticles

2.1 Synthesis of metal-oxide nanoparticles

0–2 wt% Fe-doped SnO₂ nanostructures were made from a precursor mixture by the FSP system formerly developed by our research group (Liewhiran et al., 2012, 2013; Sukunta et al., 2017; Punginsang et al., 2015; Singkammo et al., 2015). Briefly, the precursor solution (0.50 M) was made by dissolving suitable amounts of tin (II) 2-ethylhexanoate (Aldrich, 95%) and iron (III) acetyl-acetonate (Aldrich, 99.9%) in the xylene solvent (CarloErba, 98.5%). In a nominal synthesis process, the precursor was injected into a FSP chamber using a syringe pump at 5 mL/min and then spread into droplets by O₂ flow (5 L/min) defined as the 5/5 flame condition with a fixed O₂ pressure of 1.5 bar. The spray was combusted by subsidiary flamelets formed with oxygen (2.46 L/min) and methane (1.19 L/min). A sheath O₂ flow of 3.92 L/min was applied coaxially about the nozzle to regulate and control the sprayed flame. SnO₂ nanoparticles were then nucleated after vaporization and incineration of the droplets followed by Fe-doping, coagulation and coalescence. Finally, nanoparticles were attracted with the use of vacuum onto a 25.7 cm-diameter microfiber glass filter (Whatmann GF/A). The unloaded SnO₂ powder was named as P-0 while doped SnO₂ powders with 0.1, 0.2, 0.5, 1 and 2 wt% Fe were labeled as P-0.1Fe, P-0.2Fe, P-0.5Fe, P-1Fe and P-2Fe, respectively.

2.2 Gas sensor fabrication

To form a paste for spin-coating of sensing films, as-prepared flame-made 0–2 wt% Fe-doped SnO₂ nanoparticles were meticulously mixed and crushed in a mortar with the binder solution (0.30 mL) containing ethyl cellulose (Fluka, 30–70 mPa s) and α-terpineol solvent (Aldrich, 90%) (Kotchasak et al., 2018). The obtained paste was deposited by spin-coating as a thick sensing film onto Au interdigitated electrodes prefabricated on an Al₂O₃ substrate (4 mm × 5.5 mm × 0.4 mm). The sensors were stabilized in a furnace at 450 °C for 3 h to completely remove binder. The sensors coated using P-0 to P-2Fe powder samples were consequently defined as S-0 to S-2Fe.

2.3 Particle and sensing film characterizations

The phase and crystallinity of the unloaded SnO₂ and 0.1–2 wt% Fe-doped SnO₂ nanoparticles and sensing films were assessed by X–ray diffraction in the glancing incident mode (GI-XRD) (Rigaku TTRAX III diffractometer) using CuK_α X-ray source (20 kV, 20 mA) scanned with a rate of 3°/min. The pore size distributions and specific surface areas of nanoparticles were measured by nitrogen adsorption, which was analyzed using Brunauer–Emmett–Teller (BET) and Barrett-Joyner-Halenda (BJH) methods. The structural morphologies of nanopowders and sensing films were inspected by high-resolution transmission electron microscopy (HR-TEM: JSM-2100Plus, JEOL) and scanning electron microscopy (SEM: JSM-6335F, JEOL). The oxidation states of iron in nanoparticles samples were evaluated by X-ray absorption spectroscopy (XAS) and X-ray photoemission spectroscopy (XPS). Moreover, the optical band gap energy of nanopowder was measured by UV–visible spectroscopy.

2.4 Gas-sensing measurement

The gas-sensing characteristics of Fe-doped SnO₂ sensors were evaluated towards C₂H₂, C₂H₅OH, NO₂, H₂, NH₃, CO₂, NO, H₂S, CH₄, C₂H₄O, C₂H₄ and N₂O at the atmospheric pressure in a closed stainless steel chamber having a dynamic volume of ∼0.7 L. The sensors were tested by the typical flow through method at varying working temperatures from 200 to 350 °C controlled by an external dc-powered heater. In addition, the characteristics towards C₂H₂ as the main target gas were determined over the concentration range of 0.15–3 vol%. A flow of dry air as a carrier was delivered to combine with the designated concentration of gas sample balanced in dry air at a regular T-junction with a combined gas flow rate of 2 L/min. Computer-controlled mass flow controllers (5850E, Brook Instruments) were used to precisely regulate gas flow rates. The sensing devices were subjected to a target gas for 15 min at each concentration and the dry air flow was recovered for 35 min. The resistances of six sensors were simultaneously monitored at a constant bias of 10 V using a picoammeter (Keithley model 6487). The sensor response (S) was determined from R_a/R_g, where R_g was the steady-state resistance in a reducing gas (C₂H₅OH, H₂, NH₃, CO₂, H₂S, NO, CH₄, C₂H₄O and C₂H₄) and R_a was a baseline resistance in dry air. For oxidizing gas such as NO₂ and N₂O, the response function was inversed to R_g/R_a (Samerjai et al., 2012). The response time (t_res) was set as the time taken to reach 90% of the response signal and the recovery time (t_rec) was the time period to recover 90% of the baseline resistance.

