Vapour sensing properties of a cadmium oxide-antimony oxide system ceramic, Cd,Sb*O6.2 Toby J. A. R.Hitch and Colin L. Honeybourne University of the West of England, Coldharbour Lane, Bristol, UK BS16 1 QY The vapour sensing properties of a tetragonal pseudo-pyrochlore, Cd,Sb206.8, are reported. Cd2Sb206.8, prepared by a high- temperature solid-state synthesis, is an n-type semiconductor. Thick-film sensors incorporating Cd2Sb,06 8 as the active material were fabricated utilising screen printing technology. Sensors heated to 250-500 "C were exposed to low concentrations of vapours associated with food spoilage (ethanol, ethyl acetate, limonene, pinene and toluene). The sensors were monitored through the current across the electrodes as a result of a small applied voltage.Tin dioxide sensors were also fabricated to act as a comparison. The results highlight considerable selectivity differences between the Cd2Sb206.8 and SnO, sensors, particularly with regard to humidity responses where the lack of response for Cd2Sb206.8 is crucial for an operational gas sensor. An unusual feature is 'chiral' responses exhibited by Cd2Sb2O6.8 to limonene (orange oil) and pinene (pine oil) which may have practical applications in the stereochemistry field. The concept that adsorption of a gas or vapour on the surface of a semiconducting oxide may facilitate a change in the electrical properties of the oxide has been well The so-called 'gas sensor' has been extensively researched with the main body of work concentrating on tin oxide.This is unsurprising as it is cheap and responds well to small alter- ations in the surrounding atmosphere. The inherent problem with Sn02 as a gas sensor is the fact that it is largely non- selective, i.e. it yields a response to almost any impurity in the atmosphere, including water vapour. To a certain extent this problem has been overcome by the addition of catalysts6 and molecular filter^.^ More recently, other oxides have been investigated as possible gas-sensitive elements. For example, Arakawa and co-worker~~~~have investigated a range of perovskite oxides as methane sensors. In fact, sensitivity of the conductance of oxides to traces of reactive gas in air is a widespread phenom- enon. However, there is as yet no a priori system which can predict the gas sensitivity of oxides and can determine which, if any, might show advantageous properties and, in particular, selectivity of response.This paper reports and discusses the vapour sensing proper- ties of a cadmium antimony semiconducting oxide with a pseudo-pyrochlore" structure, CdzSb206.8, towards atmos- pheres containing low concentrations of a range of organic vapours. Electrical Conduction Behaviour The electrical behaviour of Cd2Sb206.8 has been studied by Li and Zhang," who found that two different mechanisms operate at high and low temperatures. At high temperature (> 560 "C), the conduction is mainly related to the defect structure. Cd2Sb20,8 has a tetragonal pseudo-pyrochlore structure.Within this structure six of every seven oxygen atoms are locked in a distorted octahedral network. The remaining oxygen is situated at the centre of the octahedron from where it can migrate easily. Therefore, oxygen vacancies can be described thus: oot+vo++02 (1) Applying the law of mass action, the following equation is t Presented at the Second International Conference on Materials Chemistry, MC2, University of Kent at Canterbury, 17-21 July 1995. reached, in which K, is the quotient of the forward and reverse rate constants implicit in eqn. (1): [VOlP8, =Kl (2) For simplicity, the concentration of occupied octahedral sites COO] can be considered a constant, which gives the following expression for the concentration of oxygen vacancies: [Val==ti, (3) where K is a constant.Because the carrier concentration is proportional to the oxygen vacancy concentration the compound displays a higher electrical conductivity under a reducing atmosphere, which is consistent with the material being classed as an n-type semicon- ducting oxide. In the low-temperature region the dominant mechanism is one of mixed valence (3:5)exhibited by the central metal ion. To meet the electroneutrality condition within the structure, some Sb5+ must be converted to Sb3+, thereby generating a mixed-valence non-stoichiometric oxide. Experimental Procedure Cd2Sb206.8 was prepared using a solid-state oxide synthesis route.12 Ground CdO and Sb205 (Aldrich, both 99.9%) were mixed in a 2 :1 ratio in methanol for 12 h, dried and pelletized.The pellets were calcined at 800 "C for 5 h (20 "C min-' ramp), reground and pelletized and then sintered at 1200°C for 5 h. An X-ray diffractometer was used to confirm the crystal structure.12 Sensors manufactured utilising Cd2Sb@6.s as the active material are illustrated in Fig. 1. A screen-printable paste was prepared containing Cd2Sb@6.8 (>75%), glass frit, surfactant and an organic medium. This was printed onto 1 in2$ (5 per square) alumina