- Email: [email protected]
High butane sensitivity and selectivity exhibited by combustion synthesized SnO2 nanoparticles
Materials Research Bulletin 65 (2015) 216–223
Contents lists available at ScienceDirect
Materials Research Bulletin journal homepage: www.elsevier.com/locate/matresbu
High butane sensitivity and selectivity exhibited by combustion synthesized SnO2 nanoparticles Suparna Banerjee 1, Pratanu Nag 1, Sanhita Majumdar, P. Sujatha Devi * Nano-Structured Materials Division, CSIR-Central Glass and Ceramic Research Institute, 196, Raja S. C. Mullick Road, Jadavpur, Kolkata 700032, India
A R T I C L E I N F O
A B S T R A C T
Article history: Received 8 January 2014 Received in revised form 5 January 2015 Accepted 7 January 2015 Available online 8 January 2015
We report here the high butane sensitivity and selectivity exhibited by nanoparticles of SnO2 prepared by a citrate–nitrate gel combustion process. Using the synthesized powders, the sensitivity towards 500, 1000 and 5000 ppm n-butane has been measured at various operating temperatures from 350 C to 500 C. The powder calcined at 600 C exhibited a sensitivity of 91 0.5% towards 1000 ppm n-butane with a recovery time of 15 s at an operating temperature of 350 C. We have explored the effect of size on both butane gas sensitivity and selectivity and found that the reduction of particle size of the order of 10 nm could result in a remarkable increase in the response of the sensor towards butane compared to other reducing gases like H2 and CH4. The most important observation that emerged from our studies is a combination of sensitivity and selectivity that has been achieved using SnO2 nanoparticles without the addition of any noble metal catalyst such as Pd or Au. ã 2015 Elsevier Ltd. All rights reserved.
Keywords: A. Oxides B. Chemical synthesis D. Surface properties
1. Introduction Tin dioxide (SnO2) is one of the most widely used semiconducting oxides for gas sensor applications due to its high sensitivity, reliability, durability and above all the lower cost [1]. Thick or thin film SnO2-based gas sensors have been extensively studied to detect a variety of toxic and explosive gases such as CO, hydrocarbons, ammonia, H2, nitrogen oxides and H2S [1–26]. The sensitivity of a thick film gas sensor depends mainly on the particle size, shape and their distribution, specific surface area and porosity of the film. Needless to say, all these properties are directly related to the method of preparation of the samples. Thus, preparation techniques that can produce controlled size and shaped particles are of paramount importance in gas sensor development. As a result, many attempts have been made to study the effect of reducing the particle size to nanometric dimension and exploring its consequent effect on the chemical sensing properties [27–32]. When the particle size (D) gets reduced to nanometric dimensions, a dramatic improvement in the gas sensing properties could be expected since a large fraction of the atoms are present at the surface or the interface regions with structure and properties different from those of the bulk. The
* Corresponding author. Tel.: +91 33 2483 8082; fax: +91 33 2473 0957. E-mail addresses: [email protected], [email protected] (P. S. Devi). 1 These two authors contributed equally. http://dx.doi.org/10.1016/j.materresbull.2015.01.025 0025-5408/ ã 2015 Elsevier Ltd. All rights reserved.
prominent effect of nano size, however, is associated with the thickness of the electron-depleted surface layer, which is defined as the Debye length ‘LD’. The Debye length ‘LD’ for a semiconducting material could be calculated as [33], p eO ekB T (1) LD ¼ e2 nd where kB is Boltzmann’s constant, e the dielectric constant, eo the permittivity of free space, T the operating temperature, e the electron charge and nd the carrier concentration. For example, LD for SnO2 has been estimated to be 3 nm. If the particle size is reduced to a size that is comparable to or lower than 2 LD, the whole crystallite is depleted of electrons and this in turn causes the gas sensitivity of the element to change dramatically with D. Since LD for SnO2 is 3 nm, the critical size at which SnO2 could exhibit the “size related nano- effect” and its consequent impact on gas sensitivity should be around 6 nm or lower. Therefore, in order to improve the gas sensitivity of SnO2 based materials it is desired to work with nanoparticles having size range within 6 nm. Although there are many reports on the synthesis of nanoparticles of SnO2 by sonochemical technique, sol–gel processing, co-precipitation technique, hydrothermal route, and gel-combustion route, the studies dealing with the butane gas sensitivity of such materials are rather scarce as evident from the cited references [2–31]. More importantly, to our knowledge most of the reported results are on the sensitivity leaving behind the importance of selectivity untouched and unexplored in most cases [27–33].
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
217
In this paper, we report the synthesis of SnO2 nanoparticles by a citrate–nitrate process (CNP) and their detailed gas sensing characteristics towards n-butane, an important component of the liquefied petroleum gas (LPG). This process could produce homogeneous high surface area powder with superior properties as demonstrated previously by other investigators [34–42].Though there has been a very few attempts to prepare SnO2 particles by combustion process [43–47], none has reported the potential of this economically viable process in preparing highly sensitive SnO2 particles for butane detection. Our interest mainly lies in exploring the sensitivity and selectivity of nanoparticles of SnO2 prepared by this economically viable process and to understand the effect of particle size on the performance of the sensor made thereof. Interestingly, in the present case, the CNP prepared SnO2 nanoparticles exhibited a high sensitivity and selectivity with a fast recovery time towards n-butane compared to SnO2 based systems prepared by us by other techniques [48,49].
The consistency of the paste and the processing variables were optimized to get final coatings of 50–60 mm thickness. Using the paste a thin coating was made on the outer surface of the alumina tubes of 3 mm length, 2 mm outer diameter and of 0.5 mm thickness attached with gold electrodes and platinum lead wires. After coating, the sensors were cured at 600 C for 2 h. so that the sensor surfaces have particles of size comparable to that of the characterized nanoparticles. The electrical resistance of the sensors in air and in n-butane was measured at different temperatures (up to 500 C) in an ambient of 60–70% relative humidity using a digital multimeter (Solartron) and a constant voltage/current source (Keithley 228 A). All the samples were initially aged at 350 C for 72 h to achieve the desired stability of the resistance value before the measurements. The percentage response or sensitivity) (%S) of the films was calculated as: DR ðRA RG Þ ð%ÞS ¼ ¼ 100 (2) RA RA
2. Experimental
RA and RG are the sensor resistances in air and measuring gas, respectively. DR for reducing gases is (RARG) where RA>RG. The response time has been measured as the time taken for a sensor to read 90% of full-scale reading after being exposed to a given gas and recovery time was the time taken for a sensor to come back to its original state when the target gas is removed.
2.1. Chemicals Tin chloride dihydrate, SnCl22H2O ([MERCK India, GR], citric acid monohydrate, ammonium nitrate, distilled water and absolute ethanol were used. All reagents used were analytically pure.
3. Results and discussion 2.2. Preparation of SnO2 nanoparticles 3.1. Thermal characterization To prepare nano particles of tin dioxide, a mixture of an aqueous solution of SnCl22H2O (0.2 M) and ammonium nitrate solutions were taken in a beaker and calculated amount of citric acid monohydrate was added to this mixed solution keeping the citrate–nitrate molar ratio as 0.3. The rest of the procedure is similar to what has been reported earlier [39–42]. The precursor ash powder obtained was calcined at 600 C for 6 h in air to remove unburnt carbonaceous material and to form a pure oxide phase. From the concept of propellant chemistry, the net oxidizing valency of metal nitrates to the net reducing valency of fuel should be unity to get the maximum exothermicity for driving the combustion reaction. In our experiments, we have optimized the citrate to nitrate ratio as 0.3 for achieving a satisfactory combustion reaction.
In order to understand the thermal decomposition behavior of the samples, simultaneous TGA-DTA measurements have been carried out on the gel and the as-prepared SnO2 powder (Fig. 1(a) & (b)). The gel sample exhibited a weight loss of 88% whereas the as-prepared powder exhibited a weight loss of only 25%. The initial weight loss of 3.26% exhibited by the gel sample between 30 C and 110 C is well supported by an endothermic peak on the DTA at 119 C. These changes are probably due to the dehydration of surface adsorbed water. From 110 C to 179 C, a weight loss of 11.24% along with a corresponding exothermic shoulder at 174 C was observed. From 179 C to 244 C, a total weight loss of 14.84% was evident. The above changes were accompanied by an
2.3. Characterization
[(Fig._1)TD$IG]
Thermogravimetric analysis (TGA) of the as-prepared gel and uncalcined powder was carried out on a Netzsch (Germany) simultaneous DTA-TGA analyser (Model No. STA 449 C) from room temperature to 1000 C at a heating rate of 10 C/min. The phase identification of the calcined powder was carried out by X-ray diffraction analysis (Philips PW1710) using Cu Ka radiation (l = 1.5406 Å). The surface area analysis of the calcined powder was measured by the Brunauer–Emmet–Teller (BET) method on Micromeritics Gemini II 2370 equipment. The particle morphology and local crystallographic structure were studied by transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM), respectively. The samples for microscopy were prepared by depositing micro-drops of the acetone-dispersed sample on a carbon coated copper grid followed by drying. The average particle size of the calcined powders was calculated using the bight field (BF) image obtained from TEM. 2.4. Gas sensing measurement For gas sensing study, a thick paste of the powder was prepared using an alumina gel as a binder. The details of the sensor fabrication and processing have been reported elsewhere [48,49].