3.1 Structural characteristics of flame-made nanopowders and sensing films

XRD data of flame-made undoped and 0.1–2 wt% Fe-doped SnO₂ (P-0 and P-0.1Fe to P–2Fe) powders are reported in Fig. 1(a). It is apparent that sharp diffraction peaks of all powders are only corresponding to the tetragonal cassiterite SnO₂ structure (JCPDS file No. 41-1445), indicating a single phase of highly crystalline SnO₂ material. In addition, all SnO₂ nanopowders show similar texturization with four major crystallographic planes at (1 1 0), (1 0 1), (2 1 1) and (2 0 0). Moreover, the peaks are slightly widened as the Fe-doping level rises from 0 to 2 wt%, implying the decrease of crystallite size. The crystallite sizes of nanoparticles calculated from Scherrer’s equation is plotted as a function of Fe-doping concentration as embedded in Fig. 1(a). From the resulting plot, the average crystallite size slightly decreases as the Fe content rises from 0 to 0.5 wt% (∼10.6 to ∼10.3 nm) and then decreases rapidly from ∼10.3 to ∼8.4 nm as the Fe content increases additionally up to 2 wt%. Hence, the Fe dopants slightly limit grain enlargement of SnO₂ crystallites, which can be attributed to the strain-induced grain-diffusion barrier effect (Kim et al., 2007). The absence of Fe or Fe₂O₃ secondary peaks may be due to either the low amount of iron species or the development of the Fe-SnO₂ solid solution. Fig. 1(b) illustrates the related XRD patterns of all sensing films (S-0 and S-0.1Fe to S–2Fe) on Al₂O₃ substrates with Au/Cr electrodes after sensing test. It confirms similar crystallinity between the sensing films and their relevant powders having the same phase with no secondary phase of Fe or FeO_x. It can be noticed that the peaks of Au electrodes (JCPDS file No. 04-0784) and Al₂O₃ substrate (JCPDS file No. 46-1212) exhibit dominant magnitudes relative to those of sensing films due to high X-ray diffraction sensitivity of large-area Au interdigit patterns and large-grain Al₂O₃ crystallites of the substrate.

Close

Fig. 1

XRD data of 0–2 wt% Fe-doped SnO₂ (a) nanoparticles (P-0 to P–2Fe) and (b) sensing films (S-0 to S–2Fe) after annealing and sensing test.

Export to PPT

Fig. 2(a)–(d) illustrate typical bright field (BF)-TEM images with selected area electron diffraction (SAED) patterns (lower inset) and related high-resolution (HR)-TEM images (right zoom) of P-0, P-0.1Fe, P-0.5Fe and P–2Fe, respectively. All sample images demonstrate the primary particles with approximately round and polygonal shapes having various dimensions (3–20 nm). Additionally, the nanoparticles seem to be slightly smaller with increasing Fe-doping concentration and all particles appear to be the same phase, indicating the absence of iron or iron oxide secondary species. The relevant SAED patterns exhibit dotted ring features, specifying that particles are mainly semicrystalline with principal diffraction rings matched with principal planes of the cassiterite SnO₂ phase, in accordance with the perceived XRD data.

Close

Fig. 2

BF-TEM images with SAED patterns (insets) and HR-TEM images of (a) undoped SnO₂ (P-0), (b) 0.1 wt% Fe-doped SnO₂ (P-0.1Fe), (c) 0.5 wt% Fe-doped SnO₂ (P-0.5Fe) and (d) 2 wt% Fe-doped SnO₂ nanoparticles (P-2Fe).