substrates over Pt/Au interdigitated electrodes and fired at 800 "C for 3 h (25"C min-' ramp). This gave film thicknesses of approximately 70 pm. The sensors were heated by means of a screen-printed Pt/Au circuit on the reverse face of the substrate.The surface of the sensor was studied by means of scanning electron microscopy (SEM) in conjunction with an energy dispersive analysis by X-rays (EDAX) probe. Sensors were also produced utilizing SnO, (99.999 YO,Aldrich) as the active material in the same way, to act as a contrast to the Cd,Sb,O,., sensors. Sensors were tested using the set-up ~~ ~ $ 1 in=2.54 cm. J. Mater. Chem., 1996, 6(3),285-288 285 -0.2 in FRONT Fig. 1 Screen-printed sensor design (+electrode bar separation = 250 pm) Fig. 2 Experimental apparatus used in vapour sensing investigations. Sensors are inserted into PCB edge connectors. illustrated in Fig. 2 in the temperature range 250-500°C in 50 "C increments. A voltage of 0.25 V was applied across the electrodes and the current was monitored.Sensors were initially run in the ambient atmosphere for 48h to acts as a 'run-in' period. During this time it was observed that the current across the electrodes increased slowly to a steady equilibrium level. Injections of known concentrations of vapour were made to the main chamber via a gas syringe. All materials used were silanated to minimize the sequestration of vapour on the walls of the chamber. The chamber was purged after each injection with 5 dm3 min-l air. When the current had recovered to its equilibrium level another injection could be performed. Results and Discussion A typical experimental run is shown in Fig. 3. Upon injection of EtOH the current across the electrodes increases sharply, i.e.the conductance increases. After the peak current is reached the response decays owing to the decreasing EtOH concen- tration in the chamber. Upon purging a large decrease in current is observed owing to a corresponding decrease in the Fig. 3 Experimental run for Cd,Sb,0,.8 sensor at 400 "Cexposed to EtOH. A, 5ppm EtOH injection; B, chamber purged with 5dm3 min-' air; C, equilibrium time; D, 10 pprn EtOH injection; E, 25 ppm EtOH injection. sensor temperature Responses are displayed in terms of I/Io, where I is the peak response current and I, is the current at ambient atmosphere. When interpreting the results, some factors must be borne in mind, although the conclusions that can be drawn from the results are relatively unaffected by them.Firstly, although the chamber atmosphere was dried, the by-products of many surface-catalysed reactions may include water vapour and this may cause changes in the patterns of response. Secondly, whilst X-ray diffraction indicated that the Cd2Sb206.8 synthesised was in a single phase, the possibility that the gas response effects might be due to minor impurities cannot be ruled out. Furthermore, there will undoubtedly be heterogeneities on a fine scale at free surfaces and grain boundaries. However, we consider that the results are not greatly affected by these complicating factors. Indeed, the results were found to be repeatable to a high accuracy (<5%). Comparison of Cd2Sb20,8 and SnO, sensors Table 1 shows a comparison between Cd2Sb20,~, and SnO, tested at 350°C to 10 ppm of a range of organic vapours.The responses are of the same order but significant differences can be seen. Cd,Sb,O,., shows a much higher response to ethyl acetate than SnO,. This is also the case for other temperatures and the difference is magnified at higher vapour concentrations. In the case of toluene the reverse situation operates, where Cd2Sb206.8 displays a smaller response than for SnO,. Again the differences in response are magnified at higher concen- trations. The responses to the terpenes are very closely related. This emphasises the fact that n-type semiconductors utilised as gas sensors do indeed show fundamental differences in response between each other, as observed by Moseley et It would be expected to be especially true for semiconductors of widely differing structures, as is the case here.The most significant differences in response are those for humidity changes. Cd2Sb206.8 yields very small responses to large humidity changes, which is also the case at higher temperatures (up to 5OOOC). This is important because one of the main by-products of many surface-catalysed reactions is water vapour, which would distort the response of moisture- sensitive materials. The sensitivity of SnO, to moisture is seen as one of the main drawbacks to the operational use of gas sensors based on SnO,. Responses to optically active terpenes Codistillation of many plant materials with steam produces a fragrant mixture of liquids called plant essential oils.These oils consist largely of cyclic hydrocarbons called terpenes. Damage/disease in plant material can cause emissions of these terpenes. In this investigation the optically active terpenes limonene (orange oil) and pinene (pine oil) are