Fig. 1. TG-DTA curves of the (a) gel, (b) as-synthesized SnO2 powder obtained from citrate–nitrate process.
218
[(Fig._2)TD$IG]
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
compounds present in the gel. From 329 C to 554 C a final weight loss of 4.01% corresponding to an exothermic peak at 504 C was also evident. In case of the as-prepared powder, from 74 C to 558 C, a single step weight loss of 24.67% accompanied by an exothermic peak centered at 504 C was observed. This has been assigned to the formation of SnO2 by the decomposition of citrate–nitrate gel derived precursor powder followed by recrystallization. Based on the above observations, a minimum calcination temperature of 600 C was fixed to prepare phase-pure SnO2 particles free from any impurity phases. 3.2. Structure and morphology
Fig. 2. XRD patterns of SnO2 powders (a) Standard (b) CNP calcined at 600 C and (c) CNP calcined at 800 C.
endotherm at 199 C and an exotherm at 234 C, respectively. From 244 C to 329 C, a further weight loss of 54.86% took place corresponding to an endothermic peak centered at 319 C. The above changes are related to probable dehydration through condensation and the heat released by decomposition of organic
The X-ray diffraction patterns of 600 C (hereafter designated as CNP1) and 800 C (hereafter designated as CNP2) calcined powders along with the XRD pattern of a commercial SnO2 (Merck grade, hereafter designated as standard) powder are shown in Fig. 2. The CNP powders exhibited all the reflections of the commercially available SnO2 powder having rutile type structure. The comparison of the plots in Fig. 2 obviously shows the difference in the widths of the diffraction peaks arising from the variation in the crystallite as well as particle size. The broad peaks observed in the XRD patterns for the nanoparticles indicate line-broadening arising from the finite number of diffracting planes and the extent of broadening was calculated using the Scherer formula. In Fig. 3a, the TEM, and Fig. 3b and c the HRTEM pictures of the CNP1 powder are presented and in Fig. 4a and b the TEM and HRTEM pictures of CNP2 and in Fig. 4c and d, the TEM and HRTEM
[(Fig._3)TD$IG]
Fig. 3. (a) TEM, (b) and (c) HRTEM (with corresponding FFT figures at the inset) and (d) SAED pattern of CNP1 powder.
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
[(Fig._4)TD$IG]
219
Fig. 4. (a) and (c) bright field TEM and (b) and (d) HRTEM (with corresponding FFT figures at the inset) of CNP2 and SnO2 standard powders, respectively.
pictures of SnO2-standard are presented. Fig. 3a indicates the presence of well-distributed very fine particles of pure SnO2 having average particle size of the order of 4 nm as evident from the size distribution shown in Fig. 5a. The average particle sizes were found to be 4.01 and 14.62 nm for CNP1 and CNP2 powders, respectively, calculated from the histogram of particle size distribution. In Fig. 3b and c, two high-resolution TEM images obtained from spherical and elliptical SnO2 particles are shown (with corresponding FFT figures at the inset) in order to further confirm the single nanocrystalline nature of these isolated particles. In Fig. 3c two overlapping (11 0) planes can be seen in an elliptical nanoparticles. The selected area electron diffraction (SAED)
pattern taken from a representative particle shown in Fig. 3d indicates lattice fringes corresponding to (11 0), (1 0 1), (2 0 0), (2 1 0), (2 11), (2 2 0) and (3 0 1) planes of tetragonal SnO2. From Fig. 4a, it is very clear that the CNP2 particles are larger than CNP1 with an average size of 14.62 nm as shown in the size distribution in Fig. 5b. SnO2 (standard) on the other hand are much higher in size and highly agglomerated with average particle size of around 99.87 nm (Fig. 5c). The corresponding HRTEM images are shown in Fig. 4b and 4d.The TEM pictures of CNP1 (Fig. 3a), CNP2 (Fig. 4a) and standard (Fig. 4c) unequivocally confirm the difference in size of the individual particles. The data shown in Table 1, further indicates the superior properties of CNP1 powder
[(Fig._5)TD$IG]
Fig. 5. Particle size distribution with Gaussian Fit derived from the bright field TEM image of the (a) CNP1, (b) CNP2 and (c) Standard powder samples with corresponding histogram.
220
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
Table 1 Physico–chemical characteristics of standard SnO2 and pure SnO2 powders prepared by CNP. SnO2ID
TC a ( C) [Å]
c [Å]
Lattice SBET volume (m2/g) [Å]3
(Standard) – 4.72 0.01 3.14 0.01 70 (CNP2) 800 4.74 0.01 3.18 0.01 71 (CNP1) 600 4.79 0.02 3.16 0.07 72
– 40 58
Average particle size (DTEM) (nm) 99.87 14.62 4.01
TC = calcination temperature; DTEM = Particle size from TEM.
[(Fig._6)TD$IG]
samples. The FTIR spectrum of standard powder having a larger particle size is broad with the presence of two weakly defined bands at around 520 cm1 and 615 cm1 respectively. The principle bands at 615–623 cm1 and 657–667 cm1 in FTIR spectra of SnO2 are the fundamental vibration frequencies of ySn–O and yO–Sn–O of tin (IV) dioxide, respectively, while the bands at around 520–530 cm1 can be assigned to the fundamental vibrations (ySn–O,T; T = terminal) of tin (IV) dioxide [51]. The discrepancies in the positions of IR spectra arise due to factors such as stoichiometry of the oxide (i.e., the presence of defects) the size and the state of aggregation of the particles [51–53]. The remarkable blue shifting of the FTIR peaks of CNP2 and CNP1 compared to standard samples and the disappearance of the splitting nature of the spectral bands in the latter sample could be due to the presence of more oxygen defects in the former sample. The presence of oxygen vacancy in CNP1 and CNP2 is expected to give rise to greater reduced mass effect resulting in comparatively shorter and stronger Sn—O bonds of O—Sn—VÄ in CNP1 and CNP2 ~ = 1/2p (k/m) where y is the wave number, k, the force samples. As y constant and m the reduced mass effect, the higher bond strength decreases m and increases the k values for the O–Sn–VÄ bonds in ~ Sn–O (CNP1) > y ~ Sn–O (CNP2) >> y ~ Sn–O (M). CNP1 and CNP2. i.e., y Moreover, the sharper and the split nature of the FTIR spectral bands reveal a higher structural disorder of the CNP powders than the standard powder. 3.4. Gas-sensing properties
Fig. 6. FTIR spectra of the standard, CNP1 and CNP2 SnO2 powders.
such as high surface area (58 m2/gm) and low particle size (around 4 nm) in comparison to CNP2 and standard powders. In Table 1, the particle size from TEM and lattice parameters obtained from the powder XRD data are shown. The lattice parameter shows a decrease in the ‘a’ value with increase in the calcination temperature. 3.3. FTIR studies on the calcined powders In Fig. 6, FTIR spectra of the standard, CNP1 and CNP2 SnO2 powders are presented. The FTIR spectrum of CNP1 powder exhibited three well defined bands at 667 cm1 and 623 cm1 corresponding to the two IR active modes namely (Eu) and a shoulder around 530 cm1, attributed to an IR active mode (A2u) of tin (IV) oxide, respectively [44,49,50]. These three bands shift to lower wave numbers of around 657 cm1, 619 cm1 and 526 cm1, respectively with less splitting for the larger sized CNP2 sample. The FTIR spectra in Fig. 6 indicate a blue shift of the IR bands with gradual decrease in particle size from Standard to CNP2 to CNP1
In order to understand the gas sensing characteristics, the fabricated sensors were exposed to 500,1000 and 5000 ppm butane, respectively at various temperatures. The observed response as a function of the operating temperature is plotted in Fig. 7a–c. For 500 ppm n-butane, at 500 C, a maximum percentage response of around 89% and 77% were obtained for the CNP1 and CNP2 powders, respectively, in contrast to the commercial SnO2 that exhibited only 40% response. At an operating temperature of 350 C, the percentage response of CNP1 increased from 60 to 91 to 93% and that of CNP2 increased from 48 to 83 to 89% on increasing the gas concentrations from 500 to1000 to 5000 ppm, respectively. It is clear from Fig. 7a–c that the gas response increases with operating temperature, reaches a maximum at 500 C in case of 500 ppm and at 450 C in cases of 1000 and 5000–ppms, respectively, and then falls with further increase in operating temperature due to desorption of the surface adsorbed metastable oxygen species from the surface of SnO2 coating. Fig. 8 shows the typical dynamic electrical response of the CNP1 sensor for eight consecutive cycles. The recovery time was around 15 s (for about 90% recovery) and a sensitivity of around 91.5% towards 1000 ppm butane gas was observed at 350 C. The fast recovery time exhibited by these sensors could be attributed to
[(Fig._7)TD$IG]
Fig. 7. Percent sensitivity (response) recorded for SnO2 thick film sensor at different operating temperatures and at different butane concentrations, (a) 500 ppm (b) 1000 ppm and (c) 5000 ppm, respectively.