Export to PPT

The related HR-TEM images shows various nanoparticles displaying lattice fringes, which can be indexed to principal planes of cassiterite SnO₂ phase. Additionally, the planes of P-2Fe tend to have slightly smaller d-spacing than those of P-0, P-0.1Fe and P-0.5Fe, signifying a smaller lattice constant due to substitutional doping of Fe atoms in the lattice. A replacement of Sn⁴⁺ by Fe³⁺ species should result in a reduced lattice constant due to its smaller ionic size (Bagheri-Mohagheghi et al., 2009; Kaur et al., 2012)) relative to that of Sn⁴⁺ (Lee et al., 2018; Sukunta et al., 2017; Kaur et al., 2012)). The Fe-Sn substitution is concordant with the Hume–Rothery principle due to the smaller difference of the Sn⁴⁺ and Fe³⁺ effective ionic radii than the upper bound of 15%. Besides, it agrees with some studies of Fe-doped SnO₂ nanomaterials, which indicate the conception of solid Fe-SnO₂ solution via the diffusion process (Bagheri-Mohagheghi et al., 2009; Beltran et al., 2010; Rao et al., 2017; Sambasivam et al., 2011).

The illustrative nitrogen adsorption–desorption isotherms with relevant pore size distributions (insets) of P-0, P-0.1Fe, P-0.5Fe and P-2Fe are exemplified in Fig. 3(a)–(d). It is observed that various isotherms similarly conform to the II-H3 IUPAC pattern, signifying the unhindered multilayer/monolayer adsorption on microporous-mesoporous adsorbents with capillary condensation (Sing et al., 1985). The respective BJH pore size distributions further reveal the dependence of multi-mode pore size distribution on the Fe-doping content. The distribution of unloaded SnO₂ powder (P-0) displays two eminent maxima around 10–20 Å (Inset of Fig. 3(a)) but that of P-0.1Fe has one dominant maxima at ∼15 Å (Insets of Fig. 3(b)). In the case of P-0.5Fe (Inset of Fig. 3(c)), the pore size distribution exhibits seven narrow peaks at the pore diameter of ∼15, ∼17, ∼19, ∼21, ∼26, ∼29 and ∼ 32 Å, respectively. At the highest Fe-doping level of 2 wt%, three notable peaks are seen ranging from ∼15 to ∼20 Å with one main maxima at ∼16 Å. Also, the peak magnitudes of P-2Fe are considerably higher than those of P-0, P-0.1Fe and P-0.5Fe. The results imply that P-2Fe has relatively high micropore density relative to P-0, P-0.1Fe and P-0.5Fe while the Fe contents lower than 2 wt% result in the combination of micropores (<20 Å) and mesopores (20–500 Å). In addition, the mesopore sizes tend to decrease with increasing Fe-doping concentration and the high Fe content results in higher pore density. Fig. 4(e) reports the BET specific surface area (SSA_BET) and total pore volume in relation to the Fe content. It can be seen that SSA_BET initially varies slightly as the Fe content rises from 0 to 0.5 wt% (around 57–58 m²/g) but then increase markedly at the high Fe contents from 1 to 2 wt% (76.3–86.1 m²/g). Similarly, the total pore volume decreases slowly from 0.56 to 0.27 cm³/g as the Fe content increases from 0 to 0.5 wt% before increasing significantly to 0.72 cm³/g when the Fe content additionally rises to 2 wt%. The large total pore volume and specific surface area may be associated with the very fine micropore features of the nanoparticle structure at high Fe contents.

Close

Fig. 3

N₂ adsorption isotherms with the corresponding BJH pore size distributions (insets) of (a) P-0, (b) P-0.1Fe, (c) P-0.5Fe and (d) P-2Fe. (e) Specific surface area: SSA_BET (red) and corresponding total pore volume: V_p (blue) vs. Fe content.

Export to PPT

Close

Fig. 4

XAS spectra of 2 wt% Fe-doped SnO₂, FeO and Fe₂O₃. Inset table: Estimated XAS parameters.

Export to PPT

The oxidation states and chemical species of Fe ions are determined from the standardized Fe K-edge XAS spectrum of 2 wt% Fe-doped SnO₂ nanoparticles relative to those of standard Fe compounds (FeO and Fe₂O₃) as demonstrated in Fig. 4. It is evident that the absorption edge of 2 wt% Fe-doped SnO₂ sample is located at 7121.41 eV, which lies much closer to that of Fe₂O₃ (7121.80 eV) compared with that of FeO (7115.95 eV). The relative percentage of Fe³⁺ and Fe²⁺ oxidation states are calculated by the interpolation of edge energies to be 93.45% and 6.55%, respectively as reported in the inset table (Fig. 4). Thus, the main oxidation state of Fe-doped flame-made SnO₂ nanoparticles is confirmed to be Fe³⁺. It is also highly likely that Fe³⁺ states substitute Sn⁴⁺ ones in the lattice because the secondary phase of Fe₂O₃ cannot be observed according to the HR-TEM image.