investigated. Table 1 Comparison of response characteristics for Cd,Sb2o6 and SnO, sensors at 350 "C upon exposure to 10 ppm of various organic vapours organic vapour Cd2Sb2068 Sn02 ethanol 2.11 2.49 ethyl acetate toluene 3.18 1.09 2.21 1.65 (S)-(-)-limonene (R)-(+)-limonene(S)-(-)-a-pinene 1.54 1.41 1.46 1.54 1.52 1.72 (R)-(+)-a-pinene (S)-(-)-fl-pinene 0-10096 R.H." 1.43 1.37 1.09 1.7 1.3 1.Y2 a R.H. =relative humidity.286 J. Mater. Chem., 1996,6(3), 285-288 The racemic forms of limonene and pinene are illustrated in Fig. 4.The responses of Cd2Sbz06,8at 450 "Cto these vapours are illustrated in Fig. 5 and 6. Fig, 5 shows that the sensors show significant variations in response to the (R)and (S)forms of lhonene, the (S) response being the greater. This is an unusual phenomenon as the two forms have almost identical volatilities and are approximately equivalent in size. One possible explanation for this is that the porosity of the films may exclude the (R)form from penetrating as deeply into the film as the (S)form. An SEM image of the Cd,Sb20,.8 sensor surface is shown in Fig. 7. The high porosity of the films can be seen clearly. However, the Sn02 films show an equivalent porosity but the chiral differences in response are not observed.It should also be noted that (S)-limonene is very slightly more volatile than (R)-limonene and this variation may be magnified on the sensor surface. These response variations become more pronounced with increasing temperature up to the maximum response temperature of 450 "C. Above this temperature the responses and the (R)and (S) differences decrease rapidly. In the case of pinene (Fig. 6), the response characteristics for the (R)-a and (S)-a forms are almost identical for the temperature range used. However, it was observed that the (S1-p form Yields significantly dlanced responses. This is Fig. 4 Racemic isomers of limonene (orange oil) and pinene (pine oil) 3.0-2.5-4O -z4 2.0-1.5-i.oJ 0 5 10 15 20 limonene (ppm) Fig.5 Limonene response data for Cd2Sb206,8 sensor at 350 "C. (a)(S)-Limonene;(b)(R)-limonene. 5 4-I/___ I0-0 5 10 15 20 pinene (ppm) Fig. 7 SEM image of Cd2Sbz06.g sensor surface (magnification x 200000) rather less surprising than with the case with limonene, because structural differences between the o! and p forms could easily account for the variation in response. This is despite the fact that, as with limonene, the volatilities of all three forms are almost identical. Again, as with limonene, the responses reach their peak at 450°C,with a subsequent sharp decline at elevated temperatures. Table 2 illustrates the variations in response for limonene and pinene (20ppm) at the peak response temperature.All the responses discussed were repeat- able to a high degree. Note that for limonene and pinene the responses vary linearly with vapour concentration. The cause of this will be discussed. Responses to other organic vapours The response data for ethanol, ethyl acetate and toluene are shown in Fig. 8. The prime feature of note is that, in contrast to the terpenes, ethanol and ethyl acetate display square-root- dependent response characteristics. Therefore it would appear that a different response and/or adsorption mechanism is in operation for the two sets of compounds. Toluene also exhibits a square-roo t response dependence, although data for higher concentrations is lacking.The maximum responses for both ethanol and ethyl acetate are seen at 400°C. Mechanistic discussion It is well known from empirical gas-sensing studiesI4 that the conductance change, Ag, induced by the introduction of a particular gas of partial pressure PRis given by the relationship: Ag =aPvR (4) Table 2 Peak response data for CdzSb206.8 sensor exposed to 20 ppm of various aromatic terpenes aromatic terpene maximum 1/1, Ta/T (S)-(-)-limonene(R)-(+)-limonene (S)-(-)-a-pinene (R)-(+)-a-pinene (S)-(-)-/?-pinene 3.64 3.18 2.78 2.8 1 5.64 450 450 450 450 450 Fig. 6 Pinene response data for Cd2Sb@,,, sensor at 350°C. (a)(S)-8-pinene; (b) 0, a(S)-a-pinene; M, (R)-a-pinene. Temperature of maximum response.J. Mater. Chem., 1996, 6(3), 285-288 287 in which a is normally a constant for a given type of substrate of order of magnitude and v is said to characterise the order of the surface reaction that dominates induced changes in surface charge-carner density This index is frequently fractional for semiconducting oxides Many workers have recently reported a square-root dependence of the electrical response upon the partial pressure of a reducing gas in dry air and moist air utilising materials such as SnO, (and metal- doped Sn0,),16 chromium titanium oxide (a p-type binary oxide),17 ZnO and ZnO/CuO contact ceramics In this work, the dependence of the electrical response upon the partial pressure of the reducing gas differs between ethanol and ethyl acetate (dependence on the square-root of the partial pressure) and the larger hydrocarbons limonene and pinene (linearly dependent upon the