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
[(Fig._8)TD$IG]
221
Fig. 8. Dynamic response curve of the thick film CNP1 sensor upon exposure to 1000 ppm butane at an operating temperature of 350 C for eight consecutive cycles.
[(Fig._9)TD$IG]
Fig. 9. The response time of the CNP1 sensor towards 1000 ppm n-butane measured at 350 C.
the superior characteristics of the combustion synthesized powder. The observed response of the CNP1 sensor towards 1000 ppm butane shown in Fig. 9 confirms that a minimum exposure time of 5 s was sufficient for the sensor to respond to the gas, albeit with a lower percent response. On exposure of 12 s, a percent sensitivity as high as 95% was observed. It is clear from Fig. 9 that the gas response increases with response time (Rs), reaches a maximum (Rs = 12 s) and falls above 12 s. To further understand and establish the novelty of the SnO2 nanoparticle based CNP1 sensor, its cross-sensitivity has also been examined in presence of other reducing gases at 350 C. In Fig. 10, the sensitivities exhibited by the CNP1 sensor towards 1000 ppm’s each of butane, methane and H2 at 350 C, respectively, are compared in order to determine its selectivity to any particular gas. It is interesting to see the differential sensitivity of the sensor (>91% towards 1000 ppm n-butane at 350 C) with respect to other gases like H2 (57%) and methane (20%) which confirms that if we have a mixture of 1000 ppm’s each of butane, methane and H2 gases, the fabricated sensor could effectively and efficiently
[(Fig._10)TD$IG]
detect and sense only butane. The selectivity coefficient, b (b = Sbutane/Sgas) varied in the order CH4 > H2 > C4H10 with 4.57 and 1.58 for methane and H2, respectively. This selective detection of butane by the fabricated sensor when there is a mixture of the above mentioned gases is very interesting since it is very difficult to fabricate SnO2 based sensors that are selective to a particular gas. In order to find the underlying reason for the high sensitivity and the extremely high butane selectivity exhibited by the CNP1 nanoparticles we have checked the cross sensitivity of all the samples which clearly illustrates the direct influence of the particle size on selectivity. The response towards n-butane decreased in the order CNP1 > CNP2 > standard with the CNP1 sample exhibiting a high butane sensitivity and selectively compared to the other two samples. In Fig. 11, the response of all the sensors at 350 C towards 5000 ppm n-butane and methane are compared. The (%) response of CNP1 and CNP2 sensors towards 5000 ppm n-butane is much higher and comparable. The response of both the samples towards same concentration of methane gas, on the other hand was poor with CNP2 exhibiting a slightly higher response than CNP1. Interestingly, the response of standard SnO2 coated sensor towards both butane and methane were comparable. We have also monitored the stability of the prepared sensors by checking the long-term drift of the resistance in air (Rair) and the percent response towards 500 ppm gas at 500 C. The measurements were performed for two successive months after initial checking with ten days interval of time. The percent response of the best CNP1 sensor increased slightly in the first month and then became almost stable. However, the resistance remained almost constant. The gas sensing mechanism [43,46] of a semiconductor depends on the adsorption and subsequent desorption reactions occurring at the surface of the semiconducting oxide, which are highly temperature dependent. Normally, the atmospheric oxygen chemisorbs on the SnO2 surface, consuming free electrons from the conduction band resulting in an electron depleted surface layer and a consequent rise in resistance of SnO2 (Eqs. (3)–(5)).
[(Fig._1)TD$IG]
Fig. 10. Dynamic response curves of the thick film CNP1 sensor upon exposure of 1000 ppm of butane, hydrogen and methane gases at an operating temperature of 350 C.
Fig. 11. Selectivity study of CNP1, CNP2 and standard powder coated sensors towards 5000 ppm n-butane and 5000 ppm methane.
222
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
O2 (gas) @ O2 (physisorbed)
(3)
O2 (physisorbed) @ e O2 (chemisorbed)
(4)
O2 +e @ 2 O (chemisorbed)
(5)
At low temperature, oxygen adsorbs on the SnO2 surface nondissociatively in the molecular form as charged O2. At high temperatures, it dissociates to 2O or O2 [33]. The reaction of the butane gas with these chemisorbed oxygen species releases the bound electrons back to the conduction band as free electrons, which decreases the resistance of the SnO2 coatings. C4H10 + 2 O (chemisorbed) ! C4H8O + H2O + 2e
(6)
The high butane sensitivity and selectivity exhibited by the SnO2 CNP1 nanoparticle is almost similar to the results reported by Xu et al. [32] reporting a strong relation between the grain diameter (D) and sensitivity. As demonstrated, when the particle size gets closer to or smaller than 2 LD, the gas sensitivity could in principle, be controlled by the grains themselves and not the gain boundaries or interfaces [32]. This in turn gives rise to a substantial increase in the gas sensitivity since in this case the entire grains will become depleted of electrons. The 600 C calcined SnO2 nanoparticles prepared by the citrate–nitrate process (CNP1) have an average particle size of 4 nm, which is lower or close to the reported critical size of pure SnO2 to exhibit the size related effect. This clearly demonstrates that the CNP1 particles with a size of the order of 4 nm shows the maximum sensitivity compared to larger sized particles, thus proving the size effect as the major cause for the high butane sensitivity. A correlation between the particle size and selectivity was reported by Chen et. al., [22,23]. The authors claim high ethanol sensitivity from 3 nm sized SnO2 nano rods, as a consequence of the size reduction down to LD [23]. A report by Panchapakesan et al. also discussed about the sensitivity and selectivity of tin oxide nanostructures on large arrays of micro hotplates [50]. Thus, it is evident from our results that butane can be detected selectively using CNP derived nano material based sensors with a high sensitivity (>91% towards 1000 ppm butane). Now the question arises as to why the CNP1 based sensors having a finer particle size and higher surface area showed high sensitivity and selectivity towards n-butane only? It is evident from Fig. 10 that the sensitivity of the CNP sensors decreased in the order butane > H2 > methane. In presence of a reducing gas, chemisorbed O2 ions oxidizes the corresponding reducing gas to form carbon dioxide and water. Hence, the overall reaction of the gas sensing mechanism of the SnO2 sensor in presence of any reducing gas is nothing but the corresponding oxidation or the combustion reaction of the corresponding gas. Thus, more sensitivity and selectivity of CNP sensors for butane (>91% towards 1000 ppm and 93% for 5000 ppm) as compared to methane or H2, indicates the greater tenancy of butane to undergo oxidation or combustion reaction, thus leading to the preferentially high butane gas sensitivity value. Further, the response and selectivity towards a particular gas however, depends also on the relative bond strengths i.e., C—H of the reducing gases. It is known that the hydrocarbons containing more carbon atoms in their molecule are expected to dissociate easier than methane. This is due to the fact that the bond dissociation energies of C—H for methane (436 KJ/mol) is more compared to that of butane (425 KJ/mol) [53]. So, CNP sensors show faster response and better sensitivity towards butane and the least sensitivity to methane since it a more stable hydrocarbon than butane. Thus, we believe that the high sensitivity exhibited by the nanoparticles of SnO2 is mainly due to the “size effect” achieved by