Fig. 5 illustrates the high-resolution XPS spectra consisting of Sn 3d, Fe 2p and O 1 s principal levels of undoped and 2 wt% Fe-doped SnO₂ powders and sensors (P-0, S-0, P-2Fe and S-2Fe) after sensing test. For the Sn element, the spin-splitting pairs (Sn 3d_5/2:Sn 3d_3/2) of P-0 and P-2Fe (Fig. 5(a) and (c)) can be independently deconvoluted into the leading pairs located at 486.3:494.7 and 486.6:495.0 eV and the other minor pairs positioned at 487.1:495.7 and 487.8:496.1 eV, respectively. In contrast, the doublet pairs of S-0 and S-2Fe (Fig. 5(a) and (c)) can be fitted with individual doublet pairs of Gaussian peaks observed at 486.6:495.0 and 486.7:495.1 eV, respectively. The highest pairs of all samples can be associated with the same Sn⁴⁺oxidation state (SnO₂) while the tiny pairs of P-0 or P-2Fe can match with the Sn⁴⁺ state of Sn(OH)₄ due to humidity adsorption (Kwoka et al., 2011). The results indicate that the states of Sn species are insignificantly affected by Fe-doping but depend on the form of materials. The materials in powder form contain Sn(OH)₄ species, which are absent in the related sensing films due possibly to humidity desorption after annealing and sensing test at elevated temperatures. Concerning the O element, the O 1s levels of P-0:P-2Fe (Fig. 5(b): (d)) can be similarly separated into four components peaked at 530.4:530.5 (main peak), 531.4:531.8, 532.3:532.8, and 533.2:534.1 eV, respectively. For the sensing films, the O 1s peak of S-0 (Fig. 5(b)) can be divided into three peaks centered at 530.1, 531.1 and 532.1 eV while that of S-2Fe (Fig. 5(d)) can be broken into four parts peaked at 530.4, 531.3, 532.5 and 533.9 eV, respectively. The major O 1s component can be attributed to surface lattice oxygen (O²⁻) of Fe-doped SnO₂ structures while the minor ones located at 531.3–531.8, 532.5–532.8 and 533.9–534.1 eV may be attributed to the superoxide species ( ${\text{O}}_2^{-}$ ), a hydroxyl (OH⁻) group from humidity and loosely-bound oxygen molecules, respectively (Watts, 1994). The results suggest that the states of oxygen species are influenced by Fe doping and the Fe-doped SnO₂ samples tend to exhibit higher contents of chemisorbed and physisorbed oxygen species on surface. In the case of Fe element (Fig. 5(e)), the Fe 2p doublet pair overlaps with the Sn 3p_3/2 peak of SnO₂ so that Fe 2p peaks must be found by the deconvolution of overlapped peaks. Due to relatively weak Fe 2p signals, only Fe 2p_3/2 peaks of P-2Fe and S-2Fe can be found at 710.8 and 711.3 eV, respectively. They can match well with the Fe⁺³ oxidation state (Fe₂O₃) (Pradhan et al., 2014; Bhargava et al., 2007). The XPS result of Fe is in accordance with the Fe⁺³ oxidation state identified by XAS. It should be remarked that the XAS and XPS data of Fe-doped SnO₂ nanoparticles with lower Fe contents are omitted because their Fe signals are either poor or absent.

Close

Fig. 5

High-energy resolution core-level XPS spectra of undoped and 2 wt% Fe-doped SnO₂ powders (P-0 and P-2Fe) and sensors (S-0 and S-2Fe): (a, c) Sn 3d, (b, d) O 1s and (e) Fe 2p.

Export to PPT

Fig. 6 shows UV–Vis absorption spectra of 0 − 2 wt% Fe-doped SnO₂ nanoparticles versus the wavelength in the range of 200–700 nm. It is seen that all samples show strong absorption edges in the UV region (300–330 nm) corresponding to the band gap transition and the edge position of Fe-doped SnO₂ nanoparticles moves toward longer wavelength as the Fe content rises from 0 to 2 wt%. The optical band gap energy (E_g) can then be calculated using Tauc’s relation (Shaikh et al., 2017): $$\alpha =C\times \frac{{(h\nu -E_g)}^n}{h\nu }$$ where C is a constant, α is the absorption coefficient taken from the UV-spectra, hν is the photon energy and n is a constant depending on the type of transition: n = 1/2, 3/2, 2, and 3 for direct allowed, direct forbidden, indirect allowed, and indirect forbidden band gaps, respectively. The absorption data can be best fitted with n = 1/2, confirming the direct transition nature of SnO₂ semiconductor. Thus, the optical band gap of Fe-doped SnO₂ nanoparticles can be computed from the intercept of the extrapolated linear part (α = 0) of (αhv)² versus hv as displayed in the inset of Fig. 6. The band gap of P-0 (3.52 eV) is close to the reported values of SnO₂ materials prepared by various techniques (Saleh et al., 2016). With increasing Fe-doping content, the band gap energy decreases to 3.47, 3.44, 3.38, 3.16 and 2.95 eV for P-0.1Fe, P-0.2Fe, P-0.5Fe, P-1Fe and P-2Fe, respectively. The result reveals that the optical energy band gap decreases with increasing Fe content, which may be attributed to the doping effect of Fe atoms that introduces additional defect-state energy levels close to the band edges (Mani et al., 2017; Rao et al., 2017).