partial pressure) Note that the large hydrocarbons can undergo catalytic dehydrogenation (as a first stage, say, in their decomposition) whereas ethanol and ethyl acetate cannot readily undergo dehydrogenation and are more likely to proceed uzaelimination of water C2HSOH+C2H, + H2O (5) or other small organic molecules Kohl and co-workers have noted the range of product^'^ that, in the case of n-type oxides,20 can be observed from the surface interactions of even small adsorbed molecules Decomposition reactions do not necessarily have a direct bearing on changes in charge-carrier density, however, oxidation of decomposition products by surface-bound oxygen anions can be of key importance 2o We can illustrate the process of direct combustion by using the model of McAleer et al k2RO,,, + n' -Re, + 0, (7) with the requirements that the rate of release of the surface oxygen monoanion (Os-) is very much less than its rate of reaction with the reducing gas, R, and that the rate of formation of Os-is very much larger than its rate of reaction with R Species other than Os-(eg O,s-and 02s2-)can also be utilised in similar simple schemes In our work R(g) could be a species in the air that we are aiming to detect However, we suggest that R(g) is more likely to be one of the species produced by catalytic decomposition (such as H2 or C2H4) that has desorbed subsequent to the decomposition process and can react readily with surface- bound oxygen ions, thereby generating 'free' electron charge carners (n') at the surface The large hydrocarbons limonene 10 91 / 8-7-6-5-41 /I Fig.8 Response data for CdzSbzOss sensor at 400°C exposed to (a)ethyl acetate, (h)ethanol and (c)toluene 288 J Muter Chem , 1996,6(3), 285-288 and pinene can dehydrogenate to release hydrogen and a hydrocarbon with more double bonds This IS especially appro- priate for limonene, for by such a process a planar aromatic molecule is formed We suggest that the observed difference in behaviour between these hydrocarbons and the small, oxygen- containing species arises from the combustion of hydrogen released by catalytic dehydrogenation of the hydrocarbons Experiments using thermal desorption mass spectrometry must be performed before any further comments can be made Conclusions Screen-printed thick-film sensors were produced utilising an n-type semiconducting ceramic, Cd,Sb,O, 8, as the active material The responses of these sensors were compared with those of SnO, Significant differences in response were obtained for ethyl acetate and toluene An important difference was the lack of sensitivity of Cd2Sb2068 to humidity changes This could prove useful in a commercial gas sensing device The responses of Cd,Sb,O, to optically active terpenes were unusual in the fact that in the case of limonene the (S) form gave larger responses than the (R)form In the case of pinene the B form gave the larger responses This behaviour was not displayed by SnO, This property could have potential practi- cal usage in distinguishing optical isomers, where polansed light is currently employed To gain a clearer picture other optically active compounds will be investigated The response characteristics of the vapours tested for were found to differ with respect to vapour concentration The terpene responses were linear with respect to concentration and the smaller hydrocarbons displayed a square-root depen- dence It is suggested that dehydrogenation could be the key to these differences This will be actively investigated using mass spectrometric methods References C Wagner, J Chem Phys , 1950,18,69 K Hauffe and H J Engell, Z Electrochem , 1952,56,366 N Taguchi, US Pat, 3 695 848,1972 H Windischmann and P Mark, J Electrochem SOC, 1979, 126, 627 5 J Watson, Sensors Actuators, 1984,5,29 6 J F McAleer, P T Moseley, J 0 W Norris, D E Williams and B C Tofield, J Chem SOC Faraday Trans 1,1988,84441 7 K Nakashima and S Suzuki, Anal Chim Acta, 1984,162,153 8 T Arakawa, Chem Ind (Dekker), 1993,50,361 9 T Arakawa, N Ohara, H Kurachi and J Shiokawa, Anal Chem Symp Ser , 1983,17,159 10 R A McCauley, J Appl Phys, 1980,52,290 11 B Li and J Zhang, J Am Ceram SOC ,1989,72,2377 12 B Li, J Am Ceram SOC , 1988,71, C78 13 P T Moseley, A M Stoneham and D E Williams, in Techniques and Mechanisms in Gas Sensing, ed P T Moseley, J 0 W Norris and D E Williams, Adam Hilger, Bristol, 1991 14 P Van Geloven, J Moons, M Honore and J Roggen, Sensors Actuators B, 1989,17, 361 15 J Vetrone, Y W Cheung, R Covicchi and S Semancik, J Appl Phys, 1993,73 8371 16 P T Moseley, Sensors Actuators B, 1991,3,167 17 D H Dawson, G S Henshaw and D E Williams, Sensors Actuators B, 1995,26,76 18 S T Jun and G M Choi, Sensors Actuators B, 1994,17, 175 19 D Kohl, W Thoren, U Schnakenberg, G Schuill and G Heiland, J Chem SOC Faraday Trans, 1991,87,2647 20 D Kohl, Sensors Actuators B 1990,1,158 21 J F McAleer, P T Moseley, J 0 W Norm and D E Williams, J Chem SOC Faraday Trans 1 1987,83,1323 Paper 5/04833H, Received 21st July, 1995