the CNP as corroborated by various experimental evidences. Though there are numerous reports on the gas sensing properties of SnO2 nanoparticles and their size effect, this report highlights the preparation of SnO2 nanoparticles with a critical size using a cost effective method that has resulted in achieving both high butane sensitivity and a prominent differential sensitivity. Thus, by controlling the particle size to the nano-regime it is possible to develop sensors with high sensitivity and selectivity for a particular gas at a specific operating temperature and this is the most important observation emerged from this study. 4. Conclusion SnO2 nanoparticles with a size of the order of 15 nm or lower have been prepared by a citrate–nitrate process (CNP). The CNP powder based sensors exhibited a sensitivity as high as >91% at 350 C operating temperature with the required selectivity towards 1000 ppm n-butane in presence of other gases like H2 and CH4. It was evident from our results that butane can be detected selectively using CNP derived nano material based sensors with a high sensitivity (>93% towards 5000 ppm butane) and selectivity coefficient. The long-term stability and recovery of this sensor are also remarkable. To conclude, this article demonstrates the effect of using nanoparticles of SnO2 in developing gas sensors with enhanced sensitivity and selectivity towards a particular gas for practical applications without the use of any noble metal catalyst. Acknowledgements PSD acknowledges the financial support from Board of Research in Nuclear Sciences (BRNS). PN is a Senior Research Fellow of the Council of Scientific and Industrial Research (CSIR), Govt. of India. The authors also thank the members of the Sensor and Actuator Division for their kind cooperation during the initial stage of this work. References [1] K. Ihokura, J. Watson, The Stannic Oxide Gas Sensor: Principles and Applications, 1st ed., CRC Press Inc., USA, 1994. [2] G. Eranna, B.C. Joshi, D.P. Runthala, R.P. Gupta, Crit. Rev. Solid State Mater. Sci. 29 (2004) 111–188. [3] L. Renard, H. Elhamzaoui, B. Jousseaume, T. Toupance, G. Laurent, F. Ribot, H. Saadaoui, J. Brotz, H. Fuess, R. Riedel, A. Gurlo, Chem. Comm. 47 (2011) 1464–1466. [4] Y. Tan, C. Li, Y. Wang, J. Tang, X. Ouyang, Thin Solid Films 516 (2008) 7840–7843. [5] X.M. Yin, C.C. Li, M. Zhang, Q.Y. Hao, S. Liu, Q.H. Li, B.L. Chen, T.H. Wang, Nanotechnology 20 (2009) 455503-1–455503-6. [6] D. Wang, P. Hu, J. Xu, X. Dong, Q. Pan, Sens. Actuators B 140 (2009) 383–389. [7] M. Xu, J. Zhang, S. Wang, X. Guo, H. Xia, Y. Wang, S. Zhang, W. Huang, S. Wu, Sens. Actuators B 146 (2010) 8–13. [8] N.V. Hieu, N.A.P. Duc, T. Trung, M.A. Tuan, N.D. Chien, Sens. Actuators B 144 (2010) 450–456. [9] F. Song, H. Su, J. Han, J. Xu, D. Zhang, Sens. Actuators B 145 (2010) 39–45. [10] Y.I. Lee, K.J. Lee, D.H. Lee, Y.K. Jeong, H.S. Lee, H.C. Yong, Curr. Appl. Phys. 9 (2009) S79–S81. [11] H. Wang, J. Liang, H. Fan, B. Xi, M. Zhang, S. Xiong, Y. Zhu, Y. Qian, J. Solid State Chem. 181 (2008) 122–129. [12] T. Kida, T. Doi, K. Shimanoe, Chem. Mater. 22 (2010) 2662–2667. [13] S. Shao, M. Dimitrov, N. Guan, R. KÅhn, J. Mater. Chem. 19 (2009) 8411–8417. [14] G. Jime’nez-Cadena, J. Riu, F.X. Rius, Analyst 132 (2007) 1083–1099. [15] H.R. Kim, K.-I.I. Choi, J.-H. Lee, S.A. Akbar, Sens. Actuators B 136 (2009) 138–143. [16] X. Lou, C. Peng, X. Wang, W. Chu, Vaccum 81 (2007) 883–889. [17] M.R. Vaezi, S.K. Sadrnezhaad, Mater. Sci. Eng. B 140 (2007) 73–80. [18] U. Kersen, L. Holappa, Anal. Chim. Acta 562 (2006) 110–114. [19] T. Hyodo, N. Nishida, Y. Shimizu, M. Egashira, Sens. Actuators B 83 (2002) 209–215. [20] M.I. Baraton, L. Merhari, H. Ferkel, J.F. Castagnet, Mater. Sci. Eng. C 19 (2002) 315–321. [21] A. Chowdhuri, V. Gupta, K. Sreenivas, R. Kumar, S. Mozumdar, P.K. Patanjali, Appl. Phys. Lett. 84 (2004) 1180–1182.
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223 [22] Y.J. Chen, L. Nie, X.Y. Xue, Y.G. Wang, T.H. Wang, Appl. Phys. Lett. 88 (2006) 083105-1–083105-3. [23] Y.J. Chen, X.Y. Xue, Y.G. Wang, T.H. Wang, Appl. Phys. Lett. 87 (2005) 233503-1–233503-3. [24] G.J. Li, X.H. Zhang, S. Kawi, Sens. Actuators B 60 (1999) 64–70. [25] G.G. Lee, S. Joong, L. Kang, Sens. Actuators B 107 (2005) 392–396. [26] Y. Liu, E. Koep, M. Liu, Chem. Mater. 17 (2005) 3997–4000. [27] N.L. Wu, S.Y. Wang, I.A. Rusakova, Science 285 (1999) 1375–1377. [28] E.R. Leite, I.T. Weber, E. Longo, J.A. Varela, Adv. Mater. 12 (2000) 965–968. [29] J. Zhu, Z. Lu, S.T. Aruna, D. Aurbach, A. Gedanken, Chem. Mater. 12 (2000) 2557–2566. [30] A. Kolmakov, Y. Zhang, G. Cheng, M. Moskovis, Adv. Mater. 15 (2003) 997–1000. [31] Y. Wang, X. Jiang, Y. Xia, J. Am. Chem. Soc. 125 (2003) 16176–16177. [32] C. Xu, J. Tamaki, N. Miura, N. Yamazoe, Sens. Actuators B 3 (1991) 147–155. [33] M.E. Frank, T.J. Koplin, U. Simon, Small 2 (2006) 36–50. [34] A. Sutka, G. Mezinskis, Front. Mater. Sci. 6 (2012) 128–141. [35] L.D. Jadhav, S.P. Patil, A.U. Chavan, A.P. Jamale, V.R. Puri, Micro Nano Lett. (2015) 6. [36] A.S. Mukasyan, P. Epstein, P. Dinka, Adv. Sci. Tech. 63 (2010) 187–196. [37] S.T. Aruna, A.S. Mukasyan, Curr. Opin. Solid State Mater. Sci. (2008) 44–50. [38] T. Mimani, K.C. Patil, Mater. Phys. Mech. 4 (2001) 134–137.
223
[39] A. Kumar, P.S. Devi, H.S. Maiti, Chem. Mater. 16 (2004) 5562–5563. [40] S. Banerjee, A. Bumajdad, P.S. Devi, Nanotechnology 22 (2011) 275506-1–275506-3. [41] A. Kumar, P.S. Devi, H.S. Maiti, J. Power Sources 161 (2006) 79–86. [42] S. Banerjee, K. Priolkar, P.S. Devi, Inorg. Chem. 50 (2011) 711–713. [43] L. Fraigi, D.G. Lamas, N.E. Walsoe de Reca, Nanostruc. Mater. 11 (1999) 311–318. [44] L.B. Fraigi, D.G. Lamas, N.E. Walsoe de Reca, Mater. Lett. 47 (2001) 262–266. [45] H. Liu, S.P. Gong, Y.X. Hu, J.Q. Liu, D.X. Zhou, Sens. Actuators B 140 (2009) 190–195. [46] G. Neri, A. Bonavita, G. Micali, N. Donato, F.A. Deorsola, P. Mossino, I. Amato, B. De Benedetti, Sens. Actuators B 117 (2006) 196–204. [47] M. Bhagwat, P. Shah, V. Ramaswamy, Mater. Lett. 57 (2003) 1604–1611. [48] S. Majumdar, S. Chakraborty, P. Sujatha Devi, A. Sen, Mater. Lett. 62 (2008) 1249–1251. [49] J. Ahmed, S. Vaidya, T. Ahmad, P. Sujatha Devi, D. Das, A.K. Ganguli, Mater. Res. Bull. 43 (2008) 264–271. [50] B. Panchapakesan, R. Cavicchi, S. Semancik, D.L. DeVoe, Nanotechnology 17 (2006) 415–425. [51] D.A. Popescu, F.B. Verduraz, Catal. Today 70 (2001) 139–154. [52] R. Summit, J. Appl. Phys. 39 (1968) 3762–3767. [53] A. Gurlo, ChemPhysChem 7 (2006) 2041–2052.