Close

Fig. 6

UV–Visible absorbance spectra as a function of wavelength for SnO₂ nanoparticles with different Fe-doping levels. Inset: Tauc plots between (αhv)² vs. photon energy.

Export to PPT

Fig. 7(a)–(f) demonstrate the cross-sectional SEM images of sensors (S-0 to S–2Fe) after sensing measurements. The images show various sensing films with thicknesses of ∼10–15 μm comprising densely collective nanoparticles well arranged on solid alumina substrates. It was found that the densification of sensing films occurred after annealing at 450 °C. From the results, the Fe-doping level does not markedly affect the film density and the evenness of the film thickness endorses the reliability of powder-pasting and spin-coating methods.

Close

Fig. 7

SEM micrographs of undoped SnO₂ and Fe-doped SnO₂ sensing films: (a) S-0, (b) S-0.1Fe, (c) S-0.2Fe, (d) S-0.5Fe, (e) S-1Fe and (f) S–2Fe on Au/Al₂O₃ substrates.

Export to PPT

3.2 Gas-sensing properties

The gas-sensing characteristics of SnO₂ sensors with different Fe concentrations were characterized in terms of sensor responses towards various environmental and flammable gases including C₂H₂, C₂H₅OH, H₂, CH₄, C₂H₄, H₂S, NH₃, CO₂, C₂H₄O, NO, NO₂ and N₂O at their typical concentrations and a working temperature of 300 °C as shown in Fig. 8. It is clear that the SnO₂ sensor (S-0) shows comparatively high response towards C₂H₅OH, modest responses to C₂H₂ and NO₂, and low responses to H₂, NH₃, CO₂, NO, H₂S, CH₄, C₂H₄O, C₂H₄ and N₂O, presenting good C₂H₅OH selectivity against these gases. With Fe-doping at the lowest content of 0.1 wt%, the C₂H₂ response increases prominently while those to C₂H₅OH, and NO₂ marginally decrease and other gas responses remain relatively very low. Thus, the small Fe-doping level causes the sensor to become selective to C₂H₂. However, the attained C₂H₂ selectivity is only guaranteed under the condition that the concentrations of C₂H₅OH and NO₂ polluting gases are usually in the ranges of lower than 1000 ppm and 10 ppm, respectively (Gouw et al., 2012; Bikov et al., 2013; Cho et al., 2006). Thus, the sensors will be applicable for C₂H₂ leakage detection under normal ambient conditions with typical polluting concentrations of C₂H₅OH and NO₂. As the Fe content additionally increases to 1 wt%, the C₂H₂ response decreases steadily but is still larger than those of other gases, implying reduced C₂H₂ selectivity. However, the NO₂ response decreases more significantly to a very low value at 1 wt% Fe, indicating improved C₂H₂ selectivity against NO₂ and the possible use of 1 wt% Fe-doped SnO₂ sensor for C₂H₂ detection when NO₂ is a potential interfering gas. With the highest Fe-doping level of 2 wt%, the responses to all gases degrade considerably to similar values and the sensor displays very low selectivity among these gases. From the data, the optimal Fe doping level of 0.1 wt% provides substantial improvement of C₂H₂-sensing characteristics at a moderate working temperature of 300 °C. Thus, only full results of C₂H₂ will be further reported.

Close

Fig. 8

The histogram of various gas responses including 1 vol% H₂, 2000 ppm NH₃, 3 vol% CO₂, 25 ppm NO, 5 ppm NO₂, 3 vol% C₂H₂, 1000 ppm C₂H₅OH, 10 ppm H₂S, 1 vol% CH₄, 30 ppm C₂H₄O, 1000 ppm C₂H₄ and 100 ppm N₂O at 300 °C of Fe-doped SnO₂ sensing films with various Fe-doping levels.