Contents lists available at ScienceDirect
Materials Research Bulletin journal homepage: www.elsevier.com/locate/matresbu
High butane sensitivity and selectivity exhibited by combustion synthesized SnO2 nanoparticles Suparna Banerjee 1, Pratanu Nag 1, Sanhita Majumdar, P. Sujatha Devi * Nano-Structured Materials Division, CSIR-Central Glass and Ceramic Research Institute, 196, Raja S. C. Mullick Road, Jadavpur, Kolkata 700032, India
A R T I C L E I N F O
A B S T R A C T
Article history: Received 8 January 2014 Received in revised form 5 January 2015 Accepted 7 January 2015 Available online 8 January 2015
We report here the high butane sensitivity and selectivity exhibited by nanoparticles of SnO2 prepared by a citrate–nitrate gel combustion process. Using the synthesized powders, the sensitivity towards 500, 1000 and 5000 ppm n-butane has been measured at various operating temperatures from 350 C to 500 C. The powder calcined at 600 C exhibited a sensitivity of 91 0.5% towards 1000 ppm n-butane with a recovery time of 15 s at an operating temperature of 350 C. We have explored the effect of size on both butane gas sensitivity and selectivity and found that the reduction of particle size of the order of 10 nm could result in a remarkable increase in the response of the sensor towards butane compared to other reducing gases like H2 and CH4. The most important observation that emerged from our studies is a combination of sensitivity and selectivity that has been achieved using SnO2 nanoparticles without the addition of any noble metal catalyst such as Pd or Au. ã 2015 Elsevier Ltd. All rights reserved.
Keywords: A. Oxides B. Chemical synthesis D. Surface properties
1. Introduction Tin dioxide (SnO2) is one of the most widely used semiconducting oxides for gas sensor applications due to its high sensitivity, reliability, durability and above all the lower cost [1]. Thick or thin film SnO2-based gas sensors have been extensively studied to detect a variety of toxic and explosive gases such as CO, hydrocarbons, ammonia, H2, nitrogen oxides and H2S [1–26]. The sensitivity of a thick film gas sensor depends mainly on the particle size, shape and their distribution, specific surface area and porosity of the film. Needless to say, all these properties are directly related to the method of preparation of the samples. Thus, preparation techniques that can produce controlled size and shaped particles are of paramount importance in gas sensor development. As a result, many attempts have been made to study the effect of reducing the particle size to nanometric dimension and exploring its consequent effect on the chemical sensing properties [27–32]. When the particle size (D) gets reduced to nanometric dimensions, a dramatic improvement in the gas sensing properties could be expected since a large fraction of the atoms are present at the surface or the interface regions with structure and properties different from those of the bulk. The
* Corresponding author. Tel.: +91 33 2483 8082; fax: +91 33 2473 0957. E-mail addresses: [email protected], [email protected] (P. S. Devi). 1 These two authors contributed equally. http://dx.doi.org/10.1016/j.materresbull.2015.01.025 0025-5408/ ã 2015 Elsevier Ltd. All rights reserved.
prominent effect of nano size, however, is associated with the thickness of the electron-depleted surface layer, which is defined as the Debye length ‘LD’. The Debye length ‘LD’ for a semiconducting material could be calculated as [33], p eO ekB T (1) LD ¼ e2 nd where kB is Boltzmann’s constant, e the dielectric constant, eo the permittivity of free space, T the operating temperature, e the electron charge and nd the carrier concentration. For example, LD for SnO2 has been estimated to be 3 nm. If the particle size is reduced to a size that is comparable to or lower than 2 LD, the whole crystallite is depleted of electrons and this in turn causes the gas sensitivity of the element to change dramatically with D. Since LD for SnO2 is 3 nm, the critical size at which SnO2 could exhibit the “size related nano- effect” and its consequent impact on gas sensitivity should be around 6 nm or lower. Therefore, in order to improve the gas sensitivity of SnO2 based materials it is desired to work with nanoparticles having size range within 6 nm. Although there are many reports on the synthesis of nanoparticles of SnO2 by sonochemical technique, sol–gel processing, co-precipitation technique, hydrothermal route, and gel-combustion route, the studies dealing with the butane gas sensitivity of such materials are rather scarce as evident from the cited references [2–31]. More importantly, to our knowledge most of the reported results are on the sensitivity leaving behind the importance of selectivity untouched and unexplored in most cases [27–33].
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
217
In this paper, we report the synthesis of SnO2 nanoparticles by a citrate–nitrate process (CNP) and their detailed gas sensing characteristics towards n-butane, an important component of the liquefied petroleum gas (LPG). This process could produce homogeneous high surface area powder with superior properties as demonstrated previously by other investigators [34–42].Though there has been a very few attempts to prepare SnO2 particles by combustion process [43–47], none has reported the potential of this economically viable process in preparing highly sensitive SnO2 particles for butane detection. Our interest mainly lies in exploring the sensitivity and selectivity of nanoparticles of SnO2 prepared by this economically viable process and to understand the effect of particle size on the performance of the sensor made thereof. Interestingly, in the present case, the CNP prepared SnO2 nanoparticles exhibited a high sensitivity and selectivity with a fast recovery time towards n-butane compared to SnO2 based systems prepared by us by other techniques [48,49].
The consistency of the paste and the processing variables were optimized to get final coatings of 50–60 mm thickness. Using the paste a thin coating was made on the outer surface of the alumina tubes of 3 mm length, 2 mm outer diameter and of 0.5 mm thickness attached with gold electrodes and platinum lead wires. After coating, the sensors were cured at 600 C for 2 h. so that the sensor surfaces have particles of size comparable to that of the characterized nanoparticles. The electrical resistance of the sensors in air and in n-butane was measured at different temperatures (up to 500 C) in an ambient of 60–70% relative humidity using a digital multimeter (Solartron) and a constant voltage/current source (Keithley 228 A). All the samples were initially aged at 350 C for 72 h to achieve the desired stability of the resistance value before the measurements. The percentage response or sensitivity) (%S) of the films was calculated as: DR ðRA RG Þ ð%ÞS ¼ ¼ 100 (2) RA RA
2. Experimental
RA and RG are the sensor resistances in air and measuring gas, respectively. DR for reducing gases is (RARG) where RA>RG. The response time has been measured as the time taken for a sensor to read 90% of full-scale reading after being exposed to a given gas and recovery time was the time taken for a sensor to come back to its original state when the target gas is removed.
2.1. Chemicals Tin chloride dihydrate, SnCl22H2O ([MERCK India, GR], citric acid monohydrate, ammonium nitrate, distilled water and absolute ethanol were used. All reagents used were analytically pure.
3. Results and discussion 2.2. Preparation of SnO2 nanoparticles 3.1. Thermal characterization To prepare nano particles of tin dioxide, a mixture of an aqueous solution of SnCl22H2O (0.2 M) and ammonium nitrate solutions were taken in a beaker and calculated amount of citric acid monohydrate was added to this mixed solution keeping the citrate–nitrate molar ratio as 0.3. The rest of the procedure is similar to what has been reported earlier [39–42]. The precursor ash powder obtained was calcined at 600 C for 6 h in air to remove unburnt carbonaceous material and to form a pure oxide phase. From the concept of propellant chemistry, the net oxidizing valency of metal nitrates to the net reducing valency of fuel should be unity to get the maximum exothermicity for driving the combustion reaction. In our experiments, we have optimized the citrate to nitrate ratio as 0.3 for achieving a satisfactory combustion reaction.
In order to understand the thermal decomposition behavior of the samples, simultaneous TGA-DTA measurements have been carried out on the gel and the as-prepared SnO2 powder (Fig. 1(a) & (b)). The gel sample exhibited a weight loss of 88% whereas the as-prepared powder exhibited a weight loss of only 25%. The initial weight loss of 3.26% exhibited by the gel sample between 30 C and 110 C is well supported by an endothermic peak on the DTA at 119 C. These changes are probably due to the dehydration of surface adsorbed water. From 110 C to 179 C, a weight loss of 11.24% along with a corresponding exothermic shoulder at 174 C was observed. From 179 C to 244 C, a total weight loss of 14.84% was evident. The above changes were accompanied by an
2.3. Characterization
[(Fig._1)TD$IG]
Thermogravimetric analysis (TGA) of the as-prepared gel and uncalcined powder was carried out on a Netzsch (Germany) simultaneous DTA-TGA analyser (Model No. STA 449 C) from room temperature to 1000 C at a heating rate of 10 C/min. The phase identification of the calcined powder was carried out by X-ray diffraction analysis (Philips PW1710) using Cu Ka radiation (l = 1.5406 Å). The surface area analysis of the calcined powder was measured by the Brunauer–Emmet–Teller (BET) method on Micromeritics Gemini II 2370 equipment. The particle morphology and local crystallographic structure were studied by transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM), respectively. The samples for microscopy were prepared by depositing micro-drops of the acetone-dispersed sample on a carbon coated copper grid followed by drying. The average particle size of the calcined powders was calculated using the bight field (BF) image obtained from TEM. 2.4. Gas sensing measurement For gas sensing study, a thick paste of the powder was prepared using an alumina gel as a binder. The details of the sensor fabrication and processing have been reported elsewhere [48,49].
Fig. 1. TG-DTA curves of the (a) gel, (b) as-synthesized SnO2 powder obtained from citrate–nitrate process.
218
[(Fig._2)TD$IG]
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
compounds present in the gel. From 329 C to 554 C a final weight loss of 4.01% corresponding to an exothermic peak at 504 C was also evident. In case of the as-prepared powder, from 74 C to 558 C, a single step weight loss of 24.67% accompanied by an exothermic peak centered at 504 C was observed. This has been assigned to the formation of SnO2 by the decomposition of citrate–nitrate gel derived precursor powder followed by recrystallization. Based on the above observations, a minimum calcination temperature of 600 C was fixed to prepare phase-pure SnO2 particles free from any impurity phases. 3.2. Structure and morphology
Fig. 2. XRD patterns of SnO2 powders (a) Standard (b) CNP calcined at 600 C and (c) CNP calcined at 800 C.
endotherm at 199 C and an exotherm at 234 C, respectively. From 244 C to 329 C, a further weight loss of 54.86% took place corresponding to an endothermic peak centered at 319 C. The above changes are related to probable dehydration through condensation and the heat released by decomposition of organic
The X-ray diffraction patterns of 600 C (hereafter designated as CNP1) and 800 C (hereafter designated as CNP2) calcined powders along with the XRD pattern of a commercial SnO2 (Merck grade, hereafter designated as standard) powder are shown in Fig. 2. The CNP powders exhibited all the reflections of the commercially available SnO2 powder having rutile type structure. The comparison of the plots in Fig. 2 obviously shows the difference in the widths of the diffraction peaks arising from the variation in the crystallite as well as particle size. The broad peaks observed in the XRD patterns for the nanoparticles indicate line-broadening arising from the finite number of diffracting planes and the extent of broadening was calculated using the Scherer formula. In Fig. 3a, the TEM, and Fig. 3b and c the HRTEM pictures of the CNP1 powder are presented and in Fig. 4a and b the TEM and HRTEM pictures of CNP2 and in Fig. 4c and d, the TEM and HRTEM
[(Fig._3)TD$IG]
Fig. 3. (a) TEM, (b) and (c) HRTEM (with corresponding FFT figures at the inset) and (d) SAED pattern of CNP1 powder.
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
[(Fig._4)TD$IG]
219
Fig. 4. (a) and (c) bright field TEM and (b) and (d) HRTEM (with corresponding FFT figures at the inset) of CNP2 and SnO2 standard powders, respectively.
pictures of SnO2-standard are presented. Fig. 3a indicates the presence of well-distributed very fine particles of pure SnO2 having average particle size of the order of 4 nm as evident from the size distribution shown in Fig. 5a. The average particle sizes were found to be 4.01 and 14.62 nm for CNP1 and CNP2 powders, respectively, calculated from the histogram of particle size distribution. In Fig. 3b and c, two high-resolution TEM images obtained from spherical and elliptical SnO2 particles are shown (with corresponding FFT figures at the inset) in order to further confirm the single nanocrystalline nature of these isolated particles. In Fig. 3c two overlapping (11 0) planes can be seen in an elliptical nanoparticles. The selected area electron diffraction (SAED)
pattern taken from a representative particle shown in Fig. 3d indicates lattice fringes corresponding to (11 0), (1 0 1), (2 0 0), (2 1 0), (2 11), (2 2 0) and (3 0 1) planes of tetragonal SnO2. From Fig. 4a, it is very clear that the CNP2 particles are larger than CNP1 with an average size of 14.62 nm as shown in the size distribution in Fig. 5b. SnO2 (standard) on the other hand are much higher in size and highly agglomerated with average particle size of around 99.87 nm (Fig. 5c). The corresponding HRTEM images are shown in Fig. 4b and 4d.The TEM pictures of CNP1 (Fig. 3a), CNP2 (Fig. 4a) and standard (Fig. 4c) unequivocally confirm the difference in size of the individual particles. The data shown in Table 1, further indicates the superior properties of CNP1 powder
[(Fig._5)TD$IG]
Fig. 5. Particle size distribution with Gaussian Fit derived from the bright field TEM image of the (a) CNP1, (b) CNP2 and (c) Standard powder samples with corresponding histogram.
220
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
Table 1 Physico–chemical characteristics of standard SnO2 and pure SnO2 powders prepared by CNP. SnO2ID
TC a ( C) [Å]
c [Å]
Lattice SBET volume (m2/g) [Å]3
(Standard) – 4.72 0.01 3.14 0.01 70 (CNP2) 800 4.74 0.01 3.18 0.01 71 (CNP1) 600 4.79 0.02 3.16 0.07 72
– 40 58
Average particle size (DTEM) (nm) 99.87 14.62 4.01
TC = calcination temperature; DTEM = Particle size from TEM.
[(Fig._6)TD$IG]
samples. The FTIR spectrum of standard powder having a larger particle size is broad with the presence of two weakly defined bands at around 520 cm1 and 615 cm1 respectively. The principle bands at 615–623 cm1 and 657–667 cm1 in FTIR spectra of SnO2 are the fundamental vibration frequencies of ySn–O and yO–Sn–O of tin (IV) dioxide, respectively, while the bands at around 520–530 cm1 can be assigned to the fundamental vibrations (ySn–O,T; T = terminal) of tin (IV) dioxide [51]. The discrepancies in the positions of IR spectra arise due to factors such as stoichiometry of the oxide (i.e., the presence of defects) the size and the state of aggregation of the particles [51–53]. The remarkable blue shifting of the FTIR peaks of CNP2 and CNP1 compared to standard samples and the disappearance of the splitting nature of the spectral bands in the latter sample could be due to the presence of more oxygen defects in the former sample. The presence of oxygen vacancy in CNP1 and CNP2 is expected to give rise to greater reduced mass effect resulting in comparatively shorter and stronger Sn—O bonds of O—Sn—VÄ in CNP1 and CNP2 ~ = 1/2p (k/m) where y is the wave number, k, the force samples. As y constant and m the reduced mass effect, the higher bond strength decreases m and increases the k values for the O–Sn–VÄ bonds in ~ Sn–O (CNP1) > y ~ Sn–O (CNP2) >> y ~ Sn–O (M). CNP1 and CNP2. i.e., y Moreover, the sharper and the split nature of the FTIR spectral bands reveal a higher structural disorder of the CNP powders than the standard powder. 3.4. Gas-sensing properties
Fig. 6. FTIR spectra of the standard, CNP1 and CNP2 SnO2 powders.
such as high surface area (58 m2/gm) and low particle size (around 4 nm) in comparison to CNP2 and standard powders. In Table 1, the particle size from TEM and lattice parameters obtained from the powder XRD data are shown. The lattice parameter shows a decrease in the ‘a’ value with increase in the calcination temperature. 3.3. FTIR studies on the calcined powders In Fig. 6, FTIR spectra of the standard, CNP1 and CNP2 SnO2 powders are presented. The FTIR spectrum of CNP1 powder exhibited three well defined bands at 667 cm1 and 623 cm1 corresponding to the two IR active modes namely (Eu) and a shoulder around 530 cm1, attributed to an IR active mode (A2u) of tin (IV) oxide, respectively [44,49,50]. These three bands shift to lower wave numbers of around 657 cm1, 619 cm1 and 526 cm1, respectively with less splitting for the larger sized CNP2 sample. The FTIR spectra in Fig. 6 indicate a blue shift of the IR bands with gradual decrease in particle size from Standard to CNP2 to CNP1
In order to understand the gas sensing characteristics, the fabricated sensors were exposed to 500,1000 and 5000 ppm butane, respectively at various temperatures. The observed response as a function of the operating temperature is plotted in Fig. 7a–c. For 500 ppm n-butane, at 500 C, a maximum percentage response of around 89% and 77% were obtained for the CNP1 and CNP2 powders, respectively, in contrast to the commercial SnO2 that exhibited only 40% response. At an operating temperature of 350 C, the percentage response of CNP1 increased from 60 to 91 to 93% and that of CNP2 increased from 48 to 83 to 89% on increasing the gas concentrations from 500 to1000 to 5000 ppm, respectively. It is clear from Fig. 7a–c that the gas response increases with operating temperature, reaches a maximum at 500 C in case of 500 ppm and at 450 C in cases of 1000 and 5000–ppms, respectively, and then falls with further increase in operating temperature due to desorption of the surface adsorbed metastable oxygen species from the surface of SnO2 coating. Fig. 8 shows the typical dynamic electrical response of the CNP1 sensor for eight consecutive cycles. The recovery time was around 15 s (for about 90% recovery) and a sensitivity of around 91.5% towards 1000 ppm butane gas was observed at 350 C. The fast recovery time exhibited by these sensors could be attributed to
[(Fig._7)TD$IG]
Fig. 7. Percent sensitivity (response) recorded for SnO2 thick film sensor at different operating temperatures and at different butane concentrations, (a) 500 ppm (b) 1000 ppm and (c) 5000 ppm, respectively.