Export to PPT

C₂H₂ responses of all sensors (S-0 to S-2Fe) evaluated at different working temperatures from 200 to 350 °C towards 3 vol% C₂H₂ are displayed in Fig. 9. It is evident that all C₂H₂ responses monotonically rise with increasing temperature from 200 to 300 °C, before slightly reducing as the working temperature additionally increases to 350 °C. In addition, the response increases rapidly as the working temperature rises from 250 to 300 °C. Thus, the undoped and Fe-doped SnO₂ sensors (S-0 to S-2Fe) have the highest responses at the optimal working temperature of 300 °C. Concerning the effect of Fe content, the sensor response increases greatly with a tiny Fe-doping level (0.1 wt%) while the higher Fe content leads to a substantial decline of C₂H₂ response. Principally, the S-0.1Fe sensor displays a high C₂H₂ response of ∼750 to 3 vol% at 300 °C. The roles of Fe dopants and the influence of working temperature will be explained further in the mechanism section.

Close

Fig. 9

The histogram of sensor response of 0–2 wt% Fe-doped SnO₂ sensing devices (S-0 to S-2Fe) to 3 vol% C₂H₂ as a function of operating temperature in the range of 200–350 °C.

Export to PPT

Fig. 10(a) displays the change in resistance of S-0 to S-2Fe sensors subjected to various C₂H₂ concentrations in backward (3–0.15 vol%) and forward (0.15–3 vol%) cycles at 300 °C. The results demonstrate that the resistance in air increases monotonically by more than a few orders of magnitude with increasing Fe content from 0 to 2 wt% and are repetitively stable in both cycles. Interestingly, the 0.1 wt% Fe-doped SnO₂ sensor exhibits notably rapid resistance decrease upon C₂H₂ exposure, confirming a common n-type gas-sensing property to a reducing gas. Furthermore, the change in resistance extends markedly with Fe-doping at the lowest content of 0.1 wt% but then deteriorates at higher Fe concentrations.

Close

Fig. 10

(a) Change in resistance of 0–2 wt% Fe-doped SnO₂ sensors (S-0 to S-2Fe) subjected to 0.15–3 vol% C₂H₂ at 300 °C and (b) the relevant sensor response and response time vs. C₂H₂ concentration.

Export to PPT

The related sensor responses and responses time of all sensors (S-0 to S-2Fe) as functions of C₂H₂ concentration at 300 °C are reported in Fig. 10(b). It is clear that the C₂H₂ response critically increases owing to the addition of Fe at the small doping concentration of 0.1 wt% but then continually declines as the Fe concentration rises from 0.1 to 2 wt%. Particularly, the optimal Fe-doped sensor (S-0.1Fe) offers excellent responses of 741.0 (backward) and 748.7 (forward) with a small response time (∼2.5 s) to 3 vol% C₂H₂ in comparison with S-0.2Fe (S = 376–383, t_res = ∼3.1 s), S-0.5Fe (S = 277–282, t_res = ∼3.5 s), S-1Fe (S = 154–158, t_res = ∼5.3 s), S-0 (S = 125–143, t_res = ∼61 s) and S-2Fe (S = 26–32, t_res = ∼113 s), respectively. Regarding the gas response function, the responses rise steadily with increasing C₂H₂ concentration following the typical power-law relationships as reported along with the labels. It can be observed that the calculated exponent values in backward and forward cycles of all sensors are similarly around one, signifying that the amount of Fe dopants insignificantly affect the type of preadsorbed surface oxygen species (Yamazoe and Shimanoe, 2008). Moreover, the optimal Fe-doped sensor also displays good responses of 169–171 and 20–22 at lower concentrations of 1 and 0.15 vol%, respectively. Correspondingly, the limit of acetylene detection is estimated to be as low as ∼25 ppm according the power equation at the threshold response of 1.1. Thus, the Fe-doped flame-made SnO₂ sensor is a potential candidate for leakage detection of acetylene, which is a widely used fuel for various industrial applications in cutting, welding, straightening and other confined heating processes (Lin et al., 2015). Furthermore, the achieved response and response time values are significantly better than other advanced C₂H₂ sensors as previously displayed in Table 1, which offer sensor response of <70 and response time of >3 s to C₂H₂ at various concentrations (<1 vol%). The superior C₂H₂ sensing performances may be credited to relatively large effective surface area and well distributed Fe dopants in the FSP-made Fe-doped SnO₂ nanoparticles.