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
[(Fig._8)TD$IG]
221
Fig. 8. Dynamic response curve of the thick film CNP1 sensor upon exposure to 1000 ppm butane at an operating temperature of 350 C for eight consecutive cycles.
[(Fig._9)TD$IG]
Fig. 9. The response time of the CNP1 sensor towards 1000 ppm n-butane measured at 350 C.
the superior characteristics of the combustion synthesized powder. The observed response of the CNP1 sensor towards 1000 ppm butane shown in Fig. 9 confirms that a minimum exposure time of 5 s was sufficient for the sensor to respond to the gas, albeit with a lower percent response. On exposure of 12 s, a percent sensitivity as high as 95% was observed. It is clear from Fig. 9 that the gas response increases with response time (Rs), reaches a maximum (Rs = 12 s) and falls above 12 s. To further understand and establish the novelty of the SnO2 nanoparticle based CNP1 sensor, its cross-sensitivity has also been examined in presence of other reducing gases at 350 C. In Fig. 10, the sensitivities exhibited by the CNP1 sensor towards 1000 ppm’s each of butane, methane and H2 at 350 C, respectively, are compared in order to determine its selectivity to any particular gas. It is interesting to see the differential sensitivity of the sensor (>91% towards 1000 ppm n-butane at 350 C) with respect to other gases like H2 (57%) and methane (20%) which confirms that if we have a mixture of 1000 ppm’s each of butane, methane and H2 gases, the fabricated sensor could effectively and efficiently
[(Fig._10)TD$IG]
detect and sense only butane. The selectivity coefficient, b (b = Sbutane/Sgas) varied in the order CH4 > H2 > C4H10 with 4.57 and 1.58 for methane and H2, respectively. This selective detection of butane by the fabricated sensor when there is a mixture of the above mentioned gases is very interesting since it is very difficult to fabricate SnO2 based sensors that are selective to a particular gas. In order to find the underlying reason for the high sensitivity and the extremely high butane selectivity exhibited by the CNP1 nanoparticles we have checked the cross sensitivity of all the samples which clearly illustrates the direct influence of the particle size on selectivity. The response towards n-butane decreased in the order CNP1 > CNP2 > standard with the CNP1 sample exhibiting a high butane sensitivity and selectively compared to the other two samples. In Fig. 11, the response of all the sensors at 350 C towards 5000 ppm n-butane and methane are compared. The (%) response of CNP1 and CNP2 sensors towards 5000 ppm n-butane is much higher and comparable. The response of both the samples towards same concentration of methane gas, on the other hand was poor with CNP2 exhibiting a slightly higher response than CNP1. Interestingly, the response of standard SnO2 coated sensor towards both butane and methane were comparable. We have also monitored the stability of the prepared sensors by checking the long-term drift of the resistance in air (Rair) and the percent response towards 500 ppm gas at 500 C. The measurements were performed for two successive months after initial checking with ten days interval of time. The percent response of the best CNP1 sensor increased slightly in the first month and then became almost stable. However, the resistance remained almost constant. The gas sensing mechanism [43,46] of a semiconductor depends on the adsorption and subsequent desorption reactions occurring at the surface of the semiconducting oxide, which are highly temperature dependent. Normally, the atmospheric oxygen chemisorbs on the SnO2 surface, consuming free electrons from the conduction band resulting in an electron depleted surface layer and a consequent rise in resistance of SnO2 (Eqs. (3)–(5)).
[(Fig._1)TD$IG]
Fig. 10. Dynamic response curves of the thick film CNP1 sensor upon exposure of 1000 ppm of butane, hydrogen and methane gases at an operating temperature of 350 C.
Fig. 11. Selectivity study of CNP1, CNP2 and standard powder coated sensors towards 5000 ppm n-butane and 5000 ppm methane.
222
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223
O2 (gas) @ O2 (physisorbed)
(3)
O2 (physisorbed) @ e O2 (chemisorbed)
(4)
O2 +e @ 2 O (chemisorbed)
(5)
At low temperature, oxygen adsorbs on the SnO2 surface nondissociatively in the molecular form as charged O2. At high temperatures, it dissociates to 2O or O2 [33]. The reaction of the butane gas with these chemisorbed oxygen species releases the bound electrons back to the conduction band as free electrons, which decreases the resistance of the SnO2 coatings. C4H10 + 2 O (chemisorbed) ! C4H8O + H2O + 2e
(6)
The high butane sensitivity and selectivity exhibited by the SnO2 CNP1 nanoparticle is almost similar to the results reported by Xu et al. [32] reporting a strong relation between the grain diameter (D) and sensitivity. As demonstrated, when the particle size gets closer to or smaller than 2 LD, the gas sensitivity could in principle, be controlled by the grains themselves and not the gain boundaries or interfaces [32]. This in turn gives rise to a substantial increase in the gas sensitivity since in this case the entire grains will become depleted of electrons. The 600 C calcined SnO2 nanoparticles prepared by the citrate–nitrate process (CNP1) have an average particle size of 4 nm, which is lower or close to the reported critical size of pure SnO2 to exhibit the size related effect. This clearly demonstrates that the CNP1 particles with a size of the order of 4 nm shows the maximum sensitivity compared to larger sized particles, thus proving the size effect as the major cause for the high butane sensitivity. A correlation between the particle size and selectivity was reported by Chen et. al., [22,23]. The authors claim high ethanol sensitivity from 3 nm sized SnO2 nano rods, as a consequence of the size reduction down to LD [23]. A report by Panchapakesan et al. also discussed about the sensitivity and selectivity of tin oxide nanostructures on large arrays of micro hotplates [50]. Thus, it is evident from our results that butane can be detected selectively using CNP derived nano material based sensors with a high sensitivity (>91% towards 1000 ppm butane). Now the question arises as to why the CNP1 based sensors having a finer particle size and higher surface area showed high sensitivity and selectivity towards n-butane only? It is evident from Fig. 10 that the sensitivity of the CNP sensors decreased in the order butane > H2 > methane. In presence of a reducing gas, chemisorbed O2 ions oxidizes the corresponding reducing gas to form carbon dioxide and water. Hence, the overall reaction of the gas sensing mechanism of the SnO2 sensor in presence of any reducing gas is nothing but the corresponding oxidation or the combustion reaction of the corresponding gas. Thus, more sensitivity and selectivity of CNP sensors for butane (>91% towards 1000 ppm and 93% for 5000 ppm) as compared to methane or H2, indicates the greater tenancy of butane to undergo oxidation or combustion reaction, thus leading to the preferentially high butane gas sensitivity value. Further, the response and selectivity towards a particular gas however, depends also on the relative bond strengths i.e., C—H of the reducing gases. It is known that the hydrocarbons containing more carbon atoms in their molecule are expected to dissociate easier than methane. This is due to the fact that the bond dissociation energies of C—H for methane (436 KJ/mol) is more compared to that of butane (425 KJ/mol) [53]. So, CNP sensors show faster response and better sensitivity towards butane and the least sensitivity to methane since it a more stable hydrocarbon than butane. Thus, we believe that the high sensitivity exhibited by the nanoparticles of SnO2 is mainly due to the “size effect” achieved by