3.3 Gas-sensing mechanisms

From the observations, the optimum Fe-doping level of 0.1 wt% delivers significantly enhanced gas-sensing properties with high, fast and selective C₂H₂ response. However, the crystallite and particle sizes only slightly change upon Fe doping according to the structural characterization results. Hence, the slight change of morphology upon Fe doping should only marginally contribute to the response enhancement. Consequently, the catalytic and electronic effects of substitutional Fe dopants should play much more important roles to the gas-sensing enhancement, which can be described with the representations as portrayed in Fig. 11. With no Fe doping (Fig. 11(a)), the SnO₂ nanoparticles display nominal n-type behaviors with a moderate conductivity from n-type structural defects, which are mainly oxygen vacancies. SnO₂ has a relatively low electronic resistivity compared with many metal oxides owing to its relatively high amount of surface oxygen vacancies. On surface, chemisorbed oxygen species ( ${\text{O}}_2^{-}$ , ${\text{O}}^{-}$ and ${\text{O}}^{2-}$ ) take electrons from SnO₂ conduction band and induce surface electron depletion regions as displayed in Fig. 11(a).

Close

Fig. 11

Representations of C₂H₂ sensing process of SnO₂ nanoparticles with (a) no Fe doping, (b) Fe doping at an optimal concentration (0.1 wt%) with inset: the detailed atomic-level reaction mechanisms at a Fe³⁺ site and (c) Fe doping at a high concentration (>0.1 wt%).

Export to PPT

With Fe doping (Fig. 11(b)−(c)), Sn⁴⁺ ions in the lattice should be substituted by Fe³⁺ ions, resulting in the creation of holes via the defect reactions (Inyawilert et al., 2017):

To initiate the reactions, C₂H₂ molecules must adsorb and then be oxidized by O⁻ or O²⁻, emancipating H₂O and CO. In addition, some CO may readsorb on surface and then oxidized again into CO₂. These processes reduce the steady-state surface coverage of oxygen species while releasing electrons into the SnO₂conduction band, leading to a lower electrical resistance. Hence, the C₂H₂ response will principally relate to the C₂H₂ dissociation rate and the amount of surface oxygen species. It can be expected that Fe doping can affect the C₂H₂ dissociation rates on SnO₂ since Fe and FeO_x are well-known catalysts for C₂H₂ dissociation widely used for the growth of nanocarbon materials (Khedr et al., 2008). In addition, Fe/SnO₂ cocatalysts have been used to synthesize carbon nanocoils from C₂H₂ source (Muneaki and Khalid, 2014). It is thus plausible that Fe-substituted SnO₂ nanoparticles can act as co-catalysts that will directly enhance C₂H₂ interaction as well as chemisorption of oxygen species as suggested by the XPS analysis discussed earlier.

The catalytic properties of Fe/SnO₂ cocatalysts for C₂H₂ detection can be quantitatively verified by considering the response rates versus Fe concentration using the Langmuir isotherm model (Boontum et al., 2018). According to the model, the adsorbed surface coverage of gas species will rise approximately linearly with increasing gas concentration at low gas concentrations or low partial gas pressures. In addition, the coverage is proportionate to the response rate at which oxygen species interact with gas molecules. The response rate corresponds to the rate of change in resistance and is the reciprocal of the response time (t_res⁻¹) (Singkammo et al., 2018). Fig. 12 displays the plots of t_res⁻¹ versus C₂H₂ concentration of Fe-doped SnO₂ sensors with various Fe contents at 300 °C in forward and backward cycles. The results demonstrate that all sensor response rates for both cycles increase approximately linearly with increasing C₂H₂ concentration, following the Langmuir isotherm model. The slope of each plot corresponds to the effective reaction rate constant $(k)$ , which is plotted versus Fe concentration as presented in the inset of Fig. 12. It is apparent that the rate constant increases significantly to a maximum value at the optimal Fe content of 0.1 wt% before dropping rapidly down almost to the starting value as the Fe content increases to 2 wt%. The complex relationship of the rate constant with the Fe content signifies compound catalytic mechanisms of Fe-doped SnO₂ cocatalyst, which accelerates the reducing reaction with C₂H₂ via multi-step electron transfers (Natkaeo et al., 2018).

Close

Fig. 12

Dependence of the sensor response rate on C₂H₂ concentration. Inset the effective reaction rate constant (k) versus the Fe concentration at 300 °C.