the CNP as corroborated by various experimental evidences. Though there are numerous reports on the gas sensing properties of SnO2 nanoparticles and their size effect, this report highlights the preparation of SnO2 nanoparticles with a critical size using a cost effective method that has resulted in achieving both high butane sensitivity and a prominent differential sensitivity. Thus, by controlling the particle size to the nano-regime it is possible to develop sensors with high sensitivity and selectivity for a particular gas at a specific operating temperature and this is the most important observation emerged from this study. 4. Conclusion SnO2 nanoparticles with a size of the order of 15 nm or lower have been prepared by a citrate–nitrate process (CNP). The CNP powder based sensors exhibited a sensitivity as high as >91% at 350 C operating temperature with the required selectivity towards 1000 ppm n-butane in presence of other gases like H2 and CH4. It was evident from our results that butane can be detected selectively using CNP derived nano material based sensors with a high sensitivity (>93% towards 5000 ppm butane) and selectivity coefficient. The long-term stability and recovery of this sensor are also remarkable. To conclude, this article demonstrates the effect of using nanoparticles of SnO2 in developing gas sensors with enhanced sensitivity and selectivity towards a particular gas for practical applications without the use of any noble metal catalyst. Acknowledgements PSD acknowledges the financial support from Board of Research in Nuclear Sciences (BRNS). PN is a Senior Research Fellow of the Council of Scientific and Industrial Research (CSIR), Govt. of India. The authors also thank the members of the Sensor and Actuator Division for their kind cooperation during the initial stage of this work. References [1] K. Ihokura, J. Watson, The Stannic Oxide Gas Sensor: Principles and Applications, 1st ed., CRC Press Inc., USA, 1994. [2] G. Eranna, B.C. Joshi, D.P. Runthala, R.P. Gupta, Crit. Rev. Solid State Mater. Sci. 29 (2004) 111–188. [3] L. Renard, H. Elhamzaoui, B. Jousseaume, T. Toupance, G. Laurent, F. Ribot, H. Saadaoui, J. Brotz, H. Fuess, R. Riedel, A. Gurlo, Chem. Comm. 47 (2011) 1464–1466. [4] Y. Tan, C. Li, Y. Wang, J. Tang, X. Ouyang, Thin Solid Films 516 (2008) 7840–7843. [5] X.M. Yin, C.C. Li, M. Zhang, Q.Y. Hao, S. Liu, Q.H. Li, B.L. Chen, T.H. Wang, Nanotechnology 20 (2009) 455503-1–455503-6. [6] D. Wang, P. Hu, J. Xu, X. Dong, Q. Pan, Sens. Actuators B 140 (2009) 383–389. [7] M. Xu, J. Zhang, S. Wang, X. Guo, H. Xia, Y. Wang, S. Zhang, W. Huang, S. Wu, Sens. Actuators B 146 (2010) 8–13. [8] N.V. Hieu, N.A.P. Duc, T. Trung, M.A. Tuan, N.D. Chien, Sens. Actuators B 144 (2010) 450–456. [9] F. Song, H. Su, J. Han, J. Xu, D. Zhang, Sens. Actuators B 145 (2010) 39–45. [10] Y.I. Lee, K.J. Lee, D.H. Lee, Y.K. Jeong, H.S. Lee, H.C. Yong, Curr. Appl. Phys. 9 (2009) S79–S81. [11] H. Wang, J. Liang, H. Fan, B. Xi, M. Zhang, S. Xiong, Y. Zhu, Y. Qian, J. Solid State Chem. 181 (2008) 122–129. [12] T. Kida, T. Doi, K. Shimanoe, Chem. Mater. 22 (2010) 2662–2667. [13] S. Shao, M. Dimitrov, N. Guan, R. KÅhn, J. Mater. Chem. 19 (2009) 8411–8417. [14] G. Jime’nez-Cadena, J. Riu, F.X. Rius, Analyst 132 (2007) 1083–1099. [15] H.R. Kim, K.-I.I. Choi, J.-H. Lee, S.A. Akbar, Sens. Actuators B 136 (2009) 138–143. [16] X. Lou, C. Peng, X. Wang, W. Chu, Vaccum 81 (2007) 883–889. [17] M.R. Vaezi, S.K. Sadrnezhaad, Mater. Sci. Eng. B 140 (2007) 73–80. [18] U. Kersen, L. Holappa, Anal. Chim. Acta 562 (2006) 110–114. [19] T. Hyodo, N. Nishida, Y. Shimizu, M. Egashira, Sens. Actuators B 83 (2002) 209–215. [20] M.I. Baraton, L. Merhari, H. Ferkel, J.F. Castagnet, Mater. Sci. Eng. C 19 (2002) 315–321. [21] A. Chowdhuri, V. Gupta, K. Sreenivas, R. Kumar, S. Mozumdar, P.K. Patanjali, Appl. Phys. Lett. 84 (2004) 1180–1182.
S. Banerjee et al. / Materials Research Bulletin 65 (2015) 216–223 [22] Y.J. Chen, L. Nie, X.Y. Xue, Y.G. Wang, T.H. Wang, Appl. Phys. Lett. 88 (2006) 083105-1–083105-3. [23] Y.J. Chen, X.Y. Xue, Y.G. Wang, T.H. Wang, Appl. Phys. Lett. 87 (2005) 233503-1–233503-3. [24] G.J. Li, X.H. Zhang, S. Kawi, Sens. Actuators B 60 (1999) 64–70. [25] G.G. Lee, S. Joong, L. Kang, Sens. Actuators B 107 (2005) 392–396. [26] Y. Liu, E. Koep, M. Liu, Chem. Mater. 17 (2005) 3997–4000. [27] N.L. Wu, S.Y. Wang, I.A. Rusakova, Science 285 (1999) 1375–1377. [28] E.R. Leite, I.T. Weber, E. Longo, J.A. Varela, Adv. Mater. 12 (2000) 965–968. [29] J. Zhu, Z. Lu, S.T. Aruna, D. Aurbach, A. Gedanken, Chem. Mater. 12 (2000) 2557–2566. [30] A. Kolmakov, Y. Zhang, G. Cheng, M. Moskovis, Adv. Mater. 15 (2003) 997–1000. [31] Y. Wang, X. Jiang, Y. Xia, J. Am. Chem. Soc. 125 (2003) 16176–16177. [32] C. Xu, J. Tamaki, N. Miura, N. Yamazoe, Sens. Actuators B 3 (1991) 147–155. [33] M.E. Frank, T.J. Koplin, U. Simon, Small 2 (2006) 36–50. [34] A. Sutka, G. Mezinskis, Front. Mater. Sci. 6 (2012) 128–141. [35] L.D. Jadhav, S.P. Patil, A.U. Chavan, A.P. Jamale, V.R. Puri, Micro Nano Lett. (2015) 6. [36] A.S. Mukasyan, P. Epstein, P. Dinka, Adv. Sci. Tech. 63 (2010) 187–196. [37] S.T. Aruna, A.S. Mukasyan, Curr. Opin. Solid State Mater. Sci. (2008) 44–50. [38] T. Mimani, K.C. Patil, Mater. Phys. Mech. 4 (2001) 134–137.
223
[39] A. Kumar, P.S. Devi, H.S. Maiti, Chem. Mater. 16 (2004) 5562–5563. [40] S. Banerjee, A. Bumajdad, P.S. Devi, Nanotechnology 22 (2011) 275506-1–275506-3. [41] A. Kumar, P.S. Devi, H.S. Maiti, J. Power Sources 161 (2006) 79–86. [42] S. Banerjee, K. Priolkar, P.S. Devi, Inorg. Chem. 50 (2011) 711–713. [43] L. Fraigi, D.G. Lamas, N.E. Walsoe de Reca, Nanostruc. Mater. 11 (1999) 311–318. [44] L.B. Fraigi, D.G. Lamas, N.E. Walsoe de Reca, Mater. Lett. 47 (2001) 262–266. [45] H. Liu, S.P. Gong, Y.X. Hu, J.Q. Liu, D.X. Zhou, Sens. Actuators B 140 (2009) 190–195. [46] G. Neri, A. Bonavita, G. Micali, N. Donato, F.A. Deorsola, P. Mossino, I. Amato, B. De Benedetti, Sens. Actuators B 117 (2006) 196–204. [47] M. Bhagwat, P. Shah, V. Ramaswamy, Mater. Lett. 57 (2003) 1604–1611. [48] S. Majumdar, S. Chakraborty, P. Sujatha Devi, A. Sen, Mater. Lett. 62 (2008) 1249–1251. [49] J. Ahmed, S. Vaidya, T. Ahmad, P. Sujatha Devi, D. Das, A.K. Ganguli, Mater. Res. Bull. 43 (2008) 264–271. [50] B. Panchapakesan, R. Cavicchi, S. Semancik, D.L. DeVoe, Nanotechnology 17 (2006) 415–425. [51] D.A. Popescu, F.B. Verduraz, Catal. Today 70 (2001) 139–154. [52] R. Summit, J. Appl. Phys. 39 (1968) 3762–3767. [53] A. Gurlo, ChemPhysChem 7 (2006) 2041–2052.