Export to PPT

The exact catalytic roles of Fe dopants on gas-sensing mechanisms of SnO₂ materials remain not clearly understood and presently under exploration. In this research, a new C₂H₂ dissociation mechanism on Fe-doped SnO₂ surface is proposed with the atomic-level reaction models as sketched in the inset of Fig. 11(b). The first step illustrates the Fe-doped SnO₂ surface with preadsorbed oxygen species on Fe³⁺ and Sn⁴⁺ in air. Upon injecting acetylene, the C₂H₂ molecule may preferentially adsorb at Fe³⁺ sites of Fe-doped SnO₂ surface possibly by exchanging with the oxygen ion (O⁻) and forming C₂ species together with a water molecule while an electron is injected to reduce Fe³⁺ to Fe²⁺. Next, the oxidation of a C₂ molecule necessarily occurs via the interactions between two adjacent O⁻ species, resulting in the break of carbon triple bonds and the formation of pairs of CO molecules that may either directly desorb or readsorb to other sites and be oxidized again into CO₂ (omitted in the inset diagram). In the process, the oxidation of C₂ is expected to be the rate-limiting stage since the oxidation of C₂ into CO is a highly exothermic process (Ard et al., 2013). At this step, electrons are released back to Sn lattice to oxidize C₂ into CO molecules. The Fe³⁺ state will only be restored after readsorption of oxygen species. The low C₂H₂ response of undoped SnO₂ is likely due to a dissimilar role played by Sn compared with Fe on the formation of carbide. Iron carbide (FeC_x) can be much more easily formed compared with tin carbide (SnC_x) due to much lower required energy and temperature (H, 1896; Janbroers et al., 2011). Therefore, Fe-doping significantly improves the C₂H₂ response of SnO₂ nanoparticles by facilitating the carbon binding and succeeding oxidation of carbon species.

Moreover, the catalytic reactivity of Fe-doped SnO₂ nanostructures is rather particular to C₂H₂ molecules, resulting in high C₂H₂ selectivity against various oxidizing and reducing gases except ethanol (C₂H₅OH) due to their similar chemical structure and NO₂ due to its very high reactivity with SnO₂. With increasing Fe content from 0 to 0.1 wt%, the high activity of Fe species provides enhanced C₂H₂ response but other gas responses are not much changed owing to their little reactivity with the nanoparticles. The attained selectivity may be associated with an optimal Fe-substituted Sn structure providing most suitable adsorption sites for C₂H₂. The activity of Fe/SnO₂ cocatalyst should depend significantly on the density of Fe sites in SnO₂ since the adsorption energy will be substantially affected by atoms surrounding the adsorption sites (Ard et al., 2013). At higher Fe-doping levels, the catalytic reactivity to C₂H₂ of Fe-doped SnO₂ nanostructures decreases rapidly due possibly to the elevating level of structural disorders particularly on surface whereas other gas responses also decrease but with lower rates because their reactivity were already low, leading to relatively low C₂H₂ selectivity. The disorders behave as surface states that capture charge carries (Zhang and Yin, 2013), reducing charge transfer from reducing or oxidizing reaction to the SnO₂ conduction band, lowering the change of surface conductivity and diminishing gas response as depicted in Fig. 11(c). Consequently, the response is optimal at a low Fe content of 0.1 wt% since active Fe³⁺ species are formed with a sufficiently low content of surface and bulk defects. At a higher Fe content, the sensor response declines due to increasing density of surface states. Nevertheless, additional experimental and theoretical studies of C₂H₂ interaction with Fe-doped SnO₂ nanostructures will be required to validate the offered mechanisms and conclude the true role of Fe dopant on the C₂H₂ sensing process.

Concerning the formation of oxygen species, the oxygen chemisorption rate depends mostly on temperature and the physiochemical properties of base metal oxides, which may be modified by additives. Hence, the characteristics of oxygen desorption and adsorption rates of a sensing material at different temperatures dictate the temperature dependency of gas-sensing properties. For pristine SnO₂, ${\text{O}}_2^{-}$ ions are prevailing below 160 °C and will start transforming to O⁻/O²⁻ species above this temperature whereas the desorption of ${\text{O}}_2^{-}$ and O⁻ or O² will follow while reducing the temperature under 150 and 560 °C, respectively (Yamazoe et al., 1979). Thus, undoped SnO₂ sensors usually show an optimal adsorption rates of O⁻ and O² at 300–350 °C. Consequently, adsorbed C₂H₂ molecules can increasingly react with oxygen species as the temperature increases up to 300 °C but the desorption of C₂H₂ as well as oxygen species will become significant at higher temperatures, resulting in the decline of C₂H₂ response. With Fe doping, Fe atoms as p-type dopants for SnO₂ can enhance the density of oxygen vacancies, facilitating the dissociation of oxygen gas molecules into oxygen atoms (Zhao et al., 2017), which can chemisorb easily on the surfaces at a raised temperature, resulting in a supplementary increase of response. However, the small amount of Fe dopants does not change the temperature dependency of adsorption/desorption behaviors of SnO₂ surface. Thus, the Fe-doped SnO₂ sensors still exhibit the same optimal working temperature of 300 °C.